focus on deepstar® global deepwater technology - satodev

Volume 47 Number 3 May/June 2013

The International, Interdisciplinary Society Devoted to Ocean and Marine Engineering, Science, and Policy

Journal

Focus on DeepStar® Global Deepwater Technology Development

EXHIBIT SPACESpace is filling up, but many good locations are still available. Visit the interactive Oceans 2013 Exhibit Hall Floor Plan at:oceans13mtsieeesandiego.org/exhibitor/floorplan and pick your spot. Or contact Sue Kingston, IEEE Exhibits Manager, for more information. e-mail: [email protected] • office: 1 (310) 937-1006 • cell 1 (310) 699-2609

• Twelve Co-participating Societies

• Academic Host: Scripps Institution

of Oceanography/UCSD

• Over 250 exhibits in two halls, plus

Academic Row

• Opening Plenary with Ocean VIPs

• Extensive technical and scientific

tracks and sessions

• Ultra-Deep Ocean Track with

renowned presenters

• Tutorials with college credit

• Student Opportunities: Discounted

Rates, Session Chairs, Poster Competition

• Golf Tournament

• Underwater Film Festival on two nights

• Gala on the Carrier U.S.S. Midway

• A celebration of the MTS 50th Anniversary

• STEM Educators Workshop

OCEANS‘13 CONFERENCE HIGHLIGHTS INCLUDE:

VISIT US ONLINEOCEANS13MTSIEEESANDIEGO.ORG

LIKE US ON FACEBOOK:

EXHIBITORSGood spaces are still left, but going fast!

ATTENDEESRegistration opens May 20!

/Oceans2013

AcAdemic SponSor:

Front Cover: DeepStar® cover graphic, artwork, and logo is a registered trademark of Chevron Intellectual Property LLC.Back Cover: All images can be found within this issue. Top row (l-r): Schmidt et al., Figure 6; Cui, Figure 4; 2nd row: Vedachalam et al., Figure 4; 3rd row (l-r): Yoon & Na, Figure 3; Vedachalam et al., Figure 15; Cooper et al., Figure 1; Bottom row (l-r): Jacobson, Figure 3; Peirano, Figure 2.

Focus on DeepStar5DeepStar®: 22 Years of Success DeepStar: A Global Deepwater Technology Development ProgramGreg Kusinski, Art J. Schroeder, Jr., James E. Chitwood

13DeepStar 11304: Laying the Groundwork for AUV Standards for Deepwater FieldsJohn Jacobson, Pierce Cohen, Amin Nasr, Art J. Schroeder, Jr., Greg Kusinski

19DeepStar Metocean Studies: 15 years of DiscoveryCortis Cooper, Tom Mitchell, George Forristall, James Stear

27Recent Development in IR Sensor Technology for Monitoring Subsea Methane DischargeMark Schmidt, Peter Linke, Daniel Esser

General Papers37Development of the Jiaolong Deep Manned SubmersibleWeicheng Cui

55Reliability-Centered Development of Deep Water ROV ROSUB 6000Vedachalam Narayanaswamy, Ramesh Raju, Muthukumaran Durairaj, Aarthi Ananthapadmanabhan, Subramanian Annamalai, Gidugu Ananada Ramadass, Malayath Aravindakshan Atmanand

Volume 47, Number 3, May/June 2013

Focus on DeepStar® Global Deepwater Technology Development

72Anchor Drop Tests for a Submarine Power-Cable ProtectorHan-Sam Yoon, Won-Bae Na

81Elements of Underwater Glider Performance and StabilityShuangshuang Fan, Craig Woolsey

99Analysis of Underwater Acoustic Communication ChannelsSadia Ahmed, Huseyin Arslan

118Wrecks on the Bottom: Useful, Ecological Sentinels?Andrea Peirano

In This Issue

The Marine Technology Society Journal (ISSN 0025-3324) is published by the Marine Technology Society, Inc., 1100 H Street NW, Suite LL-100 Washington, DC 20005

Subscription Prices (all prices include shipping): MTS members can purchase the printed Journal for $32 domestic and $110 international. Non-members and library subscriptions are $435 online only (includes 13 years of archives); $460 for online with print (includes 12 years of archives); $135 print only within the U.S. and Canada, $200 print only international. Single-issue (hardcopy) is $20 plus $7.50 S&H (domestic), $24.50 S&H (international); Pay-per-view (worldwide): $15/article. Postage for periodicals is paid at Washington, DC and additional mailing offices.

P O S T M A S T E R :Please send address changes to:

Marine Technology Society Journal1100 H Street NW, Suite LL-100Washington, DC 20005Tel: (202) 717-8705; Fax: (202) 347-4305

Copyright © 2013 Marine Technology Society, Inc.

Text: SPi Cover and Graphics:Michele A. Danoff, Graphics By Design

BOA RD OF DIREC T ORSPresidentDrew MichaelROV Technologies, Inc.President-electRichard W. SpinradOregon State UniversityImmediate Past PresidentJerry BoatmanQinetiQ North America – Technology SolutionsGroupVP—Section AffairsLisa MedeirosGeospace Offshore CablesVP—Education and ResearchJill ZandeMATE CenterVP—Industry and TechnologyRaymond TollComputer Sciences CorporationVP—PublicationsDonna KocakHARRIS CorporationTreasurer and VP—Budget and FinanceDebra KillInternational Submarine EngineeringVP—Government and Public AffairsJustin ManleyTeledyne Benthos

SEC T IONSCanadian MaritimeVacantGreat Lakes TBD Gulf CoastLaurie JuganConsultantHampton RoadsRaymond TollComputer Sciences CorporationHawaiiStu BurleyStrategic Theories UnlimitedHoustonAimee MarshEICJapanProf. Toshitsugu SakouTokai UniversityMid-Gulf Coast-Florida Erica MoultonMATEMontereyJill ZandeMATE CenterNew EnglandChris JakubiakUMASS Dartmouth-SMASTNorwayTBDNewfoundland and LabradorGary DinnMemorial UniversityOregonJeremy ChildressOregon State UniversityCollege of Oceanic & Atmospheric SciencesPuget SoundFritz StahrUniversity of WashingtonSan DiegoDavid VelascoSonTek/YSISouth KoreaDr. Seok Won HongMaritime & Ocean Engineering Research Inst.(MOERI/KORDI)Washington, D.C.Capt. Dennis “Mike” Egan, P.E.EESS LLC

P ROF E S SION A L C OMMI T T EE SIndustry and TechnologyBuoy TechnologyDr. Walter PaulWoods Hole Oceanographic InstitutionCables and ConnectorsVacantDeepwater Field Development TechnologyDr. Benton BaughRadoil, Inc.DivingMichael LombardiOcean OpportunityMichael MaxHydrate Energy International LLCDynamic PositioningHoward ShattoShatto EngineeringManned Underwater VehiclesWilliam KohnenHYDROSPACE GroupMooringsJack RowleySAICOceanographic InstrumentationDr. Carol JanzenSea-Bird Electronics, Inc.Offshore StructuresDr. Peter W. MarshallMHP Systems EngineeringRemotely Operated VehiclesChuck RichardsC.A. Richards and Associates, Inc.Ropes and Tension MembersEvan ZimmermanDelmar Systems, Inc.Seafloor EngineeringCraig EtkaScorpio Ventures, LLCUnderwater ImagingDr. Fraser DalgleishHarbor Branch Oceanographic InstituteUnmanned Maritime VehiclesRafael MandujanoVehicle Control Technologies, Inc.

Education and ResearchMarine ArchaeologyDan WarrenC & C TechnologiesMarine EducationErica MoultonMATE CenterMarine Geodetic Information SystemsDave ZilkoskiNOAAMarine MaterialsCapt. Dennis “Mike” EganEESS, LLCOcean ExplorationShane ZiglerOceanGate, Inc.Physical Oceanography/MeteorologyDr. Richard L. CroutNational Data Buoy CenterRemote SensingRyan VandermeulenUniversity of Southern Mississippi

Government and Public AffairsMarine Law and PolicyVacantMarine Mineral ResourcesDr. John C. WiltshireUniversity of HawaiiMarine SecurityDallas MeggittSound & Sea TechnologyOcean Economic Potential

James MarshUniversity of HawaiiOcean Observing SystemsDonna KocakHARRIS CorporationOcean PollutionRyan MortonAnadarko PetroleumRenewable EnergyMaggie L. MerrillMarine Marketing Services and New England MREC at UMass DartmouthKenneth BaldwinUniversity of New Hampshire

S T UDEN T SEC T IONSAlpena Community CollegeCounselor: TBDBrigham Young UniversityCounselor: TBD College of William & MaryCounselor: Heather MacDonaldDalhousie UniversityCounselor: Tejana Ross, Ph.D.Duke UniversityCounselor: Douglas Nowacek, Ph.D.Florida Atlantic UniversityCounselor: Douglas A. Briggs, Ph.D.Florida Institute of TechnologyCounselor: Stephen Wood, Ph.D., P.E.Long Beach City CollegeCounselor: Scott FraserMarine Institute (MI)Counselor: TBD Massachusetts Institute of TechnologyCounselor: Alexandra Techet, Ph.D.Memorial University (MUN)Counselor: TBD Monterey Peninsula College/Hartnell CollegeCounselor: Jeremy R. HertzbergNortheastern UniversityCounselor: Joseph Ayers, Ph.D.Oregon State UniversityCounselor: R. Kipp Shearman Texas A&M University—College StationCounselor: Robert Randall, Ph.D.Texas A&M—Corpus ChristiCounselor: Dugan Um, Ph.D.Texas A&M University—GalvestonCounselor: David Talent, Ph.D.United States Naval AcademyCounselor: Cmdr. David J. RobillardUniversity of HawaiiCounselor: Dr. Reza GhorbaniUniversity of HoustonCounselors: Raresh Pascali, P.E., Chuck RichardsUniversity of North Carolina—CharlotteCounselor: James Conrad, Ph.D.University of Southern MississippiCounselor: Stephen Howden, Ph.D. University of WashingtonCounselor: Rick RupanWebb InstituteCounselor: Matthew Werner

HONOR A RY MEMBERS†Robert B. Abel†Charles H. BussmannJohn C. Calhoun, Jr.John P. Craven†Paul M. FyeDavid S. Potter†Athelstan Spilhaus†E. C. Stephan†Allyn C. Vine†James H. Wakelin, Jr.†deceased

Marine Technology Society Officers

50 years of excellence.

Your assistance is needed!

e e co e our contri u ons o anecdotesersona re e rances and ictures

MTS Needs Your Assistance!

Ann E. Jochens, Ph.D.EditorTexas A&M University

Brian Bingham, Ph.D.University of Hawaii at Manoa

Donna KocakHARRIS Corporation

Scott Kraus, Ph.D.New England Aquarium

Dhugal Lindsay, Ph.D.Japan Agency for Marine-Earth Science & Technology

Justin ManleyTeledyne Benthos

Eugene MassionMonterey Bay Aquarium Research Institute

Jill ZandeMATE Center

EditorialDonna KocakVP of Publications

Ann E. Jochens, Ph.D.Editor

Amy MorganteManaging Editor

AdministrationDrew MichelPresident

Richard LawsonExecutive Director

Chris BarrettDirector of Professional Development and Meetings

Jeanne GloverMembership and Marketing Manager

Allan GrayFinancial Operations Manager

Jacob SobinMember Groups and Student Outreach Manager

Suzanne VoelkerSubscription Manager

Editorial BoardThe Marine Technology Society is a not-for-profit, international professional society. Established in 1963, the Society’s mission is to promote the exchange of information in ocean and marine engineer-ing, technology, science, and policy.

Please send all correspondence to:The Marine Technology Society1100 H Street NW, Suite LL-100Washington, DC 20005(202) 717-8705 Tel. (202) 347-4305 FaxMTS Journal: [email protected]: [email protected]: [email protected]: [email protected]: [email protected]: www.mtsociety.org

MEMBERSHIP INFORMATIONmay be obtained by contacting the Marine Technology Society. Benefits include:■ Free subscription to the online Marine

Technology Society Journal, with highly reduced rates for the paper version

■ Free subscription to the bimonthly news -let ter, Currents, covering events, business news, science and technology, and people in marine technology

■ Member discounts on all MTS publications■ Reduced registration rates to all MTS and

MTS-sponsored conferences and workshops■ Member-only access to an expansive Job

Bank and Member Directory■ Reduced advertising rates in MTS

publications■ National recognition through our Awards

ProgramIndividual dues are $75 per year. Life membership is available for a one-time fee of $1,000. Patron, Student, Emeritus, Institutional, Business, and Corporate memberships are also available.

ADVERTISINGAdvertising is accepted by the Marine Tech nology Society Journal. For more informa-tion on MTS advertising and policy, please contact Mary Beth Loutinsky, [email protected]

COPYRIGHTCopyright © 2013 by the Marine Technology Society, Inc. Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by the Marine Technology Society, provided that the base fee of $1.00 per copy, plus .20 per page is paid directly to Copyright Clearance Center, 222 Rosewood Dr., Danvers, MA 01923.

For those organizations that have been granted a photocopy license by CCC, a sepa-rate sys tem of payment has been arranged. The fee code for users of the Transactional Reporting Service is 0025-3324/89 $1.00 + .20. Papers by U.S Government employees are declared works of the U.S. Government and are therefore in the public domain.

The Marine Technology Society cannot be held responsible for the opinions given and the statements made in any of the articles published.

ABSTRACTSMTS Journal article abstracts, if available, can be accessed for free at http://www.ingentaconnect.com/content/mts/mtsj.

Print and electronic abstracts of MTS Journal articles are also available through GeoRef <http://www.agiweb.org/georef/>, Aquatic Sciences and Fisheries Abstracts, published by Cambridge Scientific Abstracts <http://www.csa.com/factsheets/aquclust-set-c.php>, and Geobase’s Oceanbase published by Elsevier Science.

CONTRIBUTORSContributors can obtain an information and style sheet by contacting the managing editor. Sub missions that are relevant to the concerns of the Society are welcome. All papers are sub-jected to a stringent review procedure directed by the editor and the editorial board. The Journal focuses on technical material that may not otherwise be available, and thus technical papers and notes that have not been published previously are given priority. General commen-taries are also accepted and are subject to review and approval by the editorial board.

4 Marine Technology Society Journal

I N T R O D U C T I O N

DeepStar®: 22 Years of SuccessDeepStar: A Global Deepwater TechnologyDevelopment Program
A U T H O R SGreg KusinskiChevron Energy TechnologyCompany andDeepStar Director
Art J. Schroeder, Jr.Energy Valley, Inc., andDeepStar Consultant

James E. ChitwoodChitwood Engineering andDeepStar Consultant

Introduction

The oil and gas industry has longestablished a collaborative approachto reduce risks and optimize results,especially in the area of technologydevelopment for major explorationand production (E&P) projects. In1991, a visionary group of industryoperators laid the foundation forwhat is DeepStar® today, a highlysuccessful initiative for technical col-laboration to enhance deepwater ex-ploration, drilling and production.The Chevron-managed, consensus-driven DeepStar consortium hasevolved steadily during the past twodecades to meet ever increasing tech-nical challenges of deepwater devel-opment and production. However,it has never wavered from its coremission to provide our industry withan open platform to investigatecomplex technical challenges andcollaboratively develop safe, viableso lut ions . This ar t ic le out l inesDeepStar’s strategies and overviews

its organization, operation, and direc-tion. Some additional detail is providedon historical major accomplishmentsas well as current research activitiesand a vision of the future for deepwatersuccess.

DeepStar OverviewDeepStar is an operator-driven and

funded research and development(R&D) collaboration between oilcompanies, vendors, regulators, andacademic/research institutes started in1991. Within the overlap of operatormember-driven, collaborative spacedeepwater technology needs, DeepStarserves as a well-proven tool to lever-age resources of a multitude of scarcesources. Leverage includes■ direct hard-dollar funding of third-

party R&D suppliers;■ DeepStar member internal techni-

cal expertise, subject matter experts(SMEs) from both Participants (op-erators) and Contributors (manufac-tures, service companies, universities,engineering firms, etc.);

■ in-kind contributions including pre-vious work in the field of interest;

■ potential field test/demonstrationopportunities;

■ a well-honed, effective and efficientset of processes and procedures;

■ a respected track record of success in-terfacingwith regulatory agencies andstandards-developing organizations.Figure 1 outlines the role DeepStar

occupies to meet operator members’collaborative technology needs.

May

With DeepStar membership includ-ing over 80 organizations employingwell over 1 million people and operat-ing in all of the world’s deepwater andultradeepwater basins, it is well posi-tioned to ensure members are focusedon the right technical challenges andleveraged to produce meaningful andvalue-added results. See Figure 2 for alisting of current members.

DeepStar has a proven and long-term record for delivery of value toits membership. Its mission is to facil-itate a cooperative, globally alignedeffort focused on identification anddevelopment of economically viablemethods to drill, produce, and trans-port hydrocarbons from deepwater.DeepStar’s current Phase XI programis focused on global deepwater develop-ment in water depths to 10,000+ feet.As the premier global forum to de-fine deepwater technology needs, itprovides value to its membership byleveraging financial and technical re-sources to■ define and rank important deep-

water technology needs,■ deliver technologies via a well honed

stage-gate process,■ bu i l d d e e pwa t e r t e c hn i c a l

competency,■ adopt and deploy deepwater

technologies.DeepStar has established goals to

■ enhance existing deepwater tech-nologies by improving the safety,operability, flexibility, reliability,and profitability of existing deep-water production systems;

/June 2013 Volume 47 Number 3 5

FIGURE 1

Collaborative space deepwater R&D needs.

Operator Members (Participants) have desire forcollaborative funding where facilities & deepwatertechnologies will not provide a competitive advantage(This was the original driver that started DeepStar.)

■ DeepStar is a proven well leveraged tool in a collaborativespace

■ DeepStar is an Operator-Driven Program(e30 Multi-Discipline Projects).

■ DeepStar is a recognized Industry collaborative voice.

■ There is an opportunity for members to benefit fromDeepStar by participating in fewer – but larger and “better”JIPs


FIGURE 2

DeepStar is a consortium membership; Phase XI (as of 4/23/2013).

■ develop new enabling deepwatertechnologies by advancing noveltechnologies to allow productionin areas that are currently techni-cally unproven with the ultimategoal of developing technologies re-quired for economic production in12,000+ feet water depth;

■ gain acceptance of deepwatertechnologies by regulators andindustry by facilitating the devel-opment of industry standards andpractices as appropriate and foster-ing communications with regula-tory agencies;

■ provide a forum and process fordiscussion, guidance and feed-backwith contractors, vendors, op-erators, regulators and academia,regarding deepwater technology,and promote standardization ofcomponent interfaces.These goals are accomplished via

the following DeepStar executionstrategies:■ technology development aligned

with business needs;■ transfer and apply technology to

deepwater assets;■ gain acceptance of deepwater tech-

nologies by industry, standards orga-nizations, and regulators;

■ focus on the front end of the technol-ogy development cycle by advancingcritical fundamental knowledge(science), providing proof of conceptsand performing techno-economicengineering studies.While the manufacturing and service

sector members compete fiercely, mostDeepStar work is considered a “non-competitive” area between the operatorsand thus a suitable space for collabora-tion. This also includes projects that sup-port reliable and safe operations. Froman operator perspective, there is highvalue in collaborating on these topics,as an incident with one operator can

cause a ripple affect across the entireindustry. As such, DeepStar provides anoutstanding opportunity for SMEs to col-laborate and foster the exchange of ideas.

DeepStar OrganizationDeepStar is managed by Chevron

via a contractual arrangement with itsparticipating organizations, utilizingChevron back-office support servicesincluding legal, procurement, financ-ing, accounting, and administrative.Chevron also provides a Director anda technical staff of two industry vet-erans to support the consortium.DeepStar, with 11 deepwater operat-ing companies and approximately 80contributing member companies, isthe world’s longest running deepwaterconsortium. One thousand plus tech-nical and management committee vol-unteers coupled with straightforward,well-honed and proven policies and pro-cedures also make it one of the mostsuccessful. The SMEs are organizedfunctionally into Technical Committeeswith operator chairs (and co-chairs) asshown in Figure 3.

The Technical Committees’ roleis to identify technology needs andthen generate “ideas”—captured ona cost, time, and resources (CTRs)form for Management Committee re-view. Once approved and bid out,their role is to review and recommendawards to the DeepStar Director.Once awarded, their role is to overseeand technically guide the execution ofthe work. The Technical Committeesalso table discussions/presentations ofnew technologies and/or challengesthat could potentially be addressedwithin the DeepStar program. Timespent by the various Committee mem-bers on Technical Committee activi-ties is contributed by their respectivecompanies. The Technical Commit-

May

tees are currently charged as outlinedbelow.

Geosciences Committee (×000)addresses challenges in prospectand reservoir delineation, character-ization and surveillance, which areeither specific to deepwater prov-inces or which relate to appraisal anddevelopment decisions leading tosignificant economic impact in suchprovinces.

Regulatory Committee (×100)provides liaison between DeepStartechnical committees and governmen-tal regulators such as the Bureau ofSafety and Environmental Enforce-ment (BSEE) and the U.S. Coast Guard.The objective of the committee centerson the exchange of technical informa-tion between the working technicalgroups inDeepStar and regulatory rep-resentatives. The Regulatory Commit-tee also works and communicates withleading industry organizations, such asthe Offshore Operators Committee(OOC), American Petroleum Institute(API), and others. The objective ofthis interface between DeepStar andother parties is the expedient full re-view of relevant issues involved in de-ployment of new tools or processesunder development.

Flow Assurance Committee(×200) goal is to assure reliable andeconomic production in deepwaterby appropriate design and operationthrough prediction of fluid behavior,flow management, and remediationof deposition and line plugs from thereservoir to the point of sale.

Subsea Systems Committee(×300) goal is to develop technologyand qualify equipment to enable thedeployment and IMR of “subsea facil-ities for 50+ mile tie-backs in 10,000foot water depth.”

Floating Systems Committee(×400) role is to further technology


and fill gaps related to deepwater float-ing systems and their associated moor-ings and risers.

Drilling, Completions, and In-tervention Committee (×500) mem-bers share their experiences and data toidentify technology improvements indeep water drilling, completion, andwell intervention operations.

Administration and TechnicalSupport (×600) — The ProgramDirector heads this committee andoversees DeepStar’s leadership and ad-ministration responsibilities.

Reservoir Engineering Commit-tee (×700) looks at trends that are ofsignificant generic interest to the in-dustry, while avoiding detailed reser-voir issues where participants havecompetitive concerns. The committeehas assembled the best available publicdomain information for use in learn-


ing global characteristics of deepwaterreservoirs.

Met-Ocean Committee (×800)goal is to improve knowledge andmodeling capabilities of ocean cur-rents, waves and wind, which will pro-vide more accurate facility designcriteria and reduce downtime duringdeepwater drilling operations.

Systems Engineering Committee(×900) has the responsibility for ana-lyzing existing and/or potential tech-nology gaps, especially at the facilitysystems level and bringing them to theattention of the committee best suitedto carry out further work in the specifictechnology area.

A huge benefit to members is theinterconnectivity and continuity thestaff bring by linking together the piecesof the deepwater technology puzzle asshown in Figure 4.

Operational FrameworkThe DeepStar R&D projects are

funded by operator fees, and the deminimus cash fee from nonoperatormembers helps offset meeting costs.While all member classes may proposeprojects, the final funding decisions aretaken by the operators, known withinDeepStar as Participants. This makessense not only by the fact they are theones funding projects but they are alsothe “voice of the customer” for theman-ufacturing, engineering, and servicecompany members, known as Contrib-utors. The entire project selection pro-cess provides the Contributors with avery clear understanding of their cus-tomers’ needs and priorities in the col-laborative space (Figure 5).

A l l DeepSta r p ro j e c t s mus thave an operator Champion. Cham-pions, along with DeepStar staff, are

FIGURE 3

DeepStar organization.

responsible for the execution of theirprojects and are held accountable forthe deliverables (results, schedule, andcost). A subcommittee, or WorkingCommittee, is usually formed to sup-port the Champion in executing the

project and providing peer guidance.With DeepStar staff assistance, Cham-pions and their Working Committeesfulfill the following roles for each ap-proved CTR:■ refine scope of work,

May

■ define costs,■ implement CTR,■ select contractors (bid process),■ submit justification for manage-

ment approval,■ insure contract implementation,■ arrange kickoff meeting,■ arrange for progress meetings,■ reviewmonthly reports and provide

periodic updates to the TechnicalCommittee,

■ maintain consensus,■ approve and confirm contractor

invoices,■ approve Final Report.

While DeepStar projects can spanthe development spectrum, character-istics of a typical DeepStar project areas follows:■ operator championed,■ lower technology readiness level

(i.e., TRLs 1–3 on a scale of 1–7),■ low cost (200 k–500 k USD),■ shorter duration (6–18months), and■ multiphase stage-gated.

DeepStar SuccessesDeepStar has successfully identi-

fied and executed hundreds of R&Dprojects over its 11 phases, which havesubsequently enabled member compa-nies to achieve their business objectives.Some of the successes have been in thefollowing technical arenas:■ Poly rope: development, recom-

mended practices, standards andregulatory approval;

■ FPSOs: standards and regulatoryapproval;

■ Vortex-Induced Vibration: under-standing, prediction, mitigationand control;

■ Met-Ocean: understanding, pre-diction and design practices andstandards;

■ Promulgator of standards andregulations;

FIGURE 4

DeepStar technical committees span the deepwater technology puzzle.

FIGURE 5

DeepStar project funding roadmap.


■ Flow assurance management, in-cluding modeling, operations,remediation.As DeepStar members (not Deep-

Star itself ) sell kit and services, thenonmember may have difficulty inimmediately seeing the underlyingDeepStar value-add. Outlined in Fig-ure 6 are some of the standards andrecommended practices emanatingfrom DeepStar work in the equip-ment area.

Examples of areas where DeepStarwork in the modeling and simulationarena has given rise to commercialsolutions are presented in Figure 7.

Continually being at the forefrontof deepwater developments, Deep-Star, in its various studies, has set thegroundwork on many of the safetystandards we know today. In particu-lar, DeepStar drove the developmentof the Technology Qualification Pro-cess (TQP), which much of indus-try leverages for their R&D efforts toaccurately represent the maturity ofnew technology. Additionally, muchof the DeepStar work that led to vari-ous standards and recommendedpractices has since been promulgated

10 Marine Technology Society Journa

to governmental agency rules and reg-ulations, some of which are listedin Figure 8.

Current Phase XIFocus Areas

The DeepStar process utilizes anoperator-driven CTR tool to help iden-tify projects of interest. The CTRs areoperator vetted and ranked at theTechnical Committee level and thenprioritized by the Senior Advisor Man-

l

agement Committee. Current phaseprojects can be grouped as shown inFigures 9 and 10.

DeepStar, the FutureWhile Neils Bohr is quoted as

saying, “Prediction is very difficult,especially if it’s about the future,”it is clear that DeepStar’s future isbright as it will continue to evolveand remain relevant, following astrategy finely honed over the past20 years. First, there is no shortage of“needs” for new technologies in thecollaborative space. Post-Macondo,the industry is placing even moreemphasis on building and maintain-ing sustainable Integrity Managementprograms with particular focus on in-clusion addressing low probability,high consequence events. Additionalefforts are also expected in the devel-opment of standards and protocolsleading to greater reliability, better/more effective interfaces between in-dividual components and equipmentmodules, and more effective mainte-nance procedures and practices, alladding up to a safer overall operatingenvironment. These strategies should

FIGURE 6

DeepStar equipment programs.

■ ROV Interfaces InAPI 17H■ Subsea Equipment Qualification API 17N■ HIPPs for the GOM InAPI 17O■ Qualification of Large Bore HP Valves Updated API 17D■ DW Pipeline Repair & Tie-Ins DW RUPE■ Pipeline Limit State Design Codes API RP 1111■ Subsea Multiphase Flow Meters API RP 85■ Drilling Riser Recommended PracticeUpdated API 16Q

FIGURE 7

DeepStar modeling and simulation tools.

■ Life Cycle Well Cost Control Evaluation Tool Commercial■ Well Control Software Commercial■ Reservoir Characterization and Performance Database Commercial■ GOM Current & MetOcean Forecast Models Commercial■ Hydrate Plugging in Oil Model Commercial■ Hydrate Disassociation Model Benchmarking■ Asphaltene Deposition Model and Lab Test Benchmarking■ Annulus Pressure Buildup for XHPHT Wells Commercial

also yield an overall lower total costof ownership.

DeepStar’s current Managementand Technology Committee structurewill continue as is and its time-honedprocesses have served its membershipwell, directing funds and resources to

those collaborative areas most highlyranked by the operators. Its “holistic”approach with eight Technical Com-mittees provides a synergy that isgreater than just the sum of the indi-vidual committees. Generally, Deep-Star will continue with the successful

May/J

CTR “bottoms up” need-driven pro-cess with strategic overarching top-down direction. The needs will beboth near term (to 5 years) and longerterm (to 10 years). As such, DeepStarwill continue to fund “individual proj-ects” within all the Technical Com-mittees that are operator driven andanticipates adopting a few overarch-ing areas of focus with an emphasis onStandards where appropriate. Nearterm one could also expect projects sup-porting better appraisal tools, IncreasedOil Recovery (IOR), particularly in theLower Wilcox, Subsea processing andDrilling, Completion and Interventionprojects focused on supporting saferoperations with a lower overall cost.

The Management Committeewill be encouraging “Bigger Impact”projects that are conducted in a morecollaborative manner, particularly withlarger Contributors. To better facilitatethis outcome, there will be some sensi-ble changes to the standard DeepStarIntellectual Property restrictions basedon overall balance of value delivered toproject. DeepStar traditionally providesthe collective “Voice of the Customer”to member service and manufacturingcompanies, allowing them to providebetter focused, more reliable servicesand products at a lower cost. Leveragingon this role,DeepStar is askingmembersto consider DeepStar as an alternative ofchoice versus spinning-up new or par-ticipating in other “one-off” Joint In-dustry Projects (JIPs).

DeepStar expects to continueinteraction with regulators to ensureDeepStar-developed technologies canbe readily accepted for deploymentand use. The strategy will continuepresenting a well-defined operationalneed and DeepStar’s stage-gate tech-nology development solution to regula-tors at an early enough point so as to beable to incorporate appropriate action

FIGURE 8

DeepStar safety programs.

Safety is the guiding tenet during the design, deployment and operationsphase of all DeepStar projects

■ DeepStar draft Deepwater Operating Plans BSEE DWOP■ Safe Hydrate Remediation Guideline &Operations DeepStar Guideline■ Real-Time SS Monitoring & Modeling for SS Control Commercial■ Subsea Prod. System Reliability & Risk Mgmt API 17 N■ Deepwater Pipeline Limit State Design in Codes API RP 1111■ Technology Qualification Process Standardization API 17N■ Floating System Integrity Management Technologies DeepStar Guideline■ Shallow Water Flow Management Best Practices Standard Practice

FIGURE 9

DeepStar Phase XI project types (only third-party direct spend included in budget numbers; inaddition are the value of Participants and Contributors SME time and in-kind contributions).

une 2013 Volume 47 Number 3 11

plans to address any regulatory concernwithout incurring development delays.

Conclusions and SummaryThe world’s deepwater and ultra-

deepwater basins hold tremendousresource promise to help meet globalenergy needs. DeepStar and its 80-plusmember organizations have processes,procedures and most importantly athousand-plus SMEs to help ensurethat the most appropriate technologiesare identified and then pursued andpulled through to commercialization.

DeepStar successes are numerous andhave had long-term impact due in largepart to the operator lead and operator“pull”onproject selection.Additional con-tributing factors include the following:■ successful in defining and commu-

nicating gaps and needs, especiallyin the early TRL stages;

■ R&D directly relate to operator’sfuture Major Capital Projects(MCP) developments;

■ long-term stability, proficiency,and consistency of staff● 20 years identifying and leading

deepwater technology develop-ment and application,


● $100 MM of projects and 325+technical reports,

● 1,000+ Subject Matter Experts,● 70+ member organizations pro-

vide ample opportunities forfield test and demo;

■ standardized, efficient and cost-effective processes and procedures● procurement and contracting,● portfolio vetting and selection,● project management,● technology transfer;

■ freedom to fund when and howbest to meet operators’ strategicneeds in a timely manner, no pub-lic disclosure, go at right speed,start-stop-change direction foroptimum results. Refocused every2 years.Organizations, including opera-

tors, manufacturers, service compa-nies, as well as universities interestedin defining needs and developingnew technologies to safely and prof-itably unlock the ultra-deepwater po-tential are invited to apply to join theDeepStar team. The reader of thispaper is encouraged to connect withDeepStar staff about participationdetails. For more information, pleasevisit www.DeepStar.com.

l

AcknowledgmentsThe authors gratefully acknowl-

edge the contributions of the DeepStarmember organizations for their inputand support.

FIGURE 10

The current suite of projects and how they support DeepStar’s current high-level field develop-ment scenarios.

P A P E R

DeepStar 11304: Laying the Groundworkfor AUV Standards for Deepwater Fields
A U T H O R SJohn JacobsonLockheed Martin Corporation
Pierce CohenChevron Energy TechnologyCompany and DeepStarProject Champion

Amin NasrTotal SA and DeepStarProject Champion

Art J. Schroeder, Jr.Energy Valley, Inc., andDeepStar Consultant

Greg KusinskiChevron Energy TechnologyCompany and DeepStar Director

A B S T R A C T
Emerging autonomous underwater vehicles (AUVs) developments across the
oil and gas industry now include pipeline inspection; structural survey; deepwaterinspection, repair and maintenance (IRM); and field resident systems for remote/harsh environments. As these capabilities mature, AUVs will become an increasinglyimportant tool for deepwater field operations. Early adoption of AUV standards willfacilitate more rapid deployment of AUV technologies and enable the industry toreap a wide range of safety, environmental, operational, and economic benefitsfor its deepwater fields. The development of industry standards for AUV interfaceswill facilitate more rapid implementation of AUV capabilities and lead to more cost-effective, compatible system designs by AUV vendors and field hardware manufac-turers. The development of regulatory standards for the interpretation and acceptanceof autonomous inspection results is also an essential step toward the achievementof more cost-effective operations and regulatory oversight of deepwater subsea fields.This paper describes a future vision for the use of AUVs in deepwater field operations,the benefits to be realized, and the future capabilities of AUVs that must be anticipatedand facilitated within AUV standards to achieve that vision. Additionally, this paperdescribes the goals and objectives of DeepStar Project 11304, which is laying thegroundwork to achieve accelerated standardization of AUV interfaces and the devel-opment of regulatory standards for AUV inspections.Keywords: autonomous underwater vehicle (AUV), deepwater field, interface standards,regulatory standards, inspection

oil and gas were limited primarily tobathymetric/geophysical survey, but

Introduction

Until recently, autonomous under-water vehicles (AUVs) in offshore

emerging developments across the sub-sea industry now open the potential forsafer, better, and cheaper pipeline in-spection; structural survey; deepwaterinspection, repair and maintenance(IRM); and field resident systems forremote/harsh environments. Industrydevelopments range from low-logisticsAUVs that simply “mow the lawn,” tosemiautonomous “delivery trucks” withremotely operated payloads, to AUVswith highly integrated perception re-sponse autonomy that provides sig-nificant ability to interpret sensor dataand respond accordingly. As these ca-pabilities mature, AUVs will becomean increasingly important tool for

deepwater field operations. Early adop-tion of AUV standards will facilitatemore rapid deployment of AUV tech-nologies and enable the industry to reapa wide range of safety, environmental,operational and economic benefits fornot only deepwater fields but poten-tially for all offshore developments.

Future Vision for AUVsin Deepwater Fields

The performance of IRM tasksin deepwater using current methodswith remotely operated vehicles (ROVs)can be costly and inefficient. Deep-water ROVs require large dynamically

May/J

positioning (DP) II vessels and thou-sands of square feet of deck space forsupport equipment that can weighup to 90 tons. Vessel operations arelimited by weather, mobilization/demobilization timelines, and opera-tional constraints, which can be dueto umbilical deployment in deepwater.Extended vessel deployments at highday rates are required to completeROV-based IRM operations due tothese inefficiencies, driving the costsof field operations and maintenancein deepwater to millions of dollarsper year. Under these circumstances,the development of smaller “marginalfields” or deepwater fields in remote


locations or hostile environments suchas under arctic ice may drive a prospectto low or noncommercial viability.

The implementation of AUV-basedIRM will provide significant improve-ments in safety, operating efficiency,and project economics for deepwaterfields. No longer will large DPII vesselswith expensive and cumbersome ROVspreads be required for simple IRM. Inthe near term, AUVs can be deployedfrom smaller “utility class” vessels, becapable of operations in higher seastate and current conditions, and per-form IRM tasks much more efficientlywithout the operational limitationsand equipment hazards imposed byumbilical and tether management sys-tems. Reduction in equipment com-plexity, vessel size, and crew size willalso result in improved safety, reliabil-ity, and lower environmental impact.Eventually, AUVs will become “fieldresident,” residing in the subsea fieldfor periods ofmonths or years, resultingin the elimination of surface vessels, fur-ther improvements in environmen-tal monitoring, equipment safety, andoperating efficiencies, and substantialreductions in cost.

Future capabilities for deepwaterAUVs will be substantially enhancedthrough the use of subsea docking sta-tions and local Wi-Fi “hot spots.” Theability to upload high volumes of mis-sion sensor data, download supervisoryinstructions, and recharge batterieswill extend AUV mission life to daysor weeks and eventually to monthsand/or years. The ability to have local-ized real-time high bandwidth wire-less communications for critical IRMoperations such as subsea productionequipment monitoring, sampling,valve operations, and other interven-tion operations will eliminate the needto mobilize expensive surface vesselsand large ROV spreads to accomplish


routine maintenance tasks. The valueof these capabilities for remote deep-water and/or arctic locations cannotbe overstated, since it provides imme-diate in-field access to complete tasksthat would otherwise take days orweeks to accomplish, especially consid-ering the mob-demob time associatedwith ROVs.

Current AUV DevelopmentTrends for Deepwater

There are a number of ongoingAUV developments for deepwater oiland gas that, when brought to full com-mercial capability, will have game-changing implications for deepwaterfield operations and maintenance.These include pipeline inspection,deepwater facilities inspection, andfield resident AUV capabilities.

With its HUGIN AUV, KongsbergMaritime has demonstrated autono-mous multisensor pipeline inspection,including collection of detailed digitalstill imagery of an underwater pipelineby using interferometric synthetic ap-erture sonar (SAS), pipeline trackingsoftware, a multibeam echo sounder,and a high-resolution still camera ina two-pass mission. In the first pass,side-scan data from the SAS are usedto detect and track the pipelines inreal time, extracting pipeline featuresin the sonar images. In the secondpass, the AUV conducts a low-altitudeinspection of the pipeline using itsmultibeam sonar and optical camera.Imagery is postprocessed into high-resolution sonar images and bathyme-try maps of the pipeline (Borhaug &Hagen, 2011).

The autonomous inspection vehi-cle (AIV), currently under developmentby Subsea 7, is targeted for subsea fa-cilities inspection. The AIV employsa compact, ROV-like shape to achieve

l

highmaneuverability and full hoveringcapability, which will be used to con-duct inspection of subsea infrastruc-ture such as wellheads, manifolds,jumpers, spool pieces, flowlines, andrisers. Sensors include still and HDvideo camera, lighting, 3D imagingsonar, and profiling sonar. The AIV isdeployed by an “intelligent basket” thatis lowered to the seabed and enablesautonomous IRM operations withoutsurface vessel support over a periodof days until retrieved (Jamieson et al.,2012) (Figure 1).

The Sabertooth AUV, under devel-opment by Saab SeaEye, is targetedfor autonomous, field resident, sub-sea facilities inspection. Designed forremote inspection and interventiontasks without the need for a supportship, the Sabertooth will be capableof long distance transits betweenwork sites and of docking with a subseadocking station that is deployed withinthe subsea field. The docking stationwill provide interfaces for batterycharging, high bandwidth data trans-fer, and mission planning and control.Targeted to remain submerged for upto 12 months, the system will also becapable of tethered operations using amicrofiber tether. The system can de-ploy a range of sensors and tooling forinspection and intervention on subseaproduction systems (Furuholmen et al.,2010) (Figure 2).

FIGURE 1

SubSea 7’s AIV.

The SWIMMER system, underdevelopment by Total and Cybernetix,employs a hybrid AUV-ROV approachto achieve vessel-independent, field res-ident IRM operations in remote and/ordeepwater environments. The systemfeatures a large AUV that is launchedfrom a support vessel, transits manykilometers to a docking station withina subsea field, and docks to one ofmanydedicated docking stations stagedwithin the field. Once docked, a work-class ROV is deployed from the AUVto perform a wide range of IRM tasksusing typical ROV tooling and sensors.The ROV can work within a 150-mwatch circle from the docking station;work beyond that radius requires theAUV to transit to a new docking sta-tion to perform the work. The systemis targeted for periods of up to 90 dayssubmergence before being retrievedfor maintenance (Tito & Rambaldi,2009).

The Marlin® AUV, developed byLockheed Martin, is designed to pro-vide a wide range of structural integritymanagement capabilities, includingstructural survey, subsea facilities in-spection, and pipeline inspection. Fea-turing a hydrodynamic hull form withhovering capability, an interchange-able, under-slung work package (pylon),the Marlin employs a 3D sonar to de-velop geo-registered 3D models of off-

shore production platforms, downedstructures, and debris fields in a mat-ter of minutes or hours, with typicalmodel resolution of ∼5 cm. The ca-pability to perform 3D laser inspec-tions yielding 3D models with ∼5 mmresolution is under development. Alsounder development is a 4,000-mdepth-rated version of Marlin withfull station keeping capability, whichwill be used to facilitate inspectionof subsea infrastructure. This versionof Marlin will be capable of beinglaunched from a floating productionsystem (FPS), conducting flowline andriser inspections during transits of 15–20 km, followed by inspection of in-field infrastructure such as wellheads,manifolds, jumpers, and spool pieces,and then returning to the FPS for recov-ery andmission data retrieval (McLeodet al., 2012) (Figure 3).

AUV Interface Standardsfor Deepwater Fields

Future capabilities for deepwaterAUVs will be substantially enhancedthrough the use of permanently in-stalled field infrastructure such asdocking stations, communicationsnodes, and other interfaces. Dockingstations will provide electrical and/orinductive interfaces for battery charg-ing, optical or wireless interfaces for

May/J

high bandwidth communicationsand data transfer, and mechanicalinterfaces for docking, tool storage,and interchange. Docking stationsmay also facilitate the use of tethermanagement systems for semiauto-nomous vehicle systems. Other per-manently installed interfaces thatwould leverage the benefits of AUVcapabilities include wireless subseacommunication nodes (radio frequency(RF), acoustic, optical, and/or hybrid),data harvesting nodes, subsea produc-tion equipment monitoring and sam-pling points, and intervention toolinginterfaces.

The development of industry stan-dards for AUV interfaces, similar tothe API/ISO standards for ROVs,will lead to compatible system designsby AUV vendors and field hardwaremanufacturers and will enable morecost effective development and moreefficient operations of deepwater sub-sea fields. As sensor and AUV manu-facturers design and operators deploycompatible kit, the expense (capexand opex) versus the value of subseadata collected will drop by orders ofmagnitude in a positive re-enforcingcycle. The offshore oil and gas industryhas the benefit of many lessons learnedfrom the development of ROV inter-faces over a period of approximately20 years of development from theearly 1970s to the early 1990s. Duringthe early days of ROVs, each field de-velopment project developed its ownunique installation, operations, andmaintenance procedures, which typi-cally resulted in unique interfaces forintervention tooling. As a result, pro-jects frequently paid a high price for“reinventing the wheel,” includingthe cost of nonrecurring engineeringdesign, the cost to procure uniquetooling (and the inability to reuse thetools on subsequent projects), and

FIGURE 2

Sabertooth AUV from Saab SeaEye.

FIGURE 3

Lockheed Martin’s Marlin® AUV.


the inability to invoke full competitionfor offshore services due to being tiedto a single vendor’s technology. As aresult, the costs of operations, mainte-nance, and spares frequently grew overtime due to the difficulty and complex-ity of maintaining unique tools andprocedures for each subsea field. Fur-ther adding to cost and potentiallycompromising safety, the lack of stan-dards drove the need for customizedoperator training for each installationand tended to complicate rotationaland/or multiclient deployments.Through early standardization, the in-dustry can leverage these lessons toreap the rewards with sensible, well-thought standardization for AUVs.

Regulatory Standards forAUV Inspections

In the future, AUVs will leveragea wide range of sensor technologiesto meet both end user and regulatoryrequirements for integrity monitoringof offshore infrastructure. Offshore in-frastructure that falls under regulatoryoversight includes fixed platforms, float-ing structures, pipelines, risers, mooringlines and subsea production equipment.Sensor technologies that can be em-ployed by AUVs to meet inspectionrequirements include video, photo-graphic, sonar, laser, ultrasonic, mag-netic, and others.

Today, most regulations and in-dustry standards do not address theuse of AUVs for inspection, primarilybecause the technology did not exist atthe time the regulations were devel-oped. While AUVs can be used tomeet standards where only the sensortechnology is specified (e.g., “sonarwith frequency of at least 500 kHz”),the regulations are frequently eitherambiguous or prohibitive regardingthe use of AUVs. In other instances,


the regulation specifically recognizesthe inspection method (diver or ROVinspection) where a human is “in theloop.”

Additionally, as sensor and auton-omy technologies advance, the de-ployment of autonomous inspectiontechnologies will provide a range ofnew and as yet unrecognized formatsfor inspection results, which when as-sessed using state of the art engineeringtools, have the potential to provide con-siderable improvements in inspectioneffectiveness as well as efficiency andcost. For example, the use of 3Dmodel-ing and autonomous change detectionusing 3D sonar, while not currentlyrecognized by any regulatory standard,would dramatically improve the abilityto quickly and effectively identify fea-tures that have changed since a previ-ous structural survey. The capabilityto identify, map, and measure featuressuch as free span, seabed scour, bent,missing or damaged structural mem-bers, and/or distorted or twisted beamswould provide a significant advantagein posthurricane inspection operations.Similarly, performing autonomousstructural inspection using a 3D laserwould provide the capability to quicklyand efficiently identify, map, and mea-sure cracks and other structural defectsor damage. Utilizing these advancedtechnologies and techniques will helpminimize the need for divers, par-ticularly in the initial incident triagephase where the risk of the unknownis highest.

The development of regulatorystandards for the interpretation andacceptance of autonomous inspectionresults where no diver or ROV pilotis involved is therefore an essentialstep toward achievement of a safer,more cost-effective and efficient opera-tions and maintenance of deepwatersubsea fields.

l

DeepStar Project 11304:Laying the Groundworkfor AUV Standards

DeepStar is a joint industry tech-nology development project focusedon advancing the technologies tomeet its members’ deepwater businessneeds to deliver increased productionand reserves. It provides a forum toexecute deepwater technology devel-opment projects and leverage thefinancial and technical resources ofthe offshore oil and gas industry.DeepStar’s membership is composedof companies and organizations fromacross the industry and the globe,including oil companies, vendors, reg-ulators, and academic and researchinstitutions. DeepStar runs in 2-yearfunding phases, and participation isby phase. In Phase XI, membershipincludes 11 major oil companies andover 70 contributing companies fromindustry and academia.

DeepStar has funded Project 11304,AUV Standards for Deepwater Fields,and through this project is laying thegroundwork to achieve acceleratedstandardization of AUV interfaces andthe development of regulatory stan-dards for AUV inspections. The projectcomprises two major tasks:

Task 1 – AUV Interface Standardsfor Deepwater Fields, andTask 2 – Regulatory Standards forAUV Inspections.
The primary goals of this project
are as follows:(1) identify subsea interfaces thatshould be standardized for AUVs,develop recommendations for ap-propriate standards related to sub-sea interfaces for AUVs, and beginthe process of approval/releasethrough an appropriate standardsorganization; and(2) identify current regulatory in-spections that could be conducted

with AUVs, develop draft versionsof appropriate standards for suchAUV inspections of subsea infra-structure, and make available workproducts that could help facilitateregulatory agency review and po-tential adaptation.
In order to keep the scope of the
project manageable, a single entity wasselected for each Task as the primarystakeholder with whom to hold discus-sions and to whom to submit recom-mendations on AUV standards. ForTask 1, the American Petroleum Insti-tute (API) Subcommittee 17 (SC17)was identified as the most appropriatestandards organization for the develop-ment of AUV interface standards. ForTask 2, the U.S. Bureau of Safety andEnvironmental Enforcement (BSEE)was identified as the regulatory agencyhaving most appropriate jurisdictionregarding AUV inspections.

In order to achieve the project ob-jectives, a roadmap has been developedto define top-level project activities andevents leading to the development andsubmission of work-products to APIand to BSEE. An overview of the proj-ect road map is provided in Figure 4.

The project is currently in progresswith an anticipated completion date ofMarch 2014. The scope of Task 1 andTask 2 efforts have been defined in de-tail, preliminary discussions with APIand BSEE are in progress, and industrysurveys and workshops are planned forspring 2013. Key industry stakeholderswho will be engaged in the survey andworkshop activities include FieldOperators, EPC Contractors, SubseaHardware Providers, ROV/AUV Ser-vice Contractors, AUV Developers,AUV Sensor Developers, and InterfaceTechnology Developers.

The focus of Task 1 survey/workshop activities will be on (1) iden-tifying and documenting the current

and future AUV Interface implemen-tations foreseen by the industry, (2) fa-cilitating discussion and interactionamongst industry stakeholders regard-ing AUV interface standardization,and (3) developing consensus withinthe industry on what AUV interfacesshould be standardized and what thoserecommended standards should include.The output of these activities will beshared with API SC17 in the form ofa Task Report, which could result informal initiation of the developmentof API standards for AUV Interfaces.

The focus of Task 2 survey/workshop activities will be on (1) iden-tifying and documenting the currentand future AUV inspection technologiesforeseen by the industry, (2) facilitat-ing discussion and interaction amongstindustry stakeholders regarding currentregulatory standards that could be metusing AUV inspection technologies, and(3) developing work products that willbe useful to BSEE, onwhich regulationscould be most improved by allowinginclusion of AUV inspections.

After the survey and workshopactivities are completed, project fo-cus will shift to review and refine-

May/J

ment of recommendations within theDeepStar Project Team, presentation/discussion of recommendations withAPI SC17 and BSEE, and developmentof formal work-products reflecting theconsensus opinion of the industry andthe applicable standards agency.

ConclusionThe future vision for AUVs in deep-

water field operations is compelling.AUV technologies are evolving at veryrapid pace, bringing new and powerfulcapabilities that offer significant safety,environmental, operational and eco-nomic benefits. In order to fully realizethese benefits, future capabilities ofAUVs must be anticipated and facil-itated within industry standards andregulatory allowances. DeepStar Proj-ect 11304 is laying the groundworkto achieve standardization of AUV in-terfaces and regulatory allowance forAUV inspections. By bringing industrystakeholders and standards agenciestogether, early adoption of standardscan be achieved, thereby capturing thelessons learned with ROVs and ac-celerating implementation of these

FIGURE 4

DeepStar 11304 Project Road Map.


game-changing AUV capabilities indeepwater fields.

AcknowledgmentsThe authors gratefully acknowl-

edge the contributions of the DeepStarmember organizations for their inputand support.

Corresponding Author:John JacobsonLockheed Martin Corporation10375 Richmond Avenue,Suite 400, Houston, TX 77042Email: [email protected]

ReferencesBorhaug, E., & Hagen, P.E. 2011. AUVs

Take On Pipeline Inspection. E&P Magazine

Website. http://www.epmag.com/Production-

Drilling/AUVs-On-Pipeline-Inspection_86704.

(accessed March 6, 2013).

Furuholmen, M., Johansson, B., & Siesjö, J.

2010. Seaeye Sabertooth: A Hybrid AUV/ROV

Offshore System. In: MTS/IEEE Oceans

2010 Proceedings, September, 2010. Seattle,WA:

IEEE. http://dx.doi.org/10.1109/OCEANS.

2010.5663842.

Jamieson, J., Wilson, L., Arredondo, M.,

Evans, J., Hamilton, K., & Sotzing, C. 2012.

Autonomous Inspection Vehicle: A New

Dimension in Life of Field Operations.

Offshore Technology Conference Proceedings,

April–May, 2012. Houston, TX: Offshore

Technology Conference. http://dx.doi.org/

10.4043/23365-MS.

McLeod, D., Jacobson, J., & Tangirala, S.

2012. Autonomous Inspection of Subsea

Facilities-Gulf ofMexicoTrials. In: Proceedings,

Offshore Technology Conference, April–May,

2012. Houston, TX: Offshore Technology

Conference. http://dx.doi.org/10.4043/

23512-MS.

Tito, N., & Rambaldi, E. 2009. SWIMMER:

Innovative IMR AUV. In: OTC 2009 Pro-


ceedings, Offshore Technology Conference,

May, 2009. Houston, TX: Offshore Technol-

ogy Conference. http://dx.doi.org/10.4043/

19930-MS.

l

P A P E R

DeepStar Metocean Studies:15 years of Discovery
A U T H O R SCortis CooperChevron Energy Technology Co.
Tom MitchellConsultant, Georgetown, Texas

George ForristallForristall Ocean Engineering, Inc.

James StearChevron Energy Technology Co.

A B S T R A C T
In 1998, DeepStar began the first of many successful studies that have resolved
important questions concerning meteorological and oceanographic (metocean)processes that can cause large loads or fatigue problems on deepwater facilities.In so doing, these studies have immeasurably enhanced the reliability and safety ofdeepwater structures and pushed the frontiers of ocean science that have tradition-ally been the realm of academic research. The efforts have focused on three majorphenomena: the Loop Current, Topographic Rossby Waves (TRW), and stormwinds. Much of the DeepStar effort has focused on improving numerical modelsof the respective phenomena because they can provide long historical databasesat any site—data that serve as the basis for operating and extreme criteria with rea-sonable statistical uncertainty. Studies of the Loop include the first measurementsof the Loop inflow and turbulence and evaluation of existing numerical models.Most of DeepStar’s efforts on TRWs started in 2008, and in a 5-year period, ithas developed a validated numerical model and used it to build a 50-year hindcastdatabase. Efforts are underway to use those results to build a stochastic forecastmodel. Finally, DeepStar has analyzed a large set of wind measurements taken fromthe powerful recent hurricanes and found that recommended formulas for windprofiles and spectra have significant bias and will be corrected in future recom-mended practices.Keywords: oceanography, meteorology, metocean, ocean currents, ocean waves,ocean winds

Many mysteries remain concerningmetocean variables, especially deep

Introduction

Almost all aspects of offshore facil-ities are affected by winds, waves, andcurrents, including operations and cap-ital costs. Indeed, in many deeper waterlocations, the choice of the basic facilityis heavily influenced by the meteoro-logical and oceanographic (metocean)conditions, second only to the reservoircharacteristics and water depth.

water currents and hurricane-drivenwinds and waves. Nowhere is this truerthan in theGulf of Mexico,where strongocean currents can be generated by sev-eral different processes that can varydramatically in magnitude over spacescales of a few kilometers. Several ofthese processes were first discoveredonly recently, and their quantificationhas been led by Joint Industry Projects( JIP) like DeepStar, rather than thetraditional university oceanographers.Hurricanes have also been an area ofactive industry research because theydominate the loads on most produc-tion facilities in deep water. Despitetheir importance, major uncertainty

remains concerning winds and waves,in part because few measurements havebeen made in strong hurricanes.

As a result of these myriad uncer-tainties and the importance of meto-cean criteria on safety and reliability,DeepStar IV began significant fundingof metocean studies in 1998 and hascontinued this investment since then.The following sections outline the ma-jor studies in more detail. Each sectionis focused on a particular phenome-non, e.g., Topographic Rossby Waves(TRWs), so it frequently will coverseveral studies. Each section describesthe study goals, business drivers, andmethods and summarizes the results.

May/J

Loop CurrentThe Loop Current is a strong per-

manent current that flows through theYucatan Straits, loops northward, andthen exits through the Florida Straitswhere it is renamed as the Gulf Stream.About once per year, the Loop movesnorthward of 27°N, becomes unstable,and forms a large eddy that breaksaway and drifts to the west. The Loop(which will henceforth be taken tomean the Loop proper and its asso-ciated large anticyclonic eddies) canoccasionally affect shelf waters but istypically found in water depths greaterthan 500 m. Radial speeds within theLoop can exceed 2 m/s and generate


the drag equivalent of a hurricanewave on mooring lines or generatevortex-induced vibrations that canlead to fatigue failure of risers. Theseeffects influence the design of drillingand production risers, mooring ten-sions on production and drilling rigs,connection/disconnection of drillingrisers, and installation of pipelines,mooring lines, tendons and hulls.Although no firm accounting has everbeen done, we estimate that the Loopcosts the industry on the order of$10 million/year in rig delays.

The Loop was not discovered untilthe late 1960s, and the first currentmeasurements did not occur until theearly 1980s. Given the importance ofthe Loop and its relatively recent dis-covery, it is not surprising that therewere many important unknowns thatneeded to be resolved as the industrymoved into deeper water.

In 1998, DeepStar initiated itsfirst oceanographic project by measur-ing the incoming source of the LoopCurrent—in other words, the waterinflow through the Yucatan Strait.Oceanographers had long wanted totake these fundamental measurementsbut had been frustrated by the cost andthe politics of deploying instruments inthe eastern half of the Straits controlledby Cuba. Not to be deterred, DeepStarcontracted CICESE, an oceanographicresearch institution in Mexico thathad a cooperative research agreementwith Cuban oceanographers. CICESEdeployed eight moorings across theYucatan Straits with 33 single-pointcurrent meters and eight acousticDoppler current profilers (ADCPs).In addition, they conducted four shipsurveys across the Strait during the18 months that the moorings were inplace. Results were documented inAbascal et al. (2003). The major ben-efit of the study was to provide the


first careful measurement of the inflowboundary condition for numericalmodels of the Loop. Such models arean important tool for developing designand operating conditions and, perhapsmost importantly, for forecasting.

After the Yucatan Straits measure-ments, DeepStar quickly turned to an-swering another key question aboutthe Loop Current: How turbulent isit? At the time, designers were worriedabout the ability of turbulence to ex-cite higher modes in the tendons ofTension Leg Platforms (TLPs) andspars. Current speed fluctuations canaffect these structures both by directforcing and by reducing the effective-ness of VIV suppressing strakes. Theseeffects have often been seen in modelbasins where the turbulence intensityis 10–20% of the mean velocity. Oce-anic turbulence levels were thought tobe much lower, but there was essen-tially no field data to prove that as-sumption. DeepStar filled the gap byfunding measurements in a Loop Cur-rent eddy using a unique instrumenta-tion system. The results are describedby Mitchell et al. (2007).

Measurements were made withinthe eddy and across the strong frontalboundary that separates the eddy fromthe surrounding waters. A towed vehicle,the TOMI (Towed Ocean Microstruc-ture Instrument), was equipped with aspecial 300-kHz ADCP that had its fourbeams directed fore, port, starboard, anddown. The along-beam velocities re-solved structures with wavelengths of4–60 m. The vehicle also carried shearprobes for measuring velocity fluctuationsin the dissipation range (0.5–100 cyclesper meter) and other environmentalsensors for measuring temperature,salinity, depth, and vehicle orientation.The towed body is shown in Figure 1.

Tows were conducted at 25-, 50-,100-, and 150-m depths around the

l

northern edge of the Loop Eddy in cur-rents of up to 1.7 m/s. Turbulence wasdetected with the shear probes, butmostly in the 130–150 m depth rangearound the local salinity maxima. Thelevel of turbulence was weak, and itwas distributed intermittently in bothspace and time. The most energeticevents of turbulence had eddy scalesof at most 4 m and velocity scales ofonly 1 cm/s. The typical and average val-ues were more than 10 times smaller.

After taking some basic physicalmeasurements in the Loop Current,DeepStar’s next effort was to under-stand the accuracy of available models.While many models were available atthe time, none had been rigorously val-idated. To fill this gap, DeepStar initi-ated a study in 2004 to compare fiveexisting forecast models. The modelerswere asked to run a 1-year historicalperiod for which DeepStar had a pro-prietary, detailed set of measurementsnever before seen by the modelers.Models were spun up by assimilat-ing publicly available measurementssuch as satellite altimetry. On thefirst day of each month, the modelswere run for 4 weeks without anydata assimilation. Model performance

FIGURE 1

The TOMI instrument that was used to mea-sure turbulence. The ADCP transducers aremounted on the bottom mast (forward look-ing), at the base of the (orange) upper mast(port and starboard looking), and behind thelower mast on the main body (downwardslooking).

was judged by comparing the fore-casted to the observed distance of thenearest major Loop or eddy front toseven sites scattered over much of thedeep Gulf east of 94°W. Model resultswere compared to persistence (themajor fronts were assumed to remainstationary) for the entire month. Fig-ure 2 compares the RMS (root meansquare) error accumulated for the12 runs at all seven sites. Only oneof the models was found to beat persis-tence after about 12 days, but not bymuch. Perhaps most striking was thesubstantial error exhibited by all themodels right from the start (0 week).This strongly suggests that the satelliteimagery used to spin up the modelswas far from perfect, probably becauseit failed to resolve the meanders andfrontal lobes commonly found on theLoop and its eddies. While these fea-tures may have relatively short lengthscales (order 50 km), they can signif-icantly affect the error metrics. Thefact that the models did poorly evenin a nowcast mode suggested theymay have had substantial errors evenin a hindcast mode.

The overall conclusion of the Phase 1study was that the models were tooinaccurate in forecast mode to be ofmuch value to Industry operations,but that further work was justifiedgiven the substantial benefits thatcould be had from an accurate forecast.

Towards the end of the first modelintercomparison study, a new modelappeared on the scene that was quitedifferent to the others tested in Phase 1.AEF’s model was a so-called “feature”model, which utilized proprietary drift-ing buoys as well as satellite imagery tospin-up the model. Given the promiseof this new approach, DeepStar de-cided to fund a Phase 2 study usingthe AEF model and the Oey model,winner of the Phase 1 study. Figure 3compares the error from the two mod-els with that of persistence. The AEFmodel consistently beat the Oey modeland overtook persistence at about1 week. Its forecast error stayed flatuntil the end of the second week andthen slowly climbed until it wasdouble the initial error after 4 weekswhere it remained steady until nearly7 weeks. Overall, the conclusion was

May/J

that the AEF model could provideforecasts with useful accuracy, but itssuccess depended on having access todetailed in situ measurements fromdrifting buoys or other similar sources.Such measurements cost upwards of$50,000/mo.

TRWsIn the late 1990s, British Petro-

leum (BP) measured currents reachingabout 1 m/s near the seafloor in about2,000 m of water along an underwaterfeature known as the Sigsbee Escarp-ment. While the currents were mostintense near the bottom, they remainedsubstantial for hundreds of metersabove the seafloor, finally reaching am-bient conditions at about 1,000m. TheBureau of Ocean Energy Management(BOEM; formally known as MineralsManagement Service) deployed threecurrent meters nearby for 18 monthsand observed similarly large currents.Figure 4 shows the time series of thesecurrents. Subsequent analysis of theBOEM measurements by Hamiltonand Lugo-Fernandez (2001) suggestedthat the currents were driven byTopographic Rossby waves (TRW),a 200-km-long wave with periods of10–14 days.

TRWs can generate current speedsat the bottom near the Escarpment thatfar exceed those generated by any otherphenomena. Such currents dominatethe metocean extreme and fatigueloads on pipelines and risers, especiallyflexible risers (steel catenary risers orSCRs).

DeepStar began its study of TRWsin 2003 by taking measurements ofthe cross-Escarpment variation of thewaves, a characteristic that had notbeen studied before. Eight currentmeters were placed near the seaflooracross the Escarpment at 91°08′W,

FIGURE 2

Comparison of forecast error from five Loop models with persistence.


as depicted in Figure 5. BOEM had athrough-columnmooring, L4, deployedjust to the south of the DeepStar array.Over the 1-year deployment, about ahalf-dozen TRWs were measured andshowed the strongest currents occurrednear the base of the Escarpment at S2and S3. Only about 10 km north at S6speeds dropped off rapidly to less thanhalf those observed at S2. In contrast,the reduction on the down-dip sideof the Escarpment was much smaller,e.g., L4 was about 30% less than S2.


In 2008, DeepStar restarted itsefforts on TRWs because of increasedexploration activity along the Escarp-ment and reports from drilling rigsthat were adversely affected by strongbottom currents. That year, two pro-jects were started. The first involvedtaking more current measurements,but this time with moorings spreadalong the Escarpment as well as across.Figure 6 shows the four DeepStarmoorings as well as other mooringsdeployed by Chevron and Shell, which

l

overlap in time and were later ob-tained byDeepStar. The 1 year ofmea-surements showed strong variationalong the Escarpment and confirmedthe strong cross-escarpment variationfirst observed in the 2003 DeepStarmeasurements.

The second project, begun in2008, focused on the developmentof a numerical model with the goal ofeventually using it to develop opera-tional and design criteria. Without amodel, the industry would have hadto develop criteria based on measure-ments of only a few TRWs at a fewlocations. The latter was especiallytroubling, because the measurementsshowed that TRW-generated currentsvaried significantly over length scalesof a few kilometers.

Florida State University (FSU) wascontracted to develop the model andsoon discovered that the numericaldiscretization in standard ocean cur-rent models generated substantial nu-merical errors when dealing with thesharp bathymetric gradients of theEscarpment. A more advanced numer-ical technique was implemented, andthe model was then used to hindcastthe BOEM and DeepStar measure-ments (Dukhovskoy et al., 2009).Results were encouraging so a secondmodeling phase was kicked off, cul-minating in a well-validated model assuggested in the excellent comparisonshown in Figure 7. In the process,FSU discovered that the TRWs werebeing generated by the collision of theLoop (or a recently detached eddy) onthe outer slope of the Mississippi Fan,just south of the Delta (Morey et al.,2010).

With the successful validation of themodel, DeepStar now had a tool thatcould be used to develop operationaland extreme criteria. FSU developedthe needed database by allowing the

FIGURE 3

Comparison of forecast error from two Loop models with persistence.

FIGURE 4

Time series of near-bottom current vectors measured near the Sigsbee Escarpment. The vectorspointing up are flowing towards the northeast; those pointing down are flowing southwest.

model to free run for a 50-year period.It was a daunting computation ef-fort since it involved running a high-resolution (800 m grid size) nestedmodel covering the Sigsbee Escarp-ment inside a 3.5 km model covering

the northern Caribbean and portionsof the southeast Atlantic. Figure 8shows the model domain. Results fromthe 50-year run were archived and arenow being used by the Industry todevelop design criteria.

May/J

Finally, FSU recently started toapply their model results to develop apredictive capability that can even-tually forewarn drillers and installersof major facilities, of an approachingTRW that might threaten their op-erations. Initial results have shownthat the numerical model cannot pre-dict the phase of TRWs very well sincethere are essentially no operationalmeasurements in the lower deep watercolumn available for model initializa-tion or data assimilation. Instead ofdirect use of the model, FSU is usingthe 50-year database to develop a sto-chastic model based on independentvariables like the position of the Loop.This approach will not suffer the phaseissues and should also provide uncer-tainty estimates.

Hurricane WindsIn 2010, DeepStar funded a study

to bring together all available hurricanewind data sets made in and around theGulf since 1998, quality control them,and then analyze them in an effort tocheck the validity of the present Amer-ican Petroleum Institute (API, 2012)recommended equations for hurricanewinds. The data sets included dozensof offshore platform anemometer re-cords, measurements from NationalOceanic and Atmospheric Administra-tion (NOAA) buoys, Coastal-MarineAutomated Network (C-MAN), Auto-mated Surface Observing System(ASOS), and National Ocean Service(NOS) stations, tower arrays of ane-mometers deployed along the coast,coastal weather radars, and dropsondeobservations made by hurricane hunteraircraft.

The first phase of this study wascompleted in 2012 by Applied ResearchAssociates, Inc., Texas Tech Univer-sity, and the University of Florida

FIGURE 5

Cross section of the current meters deployed across the Sigsbee Escarpment.

FIGURE 6

Location of current measurements taken during 2008-2009. The dark blue curve shows the baseof the Sigsbee Escarpment.


and identified deficiencies in the API(2012) equations for gust factors, pro-files, and spectra when applied to hur-ricanes. Figure 9 compares measuredspeed profiles to the API (2012) andEngineering Sciences Data Unit


(ESDU, 1974, 1982, 1983) calculatedprofiles. While the API (2012) pro-file compares well within a few 10 sof meters of the sea surface, a 10% dis-crepancy appears at the higher eleva-tions typical of platform deck heights

l

(30–60 m). Such a discrepancy trans-lates to more than 20% in the staticdrag force. On the other hand, the API(2012) equation for gust factors wasfound to underestimate the observa-tions as shown in Figure 10.

The second phase of the study isnow underway, with the analysis ex-tended to tropical storm wind recordsmade off the northwest coast ofAustralia, the east coast of the UnitedStates, and a reexamination of the orig-inal Norway extratropical wind mea-surements used to develop the APIrelationships. The end goal of thisphase is a revised set of wind design re-lations that may then be incorporatedinto the latest offshore standards, forboth tropical and extratropical storms.The study is anticipated to be com-pleted by late 2013.

Summary and ConclusionsIn 1998, DeepStar began what was

to become a highly insightful set of proj-ects in the field of meteorology andoceanography (metocean). The firstproject focused on measuring the

FIGURE 7

Comparison of model and observed current at three depths taken at one of the moorings shown in Figure 6.

FIGURE 8

Contours showing the sea surface height over the large-scale (3.5 km) model. Insert shows thenested model around the Sigsbee Escarpment with a resolution of 800 m.

FIGURE 9

Comparison of measured wind profiles with recommended profiles from API and ESDU for three different central pressure bins.

FIGURE 10

Comparison of measured gust factors with recommended factors from API and ESDU for three different wind speeds.

May/June 2013 Volume 47 Number 3 25

flow of the Loop Current through theYucatan Straits—fundamental infor-mation that had never been gathered.This was followed by measurements ofturbulence in the Loop Current; a studydriven by concerns about resonancein tendons, moorings, and risers. Fieldmeasurements were completed in2003, which dismissed that concern.In 2004, attention was turned to find-ing the best available forecast model ofthe Loop Current, a tool that couldsave the Industry millions of dollarsby helping it avoid downtime duringdrilling and installation of large facili-ties like spars. Two studies were donecomparing the ability of eight existingforecast models. Results showed thatmany of the models were muchworse than simply assuming that theLoop remained unchanged (persis-tence) and revealed that the modelswere primarily limited by the accuracyof their initial conditions. This knowl-edge has been used in other Industryefforts to improve forecast models. In2004, a six-phase effort was begun toquantify Topographic Rossby Waves(TRW)—a wave with a length of200 km first measured in the Gulfin 1998 and capable of generatingcurrents of 1 m/s (2 kt) near the seafloor. Phases 1 and 2 deployed arraysof current meters that recorded severalTRWs. These measurements werethen used to develop and validate anumerical model—the first to success-fully simulate the full strength of thesepowerful waves. Phase 5 used the TRWmodel to develop a 50-year hindcastdatabase that provides accurate oper-ational and extreme current criteriathroughout much of the deepwaterGulf. Phase 6 is using the model todevelop a probabilistic forecast thatcan warn drill rigs and installationoperations of an approaching TRW.Most recently, DeepStar has funded


work to analyze the wealth of winddata collected during the recent extremehurricanes. This study has revealed thatthe present Industry standard for hurri-cane wind spectra, profiles, and gustscan be improved, so revisions will soonbe adopted.

Corresponding Author:Cortis CooperChevron Energy Technology Co.Email: [email protected]

ReferencesAbascal, A.J., Sheinbaum, J., Candela, J.,

Ochoa, J., & Badan, A. 2003. Analysis of

flow variability in the Yucatan Channel.

J Geophys Res. 108(C12):3381. doi:10.1029/

2003JC001922.

American Petroleum Institute. 2012.

Derivation of metocean design and operating

conditions, API RP 2MET, First Edition.

Engineering Sciences Data Unit, ESDU.

1974. Characteristics of atmospheric turbulence

near the ground, Item No. 74031, London.


1982. Strong winds in the atmospheric

boundary layer, Part 1: Mean hourly wind

speed, Item No. 82026, London.


1983. Strong winds in the atmospheric

boundary layer, Part 2: discrete gust speeds,

Item No. 83045, London.

Dukhovskoy, D.S., Morey, S.L., Martin,

P.J., O’Brien, J.J., & Cooper, C. 2009.

Application of a vanishing, quasi-sigma, vertical

coordinate for simulation of high-speed, deep

currents over the Sigsbee Escarpment in the

Gulf of Mexico, Ocean Model. 28(4):250-65.

Hamilton, P., & Lugo-Fernandez, A. 2001.

Observations of high speed deep currents in

the northern Gulf of Mexico. Geophys Res

Lett. 28:2867-70.

Mitchell, T.M., Lueck, R.G., Vogel, M.J., &

Forristall, G.Z. 2007. Turbulence measure-

l

ments in a Gulf of Mexico warm-core ring.

In: Proc. 26th Int. Conf. on Offshore Mech.

and Arctic Eng., OMAE2007-29231.

New York: American Society of Mechanical

Engineers.

Morey, S.L., Dukhovskoy, D.S., & Cooper, C.

2010. Measurement and modeling of topo-

graphically trapped waves along the Sigsbee

Escarpment. In: Proc. Offshore Tech. Conf.,

20694. doi:10.4043/20694-MS.

P A P E R

Recent Development in IR Sensor Technologyfor Monitoring Subsea Methane Discharge
A U T H O R SMark SchmidtPeter LinkeGEOMAR Helmholtz Centre forOcean Research Kiel, Kiel, Germany
Daniel EsserCONTROS Systems & SolutionsGmbH, Kiel, Germany

A B S T R A C T
Recently developed methane sensors, based on infrared (IR) absorption tech-
nology, were successfully utilized for subsea methane release measurements.Long-term investigation of methane emissions (fluid flux determination) from nat-ural methane seeps in the Hikurangi Margin offshore New Zealand were performedby using seafloor lander technology. Small centimeter-sized seep areas could besampled at the seafloor by video-guided lander deployment. In situ sensor mea-surements of dissolved methane in seawater could be correlated with methane con-centrations measured in discrete water samples after lander recovery. Highbackscatter flares determined by lander-based Acoustic Doppler Current Profiler(ADCP) measurement indicate bubble release from the seafloor. Highest methaneconcentrations determined by the IR sensor coincided with periods of high ADCPbackscatter signals. The high fluid release cannot be correlated with tidal changesonly. However, this correlation is possible with variability in spatial bubble release,sudden outbursts, and tidal changes in more quiescent seepage phases.

A recently developed IR sensor (2,000 m depth-rated) with a detection limit formethane of about 1 ppm showed good linearity in the tested concentration rangeand an acceptable equilibration time of 10 min. The sensor was successfully oper-ated offshore Santa Barbara by a small work-class ROV at a natural methane seep(Farrar Seep). High background methane concentration of 50 nmol L−1 wasobserved in the coastal water, which increases up to 560 nmol L−1 in dissolvedmethane plumes south of the seepage area. ROV- and lander-based sensor deploy-ments have proven the applicability of IR sensor technology for the determination ofsubsea methane release rates and plume distribution. The wide concentrationrange, low detection limit, and its robust detection unit enable this technologyfor both subsea leak detection and oceanographic trace gas investigations.Keywords: methane, sensor development, natural hydrocarbon seeps, subsea leakdetection

a natural subsea hydrocarbon seep canserve as an ideal analogon for studying

Introduction

Natural marine hydrocarbon seepsare important sources of methane(CH4) to the surface sediments, thebenthic boundary layer, and eventuallyto the water column and atmosphere.CH4 is a potent greenhouse gas thatwarms the Earth about 23 times morethan carbon dioxide (CO2) when aver-aged over 100 years. Quantifying thedischarge of CH4 from the seabed,its fate in the water column and itsflux to the atmosphere has been thesubject of ongoing research on manydifferent fronts (e.g., Clark et al.,2000; McGinnis et al., 2006; Judd& Hovland, 2007; Westbrook et al.,2009; Faure et al., 2010). Furthermore,

gas leakage scenarios from subsea con-structions like gas/oil transport lines,active or abandoned wellheads, etc.(Leifer et al., 2006).Moreover, the dis-solution behavior and transport of gasin the water column at variable ocean-ographic conditions, like currents,horizontal/vertical eddies, tidal changes,or stratified water columns can be stud-ied at natural seepage sites (e.g., Leiferet al., 2006; McGinnis et al., 2006,

2011; Schneider von Deimling et al.,2010). Here, we report on the devel-opment and deployment of novelmethane sensors, based on infraredabsorption technology, which weretested in two different subsea settings.The first setting was a long-term mul-tisensor deployment with a benthiclander, which was placed for 41 h ata 670-m-deep CH4 cold-seep at theseafloor in the Hikurangi Margin,

May/J

New Zealand. The second one was ashallow water test of ROV-operatedsensor measurement at a natural hydro-carbon seep offshore Santa Barbara,California.

CH4 Sensor Deploymenton a Deep Sea Lander

A novel methane sensor HydroC™of CONTROS System & Solutions


GmbH, Germany, was deployed ona benthic lander equipped with awide range of instrumentation tostudy the role of physical parameterson exchange processes in the benthicboundary layer. Benthic landers pro-vide a stationary study platform de-coupled from the movement of theship and simultaneously measure sev-eral physical, chemical, and biologicalparameters across the sediment waterinterface. The Fluid Flux Observatory(FLUFO) was deployed for in situ fluxmeasurements of methane and oxygenin about 670-m water depth at a meth-ane seep setting known as RockGardenby local fishermen (Figure 1). This areais situated at the southern terminationof Ritchie Ridge and is uplifted by thesubduction of a seamount beneath theouter Hikurangi Margin at the eastcoast of New Zealand’s North Island.Townend (1997) estimated that morethan 20m3 of fluids are being squeezedfrom accreted and subducted sedimentsalong each meter of the HikurangiMargin every year, which results inabundant evidence of escaping gas off-shore (Faure et al., 2010; Linke et al.,2010; Naudts et al., 2010) and onshore(Campbell et al., 2008). The deploy-ment was part of a large campaign in-volving a large range of equipment andscientific disciplines to study themeth-ane seeps at the Hikurangi Margin(Greinert et al., 2010).

The observatory consists of a ti-tanium tripod frame that carries 21Benthos glass spheres for buoyancyand ballast weights attached to eachleg (Figure 2a). The release of theballast weights is controlled by twoacoustic releasers. FLUFO is equippedwith two circular benthic chambers,each covering a sediment area of651.4 cm2. A video-guided launchingsystem (LAUNCHER) allowed smoothplacement of the observatory on a


selected site at the seafloor (Pfannkuche& Linke, 2003). Two to three hoursafter deployment, the benthic fluxchambers were slowly driven into thesediment. Seabed methane emissionwas monitored with eight sequentiallywater samples taken from each cham-

l

ber by means of glass syringe sam-plers. Sampling (monitoring) periodswere about 34 (FLUFO-5) and 40 h(FLUFO-4, BIGO-4), respectively.After the in situ incubation, the bottomof the chambers was closed with a shut-ter to recover the sediments for further

FIGURE 1

Overview map showing the bathymetry of the Hikurangi Margin at the east coast of New Zealand,mapped during R/V SONNE cruise SO191 in 2007 (modified from Linke et al., 2010). The enlargedbathymetric maps depict the Rock Garden area with stations relevant for this paper; for example,the site of vigorous gas discharge (Faure bubble site) discovered during a ROV dive (Naudtset al., 2010) and two other lander stations (FLUFO-4 and BIGO-4) described in Linke et al. (2010).(Color version of figures are available online at: http://www.ingentaconnect.com/content/mts/mtsj/2013/00000047/00000003.)

analyses. After recovery, syringe watersamples retrieved during lander deploy-ment were immediately transferred intothe cold room, where subsamples wereobtained for the determination of oxy-gen and methane (Linke et al., 2010).

Next to the chambers, the landercarried the HydroC™methane sensorfor in situ measurements up to 4,000 mwater depths (Figure 2b). The high-pressure seawater side is separated bya permeable membrane from the inter-nal infrared detection unit. An internalpump system increases equilibration ofinternal partial pressure of, for example,methane with the dissolved methane inseawater. Concentrations of methanewere determined by using opticalNDIR absorption technique. Largequantities of methane accumulated inthe internal gas circuit can actively beremoved with a patented exhaust sys-tem. The sensor was calibrated to detectCH4 concentrations as low as approxi-mately 100 nmol L−1, and data werelogged by a 24-bit SmartDI controller.

The HydroC™ methane sensorwas mounted upright at the landerframe to avoid any trapping of gas bub-bles in front of the membrane inlet.Beside the methane sensor, FLUFOwas equipped with an upward-looking

Acoustic Doppler Current Profiler(ADCP; 300 kHzWorkhorse SentinelADCP, Teledyne RD Instruments,USA) and a small stand-alone memoryCTD (Conductivity, Temperature,Depth; XR420, RBR Ltd., Ottawa,Canada). The CTD was also equippedwith an optical backscatter sensor(SeaPoint), which measures light scat-tered by particles suspended in water.

The lander was deployed in thevicinity of a methane gas vent namedFaure bubble site (FLUFO-5; Figure 2).Here, bubble release occurs from dif-ferently sized depressions, which areoften aligned in NW-SE direction; thelargest depression observed by a ROVwas 50 cm in diameter and 15 cm deep(Naudts et al., 2010). These observa-tions clearly showed that the depres-sions are formed by the often violentrelease of bubbles. Naudts and cowork-ers observed that the bubbles entrainedsediment particles, which then get car-ried away by the water currents, creat-ing the depressions and a sedimentoutfall away from the venting hole.The data obtained with the HydroC™sensor depict pulses of CH4 emission(Figure 3a), ranging between 150 and200 nmol L−1. Water samples obtainedfrom the ambient bottomwater during

May/J

a parallel deployment of another lander(BIGO-4; Figure 2) at the same heightabove the sediment water interface likethe sensor showed two distinct peakswith CH4 concentrations of 189 and190 nmol L−1, respectively (Linke et al.,2010), which are in the same range ofthe measurements obtained with theHydroC™ sensor. On the other hand,the data shown here depict that thesensor needed some time of relaxationafter it had experienced high a peakof CH4 before it was able to recordanother sudden increase.

However, the pulses seen in theHydroC™ sensor data correspond withincreases in the backscatter strength de-tected in all four beams of the upward-looking ADCP (Figure 3d). The “flares”(presumed to be bubbles) persisted for10–60 min, and some of them coveredalmost the whole acoustic depth range(100 m) of the ADCP. No associatedsignal was observed in the turbidity dataobtained from the CTD (Figure 3a).The occurrence of the flares does notseem to be related to a sudden or tidalhydrostatic pressure drop (Figure 3c).In fact, some of the outburst occurredduring high tide and at maximum cur-rent velocities of more than 20 cm s−1

(Figure 3b). This is in agreement withresults of Linke et al. (2010) from an-other lander deployment next to theFaure Site (FLUFO-4; Figure 2). Theyfound CH4 concentration fluctuationsin both the ambient bottom water andthe chamber water, which coincidedwith tidally induced fluctuations ofcurrents and acoustic backscatter flaresin the ADCP record.

Furthermore, these measurementsagree very well with ROV observationsin the area reporting highly variablespatial bubble release rates and bubblesizes, with periods of low activity, alter-nating with periods of violent out-bursts (Naudts et al., 2010).

FIGURE 2

Launch of the Fluid Flux Observatory (FLUFO) with the video-guided launcher on top showing thedifferent scientific modules integrated in the back (a) and in the front (b) of the lander (modifiedfrom Linke et al., 2010).


High-Sensitive MethaneSensor (HISEM)Development

A new methane sensor, whichshould fulfill the needs for scientific


trace gas (i.e., CH4) investigations inthe oceans and for subsea leak detec-tion, is currently under development(www.martec-era.net). The sensortechnology is based on laser diode IRabsorption technology that provides

l

excellent detection limits at goodsignal-to-noise ratios. The actualconfiguration is a 2,000-m depth-ratedversion with a (Contros HydroC™)membrane-inlet configuration. Thesystem was tested in the laboratory

FIGURE 3

Physical measurements obtained simultaneously to the changes in CH4 concentration during deployment of FLUFO-5. Top to bottom: (a) turbiditychanges and CH4 concentration, (b) depth-averaged velocity time series, (c) local hydrostatic pressure, and (d) ADCP backscatter (all four beams).

against various partial pressures ofmethane dissolved in water in a tem-perature controlled (4–15°C) water-filled calibration tube. The wateris continuously equilibrated at atmo-spheric pressure with standard gas mix-tures of methane (3–200 mol-ppm) insynthetic air pressure bottles (∼200 bar).

Gas exchange with atmosphere is pre-vented in the semiclosed system. Theresponse of the sensor signal is con-tinuously recorded during testing, andequilibration of the signal is establishedwith a response time (t65) of about10 min after partial methane pressureshave been changed in the tube (Figure 4).

May/J

The detection limit of the HISEM ofabout 1 ppm could be determined witha signal-to-noise ratio of 5.

The sensor output (ppm unit) showsgood linearity (R2 = 0.99998) againstthe known pressure bottle concentra-tions given in mol-ppm with an offsetof ∼1 ppm (Figure 4). The correlation

FIGURE 4

Methane concentrations determined with the HISEM system, which was placed in a water-filled calibration tube at 4°C. The water is equilibrated withmethane by using different gas mixtures (3, 5, 11, 50, 100, and 200 mol-ppm CH4 in pressure bottles).


was confirmed by parallel determina-tion of dissolved methane concentra-tions sampled from the calibrationtube. These methane analyses wereconducted by using head space sam-pling technique and subsequent gaschromatographic analysis.

HISEM Offshore Test SiteTo test the sensor performance

for methane plume detection and itsoffshore practicability in operatingthe system with a small work-class re-motely operated vehicle (i.e., HYSUB20 ROV), a 3-day offshore campaignwas performed in November 2012near Santa Barbara, Southern Califor-nia. The offshore test site that was cho-sen is named Farrar Seep and is locatedwithin the Coal Oil Point seep area inthe inner Santa Barbara Channel (Fig-ure 5a) about 1,300 m east of the Uni-versity of California, Santa Barbaraarea (Figure 5b). The Farrar Seep is anatural hydrocarbon seep that is indi-cated by gas bubble release from theseafloor at about 22 mbsl. Naturalgas seeps in the area of the innerSanta Barbara Channel can be charac-


terized by low seepage activity, and thegas composition of bubbles emanatingfrom the seafloor consists of up to 90%of methane and 10% of higher hydro-carbons (e.g., Leifer et al., 2006). Asgas bubbles dissolve and exchangetheir gas content during uplift in sea-water, dissolved gas plumes are formedin thewater columnand the initial hydro-carbon content of the bubbles decreases(e.g., Leifer & Patro, 2002; Clark et al.,2003; McGinnis et al., 2006). Numer-ous natural gas and oil seeps exist inthe inner Santa Barbara channel, whichlead to general high background con-centrations of dissolved methane inthe area (e.g., ∼20–100 nmol L−1;Clark et al., 2000).

l

To measure dissolved methaneconcentrations during ROV dives, theHISEM system was mounted parallelbehind the upper bumper bar of theHYSUB 20 (Figure 6). The head (mem-brane inlet) of the HISEM prototypewas connected with a plastic tube to asuction inlet at the front of the ROV. Ametal filter was mounted to the suctioninlet, and the inlet area was monitoredpermanently by cameras. A secondtube connected the suction inlet by ay-adapter with a CTD and a commer-cial leak detection device (Combinationof HydroC-CH4, Fluorometer). Aconstant water flow through the tubingwas guaranteed by two Seabird pumps,which operated inline the tubes.

FIGURE 6

(a) Deployment of the HYSUB 20 ROV from the starboard site of M/V Danny C. (b) The HISEMprototype (marked by white rectangle) was mounted behind the bumper bar of the ROV.

FIGURE 5

(a) Map of the overall area of the Santa Barbara channel in Southern California. The test site is marked as a small open circle. (b) Farrar Seep offshoretest site (shaded rectangle), about 1,300 m east of the University of California Santa Barbara area.

ROV-BasedSensor Measurements

TwoN-S and fourW-E ROVdiveswere conducted at the 28th and 29thof November 2012 at the estimated lo-cation of the Farrar Seep (Figure 5b).The average length of a ROV divetrack was about 250 m. Furthermore,one vertical dive track was conductedat the estimated center of the seep (Fig-ure 5b). During all dives, water depth,temperature, conductivity (SV48CTD,Sea and Sun Technology) and methanesensor data (HISEM) were recordedcontinuously. However, the ROVstopped every 15 m for 1–2 min toincrease the total measuring time.The homogeneous temperatures ofabout 15.8°C and salinities of about33.4 PSU measured during the ROVdives indicate a well-mixed watercolumn in this coastal area duringNovember 2012. The water depth ofthe test area is about 16–30 mbsl andROV dive tracks plotted in Figure 7were performed above seafloor at ele-vations of 2 and 12 m, respectively.Due to strong currents in the area andespecially a current direction and cur-rent speed change between the 28thand 29th of November 2012 (GoletaPoint buoy data, SCCOOS.org), nav-igating the ROV was challenging anddeviations from predefined track lineswere about 15 m. The Farrar Seep loca-tion could be verified at 119°49.836′Wand 34°24.157′N (WGS84) by mea-suring a CH4 concentrationmaximum(up to 260 ppm and 334 nmol L−1,respectively) while crossing the cen-tral seepage site with the ROV at 2 melevation above seafloor (Figure 8).The minimum concentration of dis-solved methane, which was determinedin water masses in the test area, was∼50 nmol L−1.

The center of the main seepageactivity of the Farrar Seep was also

FIGURE 7

ROV dive tracks conducted in the test area “Farrar Seep.”

FIGURE 8

Methane sensor signal recorded with HISEM about 2 m above the seafloor during ROV track line 1at the 28th and 29th of November 2012. Track line 1 is the N-S profile and crosses the Farrar Seep(Figure 3).


indicated by the ship’s echo-sounder(acoustic blankening by gas bubbles)and ground-truthing by ROV (videoobservations). Note that measuringthe dissolved gas content within thebubble streams emanating from theseafloor did not increase the HISEMsensor concentration signal. The con-centration pattern of November 28thcould be verified by following the sametrack on the 29th (Figure 8). The devi-ation of methane concentration data ofabout 25 nmol L−1 measured withinthe central seep area is possibly relatedto a weakening of the local current re-gime on the 29th (http://sccoos.org/).

The dimensions of the main (dis-solved) methane plume were about


50–150 m around the seepage site(Figure 9). However, dissolved gasplumes with high methane concentra-tions (up to 560 nmol L−1) were alsoexamined towards the east and southat about 12 m above the seafloor (Fig-ure 9). In general, the dissolved meth-ane concentrations are highest towardsthe south, which could indicate a pre-ferred rotation of the methane plumedirection from east to south and thenwest (Figure 9).

ConclusionsSubsea determination of dissolved

methane concentration can be con-ducted by using infrared absorption

l

technology. The technology, mea-suring the partial pressure of methanein a separated gas chamber, is com-bined with a membrane inlet, whichseparates high-pressure conditionsof the deep sea from the gas chamberat normal pressure. Recent advancesin laser diode technology led alsoto cost-effective and high-sensitiveinfrared absorption units. The newlydesigned high-sensitive methane sen-sor (HISEM), which combines laserdiode infrared absorption with mem-brane inlet technology, closes a gapbetween the needs of small and lesssensitive methane sniffers used foroffshore leak detection (Oil & GasIndustry) and oceanographic trace gas

FIGURE 9

Spatial methane concentrations at Farrar Seep measured by HISEM during ROV dives. The concentration of methane is given in ppm. The total rangemeasured by HISEM corresponds to dissolved methane concentrations of 50–560 nmol L−1 in the test area.

determinations down to 1–2 nmol L−1

of CH4 (equilibrium concentration ofseawater with the atmosphere).

Membrane-inlet IR absorptionsensors can be used for long-termmeasurements at the seafloor (e.g.,lander-based deployment). The de-termination of, for example, varyingmethane concentrations in the vicinityof a methane seep have to be combinedwith determination of temperature,salinity, and pressure variations, aswell with current measurements (i.e.,using ADCPs). This combination isthe basic information used to deter-mine (dissolved) gas fluxes from natu-ral seeps or leaking constructions atthe seafloor.

Focused release of methane fromsubsea seeps and of rising plumes ofdissolved methane can be monitoredwith ROV-based IR sensor technol-ogy. However, the quantification ofmethane release also needs some basicoceanographic information about thelocal current regime and physicalparameters (T, S, P) along the ROVdive tracks. This could be realized dur-ing onboard CTD measurements andan upward-looking ADCP deployedat the seafloor. A miniaturization ofthe recently developed high sensitivemethane sensor (HISEM) is prere-quisite to use this technology onboardinspection class ROVs.

AcknowledgmentsThe HISEM system development

is funded by the German MinistryBMWi (Project No. 03SX301) withinthe European funding initiativeMARTECH (ERA-NET, MaritimeTechnologies). Wintershall Noordzeeis kindly acknowledged for support-ing the offshore operations and ROVadaptions.

Corresponding Author:Mark SchmidtGEOMAR Helmholtz Centre forOcean Research KielWischhofstr. 1-3,24148 Kiel, GermanyEmail: [email protected]

ReferencesCampbell, K.A., Francis, D.A., Collins, M.,

Gregory, M.R., Nelson, C.S., Greinert, J.,

& Aharon, P. 2008. Hydrocarbon seep-

carbonates of a Miocene forearc (East Coast

Basin), North Island, New Zealand. Sediment

Geol. 204:83-105. http://dx.doi.org/10.1016/

j.sedgeo.2008.01.002.

Clark, J.F., Leifer, I., Washburn, L., &

Luyendyk, B.P. 2003. Compositional changes

in natural gas bubble plumes; Observations

from the Coal Oil Point marine hydrocarbon

seep field. Geo-Mar Lett. 23:187-93. http://

dx.doi.org/10.1007/s00367-003-0137-y.

Clark, J.F., Washburn, L., Hornafius, J.S., &

Luyendyk, B.P. 2000. Dissolved hydro-

carbon flux from natural marine seeps to the

Southern California Bight. J Geophys Res.

105(11):509-11,522.

Faure, K., Greinert, J., Schneider von

Deimling, J., McGinnis, D.F., Kipfer, R., &

Linke, P. 2010. Methane seepage along

the Hikurangi Margin of New Zealand:

Geochemical and physical data from the water

column, sea surface and atmosphere. Mar


j.margeo.2010.01.001.

Greinert, J., Lewis, K.B., Bialas, J., Pecher, I.A.,

Rowden, A., Bowden, D.A., … Linke, P.

2010. Methane seepage along the Hikurangi

Margin, New Zealand: Overview of studies in

2006 and 2007 and new evidence from visual,

bathymetric and hydroacoustic investigations.

Mar Geol. 272:6-25. http://dx.doi.org/

10.1016/j.margeo.2010.01.017.

Judd, A.G., & Hovland, M. 2007. Seabed

Fluid Flow: The Impact on Geology, Biology,

and the Marine Environment. New York:

May/J

Cambridge Univ. Press. 475 pp. http://dx.doi.

org/10.1017/CBO9780511535918.

Leifer, I., Clark, J., & Luyendyk, B. 2006.

Simulation of a Subsurface Oil Spill by a

Marine Hydrocarbon Seep. MMS OCS Study

2006-050. Santa Barbara, CA: Coastal

Research Center, Marine Science Institute,

University of California. MMS Cooperative

Agreement Number 14-35-01-00-CA-31063.

81 pp.

Leifer, I., & Patro, R.K. 2002. The bubble

mechanism for methane transport from the

shallow sea bed to the surface: A review and

sensitivity study. Cont Shelf Res. 22:2409-28.

http://dx.doi.org/10.1016/S0278-4343(02)

00065-1.

Linke, P., Sommer, S., Rovelli, L., &

McGinnis, D.F. 2010. Physical limitations

of dissolved methane fluxes: The role of

bottom-boundary layer processes. Mar Geol.

272:209-22. http://dx.doi.org/10.1016/

j.margeo.2009.03.020.

McGinnis, D.F., Greinert, J., Artemov, Y.,

Beaubien, S.E., & Wuest, A. 2006. Fate of

rising methane bubbles in stratified waters:

How much methane reaches the atmosphere?

J Geophys Res. 111:C09007. http://dx.doi.

org/10.1029/2005JC003183.

McGinnis, D.F., Schmidt, M., DelSontro,

T., Themann, S., Rovelli, L., Reitz, A., &

Linke, P. 2011. Discovery of a natural CO2

seep in the German North Sea: Implications

for shallow dissolved gas and seep detection.

J Geophys Res (Oceans). 116:C03013.

http://dx.doi.org/10.1029/2010JC006557.

Naudts, L., Greinert, J., Poort, J., Belza, J.,

Vangampelaere, E., Boone,D.,…DeBatist,M.

2010. Active venting sites on the gas-hydrate-

bearing Hikurangi Margin, Off New Zealand:

Diffusive versus bubble-released methane. Mar


j.margeo.2009.08.002.

Pfannkuche, O., & Linke, P. 2003.

GEOMAR landers as long-term observatories.

Sea Technol. 44(9):50-5.

Schneider von Deimling, J., Greinert, J.,

Chapman, N.R., Rabbel, W., & Linke, P.

2010. Acoustic imaging of natural gas seepage


in the North Sea: Sensing bubbles controlled

by variable currents. Limnol Oceanogr Meth.

8:155-71. http://dx.doi.org/10.4319/lom.

2010.8.155.

Townend, J. 1997. Subducting a sponge:

minimum estimates of the fluid budget of the

Hikurangi margin accretionary prism. Geol

Soc N Zeal Newsl. 112:14-6.

Westbrook, G.K., Thatcher, K.E., Rohling,

E.J., Piotrowski, A.M., Pälike, H., Osborne,

A.H.,… Aquilina, A. 2009. Escape of methane

gas from the seabed along theWest Spitsbergen

continental margin. Geophys Res Lett.

36:L15608. http://dx.doi.org/10.1029/

2009GL039191.


P A P E R

Development of the JiaolongDeep Manned Submersible
A U T H O RWeicheng CuiHadal Science and TechnologyResearch Center,Shanghai Ocean University,Shanghai, China andChina Ship Scientific ResearchCenter, Wuxi, Jiangsu, China
A B S T R A C T
Deep sea exploration and exploitation are of increasing interest to 21st century
scientists, and both manned and unmanned deep submergence vehicles arenecessary means for deep sea exploration. To fulfill the requirements of deep seaexploration for China Ocean Mineral Resources R&D Association (COMRA), a deepmanned submersible was in the process of development in China from 2002 to2012 and was named Jiaolong in 2010. The purpose of this paper is to introducethe development process from a historical point of view, including the design,realization of components, assembly, open-water tank test, and sea trials of theJiaolong deep manned submersible. The technical difficulties encountered ateach stage, and their solutions are briefly described. The future development trendsfor deep manned submersibles are pointed out.Keywords: deep manned submersible, Jiaolong, development process, design,realization of components

and inspired people to dive below thesurface and explore the forms of life

Introduction

The deep seas have fascinated hu-mans for centuries. The flow of newideas has traveled through the centuries

that exist in the abyss. A detailed histor-ical account of human beings’ deep divesinto the oceans and discovery of its truemysteries can be found in, for example,Forman (2009), Kohnen (2009), andSagalevitch (2009). Deep sea explora-tions are indispensable, not only for theinvestigation of marine creatures, micro-organisms, minerals, and other resourceshidden under the deep water but also forgeophysical research into the structureand behavior of the earth. Deep sea ex-ploration and exploitation are of increas-ing interest to people in the 21st century,and bothmanned and unmanned deepsubmergence research vehicles are neces-sary means for deep sea exploration(Momma, 1999; Committee on FutureNeeds in Deep Submergence Science[CFNDSC], 2004;Committee on Evo-lution of the National OceanographicResearch Fleet, 2009; Fletcher et al.,2009; Barry & Hashimoto, 2009).

Although unmanned submersiblessuch as autonomous underwater vehi-

cles (AUVs) and remotely operatedvehicles (ROVs) have advantages incertain aspects, Human operated vehi-cles (HOVs) remain the central ele-ment in the selection of modern toolsat the service of knowledge acquisition(Kohnen, 2009; CFNDSC, 2004;Rona, 2000) because the sense of visionis the most important of all the fivehuman senses.

In the 1970s and 1980s, the U.S.manned submersible DSV Alvin, madea number of important discoveries inmarine scientific research (http://www.whoi.edu/). This promoted an upsurgein the international community todevelop deep manned submersibles.The first 6,000 m submersible was theU.S. Navy’s Sea Cliff. France, theformer Soviet Union and Japan thendeveloped another four 6,000-m deepmanned submersibles, namely, Nautile(Jarry, 1986), MIR I & II (Sagalevitch,2009), and Shinkai 6500 (Nanbaet al., 1990; Takagawa et al., 1995),

May/J

respectively. In the 1990s, Russiastarted to develop two more 6,000-msubmersibles, Rus and Consul, thatwere to be fully built in country. Dueto lack of funding, the sea trials ofthese two submersibles were notcompleted until 2011, and both arenow in service of the Russian Navy(http://www.rusnavy.com/news/).

The first proposal to develop a6,000-m deep manned submersiblein China was submitted by the ChinaShip Scientific Research Center(CSSRC) to the then State Commissionof Science and Technology (now theMinistry of Science and Technology)in 1992. Because the public demandfor deep manned submersibles wasnot urgent and the technical risk washigh, the proposal was not approved.

In 1999, the China Ocean MineralResources R&DAssociation (COMRA)submitted a new proposal for a deepmanned submersible, substantiatinga pressing need for such technology


to fulfill their duties. This renewedproposal was approved by theMinistryof Science and Technology of theChinese Government and marked thebeginning of the deep manned sub-mersible project now named Jiaolong.COMRA was appointed as the projectcoordinator and final owner of thissubmersible. The development for-mally began in June 2002, and onJune 30, 2012, the Jiaolong success-fully completed the final dive of itssea trials program.

The purpose of this paper is to ex-amine the development process from ahistorical point of view, including de-sign, realization of components, assem-bly, open-water tank test and sea trialsof the Jiaolong deep manned submers-ible. The paper will be divided intofour sections. After the first introductionsection, the entire development processof the submersible is described in Devel-opment Process of the Deep MannedSubmersible Jiaolong. This process in-cludes the three stages of design, reali-zation of components and assembly,as well as open-water tank test andsea trials. The technical difficultiesencountered at each stage and theirrespective solutions are also brieflydescribed. Latest Technical Specifi-cations of the Jiaolong are listed andFuture Development Trends for DeepManned Submersibles are briefly dis-cussed in section 3. The last Sectionis a summary, and some conclusionsfrom the development process havebeen drawn.

Development Processof the Deep MannedSubmersible JiaolongDesign

The main purpose of the project isto develop a usable submersible, whichis of comparative performance with


existing submersibles, so the technicalinnovations and the application of thelatest immature technologies were notthe emphasis of the project. Our de-sign team began its formal designwork in June 2002. According to therequirements of the quality controlmanual of CSSRC, the design was di-vided into three phases: preliminarydesign, technical design, and detaileddesign.

Having made an in-depth summaryof themissions and the overall technicalindicators specified in the contract, thefollowingfive essential key performancefactors were proposed:(1) the ability to reach a depth of

7,000 m;(2) good maneuverability;(3) the capability of real-time com-

munication and microtopographydetection;

(4) the capability to sample in a hov-ering state at a designated position;and

(5) safety and reliability.After developing the five key perfor-

mance factors, a list of technical chal-lenges was identified for the researchand production of the Jiaolong sub-mersible. This resulted in a selectionof key techniques for each of the per-formance requirements. The chief de-signer then divided the vehicle into12 subsystems according to the re-quirements of the overall technicalindicators and characteristics of theparticipating research units and person-nel. The 12 subsystems were (1) theoverall performance and general arrange-ment; (2) structure system; (3) outfitting;(4) ballast and trim adjustment system;(5) propulsion system; (6) electric powerand distribution system; (7) lighting,video, and VHF communication sys-tem; (8) control system; (9) acousticsystem; (10) hydraulic and operationsystem; (11) life support system;

l

(12) underwater emergency jettisonsystem. The principal challenge forsuch a division is the handling of inter-face relations among numerous subsys-tems. The interface relations betweenany two subsystems should not berepeated or omitted, and this cannotbe guaranteed simply by the chiefdesigner’s experience. Therefore, afour-element method was developedto solve this problem (Cui et al., 2008a,2008b). All the parameters of each sub-systemwere divided into four categories:input, output, support, and restraint.The designers were asked to draw afour-element diagram for each subsys-tem to check the consistency (shown inFigure 1). This allowed us to success-fully handle and reconcile the interfacerelationship among all 12 subsystemsin an orderly and systematic process,thus solving the first and most funda-mental problem for the developmentof the manned submersible.

The preliminary design was a sig-nificant challenge for the designteam, since nobody had any personalexperience with a real deep mannedsubmersible. The components selec-tion, how to arrange them to form acoordinated and functional vehicle,and the weight and displacement ofeach component were all unknown.

Almost all the important pieces ofequipment for the Jiaolong ended upbeing custom parts, which needed tobe developed specifically for us by theproducers. Through communicationand discussion with the relevant pro-ducers, the performance, weight, andvolume of each piece of equipmentgradually became clear.

In order to make each subsystemwork effectively and form a completeentity, able to complete its missionsand tasks, the general arrangement ofthe manned submersible was devel-oped with a modularized structure

and function as a guideline. The keypoint of the design process was to en-sure the reliability and maintainabilityof the submersible.

On February 26–27, 2003, the“Preliminary Design of the JiaolongManned Submersible” was submittedto an experts’ review organized by theState Oceanic Administration, and theproject entered into the phase of tech-nical design. The research and designat this stage were focused on feasibility,reliability, and maintainability consid-erations. A mature design state wasreached, in which the main technicalindicators of the submersible werefixed, through modeling and proto-type testing.

Designing the manned cabin wastechnically difficult because it wasdirectly related to the safety of thecrew. If the design was too conserva-tive, the submersible would be veryheavy, undermining its maneuver-ability and increasing the operationalcost. To find a safe balance, a seriesof problems needed to be solved, suchas calculating the stress concentration

factors at openings, analyzing the ulti-mate strength, choosing the appropri-ate safety factors, determining thefatigue load spectrum, and analyzingthe fatigue life.

Due to time constraints, the teamchose to follow the rules of the ChinaClassification Society (CCS, 1996)and the Russian Maritime Register ofShipping (RS, 2004). Our objectivewas to concurrently meet the two setsof rules. In addition, the finite elementmethod was used to analyze and calcu-late the ultimate strength of themannedcabin (Lu et al., 2004). Specific atten-tion was paid to stress concentrationsnear large openings, such as personnelaccess hatches, observation windows,and penetrators. The use of the finiteelementmethod with contact elementsto analyze this problem was partic-ularly helpful (Li & Cui, 2004).

The fatigue design load spectrumwas first proposed through a statisticalanalysis of the Alvinmanned submers-ible’s historical dive data (Li et al.,2004). Based on this, we analyzed thefatigue life of the stress-concentrated

May/J

areas of the manned cabin, using thesimple analysis method of fatigue lifeaccording to CCS rules (Li et al.,2006). This then provided the mainstructural parameters of the mannedcabin. Next, a Russian company wascontracted to produce the mannedcabin using the “melon petals weldinginto a semi-sphere” method. Afterthat, a pressure test was carried out inRussia. The pressure vessel was instru-mented with strain gauges, and mea-sured stress values were found to bein good agreement with the finite ele-ment analysis results. Based on this testdata, we determined that our designwas reasonable, the manufacturingquality was guaranteed, and the ultimatestrength and fatigue life of the mannedcabin of the Jiaolong was ensured.

However, we clearly knew thatonly a small number of manned sub-mersibles had been produced in theworld, and most of them had beendeveloped by the navy and not classi-fied by classification societies. There-fore, strictly speaking, the designstandards and safety assessment stan-dards of any classification societylacked practical experience. Thus, therationality and scientific foundationof the design standards needed to beinvestigated.

In the development of the Jiaolongsubmersible, 40% of the equipmentwas procured from international com-panies. In order to expand deep seatechnology development in China, anew project to construct a domesticallyproduced 4,500 m manned submers-ible was started in 2009. The main ob-jectives of this project were to reducethe operational cost and to further in-crease the reliability, in addition to thetechnology development.

This compelled the team to com-pare the design standards of all majorclassification societies; they found that

FIGURE 1

The four-element diagram of the system/subsystem.


the strength requirements of thesedesign standards varied widely. Infact it was noted that most existingdeep manned submersibles could notmeet the requirements of the pub-lished standards, see Table 1 andTable 2 (Pan & Cui, 2011a, 2011b).Considering that all these submersiblesoperated safely in the past years, mostclassification societies’ standards canbe considered overly conservative forhadal depth designs.

Hence, we reviewed the analysismethod of the buckling and ultimatestrength of spherical pressure hullunder external pressure (Pan & Cui,2010), unified the description meth-ods of the initial defects, made a seriesof calculations about all the parametersaffecting the ultimate strength of themanned cabin within the range ofpossible variations by finite elementanalyses, and, finally, proposed amore rational ultimate strength for-mula (Pan et al., 2010). By absorbing


the U.S. Navy’s requirements for thecontrol methods of macro stress lev-els and stress concentration (NSSC,1998), a set of complete design stan-dards for a manned cabin was pro-posed (Pan & Cui, 2011a, 2011b).These were adopted by CCS and writ-

l

ten into their “Principles of Check ofDrawing and Inspection of Deep-SeaManned Submersible” (CCS, 2011)as the classification standard for theJiaolong. The new design standard ofthe manned cabin is as follows (Pan& Cui, 2011a, 2011b).

TABLE 1

Calculation results of existing titanium pressure hulls.

Actual Design of Existing Pressure Hull

Submersible
Alvin Consul (Rus) Nautile ShinKai 6500 New Alvin Jiaolong
Depth (m)
4,500 6,000 6,000 6,500 6,500 7,000
Operating pressure fromGL2009 (MPa)

45.45
60.6 60.6 65.65 65.65 70.7
Internal diameter (m)
2.0 2.1 2.1 2.0 2.1 2.1
Design thickness (mm)
49 71 62–73 75 71–72 76–78
Safety factor
1.5 1.5 1.5 1.55 1.5 1.5
Minimum proof thickness calculated by design rules (mm)

DNV1988
59.7 78.2 78.2 79.5 83.5 88.7
BV1989
62.6 86.9 86.9 90.4 94.9 103.0
LR1989
75.6 99.0 99.0 101.0 106.1 113.4
CCS1996
51.5 72.7 72.7 77.8 78.9 85.2
RS2004
51.1 68.9 68.9 72.7 74.0 79.2
GL2009
59.4 83.2 83.2 86.1 90.2 97.8
ABS2010
72.0 97.6 97.6 103.3 105.1 112.9
TABLE 2

Check of actual designs of existing pressure hulls to the existing design rules.

Submersible
Alvin Consul(Rus) Nautilea
Shinkai6500

NewAlvin
Jiaolong
DNV1988
X X X X X X
BV1989
X X X X X X
LR1989
X X X X X X
RS2004
X √ √ √ X X
CCS1996
X √ √ X X X
ABS2009
X X X X X X
GL2009
X X X X X X aNote: Due to the lack of detailed material data for each submersible, the average material properties forTitanium TA6V4 have been employed in the calculations made in Table 1. Dr. Jean-Francois Drogou fromIFREMER pointed out in the review process that, for Nautile, the TA6V4 alloy is forged in beta phase that givesa resistance Re0.2 of more than 950 MPa and the design of Nautile hull satisfied with the ABS requirements.

The ultimate strength of the titanium alloy spherical pressure hull of deepmanned submersible is calculated by the following equations:

Pu ¼ 1� kΔR

� �σutR

þ σutRm

� �

ð1Þ

k ¼ a þ b × exp

�cΔR

� �

� d�tR

�

þ eΔR

� �2

þ f�tR

�2

� gΔR

� �3

� h�tR

�3!

ð2Þ

where a = −15.63; b = 606.6; c = 264.6; d = 72.72;e = 3 × 104; f = 882.5; g = 1.2 × 106; h = 3969;

σu = the guaranteed minimum tensile strength of titanium alloy used;t = thickness of the pressure hull;R = internal radius of the spherical pressure hull;Rm = mean radius of the hull.Δ = the deviation between actual spherical surface and regular spherical surface.The application range of this empirical equation is 0.025 ≤ t/R ≤ 0.08 and

Δ/R ≤ 0.01. The safety factor for this equation is 1.5.Besides the above ultimate strength formulae, additional stress limitations that

control the overall stress level and local stress concentrations need to be consideredin the makeup of a safety standard. The stress limitations are suggested to be asfollows:a. The average shell membrane stress at maximum operating pressure shall be

limited to two thirds of the minimum specified yield strength of the material.b. The highest combined value of average shell membrane stress and bending

stress (excluding effects of local stress concentrations) at maximum operatingpressure shall be limited to 3/4 of the minimum specified yield strength of thematerial.

c. The maximum compressive peak stress at any point in the hull, including ef-fects of local stress concentrations, shall be limited to 4/3 of the minimumspecified tensile yield strength and shall not exceed the compressive ultimatestrength of the material. The maximum tensile peak stress at any point in thehull, including effects of local stress concentrations, shall be limited to themin-imum specified yield strength of the material.

d. Currently, there is no condition to set a reliable standard for fatigue strengthassessment. It is recommended that designers and users of the submersibleshould pay attention to this issue and carry out the fatigue analysis at all stressconcentration areas using the commonly accepted approaches.This newly established safety standard was used to optimize the structural

design of the 4,500 m manned cabin (Pan & Cui, 2012).To solve the problem of a comprehensive optimized design for Jiaolong’s

hydrodynamic layout and lines, we systematically collected a comprehensiveamount of relevant data of all the submersibles in the world, analyzed their charac-teristics of shape and hydrodynamic performance, proposed assessment indicatorsof the submersible’s maneuverability, established theoretical and experimental fore-casting methods for its hydrodynamic performance (including speed and maneu-verability), clarified the main design variables of the manned submersible’s

May/J

hydrodynamic performance, deter-mined the properties of judging itscomprehensive hydrodynamic perfor-mance, and provided an optimizeddesign method for the manned sub-mersible’s hydrodynamic performanceunder constraints. When studyingthe spatial hydrodynamic characteris-tics of a deep manned submersible ina complex nonlinear deep sea environ-ment, the submersible’s hydrodynamiccharacteristic of a 6 degree-of-freedommotion was obtained. An equation ofa 6 degree-of-freedom spatial motionwas derived, the corresponding hydro-dynamic coefficients were determinedfrom wind tunnel tests, rotating armbasin tests, towing tank model tests,and a mathematical model was estab-lished, which can be used to forecastand analyze the submersible’s maneu-verability (Ma & Cui, 2004, 2005,2006a, 2006b; Cui & Ma, 2009; Panet al., 2010).

The ballast and trim regulating sys-tem consists of a variable ballast systemwith a flow rate of up to 3 L/min, oneballast tank with displacement of upto 1,500 kg and a set of trim regulat-ing systems with a flow rate of up to15 L/min. The variable ballast systemconsisted of a titanium sphere 840mmin diameter, a hydraulically operatedsuper-pressure pump, driven by an8-kW DC motor, oil-immersed sole-noid valve, pressure balance valve, etc.Commercial companies were commis-sioned to produce a super-pressurepump and DC motor. The trim regu-lating systemmainly consisted of tankslocated at the bow and aft, an oil con-trol box (containing oil pump, oilmotor and valves) and a piping system.The trim regulating system’s powercame from the submersible’s hydraulicsource; the oil motor used the powerof the hydraulic source to drive thelow-pressure oil pump to provide


low-pressure oil of 1.5 MPa, whichpumped the mercury back and forthbetween the bow and aft, thereby ad-justing the trim angle.

The propulsion system was com-posed of four main ducted propellersat the stern, two rotatable ducted pro-pellers midship, and one tunnel pro-peller at the bow. According to thecontract, the submersible had to havea maximum speed of up to 2.5 kmand a cruising speed up to 1 km.

The electrical power and distribu-tion system supplied electrical powerto the whole submersible. Thus, itshould have the following functions:providing 110 V DC to the thrustermotors, hydraulic source motor, andvarious lights outside the submersible;providing 24 V DC to underwateracoustic equipment, motion controlsystem, sensors and instruments inthe cabin, etc.; having the ability tosupply sufficient power for the wholesubmersible according to the require-ments of the power consumption ofall the equipment and time history oftypical missions; offering a distribu-tion switch and protection for themain power supply circuit. Accord-ing to the requirements of the sub-contract’s technical specifications, theelectrical power and distribution sys-tem needed to equip a set of oil-immersed silver-zinc batteries anda set of corresponding distributionsystems.

Based on the statistics of the loaddiagram, the load of the main batterywas 86.7 kwh; that of the secondarybattery was 24.2 kwh; thus, the totaloutput capacity of the main and sec-ondary batteries should be not lessthan 110.9 kwh. The emergencybattery is used if both the main andsecondary batteries fail and mainlyprovides power to emergency lighting,the underwater acoustic phone, and


the life support system. In addition,when the submersible is floating atthe surface, the spare battery providesa VHF radio transmitter and strobelights. The emergency battery electriccapacity is 5.4 kwh.

Lighting and video equipmentserve to provide a source of underwaterlighting for operators and observerswhen sailing, operating and observing.Video equipment serves to photographunderwater targets and then stores thevideos and photos. The main role ofthe VHF communication system is tooffer radio voice communication onthe surface. Originally, the Jiaolongwas equipped with one 3CCD colorvideo camera, two 1CCD color videocameras, one black-and-white low-light level video camera, one underwa-ter camera, two HMI lamps, two HIDlamps, four quartz halogen lamps, onepan and tilt, and one set of video cam-eras in the cabin. After the 3000-m-depth class sea trials, the lighting andcamera systems were upgraded. TheJiaolong is now equipped with 10 LEDlamps, 4 HMI lamps, 2 HID lamps,1 quartz halogen lamp, 2 HD videocameras, 2 1CCD video cameras,1 camera, 1 black-and-white low-light level camera, 2 pan and tilts,1 HD video recording system, and1 video camera in the cabin.

The hydraul ic sys tem is theJiaolong ’s most important sourceof auxiliary power, since it supplieshydraulic power for buoyancy adjust-ment, trim adjustment, and under-water sampling. The hydraulic systemhas rate of pressure up to 21 MPa, rateof flow up to 25 L/min, and a maxi-mum input power up to 10 kW.

The Jiaolong’s design requirementsalso included the function of a sam-pling system, to accomplish a seriesof tasks relying on the power fromthe hydraulic system. The sampling

l

system included two hydraulic manip-ulator arms, a hydrothermal sampler tokeep high-pressure liquid, a sedimentsampler, underwater drilling equip-ment, and a sampling basket.

The main function of the controlsystem is to collect information fromsensors on the submersible, displayand record the data and then controlthe implementation mechanisms. Thismakes the submersible competent tocruise, observe, and perform tasks.

The acoustic system is mainly com-posed of an underwater acoustic com-municator, high-resolution side-scansonar, acoustic Doppler speedometer,obstacle avoidance sonar, imagingsonar and long-distance ultra-shortbaseline positioning sonar. Installedon both sides of the submersible, thehigh-resolution side-scan sonar isused to measure the seabed’s micro-topography and targets in the wateror on the seabed and draw real-timethree-dimensional maps of the scene.The mechanical scanning imagingsonar is located at the bow of the sub-mersible and serves to detect the sur-rounding environment and forwardtargets in the water for the pilot’s ob-servation. On one hand, it helps thepilot to search for targets; on the otherhand, it prevents the submersible fromrunning into obstacles, thus ensuringits safety.

Seven anticollision sonars are in-stalled at the bow and on both sidesof the submersible. They are able tomeasure the distances between thesubmersible and obstacles in six di-rections, helping the pilot to avoid ob-stacles to ensure safety. They alsomeasure the distance from the sub-mersible to the seabed to control thesubmersible.

The Jiaolong is equipped witha transponder at its top; thus, it canbe positioned by the long-distance

ultra-short baseline positioning sonarinstalled on the mother ship. Further,the combination of this and the GPS isable to provide the absolute position ofthe submersible. In addition, a longbaseline system was added after the3000-m-depth sea trial.

The life support system providesoxygen (O2) in three ways: normal op-eration, emergency operation and oxy-gen masks. Each oxygen subsystem isindependent of the others in order toensure the system’s reliability. Thereis one set of normal life support sys-tems providing O2 for 12 h, one setof emergency life support systems pro-vidingO2 for 60 h, and one set of maskbreathing systems provide O2 for 12 h.These systems use high-pressure oxy-gen cylinders to produce oxygen andhigh-efficiency lithium hydroxide toabsorb carbon dioxide.

The Jiaolong is China’s first deepmanned submersible. Consequently,during the period of sea trials, Chinadid not have a rescue capability. There-fore, safety was the most importantconcern for the Jiaolong at that time.In the design phase, we proposed“able to dive downward, able to comeback” as the design principle. Theoverall guiding philosophy was thatthe Jiaolong should be fitted with allthe existing emergency means of exist-ing manned submersibles. In addition,we would ensure that the emergencyjettison system was sufficiently reli-able through numerous prototypeevaluations and sea trials. Havinglearnt from the successful experienceof existing manned submersibles, thefollowing self-rescue measures wereimplemented based on the idea of a“redundant design”:(1) A ballast jettisoning mechanism:

the maximum weight of the solidballast is 1.3 tons. As long as it jet-tisons all of them, the submersible

will certainly become positivelybuoyant and thus can go upward.The emergency jettison systemwas designed to be able to dropballasts using two independentmechanisms, one electromagneticand the other hydraulic. Ballastscan be jettisoned by a single action.What is more, as soon as a powerfailure starts, the ballast will beautomatically discarded.

(2) A main battery box jettisoningmechanism: in a situation inwhich solid ballasts cannot be dis-carded, the main battery box thatweighs 1.2 tons will be discardedso that the submersible can becomepositively buoyant. This functionis realized through the use of twoelectrical-explosive bolts (one ofthem offers redundancy) with aseries connection. As long as oneof them works, the main batterybox can be dropped.

(3) A mercury jettisoning mechanism:all the mercury weighs 480 kg andcan easily be jettisoned. However,jettisoned mercury will pollutethe environment; therefore, this isnot encouraged unless in an emer-gency. In many cases, even if every-thing else cannot be jettisoned,jettisoning the mercury will beable to provide sufficient buoy-ancy to make the submersible goupward.

(4) A mechanism used to jettison thehydraulic manipulator arms: ifthe manipulators are bound, theycan be completely released.

(5) A ballast tank system: when thesubmersible goes up to only 10me-ters below the sea surface, the pilotshould start the ballast tank systemto drain off water. This is capable ofproviding 1.5 tons of buoyancy,ensuring that the Jiaolong has a rel-atively large freeboard on the water.

May/J

(6) A rescue buoy: if the submersibleis trapped on the seabed, the rescuebuoy will be released by detonat-ing electrical-explosive bolts. Thestrobe light on the buoy will leadthe mother ship to find this buoy,and thenmother ship can catch thebuoy and drag the submersible upby a cable, as long as 9,000m, link-ing the submersible and the buoy.

(7) A mechanism used to jettison thesampling basket: The sampling bas-ket is also connected with the sub-mersible by an electrical-explosivebolt, so that if it is bound, it canbe jettisoned.On the basis of the general ar-

rangement drawings of the prelimi-nary design, a 1:1 model was made totest and confirm the maintainabilityand rationality of the general arrange-ments. The final configuration of theJiaolong’s general architecture and thepreliminary determination of its hydro-dynamic performances were firmedup through repeated attempts andmodifications.

On August 29–30, 2003, the“7000 m Manned Submersible ’sTechnical Design” was submitted tothe expert review organized by theState Oceanic Administration andreceived positive confirmation.

The detailed design phase involvedmainly the generation of constructiondrawings on the basis of the technicaldesign, establishing a reliable supplychannel of each piece of equipmentand doingmodel tests for critical pieces.In this phase, the team also establisheda joint debugging test process andimplementation method, technicalprinciples of assembly and an overalltesting process.

On Apr i l 13–14, 2004, the“Detailed Design of the 7000 mManned Submersible” was submittedto the expert review set by the State


Oceanic Administration and receivedacceptance.

Realization of ComponentsIt was very difficult for China to

produce some of the equipment forthe Jiaolong. The project team alsoknew that the manned cabin couldnot be produced in-country. However,Russia had developed two mannedsubmersibles (Rus and Consul ) andproduced three titanium alloy mannedcabins, and the Russians were willingto produce one for China. Thus, aRussian institution was commissionedto produce the Jiaolong’s mannedcabin, and the pressure acceptancetest was done at the Krylov Shipbuild-ing Research Institute. On January 29,2003, COMRA signed a contract withthe Russian Krylov Institute to pro-duce a 7,000-mmanned submersible’stitanium alloy structure.

The buoyancy foam was originallyplanned to be procured from Russiabut its specific weight was around0.65. This would have broughtthe submersible’s total weight up to25 tons, which exceeded the techni-cally allowed limit of 22 tons. Later,an American company, Emerson &Cuming, Inc., was willing to supplybuoyancy material, DS32, with a spe-cific gravity of 0.525. A contract wassigned and sent to the U.S. govern-ment for approval. The U.S. govern-ment agreed but added restrictions onthe DS32 foam and only authorizingEmerson to provide its DS35 foam,a type of buoyancy material with a spe-cific gravity of 0.561.

Contracts for other importedequipment were signed with foreignfirms, in most cases, following thetechnical design. The imported equip-ment reached Shanghai Customs insuccession from March 2005 untilNovember 2005. However, since


China’s exemption policy was notclear, the customs formalities for ourimported equipment were not com-pleted until September 2006. OnSeptember 25, 2006, the last piece ofimported equipment reached CSSRCfrom Shanghai Customs, and a pres-sure acceptance test was done on eachpiece of pressurized equipment.

AssemblyDomestically produced equipment

arrived in steady succession from theend of 2004. The submersible’s tita-nium alloy frame was made in Russiabut was not delivered until November2005. Therefore, a temporary steelframe was produced locally based onthe detailed design drawings for thetrial installation at the end of 2004in order to shorten the installationtime. In September 2006, when allthe equipment and components, im-ported or domestically produced, hadreached CSSRC, we started the generalassembly of the submersible’s hull,bracket, equipment, outfitting, piping,cable, etc., into a complete submers-ible, in order to experiment a completeintegration on a 1:1 steel frame andmade preliminary interfaces debug-ging tests.

After receiving the submersible’sfinal titanium frame at the final assem-bly site, we first measured the frame’sgeometric dimensions and then deter-mined the reference position of theinstallation. A laser was used duringthe installation process to monitorany changes in the reference position.

Unfortunately, the Russian-madetitanium frame’s mounting dimen-sions were beyond the permissible tol-erance, causing significant problemswith the installation of brackets forlight shells and equipment brackets.Because these brackets were made onthe basis of the drawings and thus

l

matched the steel frame, they did notmatch the titanium frame. Hence, theinterface between the brackets and tita-nium frame needed to be repeatedlymodified. What is more, oxygen cut-ting cannot be used for the titaniumalloy and this resulted in a tremendousamount of sanding work. This wasnot estimated beforehand, and thus,it delayed the schedule of the finalassembly.

When installing brackets weldedto the frame, we first drew hanginglines or, if necessary, made installationtemplates. The parts were then sandedto make them match each other,weighed, and then spot welded. Rele-vant work fixtures were installed toensure an accurate installation andreduce welding distortion. Later, weinspected the assembly, and if itpassed, we finally welded it for a fur-ther inspection.

Firstly, 37 pieces of buoyancyblocks were installed externally in the10 sections of the titanium frame;43 pieces of buoyancy blocks were in-stalled internally (including 13 piecesat the head); and then 161 bracketswhich connected the buoyancy blocksand titanium frame were welded orbolted to the titanium frame andbolted to the buoyancy blocks. Itshould be noted that, when installingthe buoyancy blocks on the titaniumframe, we had to ensure a fair shapeof the submersible.

Since the distortion tolerance of thewelded titanium frame was far morethan that of the machined buoyancyblocks, the thickness of each correc-tion gasket had to be modified severaltimes during the installation; this wasextremely cumbersome and heavywork.

If the installation of the buoy-ancy blocks contradicted the arrange-ment of the frame or interfered with

equipment or cables, modified lineswould be drawn on the buoyancyblocks and they would be modifiedby hand. All these buoyancy blockshad to be hand modified on site untilthey dimensionally fitted the frameand other interfaces representing a lotof added assembly work had to bedone on the site.

Having installed the external buoy-ancy blocks and adapted them to therequired positions, welders weldedthe blocks’ brackets to the frame.The welding distortion of the externalbuoyancy blocks had to be consideredcarefully, otherwise the bolts on the ex-ternal buoyancy blocks could not bescrewed into the nuts on the brackets.Our welders attempted to resolve thisproblem again and again, and thendecided to control the welding distor-tion by reducing the welding currentand discontinuously welding on bothsides.

The acoustic equipment such as thebathymetric side-scan sonar, Dopplerspeedometer and motion sensors re-quired very high installation accuracy,and trial and error methods with specialtools were used to meet the accuracyrequirement.

The onshore combined debuggingwas carried out after the final assembly.Themain purpose of the onshore com-bined debuggingwas to ensure our workcould enter the next phase, the open-water tank test, by checking whetherthe interfaces between the equipmentand subsystems were accurate, whethertheir communications were right, andwhether they were able to performtheir required functions. The onshorecombined debugging was a bridgebetween the final assembly and theopen-water tank test; it marked a stepforward from equipment to system.The following aspects were the focusof the onshore combined debugging:

(1) We confirmed that connectionswere correct and reliable by repeat-edly checking the cable connec-tions among the equipment.

(2) The onshore debugging check.All the electrical circuits and signaltransmission paths were checkedwith battery power.

(3) The results of onshore debuggingwere determined by the completeoperating equipment checkpoints.The final assembly and onshore

combined debugging tests altogethertook 1 year and lead to the open-water tank test phase that followed.

Open-Water Tank TestCSSRC has an open-water tank

shaped like a circular basin, with a di-ameter of 85 m and a maximum depthof 15 m, which is equipped with an in-door dock. This dock is 30 m × 8 m ×8 m and has a 30-ton bridge crane(shown in Figure 2). Thanks to thisfacility, we could add an open-water

May/J

tank test to Jiaolong’s Technical TestPlan process to identify problems inwater and solve them, before going tosea trials.

The open-water tank test included(1) underwater system verification,(2) checking the operation and inte-gration performances of each sub-system and their required functions,(3) confirming the practicability ofthe designed work procedure of eachmission, (4) examining the accuracyand reliability of the exchange of in-formation flow and control flow, and(5) accumulating the relevant data.

The main tasks of the open-watertank test completed include thefollowing:1. measuring the submersible’s weight

and center of gravity;2. determining drainage volume and

coordination of the center of buoy-ancy by a balancing test;

3. determining the mooring pro-pulsive force and the relationshipbetween the rotation rates of eachpropeller;

4. evaluating the submersible’s abilityto regulate the trim angle, accu-mulating experience by rehearsingaccording to “Operating Rules ofEach Post of Sea Trials (draft)” andthe rules of implementation com-piled for each dive in a sea trial;

5. debugging the control performanceof a 6 degree-of-freedom motion ofthe submersible and submersible’scomprehensive performance oflanding on the seabed;

6. debugging the submersible’s auto-matic control function;

7. simulating the task of sampling;8. debugging the interfaces between

the submersible and tools; and9. debugging the comprehensive oper-

ational capability of the submersible.The open-water tank test lasted for

110 days, from October 3, 2007 to

FIGURE 2

Open-water tank (a) and correspondingdock (b).


January 20, 2008. Underwater testswere performed 53 times during thisperiod, and 18 staff members enteredthe manned cabin. A series of malfunc-tions were discovered during thesetests, such as the malfunctions of thepropellers, trim pump, high-pressureair relief valve, anti-collision sonar,hydraulic source, and seawater valves,as well as the rupture of the compen-sation film of the main and secondarybattery boxes. These were all detectedand corrected on site.

Another task of the open-watertank test was to train the pilots and op-erational team for the sea trials. Severalstaff members participated in the train-ing program and dove the Jiaolongduring the tank tests. This includedMr. Cong Ye, who was the chiefdesigner of the general arrangementsof Jiaolong. He was also a memberof the team that participated in aSino-U.S. joint deep diving voyage,and he was the only member whoobtained two chances to dive withAlvin. He was consequently selectedto be the chief pilot of the Jiaolongin the sea trials. At the same time,the two future professional pilots,Mr. Wentao Fu and Mr. Jialing Tang,continued their training and wouldbe given as many operational chancesas possible.

On January 17, 2008, an expertpanel organized by COMRA came toCSSRC to test the Jiaolong. The testprogram consisted of a wide varietyof functional tests which included(1) displacing water from ballast tank,(2) executing the automatic directionalnavigation, (3) sample hovering at adesignated position, (4) operating thelife support system, and (5) jettisoningthe solid ballast. The results of this testsequence demonstrated to the satisfac-tion of the expert panel that the submers-ible’s functions and performance met


the relevant requirements. Two monthslater, on March 2, 2008, in Wuxi, theState Oceanic Administration held aPre-Delivery Inspection and Confirma-tion meeting. It was concluded that thesubmersible had met all the technicalqualifications required by the sea trials’outline and that the Jiaolong was readyfor ocean sea trials. More detailed intro-duction about open-water tank tests canbe found in Hu et al. (2008) and Cuiet al. (2009).

Sea TrialsThe sea trials were the final phase

of the Jiaolong’s development but alsorepresented the greatest risk. Successor failure of the entire project de-pended on the results of these tests.The purpose of the sea trials was tocomprehensively examine the techni-cal performance, operation efficiency,safety and reliability of the completedsubmersible, to determine whether itsperformance met all contract require-ments, design files and relevant norms,to ultimately decide if it could be deliv-ered, accepted and declared in opera-tional activity.

The Jiaolong’s sea trial was a com-plex system engineering process, whichnot only concerned the submersibleitself and the technical state of thesurface support system but also in-volved preparation of the sea area forthe trial, organizing the testing per-sonnel, preparing the relevant vessels,preparing a contingency plan for thesea trial, and many logistics factors.Our project team began to do the rel-evant preparation work in 2005, in-cluding drawing up sea trial files,purchasing spare parts, refitting themother ship, selecting and preliminar-ily training pilots and operators, andselecting the appropriate sea areas forthe trial. At the end of 2007, an outlineof the sea trial was approved by the

l

State Oceanic Administration, andthe organization of the sea trial wasformally established.

The supporting ship was clearlyidentified as a very important elementfor the sea trials test as well as forall future utilization. However, due torestricted project funding, it was im-possible to develop a specific mother-ship for Jiaolong in a parallel effort, soa temporary ship had to be selectedand refitted for the sea trials test. Aftervarious efforts, an older oceanographicsurvey ship, the Xiangyanghong-09,was chosen (Figure 3). A series of mod-ifications and updates were necessarysuch as adding an A frame at the sternand modernization of the living quar-ters. In the aft deck, a special platformwas constructed for the storage andmaintenance of the submersible.

The main technical characteristicsof the ship are given below.Length overall: 112.05 mBreadth moulded: 15.2 mDraft: 5.5 mDisplacement: 4,435 tonsEndurance: 60 daysCruising speed: 12 km,maximum

speed: 15 kmDeck equipment: 6,000 m electricalwinch for geology, 2,000 m hydraulicwinch for acoustic communication.Scientific investigation equipment:Sub-bottom profiler, Acoustic Doppler

FIGURE 3

Xiangyanghong-09 and its main facilities.

Current Profiler (ADCP), Conductiv-ity, Temperature, and Depth (CTD),Ultra-short Baseline (USBL)Laboratories: e.g., dry lab, wet lab,biology lab, submergence control cen-ter, network centerTotal space for crew aboard the ship:96, including 45 ship operators, 2med-ical staff and a maximum of 49 for thetest team.

Figure 3 shows a pic ture ofXiangyanghong-09 and its main facil-ities. The main challenge for the testteam was the ship’s lack of a dynamicpositioning system (DPS). All station-keeping had to rely on the ship master’sexpertise tomaintain safe operating dis-tances between the supporting ship andthe submersible. Close tracking of thesubmersible’s position was critical inorder tomaintain reliable acoustic com-munication during test dives.Otherwise,we would lose sight of the submersibleafter surfacing at the end of a dive.

The 1,000-m-level sea trials wereformally started in 2009, and thisproved the most difficult time in allthe sea trials. Firstly, the season wasnot appropriate as it was typhoon sea-son in the South China Sea. Secondly,problems due to the lack of experience,improper design and assembly errorsoccurred in almost every dive andthis continuously accrued the percep-tion of project risk. Thirdly, overallsupport for the project was precariousand the team understood that therewas no route of retreat. If this initialsea trial phase failed, the whole projectwould be canceled. Consequently, theonly option was to solve these prob-lems on site and complete the necessarysea trial objectives. A brief summaryfor all the test dives of Jiaolong sub-mersible is given in Table 3.

In the 1,000-m class tests, the mostchallenging technical problem was thecommunication system. Both the sur-

face VHF and underwater acousticcommunication systems did notwork. The second problem was causedby having to use expired silver-zincbatteries due to funding problems.The batteries failed in most of thedives and this added a significantmaintenance work load on the crew.Another major problem encounteredduring the 300-m sea trial was theleakage of watertight cables and con-nectors. Through the joint efforts ofthe test team, we successfully finishedthe 1,000-m class test and a new Chi-nese record of 1,109 m was achieved.

With this hard won initial sea trialssuccess, the ministry of science andtechnology approved the follow-onproject of “Improvement of JiaolongManned Submersible’s Key Tech-niques and Research of 3,000-m-Level Sea Trial.” The new project wasaimed at completely and comprehen-sively resolving the problems revealedin the 1,000-level sea trial, includingthe sitting on bottom ability, conve-nience and reliability of the groundingdetection system, reliability of the hy-draulic system, contacts between thestabilizers and the seabed, reliabilityof the video system, reliability of thecontrol system, and the reliability ofthe acoustic system.

FromMay 31 to July 18, 2010, theJiaolong’s 3,000-m-level sea trial wasperformed in the South China Sea, inwhich the Jiaolong made 17 dives andthe maximum dive depth of 3,759 mwas reached.

All required test missions were suc-cessfully completed in this sea trial, andthe results even exceeded our expecta-tions, especially since the submersiblehad run without any anomaly in thelast two dives. The ministry of scienceand technology saw a great possibilityof success from this sea trial, and the re-search project was opened to the media

May/J

and the public. More detailed infor-mation on the first two phases of seatrials can be found in Liu et al. (2010)and Cui et al. (2010).

After the success of the 3,000-m-level sea trial, the ministry of scienceand technology directly approved the“Technical Improvement of the JiaolongManned Submersible and 5,000–7,000 m Sea Trials.” This contract re-quired that all improvements, the 5,000–7,000 m sea trial and the full classifica-tion work of the Jiaolong should bedone within 2 years. Classificationwork had not been included in theinitial contract for the development.Before the 5,000-m-level sea trial,some comprehensive technical im-provements were made in five aspects,including (1) the upgrade of the videosystem, (2) the improvement of theoperation system, (3) grounding detec-tion system, (4) positioning system and(5) acoustic system. These improve-ments were all focused to enhance theJiaolong’s safety and capacity for work.

China’s contracted sea area of poly-metallic nodules was deliberatelychosen as the test site for the 5,000-m-level sea trials. This would simulta-neously conduct an investigation intothe natural resources and do scienti-fic research in this area, as required ofCOMRA by the International SeabedAuthority. This sea trial was done fromJuly 1 to August 18, 2011. Since thetest area was far away from China, theworking time was just 2 weeks, becausethe travel time was almost one month.The weather conditions remained terri-ble throughout the entire sea trials. Wewere only able to make five dives withdepths of 4,027 m, 5,057 m, 5,188 m,5,184 m, and 5,180 m, respectively.Apart from the first dive, each of theother four dives successfully landedthe submersible on the seabed severaltimes and all systems functioned well.


TABLE 3

Dive log of Jiaolong sea trials.

Dive No.

48 M

Date

arine Techno

Test Site

logy Society

Depth (m)

Journal

Brief Description

1
2009.8.3 Dock 1 2.4 Carried out the launch and recovery exercise with no persons in the cabin.
2
2009.8.3 Dock 1 2.4 Carried out the launch and recovery exercise with persons in the cabin. Carried out theground detection.
3
2009.8.4 Dock 1 2.4 Finished the launch and recovery test and the grounding detection.
4
2009.8.15 Dock 2 2.4 Finished the launch and recovery test and the check of submersible functions; Finishedthe grounding detection.
5
2009.8.15 Dock 2 2.4 Finished the launch and recovery test. Debugging the VHF system and acousticcommunication system.
6
2009.8.16 A1 2.4 Finished the VHF communication test but the acoustic communication system did notwork.
7
2009.8.17 A1 2.4 Carried out the functional test for the main ballast system and due to high positivebuoyancy, the submersible did not dive.
8
2009.8.18 A1 38 Finished the diving test by thrusters.
9
2009.8.20 A1 2.4 Finished the test of the acoustic communication system with the main engines of themothership stopped or operated.
10
2009.8.21 A1 41 Finished the functional test of filling and draining of the main ballast water tank; Finishedthe check of navigational functions; Finished functional test for the variable ballast systemand the trim adjustment system.
11
2009.8.23 A1 44 Finished the releasing test of the emergency buoy.
12
2009.8.30 B1 2.4 Debugging of the acoustic communication system.
13
2009.8.31 B1 213 Balance test in 200-m depth; Debugging the acoustic communication system.
14
2009.9.7 B1 100 Established the Morse code communication. Test terminated due to severe weather.
15
2009.9.12 Dock 2 2.4 Finished grounding detection; Debugging the acoustic communication system.
16
2009.9.13 B1 335 Finished the functional check for manual navigation with and without computer. Finishedfunctional test for automatic orientation, depth, height and hovering; Finished thefunctional test for trim adjustment; Finished the functional test for the variable ballastsystem; Finished sitting on bottom.
17
2009.9.18 B1 268 Finished functional test for automatic orientation, depth, height and hovering; Finishedthe test for payload; Finished the functional test for side scan sonars and hydrothermalsamplers; Finished the functional check for manual navigation without computer.
18
2009.9.20 B1 292 Finished sitting on bottom and placing markers; Finished functional test of the acousticcommunication system; Testing the automatic height function.
19
2009.9.30 Dock 2 2.4 Finished grounding detection and functional test for the variable ballast system;Debugging the acoustic communication system.
20
2009.10.3 C2 1,109 Finished un-powered diving and floating; Finished the whole process of submersiblediving operation.
21
2010.5.29 Dock 1 2.4 Finished the launch and recovery training for the operators; Finished the functional checkfor the ground inspection system.
22
2010.6.8 A1 2.4 Submersible turning when launched in water and one towing rope cut the cable of onetransducer. Fortunately the transducer was saved by the persons in the rubber boat.Underwater filming test terminated.
continued

TABLE 3

Continued

Dive No.
Date Test Site Depth (m) Brief Description
23
2010.6.9 A1 37 Finished the balance test; Finished underwater filming by a diver; Finished functionalcheck for side scan sonars, the acoustic communication system; Finished navigationalfunction check.
24
2010.6.11 B1 291 Finished solid blast weight check in 300-m depth; Finished the navigational functioncheck; Finished the functional test for automatic depth, orientation and height; Finishedthe functional check for the acoustic communication system; Finished sitting on bottom.
25
2010.6.12 B1 291 Finished the navigational function check; Finished the functional test for automatic depth;Finished sitting on bottom; Finished operational exercise for one pilot-in-training.
26
2010.6.20 D2 2,068 Finished un-powered diving and floating; Finished solid blast weight check in 2,000-mdepth; Finished the functional check for the acoustic communication system.
27
2010.6.22 D2 3,039 Finished un-powered diving and floating; Finished functional test for the acousticcommunication system; Finished the functional check for the ground inspection system.
28
2010.6.27 D2 2.4 Diving test terminated when the leakage alarm of the backup battery box was sounding.
29
2010.6.28 D2 2,104 Finished un-powered diving and floating; Finished functional check for the groundinspection system; Finished operational exercise for one pilot-in-training.
30
2010.6.29 B1 286 Finished bathymetric bottom mapping with two side-scan sonars; Finished navigationalfunction check; Finished sitting on bottom and placing the marker; Carried out targetsearch exercise.
31
2010.6.30 B1 286 Finished the search of the marker placed in the previous dive; Finished sitting on bottomand placing the new marker; Finished operational skill test for two pilots-in-training.
32
2010.7.6 A1 33 Finished video recording and photo taking of the whole sea trial scene by the helicopter;Finished functional check for ground inspection system and the updated video system ofthe submersible.
33
2010.7.8 D2 2,088 Diving terminated due to the high short-circuit current value displayed by the groundinspection system; Finished functional test for the hydraulic system.
34
2010.7.9 D2 3,757 Finished un-powered diving and floating; Finished sitting on bottom 5 times; Operatedthe hydrothermal sampler but without obtaining water samples.
35
2010.7.11 D2 1,134 Diving test terminated when the leakage alarm of the backup battery box was sounding.
36
2010.7.12 D2 3,757 Finished un-powered diving and floating; Finished sitting on bottom, taking hydrothermalsamples and placing markers.
37
2010.7.13 D2 3,759 Finished un-powered diving and floating; Finished sitting on bottom and taking samples;Finished bathymetric bottom mapping with two side-scan sonars. Finished operationalskill test for two pilots-in-training.
38
2011.6.28 Dock 1 2.4 Finished launch and recovery training for the operators; Finished functional check for theground inspection system.
39
2011.6.29 Dock 1 2.4 Finished launch and recovery training for the operators; Finished functional check for theground inspection system.
40
2011.7.20 E3 4,027 Finished un-powered diving and floating; Finished solid blast weight check in 4,000-mdepth; Finished functional check for ultra-short baseline and ground inspection system;Finished submersible functional check.
continued


TABLE 3

Continued

5

Dive No.

0 M

Date

arine Techno

Test Site

logy Society

Depth (m)

Journal

Brief Description

41
2011.7.25 E3 5,057 Finished un-powered diving and floating; Finished solid blast weight check in 5,000-mdepth; Finished functional check for ultra-short baseline; Finished submersible functionalcheck; Finished sitting on bottom and taking videos and photos.
42
2011.7.27 E2 5,188 Finished unpowered diving and floating; Finished navigational function check in 5,000-mdepth; Finished sitting on bottom and taking videos and photos; Finished bathymetricbottom mapping of 3 km with two side-scan sonars.
43
2011.7.29 E1 5,184 Finished un-powered diving and floating; Finished navigational function check in 5,000-mdepth; Finished sitting on bottom, placing markers; Finished sampling of manganesenodules and biologies; Recorded many videos and photos. Finished the whole process ofsubmersible diving operation.
44
2011.7.31 E1 5,180 Finished un-powered diving and floating; Finished functional check in 5,000-m depth;Finished sitting on bottom, placing markers; Finished biological sampling; Recordedmany videos and photos. Finished the whole process of submersible diving operation.
45
2012.6.1 Dock 1 2.4 Finished launch and recovery training for the operators.
46
2012.6.15 G1 6,671 Finished un-powered diving and floating; Finished functional check in 5,000 m and6,000-m depths; Obtained 405 ml water sample in normal pressure and 430 ml watersample in high pressure of 28 MPa.
47
2012.6.19 G1 6,965 Finished unpowered diving and floating; Finished functional check in 6,000-m depth;Finished sitting on bottom, placing markers, micro topography survey by side-scansonar; Obtained a sediment sample of 14 cm length; Obtained 435 ml water sample innormal pressure; Finished the whole process of submersible diving operation.
48
2012.6.22 G1 6,963 Finished un-powered diving and floating; Finished functional check and safety verificationin 6,000-m depth; Finished sitting on bottom, placing markers; Obtained three sedimentsamples whose lengths are 21 cm, 16 cm and 18 cm, respectively; Obtained 400 mland 435 ml water samples in normal pressure and 123 ml water sample in high pressureof 15 MPa; Caught one white sea cucumber and two pieces of nodular mineral samples.Finished the whole process of submersible diving operation.
49
2012.6.24 G1 7,020 Finished un-powered diving and floating; Finished functional check in 7,000-m depth;Finished sitting on bottom, placing markers; Finished submersible tele-communicationtest with the person in Beijing office; Obtained 440 ml and 445 ml water samples innormal pressure and 345 ml water sample in high pressure of 35 MPa; Finished thewhole process of submersible diving operation.
50
2012.6.27 G1 7,062 Finished unpowered diving and floating; Finished functional check and safety verificationin 7,000-m depth; Finished sitting on bottom, placing markers; Obtained two sedimentsamples whose lengths are 21 cm and 27 cm, respectively; Obtained 440 ml and435 ml water samples in normal pressure; Caught one white sea cucumber. Carried outthe biological trapping experiment with bait. Finished the whole process of submersiblediving operation.
51
2012.6.30 G1 7,035 Finished unpowered diving and floating; Finished submersible functional check andsafety verification in 7,000-m depth; Finished sitting on bottom and sample takingexercise; Finished functional verification of two domestic propellers by long distancecruise. Finished the whole process of submersible diving operation.
Note: Dock 1 = The Jiangyin International Container Terminal; Dock 2 = The mooring site near Sanya; A1 = South China Sea (109°09′E, 18°01′N); B1 = South China Sea(112°36′E, 19°01′N); C2 = South China Sea (110°26.8′E, 17°28.45′); D2 = South China Sea (116°28′E, 18°36′N); E1 = China poly metallic nodules contract area in thenortheast Pacific (154°12′W, 10°08′N); E2 = China poly metallic nodules contract area in the northeast Pacific (154°11′W, 8°40′N); E3 = The northeast Pacific (151°7′W,4°48′N); G1 = The exclusive economic zone (EEZ) of the Federated States of Micronesia in the southern Mariana Trench (141°58′E, 10°59′N).

Furthermore, each dive completed itsmission, including reaching the positionwith the required depth, and no invaliddives occurred as in the previous twoyears (Table 3). This indicated that theimproved submersible’s technical statewas stable, our test team had matured,and the sea trial’s plan and contentproved to be reasonable and scientific.As a difference from the previous trials,this sea trial was completely open to themedia, and several dives were broadcastlive on CCTV. A detailed introductionto the 5,000-m depth class test can befound in Cui et al. (2011a, 2011b).

The final 7,000-m depth class testwas carried out from June 3 to July 16,2012. Although it faced the uniquechallenges of 7,000 m sea trials, theteam was now fully prepared, carriedout very careful operations at sea, andcould analyze and solve all the prob-lems that occurred on-site. After verystrict inspection by the on-site accep-tance group of experts, it was concludedthat the test team completed the 7,000msea trials safely and successfully. Thedetails are reported in Cui et al. (2013).The sea trials tests showed that all thedesign requirements were fully real-ized. This 7,000-m sea trials test isthe deepest of the same type of deepmanned submersibles in the world.The success of the Jiaolong submersibledemonstrates that China’s mannedsubmersible technology has enteredthe club of the countries of the highestinternational standard.

Latest TechnicalSpecifications of the Jiaolongand Future DevelopmentTrends for Deep MannedSubmersibles

Although the technical specifica-tions of the Jiaolong submersible have

been reported in several references (Cui et al., 2008a, 2008b, 2009; Liu et al., 2008),some of them changed during the technical improvements made in the sea trialsperiod. The latest technical specifications of the Jiaolong are as follows:Maximum operational depth: 7,000 mMain dimensions: 8.4 × 4.2 × 3.4 mInner diameter of manned cabin: 2.1 mWeight in air: 22 tonsCrew: 1 pilot, 2 scientistsPower supply: Silver-zinc battery, >110 kWhControl mode: Automatic fixed height, fixed direction, fixed

depth, hovering and manual operationsNavigation system: Ultra-short baseline system, long baseline

system motion sensors + acoustic Dopplerspeedometer

Operation: Two seven-function hydraulic manipulatorarms

Lighting and video system: HD video camera system, LED lightsCommunication: Underwater acoustic communication,

underwater acoustic telephone, VHFcommunication + GPS positioning

Figure 4 is a photo of the latestdesign. Although the Jiaolong is thedeepest and latest manned submersiblewithin the second generation of three-crew operational submersibles, it can-not be claimed to represent the lateststate-of-the-art technology for deepmanned submersibles.

When comparing our Jiaolong withthe U.S.’s new Alvin (http://www.whoi.edu/alvin/), it can be found that,on the one hand, the inner diameters oftheir manned cabin are the same; theirpayload, battery power, lighting andimaging systems, maneuverability and

May/J

ability to sample hovering at an accu-rate position are almost equivalent. Theirrespective underwater working timeon the bottom are within the limit ofmanned submersibles. The Jiaolong alsopossesses similar performances in theaspects of automatic drive, underwateracoustic communication, capabilities ofnavigation and positioning.

On the other hand, based on a greatdeal of operational experience, the newAlvin has a more outstanding designof viewports, in-cabin man-machineinterface, data acquisition and instru-ment interfaces.

The future technical trends ofmanned submersibles mainly focus onthe aspects of economy, comfort and en-vironmental protection (Cui et al., 2011a,2011b). Two requirements will bringrevolutionary changes to the design ofmanned submersibles. One is to signif-icantly reduce the running costs, mak-ing manned submersibles comparablewith unmanned submersibles in thisaspect (Hawkes, 2009). The other isto miniaturize the in-cabin facilities,making a more comfortable in-cabin

FIGURE 4

Jiaolong deep manned submersible.


environment (Taylor & Lawson, 2009).Concerning environmental protection,this mainly relies on elimination of mer-cury as the substance to adjust the trim,ormodify the substance used in the emer-gency jettison system. Solid ballast ironwould not be used, or only if necessary.

These two last requirements primar-ily depend on the technique of a large-flow high-pressure sea water pump.The newAlvin’s design team once con-sidered making a breakthrough in thisaspect at the very beginning of theirdesign; however, they ultimately hadto give up this idea due to tremendoustechnical difficulties and lack of funds.In spite of their failure to make sucha breakthrough, it ought to have at-tracted sufficient attention as a directionof technical development. Another im-provement direction of great potentialimportance is themanipulator. Currentseven function manipulators obtainedcommercially are far away from the ca-pabilities of human hands and discov-ering how to develop a manipulatorsimulating the human hand wouldbring a revolutionary change to the sub-mersible community. At the same time,another direction of great interest tohuman beings is challenging the fullocean depth, which needs a break-through in improving submersibles’speed of descending and ascending.In this direction, James Cameron com-pleted his record-breaking MarianaTrench dive in March 26, 2012, withhis Solo sub Deepsea Challenger (http://deepseachallenge.com/). Another twofull ocean depth manned submersiblesare under development (http://www.doermarine.com/; http://deepflight.com/).

Summary and ConclusionsThis paper is the first one describing

the development process of the Jiaolong


deepmanned submersible from a histor-ical point of view, including the design,realization of components, assembly,open-water tank test and sea trials.The main technical challenges met ateach stage are briefly explained, supple-mented by the many references pub-lished during the process for interestedreaders. The Jiaolong submersible isbriefly compared with the new Alvin,which is regarded as being the best inthe deep sea community, and it is con-cluded that, except for the design of theviewports, in-cabin man-machine inter-face, data acquisition and instrumentinterfaces, most of the performancesand functions are equivalent in additionto the Jiaolong’s ability to dive 500-mdeeper. This indicates that the Jiaolongcan also be regarded as one of the mostadvanced deep manned submersiblesfor scientific operations. The 7,000-mdepth capability is the deepest of thesame type of deepmanned submersiblesin the world. The success of the Jiaolongsubmersible demonstrates that China’smanned submersible technology hasentered the realm of countries with thehighest international standard. Finally,future technical trends of manned sub-mersibles were also discussed.

AcknowledgmentsThe development of the Jiaolong

submersible was sponsored by theMinistry of Science and Technologyunder the contracts of Developmentof a 7,000-m Deep Manned Submers-ible (Project No. 2002AA401000);1,000-m Depth Class Sea Trials ofJiaolong Submersible (Project No.2009AA093201); Key TechnologyImprovements and 3,000-m DepthClass Sea Trials of Jiaolong Submers-ible (Project No. 2010AA094001);Operational Technology Improve-ment and 5,000–7,000 m Sea Trials

l

(Project No. 2011AA09A101). TheState Oceanic Administration (SOA)is the organization department of thesea trials of Jiaolong. China OceanMineral Resources R&D Association(COMRA) of SOA is specificallyresponsible for organizing the sea trials.In total, 96 members from 18 domesticorganizations jointly completed the seatrial tests. The author of this paper isvery grateful to all the team members.Comments, suggestions and languageimprovements from Dr. Don Walsh,Dr. Kurt Uetz, Dr. William Kohnenand another two reviewers to this paperare greatly appreciated. My M.Sc.student, Mr. Hao Li did some trans-lation for this paper and his help isalso appreciated. The work reportedin this paper was completed in my for-mer organization China Ship ScientificResearch Center, and now this authorhas been moved to Shanghai OceanUniversity.

Author:Weicheng CuiChina Ship Scientific Research Center,Jiangsu, 214082, ChinaCurrently with Hadal Science andTechnology ResearchCenter, ShanghaiOcean University, Shanghai, ChinaEmail: [email protected]

ReferencesBarry, J.P., & Hashimoto, J. 2009. Revisiting

the Challenger Deep using the ROV Kaiko.

Mar Technol Soc J. 43(5):77-8.

China Classification Society. 1996. Classifi-

cation and Construction Rules for Diving

Systems and Submersibles [S]. Beijing: People’s

Transportation Press (in Chinese).

China Classification Society. 2011. Principles

for plan approval and inspection of deep-sea

manned submersible. A supplement to

CCS1996, Internal document (in Chinese).

Committee on Evolution of the National

Oceanographic Research Fleet. 2009. Science

at Sea: Meeting Future Oceanographic Goals

with a Robust Academic Research Fleet, Na-

tional ResearchCouncil, ISBN: 0-309-14558-9,

120 pages, The National Academies Press. This

free PDF was downloaded from http://www.

nap.edu/catalog/12775.html.

Committee on Future Needs in Deep

Submergence Science. 2004. Future Needs in

Deep Submergence Science: Occupied and

Unoccupied Vehicles in Basic Ocean Research,

National Research Council, ISBN: 0-309-

52917-4, 152 pages, The National Academies

Press. This PDF is available from the National

Academies Press at: http://www.nap.edu/

catalog/10854.html.

Cui, W.C., Hu, Z., Ye, C., & Pan, B.B. 2011a.

Current status and future trends for deep

manned submersibles. J Cent South Univ T.

42(Suppl. 2):13-20 (in Chinese).

Cui, W.C., Hu, Z., Ye, C., & Xu, Q.N.

2010. First two phases of sea trials of deep

manned submersible “Jiaolong.” In: Ocean

Innovation Conference (OI2010), October

17-20, 2010. St. John’s, Canada: The Marine

Institute of Memorial University, Marketing

& Business Development Division.

Cui, W.C., Liu, F., Hu, Z., Zhu,M., Guo,W.,

& Liu, C.G. 2013. On 7000 m sea trials of

the manned submersible “JIAOLONG.” Mar

Technol Soc J. 47(1):67-82. http://dx.doi.org/

10.4031/MTSJ.47.1.2.

Cui, W.C., Liu, F., Hu, Z., Zhu,M., Guo,W.,

& Wang, C.S. 2011b. 5000 m sea trials test of

the deep manned submersible “JIAOLONG.”

Shipbuilding of China. 52(3)(Serial No. 197):

1-14 (in Chinese).

Cui, W.C., Liu, Z.Y., & Xu, Q.N. 2008a. The

four-element method for the design of large

complex engineering system. Shipbuilding of

China. 49(2)(Serial No. 181):1-12 (in Chinese).

Cui, W.C., & Ma, L. 2009. Hydrodynamic

problems involved in submersible design.

In: Proceedings of the 9th National

Congress on Hydrodynamics and 22nd

National Conference on Hydrodynamics,

9-29 (in Chinese). Beijing: Ocean Press.

Cui, W.C., Xu, Q.N., Liu, F., Hu, Z., Wang,

X.H., & Zhu, M. 2009. Design, construc-

tion and open water tank test of the deep

manned submersible “Harmony.” In: Keynote

paper presented in the Sixth International

Symposium on Underwater Technology

UT2009, Wuxi, China, April 2009. Wuxi:

China Ship Scientific Research Center.

Cui, W.C., Xu, Q.N., Liu, T., Hu, Z., Yang,

Y.N., Hu, Y., & Ma, L. 2008b. Design and

construction of the deep manned submersible

“Harmony.” J Ship Sci Technol. 30(1):17-25

(in Chinese).

Fletcher, B., Bowen, A., Yoerger, D.R., &

Whitcomb, L.L. 2009. Journey to the

Challenger Deep: 50 years later with the

Nereus hybrid remotely operated vehicle.

Mar Technol Soc J. 43(5):65-76. http://

dx.doi.org/10.4031/MTSJ.43.5.26.

Forman, W. 2009. From Beebe and Barton

to Piccard and Trieste. Mar Technol Soc J.

43(5):27-36. http://dx.doi.org/10.4031/

MTSJ.43.5.14.

Hawkes, G. 2009. The old arguments of

manned versus unmanned systems are about

to become irrelevant: New technologies are

game changers. Mar Technol Soc J. 43(5):

164-8. http://dx.doi.org/10.4031/MTSJ.

43.5.21.

Hu, Z., Wang, X.H., Zhu, M., Xu, Q.N.,

& Cui, W.C. 2008. Tank test of the deep

manned submersible “Harmony”[C]//. In: the

3rd Asian-and Pacific Diving and Underwater

TechnologyConference, Shanghai, June 13–15,

2008. Shanghai: Shanghai Society of Naval

Architects and Marine Engineers.

Jarry, J. 1986. SAR, NAUTILE, SAGA,

ELIT-four new vehicles for underwater work

and exploration: The IFREMER approach.

IEEE J Oceanic Eng. OE-11(3):413-7. http://

dx.doi.org/10.1109/JOE.1986.1145192.

Kohnen, W. 2009. Human exploration of

the deep seas: Fifty years and the inspiration

continues. Mar Technol Soc J. 43(5):42-62.

http://dx.doi.org/10.4031/MTSJ.43.5.30.

Liu, F., Cui, W.C., & Li, X.Y. 2008.

Introduction to design and construction of

May/J

deep manned submersible “Harmony.” In:

Paper presented in International Conference

of INMARTECH2008, All Seasons Hotel,

Toulon, France, October 8–10, 2008. Toulon:

Ifremer.

Liu, F., Cui, W.C., & Li, X.Y. 2010. China’s

first deep manned submersible. Jiaolong. Science

China—Earth Sci. 53(10):1407-10. http://

dx.doi.org/10.1007/s11430-010-4100-2.

Li, X.Y., & Cui, W.C. 2004. Contact finite

element analysis of a spherical hull in the deep

manned submersible. J ShipMech. 8(6):85-94.

Li, X.Y., Cui, W.C., & Zhang, W.M. 2006.

Fatigue life reliability analysis of titanium

manned spherical shell. J Ship Mech.

10(2):82-6 (in Chinese).

Li, X.Y., Liu, T., Huang, X.P., & Cui, W.C.

2004. Determination of fatigue load spectrum

for pressure hull of a deep manned submers-

ible. J Ship Mech. 8(1):59-70 (in Chinese).

Lu, B., Liu, T., & Cui, W.C. 2004. Ultimate

strength of pressure spherical hull in deep-sea

manned submersibles. J Ship Mech. 8(1):

51-8 (in Chinese).

Ma, L., & Cui, W.C. 2004. Simulation of

dive motion of a deep manned submersible.

J Ship Mech. 8(3):31-8.

Ma, L., & Cui, W.C. 2005. Path following

control of a deep-sea manned submersible

based upon NTSM. China Ocean Eng.

19(4):625-36.

Ma, L., & Cui, W.C. 2006a. Heading control

study of a deep manned submersible at a low

speed and large drift angles based on fuzzy

sliding mode control. Ocean Eng. 24(3):74-8

(in Chinese).

Ma, L., & Cui, W.C. 2006b. Dynamic

system identification of manned submersible

in horizontal plane. Shipbuilding of China.

47(2):76-81 (in Chinese).

Momma, H. 1999. Deep ocean technology at

JAMSTEC. Mar Technol Soc J. 33(4):49-64.

http://dx.doi.org/10.4031/MTSJ.33.4.6.

Nanba, N., Morihana, H., Nakamura, E., &

Watanabe, N. 1990. Development of deep

submergence research vehicle “SHINKAI


6500.” Technical Review of Mitsubishi Heavy

Industries, Ltd. 27(3):157-68.

Naval Sea Systems Command (NSSC).

1998. System Certification Procedures and

Criteria Manual for Deep Submergence

Systems. Revision A, SS800-AG-MAN-010/

P-9290. Published by Direction of Commander,

Naval Sea Systems Command, 0910-LP-

018-2840.

Pan, B.B., & Cui, W.C. 2010. An overview

of buckling and ultimate strength of spher-

ical pressure hull under external pressure.

Mar Struct. 23(3):227-40. http://dx.doi.org/

10.1016/j.marstruc.2010.07.005.

Pan, B.B., & Cui, W.C. 2011. A comparison

of different rules for the spherical pressure hull

of deep manned submersibles. J Ship Mech.

15(3):276-85.

Pan, B.B., & Cui, W.C. 2011. On an appro-

priate design and test standard for spherical

pressure hull in a deep manned submersible.

In: Proceeding of 2011 IEEE Symposium on

Underwater Technology (UT 2011) and 2011

Workshop on Scientific Use of Submarine

Cables and Related Technologies (SSC 2011),

Tokyo, Japan, 5-8 April 2011, IEEE Catalog

Number: CFP11UTS-PRT, ISBN: 978-1-

4577-0165-8, pp. 436-442. Tokyo: Institute

of Industrial Science, University of Tokyo

Pan, B.B., & Cui, W.C. 2012. Structural

optimization for a spherical pressure hull of

a deep manned submersible based on an

appropriate design standard. IEEE J Oceanic

Eng. 37(3):564-71. Color versions of one or

more of the figures in this paper are available

online at http://ieeexplore.ieee.org. Digital

Object Identifier. 10.1109/JOE.2012.

2190809.

Pan, B.B., Cui, W.C., Shen, Y.S., & Liu, T.

2010. Further study on the ultimate strength

analysis of spherical pressure hulls. Mar

Struct. 23(4):444-61. http://dx.doi.org/

10.1016/j.marstruc.2010.11.001.

Rona, P.A. 2000. Deep diving manned

research submersibles. Mar Technol Soc J.

33(4):13-25. http://dx.doi.org/10.4031/

MTSJ.33.4.3.


Russian Maritime Register of Shipping.

2004. Rules for the Classification and

Construction of Manned Submersibles, Ship’s

Diving Systems and Passenger Submersibles [S].

Moscow: RS Press.

Sagalevitch, A. 2009. From the bathyscaph

Trieste to the submersibles Mir. Mar Technol

Soc J. 43(5):79-86. http://dx.doi.org/10.4031/

MTSJ.43.5.3.

Taylor, L., & Lawson, T. 2009. Project

Deepsearch: An innovative solution for

accessing the oceans. Mar Technol Soc J.

43(5):169-77. http://dx.doi.org/10.4031/

MTSJ.43.5.28.

Takagawa, T., Momma, H., & Hotta, H.

1995. Advanced technology used in Shinkai

6500 and full ocean depth ROV Kaiko.


l

P A P E R

Reliability-Centered Development of DeepWater ROV ROSUB 6000
A U T H O R SVedachalam NarayanaswamyRamesh RajuMuthukumaran DurairajAarthi AnanthapadmanabhanSubramanian AnnamalaiGidugu Ananada RamadassMalayath Aravindakshan AtmanandNational Institute of OceanTechnology, Chennai, India
A B S T R A C T
This paper presents the reliability-centered development of a deep water
Remotely Operable Vehicle (ROV) ROSUB 6000 by the National Institute of OceanTechnology (NIOT), India. ROV operations are required during deep water interven-tions, such as well head operations, emergency response situations, bathymetricsurveys, gas hydrate surveys, poly-metallic nodule exploration, and salvage oper-ations. As per our requirements, the system needs to be capable of deep wateroperation for a period of 300 h/year and to be extremely reliable. Methodologiesapplied during the development and enhancement phases of electrical and controlsystems, taking into consideration the cost, space, and time constraints to attain thebest possible reliability are detailed. Reliability, availability, and maintainability(RAM) studies are carried out to identify possible failure cases. It is found thatROV-Tether Management System (TMS) docking failure could be detrimental tothe ROV system, and manipulator system failure could be detrimental to subseaoperations. It has been calculated and found that the improved design has amean time between failure (MTBF) of 4.9 and 6.2 years for ROV-TMS dockingand manipulator system operations, respectively. The importance of monitoringtether cable healthiness during normal and winding operations and the systems im-plemented for effectively monitoring and maintaining the tether cable operationaland functional healthiness, using the tether cable pay-out, vehicle heading, electricinsulation, and optical performance sensors with the aid of a sea battery, are dis-cussed. Maintenance decision support tables, which detail the operational person-nel for the upkeep of the systems during the indicated interval so that the highestpossible reliability is maintained during the mission period, are also presented.Keywords: remotely operated vehicle, reliability, probability of failure, mean timebetween failure, FMECA analysis

tion, control and electrical system,Hydro Acoustic Navigation System

Introduction

The remotely operable submersibleROSUB 6000 for deep sea operation isan electric work class Remotely Oper-able Vehicle (ROV), equipped withtwo manipulators and having an addi-tional pay load capability of 150 kgfor mounting scientific and mission-oriented subsystems. The ROSUB sys-tem comprises the main components,namely the ROV, Tether Manage-ment System (TMS), Launching andRecovery System (LARS), ship sys-tems, control console, instrumenta-

(HANS), and the control and opera-tional system (Manecius et al., 2010;Ramesh et al., 2010). Figure 1 showsthe overall view of the ROV withLARS in the ship.

Figure 2 shows the system launch-ing phase, where the work class ROVand the TMS are docked together andlaunched from the mother vessel usingthe LARS. The storage winch housesthe 7,000-m-long umbilical cable,and its operation is synchronizedwith the LARS and other deck gear.

The umbilical is an electro-optic in-terface between the TMS and the shipand is used to carry the weight of theTMS and the ROV.

The LARS takes the ROV-TMSdocked system below the splash zoneand undocks it. The ROV-TMS sys-tem is lowered to the location of inter-est. As the docked system gets closerto the target of interest, the ROV isreleased from the TMS. The ROV isequipped with thrusters and can beoperated on the pilot’s command

May/J

from the launching vessel. Manipu-lators mounted on the ROV are usedto carry out the required subsea opera-tion. After completion of the task, theROV is docked back to the TMS in thesubsea location, and the system is takenback to the ship.

Figure 3 indicates the electrical andcontrol system architecture in the TMS,ROV, and the ship. The ship powerat 415 V at 50 Hz is converted into6,600 V at 460 Hz, using a combina-tion of a standard frequency converter


and a step-up transformer. The electro-optical connectivity between the shipand the TMS is achieved by a 7,000-mumbilical cable (Manecius et al., 2010;


Ramesh et al., 2010), while the con-nectivity between the TMS and theROV is obtained by a 400-m-longtether cable. Subsea power converters

l

in the TMS and the ROV convertpower at 6,600 V at 460 Hz to thepower level requirements of the sub-systems. The voltage of 6,600 V isselected as a tradeoff between the sizeof the umbilical/tether cables, whilethe voltage drop is managed in the6,000 m transmission cable and thepower requirements of the system. Afrequency of 460 Hz is used to reducethe size and weight of the magneticcomponents in the ROV and TMS.

Electronic controllers with input-output modules are used for interfac-ing with the field devices in the TMSand the ROV. Data formats are man-aged by using multiplexers and mediaconverters. The ROV andTMS systemscommunicate with the ship systemusing the fiber optic Ethernet linkthrough the slip rings located in theTMS and deck storage winches. Themanipulators can be operated fromthe control console, and the controlis interfaced to the ROV and the shipside controllers at either end.

RAM Methodology andthe Standards Followed

As the ROSUB 6000 system is to beinvolved in mission critical applications,reliability, availability, andmaintainabil-ity (RAM) are of utmost importance. Aspart of the RAM studies, Failure ModeEffects Analysis (FMECA) studies weredone to identify the possible failures.

Failures were classified into fourcategories based on their impact on theoverall system, and the most criticalfailure surfacing in each category istaken up for improvements.a. Type 1 failures that have a minimal

impact on the mission objective andthe system, e.g., failure of one ofthe lamps or cameras in the ROV.

b. Type 2 failures that could defeat themission objective, but not damaging

FIGURE 1

View of the system architecture of ROSUB 6000.

FIGURE 2

View of the ROSUB 6000 system ready to be launched.

to the systems involved, e.g., failureof the subsea positioning system.

c. Type 3 failures that cause damage tothe system itself, e.g., ROV-TMSdocking failure.

d. Type 4 failures that could causedamage to the system as well asthe subsea installation, e.g., manip-ulator failure while in operation.The following Types 3 and 4 fail-

ures are described in this paper:

1. Failure of the ROV to dock withthe TMS underwater.

2. Failure of the manipulator whencarrying out a subsea task.

Reliability trees are drawn to find out theprobability of failure for the identified twocritical operations. Calculations were per-formed to find out the mean time be-tween failure (MTBF) for a 5-year period,with the system clocking deep wateroperation for a period of 300 h/year.

May/J

The following are the list of themajor standards followed:1. FIDES Guide for Reliability Meth-

odology (FIDES Guide, 2009) forelectronic systems, which deals withthe mission and environment spe-cific analysis.

2. MIL HDBK 217F Military Hand-book for Reliability Estimation(U.S. Department of Defense,1991) of the Electronics Equipment.

3. OREDA Handbook (OREDAAUTOR, 2003) for OffshoreReliability Data.

4. IEEE 493 IEEE RecommendedPractices (IEEE Standard, 1997)for theDesign of Reliable Industrialand Commercial Power Systems.

5. Reliability of Valve Regulated LeadAcidBatteries (DeAnda et al., 2004),a study by the U.S. Department ofEnergy, Energy Storage Systemsprogram.Failure rate determination is done

by the following methods:a. Based on the manufacturer’s data

and interpretation suitable for themission profile.

b. For systems and circuit boards,where schematics are available,using component failure data fromthe respective standards, and calcu-lating the failure rates taking themission profile, operating condi-tions and stresses into consideration.

c. For commercially sourced circuitboards, where there is no adequateinformation, based on the functionalspecification, the failure rate for themission profile is calculated usingstandards.The FIDES approach is based on

physics and the failures supported bythe analysis of test data, feedbackfrom operations and existing models,along with the statistical interpreta-tions over the normal operating lifeperiod of the involved systems.

FIGURE 3

Electrical and control architecture of the ROSUB 6000 system.


The standard considers the in-fluence of the operating temperature,amplitude and frequency of the tem-perature changes, vibration amplitude,humidity and operating stress factors.For example, the impact on the life-time due to the operating temperature,thermal cycling and vibration stressesare based on the Arrhenius, Norrisand Basquin laws (FIDES, 2009).The standards make provision to ac-count formanufacturing and integratingquality factors into the calculations,taking into account the environmentalconditions and stresses during differ-ent mission profiles. The standardalso provides the commercial off-the-shelf approach for calculating the fail-ure rates of commercially availablecircuit boards for the defined missionprofile with the functional require-ments and mission profile as an inputfrom the user.

The failure rate of the componentsis not a linear function with time.Failure rates are usually mentionedin terms of the number of failures perbillion hours (FIDES, 2009), abbre-viated as failure-in-time (FIT). Asan example, Table 1 populates thecalculated failure rate of a thruster

T

F(


motor power electronic controller.It can be seen that the device has a fail-ure rate of 77 and 446 FIT for a mis-sion profile of 300 and 8,760 h/year,respectively.

The failure rates of the componentsare calculated at the circuit componentlevel and incorporated in the reliabil-ity trees. Table 2 gives the standardsfollowed for the major systems andcomponents.

The TOTAL-SATODEV ’s GRIFsoftware is used for realizing the treesfor calculating the probability of failurefor a specific time period, with provi-sions to incorporate themean time to re-pair (MTTR), and the options to selectthe reliability laws based on the avail-ability of field failure data. In thisstudy, exponential laws are applied, asthe field failure data are very limitedfor subsea and similar systems.

Reliability and AvailabilityType 3 Failure Scenario:ROV-TMS Docking Failure;Normal ROV-TMS Recovery

When the ROV completes theidentified mission, it has to be dockedwith the TMS, and the docked systemhas to be brought to the subsurfaceclose to the ship, so that the ROV-TMS system can be docked with theLARS in the splash-free subsurface

l

region and then recovered to the deck.Figure 4 shows the pilot view of theROV-TMS docking, captured by theilluminated camera located in the TMSduring ROV deep water operations in2008, where the navigation parametersare displayed on the pilot screen in theship’s control console.

The camera and lights located inthe TMS will aid the pilot in carryingout the ROV-TMS docking. Givenbelow is the process sequence in carry-ing out a successful subsea dockingprocess.a. Pilot the ROV close to the TMS

location.b. Wind the already released tether

cable back in the TMS by operat-ing the winch in the TMS.

c. Pilot the ROV vertically down theTMS and operate the top thrust-ers to produce a downward thrust(normally ROV is upward buoyantby 20 kg) in such a way that thetether cable is held under minimumtension.

d. With the minimum tension in thecable, operate the TMS winch andwind the tether cable in a way thatthe cable is without any slack.

e. Continue the winding at a slowerpace until the ROV enters the TMScone and gets onto the latches (whichwill be indicated by the limit switchesinside the latches).

TABLE 2

Major standards followed for study.

Component
Standards
CPU, AC-DC converters, DC-DC converters, fuses, electronicsand optical connectors, Ethernet converters, data and videomultiplexers, input and output modules for data acquisition cards

FIDES

HF converters and transformers, isolators, motors
MIL and IEEE
Power contactors, halogen lamps
MIL
Umbilical and tether cables, terminations, subsea sensors
OREDA
ABLE 1

ailure-in-time date for brushless direct currentBLDC) motor power electronics controller.

Hours of Operationper Year

Failure-In-Time(FIT)

1
64
100
68
300
77
500
85
3,000
195
5,000
282
8,000
413
8,760
446

Docking FailureThe following failures could lead

to docking failure:a. TMS docking vision support,b. TMS tether winch operation,c. ROV thrusters for the docking

operation, andd. Tether cable twist indication from

the TMS.The failure in any of the above results

in the ROV-TMS docking failure.Figure 5 shows such a condition.

In such a scenario, the ROV hasto be salvaged after bringing it to thewater surface. This is done by windingthe deck umbilical cable, until theTMS surfaces and is docked with theLARS. As the ROV is linked with

the tether cable and as it is positivelybuoyant, it tends to surface. This sur-facing is detrimental to the ROV andthe released tether cable, as there areincreased chances of entanglementwith the ship’s systems such as thrust-ers. To minimize the chances, theprobability of the ROV-TMS dock-ing failure has to be minimized to thelowest possible level.

The base design case indicates thepreliminary system design. Duringthe operational phases, based on theexperience acquired, the systems areimproved. The improvements areincorporated in the base case, andtheir benefits in terms of the MTBFare detailed.

May/J

a. Base design caseFailure trees are drawn to calcu-

late the probability of failure for a5-year period, with the system in oper-ation in deep waters for a period of300 h/year. From the tree shown inFigure 6, it can be seen that the prob-ability of failure is 92.48% and theMTBF works out to 1.9 years. Theforthcoming sections detail the tech-nical improvements made to attain aMTBF of 5 years.b.Maintaining spares for ship-basedelectric systems

The failure trees are analyzed forthe contribution made by the deck-based systems. It is calculated that fail-ure probability of 6 kV power supplyinput to the TMS is 64.65%. Thispower supply input is used by all thesystems in the TMS and ROV. It isfound that the following componentswhich have higher failure rates arereparable within 10 h by having a hotstandby on board the deploymentvessel:a. High-frequency (HF) converterb. Umbilical storage winch slip ringc. Medium-voltage (MV) transformerd. Isolatorse. Ship-side data multiplexer

FIGURE 4

Illuminated camera view of the ROV docking to the TMS.

FIGURE 5

View indicating system recovery after a docking failure.


f. Real-time controllerg. Photonic inertial navigation systemh. BLDC motor controllerBy means of having spares in theship, the probability of failure of thepower supply input to the TMS canbe reduced from 64.65% to 1.38%.Hence, the MTBF can be increasedto 3.3 years.c. By introducing black-dock sensors

As shown in Figure 4, the ROV-TMS docking operation is supportedby a camera, lights, associated con-trols and power systems located inthe TMS. It is calculated that theTMS docking vision support has aprobability of failure of 20.51%.When this support system fails, thepilot cannot dock back the ROV tothe TMS, even if the other con-ditions required for proper dockingare maintained. To overcome this,black-dock sensors are introduced.The sensors are used to measure thefollowing:a. Tether cable tension.


b. Cable winding counter to ensurethat the already released cable iswound back.The black-dock sensors enable the

pilot to carry out the docking opera-tion in the event of vision system fail-ure. From the failure trees in Figure 7,it can be seen that, by using the black-dock system, the probability of failureis reduced to 16.71% from 20.51%.

By means of this, the MTBF of theROV-TMS docking process can beincreased to 3.6 years.d. By operating the ROV isolationswitchgear in the TMS as a switch

From standards (U.S. Departmentof Defense, 1991), the FIT of theelectric power contactor opening underload is 10 times higher than when it isoperated without load. The ROV op-erates with a power of nearly 60 kW atnear unity power factor for propulsionand control systems. The remotely op-erable isolating device in the TMS hasto handle a maximum current of 10 Aat 6,600 V. Using the conventional

l

industrial standard vacuum circuitbreaker (VCB) is not an attractive solu-tion as a typical VCB of minimum rat-ing of 400 A at 7.2 kV weighs around25 kg and has a dimension of ap-proximately 0.4 m (L) × 0.21 m (W) ×0.32 m (H). In addition to their hugefootprints, such circuit breakers shouldbe operated inside pressure-rated en-closures with feed through. This in-creases the size, weight, and cost ofthe assembly, which are not preferredsolutions. As a solution to this chal-lenge, an industrial standard 660 V,100 A motor air break low voltagecontactor is selected and used in anoil-filled, pressure-compensated envi-ronment. As the dielectric capacity ofoil is 10 times that of air, this encour-ages the use of the 660 V air contactorat 6,600 V in a pressure compensatedmode. The dimension of the selectedcontactor used to carry out the me-dium voltage (MV) switching opera-tion is approximately 0.2 m (L) ×0.11 m (W) × 0.1 cm (H).

FIGURE 6

Tree for calculating the docking failure MTBF for the base design case.

The reliability of the MV switch-gear is of utmost importance in theoverall system, as the failure of thiscontactor leads to the paralysis of thewhole system. When the switchgearopens, an electrical arc is produced,and this is extinguished by the arc-chute mechanism and insulating oil.

As the arc generates high tempera-ture, it decomposes a portion of the oilinto gases (Thomas, 2008) composedof 70% hydrogen, 20% acetylene andcarbon particles. Carbon particles areconductive in nature. The quantity ofdecomposed carbon is a function ofthe arc intensity, duration of the arcand the temperature of the arc plasma.As the switchgear is in a closed oilsystem, prolonged usage or excessivegeneration of carbon particles leadsto oil contamination and subsequentfailures. Therefore, the longevity andhence the reliability of the assemblycan be improved by ensuring that theswitchgear opens under a no-load con-

dition, resembling the required con-dition of a MV switch.

FMECA studies revealed that pos-sible situations leading to the openingof the MV switch TMS located underload are as follows:1. Input power failure in the TMS

alone.2. Control system failure in the TMS

alone.3. Accidental open command to the

MV switch.To overcome the above situations,

a possible solution would be to delaythe opening of the switch by a few sec-onds, during which time the electricalload on theMV switch will be reduced.This requires reliable localized energystorage in the TMS system. Eventhough batteries could be attractive interms of size, they are not so attractivein terms of reliability and safety. Supercapacitors are found to be the besttrade-off (Lemofouet & Rufer, 2006;Mallika & Saravanakumar, 2011;

May/J

Paulo et al., 2001; Shukla et al., 2001).They are electrochemical, double layercapacitors (Barrade et al., 2003; Rufer,Hotellier, & Barrade, 2004; Weddellet al., 2011; DeAnda et al., 2004),which are preferable because theyoffer high power and energy densities.

The MV switch operates on 24VDC supply and requires a maximumcoil current of 0.2 A. The design criteria(Al-Ramadhan&Abido, 2012) involveensuring a minimum voltage of 12 Vfor continuous holding of the coil forduration of 60 s. Based on these param-eters, we have calculated the require-ments of energy (W) and minimumcapacitance (C). Twelve capacitors areconnected in a series to get the effectivecapacitance of 0.833 F that can hold theMV switch up to 60 s. The designed cir-cuit incorporating the required protec-tion circuits is integrated into the TMScontrol system. When a 24-V controlcommand from the TMS is given, theMV switch coil gets energized. The

FIGURE 7

MTBF improvement by incorporating black-dock sensors.


super capacitor is charged through a50 Ω, 10-W resistor. In the event ofthe 24 V control command failure,the super capacitors will continue tohold the MV switch coil by supplyingthe stored energy through the diode.The super capacitors shall continueto energize the contactor coil and, inturn, hold the power contacts of theswitch in a closed condition for a periodof 60 s. The control system utilizes thisperiod to reduce the electrical load onthe MV switch, so that it operates atno-load. The required logic is imple-mented in the logic controllers in theROV, the TMS and the ship. By meansof this, the probability of power fail-ure to the ROV is reduced from 26.4%to 12.8%.

By means of this, the MTBF ofthe ROV-TMS docking process canbe increased to 3.8 years.e. By having redundancy in the Ship-ROV optic link and a serial linkbetween the ROV and the TMS

The ROV and TMS communi-cate with the ship system through theoptical-based Ethernet network. Thetopology in the base and improvedcases is shown in Figure 8.

In the base case, the ROV-Shipnetwork spans through the opticalcore in the tether cable and in the um-bilical cable, with one pass in the TMSwinch slip ring, and one in the deckstorage winch slip ring and to theship console. The TMS-Ship networkspans through one pass in the TMSwinch slip ring and in the deck storagewinch slip ring and one core in theumbilical cable and then to the shipconsole.

The improved design incorporatesthe following:a. Redundancy is introduced in the

ROV-Console communicationlink, by having an additional opti-cal pass in the TMS tether and deck


storage winch slip rings, and addi-tional optical cores in the tetherand umbilical cables.

b. A new RS485 link between theROV and the TMS processorsusing the twisted pair electric cablein the tether cable is established.From Figure 9, it can be seen that

the ROV-Ship Console communicationfailure probability is 2.59%. By meansof the improved architecture, the prob-ability of failure is reduced from 2.59%to 0.434%.

With this improvement, theMTBFof the ROV-TMS docking process isincreased to 4.6 years.f.With voltage control devices on thetop side and back-emf capacitors inTMS and ROV 300 V circuits

Thruster motors are operated usingBLDC motor electronic controllerslocated in the ROV. Tether winchhydraulic motors are also operated bythe BLDC motor controllers locatedin the TMS. The controllers are ener-gized by 300-V power supplies in theROV and the TMS. The BLDCmotor power electronic systems aresensitive to excess voltage. To improvethe reliability of the electronic control-lers and subsequently the system reli-ability, voltage management systemsare required.

Voltage management from the shipside is implemented as follows:

In the ROSUB 6000 system, theship power at 415 V at 50 Hz is con-verted to 6,600 V at 460Hz and trans-mitted through the 7,000-m-longumbilical cable and further to theROV through the 400-m-long tethercables. The voltages are stepped downand rectified so as to obtain 300 Vand 24 V DC, suitable for operatingthe propulsion thrusters and controlsystems. The ROV has an electricalload of 70 kW and the TMS has atotal load of 20 kW.

l

When the ROV and TMS operateat full electric load, the system is de-signed in such a way that the shipsidevoltage is 6,600 V and the ROV andTMS end voltage is 6,000 V, with anumbilical cable voltage drop of 600 V.At no load condition, the TMS and theROV systems experience a voltage of6,600 V. This results in 10% increasein the 300 V circuits, which is not ahealthy condition for the thrustermotor power electric speed controllersystem.

This overvoltage condition is man-aged by implementing the automaticvoltage compensation system on theship side, using a high-frequency (hF)power converter, which works on theprinciple of v/f control philosophy.The control methodology is imple-mented using a proportional-integral-derivative (PID) controller, and thesame is shown in Figure 10. At noload condition, the system draws onlythe reactive current required to chargethe cable capacitance. When the ROVsystem gets loaded, there is an increasein the active component of the loadcurrent. The PID controller is tunedto provide the control input to the gatedriver of the HF converter in such away that 300 V is maintained constantat the ROV and TMS ends.

Voltage management on the subseaside is implemented as follows:

Thrusters and pump motors in theROV and TMS are operated withbrushless DC motors, which are con-trolled by power electronics. The con-trollers help in operating the systems atthe required speeds. A BLDC motorin operation produces back-emf volt-age (Brushless Direct Current Motormanufacturer recommendations forthruster Power System reference), atthe pulse width modulation (PWM)frequency. This back-emf has the ten-dency to raise the overall 300 V DC

FIGURE 8

Schematic of the network.


bus voltages to more than 600 V. Thisis detrimental to all the power elec-tronics controllers connected to the300 V DC network. To counter thisback-emf, the DC capacitor banks of


80 and 20 mF are connected acrossthe DC buses permanently in theROV and TMS. In the base case,each BLDC motor power electronicscontroller has a probability of failure

l

of 19.52%. By ensuring proper operat-ing voltage with the aid of the de-scribed voltage management devicesthe probability of failure for eachBLDC motor controller is reducedto 16%. With this improvement, theMTBF of the ROV-TMS dockingprocess is increased to 4.8 years.g. With redundant TMS pump

In the base case, the TMS systemis equipped with one electric motordriven hydraulic pump, which is usedfor operating the tether cable winchdrum. With a redundant pump, thewinch operation due to the electricaland corresponding control systemfailure is reduced from 9.96% to 1%.The same can be seen from the failuretrees in Figure 11.

With this improvement, theMTBFof the ROV-TMS docking process isincreased to 4.9 years. The same canbe seen from Figure 12. The stage bystage improvement in the probabilityof failure and the correspondingMTBF can be found from Table 3.

Type 4 Failure Scenario:Manipulator Operation FailureNormal Working Scenario

The ROV in the ROSUB 6000 isequipped with two manipulator armswith five and seven functions for carry-ing out subsea operations. The manip-ulators are operated by the personnelfrom the control console using joy-sticks. In addition to other vision sys-tems, manipulator arms are equippedwith a camera, which will give a real-time video feedback to the operatingpersonnel. Figure 13 shows the workclass ROVs withmanipulators involvedin different kinds of tasks.

The subsea operations depend onthe mission objective and may involvetasks such as electric wet mate con-nector mate/de-mate, pipe line valve

FIGURE 9

Tree indicating improvements in TMS-Console.

FIGURE 10

(a) Automatic Voltage Compensation implementation methodology. (b) Location of capacitors inthe subsea side.

May/J

operations and pipe line sacrificialanode fixing.

Failure ScenarioThe risk involved in these subsea

operations varies widely (API RP14B,2012; NORSOK standard, 2001).Figure 12 shows the ROV holdingone end of the wet mate cable connec-tor for mating with the fixed ROVpanel. The other end of the cablecould be a part of the permanent sub-sea installation. When the manipula-tor operation fails in this condition,the forced retrieval of the ROV bywinding back the tether or main um-bilical winch could damage the subseainstallation. Such damages could turndisastrous for subsea oil and gas well-heads and manifolds. Thus, the ROVwill be anchored to the subsea installa-tion, with other systems connected tothe ROV deployment vessel. To miti-gate this situation, another intervention

FIGURE 11

Improvement in the failure rate with redundant pump improved case.

FIGURE 12

Tree indicating improvement in the overall MTBF after introducing the redundant pump.


vehicle has to be dispatched for crisismanagement. A similar conditioncould occur if the ROV operates anoil pipe line isolation valve, whichleads to valve damages and consequentoil leakages, if double barrier isolationis not provided. The same could berisks involved during salvage opera-tions. Therefore, the reliability of themanipulator system has to be high.

Base CaseThe following could lead to manip-

ulator operations failure:a. Manipulator functional failure.b. Hydraulic pump motor electric

system failure.c. Manipulator operation vision sup-

port failure.In the base case, the ROV manipula-

tor control failure is calculated to be


90.27% with a corresponding MTBFof 2.1 years. The network and hard-ware used for the ROV-TMS dockingfunction is shared by the manipulatorsystem. Therefore, the system improve-ments discussed for Type 3 ROV-TMSdocking failure apply equally for reduc-ing the manipulator operation failure.

It can be seen from Figure 14 thatthe manipulator operation failure prob-ability is reduced to 55% with a cor-responding MTBF of 6.2 years. Theimprovement with each change islisted in Table 4.

MaintainabilityImportance of Cable TwistMonitoring Mechanism

When the ROV is undocked fromthe TMS, the pilot will maneuver the

l

ROV to the location of interest. Dur-ing the process, based on the ROVheading with respect to the TMS, theKevlar-armored, torsionally balancedtether cable will be subjected to twist-ing, and the number of twists dependon the skill of the pilot, nature of theoperation and the condition of the seawater currents. Further, as the TMS isconnected to the deployment vessel,which is maintained in a dynamicpositioning mode, it may also createheading changes for the TMS withrespect to the ROV. Thus, the num-ber of twists depends on the relative360 deg. heading counts. When theROV has completed the mission, itwill be docked back to the TMS afterwinding back all the reeled out cable.Tether cables are designed to with-stand specific twists per meter length(Buckham et al., 2003; Sakamot et al.,2008). When the number of twists/meter exceeds the design limit, thecable sheath and, hence, the cablewill be damaged. Thus, to avoid cabledamage, the pilot has to ensure thatthere is no residual twist on the cablebefore docking.

To aid in a hassle free dock, thefollowing parameters are monitoredby the control system:a. ROV heading counts provided by

the heading sensor in the ROV.b. TMS heading counts provided by

the heading sensor mounted inthe TMS.

c. Length of the cable reeled out ofthe drum, which is provided bythe tether cable layer counter.

d. Cable tension which is provided bythe load cell sensor mounted in thewinch.As the heading sensors and electronic

counting are done in the ROV andTMS, a power interruption to theROV and TMS will result in the lossof twist count information. Even

TABLE 3

MTBF Improvements with modifications for TMS-ROV docking failure case.

Improvement
Probability of Failure MTBF
Base design case
92.48% 1.9 years
By means maintaining spares for ship basedelectric systems

78.9%
3.3 years
By operating ROV isolation circuit breaker as aswitch

73.13%
3.8 years
By having redundancy in Ship-ROV optic link and aserial link between ROV and TMS

66.64%
4.6 years
With voltage control devices in top side andback-emf capacitors in ROV 300 V circuits

64.97%
4.8 years
With redundant TMS pump
64.05% 4.9 years
FIGURE 13

Typical ROV manipulator in operation (Courtesy: WHOI and MBARI websites).

though the last logged value could bemade available in the shipside, oncethe power resumes, the number ofheading changes undergone by theROV during the interruption periodwill not be known. Therefore, theresidual twists on the cable will notbe available to the pilot during thefinal docking stage. To overcome thisproblem and to avoid power interrup-tion to the heading sensor and moni-toring electronics, a sea battery isused as a backup power in the ROV.

Figure 15 shows the heading countsmonitoring mechanism and the seabattery in the ROV.

By means of introducing the seabattery, the probability of powerfailure to the ROV control systems isreduced from 15% to 5%. Thus, thetwist feedback information failureprobability to the pilot is reducedfrom 60% to 55%.

Table 5 shows the MTBF of theROV-TMS docking process andmanipulator operation with different

May/J

failure rates of the tether cable. Thisis given to indicate the impact of thetether cable failure on the overall sys-tem MTBF.

As it is understood that the tethercable is an important componenthaving a significant role in the MTBF,systems are incorporated that canmeasure the cable electric insulationand optical losses in the system online.As the cable performance depends onthe operating ambient condition, theinclusion of this mechanism will beuseful for the system safety as well asfor preventive maintenance action.

Preventive MaintenanceDecision Support

Table 6 shows the probability offailure in percentage for the ROV-TMS docking process for differentperiods. The table also details the roleof the subcomponents.

The cells with data in bold fontsindicate the failure rates of the com-ponents whose failure probabilitiesare higher.

FIGURE 14

ROV manipulator control failure for the base and improved cases.

TABLE 4

MTBF improvements with modifications for manipulator function failure case.

Improvement
Probability of Failure MTBF
Base design case
90.27% 2.1 years
By means of maintaining spares for ship basedelectric systems

71%
4.1 years
By operating ROV isolation circuit breaker as aswitch

64.7%
4.8 years
By having redundancy in Ship-ROV optic link anda serial link between ROV and TMS

64%
5.0 years
With voltage control devices in top side andback-emf capacitors in ROV 300 V circuits

55.11%
6.2 years


Preventive MaintenanceSchedule Summary

Umbilical and Tether CablesThe longevity of the tether cable

depends on the handling and electricalstresses. Handling stresses are due totwisting and can be optimized by skilledpiloting. The failure rates of the umbil-ical and tether cables were obtainedfrom OREDA and are used for thecalculations. Andrew Bowen et al ofWoods Hole Oceanographic Institu-tion (WHOI) and Ed Mellinger et al.of Monterey Bay Aquarium ResearchInstitute (MBARI) performed the insu-lation life testing of field-aged samplesof cables used inMBARITiburonROVand WHOI Jason ROVs, which havean accumulated time of 6,700 h in230 dives. The tests identified factorsaffecting the insulation stress at the op-erating voltage and high hydrostatic pres-sures. It was calculated that the cablecould survive for 22.8 years under dryaging conditions but would survive onlyfor 7.6 years when operated at 400 Hz

FIGURE 15

ROV and TMS having independent heading sensing systems and the location of the sea battery inthe ROV.

TABLE 5

Impact of tether cable failure in failure cases under study.

Times TetherFailure FIT

ROV-TMS Docking
Manipulator Operation
Probability ofFailure (%)

MTBF(years)

Probability ofFailure (%)

MTBF(years)

×1
64 4.9 55 6.3
×2
65 4.8 56 6.1
×5
67 4.5 58.5 5.7
×10
70 4.1 62.4 5.1
×20
75 3.6 67 4.5
TABLE 6

Maintenance decision support table.

System/Component

Probability of Failure in %

1 year
3 years 5 years 10 years 15 years
ROV-TMS docking
18.1 45.4 64.0 87.6 93.5
Subcomponents

Cable termination
0.064 0.18 0.30 0.60 0.90
Tether cable
0.38 1.14 1.9 3.76 5.59
Umbilical cable
0.21 0.63 1.06 2.11 3.15
HF converter
16.1 41.1 58.6 82.9 92.9
LV isolator
0.32 0.98 1.6 3.24 4.82
HV transformer
0.58 1.7 2.9 5.72 8.46
BLDC motors
0.39 1.18 2.01 3.9 5.8
Electrical slip ring
1.54 4.56 7.48 14.4 20.8
Power contactor
0.4 1.3 2.2 4.28 6.35
Plasma display
0.44 1.2 2.31 3.18 4.73
continued

in a continuous deepwater environment.Based on this experience, we have esti-mated the tether cable replacement pe-riod for the ROSUB6000 as 10 years.

Sea BatteryA sea battery has a failure rate of

8.8% in the first year of operation.

The failure rates are used from a stan-dard obtained from the studies madefor the U.S. Department of Energy-Energy Storage Studies program(DeAnda et al., 2004), which consid-ers battery failure rates, when used inan UPS and similar applications. Leadacid battery life depends highly on the

May/J

number of charge-discharge cycles andthe depth of discharge. In our appli-cation, the battery is subjected to veryminimal stresses in terms of the depthof discharge and duty cycle. Based onthis experience, we have estimated thesea battery replacement period for theROSUB 6000 as 3 years.

TABLE 6

Continued

System/Component

Probability of Failure in %

1 year
3 years 5 years
une 2013 V

10 years

olume 47 Numb

15 years

Computers
0.45 1.35 2.2 4.45 6.6
PLC processor with memory
0.3 0.71 1.05 1.8 2.5
NI CPU IO modules
0.021 0.06 0.1 0.2 0.31
Thrusters BLDC motor power electronic controller
0.37 0.82 1.03 2.03 2.31
Fuses
0.008 0.026 0.043 0.087 0.13
Halogen lamp
0.017 0.052 0.087 2.13 2.59
Multiplexer power board
0.025 0.076 0.12 0.025 0.38
Multiplexer data board
0.079 0.23 0.39 0.79 1.18
Multiplexer video board
0.007 0.02 0.03 0.07 0.1
Real-time controller
1.11 3.31 5.46 10.64 15.53
Sea battery
8.8 24 37 60 75
Proximity sensor
0.052 0.15 0.26 0.52 .78
Fiber optic rotary joint
1.92 5.66 9.26 17.6 25.3
Photonic inertial navigation sensor
25.3 58.4 76.7 96.4 98.8
TABLE 7

Summary of failures with probabilities of failure.

Failure Type
Failure Description
Base Case
Improved Case
Probability of Failure
MTBF Probability of Failure
e

MTBF

1
Camera not available for operationin ROV
83.51%
2.8 years 7.25% >10 years
1
Halogen Lamp not available foroperation in ROV
83.40%
2.8 years 7.12% >10 years
2
Visual survey failure 98.86% 1.1 years 53.34% 6.6 years
2
SONAR bathymetric survey failure 97.87% 1.2 years 78.05% 3.3 years
2
Gas hydrate survey failure 98.89% 1.1 years 53.73% 6.5 years
3
ROV-TMS subsea docking failure 92.48% 1.9 years 64.05% 4.9 years
4
ROV manipulator operation failure 90.27% 2.1 years 55.11% 6.2 years
r 3 69

Summary and ConclusionsThis paper describes the method-

ology of RAM studies carried on theROSUB 6000 deep water work classROV system. Table 7 summarizes alltypes of failures studied, the probabilityof failures and the MTBF in the basecase and improved cases.

The technical challenges in achiev-ing the best possible MTBF for thecritical failures are explained in detail.The maintenance decision support tabledetails the operational personnel for theupkeep of the systems during the indi-cated maintenance interval, so that thehighest possible reliability ismaintainedduring the mission period. The impor-tance of monitoring the tether cablehealthiness during normal and wind-ing operations, and the systems imple-mented for effectively monitoring andmaintaining the tether cable’s opera-tional and functional healthiness, usingthe tether cable pay-out, vehicle head-ing, electric insulation and optical per-formance sensors with the aid of a seabattery, is also discussed.

AcknowledgmentsThe authors gratefully acknowledge

the support extended by theMinistry ofEarth Science, Government of India infunding this research. The authors alsowish to thank the other staff involvedin the project for their contribution andsupport.

Corresponding Author:Narayanaswamy VedachalamNational Institute ofOceanTechnologyChennai, IndiaEmail: [email protected]

ReferencesAl-Ramadhan, M., & Abido, M.A. 2012.

Design and simulation of super capacitor en-


ergy storage system. In: International Confer-

ence on renewable energies and power quality

(ICREPQ), Spain, 28–30March 2012, pp. 1-4.

Spain: European Association for the Develop-

ment of Renewable Energies, Environment

and Power Quality (EA4EPQ).

API RP14B, American Petroleum Institute.

2012. API Recommended Practices for Design,

Installation, Repair and Operation of Sub-

surface System. Pennsylvania: American

Petroleum Institute. 30 pp.

Barrade, P., & Ruffer, A. 2003. Current

capability and power density of supercapaci-

tors: Considerations on energy efficiency,

ref. ISBN 90-75815-07-7. Pages 3,4.

Buckham, B.J., Driscoll, R.F., Nahon, M., &

Radanovic, B. 2003. Three dimensional dy-

namic simulation of slack tether cable in an

ROV system. In: Proceedings of ISOPE 2003,

Hawaii, USA, Department of Mechanical

Engineering, Vol. 2, pp. 2-3. Victoria BC,

Canada: University of Florida.

DeAnda, M.F., Miller, J., Moseley, P., &

Butler, P. 2004. Reliability of Valve-Regulated

Lead-Acid Batteries for Stationary Applica-

tions. California: Sandia National Laboratories.

68 pp. http://dx.doi.org/10.2172/918779.

FIDES Guide. 2009. Reliability Methodology

for Electronic Systems. FIDES group. 465 pp.

IEEE. 1997. IEEE Standard 493-1997, Recom-

mended Practice for the Design of Reliable In-

dustrial and Commercial Power Systems, IEEE

Standards Boards. Available at: http://standards.

ieee.org/findstds/standard/493-2007.html.

Lemofouet, S., & Rufer, A. 2006. Hybrid

Energy Storage system based on compressed

air and supercapacitors with maximum effi-

cient point tracking. IEEE T Ind Electron.

53(4):1105-15.

Mallika, S., & Saravanakumar, S. 2011.

Review of ultracapacitor—Battery interfaces for

energy management system. Int J Eng Technol.

3(1):2-3.

Manecius, S., Ramesh, R., Subramanian, A.N.,

Sathianarayanan, D., Harikrishnan, G.,

Jayakumar, V.K., … Amiragov, A. 2010.

Technology tool for deep ocean exploration—

remotely operated vehicle. In: Twentieth

l

International Offshore (Ocean) and Polar

Engineering Conference, pp. 206-212. Beijing,

China: International Society of Offshore and

Polar Engineering (ISOPE).

Mellinger, E., Aabo, T., Bowen, A., Katz, C.,

& Petitt, R.A., Jr. 2010. High Voltage Testing

of an ROV electro-optic tether cable. MTS.

28(9):4-5.

NORSOK standard. 2001. Risk and Emer-

gency Preparedness Analysis. Norway:

Norwegian Technology Centre.

OREDA AUTOR. 2003. OREDA. SINTEF.

835 pp.

Paulo, F.R., Brian, K.J., Maiiesa, L.C.,

Ayşen Basa, A., & Yilu, L. 2001. Energy

Storage Systems for Advanced Power Appli-

cations. IEEE Proceedings. 89(12):3-4.

Ramesh, R., Jayakumar, V.K., Manecius, S.,

Doss Prakash, V., Ramadass, G.A., & Atmanand,

M.A. 2010. Distributed Real Time control

systems for deep water ROV (ROSUB 6000).

In: 7th International Conference on Com-

putational Intelligence, Robotics and Auton-

omous Systems (CIRAS) in 13th FIRA Robot

World congress, Proceedings, pp. 33-40.

New York: Springer.

Rufer, A., Hotellier, D., & Barrade, P. 2004.

A super capacitor based energy storage sub-

station for voltage compensation in Weak

Transportation networks. IEEE T Power Deliv.

19(2):629-36.

Sakamot, K., Fujimoto, Y., & Osawa, H.

2008. Development of fatigueless umbilical

cable for full ocean depth 12000 m.

In: Marine cable Development Section

Engineering Department, Furukawa Electric

Co. Ltd, Japan, Advanced Underwater

Vehicle R and D Group, JAMSTEC, Japan

in International wire and cable 57th Sym-

posium proceedings, pp. 2-4. Eatontown, NJ:

IWCS Inc.

Shukla, A.K., Sampath, S., & Vijayamohanan,

K. 2001. Electro-chemical capacitors: Energy

Storage and beyond batteries. Curr Sci.

79(12):37-38,41.

Tecnadyne Application Note AN601. 2006.

General Power System Wiring Practices

Applied to Tecnadyne DC Brushless Motors.

San Diego, CA: Tecnadyne, Inc.

Thomas, A.P. 2008. Oil Circuit Breaker

Diagnostics. In: 7th Annual Wiedmann

Technical conference, pp. 1-2, 5. New Orleans:

Weidmann Inc.

U.S. Department of Defense. 1991. Military

Handbook, Reliability Prediction of Elec-

tronic Equipment. Washington, DC: Author.

Weddell, A.S., Merrett, G.V., Kazmierski,

T.T., & Al-Hashimi, B.M. 2011. Accurate

supercapacitor modeling for Energy harvesting

wireless sensor nodes. IEEE T Circuits-II.

58(12):2. http://dx.doi.org/10.1109/TCSII.

2011.2172712.


P A P E R

Anchor Drop Tests for a SubmarinePower-Cable Protector
A U T H O R SHan-Sam YoonResearch Center for OceanIndustry and Development,Pukyong National University,Busan, South Korea
Won-Bae NaDepartment of Ocean Engineering,Pukyong National University,Busan, South Korea


A B S T R A C T

l

Submarine power cables are widely used for power transmission, such asbetween mainlands and offshore islands and from offshore wind farms to on-landsubstations. There are several ways to protect power cables from accidental loads.Protection includes concrete blankets, sand bags, bundles, tunnel-type protectors,and trenching. However, no design standard for power-cable protectors is currentlyavailable because of the varieties of cable protection solutions and man-made ornatural hazards to submarine power cables. Thus, this paper presents anchordrop tests for a newly designed, matrix-type submarine power-cable protectorassembled with reinforced concrete blocks, to make a safety assessment. Marineenvironments were surveyed at the target site and simulated in the test set-up. A2-ton stock anchor was selected as the colliding object, and a 25-ton crane wasprepared to drop the anchor. Preliminary tests were performed to investigate theeffect of soil composition and protector arrangements on the test results. Finally,four field anchor drop test scenarios were designed, carried out, and analyzed, and asafety assessment was made for the submarine power cable. From the tests, it wasfound that, in addition to falling distances, the soil composition and saturation weresignificant factors for the settlement depth and damaged areas. Considering thesettlement depth of soils, the damaged areas of the concrete blocks, and the dam-aged state of the pipes (safety zone), all of the test results showed that the mattressfailed to protect the power cable from the anchor collision. The deformation, dam-age, and breakage of the pipe, which simulated the safety zone of the power cable,gave clues as to the reasons for the failure.Keywords: anchor drop, field test, submarine power-cable protector, stock anchor

and their use has not spread rapidly insome developing and even industrialized

Introduction

Anchoring activities—collidingand dragging—have become the larg-est cause of submarine cable faultsin recent years (International CableProtection Committee, 2009). To pre-vent power cable failures from thiscause, the International Cable Protec-tion Committee (ICPC) suggests theuse of automatic identification systems(AISs) for vessels. Nevertheless, the useof AISs is not applicable to small ships,

countries (Coffen-Smout & Herbert,2000; International Cable ProtectionCommittee, 2009; Wagner, 1995).This is a serious issue, because cablefailures can lead to significant inconve-nience to residents, cutting communi-cations, power, and water supplies. Forexample, in 2010 the submarine powercable linking Shanghai and ShengsiIsland, Zhejiang, was fouled and brokencompletely as a result of ship anchor-ing. The island lost its power supplyfor 4 days, and then electricity was onlypartially restored to about 90,000 resi-dents using diesel generators (Jie &

Yao-Tian, 2012). Similarly, in 2006,one of two submarine power cablesconnecting the main Korean Peninsulato Jeju Island was damaged by ship an-choring activities. This event caused apower-supply blackout for the whole ofJeju Island, resulting in severe inconve-nience to nearly 570,000 residentsand 30,000 tourists. Fortunately, noloss of human life was reported dur-ing that period, likely because theoutage lasted only 2.5 h during thedaytime (Woo et al., 2009). How-ever, it was certainly undesirable be-cause tourism is important to the

island, and the power-supply black-out caused a simultaneous suspensionof water availability.

Such accidents are also applicableto the power cables transporting off-shore wind power to a national gridon land. Moreover, given the currentincreasing trends in offshore wind-energy projects, offering security againstmany energy-supply emergencies—whether natural or man-made—is asignificant issue. Thus, there has beenmuch work to assess the causes of dam-age and to improve cable protection.The possible causes—anchor collision

and dragging—have been investigatedand design criteria for submarinepower-cable protectors have been im-proved in South Korea (Woo et al.,2009). This extensive research has in-cluded computer simulations of an-chor collision and dragging on cableprotectors (Woo & Na, 2010). How-ever, numerical simulations alone arenot fully reliable because it is not easyto make definite safety or failure as-sessments based solely on numericalresults, such as stress contour and de-formation patterns.

Thus, in this study, field anchordrop tests of a newly designed subma-rine power-cable protector were inves-tigated. For this purpose, the followingworks were carried out: (1) materialpreparation and test set-up, (2) fieldanchor drop tests, and (3) safety as-sessment. For material preparation, amattress-type power-cable protectorwas selected, designed to allow adjust-ment of its horizontal and verticallengths by controlling the number ofreinforced concrete blocks used. Thisadjustability allows the blocks to be as-sembled in the field, using hooks andlinks, into different sizes of protectorsin a designated field environment afterinstalling the power cables on the sea-bed. In the field tests, first, the termi-nal velocity of a 2-ton stock anchor inwater was calculated and converted thevelocity into a corresponding fallingdistance in air. Second, the soil proper-ties were surveyed at the target site—Haenam, in the Southwest corner ofthe Korean Peninsula. Third, by rec-ognizing the fundamental behaviorsof the protector, an experimental set-upwas designed.

This study should be relevant tothe offshore wind power industry,where submarine cables take powerto an offshore transformer and the off-shore transformer converts the elec-

tricity to a high voltage (about 33 kV)and sends it 8–16 km to a nationalor international grid at a substationon land (Byrne & Houlsby, 2012;Esteban et al., 2011; Yamabuki &Kubori, 2012). Because wind speedstend to increase significantly with dis-tance from land, the offshore wind re-sources can generate more electricitythan wind resources at adjacent land-based sites. This means the length ofthe power cables will increase, and sowill the need for cable protection.The test process and results can pro-vide a valuable checkpoint for manag-ing the electrical power supply fromthe mainland to an island and froman offshore wind farm to an on-landstation.

Current PracticeIn South Korea, there is no a spe-

cific design guideline and standardfor power cable protectors. After KTSubmarine (a Korean company) wasestablished in 1994, submarine powercable installation and maintenancehave been practiced under the super-vision of the Korea Electric PowerCorporation (KEPC). However, thedesign practice has not yet been estab-lished because of the lack of data andstudies. Although all of the powercables in South Korea have been oper-ated and maintained by KEPC since1979, the first major issue on powercable design came up in 2006, afterthe failure of the cable between Henamand Jeju. From the accident, it wasfound that the concrete mattresses in-stalled were not good enough to en-dure current anchor specificationsand activities. That is why this studyinvestigated the newly designed cableprotector. With the series of field an-chor collision and dragging tests, theKorean government is focusing on a

May/J

safety evaluation guideline for sub-marine power cable protection, and as-sociated projects will launch in 2013.The safety guideline will be completedin 2015.

There is no international designguideline and standard. At the SubseaPower Cables Conference in London(April 2012), a roundtable discussiondealt with the necessities of interna-tional standards set for cable installa-tion projects. The burial protectionindex was suggested as a power cableprotection solution, but this methodis conceptual. Sharples (2011) empha-sized the fact that different cable pro-tection technologies developed overtime lend themselves to specific condi-tions and scenarios. These includerock dumping, preformed concretemattress, assisting natural sedimenta-tion with frond mats, durable paddingor grout-filled bags, external plastic(polyurethane) or metal sheathing onthe cable, heavy armoring built intothe cable itself, and custom solutions.Det Norske Veritas (DNV) publishedthe recommended practice DNV-RP-F401 (Electrical Power Cables in SubseaApplications, 2012). The recom-mended practice (RP) focuses not oncable protection but on the electricalpower cables themselves. Because ofthe varieties in cable protection solu-tions and man-made or natural haz-ards to submarine power cables, it ishard to set the design guideline andstandard. Therefore, at the moment,most work focuses on mitigating therisk of cable damage through cableprotection solutions and planning thebest cable route, etc.

However, Det Norske Veritas pub-lished a related recommended practiceDNV-RP-F107 (Risk Assessment ofPipeline Protection, 2010). This recom-mended practice has a chapter called“Pipeline and Protection Capacity,”


which deals with steel pipeline, flexiblepipeline, umbilicals, and different pro-tection methods. Here, umbilicals aretypically a complex compound of tub-ing, electrical wires, reinforcement,and protective layers. In other words,umbilicals consist of hydraulic linesused to transfer chemicals and fluidsand electric and fiber optic componentsto monitor pressure, temperature, flowand vibration, while submarine powercables are used for transmission of elec-tric power. Therefore, the recommendedpractice can be useful to develop thedesign practice of submarine powercable protection.

Material Preparationand Test Set-up

A dimensional description of thepower-cable protector is shown in Fig-ure 1. The proposedmattress-type pro-tector was assembled from 70 blocks(5 EA × 14 EA), as shown in Figure 2a.Here, the first notation (5 EA) indicatesthe number of blocks in the y direction(width), and the secondnotation (14EA)indicates the number of blocks in thex direction (length). There were threetypes of blocks—left, basic, and right—meaning that each column consistedof one left, one right, and three basic


blocks between the left and right blocks.These arrangements were designed tochange the size of the protector. Theassembly could be performed on a ves-sel using circular links and arch hooks(Figure 2b).

Each block was 0.5 m (length) ×0.5 m (width) × 0.1 m (depth) and wasmanufactured from reinforced concreteto increase its compressive and ten-sile strengths. The design concretecompressive strength was 52 MN/m2.The reinforcement array is shown inFigure 2c. Here, the tensile strengthof the reinforcing bars (SD 400) was560 MN/m2, the diameter was 13 mm,and the entire length of the reinforc-ing bars was 6.274 m in each concreteblock. The hooks were zinc-coated,cold-drawn, round steel bars with atensile strength of 686 MN/m2 and adiameter of 16 mm. Once the rein-forced concrete blocks were manufac-tured, these arch-shaped hooks werelinked by the circular links (Figure 2b)so that the blocks were assembled andextended to the mattress. These arch-

l

shaped hooks were connected to thereinforcing bars (Figure 2c).

Concrete is often considered tothe ultimate construction material, be-cause of its considerable strength, rela-tively low cost, and virtually limitlessdesign capabilities. When subjectedto an impact or collision, concrete isgenerally submitted to both a multi-axial state of stress characterized by ahigh mean stress and high strain rates.It is known that the only material pa-rameter governing impact resistance ofconcrete structures is based on the uni-axial unconfined compressive strengthof concrete (Daudeville & Malécot,2011). Therefore, it is necessary forconcrete to have a significant increasein its toughness, hence impact resis-tance (Liu et al., 2012).

Field drop tests on the protectorwere carried out in an open space.For the field-test set-up, a 2-ton stockanchor (KSV3311, 2006) and a 25-toncrane (Figures 2d and 2e) were pre-pared. To analyze the movements ofthe protector and anchor, a camcorderwas used.

It should be noted here that, insteadof a water environment, tests were car-ried out on land because otherwise itwould be quite difficult to observe thestate of the protector during the anchordrop process. To accommodate thewater environment, the effect of waterresistance on the anchor was consideredin the test by facilitating the soil condi-tions of the target sites and convertingthe terminal velocity of the anchorin water into a falling distance of theanchor in air.

To simulate the soil conditions onthe target seabed (Haenam), the soilswere surveyed and the soil composi-tions were obtained using the AmericanAssociation of State Highway andTransportation Officials (AASHTO)soil classification method (Table 1).

FIGURE 1

Dimensional description of the mattress-typepower-cable protector.

FIGURE 2

Materials and facilities: (a) assembly of 70blocks (5 EA × 14 EA), (b) circular links andarch hooks used to assemble each block,(c) array of reinforcing bars and hooks ineach concrete block, (d) 2-ton stock anchor,and (e) 25-ton crane.

However, in the field tests, approximatevalues were used instead of the exactpercentage of each sediment texture(Table 1). Case 1 had 50% gravel and50% sand, while Case 2 had 3% graveland 97% sand. In addition to the com-position, the soils were classified intodry and saturated states by adjustingtheir water content. However, the de-gree of saturation was not recorded indetail; only rough descriptions—dryand saturated—were used in the tests.

Considering the tidal effects on thewater depths at the target site, it wasassumed that the terminal velocitywas the design velocity of the stockanchor. However, in the calculations,it was not easy to determine the dragcoefficient of the stock anchor becauseof its unusual shape. Thus, a rangeof drag coefficients, from 0.7 to 1.7(Table 2), was introduced to investi-gate a range of terminal velocities. Bymaking use of the principle of energyconservation, the falling distances inair were calculated, ranging from 1.78to 4.32 m. Because of the maximumvalue of 4.32 m (4.96 m in water), afalling distance of 5 m was consideredin the field tests. This selection wasconservative in the sense that the cor-responding drag coefficient to themaximum falling distance was 0.7,which is much smaller than the dragcoefficients of a C-section (2.30 or1.20), square prism (2.05 or 1.05),and disk (1.17) (Fox et al., 2004).However, it should be emphasizedhere that the protector was designed

to be located not only at the targetsite but also at other sites. Thus, inaddition to 5 m, 9 m was used as afalling distance in the tests.

Considering the added mass effectson the tests, it should be noted that thestock anchor reaches the falling statewith a constant falling velocity (the ter-minal velocity) in the target sites. Inother words, the anchor is not acceler-ating before colliding. As this happens,the effect of the added mass on theanchor is not significant. Therefore,the effects of the added mass on thetest results were not considered in thetest. In addition, because of the pres-ence of water, pressure effects in deepwater and the resulting soil behaviorare other factors to be considered inthe tests. However, their effects on

May/J

the test results were assumed minor;hence, these were not considered inthe tests.

Before carrying out the field tests,two preliminary tests were performed,designed as shown in Table 3, topinpoint the measuring points anddeformation patterns. The effects ofvariations in the soil composition(Case 1 and Case 2) and mattress spec-ifications (3 EA × 3 EA and 5 EA ×14 EA) were also investigated. Otherfactors, such as the soil condition andfalling distance, were fixed as dry and10 m, respectively. In the preliminarytests, the safety assessment was madethrough the failure of blocks and theresulting settlement depth and shape.Unlike the main test program, a plas-tic pipe under the mattress was notinstalled.

Finally, a criterion was decided forthe safety or failure assessment of theprotector after collision with the 2-tonstock anchor for the main test program.For the assessment, a safety zone wasintroduced along the center of thepower-cable location. To simulate this,a plastic pipe was embedded below the

TABLE 1

Soil composition at the target site.

Ground

Sediment Texture (%)

Gravel
Sand Silt and Clay
Case 1
49.62 (≈50) 49.88 (≈50) 0.5 (≈0)
Case 2
2.89 (≈3) 96.81 (≈97) 0.3 (≈0)
TABLE 2

Terminal velocity and falling distance with respect to drag coefficient.

Drag Coefficient
Terminal Velocity (m/s)
une

Energy (J)

2013 Volume

Falling Distance (m)

0.7
9.856 84755 4.32
0.8
9.219 74153 3.78
0.9
8.692 65918 3.36
1.0
8.246 59326 3.03
1.1
7.682 53930 2.75
1.2
7.528 49445 2.52
1.3
7.232 45633 2.33
1.4
6.969 42374 2.16
1.5
6.733 39553 2.02
1.6
6.519 37078 1.89
1.7
6.324 34893 1.78
47 Number 3 75

mattress. The mattress was assumed tohave failed if the anchor and brokenconcrete blocks delivered a compres-sive force to the pipe. In total, fourscenarios were designed to provide asafety assessment of the protectorunder anchor drops with the two-tonstock anchor (Table 4).

Drop Test ResultsThe field tests, described inTables 3

and 4, were carried out on open land atPukyong National University, locatedin Busan. As described earlier, a 25-toncrane was used to drop the two-tonstock anchor to the protector. Beforecarrying out the four drop scenarios(Table 4), preliminary drop tests wereperformed to investigate damage pat-terns, determine measuring points,and assess the effects of soil composi-tions. Figure 3 shows a typical collidingsequence or process. In all of the tests,the same colliding process occurred, asfollows.


A detailed description of P1 is pro-vided in Table 3. For the test, first, thedry soil composition of Case 1 was pre-pared below the mattress and the stockanchor was dropped from 10m. In thiscase, a 3 EA × 3 EA mattress was pre-pared instead of a 5 EA × 14 EA. As aresult (Figure 4), the blocks near thecollision area were broken, and threeblocks were crushed (Figure 4a). Afterremoving the protector (Figure 4b),the measured maximum vertical soilsettlement was 7 cm. Despite the rela-tively shallow settlement, the protectorfailed to protect the power cable be-

l

cause of the broken blocks under themattress and because of the settlementdepth and shape.

A detailed description of P2 is pro-vided in Table 3. For the test, the drysoil composition ofCase 2was preparedbelow the mattress and the stock an-chor was dropped from 10 m. Here, a5 EA × 14 EA mattress was prepared.The protector was damaged (Figure 5),and the blocks near the collision areawere broken. Two blocks were totallycrushed (Figures 5a and 5b). After re-moving the protector (Figures 5c and5d), the measured maximum verticalsoil settlement was 37 cm. The verticalsettlement differed from that in P1, al-though the falling distance of T1 andT2 was the same. It happened becauseof the difference in the soil composi-tion between T1 and T2. That is, thesoil composition of Case 2 (gravel 3%and sand 97%) was much weaker than

TABLE 3

Preliminary drop tests.

Identity
SoilCondition
SoilComposition

FallingDistance

Mattress Specifications

No. of Rows
No. of Columns
P1
Dry Case 1 10 m 3 3
P2
Dry Case 2 10 m 5 14
TABLE 4

Anchor drop test scenarios.

Identity
SoilCondition
SoilComposition

FallingDistance

Mattress Specifications

No. of Rows
No. of Columns
T1
Dry Case 1 5 m 5 14
T2
Dry Case 1 9 m 5 14
T3
Saturated Case 2 5 m 5 14
T4
Saturated Case 2 9 m 5 14
FIGURE 3

A typical anchor drop process between theanchor and protector.

FIGURE 4

Test result P1 (falling distance: 10 m, soil:Case 1, mattress 3 EA × 3 EA).

FIGURE 5

Test result P2 (falling distance: 10 m, soil:Case 2, mattress 5 EA × 14 EA).

that of Case 1 (gravel 50% and sand50%). For Case 2, the mattress pro-tector also failed to protect the powercable because of the broken blocksunder the mattress and the settlementdepth and shape.

Considering T1 (Table 4), the drysoil composition of Case 1 was pre-pared below the mattress and the stockanchor was dropped from 5 m (Fig-ure 6a). The mattress arrangementwas 5 EA × 14 EA. Blocks near thecollision area were broken (Figure 6b),resulting in a damaged area that was90-cm long and 17-cm wide, for atotal area of 1,530 cm2 (Figure 6c).After removing the protector, it wasfound that the measured maximumvertical soil settlement was about5 cm and the embedded pipe was dam-aged (Figure 6d).

Considering T2 (Table 4), the drysoil composition of Case 1 was pre-pared below the mattress and thestock anchor was dropped from 9 m(Figure 7a). In this case, a 5 EA ×14 EA mattress was prepared. Theblocks near the collision area were bro-ken (Figure 7b), resulting in a damagedarea that was 90-cm long and 28-cmwide, for a total area of 2,520 cm2 (Fig-ure 7c). After removing the protector,it was found that the measured maxi-

mum vertical soil settlement was 7 cmand the embedded pipe was severelydamaged (Figure 7d).

Considering T3 (Table 4), the sat-urated soil composition of Case 2 wasprepared below the mattress and thestock anchor was dropped from 5 m(Figure 8a). The mattress arrangementwas 5 EA × 14 EA. The blocks near thecollision area were broken (Figure 8b),resulting in a damaged area that was96-cm long and 32-cm wide, for atotal area of 3,072 cm2 (Figure 8c).After removing the protector, it wasfound that the measured maximumvertical soil settlement was 13 cmand the embedded pipe was severelydamaged (Figure 8d).

May/J

Considering T4 (Table 4), the sat-urated soil composition of Case 2 wasprepared below the mattress and thestock anchor was dropped from 9 m(Figure 9a). The mattress arrangementwas 5 EA × 14 EA. The blocks near thecollision area were broken (Figure 9b),resulting in a damaged area that was130-cm long and 50-cm wide, for atotal area of 6,500 cm2 (Figure 9c).After removing the protector, it wasfound that the measured the max-imum vertical soil settlement was35 cm and the embedded pipe wasseverely damaged (Figure 9d).

Safety AssessmentThe anchor drop test results are

summarized in Table 5. As shown inthe table, all of the test results showedthe failure of the mattress power-cableprotector under the anchor drop sce-narios tested. This failure assessmentwas due to the excessive compressiveforce on the pipe under the mattress,with permanent deformation andbreakage of the pipe. The collisionforces on the reinforced concrete blockstransferred to the soil and obviously tothe pipe. To help assess the test results,Figure 10 shows the settlement depthand damaged area, respectively, versus

FIGURE 6

Test result T1 (falling distance: 5 m, soil:Case 1, mattress 5 EA × 14 EA).

FIGURE 7


FIGURE 8


FIGURE 9

Test result T4 (falling distance: 9 m, soil:Case 2, mattress 5 EA × $214 EA).


the falling distances (5 and 9 m). Fromthe figures, the increase in falling dis-tance increased the settlement depthand damaged area, although the slopesdiffered. Considering the effect of thesoil conditions and composition, thesaturated Case 2 (gravel 3% and sand97%) gave deeper settlement depthsand wider damaged areas than thedry Case 1 (gravel 50% and sand50%) did. It was shown that this wasmainly due to the soil composition ef-fect in addition to soil saturation be-cause, in the preliminary tests, Case 2gave a deeper depth (37 cm) thanCase 1 (7 cm) did, although the soilconditions were dry.

In the anchor drop tests, anchorvelocity is the most important factor


because it is directly related to the im-pact energy. The anchor velocity inwater was converted to the velocity inair (Woo & Na, 2010). In that sense,the water effect on the impact was con-sidered. In tunnel-type protectors,power cables do not have to be embed-ded in the seabed. In most cases, theprotectors cover the power cable,which is located on the seabed. Thus,the main energy dissipation is accord-ing to the toughness of the concretemattress. However, as shown in thetest results, soil composition and soilsaturation had effects on the test re-sults. In that sense, the water effecton the soil was partially considered inthe tests. Regarding to the water effecton the mattress, it should be empha-

l

sized here that reinforced concretestructures in water generally deterio-rate as time goes. Thus, it is expectedthat the protector in water will havedifferent reaction to anchor collisionduring its service life. However, Kimet al. (2008) reported that reinforcedconcrete reefs, immersed in seawaterfor 18–25 years, have sound physicaland chemical properties. Nonetheless,in this field study, there were only lim-ited test cases to distinguish the effectsof soil composition and saturation onthe test results.

Considering the test results, thenewly designed protector should bemodified and the burial condition ofthe power cable should be specifiedto dissipate impact energy. A series of

TABLE 5

Failure assessment.

Identity
SoilDescription
FallingDistance (m)

DamagedArea (cm2)

SettlementDepth (cm)
Pipe State
FailureAssessment

T1
Dry Case 1 5 1,530 5 Damaged Failed
T2
Dry Case 1 9 2,520 7 Damaged Failed
T3
Sat. Case 2 5 3,072 13 Damaged Failed
T4
Sat. Case 2 9 6,500 35 Damaged Failed
FIGURE 10

Settlement depth (a) and damaged area (b) versus falling distance obtained from T1 to T4.

field anchor drop and dragging testswill be scheduled to investigate themechanism and accordingly to establisha safety evaluation guideline of sub-marine power cable protection in thenear future. The associated projectsin South Korea will be launched in2013 and are expected to be done in2015.

ConclusionsThis study presents anchor drop

tests on a newly designed matrix-typesubmarine power-cable protector as-sembled from reinforced concreteblocks to make a safety assessment.For the study, the conditions of themarine environments, such as waterdepths and soil specifications, weresurveyed at the target site and sim-ulated in the test set-up. A 2-tonstock anchor was selected as the collid-ing object, a 25-ton crane was preparedto drop the anchor, and recordingunits were set up to capture the col-lision process and damage patterns.Before carrying out designated test sce-narios, preliminary tests were per-formed to investigate the effects ofthe soil composition and mattress ar-rangement on the test results. Finally,four anchor drop test scenarios weredesigned, carried out, and analyzed,and then a safety assessment wasmade for the matrix-type submarinepower-cable protector.

From the tests, in addition to thefalling distance, it was found that soilcomposition was a significant factorin the settlement depth and damagedarea. Case 2 (3% gravel and 97%sand) gave deeper settlement depthsand wider damaged areas than Case 1(50% gravel and 50% sand) did. Thisobservation was also made in the pre-liminary tests; thus, in addition to thesoil conditions (dry and saturated), it

was shown that soil composition didprovide a representative effect on thesettlement depth and damaged area.Considering the settlement depth ofthe soils, the damaged areas of the con-crete blocks, and the damaged state ofthe pipes (safety zone), all of the testresults showed that the mattress failedto protect the power cable from the an-chor collision forces. The deformation,damage, and breakage of the pipe, sim-ulating the safety zone of the powercable, gave clues as to the failure. How-ever, it should be noted here that thefailure assessment in this study doesnot necessarily ideally represent thereal failure of the power cables in thetarget marine environment because ofreal power cable specifications, such asthe presence of armor layers. Also, inthe calculations of falling distances, aconservative selection would result ina worse failure assessment. Never-theless, this study shows the collisionprocess of the stock anchor on themat-tress, damaged patterns of the concreteblocks, soil settlement, and effect ofsoil composition on the test results.Thus, it was shown that the test pro-cess and results provide valuable infor-mation to design further anchor droptests, specify the soil conditions, andestablish a safety evaluation guidelineof submarine power cable protectionin the near future.

AcknowledgmentsThis study was supported finan-

cially by the Singangjin Power Stationof the Korea Electric Power Corpora-tion (KEPCO). We sincerely thankDr.Hongjin Kim for his technical sup-port. Additionally, material prepara-tion by Samsung Industry is greatlyappreciated. It should be noted herethat this paper does not necessarilyrepresent the views of the funding

May/J

agency or its officers, agents, andemployees.

Corresponding Author:Won-Bae NaDepartment of Ocean Engineering,Pukyong National University,Busan, South KoreaEmail: [email protected]

ReferencesByrne, B.W., & Houlsby, G.T. 2012.

Foundations for offshore wind turbines.

Philos Trans R Soc A. 361:2909-30. http://

dx.doi.org/10.1098/rsta.2003.1286.

Coffen-Smout, S., & Herbert, G.J. 2000.

Submarine cables: A challenge for ocean

management. Mar Policy. 24:441-8. http://

dx.doi.org/10.1016/S0308-597X(00)00027-0.

Daudeville, L., &Malécot, Y. 2011. Concrete

structures under impact. Eur J Environ

Civil Eng. 15:101-40. http://dx.doi.org/

10.1080/19648189.2011.9695306.

DNV-RP-F107. 2010. Risk Assessment of

Pipeline Protection. Katy, TX: Det Norske

Veritas.

DNV-RP-F401. 2012. Electrical Power Cables

in Subsea Applications. Katy, TX: Det Norske

Veritas.

Esteban, M.D., Diez, J.J., López, J.S., &

Negro, V. 2011. Why offshore wind energy?

Renew Energ. 36:444-50. http://dx.doi.org/

10.1016/j.renene.2010.07.009.

Fox, R.W., McDonald, A.T., & Pritchard, P.J.

2004. Introduction to Fluid Mechanics.

Hoboken, NJ: Wiley.

International Cable Protection Committee.

2009. Damage to submarine cables caused by

anchors. Loss Prevention Bull. Available at:

http://www.iscpc.org/publications/Loss_

Prevention_Bulletin_Anchor_Damage.pdf.

Jie, W., & Yao-Tian, F. 2012. Study on safety

monitoring system for submarine power cable

on the basis of AIS and radar technology.


Phys Proc. 24:961-5. http://dx.doi.org/

10.1016/j.phpro.2012.02.144.

Kim, H.S., Kim, C.G., Na, W.B., Woo, J.,

& Kim, J.K. 2008. Physical and chemical

deterioration of reinforced concrete reefs in

Tongyeong coastal waters, Korea. Mar Technol

Soc J. 42(3):110-8.

KS V3311. 2006. Anchor. Korean Standards

Association.

Liu, F., Chen, G., Li, L., & Guo, Y. 2012.

Study of impact performance of rubber

reinforced concrete. Constr Build Mater.

36:604-16. http://dx.doi.org/10.1016/

j.conbuildmat.2012.06.014.

Sharples, M. 2011. Offshore electrical cable

burial for wind farms: state of the art, standards

and guidance & acceptable burial depths,

separation distances and sand wave effect.

Washington, DC: Bureau of Safety and Envi-

ronmental Enforcement.

Wagner, E. 1995. Submarine cables and

protections provided by the law of the sea.

Mar Policy. 19:127-36. http://dx.doi.org/

10.1016/0308-597X(94)00002-A.

Woo, J., Na, W.B., & Kim, H.T. 2009.

Numerical simulation of arch-type submarine

cable protector under anchor collision.

J Ocean Eng Technol. 23:96-103.

Woo, J., & Na, W.B. 2010. Analyses of the

maximum response of cylinders-connected

protector under anchor colliding and dragging.

J Ocean Eng Technol. 24:81-7.

Yamabuki, K., & Kubori, K. 2012. A tran-

sient analysis of a scaled model for a sub-

marine cable connected with an offshore wind

farm. Electr Pow Syst Res. 85:59-63. http://

dx.doi.org/10.1016/j.epsr.2011.07.009.


P A P E R

Elements of Underwater GliderPerformance and Stability
A U T H O R SShuangshuang FanThe State Key Lab of Fluid PowerTransmission and Control,Zhejiang University
Craig WoolseyDepartment of Aerospace andOcean Engineering, Virginia Tech

A B S T R A C T
Underwater gliders are winged autonomous underwater vehicles (AUVs) that
can be deployed for months at a time and travel thousands of kilometers. Aswith any vehicle, different applications impose different mission requirementsthat impact vehicle design. We investigate the relationship between a glider’sgeometry and its performance and stability characteristics. Because our aim is toidentify general trends rather than perform a detailed design optimization, we con-sider a generic glider shape: a cylindrical hull with trapezoidal wings. Geometricparameters of interest include the fineness ratio of the hull, the wing positionand shape, and the position and size of the vertical stabilizer. We describe theresults of parametric studies for steady wings-level flight, both at minimum glideangle and at maximum horizontal speed, as well as for steady turning flight. Wedescribe the variation in required lung capacity and maximum lift-to-drag ratiocorresponding to a given vehicle size and speed; we also consider range and en-durance, given some initial supply of energy for propulsion. We investigate how theturning performance varies with wing and vertical stabilizer configuration. Tosupport this analysis, we consider the glider as an 8-degree-of-freedom multibodysystem (a rigid body with a cylindrically actuated internal moving mass) anddevelop approximate expressions for turning flight in terms of geometry and controlparameters. Moving from performance to stability and recognizing that a glider’smotion is well described in terms of small perturbations from wings-level equilib-rium, we study stability as an eigenvalue problem for a rigid (actuators-fixed) flightvehicle. We present a number of root locus plots in terms of various geometric param-eters that illuminate the design tradeoff between stability and control authority.Keywords: multibody dynamics, design analysis, range and endurance, buoy-ancy propulsion

ocean sampling missions. More recentdevelopment efforts have produced var-

1. Introduction

For at least two decades, experi-mental ocean scientists have envi-sioned a dense distribution of mobileocean sensor platforms to provide per-sistent and pervasive ocean monitoring(Curtin et al., 1993). Within this vi-sion, there is an economic argumentfor using underwater gliders to samplethe deep ocean over large spatial andtemporal scales (Rudnick et al., 2004).The first generation of underwatergliders, such as Slocum (Webb et al.,2001), Seaglider (Eriksen et al., 2001),and Spray (Sherman et al., 2001), weredesigned for such long-endurance, deep

iants that are suitable for both deeper(Osse & Eriksen, 2007) and shallower(Teledyne Webb Research, 2013) op-erations. Although glider endurancesuffers in shallower water, it compareswell with that of conventional under-water vehicles, and gliders are more ro-bust to biological fouling and debris.Moreover, advanced perception, plan-ning, and control algorithms that exploitnatural vehicle and environmental dy-namics can improve performance. Ashallow-water glider developed recently

at Virginia Tech provides unique newcapabilities for testing such algorithms(Wolek et al., 2012).

Earlier studies have addressed vari-ous stability and performance tradeoffsinvolved in underwater glider design.A sweeping 2003 analysis sponsoredby the U.S. Office of Naval Researchprovided insightful comparisons be-tween the propulsive efficiency of un-derwater gliders and both biologicaland engineered flyers ( Jenkins et al.,2003). One conclusion of that study

May/J

was that a blended wing-body gliderwould yield greater flight efficiencythan “legacy” configurations. The pri-mary focus of the 2003 glider studywas performance; earlier investigationshad considered stability. Graver et al.(1998) investigated the stability ofwings-level flight with respect to a“bottom-heaviness” parameter relatedto the glider’s metacentric height.Galea (1999) provided a general staticstability analysis for glider wings-levelflight and turning motion. Graver


(2005) analyzed scaling rules for steadyglide performance (e.g., the glide speedversus ballast loading, glide angle andglider volume), and he provided apreliminary overview of glider designrequirements, including sizing, body,and wing geometry. Graver (2005)also discussed longitudinal stability,focusing especially on the role of thephugoid mode, a mode that had beenrecognized in aircraft motion a cen-tury before by Lanchester (Lanchester,1908;Mises, 1959). Nonlinear glidingstability is discussed in Bhatta andLeonard (2008), where a compositeLyapunov function is constructed toprove gliding stability. One advantageof Lyapunov analysis is that it pro-vides a tool for estimating the regionof attraction for asymptotically stableequilibria; such information can beuseful in scheduling transitions be-tween steady motions.

In any flight vehicle design effort, itis important to understand the rela-tionship between vehicle geometryand performance and stability (Pamadi,2004). To simplify our parametric anal-ysis, we describe the hull and winggeometry using several nondimen-sional parameters: the hull finenessratio, wing aspect ratio, wingspanratio, and wing thickness ratio. In ad-dition to these parameters, the wingand vertical stabilizer configurations(shape, size, and location) are also con-sidered. As hydrodynamic coefficientsare determined by the glider geometryand the flow characteristics, it is neces-sary to obtain a hydrodynamic modelfor the glider. For generic vehicleshapes, there are well-known semi-empirical expressions relating geometryto aerodynamic (or hydrodynamic) coef-ficients (Etkin & Reid, 1996; USAFStability and Control DATCOM).Because our focus is to identify generaltrends rather than perform a detailed


design optimization, we use simplemethods such as these to develop ageneric glider hydrodynamic modelthat supports performance and stabilityanalysis. We consider two wings-levelflight conditions (minimum glide angleand maximum horizontal speed) as wellas turning motion. To support our anal-ysis of turning motion for a vehiclewith a cylindrically actuated internalmoving mass, we follow the regularperturbation method described inMahmoudian et al. (2010). The per-formance analysis covers preliminarydesign considerations, such as sizing,range and endurance, and turning capa-bility. For stability analysis, we view theglider as a rigid body and investigatethe effect of geometric parameters onthe longitudinal and lateral-directionaleigenmodes. Specifically, we considerthe effect of changes in wing shape,wing longitudinal location, and verti-cal stabilizer size and location.

In this paper, we consider a con-ventional glider configuration compris-ing a cylindrical hull with a trapezoidalwing and vertical stabilizer. The anal-ysis makes use of a new multibodydynamic model for a glider with a cy-lindrically actuated moving mass. Thisactuation scheme matches that of sev-eral existing gliders including Seaglider,Spray, and the Virginia Tech underwaterglider (Wolek et al., 2012). The paperis organized as follows. Section 2 pre-sents the derivation of an 8-degree-of-freedom underwater glider dynamicmodel with a cylindrically actuatedmoving mass. In Section 3, we definethe nomenclature associated with glid-er geometry and with vehicle motionand we review relevant glider hydrody-namic characteristics. Section 4 presentssome analytical results concerning per-formance in steady wings-level flightand in steady turning motion. Sec-tion 5 describes how the longitudinal

l

and lateral-directional eigenmodes fora rigid glider vary with certain geomet-ric parameters. We summarize andconclude in Section 6.

2. Underwater GliderDynamics With aCylindrically ActuatedMoving Mass

The 8-degree-of-freedom glider ismodeled as a rigid body (mass mrb)with a single moving point mass (mp)that is offset from the centerline. Thepoint mass can move longitudinallyand can revolve around the vehicle cen-terline. Earlier multibody dynamicmodels incorporated one or morerectilinear mass particles for attitudecontrol, but the cylindrical actuationscheme better reflects the actual con-trol mechanism in several existinggliders.

The vehicle also includes a variableballast actuator whose effect is repre-sented by a point mass (mb) with vari-able volume (Vb), which is fixed at thebody frame origin. Inmodeling a buoy-ancy control system, one may assumeeither that the mass varies or the dis-placement, depending on whether the“control volume” that contains thevehicle is fixed or deformable. In eithercase, the net effect is a change in thevehicle’s density. Thus, in our model,when the variable ballast actuator de-creases the vehicle’s volume sufficiently,the vehicle becomes denser than thesurrounding water and descends. Con-versely, when the actuator increases thedisplaced volume, the vehicle becomesless dense than the surrounding waterand ascends.

The total vehicle mass, whichremains fixed, is

m ¼ mrb þ mp þ mb

2.1. KinematicsDefine a body-fixed, orthonormal

reference frame centered at the geo-metric center of the vehicle (the cen-ter of buoyancy) and represented bythe unit vectors b1, b2, and b3. Thevector b1 is aligned with the longi-tudinal axis of the vehicle, b2 pointsout the right wing, and b3 completesthe right-handed triad (see Figure 1).Define another orthonormal referenceframe, denoted by the unit vectors i1,i2, and i3, which is fixed in inertialspace such that i3 is aligned with theforce due to gravity. The relative orien-tation of these two reference frames isgiven by the proper rotation matrixRIB, which maps free vectors fromthe body-fixed reference frame intothe inertial reference frame. Let eirepresent the standard basis vectorfor R3, where i Î{1, 2, 3}, and let thecharacter � denote the 3 × 3 skew-symmetric matrix satisfying ab ¼a× b for vectors a and b. Then, interms of conventional Euler angles

(roll angle φ, pitch angle θ, and head-ing angle ψ), we have

RIB φ; θ;ψð Þ ¼ ee3ψee2θee1φ

whe r e e ( · ) d eno t e s th e ma t r i xexponential.

The origin of the body frame sits atthe vehicle’s center of buoyancy (CB),which remains fixed under our as-sumptions. The center of mass (CM)of the glider (neglecting the contri-bution of mp) is located at rrb (seeFigure 1). Let the inertial vector X =[x, y, z]T represent the position vectorfrom the origin of the inertial frame tothe origin of the body frame. Let v =[u, v, w]T and ω = [p, q, r]T representthe translational and rotational veloci-ty of the body with respect to the iner-tial frame, but expressed in the bodyframe. The kinematic equations are

X� ¼ RIBv ð1Þ

R�IB¼ RIB

^ω ð2Þ

May/J

In addition to the 6 degrees of free-dom associated with the vehicle’s trans-lation and rotation, there are 2 degreesof freedom associated with the movingmass, which is modeled as a particle. Todescribe the pointmass position, we de-fine a third orthonormal “actuator” triad{a1, ar, aμ}, where a1 is parallel with b1(see Figure 1). The vector ar points inthe radial direction from the vehiclecenterline through the point mass. Thevector aμ completes the right-handedframe. The proper rotation matrix RBP

maps free vectors from this particle-fixed frame to the body frame,

RBP ¼1 0 00 �sinμ �cosμ0 cosμ �sinμ

0

@

1

A

where μ is the rotation angle of themoving mass about the longitudinalaxis of the vehicle. Let

∼rp ¼ rpxa1 þ Rpar ¼rpxRp

0

0

@

1

A

p

denote the particle’s position with re-spect to the body frame origin, expressedin the particle frame. Correspondingly,define the body vector

rp ¼ RBP∼rp¼

rpx�Rp sinμRp cosμ

0

@

1

A

b

Then the kinematics of the movingmass particle relative to inertial space is

vp ¼ vþ ω þ μ� b1Þ× rp þ r�pxb1Þ ð3Þ��

Note that we may also write

vp ¼�

I �^rp e1 �^rpe1�

η

where η ¼

vωr�pxμ�

!

:

FIGURE 1

An underwater glider with a cylindrically actuated moving point mass.


2.2. DynamicsTo derive the equations of motion,

one first defines the total kinetic energyof the body/particle/fluid system,

T ¼Tp þ Tb þ Trb þ Tf ¼ 12ηTM rp

� �η

where

M rp� � ¼ Mp rp

� �þMb þMrb þMf

Explicit expressions for the generalizedinertiamatrices are given in Appendix A.

The generalized momentum is

phppxhpx

0

BB@

1

CCA ¼ ∂T

∂η

where p is the total body translationalmomentum, h is the total body an-gular momentum, and ppx and hpx ap-pear in the point mass momentum asfollows:

pp ¼ RBP

ppxppr

hpx=Rp

0

@

1

A

p

The dynamic equations in the bodyframe are

p� ¼ p ×ω þ ∼mgζþ fv ð4Þ

h� ¼ h ×ω þ p × vþ mrbgrrb × ζ

þ mpgrp × ζþmv ð5Þ

p�px¼ e1⋅

�

RTBP

�pp ×

�ωþ μ� b1

�

þ mpgζ��þ upx ð6Þ

h�px¼ Rpe3⋅

�

RTBP

�pp ×

�ω þ μ� b1

�

þ mpgζ��þ τpx ð7Þ


where

ζ ¼ RT i3

is the “tilt vector,” the body frame unitvector in the direction of gravity, fvand mv are the viscous force and mo-ment acting on the glider, expressedin the body frame, and upx and τpx arethe input force and moment, whichadjust the moving point mass locationto control vehicle attitude.

Letting ub denote the rate of changeof displaced volume, an input, we have

V�b ¼ ub ð8Þ

Equations (4) through (8) describe thedynamics of the cylindrically actuatedglider. Together with the kinematicequations (1), (2), and (3), these equa-tions completely describe the motionof the vehicle in still water.

Having obtained a multibody dy-namicmodel, onemay compute steadymotions, examine their stability, de-velop feedback control schemes, andimplement them in simulation. Forexample, one may investigate flightcontrol methods that involve stitch-ing together sequences of stable glid-ing motions, as in Mahmoudian andWoolsey (2013).

3. Glider Geometryand HydrodynamicCharacteristics3.1. Geometry and Nomenclature

Motivated by the challenge of de-signing an underwater glider to meetparticular mission requirements, weinvestigate the general relationship be-tween glider geometry and the stabilityand performance characteristics. Indoing so, we consider a generic wing-and-cylinder configuration, where the

l

cylindrical fuselage has diameter d andlength l. The wingspan (tip to tip) is b.We consider an untapered, trapezoidalwing with a constant cross-sectionalshape. The wing planform area is de-noted S. The mean aerodynamic chordlength (i.e., the average width of a rect-angular wing with equivalent aerody-namic properties) is c and S ¼ b⋅c.Another important geometric param-eter is the maximum wing thickness t.

The hull and wing geometry can becharacterized by several nondimen-sional parameters: the hull finenessratio ( f ), the wing aspect ratio (AR),the wingspan ratio (κ), and wing thick-ness ratio (∼t ).

f ¼ ld; AR ¼ b2

S; κ ¼ b

d; and

∼t ¼ tc

We also consider a few other con-figuration parameters for the wingand vertical stabilizer, including winglongitudinal position lw, vertical stabi-lizer longitudinal position lv and areaof the vertical stabilizer Sv. The aerody-namic center of the wing and the ver-tical stabilizer, which is located nearthe quarter-chord line, are

∼lw and

∼lv

(normalized by the fuselage length l )aft of the center of buoyancy, respec-tively, while the area of the vertical sta-bilizer Sv is normalized by the fuselagefrontal area Sf . All of the geometricparameters that we consider are indi-cated in Figure 2.

A critical parameter for perfor-mance analysis is the buoyant lungcapacity η, given as a percentage ofthe neutrally buoyant displacement.Buoyancy actuators include piston-cylinder arrangements, oil-filled blad-ders, and pneumatic bladders. Forsimplicity, we consider the case of apiston-cylinder actuator and assume a

cylindrical hull with constant cross-sectional area (see Figure 3). Recallthat the dynamic model presented ear-lier assumes changes in buoyant vol-ume occur at the origin of the bodyframe. Although a change in buoyancygenerated with the arrangement shownin Figure 3 would induce a moment,we do not account for it here; our aimis merely to relate changes in buoyancyto the vehicle geometry. Let η denotethe percentage difference between thedisplacement required for neutral buoy-ancy (VNB) and the maximum displace-ment (V). That is,

V ¼ ð1þ η ÞVNB and η ¼ V=VNB � 1

where VNB = W /(ρg) and W is thevehicle’s dry weight. The “net weight”of the glider is denoted

∼W ¼ W � B,

where B is the buoyant force.

Let η denote the fractional dis-placement from neutral buoyancy,η ∈��η; η

�, defined such that η > 0

corresponds to an increase in buoy-ancy. Then

B ¼ ρg 1þ ηð ÞVNB

Thus,

∼W ¼W � B ¼ ρgVNB � ρg 1þ ηð ÞVNB

¼ �ηρgVNB

¼ � η1þ η

� �

ρgV

3.2. Hydrodynamic CharacteristicsTo assess glider performance and

stability, it is necessary to develop ahydrodynamic model. The hydro-dynamic forces (lift, side force, and drag)are expressed in the “current” referenceframe, defined such that the 1-axis isaligned with the velocity vector. Let α =arctan(w/u) denote the vehicle’s “angleof attack” and let β = arcsin(v/||v||)denote the “sideslip angle.” The cur-rent frame is related to the body framethrough the proper rotation matrix

RBC α; βð Þ ¼ e�e2αee3β

May/J

The hydrodynamic force and momentare typically represented using non-dimensional coefficients as follows:

fv ¼ RBC

LSFD

0

@

1

A¼ 12ρV 2S

CL

CSF

CD

0

@

1

A

mv ¼KMN

0

@

1

A¼ 12ρV 2S

bCl

cCm

bCn

0

@

1

A

The nondimensional coefficients de-pend on the vehicle geometry and theflight condition. Based on physicalconsiderations, such as left/right sym-metry of the outer mold line, we assumethese coefficients have the followingdependencies

CL ¼ CL�α; ∼q�; CSF ¼ CSF

�β; ∼p; ∼r

�;

CD ¼ CD�α; ∼q�; Cl ¼ Cl

�β; ∼p; ∼r

�;

Cm ¼ Cm�α; ∼q�and Cn ¼ Cn

�β; ∼p; ∼r

�

Here, ∼p; ∼q; ∼r are nondimensional roll,pitch and yaw rates, which are definedas

∼p¼ p= 2V0=bð Þ; ∼q ¼ q=�2V0=c

�

and ∼r ¼ r= 2V0=bð Þ

where V0 is the nominal velocity of thevehicle. (Please note that while V rep-resents volume, V represents speed.)For aircraft flight dynamic models, de-pendence of the forces and momentson vehicle acceleration is typically in-cluded in these nondimensional coeffi-cients as well (e.g., dependence of Cm

on α� ). Here, these effects are insteadincluded in the added mass and inertiaterms. Dependence of the momentson inertial attitude, due to the meta-centric moment, is also incorporatedelsewhere in the formulation.

FIGURE 2

Geometric parameters of the underwater glider model.

FIGURE 3

Notional schematic of the buoyancy actuationsystem.


Following convention, we assumethat the nondimensional hydrodynamicforces and moments above dependlinearly on their arguments (e.g., Cn ¼Cnβ β þ Cnp

∼p þ Cnr∼r ), with the excep-

tion of the drag coefficient, which var-ies quadratically with lift,

CD ¼ CD0 þ KC 2L

The parameterCD0 represents the par-asitic drag due, for example, to skin fric-tion. The parameter K accounts for dragthat is induced when the wing gener-ates lift. It scales inversely with theaspect ratio AR ; induced drag vanishesfor a wing of infinite span.

The hydrodynamic coefficients (suchas Cnβ ) are determined in part by theglider geometry, such as the slender-ness of the fuselage, the shape andlocation of the wing, the size and loca-tion of the vertical stabilizer, etc. Forgeneric flight vehicle shapes, there arewell-known semiempirical expressionsrelating geometry to aerodynamic (orhydrodynamic) coefficients (Etkin &Reid, 1996;USAF Stability andControlDATCOM), although these methodsare imprecise. One can obtain higherfidelity approximations using compu-tational fluid dynamics methods, rang-ing from simple vortex lattice methodsto full Navier-Stokes solvers. However,semiempirical methods provide a sim-ple way to investigate the first-ordereffect of changes in vehicle geometryon performance and stability.

While hydrodynamic coefficientsdepend on glider geometry, they alsodepend on the flow characteristics,which correlate to the Reynolds number

Re ¼ Vxlengthν

where xlength is some characteristic lin-ear dimension (e.g., vehicle length,


hull diameter, or wing chord). In fact,the hydrodynamic coefficients are espe-cially sensitive to Reynolds number forspeeds at which underwater gliderstypically operate. For motion throughwater, these speeds correspond to a crit-ical Reynolds number range in whichflow over the vehicle may (or may not)transition from laminar to turbulent.

4. Performance Analysis4.1. Steady Motions

In this section, we discuss two typesof underwater glider steady motions:wings-level flight and steady turningmotion.

4.1.1. Wings-Level FlightOur discussion of wings-level per-

formance begins with a general over-view and then focuses on two specificflight conditions: flight at minimumglide angle and flight at maximumhorizontal speed.

Figure 4 shows the free body dia-gram for a flight vehicle in wings-level gliding flight. In this sketch,

∼W

denotes the vehicle’s netweight (weightminus buoyant force). Here, we adopta “performance model”—we treat thevehicle as a point mass, ignoring rota-tional dynamics. As explained earlier,the lift and drag force are typically ex-pressed in terms of nondimensional

l

coefficients CL and CD. The coefficientsCL andCD are related to the descent angleξ as shown in the triangle at the right inFigure 4. Summing forces in Figure 4along the velocity direction gives

CD12ρV 2S

� �

¼ ∼W sinξ

The descent angle is the angle from theinertial horizontal plane down to the ve-locity vector; it is the negative of theflight path angle γ that is commonlyused in aircraft flight dynamics (Etkin& Reid, 1996). Solving for speed, onefinds that

V ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2∼W sinξρSCD

s

ð9Þ

The horizontal and vertical componentsof speed are

Vx ¼ V cosξ

Vz ¼ V sinξ

Referring to Figure 4, we note the fol-lowing relationships:

sinξ ¼ CDffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC2D þ C2

L

p and

cosξ ¼ CLffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC2D þ C2

L

p ð10Þ

FIGURE 4

Free body diagram for wings-level gliding flight.

Recall that

∼W ¼ � η

1þ η

� �

ρgV

As mentioned earlier, we assume thatbuoyancy is controlled by a single-stroke (piston-cylinder) actuator. Wetake the piston diameter to be the hulldiameter, and we assume the buoyancyactuation is concentrated at the originof the body frame (i.e., we ignore mo-ments due to changes in buoyancy).When the vehicle’s net weight is maxi-mum, we have η ¼ �η so that

∼W ¼ η

1þ η

� �

ρgV

¼ η1þ η

� �

ρgπ4d 2l

� �

ð11Þ

Substituting equation (11) into equa-tion (9), we find that

V 2 ¼ 2∼W sinξρSCD

¼ πgl2

d 2

Sη

1þ η

� �sinξCD

¼ πgl2

ARκ2

η1þ η

� �1

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC2D þ C2

L

p

ð12Þ

Thus, we find that speed scales withthe square root of linear dimensionl and inversely with the wingspanratio κ. The dependence of speed onaspect ratio is more subtle, since ARaffects CD through the induced dragparameter K.

A. Minimum Glide Angle Flight. Instill water, flight at the shallowest glidepath angle results in the maximumhorizontal range. Referring to Figure 4,the minimum glide angle occurs whenthe lift-to-drag ratio is maximum. Inpowered flight (e.g., for aircraft flying

at a constant altitude), this flight condition corresponds to the minimum dragforce; the values of the drag and lift coefficient in this condition are

CDmd ¼ 2CD0 and CLmd ¼ffiffiffiffiffiffiffiffiCD0

K

r

(Adopting aircraft nomenclature, the subscript “md” stands for “minimumdrag.”)These expressions can be obtained by differentiating the right-hand side of tan ξ =(CD0 +KCL

2)/CLwith respect toCL and finding the roots. Referring to equation (12),we find that the speed for minimum drag (the “speed-to-fly”) is

Vmd ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

πgl2

ARκ2

η1þ η

� �, ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

C2Dmd

þ C2Lmd

qvuut

¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

πgl2

ARκ2

η1þ η

� �, ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiCD0

K1þ 4KCD0ð Þ

rvuut

ð13Þ

Theminimum glide angle condition is illustrated in the plot ofCL versusCD inFigure 5, a curve sometimes called the drag polar. Note that the lift-to-drag ratio ismaximum when a line extending from the origin is just tangent the drag polar.This flight condition corresponds to the shallowest achievable glide angle, which is8° in the case of the Slocum model given in (Graver, 2005; Bhatta, 2006).

B. MaximumHorizontal Speed.Using the lift-and-drag triangle in Figure 4, wefind that

Vx ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2∼WρS

CL

C2L þ C2

D½ �3=4s

FIGURE 5

Drag polar for a model of the Slocum glider.


The value is maximum when

f CLð Þ ¼ CL

C2L þ C2

D½ �3=4

is maximum. Defining σ ¼ 2KCD0 , one finds (after some effort) that

CLmaxVx ¼ffiffiffiffiffiffiffi24K

r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

� 1þ σð Þ þffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1þ σð Þ2 þ 8σ2

qr

CDmaxVx ¼ CD0 þ18K



q �

Noting from Figure 4 that

tanξ ¼ CD

CL;

we find that for wings-level gliding flight at maximum horizontal speed,

tanξ ¼ CDmaxVx

CLmaxVx

¼ffiffiffi2

pσþ 1

ffiffiffi8

p�



q

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


1þ σð Þ2 þ 8σ2q

r

Figure 6 shows the drag-to-lift ratio and the descent angle. Note that in the limitσ → 0, we have ξ → arctan

�1=

ffiffiffi2

p �≈ 35°.

For well-designed flight vehicles, for which the product KCD0 is reasonablysmall, the descent angle for maximum downrange speed is well approximated


by its theoretical limit arctan 1=ffiffiffi2

p� �.

Further, one may verify that in thelimit KCD0 → 0, the lift and dragcoefficients for maximum horizontalspeed are

CLmaxVx ≈ffiffiffi2

pCD0 and

CDmaxVx ≈ CD0 1þ 2KCD0ð Þ

4.1.2. Turning MotionMahmoudian et al. (2010) devel-

oped analytical approximations forsteady turning motion of an under-water glider using regular perturbationtheory, with the vehicle turn rate as theperturbation parameter. The analysisassumed rectilinear motion of the mov-ing point mass. Using the same ap-proach, we rederive the steady turningflight approximation with a cylindricalmoving point mass actuator. We treatturning flight as a small perturbationfrom a nominal, wings-level equilibriumflight condition at speed V0 and pitchangle θ0, at some corresponding angleof attack α0. In wings-level flight, themoving mass rotation angle μ = 0, sothat the roll angle ϕ and the sideslipangle β are both zero. We hold thepitch angle constant at θ0 and perturbthe steady wings-level flight such thatthe body angular velocity vector ω isvertical with a small magnitude ɛωn

where ωn is some characteristic fre-quency (e.g.,

ffiffiffiffiffiffiffig=l

por V0/l ), and ɛ is

the perturbation parameter. To firstorder in ɛ, one finds

V ≈ V0; φ ≈ ɛφ1;

α ≈ α0; β ≈ ɛβ1;

∼m ≈ ∼m0 and μ ≈ ɛμ1

Explicit expressions for φ1, β1, andμ1 are given in equations (14), (15),

FIGURE 6

Drag-to-lift ratio and descent angle versus 2KCD0.

and (16); these analytical expressions enable one to investigate the role of vehicle parameters in a vehicle’s turningperformance.

ξ 1 ¼�1��2gmp

∼m0RpÞÞV0ωn��Cnr l

2 ∼m0Sρþ 2� ∼m0Nv� þ mprpx

�m� Xu�

��cosα0

�cotθ0

� 2� ∼m0Mw� þ mprpx

��mþ Zw��sinα0 �

��S�Cnβ l

∼m0 þ mprpx�CD0 þ CSβ þ C2

LαK α20��ρ

þ 2 ∼m0��Xu� þ Yv� Þcosα0Þcscθ0

��Cnr l


�m� Xu�

��cosα0Þcosθ20

þ �Clp l2 ∼m0Sρþ 2 mrbrrbz þ mpRp

� �m� Zw�ð Þsinα0Þsinθ20 þ

��mrbrrbz þ mpRp

��m� Xu�

�cosα0

þ � ∼m0Nv� þ mprpx�m� Zw�

��sinα0

�sin2θ0

��S�Cnβ l

∼m0þmprpx�CD0 þ CSβ þ C2

LαK α20��ρ

þ 2 ∼m0��Xu� þ Yv�

�cosα0

�cosθ0 þ

�S��Clβ l

∼m0 þ�mrbrrbz þ mpRp

��CD0 þ CSβ þ C2

LαK α20��ρ

þ 2 ∼m0�Yv� � Zw�

�sinα0

�sinθ0

�� ð14Þ

φ1 ¼�1=�2g ∼m0

��V0ωnsecθ0

�2�mþ ∼m0� Xu�

�cosα0cosθ0 þ 2

�mþ ∼m0�Zw�

�sinα0sinθ0

� �S�CD0 þ CSβ þ C2LαK α20

�ρ��Cnr l


�m� Xu�

��cosα0

�cosθ20

þ�Clp l2 ∼m0Sρþ 2

�mrbrrbz þ mpRp

��m� Zw�

�sinα0

�sinθ20 þ


��m�Xu�

�cosα0

þ� ∼m0Nv� þmprpx�m�Zw�

��sinα0

�sin2θ0

��=��S�Cnβ l

∼m0þmprpx�CD0 þCSβ þC 2

LαK α20��ρ

þ 2 ∼m0��Xu� þ Yv�

�cosα0

�cosθ0 þ

�S��Clβ l

∼m0þ�mrbrrbz þ mpRp

��CD0 þ CSβ þ C2

LαK α20��ρ

þ 2 ∼m0�Yv� � Zw�

�sinα0

�sinθ0

�� ð15Þ

β1 ¼ ��ωn��Cnr l


�m� Xu�

��cosα0

�cosθ20

þ �Clp l2 ∼m0Sρþ 2

�mrbrrbz þ mpRp

��m� Zw�

�sinα0

�sinθ20 þ


��m� Xu�

�cosα0

þ � ∼m0Nv� þ mprpx�m� Zw�

��sinα0

�sin2θ0

��=�V0��S�Cnβ l

∼m0þ mprpx�CD0 þCSβ þC2

LαK α20��ρ

þ 2 ∼m0��Xu� þYv�

�cosα0

�cosθ0þ

�S��Clβ l

∼m0 þ�mrbrrbz þ mpRp

��CD0 þ CSβ þ C 2

LαK α20��ρ

þ 2 ∼m0�Yu� � Zw�

�sinα0

�sinθ0

�� ð16Þ

4.2. Parametric Performance AnalysisHere, we examine performance in steady wings-level and turning flight.

4.2.1. Performance Analysis of Wings-level FlightA vehicle’s geometry defines its performance. We first consider the effect of various geometric parameters on the speed at

minimum glide angle (also called the “speed to fly”) and the maximum lift-to-drag ratio. We also consider range, endurance,and number of dives, for a given propulsive energy budget. We focus on a few key parameters such as hull length and finenessratio, wing aspect ratio, the wingspan ratio, and also the buoyant lung capacity η .

Size Analysis. Because underwater gliders operate at conditions where the flow over the vehicle and its components maybe laminar or turbulent, it is important to keep track of the Reynolds number, which characterizes the flow over a body. Wedefine

Rel ¼ Vlν

and Rec ¼V cν

¼ Relcl¼ Rel

κfAR


Referring to equation (12), we may write

Re2l ¼�lν

�2 πgl2

ARκ2

�η

1þ η

�1

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC 2D þ C2

L

p

Note that the Reynolds number Rel affects the parasitic drag coefficient CD0 andthus appears on both sides of the equation above. Solving for η we find

η ¼ 1Φ� 1

where Φ ¼ 1Re2l

π2gl 3

ν2

ARκ2

1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC2D þ C2

L

p ð17Þ

For flight at the shallowest descent angle, the parameters CD, CL, and Rel are

CDmd ¼ 2CD0 ;CLmd ¼ffiffiffiffiffiffiffiffiCD0

K

r

and Relmd ¼Vmdlν

:

The maximum lift-to-drag ratio at this flight condition can be described as

LD

¼ CLmd

CDmd

¼ffiffiffiffiffiffiffiffiffiffiffiffiffi

14KCD0

s


Referring to (Hoerner, 1965; Stengel,2004), one finds that CD0 depends onReynolds number, so CD0 varies withspeed and length.

Figures 7 and 8 present surfaceplots of the variation in required lungcapacity and the maximum lift-to-dragratio for a glider with the followingparameter values

l ¼ 1:5m; AR ¼ 11; κ ¼ 6;

f ¼ 7; and ∼t ¼ 0:02

These parameter values are based on amodel of Slocum used in earlier studies(Graver, 2005; Bhatta, 2006).

Figure 7 shows the variation in lungcapacity required for flight of a 52-kgglider over a range of minimum glideangle speeds, given variations in one ofthe four geometric parameters l, AR,κ, or f. (In each plot, the nonvaryingparameters take the nominal valuesgiven earlier.) To reduce the lung ca-pacity required for a given speed at theminimum glide angle, one should in-crease the vehicle size (parameterizedby hull length l ) and fineness ratio f,and decrease the wingspan ratio κ. Therequired lung capacity does not varywith the wing aspect ratio AR.

Figure 8 shows the variation inmaximum lift-to-drag ratio of a 52-kgglider over a range of minimum glideangle speeds, given variations in oneof the four geometric parameters l,AR, κ, or f. (Again, the nonvaryingparameters take the nominal valuesgiven earlier.) The apparent disconti-nuities occur when the flow over thewings transitions between laminar andturbulent. To increase the lift-to-dragratio for a given speed to fly within agivenflow regime (laminar or turbulent),one should increase the vehicle size(i.e., the hull length l ) and the wing-span ratio κ. The lift-to-drag ratio ap-pears less sensitive to changes in hull

FIGURE 7

Variation of the required lung capacity with the given minimum glide angle speed and size.(a) Variation of lung capacity with Vmd and l. (b) Variation of lung capacity with Vmd and AR.(c) Variation of lung capacity with Vmd and κ . (d) Variation of lung capacity with Vmd and f.

fineness ratio or wing aspect ratio,within a given flow regime.

Figure 9 shows the variation indisplaced mass (for neutral buoyancy)with length, for the nondimensionalparameter values above.

Range and Endurance. Referring toFigure 4 and considering saw-toothprofiles to a depth Δz at the minimumglide angle flight condition, the per-diverange is approximately

Δx ¼ 2CLmd

CDmd

Δz ¼ ΔzffiffiffiffiffiffiffiffiffiffiffiKCD0

p

Note that we are ignoring the effects oftransition between gliding flight con-ditions, on the assumption that suchtransitions can be effected quickly com-pared with the total dive time and

depth. The energy required to executea single saw-tooth dive is given (ap-proximately) by the pressure-volumework required at depth. Assume that

May/J

the vehicle displacement varies fromminimum to maximum

2ηVNB ¼ 2

�η

1þ η

�

V

Correspondingly, the piston of thebuoyancy actuator is moved from itsmaximally retracted position to its max-imally extended position. The pistonmust counter a hydrostatic pressureρgΔz acting on the piston face, whichhas area πd2/4. Thus, we find that

ΔE ¼ π2

�ρgΔz

��

η1þ η

��l3

f 2

�

In keeping with other simplifying as-sumptions, we ignore the variation inpressure within the dry volume due tothe piston displacements. The time re-quired to execute this dive, assuming thesymmetrical flight condition in ascent, is

ΔT ¼ 2Δzz�

where, recalling equations (10) and (13),

z� ¼ Vmd sin ξmd

¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

πgl2

ARκ2

�η

1þ η

�s

CDmd�C2Dmd

þ C2Lmd

�3=4

Given an amount of stored energyEp allocated for propulsion, the total

FIGURE 9

Neutral displacement versus length, at f = 7.

FIGURE 8

Variation of maximum lift-drag ratio with the given minimum glide angle speed and size. (a) Vari-ation of maximum lift-to-drag ratio with Vmd and l. (b) Variation of maximum lift-to-drag ratio withVmd and AR. (c) Variation of maximum lift-to-drag ratio with Vmd and κ . (d) Variation of maximumlift-to-drag ratio with Vmd and f.


number of dives that can be executed todepth Δz is

N ¼ f loor

�EpΔE

�

where “floor” indicates the greatestlower integer. The total range that canbe traversed (in still water) is

R ¼ NΔx

and the total endurance is

E ¼ NΔT :

Figure 10 shows contour plots indi-cating range and endurance and num-ber of dives for a glider of given sizeand geometry when repeatedly divingto a given depth at a given nominalspeed. In these plots, the vehicle lengthis 1.5 m, which corresponds to a massof 52 kg. The total power available


for propulsion is 540 kJ, equivalentto a single 5 Ah, 30 VDC battery.In generating Figure 10, we assumedflight at the minimum glide anglespeed with the net weight (or buoy-ancy) fixed at the maximum value, asdetermined from equation (17).

As indicated in Figure 10(a), thenumber of dives that can be executedinvolves a tradeoff between nomi-nal speed and dive depth. Range andendurance diminish nonlinearly withincreasing nominal speed but are inde-pendent of dive depth, a consequenceof our assumptions that hydrostaticpressure is linear in depth (i.e., thewater density is constant) and thatpump efficiency does not vary withdepth. Deep diving gliders typicallymodulate their buoyancy using apump-driven, oil-filled bladder whoseefficiency varies with ambient pressure.

l

Moreover, for deep dives, variation inwater density is significant (Osse &Eriksen, 2007); ocean stratificationand hull compressibility have a markedeffect on glider performance (Jenkinset al., 2003). Considering these andother factors, such as the time spentin transition between equilibrium flightconditions, would alter the range andendurance contours, but the qualitativetrends would persist.

4.2.2. Performance Analysisof Turning Motion

Based on the approximate expres-sions of turning motion presented inSection 4.1.2, we examine how thewing geometry (sweep angle, dihedralangle, and vertical placement) andthe vertical stabilizer geometry (locationlv and area Sv) affect turning capabil-ity here. Variations in the wing sweepangle, dihedral angle, and vertical place-ment affect the roll moment due tosideslip, as expressed by the hydrody-namic coefficient Clβ (Etkin & Reid,1996). Therefore, we vary the coef-ficient Clβ to investigate the effect ofwing configuration parameters on turn-ing motion. Note that the vertical sta-bilizer parameters lv and Sv do notappear explicitly in equations (14), (15),and (16). However, they do affect thehydrodynamic coefficients, such asCSβ ,Cnr and Cnβ (Nelson, 1998) and thusinfluence turning performance.

Recalling that turning motions areparameterized by the moving mass ro-tation angle (μ), the glider roll angle(φ), and the sideslip angle (β), we ex-amine the variation in the first-ordersensitivities μ1, φ1, and β1 with lv, Sv,and Clβ for a given turn rate. Recallthat approximate values of μ, φ, andβ may be obtained by multiplying theparameters μ1, φ1, and β1 by the turnrate. In Figures 12(a), 11(b), and 11(c),which were generated using expressions

FIGURE 10

Contours of range (in km), endurance (in days), and number of dives for a 540-kJ glider. (a) Divecontours. (b) Range contours (km). (c) Endurance contours (days).

(14), (15), and (16), we see that when lvor Sv (or their product, the “vertical sta-bilizer volume”) is small and/or the di-hedral parameter Clβ is less negative, asmaller moving mass rotation μ androll angle φ are required to establisha given steady turn rate. Correspond-ingly, however, the vehicle’s steady turnrate is more sensitive to errors in theroll angle. While Figure 11 describesthe effect of wing and vertical stabilizergeometry on the steady turning flightcondition, it says nothing of stability,which is the topic of Section 5.

5. Stability AnalysisA glider’s geometry affects the hy-

drodynamic coefficients that appearin stability analysis. In this section,we consider the natural modes of glidermotion and use root locus analysis to

examine the effects of a few criticalparameters on glider stability.

5.1. Eigenmode AnalysisConsidering a glider with fixed ac-

tuators as a rigid body, we linearizethe dynamic equations about a wings-level equilibrium condition. Specifically,we consider the case of steady flight atmaximum horizontal speed. Given adynamic model, one may obtain anequilibrium flight condition with thehelp of a numerical trim solver. In ourcase, using Matlab’s f solve subroutine,we found the following equilibrium:

Veq ¼ 0:77m=s; ∼meq ¼ 0:2130 kg;

γeq ¼ �35°; rpxeq ¼ 0:2826m and

αeq ¼ 1:2°

May/J

Linearizing the dynamic equations aboutthe above wings-level equilibrium, onefinds that the resulting equations de-compose into longitudinal and lateral-directional components. Ignoring certainkinematic variables, one obtains twosets of four, first-order equations,

X�long ¼ AlongXlong þ Blongulong and

X�lat ¼ AlatXlat þ Blatulat

where

Xlong ¼ Δu;Δw;Δq;Δθ� �T

and

Xlat ¼ Δv;Δp;Δr;Δφ½ �T

and where the elements of the stateand input matrices depend on thesteady flight condition and on theglider geometry, through the stabilityderivatives. Eigenvalues λ and non-dimensional eigenvectors ν (in am-plitude and phase form) for the statematrices Along and Alat are listed inTables 1 and 2.

For the maximum horizontal-speed flight condition consideredhere, the longitudinal eigenvaluesshown in Table 1 include two realeigenvalues and a complex conjugatepair. Examination of the eigenvec-tors associated with these eigenvaluesindicates the following character-istic modes:■ a quickly converging pitch rate

mode,■ a slowly converging speed mode,

and■ an underdamped mode in which

the pitch angle is coupled with theangle of attack.
The lateral-directional eigenvalues
shown in Table 2 also include tworeal eigenvalues and a complex conju-gate pair. Examination of the eigenvec-tors associated with these eigenvalues

FIGURE 11

Variation in turning parameters μ1, ϕ1 and β1 with lv, Sv and Clβ . (a) Variation of turning parameterswith lv. (b) Variation of turning parameters with Sv. (c) Variation of turning parameters with Clβ.


indicates the following characteristicmodes:■ a slowly converging, coupled roll

and yaw rate mode,■ a slowly converging mode involv-

ing roll angle and side velocity, and■ an underdampedmode in which the

roll rate and yaw rate are stronglycoupled.

5.2. Glider Stability VariedWith Geometry

Here, we investigate the effect ofspecific geometric parameters on theeigenvalue distribution. We considertwo specific flight conditions: maxi-mum horizontal speed and minimumglide angle. We present root locusplots for the longitudinal modes, interms of the longitudinal wing loca-tion lw, and for the lateral-directionalmodes, in terms of vertical stabilizerlocation lv and area Sv, as well the dihe-dral parameter Clβ (which accountsfor the effect of wing sweep, dihedralangle, and vertical placement). Thevariation of parameters lw, lv, and Sv


changes the hydrodynamic coeffi-cients, which affects the eigenmodesof the glider dynamics. There are sev-eral methods to obtain hydrodynamiccoefficients, including analytical,experimental, and computational ap-proaches (Geisbert, 2007). For anal-ysis of general trends, as describedhere, a semiempirical approach is suf-ficient (Nelson, 1998).

Again, the physical and hydro-dynamic characteristics used in this

l

analysis are based on the Slocummodel used in earlier studies (Graver,2005; Bhatta, 2006). Stability ofsteady motion is obviously affectedby vehicle speed; in order to makereasonable comparisons between dif-ferent flight conditions, we fix the flightspeed at V = 0.77 m/s, as in previousstudies.

Figure 12 shows root locus plots forlongitudinal modes in wings-levelflight at maximum speed (left) and at

TABLE 1

Eigenvalues and eigenvectors of longitudinal motion.

Longitudinal
λ1 ¼ �3:28 λ3 ¼ �0:85þ 0:27i λ3 ¼ �0:85� 0:27i λ4 ¼ �0:06
Δu
ν11 ¼ 0:0315∠0° ν21 ¼ 0:0661∠�126° ν31 ¼ 0:0661∠126° ν41 ¼ 1∠0°
Δw
ν12 ¼ 0:0631∠0° ν22 ¼ 0:6694∠0° ν32 ¼ 0:6694∠0° ν42 ¼ 0:0231∠180°
Δq
ν13 ¼ 1∠180° ν23 ¼ 1∠84:5° ν33 ¼ 1∠� 84:5° ν43 ¼ 0:0005∠180°
Δθ
ν14 ¼ 0:1306∠0° ν24 ¼ 0:4793∠�78° ν34 ¼ 0:4793∠78° ν44 ¼ 0:0038∠0°
TABLE 2

Eigenvalues and eigenvectors of lateral-directional motion.

Lateral-directional
λ1 ¼ �1:33þ 1:75i λ2 ¼ �1:33� 1:75i λ3 ¼ �1:81 λ4 ¼ �0:17
Δv
ν11 ¼ 0:0363∠� 131° ν21 ¼ 0:0363∠131° ν31 ¼ 0:1313∠0° ν41 ¼ 0:7647∠0°
Δp
ν12 ¼ 1∠0° ν22 ¼ 1∠0° ν32 ¼ 0:9125∠0° ν42 ¼ 0:4∠0°
Δr
ν13 ¼ 0:5026∠150° ν23 ¼ 0:5026∠�150° ν33 ¼ 1∠0° ν43 ¼ 0:0235∠180°
Δϕ
ν14 ¼ 0:1969∠� 127° ν24 ¼ 0:1969∠127° ν34 ¼ 0:2125∠180° ν44 ¼ 1∠180°
FIGURE 12

Root locus plots for longitudinal modes with the parameter lw. (Root locus branches begin at redcircles and end at blue squares.) (a) Maximum speed. (b) Minimum glide angle.

minimum glide angle (right) in termsof the parameter lw. In these plots,lw varies from zero (in which casethe wing is aligned with the centerof buoyancy) to 0.2 l. Note that thefarther aft the wing is located, themore stable the glider longitudinaldynamics become, due in part to theincreased pitch damping.

Figure 13 shows root locus plots forthe lateral-directional modes in wings-level flight at maximum speed (left)and at minimum glide angle (right)in terms of the parameter lv. In theseplots, lv varies from 0.5 l to l, that is,from the stern of the hull to a half-vehicle length aft of the stern. Thefarther aft the vertical stabilizer islocated, the more stable the gliderlateral-directional dynamics become,due to increasing yaw stiffness anddamping. Figure 14 shows root locusplots for the lateral-directional modesin wings-level flight at maximumspeed and at minimum glide anglein terms of the parameter Sv, whichvaries from half the hull frontal area to1.2 times the hull frontal area. Notethat a larger vertical stabilizer areayields more stable lateral-directionalmodes, by increasing yaw stiffnessand damping. Figure 15 shows rootlocus plots for the lateral-directionalmodes in wings-level flight at maximumspeed and at minimum glide angle interms of the parameter Clβ , which var-ies from “more negative” (−0.8) to “lessnegative” (−0.2). Note that larger neg-ative values of this parameter (corre-sponding to wings with larger sweepangle, larger dihedral angle, or highervertical placement) yield more stablelateral-directional dynamics. Of course,greater stability implies that greatercontrol authority is required to effecta maneuver, such as a turn. Tradeoffsbetween control authority and stabil-ity are fundamental in vehicle design

FIGURE 13

Root loci for lateral-directional modes with parameter lv. (Root locus branches begin at red circlesand end at blue squares.) (a) Maximum speed. (b) Minimum glide angle.

FIGURE 14

Root loci for lateral-directional modes with parameter Sv. (Root locus branches begin at red circlesand end at blue squares.) (a) Maximum speed. (b) Minimum glide angle.

FIGURE 15

Root loci for lateral-directional modes with parameter Clβ . (Root locus branches begin at redcircles and end at blue squares.) (a) Maximum speed. (b) Minimum glide angle.


and they suggest the need for guide-lines. For manned vehicles, such guide-lines are obtained based on crew andpassenger comfort, but for unmannedvehicles they must be derived fromother considerations such as payloadmotion.

6. ConclusionConsidering a conventional under-

water glider configuration comprisinga cylindrical hull, a simple trapezoidalwing and a vertical stabilizer, we haveprovided analysis to describe the effectof a few important geometric param-eters on steady wings-level flight andsteady turning flight. Starting froman 8-degree-of-freedom model thatincorporates a cylindrically actuatedmoving point mass, we consideredtwo important wings-level gliding con-ditions: minimum glide angle andmaximum horizontal speed. We alsodeveloped approximate analytical ex-pressions for steady turning motion interms of the cylindrical actuationscheme and investigated the relationshipbetween glider geometric parametersand the vehicle’s performance and sta-bility characteristics. The geometricparameters characterized the slender-ness of the hull, the position andshape of the wing, and the size and po-sition of the vertical stabilizer. Wefound, for example, that for a gliderof given mass and buoyant lung capac-ity, higher speeds are attainable usinglonger hull lengths, smaller wingspans, larger hull fineness ratios, andhigher wing aspect ratios. To maxi-mize the lift-drag ratio at a given min-imum glide angle speed, on the otherhand, one should decrease the hulllength and increase the wingspanratio. Adapting a regular perturba-tion approach to develop analytical ex-pressions for steady turning flight, we


found that a smaller roll angle is required to effect a steady turn at a given ratewhen the vertical tail volume and/or the dihedral effect is smaller. We also pro-vided simple methods to estimate range and endurance; a more sophisticatedanalysis would incorporate details of the buoyancy lung mechanization as wellas the variation of water density and hull compressibility.Turning to stability,we used root locus analysis to examine the variation in longitudinal and lateral-directional eigenvalues with changes in wing location, wing shape, and vertical sta-bilizer size and location. We found that the farther aft the wing is located, the morestable the glider longitudinal dynamics become, due in part to the increased pitchdamping. The farther aft the vertical stabilizer is located and/or the larger the verticalstabilizer area is, the more stable the glider lateral-directional dynamics become, dueto the increasing yaw stiffness and damping. Also, a larger dihedral effect yieldsmore stable lateral-directional modes; dihedral effect can be increased by increas-ing sweep or dihedral angle and by placing the wings higher on the hull. Increasedstability may provide better response to disturbances, but may also limit controlauthority, a tradeoff that must be considered in vehicle design.

AcknowledgmentsThe authors gratefully acknowledge the
sponsorship of theOffice of Naval Research
under Grant #N00014-08-1-0012. The first author would like to thank the ChinaScholarship Council (CSC) for financial support during her visit to Virginia Tech.

Corresponding Author:Shuangshuang Fan
The State Key Lab of Fluid Power Transmission and Control,Zhejiang University, Hangzhou, China Email: [email protected]
Appendix A. Body/Particle/Fluid System Energy
To derive the equations of motion for an underwater glider with amovingmass
particle, one first determines the kinetic energy of the body/particle/fluid systemin order to compute the momenta. The particle kinetic energy is

Tp ¼ 12mpν

Tp νp ¼ 1

2ηTMp

�rp�η

where

Mp

�rp� ¼ mp

IrpeT1eT1 rp

0

BBB@

1

CCCA

IrpeT1eT1 rp

0

BBB@

1

CCCA

T

¼

mpI �mprp mpe1 �mprpe1

mprp �mprprp mprpe1 �mprprpe1

mpeT1 �mpeT1 rp mp 0

mpeT1 rp �mpeT1 rprp 0 �mpeT1 rprpe1

0

BBBBBB@

1

CCCCCCA

Here and elsewhere in the text, 0 represents a matrix of zeros whose dimensionsare clear from context.

l

The kinetic energy Tb of the massparticle representing the buoyancycontrol system is defined similarly.The generalized mass matrix is

Mb ¼mbI �mbrb 0

mbrb �mbrbrb 0

0 0 0

0

BB@

1

CCA

Where rb denotes the location ofthe mass particle in the body frame.Since we have assumed the buoyancycontrol system is at the body frameorigin, however, rb = [0, 0, 0]T.

The kinetic energy of the rigidbody portion of the glider is also qua-dratic in η, with the generalized massmatrix

Mrb ¼mrbI �mrbrrb 0

mrbrrb J rb 0

0 0 0

0

BB@

1

CCA

where Jrb is the matrix of rigid bodymoments of inertia.

Finally, define the generalized addedmass matrix

Mf ¼M f CT

f 0Cf J f 00 0 0

0

@

1

A

where the component submatrices Jf,Mf, and Cf represent added inertia,added mass, and hydrodynamic cou-pling between the translational and ro-tational motion of the rigid body.

The total kinetic energy of the rigidbody/particle/fluid system is

T ¼Tp þ Tb þ Trb þ Tf ¼ 12ηTM

�rp�η

where

M�rp� ¼ Mp

�rp�þMb þMrb þMf :

ReferencesBhatta, P. 2006. Nonlinear stability and

control of gliding vehicles. Ph.D. thesis,

Princeton University. 206 pp.

Bhatta, P., & Leonard, N.E. 2008. Nonlinear

gliding stability and control for vehicles

with hydrodynamic forcing. Automatica.

44(5):1240-50. http://dx.doi.org/10.1016/

j.automatica.2007.10.006.

Curtin, T.B., Bellingham, J.G., Catipovic, J.,

& Webb, D. 1993. Autonomous oceano-

graphic sampling networks. Oceanography.

6(3):86-94. http://dx.doi.org/10.5670/

oceanog.1993.03.

Eriksen, C.C., Osse, T.J., Light, R.D.,Wen, T.,

Lehman, T.W., Sabin, P.L.,… Chiodi, A.M.

2001. Seaglider: A long-range autonomous

underwater vehicle for oceanographic research.

IEEE J Oceanic Eng. 26(4):424-36. http://

dx.doi.org/10.1109/48.972073.

Etkin, B., & Reid, L.D. 1996. Dynamics of

Flight: Stability and Control. New York:

John Wiley and Sons. 400 pp.

Galea, A.M. 1999. Optimal path planning

and high level control of an autonomous

gliding underwater vehicle. Master thesis,

Massachusetts Institute of Technology.

67 pp.

Graver, J., Liu, J., Woolsey, C.A., & Leonard,

N.E. 1998. Design and analysis of an under-

water vehicle for controlled gliding. In: Proc

CISS. pp. 801-6. Princeton, NJ: IEEE.

Graver, J.G. 2005. Underwater gliders:

Dynamics, control and design. Ph.D. thesis,

Princeton University. 273 pp.

Geisbert, J.S. 2007. Hydrodynamic modeling

for autonomous underwater vehicle using

computational and semi-empirical methods.

Master thesis, Virginia Tech. 87 pp.

Hoerner, S.F. 1965. Fluid Dynamic Drag.

Midland Park, NJ: Author. 452 pp.

Jenkins, S.A., Humphreys, D.E., Sherman, J.,

Osse, J., Jones, C., Leonard, N.E.,…,Wasyl, J.

2003.Underwater glider system study. Technical

Report 53, Office of Naval Research (ONR).

240 pp.

May/J

Lanchester, F.W. 1908. Aerodonetics.

London: A. Constable and Co. 468 pp.

Mahmoudian, N., Geisbert, J., & Woolsey,

C.A. 2010. Approximate analytical turning

conditions for underwater gliders and impli-

cations for path planning. IEEE J Oceanic

Eng. 35(1):131-43. http://dx.doi.org/10.1109/

JOE.2009.2039655.

Mahmoudian, N., & Woolsey, C.A. 2013.

An efficient motion control system for under-

water gliders. Nonlinear Engineering: Modeling

and Application, 2013. (To appear.)

Mises, R.V. 1959. Theory of Flight.

New York: Dover. 672 pp.

Nelson, R.C. 1998. Flight Stability and Auto-

matic Control. New York: McGraw-Hill

Higher Education. 441 pp.

Osse, T.J., & Eriksen, C.C. 2007. The

deepglider: A full ocean depth glider for

oceanographic research. In: Proc MTS/IEEE

OCEANS. pp. 1-12. Vancouver, BC: IEEE.

Pamadi, B.N. 2004. Performance, Stability,

Dynamics and Control of Airplanes. Reston,

VA: AIAA. 780 pp. http://dx.doi.org/10.2514/

4.862274.

Rudnick, D.L., Davis, R.E., Eriksen, C.C.,

Fratantoni, D.M., & Perry, M.J. 2004.

Underwater glider for ocean research.


Sherman, J., Davis, R.E., Owens, W.B., &

Valdes, J. 2001. The autonomous under-

water glider “Spray”. IEEE J Oceanic Eng.

26(4):437-46. http://dx.doi.org/10.1109/

48.972076.

Stengel, R.F. 2004. FlightDynamics. Princeton,

NJ: Princeton University Press. 845 pp.

Teledyne Webb Research. 2013. Electric

glider. http://www.webbresearch.com/

electricglider.aspx, April 2013.

USAF Stability and Control Datcom. 1978.

Flight Control Division, Air Force Flight

Dynamics Laboratory. Fairborn, OH:

Wright Patterson Air Force Base.

Webb, D.C., Simonetti, P.J., & Jones, C.P.

2001. SLOCUM: An underwater glider


propelled by environmental energy. IEEE J

Oceanic Eng. 26(4):447-52. http://dx.doi.org/

10.1109/48.972077.

Wolek, A., Burns, J.,Woolsey, C.A., Techy, L.,

Quenzer, J., & Morgansen, K. 2012. A

maneuverable, pneumatic underwater glider.

In: Proc MTS/IEEEOCEANS. October 14–19,

2012. Virginia Beach, VA: IEEE.


P A P E R

Analysis of Underwater AcousticCommunication Channels
A U T H O R SSadia AhmedHuseyin ArslanDepartment of ElectricalEngineering, University ofSouth Florida
A B S T R A C T
The underwater acoustic communication (UAC) channel presents many difficul-
ties such as high frequency, space, and time selectivity, frequency-dependentnoise, and significant range and band limitation on transmission. Traditional UACchannel models that model such channels primarily include environmental modelsbased on experimental data; models that are developed using mathematical equa-tions such as wave equations, modal methods, and parabolic equations; and usingstatistical distributions. These methods/models are often limited in their coverageand accurate representations of every possible UAC channel environment. It is alsophysically impractical and cost ineffective to try to measure/estimate each channelto determine its model. In this paper, the authors will present the analysis of UACchannels according to the UAC channel environments classified and presented ina prior work by the authors, in which cognitive intelligence is used in the selectionof the appropriate channel representations according to each sensed environment.To the best knowledge of the authors, this type of analysis and representation ofUAC channels with respect to each UAC environment has not been addressed in theliterature to date and therefore presents a significant contribution.Keywords: channel representation, deep channel, shallow channel, underwaterchannel environment

quires representing of such channelaccording to its channel environment,

Introduction

The underwater acoustic commu-nication (UAC) channel is vast andvarying in nature. The existing UACchannel models may not be sufficientto represent every single UAC chan-nel scenario. Moreover, the choice ofmodels from the existing ones maynot be accurate to represent the truechannel environment. Thus, the vary-ing nature of the UAC channel re-

which in turn is specified by sets ofchannel parameters. If the channelparameters and hence the channel en-vironment is accurately determined,the environment can be mapped ontothe appropriate UAC channel repre-sentation (Ahmed & Arslan, 2009a,2009b). The same channel represen-tations can be reapplied in same orsimilar environments. The parametervalues can be adaptively measured/estimated or derived according to theUAC environment. Therefore, the rela-tionship between UAC channel repre-sentations and the UAC environmentsallows wider representation and cover-age of the varying UAC channels.A high level classification of the UACchannel environments according toliterature is presented in Figure 1.

The contributions of this paper areas follows:■ This paper presents an analysis of

UAC channel impulse responseaccording to each broadly catego-rized UAC channel environment(Figure 1).

■ Fading characteristics of the re-ceived signal and signal-to-noiseratio (SNR) of UAC channel ineach channel environment andthe relationship between envi-ronment parameters and differ-ent types of fading are analyzedand discussed.

■ The channel representations aregenerated using the AcTUP soft-ware module.
The rest of the paper is organized
as follows. Related Works presents

May/J

a brief literature review of UACchannel models. Mapping of UACChannel Parameters to UAC ChannelEnvironments discusses the mappingof UAC channel environment param-eters onto UAC channel environments.UAC Channel Representation Accord-ing to Environment presents the anal-ysis of channel impulse response andfading characteristics of UAC channelsaccording to channel environments.SNR Analysis provides the SNR analy-sis of the channel representations. Ben-efits of Channel Classification andSelection of Channels briefly presentsthe benefits of channel selection andthe channel selection methods. Simula-tion Results presents the simulation re-sults, and Conclusion addresses theconcluding remarks.


Related WorksThe UAC channels are presented

using stochastic models in some litera-ture. In Galvin and Coats (1996), theamplitude and phase fluctuation ofreceived signal is presented first by lin-ear and then nonlinear transformationsof Gaussian variables. In Morozov(1995), the estimation of slow phasefluctuations of channel impulse re-sponse is described asMarkov’s process,and the accumulation of impulse chan-nel responses is decided by the Viterbialgorithm. In Byun et al. (2007), a chan-nel model is proposed to present thetime-varying UAC channel, where thetime variation is produced by trans-

100 Marine Technology Society Journ

mitter and receiver motion. In thismodel, the eigen rays correspondingto direct/reflected multipaths are de-termined using ray tracing while therandom diffusive multipaths are mod-eled using Rayleigh fading.

In the current paper, the randomfluctuation due to diffusive paths arerepresented as Rayleigh or Rician vari-ables according to the environment.

In Appleby and Davies (1998),time and frequency dispersion ofUAC channel is represented by amodel, which encompasses two sepa-rate modeling stages. First, the envi-ronment model is developed usingthe environmental data such as windspeed, sound speed profile, bottom

al

loss characteristics and incorporatesfeatures such as surface wave model,internal wave model, tidal currentsmodel, two-layer bottom interactions,and ambient noise. Second, 3-D raytracing is used to develop the propaga-tion model that includes effects such asrange dependent bathymetry, branch-ing at the water/sediment interface,propagation through the sedimentlayer, range-dependent sound speed,caustics, reflection from the movingsea surface, and motion of transmitter/receiver platforms.

In the current paper, the effect ofenvironment parameters and the prop-agation characteristics are incorpo-rated in the deterministic and stochasticcomponents of the arrived paths.

In Geng and Zielinski (1995), therandom fluctuation of dominanteigen paths between transmitter andreceiver is represented by the presenceof smaller sub-eigen paths. The com-bination of eigen and sub-eigen pathsis represented by Rician fading. InBjerrum-Niese and Lutzen (2000),signal-to-multipath ratio and signalfading statistics is presented. Turbu-lence of water causes change in signalphase and amplitude. Ray tracing isused to determine the deterministiccomponent and phase and amplitudevariation is placed on top of the deter-ministic components.

In the current paper, the determin-istic eigen components are first derivedusing Ray tracing and then the randomfluctuation causing sub-eigen compo-nents are placed on top of the determin-istic components. The combined effectof eigen and sub-eigen is represented asGaussian, Rayleigh, or Rician variables.

In Walree et a l . (2008) andSocheleau et al. (2011), data are mea-sured and the measured data are inter-polated to determine channel impulseresponse (CIR). Both uncorrelated and

FIGURE 1

Identification and high-level classification of UAC channel environments.

correlated taps are considered. Pseudo-random binary signal is used as probesignal in the measurements.

Mapping of UAC ChannelParameters to UACChannel Environments

The cognitive intelligence (CI) al-gorithms to map UAC channel param-eters to environments and to mapUAC environments to channel modelshave been presented in Ahmed andArslan (2009a, 2009b), respectively.

All possible UAC channel environ-ments may be determined from thesensed channel environment parame-ters. The channel representations, oncemapped from the environments, willbe more accurate since they are derivedfrom the actual channel environments.

UAC ChannelRepresentation Accordingto Environment

In the proposed channel analysis,the following assumptions are made.Channel Assumptions■ The channel equations are derived

in the time domain. The transmis-sion frequency and bandwidth areconsidered as parameters definingthe UAC environment (Ahmed &Arslan, 2009a).

■ Power in terms of path loss in dBis considered for each path in thepower delay profile.

■ Time variation of channel is appliedon top of the power delay profile.

■ The individual scatters are assumedto be independent and identicallydistributed (i.i.d.) Gaussian randomvariables in the case of uncorrelatedtaps.

■ Each dominant and its correspond-ing sub-eigen paths arrive at thereceiver at the same time.

■ Line-of-sight (LOS) path is con-sidered to be present in each UACenvironment.

The rest of this section is organized asfollows. Channel Impulse Responseof UAC Channels presents the CIRand Fading Characteristics, ProbabilityDistribution, Relationship BetweenFading and Channel Parameters forWide Sense Stationary (WSS) Un-correlated Tap Channel presents thefading characteristics of different UACchannels according to each UACenvironment. Frequency Dependentand Independent Noise briefly toucheson the noise components and FadingCharacteristics, Distribution, Rela-tionship Between Fading and Chan-nel Parameters for Quasi-stationaryChannels on quasistationarity of UACchannels.

Channel Impulse Responseof UAC Channels

Let the transmitted signal be denotedby an impulse, δ(t), where δ represents aDirac delta function. The UAC channelwill produce multipaths that may varyin time. In a time invariant channel,the channel impulse response can be ex-pressed as h(τREF)=

PL�1l¼0 αlδ(τREF −τl).

The variables αl and τl denote theattenuation coefficient and the timedelay of the lth path respectively. Thevariable L denotes the total number ofmultipaths, τREF is a time reference.In a time variant channel, this equa-tion can be modified as below,

h τREF ; tð Þ ¼XL�1

l¼0

αl tð Þδ τREF � τl tð Þð Þ:

ð1Þ

In an UAC channel, group of multi-paths may consist of dominant paths,for examples, originating from reflec-tions on the boundaries, also knownas eigen paths (Gutierrez et al., 2005;

May/Ju

Geng&Zielinski, 1995) accompaniedby less dominant paths or sub-eigenpaths, for example, originating fromvarious scatters.

Shallow Channel RepresentationAccording to Environment

A generic shallow water channel isillustrated in Figure 2. The shallowwater UAC channel multipaths canbe grouped according to their origina-tion. In a shallow UAC channel envi-ronment, there may be five groups ofeigen and sub-eigen paths. The firstgroup of paths may result from surfaceonly reflection and scatter, while thethird group may result from bottomonly reflection, bottom refraction,and scatter. The second group may in-clude paths that reflect on both surfaceand bottom but the first reflection ison the surface. The fourth group mayalso include paths reflecting on bothsurface and bottom but the first reflec-tion is on the bottom. There may bescatters associated with these pathsas well. Depending on the transmis-sion range and depth, sound refrac-tion through water layers may result in

FIGURE 2

Underwater acoustic shallow water channel.

ne 2013 Volume 47 Number 3 101

another group of eigen and sub-eigen paths. This fifth group may also includepaths resulting from reflection, refraction, and scatter on marine life and otherobjects in the water medium. In addition, there may be a direct LOS that maydepend on transmission range and depth. Equation (1) can be expanded to repre-sent channel impulse response for all these five groups of paths as below.

hðtÞ ¼ α0ðtÞδðtÞ þXII

i¼1

�αiðtÞδðt � τiðtÞÞ

�

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP1

þXI

ii¼IIþ1

�αiiðtÞδðt � τiiðtÞÞ

�

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP2

þXJJ

j¼Iþ1

�αjðtÞδ

�t � τjðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP3

þXJ

jj¼JJþ1

�αjjðtÞδ

�t � τjjðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP4

þXL�1

k¼Jþ1

�αkðtÞδðt � τkðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP5

: ð2Þ

Table 1 presents examples of dominant paths in these groups.Each group of (2) can be expanded into many eigen paths and their corre-

sponding sub-eigen paths as in (3). Each eigen and its sub-eigen paths arrive at thesame time at each delay.

GROUP1 ¼"

α1ðtÞδ�t � τ1

�þXM1�1

m1¼2

α1m1ðtÞδ�t � τ1

�#

þ"

α2ðtÞδ�t � τ2

�þXM2�1

m2¼2


�#

:::

:::þ"

αII ðtÞδ�t � τII

�

|fflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflffl}

Eigenpath

þXMII�1

mII¼2

αIImII ðtÞδ�t � τII

�

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Sub�eigenpaths

#

:

ð3Þ

Figure 3 illustrates one single dominant path originating in each group along withless dominant paths due to scatters.

In an ideal scenario, with a smooth surface and a smooth bottom boundary,the channel generated multipaths can be described as arriving from image sources(Lurton, 1996). Under this scenario, the range traversed by each path from sourceto destination denoted byRangewv can be expressed in terms of the distance between


source and destination denoted by D,source and destination depth from sur-face denoted by zs and zr respectively,and water column height between sur-face and bottom denoted by H. Thereare four such image sources and aregiven by Lurton (1996).

z1v ¼ 2ðv � 1ÞH þ zs � zrz2v ¼ 2ðv � 1ÞH þ zs þ zrz3v ¼ 2vH � zs � zrz4v ¼ 2vH � zs þ zr :

ð4Þ

Then the time delay of each multipathat the destination can be expressed as

τwv ¼ Rangewvc

; ð5Þ

where w = 1, 2, 3, 4 denotes the imagesource types and c denotes the soundvelocity. The values of v can be v = 1,2, 3…. If only path loss is consideredand the sound velocity is consideredconstant (the value is provided in sim-ulation), each multipath from indi-vidual image source will go throughpath loss given by Lurton (1996),

TLwv ¼ 20logRangewv þ βRangewv; ð6Þ

where β is the water absorption coeffi-cient with units of dB/Km. In this idealscenario, in the absence of scatters, each

FIGURE 3

Single path originated from each group in UACshallowchannel. Solid line represents eigenpaths,and dashed line represents sub-eigen paths.

TABLE 1

Shallow water channel: The five groups and their respective paths.

Group
Paths in the Group
GROUP1

GROUP2

GROUP3

GROUP4

GROUP5

path arriving from each image source will only consist of a dominant path. The amplitude of each path derived from (6) can beexpressed as

Awv ¼ 10TLwv10 : ð7Þ

In case of a smooth surface and a smooth bottom and in the absence of sub-eigen paths, there will only be dominant paths; theCIR will consist of one LOS, one surface reflected path (GROUP1), one bottom reflected path (GROUP3), multiple paths inGROUP2, and multiple reflected paths in GROUP4. Therefore, in the absence of sub-eigen paths, (2) becomes

hðtÞ ¼ α0ðtÞδðtÞ þ�α1ðtÞδ

�t � τ1

��

|fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP1

þXI

ii¼IIþ1

�αiiðtÞδ

�t � τiiðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP2

þ �αJJ ðtÞδ

�t � τJJ ðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP3

þXJ

jj¼JJþ1

�αjjðtÞδ

�t � τjjðtÞ

��


GROUP4

þXL�1

k¼Jþ1

�αkðtÞδ

�t � τkðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP5

ð8Þ

The above ideal scenarios generate deterministic path components.Now, let the smooth surface and smooth bottom be replaced by surface waves and a rough bottom, respectively. Let the

following conditions be assumed: (a) the transmitter and the receiver are stationary; (b) the water column height (verticaldistance between surface and bottom) remain the same for both transmitter and receiver; (c) the acoustic wave length ismuch larger than the wave surface and bottom reflector elements; (d) the dominant paths arrive at the receiver; and (e) thedelays of dominant paths are calculated using image sources, although the surface movement due to surface waves and thepresence of rough bottom may vary these dominant path delays.

The scatters induced by surface waves and rough bottommay produce sub-eigen paths that can be represented as stochasticpath components along with each dominant path. Then (8) becomes

hðtÞ ¼ α0ðtÞδðtÞ þ�α1ðtÞδ

�t � τ1

�þXM1�1

m1¼2


��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP1

þXI

ii¼IIþ1

�αiiðtÞδ

�t � τiiðtÞ

�þXMMii�1

mmii¼2

αiimmii ðtÞδ�t � τiiðtÞ

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP2

þ �αJJ ðtÞδ

�t � τJJ

�þXPJJ�1

pJJ¼2

αJJpJJ ðtÞδ�t � τ3

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP3

þXJ

jj¼JJþ1

�αjjðtÞδ

�t � τjjðtÞ

�þXPPjj�1

ppjj¼2

αjjppjj ðtÞδðt � τjjðtÞ��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP4

þXL�1

k¼Jþ1

�αkðtÞδ

�t � τkðtÞ

�þXQk�1

qk¼2

αkqkðtÞδ�t � τk

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP5

: ð9Þ


Let the combination of each dominant and its corresponding sub-eigen paths be denoted as R′index(t). Using (9), h(t) can beexpressed as the sum of variables as below,

hðtÞ ¼ hSBET ðtÞ ¼ α0ðtÞδðtÞ þ R1ðtÞ|ffl{zffl}

GROUP1

þ RIIþ1ðtÞ þ RIIþ2ðtÞ þ :::þ RI ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP2

þ RJJ ðtÞ|fflffl{zfflffl}

GROUP3

þRJJþ1ðtÞ þ RJJþ2ðtÞ þ :::þ RJ ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP4

þRJþ1ðtÞ þ RJþ2ðtÞ þ :::RL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP5

; ð10Þ

where GROUP1 denotes the surface only reflected/scattered paths, GROUP3 denotes bottom only reflected/scattered paths,and GROUP5 denotes paths corresponding to refraction in the medium and refraction/reflection/scatter on other objects.The GROUP2 represents the surface/bottom reflected/scattered paths, while GROUP4 represents the bottom/surfacereflected/scattered paths. It is to be noted that arrival time of groups of paths and paths within the groups may vary.

According to Figure 1, the shallow channel can be classified into three broad categories. They are close to surface, closeto bottom, and in between channels. The surface and bottom roughness dictates further classification of shallow channelenvironment. The channel representations for these environments are given below.

A channel that represents a shallow channel communication environment in between surface and bottom can be expressedby (10), where hSBET (t) denotes the channel impulse response. Depending on how close to the surface and/or to the bottomthe source and destination are, both CIR, hSS and hSB , for close to surface and close to bottom environment respectively, mayhave paths in GROUP1 through GROUP4.

1) Shallow Channel Close to Rough Surface: Similar to a channel somewhere in between surface and bottom, a close torough surface shallow channel can be expressed as (10). If however the bottom is far away from the surface in a close to roughsurface communication environment, strictly ideally, the bottom reflected dominant paths may disappear due to higher lossand/or longer delay in arriving. Under such scenario, GROUP3 will be empty, and in GROUP2 and GROUP4 only surfacescattered sub-eigen paths will dominate. Let the combination of scattered sub-eigen paths generated only from the surfacecorresponding to each negligible surface/bottom reflected dominant paths be denoted by the variables R′index. In that case,(10) will reduce to hSS (t) as below,

hSS ðtÞ ¼ α0ðtÞδðtÞ þ R1ðtÞ|ffl{zffl}

GROUP1

þR′IIþ1ðtÞ þ R′IIþ2ðtÞ þ :::þ R′I ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP2

þR′JJþ1ðtÞ þ R′JJþ2ðtÞ þ :::þ R′J ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP4

þ RJþ1ðtÞ þ RJþ2ðtÞ þ :::RL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP5

: ð11Þ

The CIR in a close to a rough surface for long, short, and medium range transmission can be expressed as hSRS=L(t), hSRS=S (t),and hSRS=M (t), respectively. The expressions for these are derived from (11).

In a long range close to rough surface transmission, the sub-eigen paths due to scatter are negligible and hence may resultin GROUP2 and GROUP4 of (11) to be empty. Surface reflected dominant paths may dominate over the surface scatteredsub-eigen paths. Let R″

1(t) denote the combination of surface reflected dominant path and negligible sub-eigen scatters. Theremay be horizontal variation of sound speed resulting in scatter and/or dominant paths due to refraction. Even though somereflection and scatters may occur due to objects in the medium, in a long range these may/may not be negligible. HenceGROUP5 may not be empty. Using (11), the CIR for transmission can be written as

hSRS=LðtÞ ¼ α0ðtÞδðtÞ þ R″1ðtÞ|ffl{zffl}

GROUP1


GROUP5

: ð12Þ

In a short range close to rough surface transmission, surface scattered sub-eigen paths will be dominating. Therefore,GROUP2 and GROUP4 will not be empty. Due to short range, there may not be horizontal sound velocity variation


but the presence of other objects in the water may result in reflection and scatters. If the range is short, the reflection/scatteron objects in the medium may be significant. Therefore, GROUP5 may not be empty. The resulting CIR, hSRS=S (t) will be assame as (11) with R1(t) having more dominant sub-eigen paths.

In a medium range transmission, there may be dominant paths due to reflection and sub-eigen paths due to scatter. If therange is medium, the reflection/scatter on objects in the mediummay be significant. The CIR of such a channel, hSRS=M (t) willbe as same as (11).

2) Shallow Channel Close to Smooth Surface: In case of smooth surface, there will be no sub-eigen paths due to scatterson surface or bottom. Hence when the bottom is far away from the surface, ideally, GROUP2 and GROUP4 will be empty.For medium range transmission, (11) can be expressed as,

hSSS=M ðtÞ ¼ α0ðtÞδðtÞ þ R′″1 ðtÞ|fflffl{zfflffl}

GROUP1


GROUP5

; ð13Þ

where R′″1 denotes surface only reflected dominant path without any sub-eigen paths. Equation (13) also represents shortrange and long range close to smooth surface channels.

3) Shallow Channel Close to Rough Bottom: Similar to channels close to the surface, a close to rough bottom channelcan be expressed as (10). If the surface is far away from the bottom, ideally, the surface reflected dominant paths will disappeardue to higher loss and/or due to delay in arriving and (10) will reduce to hSB (t) as shown in (14). VariableR′Bindex(t) denotes thecombination of sub-eigen scatters generated only from the bottom corresponding to each negligible surface/bottom reflecteddominant path.

hSB ðtÞ ¼ α0ðtÞδðtÞ þ R′BIIþ1ðtÞ þ R′BIIþ2ðtÞ þ :::þ R′BI ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP2

RJJ ðtÞ|fflffl{zfflffl}

GROUP3

þ R′BJJþ1ðtÞ þ R′BJJþ2ðtÞ þ :::þ R′BJ ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP4

þ RJþ1ðtÞ þ RJþ2ðtÞ þ :::RL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP5

: ð14Þ

The CIR in a close to a rough bottom for long-, short-, and medium-range transmission can be expressed as hSRB=L , hSRB=S ,and hSRB=M , respectively. In a close to rough bottom shallow channel, in addition to bottom reflection, different bottom com-position may produce refracted scatters in GROUP2, GROUP3, and GROUP4.

In long-range transmissions, these scatters may not have significant contribution and so, GROUP2 andGROUP4may beempty. Then (14) becomes

hSRB=LðtÞ ¼ α0ðtÞδðtÞ þ R″JJ ðtÞ|fflffl{zfflffl}GROUP3


GROUP5

; ð15Þ

where R″JJ (t) denotes combination of bottom reflected dominant path and negligible sub-eigen scatters.Different marine life and other objects on the ocean floor may produce scatters due to reflection and refraction. In long-

range transmission, these scatters may not be significant but in short- and medium-range transmission, these scatters may bedominant. In short and medium range transmissions, strong bottom reflected/refracted scatters may also be dominant.Therefore, short range close to rough bottom transmission can be expressed as (14), denoted by hSRB=S (t), where RJJ(t) willcontain more dominant sub-eigen paths. A medium range close to rough bottom transmission can also be expressed as (14)and denoted by hSRB=M (t).

4) ShallowChannel Close to Smooth Bottom: In case of smooth bottom, there will be no sub-eigen paths due to scatterson the bottom. Hence for medium range transmission, (14) can be expressed as

hSSB=M ðtÞ ¼ α0ðtÞδðtÞ þ R′″JJ ðtÞ|fflffl{zfflffl}GROUP3


GROUP5

; ð16Þ


where R′″JJ (t) denotes bottom only reflected dominant path without any sub-eigenpaths. Equation (16) also represents short range and long range close to smoothbottom channels denoted by hSSB=S (t) and hSSB=L(t), respectively.

Deep Channel Representation According to EnvironmentIf only path loss is considered, in a deep channel, each multipath will go

through path loss given by Lurton (1996) and will arrive at various delays.In a deep channel, when the sound velocity is not constant, the first group of

multipaths represents the steepest paths, the second group represents grazingpaths for isothermal channel or axis paths for SOFAR channel, and the thirdgroup represents bottom reflected multipaths (Lurton, 1996). These paths arepresented below.

hDðtÞ ¼XID

iD¼1

�αiDðtÞδ

�t � τiD

�þXMDiD

�1

mDiD¼2

αiDmDiDðtÞδ�t � τiD

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP1

þXJD

jD¼IDþ1

�αjDðtÞδ

�t � τjD

�þXPDjD �1

pDjD ¼2

αjDpDjD ðtÞδ�t � τjD

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP2

þXLD�1

kD¼JDþ1

�αkDðtÞδ

�t � τkD

�þXQDkD

�1

qDkD¼2

αkDqDkDðtÞδ�t � τkD

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP3

: ð17Þ

Although (17) considers both eigen and sub-eigen components, the existence of afewer number of scatters in a deep channel may only cause dominant eigen com-ponents to exist. The deep channel can be broadly categorized into SOFAR andother types of deep channels.

1) SOFAR Channel: The CIR in a SOFAR channel for long-, medium-, andshort-range transmission can be derived from (17). For long-, medium-, andshort-range transmission, not close to surface or bottom, the scatters may beinsignificant resulting in only dominant components as below,

hDSOF=L;M ;S ðtÞ ¼XID

iD¼1

�αiDðtÞδ

�t � τiD

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GROUP1

þXJD

jD¼IDþ1

�αjDðtÞδ

�t � τjD

��

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}GROUP2

þXLD�1

kD¼JDþ1

�αkDðtÞδ

�t � τkD

��


GROUP3

: ð18Þ

However, for SOFAR channels close to surface or bottom, there may be somesignificant scatters that would result in (17). In both cases, GROUP1 denotes the


steepest paths, GROUP2 denotes theaxis paths, and GROUP3 denotes thebottom reflected paths.

2) Other Deep Channels:Isothermal Deep Channel: TheCIR in isothermal channels for long-,medium-, and short-range transmis-sion can be expressed as hDISO=L;M ;S

(t).For long-, medium-, and short-rangetransmission, not close to surface orbottom, the scatters may be insignif-icant resulting in mostly dominanteigen paths similar to (18). For chan-nels close to surface or bottom, thepresence of significant scatters may re-sult in an equation similar to (17). Thefirst group of paths denotes the steepestpaths, the second one denotes grazingpaths, and the third one denotes thebottom reflected paths.

Fading Characteristics,Probability Distribution,Relationship Between Fadingand Channel Parameters forWide Sense Stationary (WSS)Uncorrelated Tap Channel

In determining the fading char-acteristics and probability distributionof various UAC channels, the follow-ing assumptions are made:Assumptions:■ Channel paths are considered at

each sample time.■ In the absence of a natural path

at any sample, a path is gener-ated by interpolating neighboringpaths.

■ The symbol period is consideredlarge enough to encapsulate all thesignificant arrived multipaths.

After convolving with the channel thereceived signal, using (1), can be ex-pressed as

rðtÞ ¼ α0ðtÞsðtÞ þXL�1

l¼1

αl ðtÞs�t � τl

�;

þ n1ðtÞ þ n2ðtÞ; ð19Þ

where the variables s(t), n1(t) and n2(t)denote the transmitted signal, frequency-independent and frequency-dependentnoise, respectively.

When each arrived path is i.i.d.Gaussian, the following conditionsmay occur. Case a)When the dominantpath to less dominant multipath ratio,which is denoted by signal to multipathratio, SMR is less than 1, i.e., SMR ≪ 1the amplitude of the envelop of pathsfollows Rayleigh distribution (Geng &Zielinski, 1995). Case b) When theeigen path component dominates, asin the case of direct LOS, SMR ≫ 1,and the amplitude of the envelop ofthe paths follows Gaussian distribution.Case c) When the dominant path iscomparable to the less dominant paths,such that the dominant path exists asone single large path compared to theless dominant paths, i.e., SMR ≂ 1the amplitude of the envelop of thepaths follows Rician distribution.

The symbol is sampled at each ar-rivedmultipath. Each path at each sam-ple location is a combination of eigen andsub-eigen paths. At each sample location,these simultaneous paths can add con-structively or destructively. The ampli-tude of the resultant path at each samplelocation can be subjected to one of thethree conditions mentioned above result-ing inGaussian,Rayleigh orRician fadingdistribution. Let each sample represent

a variable, G, RL, or RC identifying an amplitude with either Gaussian, Rayleigh, orRician distribution, respectively.

These variables are an indication of the channel environment and are directlyrelated to the environment parameters. For example, the value of the range param-eter (short/medium/long) and/or depth parameter (close to surface/close to bot-tom, etc.) may determine the variable at each sample location to be G, RL, or RC.

If the channel delay spread is long and some paths arrive outside of the symbolduration, those outside paths will also contain combination of dominant and sub-eigen paths resulting in an amplitude variation, the distribution of which can bepresented by a Gaussian, Rayleigh, or Rician variable according to the channel en-vironment. Figure 4 illustrates the relationship between the paths, the symbol du-ration, and the sample time. The received signal representation in terms ofGaussian, Rayleigh, or Rician fading for the UAC channels described in the lastsubsection are given below. The SMR for each path can be expressed as

SMRPath ¼jEigenpathj

jP SubEigenPathj : ð20Þ

Depending on the value of SMRPath of each path, the combination of eigen andsub-eigen components will follow Rayleigh, Gaussian, or Rician distribution. Ifthe symbol duration encapsulate all paths, then

SMRSymbol ¼ jLOSjjPOtherPathj : ð21Þ

Depending on the value of SMRSymbol, the combination of all paths (each path acombination of eigen and sub-eigen components) within the symbol duration willfollow Rayleigh, Gaussian, or Rician distribution.

Fading characteristics of the shallow channel will be addressed first below, fol-lowed by the fading characteristics of the deep channel.

(i) Fading characteristics of shallow channel:1) ShallowChannel Close toRough Surface:The received signal in a close to a

rough surface for long-, short-, and medium-range transmission can be expressed asrSRS=L(t), rSRS=S (t), and rSRS=M (t), respectively. Using (12), in long-range transmission,

rSRS=LðtÞ ¼ G0 þh

RC″1ðtÞ|fflfflffl{zfflfflffl}

RicianðGROUP1Þ

i

þh

RCJþ1ðtÞ þ RCJþ2ðtÞ þ :::RCL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RicianðGROUP5Þ

i

n1ðtÞ þ n2ðtÞ; ð22Þ

where RC″1(t) represents the Rician envelop corresponding to surface reflected dom-inant path and sub-eigen paths. Each of RCJs represent a Rician envelop corre-sponding to scatter/dominant paths due to refraction from sound speed variationand/or reflection/refraction/scatter on objects in the medium. The surface reflectedpath will produce Rician fading if the scatters are not negligible otherwise it willproduce Gaussian fading.

FIGURE 4

Channel paths, symbol duration, and sampletime: The bold lines represent dominant eigenand dashed lines represent sub-eigen paths.Each arrived path within symbol duration andoutside of symbol duration may consist of onedominant and/or one/multiple sub-eigen paths.


In a short-range transmission, using (11), the received signal can be expressed as

rSRS=S ðtÞ ¼ G0 þh

RL1ðtÞ|fflfflffl{zfflfflffl}

RayleighðGROUP1Þ

i

þh

RL′IIþ1ðtÞ þ RL′IIþ2ðtÞ þ :::þ :::þ RL′I ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP2Þ

i

þh

RL′JJþ1ðtÞ þ RL′JJþ2ðtÞ þ :::þ RL′J ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP4Þ

i

þh

RLJþ1ðtÞ þ RLJþ2ðtÞ þ :::RLL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP5Þ

i


where RL1(t) represents the Rayleigh envelop corresponding to surface reflected path and dominant scattered sub-eigen paths.The RL′II s and RL′JJ s represent Rayleigh envelop of combination of surface scattered sub-eigen paths corresponding to eachnegligible surface/bottom reflected (first reflection on surface for GROUP2 and first reflection on bottom for GROUP4) dom-inant path. Each of RLJs represent a Rayleigh envelop corresponding to scatter/dominant paths due to refraction from soundspeed variation and/or reflection/refraction/scatter on objects in themedium. In a medium range, paths inGROUP5will resultin some Rician variables and some Rayleigh variables. Therefore, using (11),

rSRS=M ðtÞ ¼ G0 þh

RC1ðtÞ|fflfflffl{zfflfflffl}

RicianðGROUP1Þ

i

þh

RL′IIþ1ðtÞ þ RL′IIþ2ðtÞ þ :::þ RL′I ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP2Þ

i

þh

RL′JJþ1ðtÞ þ RL′JJþ2ðtÞ þ :::þ RL′J ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP4Þ

i

þh

RCJþ1ðtÞ þ RCJþ2ðtÞ þ :::RCJþJ′ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RicianðGROUP5Þ

þh

RLJ′þ 1ðtÞ þ RLJ′þ 2ðtÞ þ :::RLL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP5Þ

i


where RC1(t) represents the Rician envelop of dominant surface reflected path and sub-eigen scatters.2) Shallow Channel Close to Smooth Surface: In case of smooth surface, the received signal in long, short, and medium

range can be denoted as rSSS=L(t), rSSS=S (t), and rSSS=M (t), respectively. The equations for long, short, and medium range will besame and using (13) can be expressed as

rSSS=L;M ;S ðtÞ ¼ G0 þh

G′″1ðtÞ|fflfflffl{zfflfflffl}

GaussianðGROUP1Þ

i

þh

GJþ1ðtÞ þ GJþ2ðtÞ þ :::GL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GaussianðGROUP5Þ

i


whereG′″denotes the Gaussian variable corresponding to surface only reflected dominant path without any sub-eigen paths.The variableGJs denote Gaussian variables representing dominant paths due to refraction from sound speed variation and/orreflection/refraction on objects in the medium.


3) Shallow Channel Close to Rough Bottom: The received signal in a close to a rough bottom for long-, short-, andmedium-range transmission can be expressed as rSRB=L , rSRB=S , and rSRB=M , respectively. In a long-range transmission, the bottomreflected/refracted paths will produce Rician fading, if the scatters are not negligible otherwise it will produce Gaussian fading.Therefore, using (15),

rSRB=LðtÞ ¼ G0 þh

RC ′″JJ ðtÞ|fflfflfflffl{zfflfflfflffl}

RicianðGROUP3Þ

i

þh

RCJþ1ðtÞ þ RCJþ2ðtÞ þ :::RCL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RicianðGROUP5Þ

i


where RC″JJ represents the Rician envelop corresponding to bottom reflected dominant path and negligible sub-eigen scatters.Using (14),

rSRB=S ðtÞ ¼ G0 þh

RL′BIIþ1ðtÞ þ RL′BIIþ2ðtÞ þ :::þ RL′BI ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP2Þ

i

þh

RLJJ ðtÞ|fflfflffl{zfflfflffl}

RayleighðGROUP3Þ

i

þh

RL′BJJþ1ðtÞ þ RL′BJJþ2ðtÞ þ :::þ RL′BJ ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP4Þ

i

þh

RLJþ1ðtÞ þ RLJþ2ðtÞ þ :::RLL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP5Þ

i


where RL′BII s and RL′BJJ s represent Rayleigh envelops of combination of bottom scattered sub-eigen paths correspond-ing to each negligible surface/bottom reflected path (first reflection on surface for GROUP2 and first reflection on bottomfor GROUP4). In a medium range, paths in GROUP5 will result in some Rician variables and some Rayleigh variables.Therefore, using (14)

rSRB=M ðtÞ ¼ G0 þh

RL′BIIþ1ðtÞ þ RL′BIIþ2ðtÞ þ :::þ RL′BI ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP2Þ

i

þ RCJJ ðtÞ|fflfflffl{zfflfflffl}

RicianðGROUP3Þ

þh

RL′BJJþ1ðtÞ þ :::þ RL′BJ ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP4Þ

i

þh

RCJþ1ðtÞ þ RCJþ2ðtÞ þ :::RCJþJ ′ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RicianðGROUP5Þ

þ RLJ ′þ1ðtÞ þ RLJ ′þ2ðtÞ þ :::RLL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

RayleighðGROUP5Þ

i


where RCJJ (t) represents the Rician variable corresponding to the bottom reflected dominant path and sub-eigen scatters.4) Shallow Channel Close to Smooth Bottom: In case of smooth bottom, the equations can be expressed as rSSB=L , rSSB=S ,

and rSSB=M . Expressions for short, medium, and long range will be same and using (16) can be expressed as

rSSB=S ðtÞ ¼ G0 þh

G′″JJ ðtÞ|fflfflffl{zfflfflffl}

GaussianðGROUP3Þ

i

þh

GJþ1ðtÞ þ GJþ2ðtÞ þ :::GL�1ðtÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

GaussianðGROUP5Þ

i


where G′″JJ represents the Gaussian variable corresponding to bottom reflected dominant path.


5) Shallow Channel in BetweenSurface and Bottom: The receivedsignal in an environment between sur-face and bottom will be affected byboth surface and the bottom. The typeof channel for long, short, and mediumrange transmission can be expressed asrSBET =L

, rSBET =S, and rSBET=M

respectively.The equation for rSBET =L can be expressedas a combination of rSSS=L and rSSB=L ,when both surface and the bottom aresmooth. If either or both are rough,the equations can bemodified to includemore scatters. Similarly, rSBET =S will bea combination of rSSS=S and rSSB=S , andrSBET =M will be a combination of rSSS=Mand rSSB=M . Rough surface/bottom willadd scatters and equations will changeaccordingly. In a clear, unobstructedchannel, the increase of transmissionrange and water column depth willproduce dominant components. Ifthe transmission is close to surface/bottom or there are smaller objects inthe water, the resulting scatters com-bined with the dominant componentmay produce Rician fading. Withfarther increase of range and depth,scattering effect may be minimal andthe Rician may become Gaussianonly fading. In a clear, unobstructedshort range and low water columndepth channels, there will be dominantcomponents. But due to the closenessof the two boundaries and the shorterrange to travel, the scattering com-ponents will be significant and thescatters will be comparable to thedominant components, resulting inRayleigh fading. Figure 5 illustratesthis characteristics.

(ii) Fading characteristics in deepchannel: Deep channel transmissionwill result in Gaussian fading due tothe presence of noise, if the transmis-sion happens far from surface/bottomboundaries, and the objects in thetransmission path are scarce. If however


the deep channel is close to surface/bottom, the fading characteristics ofclose to surface, close to bottom or inbetween channels may apply.

Frequency Dependent andIndependent Noise

Table 2 presents the possible noisein UAC channel. Some of these noiseis frequency independent and canbe considered under, n1(t) as AdditiveWhite Gaussian Noise (AWGN) withzero mean and normal distribution.However, some of the noise in Table 2can be frequency dependent, and if theoperating transmission frequency is assame as the frequencies of any of thesenoise sources, the transmitted signal

al

will be affected. Frequency-dependentnoise is included under n2(t).

Fading Characteristics,Distribution, RelationshipBetween Fading andChannel Parameters forQuasi-stationary Channels

In UAC channel, wide sense sta-tionary (WSS) assumption is not alwaystrue (Bjerrum-Niese & Lutzen, 2000;Socheleau et al., 2011; Walree et al.,2008). When the transmitter and thereceiver or their platforms are moving,the various taps can be correlated atdifferent delays. There are different ap-proaches to address non-WSS chan-nels. One method is to consider thechannel as quasi-WSS, i.e., the channelis WSS within a certain time and bandof frequency. Another way of dealingwith non-WSS is to define time and fre-quency dependent scattering functions(Walree et al., 2008). In this paper, thechannels considered are WSS.

SNR AnalysisSignal-to-Noise Ratio for WSSUncorrelated Tap Channel

From the equations (22) through(29), it can be concluded that the receivedsignal in any underwater environmentof Figure 1 is a summation/combinationof Gaussian random variables (RV),

TABLE 2

Self and ambient noise in UAC.

Self-Noise
MachineryNoise Flow Noise Cavitation
Pumps,reduction gears,power plant,etc.

Relative motionbetween objectand the water

Bubble collapse

Ambient Noise
Hydrodynamic Seismic Biological Ocean traffic
Movementof water

Movement ofearth

Marine life
Other boats
FIGURE 5

Shallow channel: (a) Long-range, high-depth,and low-frequency Rician fading zone. (b) Short-range, low-depth, and high-frequency Rayleighfading zone.

Rician RVs, and Rayleigh RVs. Let thesummation of Gaussian RVs, RicianRVs, and Rayleigh RVs be denoted asGSUM, RSUM, and RLSUM, respectively.Therefore, the pdf of the received signalwill be equal to the pdf of an RV thatis a summation of GSUM, RSUM, andRLSUM or any combination of them.If the RV representing the received sig-nal is denoted by RVREC, then

RVREC ¼GSUM þ RSUM þ RLSUM ð30Þ

Although RVREC is expressed in (30)as summation of Gaussian, Rayleigh,or Rician variables, RVREC representsthe resultant after the vectorial additionof all paths within symbol duration.Since every path is an i.i.d. Gaussian,the resultant amplitude, i.e., |RVREC|will follow any one of Gaussian,Rayleigh, or Rician distributions. Ricianfading is governed by two parameters,K, and Ω, which can be expressed as,

K ¼ PdPLL�1

pi¼1 Ppi; ð31Þ

Ω ¼ Pd þXLL�1

pi¼1

Ppi; ð32Þ

where Pd represents the power of thedominant path, and Ppi, the power ofeach less dominant path. Clearly,when the power of any path is deter-mined by the transmission loss in (6)and by the equation for α, the pathpowers will depend on R, H, and f.According to the UAC environment,R, H, f will vary resulting in differentvalues of K andΩ. When K≂ 0 or K≪1, the denominator or the sum ofpowers of less dominant paths will begreater than the power of the domi-nant path, the received signal will fol-low Rayleigh fading. For K ≥ 1, thereceived signal will follow Rician statis-

tics. For K ≫ 1, the signal will followGaussian statistics.

The signal to noise ratio (SNR) canbe expressed as,

SNR ¼ SignalPowerNoisePower

¼ jP Pathj2NoisePower

:

ð33Þ

Let the AWGN noise variance and thefrequency dependent noise variance bedenoted as σ2 and Notr, respectively.Therefore, (33) can be expressed as

SNR ¼ ðjRVREC jÞ2σ2 þ Notr

ð34Þ

or using (19)

SNR ¼ ðjPL�1l¼0 αl ðtÞsðt � τl ÞjÞ2

σ2 þ Notr: ð35Þ

The eigen and sub-eigen paths will varyin different environments. Hence, theSNR will vary according to each en-vironment. The numerator of (35) willbe the square of Rayleigh or Rice orGaussian variable and the denomina-tor will be the summation of Gaussianvariance and variance of noise com-ponent, Notr. The square of Rayleighvariable gives exponential distributionand the square of Rician variable givesnon-chi-square distribution and hencecan be expressed as RVEXP and RVCHI,respectively. Therefore, when RVREC isa Rayleigh or Rician variable, SNR canbe expressed as SNREXP and SNRCHI

respectively as shown below.

SNREXP ¼ RVEXP

ðσ2 þ NotrÞ ð36Þ

SNRCHI ¼ RVCHI

ðσ2 þ NotrÞ ð37Þ

May/Ju

where the mean and the variance of theabove SNR will be as follows,

μEXP ¼ RVEXP

ðσ2 þ NotrÞ ;

σ2EXP ¼ VARðRVEXPÞ

ðσ2 þ NotrÞ2ð38Þ

μCHI ¼RVCHI

ðσ2 þ NotrÞ ;

σ2CHI ¼

VARðRVCHI Þðσ2 þ NotrÞ2

ð39Þ

Using the values of exponential meanand variance and non-central chi-squaremean and variance, equations (38) and(39) become

μEXP ¼ λ�1

ðσ2 þ NotrÞ ;

σ2EXP ¼ λ�2

ðσ2 þ NotrÞ2; ð40Þ

μCHI ¼ðK ′þ λÞðσ2 þ NotrÞ ;

σ2CHI ¼

2ðK ′þ 2λÞðσ2 þ NotrÞ2

; ð41Þ

where K ′ > 0 represents the degrees offreedom and λ > 0 represents the ratein the first equation and noncentralityparameter in the second equation.

Benefits of ChannelClassification andSelection of Channels

In this section, first, the benefits ofchannel classification will be providedbriefly and, second, selection of chan-nels will be discussed.

Benefits of ChannelClassification

It is beneficial to classify the UACchannels according to their environ-ments. Once the UAC environments


are determined, the appropriate chan-nel representation can be used.

In shallow channel environments,close to surface channel may haveLOS path (if both the transmitterand receiver are in sight of each other)arriving first, followed by one timesurface reflected path, then one timebottom reflected path, followed bybottom-surface one time reflectedpath, surface-bottom one time reflectedpaths, followed by other multiplereflected paths. In close to bottomchannel, the paths will be in theorder of LOS, bottom reflected, sur-face reflected, bottom-surface onetime reflected path, surface-bottomreflected path followed by other mul-tiple reflected paths. In in-betweenchannels, if the difference |zs − zr| issmall the order of the paths will beLOS, surface, bottom, surface-bottomone time reflected, bottom-surface onetime reflected path followed by multi-ple reflected paths. If the difference|zs − zr| is large, the paths will be LOS,surface, bottom, bottom-surface,surface-bottom followed by rest of themultiple reflected paths. Once thechannel environment and hencethe path origins are identified, it maybe easier to retrieve the signal infor-mation. For example, a surface only re-flected path will be affected by surfacewave and other surface agitations.Knowledge of path origin will allowapplying noise cancellation and small-scale fading compensation techniquessuited for surface boundary agitation.

The UAC channels can be affectedby geographical location of commu-nication (longitude/latitude), trans-mitter and receiver distances from thesurface and the bottom boundaries,water column height, transmissionrange, frequency, sound velocity, thewater environment, etc. Channel param-eters unique to each UAC environment


is used to differentiate each environ-ment and its corresponding channelequations. For example, the longitude/latitude of location, time of year or sea-son, wind velocity will determine theroughness of the sea surface. Knowingthese parameters in addition to thetransceiver depth from surface will de-termine if the channel is close to roughor smooth surface. Once rough sur-face is identified, scatters due to roughsurface may be taken into account, andthe channel equations correspondingto close to rough surface environmentmay be utilized.

Selection of Channel ModelsThe CI can sense channel envi-

ronment parameters and map themto accurate UAC channel environmentaccording to parameter values (Ahmed& Arslan, 2009a). Once the UACchannel environment is determined,the UAC channel environment can bemapped (Ahmed & Arslan, 2009b)onto any one of the appropriate chan-nel representations that were presentedin the previous sections.

Channel Representation of ClosestFit: In an event, when the channel data-base does not contain an exact matchof a UAC environment, CI can choosea channel representation that fitsclosest to the UAC environment.This may happen when a significantnumber of channel parameters and/orparameter values defining the spe-cific UAC environment differ from aknown UAC environment in the data-base. For example, the depth value in anew UAC environment may be largerthan the depth threshold for close tosurface environment in the database,the temperature, and the location inthe world may indicate the new envi-ronment to be in between the sur-face and the bottom. But the type ofocean traffic and the type of marine

al

life concentration may indicate thenew UAC environment to be a closeto surface channel environment. Insuch a case, depending on the devia-tion of parameter values of the newUAC environment from the thresh-olds of parameter values of a knownUAC environment in the channel data-base, the in between surface and bottomenvironment may be selected. Corre-spondingly, the channel representationfor in between surface and bottom canbe chosen for communication.

Simulation ResultsThe simulation is carried out using

the parameters in Table 3, MATLAB,and the Bellhop model (Porter, 2011)used and provided by the AcTUPsoftware module (Maggi & Duncan,2005).

The range to depth ratio can definean underwater channel to be shallowor deep. A range to depth ratio of10:1 may define a channel as shallow(Bessios & Caimi, 1994) and the trans-mission can be considered horizontal.The range of transmission can definea channel to be short, medium, orlong. A long-range channel over severaltens of kilometers is limited to a fewkilohertz; a medium-range channelover several kilometers has a band-width in the order of tens of kilohertz,while a short range channel over severaltens of meters may have hundreds ofkilohertz (Stojanovic, 1996) available.

First, the simulation results of thechannel impulse response are presented.Second, the results of the fading char-acteristics of various channels arepresented.

A) Channel Impulse Responsein Different UAC Environments■ Shallow Channel (Constant Sound

Velocity), Smooth Surface andBottom:

In a shallow in between surfaceand bottom channel environment,when the boundaries are smooth, theCIR will contain paths from surfaceonly reflection, multiple surface bot-tom reflections, bottom only reflection,multiple bottom surface reflections,and paths due to other reflections/refractions/scatters from in betweenobjects as in (10). The CIR of in be-tween channels in Figures 8, 11, and14 contain all these groups of paths.

Figures 6, 9, and 12 illustrateclose to smooth surface channels. Ifthe paths reflected from the bottomare weak, they can be neglected. Fig-ures 7, 10, and 13 illustrate close tosmooth bottom channels. If the pathsreflected on the surface are weak, theycan be neglected.

In close to surface smooth chan-nels, the values of source depth zs andreceiver depth zr will be smaller com-pared to water column height H ortransmission range D. When both Hand D are small, in case of short range,the paths LOS, surface only, bottom-only, bottom-surface, surface-bottom,paths will be more distinctly apart fromeach other. IfD is high as in long-rangechannels, the above paths will arrive

close to each other and will be lessdistinct.

For short and medium range, ifH is higher, multiple boundary re-flected paths will be grouped togetherand will almost overlap. For example,paths resulting from 2Hwill be grouped

May/Ju

together, 3H will be grouped together(using (4)).

Equations (4) and (6) are used togenerate delay and transmission lossamplitudes of Figure 15. In this figure,only the LOS, surface-only, bottom-only,bottom-surface, and surface-bottom

TABLE 3

Values of depth, transmission distance, transmitter and receiver depth for shallow channel scenario: Long-, medium-, and short-range channels.

Depth (m)
Distance (m) Transmitter Depth (m) Receiver Depth (m) Channel Type
ne 2013 Volume 47 N

Frequency (kHz)

1,000
10,000 995 995 Long, Close to bottom 1
5
5 Long, Close to surface 1
5
995 Long, In between 1
100
1,000 95 95 Medium, Close to bottom 10
5
4 Medium, Close to surface 10
5
95 Medium, In between 10
9
90 8.5 8.5 Short, Close to bottom 100
1
1 Short, Close to surface 100
1
8.5 Short, In between 100
FIGURE 6

Short range, close to surface, shallow channel (smooth surface and bottom): transmission loss/amplitude at various delays and at various ranges, depth = 9 m, transmission distance = 90 m,frequency = 100 kHz.

umber 3 113


F

Msaf

F

Msaf

FIGURE 7

Short range, close to bottom, shallow channel (smooth surface andbottom): transmission loss/amplitude at various delays and at variousranges, depth = 9 m, transmission distance = 90 m, frequency = 100 kHz.

FIGURE 8

Short range, in between smooth surface and smooth bottom, shallowchannel (smooth surface and bottom): transmission loss/amplitude atvarious delays and at various ranges, depth = 9 m, transmission distance =90 m, frequency = 100 kHz.

IGURE 10

edium range, close to smooth bottom, shallow channel (smoothurface and bottom): transmission loss/amplitude at various delaysnd at various ranges, depth = 100 m, transmission distance = 1,000 m,requency = 10 kHz.

IGURE 9

edium range, close to smooth surface, shallow channel (smoothurface and bottom): transmission loss/amplitude at various delaysnd at various ranges, depth = 100 m, transmission distance = 1,000 m,requency = 10 kHz.

F

Lcvd

F

Lavf

FIGURE 11

Medium range, in between smooth surface and smooth bottom, shallowchannel (smooth surface and bottom): transmission loss/amplitudeat various delays and at various ranges, depth = 100 m, transmissiondistance = 1,000 m, frequency = 10 kHz.

FIGURE 12

Long range, close to smooth surface, shallow channel (smooth surfaceand bottom): transmission loss/amplitude at various delays and atvarious ranges, depth = 1,000 m, transmission distance = 10,000 m,frequency = 1 kHz.

IGURE 14

ong range, in between smooth surface and smooth bottom, shallowhannel (smooth surface and bottom): transmission loss/amplitude atarious delays and at various ranges, depth = 1,000 m, transmissionistance = 10,000 m, frequency = 1 kHz.

IGURE 13

ong range, close to smooth bottom, shallow channel (smooth surfacend bottom): transmission loss/amplitude at various delays and atarious ranges, depth = 1,000 m, transmission distance = 10,000 m,requency = 1 kHz.


paths are displayed. For short range chan-nel (a), D is small and H is small, hencethe LOS and surface only paths aredistinct. For medium-range (b) andlong-range (c) channels, both D andH are high and these two paths almostoverlap each other. Paths interactingwith the bottom boundary arrive laterbut these paths are close to the LOSand surface only reflected paths interms of loss and delay and hence cannotbe neglected. If zs and zr are smallenough andH andD are large enough,the paths interacting with the bottomwill have larger delay and loss and hencecan be neglected as seen in equation (13).

In a short range, shallow, closeto smooth surface channels, as inFigure 6, the paths are more spreadout at each delay, i.e., at each delay,neighboring paths arrive without over-lapping. In Figure 8, short range, inbetween surface and bottom channel,the paths are less spread out comparedto closed to surface case. In a close tobottom channel, as in Figure 7, thepaths are least spread out among thethree short range channels.


Compared to the short range chan-nels, the medium range channel pathsare much less spread out and are al-most overlapping, and are arriving asan overlapped bundle at each delay.This can be observed in Figures 9, 10,and 11. In long-range channels, theneighboring paths completely overlapeach other. As a result, at each delay,there is only one path instead of spread-ing of paths. Figures 12, 13, and 14illustrate this.

In the shallow channel environ-ments presented above, the paths in dif-ferent groups arrive very close to eachother. Because of this reason, the pathsin different groups are not separatelyshown in the figures.■ Shallow Channel (Constant Sound

Velocity), Rough Surface andBottom:
If the bottom or the surface varies in
shallow channels, the power delay pro-file varies drastically, where only a fewpaths arrive at the receiver at differentdelays and at different power levels.

For a varying surface, the num-ber of paths, their amplitudes, and de-

al

F

Tcs

lays in Figures 6, 9, and 12 will vary.This variation will be more prominentin short and medium range channels.

For a varying bottom, the numberof paths, their amplitudes, and delaysin Figures 7, 10, and 13 will vary.This variation will be more prominentin short- and medium-range channels.■ Deep Channel Simulation (Varying

Sound Velocity Profile):
Sound velocity will vary in a deep
channel. Considering a variable soundvelocity profile simulation can be car-ried out in SOFAR and other deepchannels. Simulation results will besimilar to that of the shallow channelfigures as presented above. Because ofhigh depth and long range that charac-terize the deep channels, there will behigh loss associated with each path ar-riving at very long delays.

B) Fading Characteristics ofWSS Uncorrelated Tap Channel

In Figure 16, the power (powercalculated in terms of path loss) ratiosof direct path to other paths, in variousranges, in shallow water close to bot-tom channels are illustrated. The power

FIGURE 15

Shallow channel, close to surface (smooth surface and bottom): trans-mission loss at various delay. (a) Depth = 9m, distance = 90m, transmitterdepth = 1 m, receiver depth = 1 m, frequency = 100 kHz. (b) Depth = 100m, distance = 1,000 m, transmitter depth = 5 m, receiver depth = 4 m,frequency = 10 kHz. (c) Depth =1,000 m, distance = 10,000 m, trans-mitter depth = 5 m, receiver depth = 5 m, frequency = 1 kHz.

IGURE 16

op: Short range, close to bottom, shallow channel; Middle: Medium range,lose to bottom, shallow channel; Bottom: Long range, close to bottom,hallow channel.

values are taken from Figures 7, 10,and 13. All the arrived paths are con-sidered. In medium- and long-rangechannels, the ratio is greater than 1 atall three distances, indicating Rician orGaussian fading. At 6,000 m in longrange and at 600 m in medium range,it is Gaussian fading. At 3,000 m inlong range and at 200 m in mediumrange, it is also Gaussian fading. Atlonger distances, the direct path losespower significantly and becomes com-parable to the less dominant paths.Hence at 1,000 m in medium and at10,000 m in long range channels,it becomes Rician fading. In shortrange channels, the indirect pathshave significant power and hence re-sults in mostly Rayleigh fading at allthree distances.

ConclusionOnce the UAC environments are

determined from the channel parameters,the channel representations can be de-rived according to the environments.In this paper, a high level classificationof the UAC channel environments arepresented and channel representationsare proposed, which are derived accord-ing to the classified UAC environments.The channel analysis is shown in termsof channel impulse response equationsand fading characteristics of the receivedsignals. As part of future research, cor-related taps and non-WSS channelscan be considered.

Authors:Sadia AhmedHuseyin ArslanDepartment of Electrical Engineering,University of South Florida,4202 East Fowler Avenue,ENB 118, Tampa, FL 33620Email: [email protected];[email protected]

ReferencesAhmed, S., & Arslan, H. 2009a. Cognitive

intelligence in UAC channel parameter

identification, measurement, estimation, and

environment mapping. In: Proc. MTS/IEEE

Oceans. Bremen, Germany. doi: 10.1109/

OCEANSE.2009.5278161.

Ahmed, S., & Arslan, H. 2009b. Cognitive

intelligence in the mapping of underwater

acoustic communication environments to

channel models. In: Proc. MTS/IEEE Oceans.

Biloxi, Mississippi, USA.

Appleby, S., & Davies, J. 1998. Time,

frequency and angular dispersion modeling

in the underwater communications channel.

In: Proc. MTS/IEEE Oceans. Nice, France.

doi: 10.1109/OCEANS.1998.724318.

Bessios, A., & Caimi, F. 1994. Multipath com-

pensation for underwater acoustic communica-

tion. In: Proc.MTS/IEEEOceans. Brest, France.

doi: 10.1109/OCEANS.1994.363905.

Bjerrum-Niese, C., & Lutzen, R. 2000.

Stochastic simulation of acoustic commu-

nication in turbulent shallow water. IEEE J

Ocean Eng. 25(4)523-32. doi: 10.1109/

48.895360.

Byun, S., Kim, S., Lim, Y., & Seong, W.

2007. Time-varying underwater acoustic

channel modeling for moving platform.

In: Proc. MTS/IEEE Oceans. Vancouver, BC.

doi: 10.1109/OCEANS.2007.4449361.

Galvin, R., & Coats, R. 1996. A stochastic

underwater acoustic channel model. In: Proc.

MTS/IEEE Oceans. Fort Lauderdale, FL.

doi: 10.1109/OCEANS.1996.572599.

Geng, X., & Zielinski, A. 1995. An eigenpath

underwater acoustic communication channel

model. In: Proc.MTS/IEEEOceans. SanDiego,

CA. doi: 10.1109/OCEANS.1995.528591.

Gutierrez, M.A.M., Sanchez, P.L.P., & Neto,

J.V.V. 2005. An eigenpath underwater acoustic

communication channel simulation. In: Proc.

MTS/IEEE Oceans. Washington, DC, USA.

doi: 10.1109/OCEANS.2005.1639788.

Lurton, X. 1996. An Introduction to

Underwater Acoustics: Principles and

Applications. UK: Springer.

May/Ju

Maggi, A., & Duncan, A. 2005. Underwater

acoustic propagation modeling software-actup

v2.2l. Center forMarine Science and Technology

(CMST). January 25, 2012.

Morozov, A.K. 1995. The sequential analysis

application in the problem of estimation

of stochastic channel impulse response.

In: Proc. MTS/IEEE Oceans. San Diego, CA.

doi: 10.1109/OCEANS.1995.528570.

Porter, M.B. 2011. The bellhop manual and

user’s guide: Preliminary draft. Heat, Light,

and Sound Research, Inc. May 30, 2012.

Socheleau, F., Laot, C., & Passerieux, J.

2011. Stochastic replay of non-WSSUS under-

water acoustic communication channels

recorded at sea. IEEE Trans Signal Process.

59(10)4838-49. http://dx.doi.org/10.1109/

TSP.2011.2160057.

Stojanovic, M. 1996. Recent advances in

high-speed underwater acoustic communica-

tions. IEEE J Ocean Eng. 21(2):125-36.

http://dx.doi.org/10.1109/48.486787.

Walree, P., Jenserud, T., & Smedsrud, M.

2008. A discrete-time channel simulator driven

by measured scattering functions. IEEE J

Selected Areas Commun. 26(9)1628-37.


P A P E R

Wrecks on the Bottom: Useful,Ecological Sentinels?
A U T H O RAndrea PeiranoENEA-Marine EnvironmentResearch Centre, La Spezia, Italy

A B S T R A C T

al

Wrecks play an important role in enhancing marine biodiversity. SCUBA divingvideo-samplings were performed on eight wrecks, including seven shipwrecks anda sunken airplane, scattered over 180 km along the Ligurian coastline in the north-western Mediterranean Sea, in water depths of 30–65 m. Differences in composi-tion of macrobenthic communities were found to be related to the bottomsedimentology and the composition and geometry of the investigated structures.The iron, flat, and even substrata of shipwrecks were dominated by Bivalvia (Ostreaedulis) and Anthozoa (Corynactis viridis), whereas the aluminum, cage-like struc-ture of the airplane was dominated by massive sponges and bryozoans. Further-more, abundances of some macrobenthic invertebrates were greater on thewrecks than those observed on natural, rocky substrates. Because of the increasingpopularity of use of video and photo digital apparatuses among recreational divers,combined with wreck resistance to corrosion, artificial geometries and availablesurfaces that facilitate species’ settlement and growth, data from wrecks consistingof images and related metadata (depth, date, water temperature, etc.) could providea new opportunity for monitoring Mediterranean marine biodiversity.Keywords: wrecks, macrobenthos, video-sampling, CCR scientific diving

for the larve of benthic organisms,which otherwise would not be able

Introduction

Wrecks are artificial habitats thatmay enhance marine biodiversity(Baine, 2001; Svane & Petersen, 2001;Massin et al., 2002; Mallefet et al.,2008). Their geometrical complexi-ties, structures, and textures providesuitable substrates for settlementsof several benthic species (Svane &Petersen, 2001). On soft bottoms,wrecks especially play an importantrole by offering suitable surfaces

to settle.Given their importance as promot-

ers of marine biodiversity, old shipsand sunken oil and gas platforms,both of which have become popularas dive attractions, have been includedwithin Marine-protected areas (MPA)of the world (Bonshack & Sutherland,1985; Baine, 2001; Walker et al.,2007; Kaiser & Kasprzak, 2008).World War II ships are the most com-mon and best preserved wrecks be-cause of the thickness of their hullsand the armoring of their superstruc-tures. Moreover, some of them havebeen included on lists of archaeologicalimportance to ensure that their histor-ical importance is passed on to futuregenerations (UNESCO, 2001; Churchet al., 2007; Wessex ArchaeologyLimited, 2008; Underwood, 2009).

Because of the function and use ofthese artificial structures, however,the presence of pollutants, such asheavy metals in cables and paintsand oil and explosive residuals (Churchet al., 2007) may be found withintheir compartments. Thus, beforebeing eligible to be positioned inten-tionally on the seabed, these struc-tures should be cleaned and treatedto remove any pollutant representingserious threat to marine life. Further-more, their scuttling should be limitedby appropriate regulations (Bell et al.,1997; Helton, 2003).

On the Ligurian Sea seabed in thenorthwest Mediterranean Sea, morethan 50 wrecks were found alongthe coastline in water depths of 20–100 m (Carta, 2000; Carta et al.,2008; Mirto et al., 2009). The major-

ity are ships that were sunk duringWorld War II. Airplanes that werelost in the war often have been de-stroyed and dispersed by trawlers.The interest in Italian wrecks has gen-erally come from archeologists becauseof their importance as sites of nationalheritage (http://www.archeomar.it).However, except for a few generalobservations, the importance of theLigurian wrecks as a possible promoterof Mediterranean marine biodiversityhas not been taken into account.

The aim of the present work is toanalyze and compare the characteris-tics and composition of macrobenthiccommunities colonizing eight wrecks,taking into account the structural ge-ometry, composition, and texture ofthe materials composing the wrecks.Comparisons between macrobenthic

taxa on wrecks with those living onneighboring natural substrates alsowere made to elucidate factors influ-encing colonization and the wrecks’role in contributing to coastal com-munity richness and conservation.The wrecks as tools for monitoringmarine biodiversity are discussed.

Material and MethodsStudy Site

SCUBA video surveys were per-formed in 2008–2009 on eight“wrecks” lying on the Ligurian seabed(Figure 1), scattered over 180 km alongthe coastline. The Liguria region isdominated by a large and well-definedanticlockwise circulation and by bio-diversity hot spots promoted by thepresence of a coralligenous commu-nity, usually growing below 25 m indepth (Cattaneo-Vietti et al., 2010).With the purpose of analyzing onlymature communities, wrecks that sank30 or more years ago (Perkon-Finkelet al., 2006; Relini et al., 2007) wereselected for this study. Among these

artificial structures, only those lyingbetween 30 and 65 m in depth weretaken into account to reduce the im-pact caused by wave action and theseasonal thermocline. Seven of theeight wrecks investigated were ships;six were iron ships that sank between1917 and 1967 (MohawkDeer, Genova,Marcella, Vittoria, Equa,Concordia) andone was a wooden ship (the Cycnus)sunk in 1981. The other wreck, theBR.20, was an Italian bomber shotdown near Imperia in 1941. TheCycnus and the bomber were situatedon sandy bottoms close to Posidoniaoceanica seagrass beds (Bianchi &Peirano, 1995). One of the iron ships,theMohawk Deer, lay on a detritic sea-bed with the bow against a rocky walldominated by a rich coralligenous as-semblage. The other shipwrecks werepositioned on muddy bottoms.

Video-SamplingThe videos were taped by means of

SCUBA diving performed using amixed-gas, closed-circuit rebreather(CCR) apparatus APD “Inspiration”®

May/Ju

(Figure 2). Compared to an open-circuit (OC) apparatus, CCR alloweddives at greater depths with longerstays and a reduced number and dura-tion of decompression stops (Parrish& Pyle, 2002). A Sony full-HD(1,920 × 1,080 pixels) video camerawas enclosed in an underwater housingwith a wide-angle lens and two lamps(50W each). A frame (25 × 25 cm) wasfixed on the camera at distance of30 cm (Figure 3). Short videos (videoclips) were shot at random in differentparts of each wreck, as hull, cranes, andpropeller, etc. In the laboratory, videos

FIGURE 1

Wrecks’ distribution along the Liguria region. 1 = Cycnus, 2 = BR.20 Bomber, 3 = Mohawk Deer,4 = Genova, 5 = Marcella, 6 = Vittoria, 7 = Equa, 8 = Concordia.

FIGURE 2

Exploring the wreck of the Equa using themixed-gas closed-circuit rebreather (CCR)apparatus APD Inspiration®.

FIGURE 3

One video image from the Concordia star-board side of the ship, colonized by Ostreaedulis and Corynactis viridis (dark points).During the video shootings, the frame allowedthe diver to maintain the video camera per-pendicular to and at a fixed distance from thesurface.


were analyzed, and for each wreck,several variables, divided in differentcategories, were evaluated. Data ondepth range (30–40 m, 40–50 m,50–60 m), bottom/seabed (sediment)type (1 = sand, 2 = mud, 3 = detritic,4 = rock), wreck zone (1 = internal, 2 =external), wreck material (1 = wood,2 = iron, 3 = aluminum), surface tex-ture (1 = even, 2 = rough, 3 = grid-like,4 = with cracks), geometrical shapeof the structure (1 = round, 2 = flat,3 = angular, 4 = complex), and struc-ture slope (1 = horizontal, 2 = subhori-zontal, 3 = vertical, 4 = overhanging)were extracted from each video clip.

Among benthic organisms foundon the wrecks, only the long-livingmacrobenthic taxa were consideredin the present study because of theirmultiannual lifespan of colonization.Taxonomical identification of themacrobenthic species was facilitatedby using HD digital video softwares(Picture Motion Brower® by SonyandMedia Player Classic® by Gabest).Macrobenthic species abundance wasevaluated in terms of cover on still-frames selected randomly from eachvideo clip. The minimum surface ana-lyzed in each video clip was 1 m2. Thetotal abundance of a species was esti-mated following semiquantitativescale adopted in cover visual estimation(Boudouresque, 1971; Kenchigton,1978; Bianchi et al., 2004; Worket al., 2008): <1% = rare, 1–10% =present, 10–50% = abundant, > 50% =very abundant. The resulting matrix(wrecks × variables) was analyzed byusing the principal component analysis(PCA).

Data on species abundance re-corded on wrecks were comparedwith those collected with video and/orphoto sampling on natural substratanearby. Differences between wrecksand natural substrata were examined.


ResultsA minimum dive-time of 20 min

allowed a careful exploration of eachwreck. The resulting high-definition(HD) videos extracting single frames(maximum resolution of about 5 mega-pixels each) guaranteed suitable imagesfor identifying macrobenthic speciesmore than 1 cm in size. Macrobenthicspecies were classified at species orfamily level. A total of 35macrobenthicorganisms (7 Algae, 11 Porifera,9 Anthozoa, 1 Bivalvia, 2 Anellida,4 Bryozoa) were identified (Table 1).

Principal component analysis(PCA) revealed that the first two com-ponents were both significant: the first(PC1) explained 10.5 % of the vari-ance, the second (PC2) explained7.9 %. (Figures 4a and 4b). The anal-ysis of the variable importance, rep-resented in Figure 5, showed thatspecies ordination was mainly relatedto the type of the bottom (muddy bot-toms vs. detritic/sandy bottoms) andto the type of material composing thewreck (aluminum vs. iron). Hence,variables were divided into two maingroups, represented on the right andon the left sides of the PCA diagramrespectively (Figure 4a). Flat and verti-cal surfaces of the iron shipwrecks,lying on muddy bottoms, were charac-terized by the highest cover of the bi-valve Ostrea edulis and the anthozoanCorynactis viridis, reaching the 90%of cover (Figure 4a). Subhorizontal,overhanging rough structures favoredsettlement and growth of low-light tol-erant species such as the anthozoansParamuricea clavata, Leptopsamiapruvoti, Eunicella cavolinii, bryozoans(Aedonidae spp.), serpulids, and redencrusting algae (Corallinales spp.), asshown in the upper right side of thediagram (Figure 4a). In contrast,other species such asCaulerpa racemosav. cylindracea, Chartella sp., Codium

al

vermilara, Flabellia petiolata and Goio-cladia sp. clustered around the centerof the PCA diagram were only foundoccasionally. The aluminum hullof the BR.20 bomber was dominatedby sponges (Dictyoceratida sp. 2,Aplysina aerophoba), and one bryozoan(Celleporidae sp.) was found (Fig-ure 4a); see the lower left quadrant ofthe PCA diagram. The scatterplot ofrow scores (wrecks) confirmed theordination of variable; almost all taxagrowing on the BR.20 bomber weregrouped in the lower left quadrant(Figure 4b).

The three types of seabed identifiedduring the present study were mudwith >88 % of fines and sand and det-ritic seabed each with <10 % of fines.Fines included clay + silt followingBertolotto et al. (2005). Neither theamount of fines (R = −0.14; p > 0.05)nor the distance of the wrecks fromthe coast (R = 0.23, p > 0.05) were cor-related with the number of speciesfound.

Data collected on “wrecks” werecompared with those extracted fromthe literature and referring to naturalsubstrata (Figure 6). An increase inthe number of different benthic spe-cies on natural substrata was foundfrom the Western to Eastern side ofthe Ligurian coast. In contrast, thenumber of species colonizing wrecksshowed a weak negative trend fromwest to east. In the cases of theMohawkDeer (114%) andGenova (111%), spe-cies abundances were higher than thoserecorded on the nearby natural sub-strates (Table 2).

DiscussionThe combination of close-circuit

rebreather (CCR) diving with high-definition (HD) video shooting wasuseful in the deep-diving surveys. As

TABLE 1

List of species found on wrecks. Wreck numbers are referred to Figure 1. Numbers for bottom type, material structure, surface typology and slope arereported in material and methods.

Wreck number
1 2 3 4 5
May/June 2

6

013 Volume 4

7

7 Number 3

8

Distance from the coast (nm)
0.5 1.5 0 1 0.9 8.6 2.5 14
Video samples (n)
5 12 16 9 12 10 8 7
Depth range (m)
35–38 46 30–46 52–55 42–60 35–38 38–40 40–42
Bottom type
1 4 4 2 2 2 2 2
Wreck zone
1 1 1–2 1 1 1 1 1
Material
1–2 2–3 2 2 2 2 2 2
Surface texture
1–2–4 1–2–3–4 1–2 1–2–3–4 1–2 1–3–4 1–2–3 1–3
Structure shape
1–2–3 1–2–3–4 1–2–3 1–2–3–4 1–2–3–4 1–2–3 1–2–3–4 1–2–3
Structure slope
1–2–3 1–2–3 1–2–3–4 1–2–3 2–3–4 1–3–4 1–2–3 1–3
Algae

1 Corallinales sp.1
+ + + +
2 Corallinales sp.2
+
3 Gloiocladia sp.
+
4 Peyssonellia sp.
+
5 Caulerpa racemosa
+
7 Codium vermilara
+
8 Flabellia petiolata
+
Porifera

9 Haliclona sp.
+
10 Dictyoceratida sp.1
+ +
11 Dictyoceratida sp.2
+
12 Dysidea sp.
+ + + +
13 Aplysina aerophoba
+ + +
14 Porifera indet. sp.1
+ +
15 Porifera indet. sp.2
+ + + + + +
16 Porifera indet sp.3
+ +
+
+ + + + + +
+
Anthozoa

20 Eunicella cavolinii
+
21 Eunicella singularis
+
22 Eunicella verrucosa
+ + +
23 Leptogorgia sarmentosa
+ + +
24 Paramuricea clavata
+
continued

121

a noninvasive sampling method,video-sampling had no impact on theinvestigated organisms. The only pos-sible disturbance from sampling wasthe bioturbation effect that was pre-vented by using CCR. Bubble emis-sion during the diver’s exhalation,normally occurring with the open


circuit (OC) diving apparatus that en-hances turbidity during video shootingon vertical and below overhangingstructures, was drastically reduced bythe use of CCR. By comparing visualand photo-sampling on macrobenthicorganisms of Mediterranean hard sub-strates, Parravicini et al. (2009) did not

al

find any differences between the twomethods when photographs had a sig-nificant area coverage and adequateresolution. The video procedure usedin the present work assured the anal-ysis of a minimum area larger than0.25 m2, in accordance with methodssuggested for the sampling of Mediter-ranean hard bottoms and artificial sub-strata such as artificial reefs (Weinberg1978; Bianchi et al., 2004, Relini2004; Parravicini et al., 2009), whileHD resolution proved its potential inspecies identification.

In the present study, PCA analysisrevealed that bottom type is fun-damental in epibenthic coloniza-tion of wrecks and it has a key role indetermining marine community struc-tures. A connection between speciesoccurrence and sediment typology isin agreement with findings of Zintzenet al. (2008) and Guerin (2009), whoshowed a relationship among epifaunalcommunities on wrecks and oil plat-forms, and the bottom sediment of

TABLE 1

Continued

Wreck number
1 2 3 4 5 6 7 8
25 Corynactis viridis
+ + + +
26 Balanophyllia europaea
+ + +
27 Leptpsammia pruvoti
+
28 Phyllangia sp.
+
Bivalvia

29 Ostrea edulis
+ + + + + + +
Anellida

30 Sabella spallanzanii
+ +
31 Serpulidae spp.
+ + +
Bryozoa

32 Aedonidae sp.
+ +
33 Watersiporidae sp.
+
34 Celleporidae sp.
+ +
35 Chartella sp.
+
FIGURE 4

PCA of the loading factors between principal components axis (PC1 and PC2): a) scatterplots ofcategorical variable categories (depth, bottom type, wreck material, and structure characteristics)and continuous variables (species); b) scatterplots of wrecks’ scores. Wrecks and species num-bers are reported in Table 1.

the Baltic Sea. In Liguria this is alsotrue for natural, rocky bottoms wherelittoral photophilic communities areinfluenced by substratum mineralogy(Cattaneo-Vietti et al., 2002).

Also, wreck material is fundamen-tal in macrobenthos colonization. Ex-periments comparing natural habitats

and artificial substrata such as con-crete, rocks, ceramic, and earthenwarematerial demonstrated that speciesrecruitment and survival were strictlydependent on the substrate-type(Burt et al., 2009; Perkol-Finkel &Benayahu, 2006; Antoniadou et al.,2010). Relini et al. (2007) showed

May/Ju

in Liguria that wooden barges andshipwrecks were less appropriate forcolonization, in agreement with videoshot on the Cycnus. Iron structures ofwrecks are usually damaged by corro-sion, which first attacks the thinnestparts (Relini et al., 2007) and at theend, may contribute to the collapseof the whole structure (Bell et al.,1997). This damage is enhanced bybubbles emitted by SCUBA open-circuit apparatus and by chains usedto anchor diving buoys. The illegalpredation of wrecks further contrib-uted to their destruction (UNESCO,2001; Tilburg, 2006). Videos of thisstudy showed that the aluminum onthe aircraft BR-20 and the massivestructures such as the propellers,cranes and portholes of the iron shipswere to be the most resistant materialsto the effects of corrosion.

Studying wrecks in the North Sea,Australia and Red Sea, Guerin (2009),Walker et al. (2007) and Perkol-Finkelet al. (2005, 2006) also agreed that thegeometrical complexity of wreck struc-tures plays a key role in determin-ing species colonization. On muddybottoms of the Ligurian Sea, ver-tical surfaces of studied shipwreckswere dominated by Ostrea edulis andCorynactis viridis. The cage-like struc-ture of the bomber BR-20 probablylimited sediment accumulation by al-lowing enough water circulation to re-move sediments. These environmentalconditions determined the settle-ment and colonization of Poriferaand Bryozoa, and the two taxa becamethe dominant filter-feeders colonizingthe airplane remnants.

Wrecks are not only devices for at-tracting divers but play an importantrole in increasing production normallyassociated with bare bottoms (mudand sandy) (Svane & Petersen, 2001;Massin et al., 2002; Church et al.,

FIGURE 5

Plot of variable importance in scatterplot of Figure 4. Variable importance varies from a maximum(1) on the left to a minimum (0) on the right.

FIGURE 6

Number of species found on wrecks (present work) versus nearby natural substrata (from Peirano& Sassarini, 1991; Cocito et al., 2002; Cattaneo-Vietti, 2004).


TABL

E2

Comparison

amongtaxa

compositionof

macrobenthicassemblages

foundon

wrecksstudiedinthepresentw

ork(pw)and

thosereporte

don

naturalsubstratainnearby

sitesinprevious

papers

(1=Cattaneo-Vietti,

2004;2

=Peirano

&Sassarini,1991;3

=Co

cito

etal.,2002).

Reference

number

pwpw

pw(1)

pw(1)

pwpw

(2)

pw(2)

pw(3)

area

Imperia

Imperia

Camogli

Camogli

Paraggi

Paraggi

Fram

ura

Fram

ura

P.Mesco

P.Montenero

P.Montenero

LaSp

ezia

LaSp

ezia

Wreck/sam

ple

number

12

3P0

C4

C45

6–

7–

8–

Method

Video

Video

Video

Photo

Video

Photo

Video

Video

Photo

Video

Photo

Video

Photo

Depth

range(m

)30

4630–46

24–33

30–52

12–16

42–60

3818

–27

3818

–27

4215

–24

Botto

m(sediment)

type

13

34

24

22

42

42

4

Samples

(n)

512

1610

910

1210

78

47

30

Taxa

Algae

11

44

61

13

24

110

Spongiae

23

22

61

97

123

74

23

Anthozoa

16

51

24

17

14

413

Bivalvia

11

11

11

1

Anellida

11

12

13

Bryozoa

11

23

11

21

9

Ascidiacea

21

1

Species(n)

66

1614

109

1711

277

1710

59

Wrecksvs.n

atural

habitat(%

)-

-114

111

6341

4117


2007; Mallefet et al., 2008). Theirimportance in the Ligurian Sea waspreviously highlighted by Relini et al.(2007). The author analyzed 15 yearsof benthos colonization on artificialreefs along the Ligurian coast con-cluding that the concrete blocks actedin the same way as natural, rocky bot-toms. Data from the present workshowed that the abundance of macro-benthos on shipwrecks may be similarand, in some cases, greater than thatfound on natural substrates. Thesefindings are in agreement with theobservations of Perkol-Finker et al.(2006), which reported values forepifaunal colonization on a shipwrecksimilar to those found on neighboringnatural bottoms in the Red Sea. More-over, Church et al. (2007), studyingan historical shipwreck in the Gulf ofMexico, showed not only that the epi-benthos of the artificial structure wasricher than that of the surroundingnatural areas, but also that it could bethe source for the colonization of theneighboring substrata.

Because of their role as historicalattractions for recreational divers, ship-wrecks could become good “sentinels”of biological changes affecting marineenvironments. Large surfaces of ship-wrecks may favor, for example, inva-sive species colonization, as observedby Work et al. (2008) in the CentralPacific and by Perkon-Finkel et al.(2006) in the Red Sea. In this study,the presence of the green alga Caulerparacemosa var. cylindracea, one invasivespecies that colonized almost all theMediterranean coasts in one decade(Piazzi et al., 2005), was recorded onthe Mohawk Deer shipwreck.

Massive wreck structures such aspropellers or cranes, which are moreresistant to deterioration, are easily rec-ognizable in monitoring studies. Bycomparing video of the Mohawk Deer

with previous ones taped 13 years be-fore by recreational divers, in the sameareas of the wreck a reduction in thenumber of the gorgonians Eunicellacavolinii and Paramuricea clavata wasevidenced. These observations couldbe related with the mortality eventthat affected the gorgonian species liv-ing on natural rocky walls in the latesummer 1999 (Cerrano et al., 2000).In order to heighten the general aware-ness of the ecological and historicalimportance of wrecks and to obtainlow-cost data on their associated biodi-versity, ‘wreck monitoring programs’involving recreational divers shouldbe promoted (UNESCO, 2001;Wessex Archaeology Limited, 2008;Underwood, 2009). The increasinginterest in wrecks, mainly due to thespread of high technology divingequipment and low-cost HD photo-graphic and video apparatuses, com-bined with the use of the Internetas a source of information exchange,delivers possibilities to easily gain eco-logical data that will considerablyhelp scientists in monitoring marinebiodiversity.

AcknowledgmentsThe author is grateful to his friend

and rebreather instructor G. Paparofor his assistance during the dives onwrecks, to his colleagues C. Lombardiand S. Cocito for their help withbryozoan taxonomy and the criticalcomments on the manuscript, and toE. Taylor, A. Jochens, and N. Lucey fortheir suggestions and language revision.

Author:Andrea PeiranoENEA-Marine Environment ResearchCentre, C.P. 224, La Spezia (Italy)Email: [email protected]

May/Ju

ReferencesAntoniadou, C., Voultsiadouì, E., &

Chintiroglou, C. 2010. Benthic colonization

and succession on temperate sublittoral rocky

cliffs. Est Coast Shelf Sci. 82:426-32.

Baine, M. 2001. Artificial reefs: a review of

their design, application, management and

performance. Ocean Coast Manage. 44:241-59.

http://dx.doi.org/10.1016/S0964-5691(01)

00048-5.

Bell, M., Buchanan, M., Culbertson, J.,

Dodrill, J., Kasprzak, R., Lukens, R., &

Tatum, W. 1997. Guidelines for marine

artificial reef materials. Artificial Reef Subcom-

mittee of the Technical Coordinating

Committee, Gulf States Marine Fisheries

Commission. Ocean Springs: Gulf States

Marine Fisheries Commission office. 118 pp.

Bertolotto, R.M., Tortarolo, B., Frignani, M.,

Bellucci, L.G., Albanese, S., Cuneo, C., …

Gollo, E. 2005. Heavy metals in surficial

coastal sediments of the Ligurian Sea. Mar

Pollut Bull. 50:344-59. http://dx.doi.org/

10.1016/j.marpolbul.2004.12.002.

Bianchi, C.N., & Peirano, A. 1995. Atlante

delle Fanerogamemarine della Liguria: Posidonia

oceanica e Cymodocea nodosa. La Spezia: Enea-

Centro Ricerche Ambiente Marino. 146 pp.

Bianchi, C.N., Pronzato, R., Cattaneo-Vietti, R.,

Benedetti-Cecchi, L., Morri, C., Pansini, M.,…

Bavestrello, G. 2004. Mediterranean marine

benthos: a manual of methods for its study and

sampling. Biol Mar Medit. 10(1):185-215.

Bohnsack, J.A., & Sutherland, D.L. 1985.

Artificial reef research: a review with recom-

mendations for future priorities. Bull Mar Sci.

37(1):11-39.

Boudouresque, C.F. 1971. Méthodes d’étude

qualitative e quantitative du benthos (en parti-

culier du Phytobenthos). Tethys. 3(1):79-104.

Burt, J., Bartholomew, A., Bauman, A., Saif, A.,

& Sale, P.F. 2009. Coral recruitment and early

benthic community development on several

materials used in the construction of artificial

reefs and breakwaters. J Exp Mar Biol Ecol.

373:72-8. http://dx.doi.org/10.1016/j.jembe.

2009.03.009.


Carta, E. 2000. Navi e relitti tra Montecarlo e

il promontorio di Portofino. Rapallo: Azienda

grafica Busco editrice. 143 pp.

Carta, E., Maggiori, A., & Penco, A. 2008.

Navi e relitti tra Sestri Levante e la Spezia.

Rapallo: Azienda grafica Busco editrice. 143 pp.

Cattaneo-Vietti, R., Albertelli, G., Bavestrello,

G., Bianchi, C.N., Cerrano, C., Chiantore,

M., … Schiaparelli, S. 2002. Can rock com-

position affect sublittoral epibenthic com-

munities? PSZNI: Mar Ecol. 23(1):65-77.

http://dx.doi.org/10.1111/j.1439-0485.2002.

tb00008.x.

Cattaneo-Vietti, R. 2004. Progetto di studio

per la valutazione e valorizzazione delle emer-

genze naturalistiche dell’Area Naturale Marina

Protetta del Promontorio di Portofino. Parte I:

Valutazione della struttura delle principali

cenosi presentinell’area anche allo scopo di

definirne la ricchezza specifica (biodiversità).

Relazione finale. Genova: URL-CoNISMa-

Università degli Studi di Genova - Dipteris.

36 pp. (http://www.portofinoamp.it/it/prodotti-

editoriali/relazioni-tecniche/699-2004-

qvalutazione-e-valorizzazione-delle-emergenze-

naturalistiche-dellamp-portofinoq.html).

Cattaneo-Vietti, R., Albertelli, G., Aliani, S.,

Bava, S., Bavestrello, G., Benedetti-Cecchi, L.,…

Wurtz, M. 2010. The Ligurian Sea: Present

status, problems and perspective. Chem

Ecol. 26:319-40. http://dx.doi.org/10.1080/

02757541003689845.

Cerrano, C., Bavestrello, G., Bianchi, C.N.,

Cattaneo-Vietti, R., Bava, S., Morganti, C.,…

Sponga, F. 2000. A catastrophic mass-mortality

episode of gorgonians and other organisms

in the Ligurian Sea. Ecol Lett. 3:284-93.

http://dx.doi.org/10.1046/j.1461-0248.2000.

00152.x.

Church, R., Warren, D., Cullimore, R.,

Johnston, L., Schroeder, W., Patterson, W.,…

Moore, J. 2007. Archaeological and Biological

Analysis of World War II Shipwrecks in the

Gulf of Mexico: Artificial Reef Effect in Deep

Water. U.S. Dept. of the Interior, Minerals

Management Service, Gulf of Mexico OCS

Region, New Orleans: LA. OCS Study MMS

2007-015. 387 pp.


Cocito, S., Bedulli, D., & Sgorbini, S. 2002.

Distribution patterns of the sublittoral

epibenthic assemblages on a rocky shoal in the

Ligurian Sea (NW Mediterranean). Sci Mar.

66(2):175-81.

Guerin, A.J. 2009. Marine Communities of

North Sea Offshore Platforms, and the Use of

Stable Isotopes to Explore Artificial Reef Food

Webs. Thesis for the degree of Doctor of

Philosophy University of Southampton, Fac-

ulty of Engineering, Science and Mathematics.

Southampton: School of Ocean and Earth

Sciences. 226 pp.

Helton, D. 2003. Intentional scuttling

of vessels as a response alternative. IOSC

papers, 420:1-15. http://www.iosc.org/papers/

IOSC%202003%20a272.pdf.

Kaiser, M.J., & Kasprzak, R.A. 2008. The

impact of the 2005 hurricane season on the

Louisiana Artificial Reef Program. Mar Policy.

32:956-67. http://dx.doi.org/10.1016/j.marpol.

2008.02.006.

Kenchigton, R.A. 1978. Visual surveys of large

areas of coral reefs. In: Coral reefs: research

methods. UNESCO - Monographs on oceano-

graphic methodology, eds. Stoddart, D.R., &

Johannes, R.E., 149-62. Paris: UNESCO.

Mallefet, J., Zintzen, V., Massin, C., Norro,

A., Vincx, M., DeMaersschalck, V., …

Cattrijsse, A. 2008. Belgian shipwreck: hotspots

for marine biodiversity (BEWREMABI). Final

Scientific Report. Brussels: Belgian Science

Policy. 153 pp.

Massin, C.L., Norro, A., & Mallefet, J. 2002.

Biodiversity of a wreck from the Belgian

continental shelf: monitoring using scientific

diving. Preliminary results. Biologie. 72:67-72.

Mirto, G., Pivetta, S., & Spazzapari, G. 2009.

Relitti e navi sommerse. Liguria e Toscana.

Milano: Magenes Editoriale srl. 510 pp.

Parravicini, V., Morri, C., Ciribilli, G.,

Montefalcone, M., Albertelli, G., & Bianchi,

C.N. 2009. Size matters more than method:

Visual quadrats vs photography in measuring

human impact on Mediterranean rocky reef

communities. Est Coast Shelf Sci. 81:359-67.

http://dx.doi.org/10.1007/s10750-006-0450-3.

al

Parrish, F.A., & Pyle, R.L. 2002. Field com-

parison of open-circuit SCUBA to closed-circuit

rbreather for deep mixed-gas diving operations.

Mar Technol Soc J. 30(2):13-22. http://dx.doi.

org/10.4031/002533202787914052.

Peirano, A., & Sassarini, M. 1991. Analisi

delle caratteristiche distributive di alcune facies

di substrato duro dei fondali delle Cinque

Terre (Mar Ligure). Oebalia. 17(2):523-8.

Perkol-Finkel, S., & Benayahu, Y. 2006. The

role of differential survival patterns in shaping

coral communities on neighboring artificial

and natural reefs. J Exp Mar Biol Ecol.

369:1-7. http://dx.doi.org/10.1016/j.jembe.

2008.09.016.

Perkol-Finkel, S., Shashar, N., Barnea, O.,

Ben-Daviv-Zaslaw, R.,Oren,U., Reichart, T.,…

Benayahu, Y. 2005. Fouling reefal commu-

nities on artificial reefs: does age matter?

Biofouling. 21:127-40. http://dx.doi.org/

10.1080/08927010500133451.

Perkol-Finkel, S., Shashar, N.,&Benayahu, Y.

2006. Can artificial reefs mimic natural reef

communities? The roles of structural features

and age. Mar Environ Res. 61:121-35. http://

dx.doi.org/10.1016/j.marenvres.2005.08.001.

Piazzi, L.,Meinesz, A., Verlaque,M., Akçali, B.,

Antolić, B., Argyrou, M., … Ceccherelli, G.

2005. Invasion of Caulerpa racemosa var.

cylindracea (Caulerpales, Chlorophyta) in the

Mediterranean Sea: An assessment of the

spread. Cryptogam Algol. 26(2):189-202.

Relini, G. 2004. Mediterranean marine ben-

thos: a manual of methods for its study and

sampling. Biol Mar Medit. 10(1):267-307.

Relini, G., Relini, M., Palandri, G., Merello,

S., & Beccornia, E. 2007. History, ecology

and trends for artificial reefs of the Ligurian

sea, Italy. Hydrobiologia. 580:193-217.

http://dx.doi.org/10.1007/s10750-006-0453-0.

Svane, I., & Petersen, J.K. 2001. On the prob-

lems of epibioses, fouling and artificial reefs,

a review. PSZN I: Mar Ecol. 22(3):169-88.

Tilburg, H.V. 2006. Japanese Midget Sub at

Pearl Harbor: Collaborative Maritime Heri-

tage Preservation. In: Heritage at risk. Special

edition. Underwater cultural heritage at risk:

Managing natural and human impacts, eds.

Grenier, R., Nutley, D., & Cochran, I., 67-9.

Munchen: ICOMOS - International Council

on Monuments and Sites.

Underwood, C. 2009. Public archaeology. In:

Shared heritage: Joint responsibilities in the

management of British warships wrecks overseas,

ed. Gallagher, S., 93-109. Worverhampton:

University of Worverhampton International

seminar.

UNESCO. 2001. Underwater cultural heritage.

Text of the 2001 Convention. Paris: UNESCO.

http://ww.unesco.org/new/en/unesco/

themes/underwater-cultural-heritage/the-2001-

convention/official-text/.

Walker, S.J., Schlacher, T.A., & Schlacher-

Hoenlinger, M.A. 2007. Spatial heterogeneity

of epibenthos on artificial reefs: fouling com-

munities in the early stages of colonization

on an East Australian shipwreck. Mar Ecol.

28:435-45.

Weinberg, S. 1978. The minimal area prob-

lem in invertebrate communities and the

abiotic environment. Mar Biol. 49:41-57.

http://dx.doi.org/10.1007/BF00390728.

Wessex Archaeology Limited. 2008. Aggre-

gate levy sustainability fund marine aggregates

and the historic environment. Wrecks ecology

2007–08. Report Ref: 57456.02. Salisbury:

Wessex Archaeology. 117 pp.

Work, T.M., Aeby, G.S., & Maragos, J.E.

2008. Phase shift from a coral to a corallimorph-

dominated reef associated with a shipwreck on

Palmyra Atoll. PLoS ONE. 3(8):e2989. http://

dx.doi.org/10.1371/journal.pone.0002989.

Zintzen, V., Norro, A., Massinand, C., &

Mallefet, J. 2008. Spatial variability of epifaunal

communities from artificial habitat: Shipwrecks

in the Southern Bight of the North Sea. Est

Coast Shelf. 76:327-44. http://dx.doi.org/

10.1016/j.ecss.2007.07.012.


The MTS Journal

includes technical

papers, notes and

commentaries

of general interest in

most issues.

Contact Managing EditorAmy Morgante

([email protected])

if you have material you

would like considered.

Specifications for

submitting a manuscript

are on the MTS website

under the Publications menu.

www.mtsociety.org

E-mail:

[email protected]

Or visit our homepage at

www.mtsociety.org

UPCOMING MTS JOURNAL ISSUES CALL FOR PAPERS

July/August 2013Marine Renewable Energy Technologies: Outstanding Papers from the 4th Annual Marine Renewable Energy Technical ConferenceGuest editors: Kenneth Baldwin, Stephan Grilli, Brian Howes,Daniel MacDonald, and Maggie L. Merrill

September/October 2013The State of Technology in 2013Guest editor: Donna Kocak

November/December 2013Technologies and Techniques for 21st Century Wet DivingGuest editor: Michael Lombardi

January/February 2014 Marine Technology Developments in AsiaGuest editors: Dhugal Lindsay and Hiroyuki Nakahara.Manuscripts due end of August 2013.

Check the Society website for future Journal topics.

www.mtsociety.org

C ORP OR AT E MEMBE RSABCO Subsea

Houston, TexasAMETEK Sea Connect Products, Inc.

Westerly, Rhode IslandBluefin Robotics

Quincy, MAC & C Technologies, Inc.

Lafayette, LouisianaCDL, Inc.

Houston, TexasC-MAR Group

Houston, TexasCompass Publications, Inc.

Arlington, VirginiaDOF Subsea USA

Houston, TexasDynacon, Inc. Bryan, TexasEMAS AMC

Houston, TexasFluor Offshore Solutions

Sugar Land, TexasForum Energy Technologies

Houston, TexasFugro Atlantic

Norfolk, Virginia Fugro Chance, Inc.

Lafayette, LouisianaFugro Geoservices, Inc.

Houston, TexasFugro-McClelland Marine Geosciences

Houston, TexasFugro Pelagos, Inc.

San Diego, CaliforniaGeospace Offshore Cables

Houston, TexasGE Energy Power Conversion

Houston, Texas GL Noble Denton

Houston, Texas HARRIS CapRock Communications

Melbourne, FloridaHydroid, LLC

Pocasset, MassachusettsInnerspace Corporation

Covina, CaliforniaINTECSEA

Houston, TexasInterMoor, Inc.

Houston, TexasJ P Kenny, Inc.

Houston, TexasKongsberg Maritime, Inc.

Houston, TexasL-3 Communications

Houston, TexasL-3 MariPro

Goleta, CaliforniaLockheed Martin Sippican

Marion, MassachusettsMacArtney, Inc.

Houston, Texas Marine Cybernetics AS

Trondheim, NorwayMitsui Engineering and Shipbuilding Co. Ltd.

Tokyo, JapanOceaneering Advanced Technologies

Hanover, MarylandOceaneering International, Inc.

Houston, TexasOceanWorks International

Houston, TexasOdyssey Marine Exploration

Tampa, FloridaOil States Industries, Inc.

Arlington, TexasOceanWorks International

Houston, Texas Petrustech Oil & Gas

Houston, Texas Phoenix International Holdings, Inc.

Largo, MarylandPratt & Whitney Rocketdyne Inc.

Beverly, MAQinetiQ North America – Technology Solutions Group

Slidel, LouisianaRaytheon Technical Service Company, LLCSaipem America, Inc.

Houston, TexasSBM Atlantia

Houston, TexasSchilling Robotics, LLC

Davis, CaliforniaSEA CON Brantner and Associates, Inc.

El Cajon, California

Seanic Ocean Systems Houston, TexasSiemens Energy, Inc. Houston, TexasSonTek/YSI, Inc. San Diego, CaliforniaSouth Bay Cable Corp. Idyllwild, CaliforniaStress Engineering Services, Inc. Houston, TexasSubsea 7 (US), LLC Houston, TexasTechnip Houston, TexasTeledyne Benthos North Falmouth, Massachusetts Teledyne Oil & Gas Daytona Beach, FloridaTeledyne RD Instruments, Inc. Poway, CaliforniaUniversalPegasus International, Inc. Houston, TexasWachs Subsea LLC Houston, Texas

BUSINE S S MEMBE RSAanderaa Data Instruments, Inc. Attleboro, MassachusettsAshtead Technology, Inc. Houston, TexasBaker Marine Solutions Mandeville, LouisianaBastion Technologies, Inc. Houston, TexasBioSonics, Inc. Seattle, WashingtonBraemar Technical Services, Inc. (Engineering) Houston, TexasC.A. Richards and Associates, Inc. Houston, TexasCochrane Technologies, Inc. Lafayette, LouisianaContros Systems & Solutions GmbH Kiel, GermanyDeepSea Power and Light San Diego, CaliforniaDeepwater Rental and Sypply New Iberia, LouisianaDPS Offshore Inc. Houston, TexasEmbient GmbH Halstenbek, GermanyEnergy Sales, Inc. Redmond, WashingtonFalmat, Inc. San Marcos, CaliforniaFASTORQ New Caney, Texas Fugro Atlantic Norfolk, VirginiaFugro GeoSurveys, Inc., a division of Fugro Canada Corp. St. John’s, Newfoundland and Labrador, CanadaGAC Shipping (USA) Inc. Houston, TXHorizon Marine, Inc. Marion, MassachusettsHortonGMC Houston, TexasHuisman North America Services Rosenberg, TexasISR Group Savannah, TennesseeiXBlue, Inc. Cambridge, Massachusetts Knight, LLC Lake Forest, CaliforniaInternational Submarine Engineering, Ltd Port Coquitlam, Britsh Columbia, CanadaLiquid Robotics, Inc. Palo Alto, CaliforniaLondon Offshore Consultants, Inc. Houston, Texas Makai Ocean Engineering, Inc. Kailua, HawaiiMaritime Assurance & Consulting Ltd. Aberdeen, United KingdomMatthews-Daniel Company Houston, TexasOceanic Imaging Consultants, Inc. Honolulu, HawaiiOffshore Inspection Group, Inc. Houston, TXPoseidon Offshore Mining Oslo, Norway

QPS Portsmouth, New HampshireQuest Offshore Resources Sugar Land, TexasRemote Ocean Systems, Inc. San Diego, CaliforniaRRC Robotica Submarina Macaé, BrazilSeaBotix San Diego, CaliforniaSeaLandAire Technologies, Inc. Jackson, MississippiSEATECHRIM Moscow, RussiaSeaView Systems, Inc. Dexter, MichiganSonardyne, Inc. Houston, TexasSIMCorp Marine Environmental, Inc. St. Stephens, CanadaSound Ocean Systems, Inc. Redmond, WashingtonStress Subsea, Inc. Houston, TexasSURF Subsea, Inc. Magnolia, TexasTechnip – Global Industries Houston, TexasTechnology Systems Corporation Palm City, FloridaTeledyne Impulse San Diego, CaliforniaTension Member Technology Huntington Beach, CaliforniaVideoRay, LLC Phoenixville, PennsylvaniaW2S2, LLC Houston, Texas WET Labs, Inc. Philomath, OregonWFS Technologies Ltd Livingston, United KingdomXodus Group Houston, Texas

INS T I T U T ION A L MEMBE RSAssociacio Institut Ictineu Centre, Catala De Recerca Submarina Barcelona, SpainCanadian Coast Guard St. John’s, Newfoundland and Labrador, CanadaCity of St. John’s Newfoundland and Labrador, CanadaCLS America, Inc. Largo, MarylandConsortium for Ocean Leadership Washington, DCDepartment of Innovation, Trade and Rural Development St. John’s, Newfoundland and Labrador, CanadaFundação Homem do Mar Rio de Janeiro, BrazilHarbor Branch Oceanographic Institute Fort Pierce, FloridaInternational Seabed Authority Kingston, JamaicaJeju Sea Grant Center Jeju-City, South Korea Marine Institute Newfoundland and Labrador, Canada Monterey Bay Aquarium Research Institute Moss Landing, CaliforniaNational Research Council Institute for Ocean Technology St. John’s, Newfoundland and Labrador, Canada Naval Facilities Engineering Service Center Port Hueneme, CaliforniaNAVFAC Atlantic Norfolk, VANewcastleGateshead Initiative Gateshead, United Kingdom Noblis Falls Church, VirginiaOceanGate, LLC Miami, Florida Oregon State University College of Oceanic and Atmospheric Sciences Corvalis, OregonRichard Stockton College of New Jersey Port Republic, New Jersey Schmidt Ocean Institute Lake Forest Park, WashingtonU.S. Naval Academy Library Annapolis, MarylandUniversity of Massachusetts–Dartmouth New Bedford, Massachusetts

The Marine Technology Society gratefully acknowledges the critical support of the Corporate, Business, and Institutional members listed.Member organizations have aided the Society substantially in attaining its objectives since its inception in 1963.

Marine Technology Society Member Organizations

1100 H Street NW, Suite LL-100Washington, DC 20005

Postage for periodicalsis paid at Washington, DCand additional mailing offices.

focus on deepstar® global deepwater technology - satodev

Documents