0,8-,U,, THE SOCIETYOF NAVALARCHITECTSAND MARINE ENGINEERS. ‘*, 74 TrinityPlace,New York,N,Y,,10006●+’

8 0103 Papertobe Dresm?ed.9?theShipStructureSymposium

: Washin@o”,D.C.,October6.8,1975Z*% 2$>”,.,...*’

Instrumentation-the Only WayJohn W. Boylston, Member, Sea-Land Service,Inc.,Elizabeth,N,J,

Rudolph R. Boentgen, Visitor,Teledyne MaterialsResearch, Waltham, Mass.

James W. Wheaton, Visitor,Teledyne MaterialsResearch, Waltham, Mass,

@ copyright,975L,yTheSoc;ety.fNava,~rch’~ect,,ndMarl.,Engineer,

ABSTRACT

Communication between the theoret-ical analyst of ship structures and thepracticing naval architect can be im-prowed hy the mutual use of experimentalfull-scale data from instrumented ship?,.Such data can prcwide information onseaway loads , ship responses , and thetransfer function between them. lnaddition, full-scale data can be usedto verify or modify theory, investigateOperational problems , and determine ,through a calibration experiment, shipresponses to applied loads in structuralregions where calculations are difficultor impossible. Three examples takenfrom instrumentation projects undertakenon the SL-7 Class Containerships arepresented.

lNTRODUCTION

Tbe title of this paper and today,?,general beading of ‘,Yesterday, s Tech-nology, Today’s Ships” probably sound abit extreme. While these titles may beOverdramatic, the authors feel that theydo express tbe sense of frustration overthe time required for current theory tobe placed in practice, and over the lackof data available to assist a ship de-signer in making judgments as to whichanalytical technique will most accurate-ly predict tbe characteristics of newconstruction.

The main problem, of course, iscommunication. The theoretical analystusually has some disdain for the shi~design practitioner who wants the “analyst ts formulae reduced to plug-in,tabular form so that he, the practition-er, can use them. On the other hand,the practitioner may not grasp thesignificance of the analyst, s advancedtheories, and may lack the motivationto understand them. A median groundllUIStbe Sought , therefore , “here the an-alyst brings himself to a semi-practicallevel and the practitioner attempts topull himself out of the handbooks.One excellent common ground where bettercommunication can result is in the useof experimental data. Fu1l-scale data

collected from a properly instrumentedship presents a set of numbers whichshould be understandable by bothparties.

Collection of full-scale data isnecessary from a number of standpoints.Any predictions resulting from mathe-matical analyses of experimental modelsmust accurately characterize the actualstructure, or must be correctable in aknown way to correlate the technique tothe actual structure. Full-scale data,properly interpreted, provide the cri-teria against which all predictive tech-niques of structural response must bejudged. A second but equally importantuse of full-scale data is to provide anestimate of the input loads which formthe basis of the rational design. Suchloading data can be gathered directlyfrom a characterization of observedservice conditions, such as wind andwave probability distributions, or in-ferred from the response of the vesselto the combination of these conditions.The latter scheme requires a knowledgeof the strwture 1s input-output ortransfer function which again can beprovided by adequate full-scale datadescribing actual loads and responses.1“ sum, full-scale data can providethree indispensable parts of rationaldesign: input loads, responses, andthe derived characteristics of the linkbetween the two.

Full-scale data collection soundseasy, but as with any research projectcertain basics must be applied in orderto obtain credible data. We list thesegeneral steps that should be taken inany instrumentation project:

1. The analytical community shouldspecify where the theoreticalmodeling may be weak and whatdata are needed for verifica-tion or for use as inputs tothe model. In conjunction withthe practitioner a usefulprogram can then be formulated.

2. An experienced instrumentationteam should design and install

#---E-1

.-

the data acquisition systemand follow the project throughdata reduction.

3. Enough data should be collectedt.oanswer the questiozs posedwithin some agreed-upon limitsof accuracy, and to eliminatesecondary or extraneous ln-fluences.

This sequence is exactly what theShip Structure Committee has done in thepast and is doing today. Every ShipStructure Committee project receivestechnical supervision from a Project Ad-visory committee of the National Academyof Sciences/National Research COunCil,which provides inputs from many relateddisciplines. This paper deals with oneparticular current project of SSC: theinstrumentation program for the SL-7Containerships (1, 2).

This pIogram, a.jointly fundedundertaking of Sea-Land Ser”ice, Inc. ,the American Bmeau of Shipping and theShip Strwcture committee, representsan ~xcellent example of coop~ration be-tween private industry, regulatoryauthority and government. The goal ofthe program is to advance understandingof the performance of ship hull struc-tures and the effectiveness of the

analytical and experimental methods usedin their design. While the experimentsand analyses of the program are keyed tothe SL-7 Containers hip and a considerablebody of data will he developed relatingspecifically to that ship, the conclu-sions of the program will be completelygeneral, and thus applicable to anysurface ship structure.

The program includes measurement ofhull stresses, accelerations and environ-mental and operatinq data on the SS SEA-LAND McLEAN, development and installationof . microwave radar wavemeter for meas-uring the seaway encountered by the ves-sel, a wave tank model study and atheoretical hydrodynamic analysis whichrelate to the wave-induced loads, astructural model study and a finite ele-ment structural analysis which relate tothe structural response, and installationof long-temn stress recorders on each ofthe eight vessels of the class. ln addi-tion, work is underway to develop theinitial correlations of the results ofthe several program elements.

It will be some time before the“ariow correlations and comparisons aremade and the final judgments are in con-cerning the relationships of the variouspredictive technique: to the behavior ofthe real ship. The ntent of this paper

Fig. 1 SS SEA-LAND MCLEAN – First SL-7 Containership in a Class of Eight

E–2 .

—.

is to provide a practical perspective onthe use of full-scale experimental dataas an aid to ship design by discussingseveral examples.

THEORY VER1FICATION

One of the prime uses of instrumen-tation is the verification of theory.The analytical community tends to lookupon experimental data as ,,good,,if itagrees with theory, and ,’bad,’if it doesnot. It is therefore important, as men-tioned, that the analytical communityunderstand and accept the method of in-strumentation and its application to theparticular design problem involved.

In order to illustrate the de-signervs concern with theory, we wouldlike to present an actual design problemassociated with containerships of theSL-7 Class illustrated in Figure 1. Oneof tbe peculiar design problems asscm-iated with the containership open hullstructure is the consideration of tor-sional moment effects. While the hul1torsional moment does have a minor effecton the total vertical and lateral bend-ing stress experienced by the upperstructure in tbe ship in open areas, itcauses major detail design problems atthe intersection of transverse hatchgirders and the longitudinal box girders,and at the ends of tbe ship. The gener-ally unrestrained midsection of the ves-sel free to deflect in torsion is, ofcourse, restrained by the decked-overends of the ship. Severe restxaint-of -warping stresses occur at the ends of theship and similar high stresses will befound at the intersection of longitudinalam3 transversal deck structures. Asidefrom the high local stresses, deflectionof the hull structure with relation tothe hatch coaming/longitudinal hatchcovers produces an extremely complicateddesign problem. An excellent RINA paper(3) describes some of these problems asthey applied to the OCL ships. Most ofthese stresses and deflections aredirectly related to the magnitude of thetorsional moment and, more importantly,Its absolute value at any point along thelength of the ship.

With this in mind, we call yourattention to Figure 2 in which the cal-culated values of the torsional momentfor the SL-7 Containership based on themethods of De Wilde and Grim are sho”n.Grimes equation allows more relativewave directions to be calculated, andhis results seem more plausible “ith re-gard to the shape of the ship, whileDe Wilde gives only a simple cosine dis-tribution. These curves have been plot-ted for wave height of 23.3 feet (thelargest wave height required by ABS atthe time of the original calculation)and 38.0 feet for demonstration purposes.Waves have been encountered in excess of38.0 feet, so that additional curves

8

;&‘- DeWILDE

B6

c

4---

, . //;?< - -

0 -’/

. . .\

L18 14 10 6

STATION

Key: A-Bow Sea, 38 ft WaveB-Bow Sea, 23.3 ft WaveC-Quartering Sea, 38 ft WaveD-Quartering Sea, 23.3 ft WaveE-Bow Sea, 38 ft WaveF-Bow Sea, 23.3 ft Wave

Fig. 2 Some Predictions of TorsionalMoment Distributions for theSL-7 Class Ships

could be drawn to cover all cases. Usingthe top curve (Grim, bow sea, 38.0 feet)as an envelope curve, we think it can beagreed that this is a gross a&proxirna-tion at best. With a good finite ele-ment model and a proper torsional momentdistribution, restraint-of-warpingstresses and hull deflections could befairly accurately predicted. As it stoodSOme 5 years ago , howe”er , the finiteelement model predictions could be off bya factor of 3 depending on what theoryand nave height assumption was used.

During the second season of dataacquisition from the SS SEA-LAND MCLEAN,a severe storm with waves up to 50 feetheight “as encountered. In response to alarge slam, a midship torsional moment inexcess of 100 Kft-tons was measured asdetermined by extrapolation from the cal-ibration data. This “as an instance of adynamic stress-producing load measured inservice but nbt covered hy the theoreti-cal predictions. Under the same condi-tions, stresses in the deck just forwardOf the Aft Ho”se (Frame 142) were re-corded which were close to the highestlevels predicted by a Finite ElementMethod (FEM) analysis (see Figure 3).The FEM analysis (4) found this to be themost highly stressed area on the ship.SeaWaY measurements, however, found thehatch corner just aft of the ForwardHouse (Frame 290) , which has a geometri-cally similar cut-out, to exhibit an evenhigher stress. It is interesting to notethat although no analysis combining ver-tical, lateral, and torsional loading pre-dicted high deck stresses in this area,it is the only point of structural hullfailure encountered on all eiqht vesselsof the class. There are seveial reasonswhy these stresses are highex on the real

Longitudinal Stress

lnb td

FYB

\,

—FR 142 FEMAnalysis

-s-FR 142Measured

-A- FR 290 Measured(Longitudinal)

–m. FR 29o Measured(Circurnferen-

Ivtial)

1000 2000 kg/cm2 (approx)

~-O;i

Fig. 3 A Sample Comparison BetweenFEM Stress Prediction andMeasured Stresses for ‘IWOHatch Corners

ships than in the analysis:

3.. Secondary dynamic wave loadingsdue to high ship speed in heavyseas may produce high short-termstresses throughout the bowarea. Since the forward end ofHatch No. 1 has the leastmodulus in this area becausethe box girder area is reduced,momentary stress extremes dooccur.

2. Warping stresses due to tbe rollof the vessel in a seaway may beincreased by superposition oflateral bending mode stresses.

3. Although more prevalent in aquartering sea, the directionalshearing of the bow and result-ing compensating steeringmaneuver may add to lateraland torsional moment.

With these factors in “ind, thethird season data acquisition programwas restructured to investigate some ofthese areas. Very high circumferentialstresses at the hatch corners were meas-ured as shown in Figure 3. Perhaps evenmore significant is tbe frequency of oc-currence of these high stress levels,

and their correlation to factors otherthan wave height and direction. Simul-taneous analog magnetic tape recordingallows expansion (or compression) of thetimebase of such data for interpretation.?+nexample of such a record from thethird season data is presented in Figure4. During this period quartering wavesand swells were only three to five feetin height. Yet, peak-to-trough stresslevels (excluding still-water components)were frequently measured in excess of 30Ksi. Such high stresses, so frequentlyexperienced, could induce fatigue crack-ing failures even though their absolutelevels may be well below yield. Whencracking did appear in these areas, itwas possible, based on the knowledge ofthe recorded data for outboard locations,to demonstrate that these were self-limiting localized phenomena and not thefirst indication of a possible catas-trophic failure.

A further review of data such aspresented in Figure 4 provides insightrelative to what is causing the highdeck stress levels. Notice the high cor-relation between roll, rudder angle, andstress. Apparently, due to the quarter-ing sea, course corrections induce rollswhich, due to the extremely fine bow,cause significant combinations of tOr-sional and longitudinal horizontal “bend-ing. It is expected that an understand-ing of this interaction mechanism willresult, not only in lower stress levels,b“t in revised operating procedures forimproved motion control and steering.

The authors feel, therefore, thateven with the degree of analyticalsophistication available today notenough is known about the dynamic load-ing any one particular ship encounters.Moreover, it is doubtful that the com-plete loading spectrum of any one vesselwill ever be known. If the designer hasdetailed design problem areas, instru-mentation of preceding ships is “the onlyway” to obtain order-of-magnitudestresses and deflections. New ship de-signs can then be estimated more accu-rately by using recorded experimentaldata and modifying to fit, or by theore-tical extrapolation of the recorded data.With sufficient data, of course, newtheory can be postulated with increasedconfidence.

The cost, carrying capacity, speed,and other parameters of today’ s shipsare so sensitive to ship weight that thedesigner can no longer afford to designon yield strength with a safety factorof 5.0. The stress values shown in thispaper indicate that on the SL-7, interms of stress level, not too muchmaterial has been wasted in the uppergirder structure of the ship.

E-4

-4040 psi

(c)

+

(D)

(E)

‘F

.?=j.: (A) Midship Longitudinal Vertical Bending Stress (E) Roll Angle

(B) Forward Longitudinal Vertical Bending Stress (F) Rudder Angle

(C) Midship Longitudinal Lateral Bending Stress (G) Hatch Corner Gage FYB

(D) Torsional Midship Shearing Stress (+ CW Fwd) (H) Hatch Corner Gage FYC

Fig. 4 Sample Analog Traces for One Instant of Ship Responseto Quartering Sea with 3-5 Foot Waves and Swells

E-5+

-.’

1NVBST1GATION OF OPERATIONAI

A second general use oftion is the determination of

PROBLEMS There are an unbelievable number ofvariables which one could speculate as

instrumenta - being associated with this problem, suchthe dominant as:

variables in a complex design problem.The example we would like to use here is 1.the strut bearing and shaft problems en-countered with these new SL-7 container-ships.

As originally designed, the inter-mediate strut bearings (see Figure 5)were a water-lubricated phenolic type,while the main strut bearings were (andare) oil-lubricated Babbitt. Some wiping 2.of the main strut bearings after onlyshort service was attributed to an over-loading of this oil bearing caused byexcessive wear of the intermediate phe-nolic bearing and the resulting change inshaft alignment. The intermediate bear-ing was then converted to an oil-lubri-cated Babbitt type. Although this 3.particular problem was solwd in this man-ner for some time, considerable oil leak-age from the strut bearing seals is nowexperienced, and thus major bearingproblems still exist. 4.

The change in shaft alignmentbetween low- and high-poweroperation, and the change inbearing loads, strut and shaftstresses, shaft and strut de-flections and bearing sealpressures associated with thischange in alignment.

The dynamic load and deflectionin the shafting and bearings re-sulting from resonant and off-resonant response of the shaft-ing system to the harmonicforces and moments generatedat the propeller.

The relative movement of theshaft seals resulting from thefore-and-aft dynamic motionsof the strut.

The effects of ship motion anddeflections in the seaway upon

‘\

\\\ Fw~nml

Key , (a)(b)(c)(dj(e)(f)(9)

Hull Penetration (2 places) (h)Ribbon Cable Bonded to strut (i)Intermediate Strut (j)Aft Bearing Seal (k)Intermediate Shaft (1)Fwd Main Bearing Seal (m)Main Strut (n)

Fwd FairWater

Fwd Bearing SealBearing Ho”singAft FairwaterAft Main Bearing SealRope GuardPropeller

Fig. 5 Schematic of Outboard Instrumentationfor Shaft Seal Investigation

E-6

i .

.-

bearing reactions and shaftstresses.

5. The influences of the hullstructure to which the strutsare attached upon the stiffnessof the struts.

6. The changes in bearinq pres-sures and journal clearancesresulting from the shaft carry-ing different steady radialloads and steady moments anddeflections at slow speed andat rated speed.

7. The shaft motions and bearingpressures generated by therotating shaft excited by har-monic radial shaft forces andharmonic moments as applied tothe main and intermediate bear-ings.

8. The thermal stresses associatedwith a shaft having a high heatinput for a given length andcooled by water on both sidesof the heated section.

9. The dynamics at a bearing of s“d-denly stopping the shaft, andthe stresses in shafting assoc-iated with such sudden stopping.

Analyzing all of these variables andconservatively coupling (probably adding)them would produce loads , deflections andenough suspected problem areas to throwterror into the hearts of owners. It ISpossible that major problem areas wouldbe identified, but the actual valuesneeded for corrective design could not beobtained. Instrumentation showed thatmany of the above factors were minimal.More importantly, it showed an internallube oil pressure oscillation maqnitudethat could not be explained initially.

As noted above, many hypotheses hadbee” advanced to attempt to explain thefailure of the seals and bearings. Amongthe most reasonable of these was theexistence of various critical vibratorymodes of the shaft and various combina-tions of bearing oil pressure and varyingseawater pressure. Based on thesehypotheses an instrumentation system wasdes~gned (see Figure 5) which incorpo-rated extensi”e pressure instrumentationand some displacement sensors.

Although the conditions which werethought to exist in the area of thebearings could be monitored by conven-tional gauging, it was decided thattransducers capable of extended fre-quency response and a larger dynamicrange would be good insurance against

Fig. 6 Completed Transducer Installation on Struts

E-7d.

unexpected conditions. Pressure trans-ducers capable of responding to fre-quencies ten times greater than expectedand pressure five times greater thanexpected were selected. In addition,the noncontacting proximity sensors usedhad a linear range twice the nominalbearing clearance, and a usable fre-quency response in excess of ten timesgreater than anything which could beencountered in service.

Since the struts were external totbe hull in an area which could be sub-jected to cavitation due to the highship speed, a special method of hard-wiring the transducers to the instru-mentation inside the hull was required.After some laboratory experimentation,it was judged that adhesive bonding ofribbon-type cable along the bearinghousing and up the strut to a hullpenetration away from critical cavita-tion areas would offer the best trade-off between reasonable protection forthe wiring and minimum surface profiledisturbance. The completed installa-tion is shown in Figure 6.

ln addition, a system of installa-tion techniques and sealing was de-veloped which met all requirements foradeauate Protection. schedule considera-

Data were taken during a regularPacific crossing from which tbe follo”-ing conclusions were de”eloped:

1. The lube oil pressure, at theends of the main bearing, oscil-lated at a frequency exactlyfive times the shaft rotationalrate (propeller blade frequency) .The magnitude of the oscilla-tion varied with shaft speedand the applied oil bearinqpressure. Below 80 RPM tbeoscillations were not signi-ficant (see Figure 7).

2. The seawater pressures observedon the forward face of thebearing seal were greater thanthose on the after face, andthis differential increased withincreasing ship speed.

3. Oscillatory transverse andvertical deflections of theshaft relative to the mainbearing seal housing were smallcompared to the bearing clear-ance and no resonant modes werenoted. The motion of the shaftcenter was determined at allspeeds including the conditionof tbe shaft on iackinc! uear.. .

tions and operational secuxity.

MEASUREMENT(at 132 RPM At Intermediate Strut At Main Strut

and 33 knots)Forward Bearing Seal Aft Bearing Seal

Shaft *Vertical 0.001Displacement o.o+o-’’yJvw

(inches) -?-

ShaftLateralDisplacement 0.0$ ~ o.o$o~

(inches)

o

Sea Water 10

f

,jy~~, :“

Pressure(psi) 20 i —,— 20

0InterSeal 4Cavity

‘re’su~si) 6f T—8

0Bearing Sump loLube OilPressure

f — i;~

(psi) 20

Fig. 7 Measured Displacement and Pressure Data During 33-knot Run

E-8

4. Steady state shaft position issignificantly affected by shipmotion and changes of shaftspeed in the lower RPM ranges.

The experimental data showed muchhigher pressure fluctuations in the mainstrut as opposed to the intermediatestrut . As longitudinal movement of theshaft would have an equal effect in bothbearings, it was deduced that an un-expected condition existed: that themain bearing housing oscillated withrespect to the shaft in the axial direc-tion. Due to the geometry of the bear-ing assembly, severe cyclic fluctua-tions in the bearing oil p=essure weredeveloped. Such conditions were appar-ently beyond the acceptable range ofpressures which could be endured by theseal. In addition, due to the design ofthe lubrication system it was not pos-sible to diminish these pressure fl”ctwations. Decreasing the internal oilpressure would allow water into the bear-ings; increasing the oil pressure onlyresulted in a heavy loss of oil.

taken aimed at determining the loadsand deflections which existed at themain bearing. A set of strain gages wasused to instrument the main strut duringa later drydocking of another ship inthe series.

Before this installation was made,however, a retrofit of the problem areawas designed which took out the strutoscillatory forces by connecting all ofthe struts to tbe stern tube. A similararrangement of instrumentation of pres-sure transducers and strain gages wasinstalled on this second vessel. Asexpected, the previously observed oilpressure fluctuation conditions werefound not to exist when this vessel wasagain placed in service. After a weekin service, however, the weld holdingthis new tube restraining assemblyfailed allowing again partial movementof the main strut. upon monitoring thereactivated instrumentation, it was foundthat the oscillations in lube oil pres-sure had returned (Figure 8). The straingages provided additional valuable in-formation.

As a result of these findings, afurther experimental effort was under-

NEASURENENT(at 133 RPM Before Weld Cracking After Weld Cracking

and 33 knots) (voyage 55E) (voyage 59W)

Stress in L L

Leading Edge 5 ksi

of Main Strut5 k’i~

T-i-

Stress in -i- ~APTZ;fi1

Trailing Edgeof Main Strut 5 ksi 5 k’i~

T T

Sea Water 1L

Pressure25

5(psi)

~~

-r

lnterseal-i

CavityPressure (psi) +

H ‘:ti75

* LBearing SumpLube Oil

255 ~~

Pressure (psi) ~X5

Fig. 8 Comparison of Measured Data Taken Beforeand After Weld Failure in Restraining Tube

E-9

Lessons to be drawn from the aboveexperience include:

1. Unless the problem is absolute-ly defined, an excessive speci-fication in instrumentation canbe valuable.

2. When laying out an instrumenta-tion system, as much data aspossible should be gathered.The additional cost, once anyinstallation is made, in addedchannels or capability isusually minimal.

3. The recording medium should beas flexible as possible. Al-though a manual reading of ameter is sometimes adeauate,any waveform or dynami~ datais lost. Similarly, althoughan oscillograph will recordrelative phase and waveformdata, no time-base expansion ispossible later, and such a for-mat does not lead itself tofurther automated data analysis.Analog magnetic tape recordingis an excellent method when thenumber Of data channels isrelatively small and there isuncertainty about the nature ofthe data.

4. A thorough knwledge of state-of-the-.art in instrumentationtechniques along with an ap-preciation of tbe system beingmeasured is essential for goodresults and a cost-effectiveprogram.

5. An on-board observer (for re-pair, maintenance, and en-~:gyu~tal observations) is

, if not essential, forone-shot efforts involving aplanned experimental program.Programs can also be conductedfolly automatically if plansprovide for the acquisitionof the other data (shaft RPM,lube oil feed conditions, etc. ).

CAL1BRATION OF THE SS SEA-LAND McLEAN

This experiment is a third illus-tration of the uses of instrumentationdata in ship design. Comparing meas-ured ship stress variations with valuescalculated to result from a known changein loading can provide basic informationof great value. And, measuring theresponse in areas “here calculationsare impossible can provide informationotherwise unobtainable. In addition,a calibration of an instrumented shipwill provide verification of instrumenta-tion system performance, and thus ofthe accuracy of seaway data.

——

The McLEAN calibration experimentwas conducted on 9-10 April 19;3 inRotterdam (5). Six loading conditionswere specified in the course of unload-ing the ship in a manner which maximizedboth vertical bendimg and torsional loadchanges. At each loading condition allstrain gages were recorded, .?.swere theexisting container “eights and locations,and the drafts along the ship. Calcula-tions of vertical bending moment andtorsional moment for each condition wereprovided by the American Bureau of Ship-ping, and the related stress changes havebeen compared with the measured values inFigures 9 and 10.

4

1ksi O ASS Calculated Data

3 . TMS Measured Data

P(L.C. 2 Omitted due to A

2schedule constraints) ,j

~

g,,$”

,Y””:

0~

l\3m /4567

-1 ‘\ ,’LOAD CONDITIONw

-2J

Fia. 9 Chanae in Measured and Calculated.Midsiip Longitudinal VerticalBending Stress vs. Load Condition

40

kft-t ,j“\.30

20

10

0

Fig. 10

-1.5 II\

ksi ,’ ,/

~ -1.0

2!3

-0.5

:1.0

-1.5

---- ABS Calculated Moment---- Measured TSM Stress+ Measured LHB Stress

Measured Torsional Shear Mid-ships (TSM), LongitudinalHorizontal Bending (LHB) andCalculated Torsional Nomentvs. Load Condition

E-10 +

ln the absence of detailed section-al information suitable for calculatingshear stresses using the calculated tor-sional moments, the moments themselveshave been plotted along with the meas-ured shear data. The comparison isgenerally good. Virtually no output isindicated until the start of the torsion-al loading condition. Although there isno change in the horizontal bendingsensor output for Conditions 1 through 4,an increasing output is indicated forConditions 5 and 6. This correspondsto the torsional stress distribution(restraint of torsional warping result-ing in symmetrically opposite normalstresses about the centerline and tor-sional neutral axis) . It is also pos-sible that some of this is d“e tothermally-induced horizontal bendingwhich is restrained by the constant-temper.atureship bottom.

There is a linear relationship be-tween calculated vertical bending stresschanges and measured values, with meas-ured values consistently about 80 per-cent of these calculated, a?,shown inFi.guxe11.

4.W

ElOH 3 /~w /~ ~ ksi~ti 2 //67

~~

51 -LC6~/

u . C5LC47 /

-2 -’/” LC1 1 2 3 ksi4 5

LC3 ./ -1 -CALCULATED STRESS

/

-21Fig. 11 Midship Longitudinal Vertical

Bending Stress: Measured vs.Calculated

The instrumentation of the MCLEANwas planned so that the full-scale meas -uxments could be compared with Variousexperimental and analytical models.The calibration calculations, however.,provided data for comparison at only afew transducer locations. Never the.less, the response of othex strain gagesto the applied loadings is of greatinterest.

Most of these gages were placed inthe regions of hatch corners, especiallyat transitions from a wider to a nar-rower hatch width. For example, a setof strain gage rosettes was placed pork

and starboard just aft of the ForwardHouse, near the Hatch No. 1 corners.Since all of the cargo used to apply thevertical bending and torsional momentswas aft of this section, one might ex-pect negligible stress changes. Rela-tively significant longitudinal stresschanges were exhibited, however.These are associated more with therestraint-of-warping stresses than withthe bending moment changes . Both theForward and Aft Houses restrain thefree action of the open Cell torsionaldeflections, thus giving rise to signi-ficant (in comparison with thoseinduced by vertical bending) longitudinalstress components. These components areespecially important at hatch cornersnear the house structures because thehouse structure geometries further in-crease their magnitudes.

Three gages (SY) were located cir-cumferentially about the hatch cornerreinforcement on the starboard side justforward of the Aft House (Ratch No. 9).The first of these gages, SYA, displayedthe highest recorded strain change ofany gage during the calibration. Thisgage was located 22 1/2 degrees fromthe longitudinal direction around thecutout ring. These gages were installedespecially for the calibration, andwere read with a strain indicator.

As presented above in this paper,significant stress variations have beenrecorded underway from a number ofnew gages installed around the peripheryof the hatch corners during the thirdseason of data acquisition which endedin March 1975. Complete informationabout all underway data will be pub-lished by the Ship Structure Committeein the repoxt of the data reductionfrom the third season.

The calibration of the McLEAN hasgenerated data which not only givesconfidence in the seaway data, but alsoprovides information on loads andresponses not otherwise obtainable.The relatively high stresses measuredat the hatch corner during the calibra-tion accentuated the concern for thestructure in this region, concern whichwas justified by a subsequent historyof cracking. The available seaway datanow provide a measurement of the actualstxess levels being experienced, valueswhich the theoretical predictions didnot show.

CONCLUS1ON

This paper has presented threeexamples indicating that there areseveral uses of experimental data inpredicting and understanding the be-havior of real ships. No amount ofcomputer or model studies can everprovide the level of confidence in theanswers that such real data under real

E-n &--

,,-

conditions provide. We firmly believethat instrumentation is “the only way”in many cases, and ur~ that experi-mental data be incorporated as swiftlyas possible into the notebooks of boththe academic researcher and the prac-ticing naval architect.

REFERENCES

1. R.

2. R.

3. M.

4. A.

5. R.

A. Fain, “Design and instal-lation of a Ship ResponseInstrumentation SystemAboard the SL-7 Class CcIn-tainership SS SEA-LANDMcLEAN,,, ship StructureCommittee Report SSC-238(SL-7-1), 1973.

R. Boentgen, R. A. Fain, andJ. W. Wheaton, ,!FirstSeason Results from ShipResponse InstrumentationAboard the SL-7 Container-ship Ss SEA-LAND MCLEANin North Atlantic Ser”ice, ,,Ship Structure Committee Re-port.

Meek, R. Adams, J. C. Chapman,H. Reibel, and P. Wieske,“The structural Design ofthe O.C.L. ContainerShips, ” Paper No. 4, RoyalInstitution of NavalArchitects, Spring Meet-ings 1971.

M. Elbatouti, D. Liu, andH. Y. Jan, “StructuralAnalysis of SL-7 COn-tainership Under CombinedLoading of Vertical,Lateral and TorsionalMoments Using Finite Ele-ment Techniques, ‘uShip Structure CommitteeReport sSC-243 (SL-7-3),1974.

R. Boentgen and J. W. Wheaton“Static Structural Calibra-tion of Ship ResponseInstrumentation system

Aboard the Ss SEA-LAND

McLEAN, ” Ship StructureCommittee Report.

E-12

D1SCUSS1ON

Pin YU Ch.ng, Member‘rbeauthorsas well as the membersin the

Ship StructureCommitteehave conducteda muchneeded and useful investigation for the Societyof Naval Architectsin the Sea Land 7 instrumen-tationproject. This paper is a usefuladditionto the literatureand 1 expectthatmuch moreinformation will be forthcoming as the project

p~..e.d,.1 fully agree with the authors that com-

munication between the theoretical analyst ofship structures and the practicing naval archi-tects c.. be improved by using experimental full-scale data and that full-scale data properlyinterpreted provides the criteria agaiust which.11 predictive techniques of structural responsenmst he judged. There are,however,severalpoints1 would like to bring o“t for fu.rh.rdism,,im.

First,1 do not agreethat the theoreticalanalysthas disdainfor the designerwho preferssimpleformulae. Membersof PanelHS-3 are nowworkingon a manualfor structuralstability.‘l’hemain goal is to providesimpleformulae,tablesor chartsfor the designers.

secondly,1 do not agreewith the authorsthat the analyticalcomnmnitytendsto lookuponexperimentaldata as “good”if it agreeswiththeoryor as “bad”if it doesnot agree. A typi-cal exampleis the stabilityof shells. ~eo-retfcalstresspredictionsare wually muchhigherthanexperimentalresults. But n. analysthas ever said that the experiments are bad.

ln my opinion, an experiment is bad if itis plamed improperly andlor executed improperly,not became. the correlation i. poor.

At the present stage of ship strengthanalysis, we still don’t know much about a nur–be. of irnporta”tfactors about the structuralloading and response of ships. The only way toincrease our knowledge about these importantfactors is for both the analysts and the experi-me”talists to closely work together. Experi-ments or instrumentationwithout the guidance ofthe theory ha. little value. Theory withoutthe guidance of experfmencs is impractical.Detailed and careful interpretation of thecollected data is also essential to obtaincorrect conclusions.

According to the author. “It will he mm.time hef.re the various correlations and com-p..i..n. ... m.de .nd the fi..1 judg.m.nts arei“ concerning the relationships of the various

p..di.tive t..hniqu.s to the behavior of thereal ship.‘, It is difficult to mderstand howthe authors can conclude that instrumentation isthe cmly way if the results have not been

analyzed.one of the reasons which the a.tho.s used

to draw this conclusion is that the theoreticalprediction. hy ABS are different from the s.a-w.y m..s.rements. 1 would like to ask theauthors whether these diffe.e”ces ..... underthe same loading conditions? It seems to methat O“lY 6 loading conditicms have he.. .on-sidered in the predictions and the experirrmtscover many more loading conditions.

ln fact very good correlation has been.htained between finite element analysis andstrain-gage mess.renrats i“ the past by ABSand Det Norske Verita. for other ships. TheEs.. Norway is .“. example.

It is true that the theory for predictingthe seaway loading i. not yet very accurate.But the qwesti.” is, can instrumentation alonesolve this problem?

In claiming that instrumentation is theonly way, the authors seem to have neglected thedifficulties and limitations of experiments.

First, .11 the measurements from strain

gag.. .re Only relative. me data shOws OnlYthe change in strainfromthe calibrationcondi-tion. Secondly,if the absolutestressis be-yond the yield pointof the steel,the st,e.,–strainrelaticmdepend. also on the loadinghistory. 1“ other words, if the absolute stressat one joint w.. -25,000 psi, then what i. in-dicated hy the measurements to be 50,000 psi isreally O“lY 25,000 Psi tensile. similarly .point with measured stress of 10,000 psi may besubjected to a“ absolute stress of 30,000 psi.TMs is .“ the assumption that all strain gagesperformed properly.

Thirdly, there is the factor of residualetress which cannot be predicted by the theoryno. by the instrumentation. With the.. limita-tion. it seems to be too early to say thati“strumentation is the only way.

IIIview of the.. diffic”ltie, and ““certain-ties, it is i-ecomended that all the collecteddata be s“bjected to .aref”l study and properinterpretation not only from the theoreticalvie”point but .1s. from the practical point of“iew. This is no easy task. The interpret..must be able to separate the effect of manyfactors, and must have profound knowledge aboutthe relationships hetwee” all the factors in-“olved. The SSC, Sea Land Service, and Teledynehave done a service to the industry in providingus with these valuable data. But we have t.interpret and organize them properly before wecan “se then for the benefit of the shipbuildingindustry.

E.]3 !---

.,.

7 -—

AUTHORS’CLOsuRE

We wish to thank the contributors of boththe written and oral comments on ... paper.Their remarks present an opportunity to clarifyand amplify Cbe points which were not adequatelycovered in the paper due to the requirements ofbrevity.

It was not our intention to imply an adve.-sary relationship between the theorist and the

p...tingng naval architect. we wish to .mpha.size, however,that the ship designerrequirestimelyanswersthathave a high degreeof con-fide”... The work of the HS-3 Panel i“ ration-alizing design data will be . large step in therixbt direction.h,hen analysis technique is used,designersrequirementsare “ot being met. I’&-exmple ofthe shafting instrumentation given in the paperi. such a case. Experirme”talf“11.scale meas-urements do not involve modeling assumptions, donot rely o“ theoretical approximations and doinclude .11 factors whether explicitly retog.nized or not. Such was the case in tbe examplep.es..ted; a phenomenon was present which hadnot bee” predicted or even included in themodels. Of course, if poorly designed or ifthe scope of the problem is not appreciated,eve” full-scale instrwnentatio” can he mislead–ing. Here is the plate for close cooperationbetween the theorist a“d experimenter.

1“ tbe matter of shell theory, the consi.t–e“tly lower buckling loads observed in experi-

ments is . reflectionof the “red wo=ld,,, .=what is attainablein practice,regardless ofthe as..mptions made by theory with respect to*.iCial straightness,end fixlty,miformity ofloading,etc. 1“ the ff”al analysis, tbe.qshould predict what is observed in .eaI Iife ina way that is useful. If theoty cannot ,. ~re-dict and tbe actual structure is available, the”instrumentation i. the only way to obtin real-istic data.

Whether or “ot the loading c.nditiom as-sumed i“ the finite-elem.nt analysis matchedthat measured i“ a seaway is not tbe point. Tbelocation a“d magnitude of the maximum stresses

p.,di.t.d w... different from those actuallyobserved, Additiomslly, the magnitude a“dfrequency of the cyclic stresses “.., the hatchcorners were bigb e“o”gh to induce fatiguecracking eve. under s.. conditions much lowerthan considered i“ the FEM solution. If thedesigner had relied exclusively o“ such pre-dictions, without regard for their limitationsand sboxtcmni”gs, serious errors would have beenintroduced. Wlile the FEM is a powerful a“di“v.luable tool, it, too, must b. tested againstthe criteria of its ability to reliably predictdesign parameters, regardless of the a.s.mptio”si.here”t i“ making the a“.lysis. Here againi“strmw”tation provides the a“swe.s for thejudgments and feedback for further developmentof theory,

A referer,ceby a discusser was made toresidual stress which, it was said, could n.t b.p.edi.ted by tbe.ry or instrumentation. I“ fact,certain techniques, such .s trepan”i”g, ..”determine the residual state of stress. Son.ND’Itech”iqwas can ,1.. be used to measure

.esid..1 surface stresses. Perhapsthis is anexampleof whereknowledgeof availabletech-nique.by an experimentercouldaid the theorist.

1. sum, the authorsdo not refrainfromusingtheory,especiallywhen it indicatestbePar~eterS influencinga problem. And, ofcourse, nmcb theoreticalanalysisis well-proven(e.g.,al..lationof ‘rstatic”midshipbendingmoments)and i“ “o med of f.rtberrefinement.We do maintain,however,that formany of tbepr.hlemsconfronting tbe naval architect, theoryin many cases is i“s.fficient or too ctm2ber-s.me, and past experience or current instrumen-tation is, in practice,the ,,onlyway,,.

ESRATA

On page E-3 of o“. paper, the units ofthe vertical axis of Fig. 2 should be in tensof Kft-t or 10 Kft–t (instead of Kft-t). On

Page E-10, the horizontal axis of Fig. 10should be labeled in a manner identical to thati“ Fig. 9.

E.14 +

“,