global offices - cd-adapcomdx2.plm.automation.siemens.com/.../files/magazine/pdf/dynamics2… ·...

Europe Asia-Pacific

CDAJ Japan37/F Yokohama Landmark Tower2-2-1-1 Minato-Mirai · Nishi-kuYokohama 220-8137 · JAPAN Tel: (+81) 45 683 [email protected] www.cdaj.co.jp

ChinaCD-adapco Japan Co. LtdBeijing [email protected]

CD-adapco KoreaSeoul [email protected]

CD-adapco IndiaBangalore [email protected]

HeadquartersCD-adapco • New York office60 Broadhollow RoadMelville, NY 11747, USATel.: (+1) 631 549 [email protected]

Austin, TXTustin, CALebanon, NHAtlanta, GACincinnati, OHDetroit, MISeattle, WA Tulsa, [email protected]

For S. America - please contactMelville Office

HeadquartersCD-adapco • London office200 Shepherds Bush RoadLondon, W6 7NL, UKTel.: (+44) 20 7471 6200 [email protected] www.cd-adapco.com

FranceParis officeLyon [email protected]

GermanyNürnberg [email protected]

ItalyTurin officeRome [email protected]

Global offices - CD-adapcoAmericas

MalaysiaNumac Systems Technologies S/[email protected]

New ZealandMatrix Applied Computing [email protected]

[email protected]

South AfricaAerotherm Computational [email protected]

TaiwanFLOTREND [email protected]

TurkeyA-Ztech [email protected]

Resellers

AustraliaVeta [email protected]

[email protected]

IndiaCSM Software Pvt [email protected]

Your CAE Partner for Successwww.cd-adapco.com Hurricane

resistant platform designFSIfor membrane structures

Aerodynamicsrace faster / less energy

Airbusventilation system design

Features

Dynamics #272

..::INTRODUCTION Dr. Richard Johns

In the early days of CFD, typical model sizes were comparable tothose used for Finite Element solvers in structural analysis. Overthe years, however, the size of CFD models has overtaken FEmodels. CFD models used for F1 aerodynamics are routinely over100 million cells and, more recently, the feasibility of a 1 billioncell model has been demonstrated. On the other hand, a 1million Degree of Freedom (DOF) FE model would be consideredlarge.

So, does size matter? The short answer is “Yes” – it is well knownthat numerical errors diminish as mesh size reduces and this,together with the need to resolve small-scale geometric detail, isthe reason why local models (the practice of extracting parts of alarger FE model for fine-mesh “local” calculations at higherresolution) are used routinely.

Would it be advantageous if structural analysts could use thesame level of mesh detail as their CFD colleagues? Again, theanswer is “Yes” – and now they can! Would they describe a non-linear stress calculation of an engine structure containingover 4 million polyhedral cells (around 13 million DOFs) runningon a 32 CPU Linux Cluster in 15 minutes a breakthrough?We think they would.

In this copy of Dynamics there are a number of articles on theFlow, Thermal, Stress theme and you will hear the latest if youcan attend one of the 2007 STAR Meetings in Europe, the FarEast or the USA. If you can’t, please tell your Structural Analysiscolleagues in the next office and they can come and hear it forthemselves.

Dr. Richard JohnsAutomotive DirectorCD-adapco

It’s not often that the word “breakthrough” can be used legitimately in thebusiness of powertrain analysis. Having worked for over 35 years in the CAE andsoftware departments of OEMs, engineering consultancies and CAE softwarecompanies I can confidently attest that methodology improvements generallycome through the long and usually painful process of evolving current techniquesbased on shortcomings revealed through testing.

Does size matter?Introduction by Dr. Richard Johns

! EMAIL [email protected]

Would they describea non-linear stresscalculation of anengine structurecontaining over 4million polyhedralcells (around 13million DOFs)running on a 32 CPULinux Cluster in 15minutes abreakthrough?

We think theywould.

“

”

Introduction01 Does size matter?

Introduction from Dr. Richard Johns

Product News02 STAR-CD 4.02:

Flow, thermal & stress simulation in a single solver

03 Best practicesGuidelines on CAD for Simulation

Flow, Thermal Stress04 - 05 Simulation in STAR-CD V4

06 - 07 Water jacket optimization using CFD & FEM

08 - 09 FSI for membrane structures

10 - 12 Approaches to Industrial Fluid-Structures Interaction

Workshops13 Ongoing Workshops

Oil & Gas14 - 15 Computer flow simulation

Providing new insight into hurricane resistant platform design

16 - 17 Repair sensitivity studyfor compressor inlet labyrinth seal using Computational Fluid Dynamics

18 - 19 Safer flare design with CFD

Aerospace20 - 21 STAR-CCM+

enables ICON to develop complete-cabin CFD simulation process for Airbus

22 - 23 Flapping Airfoil Analysis of Micro Air Vehicles using STAR-CD

Sport24 - 25 CD-adapco help Felt Racing

to design “the most aerodynamic, UCI-legal bicycle frame ever created”

Biomedical26 - 27 STAR-WORKS helps VIASYS healthcare breathe easier

Pharmaceutical28 - 29 Medicine Manufacture with STAR-Pro/E

Manesty

Chemical Process30 - 31 Operational Optimization of a Municipal Waste Incinerator

32 - 33 Big in Japan

Building & Environment34 - 35 Large Eddy Simulations with STAR-CCM+

36 - 37 Urban Scale Weather Modeling NCAS and CD-adapco Partnership

Regulars38 Upcoming events

39 Training at CD-adapco

40 - 41 Dr Mesh on Modern Art & Surface Wrapping

04

Contents03

22

14 18

08

24

10

32

20

Why?Because the requirements of CAD for prototyping/manufactureand CAD for simulation often conflict. To give one example, verysmall features – like bolts or the maker’s logo – are needed formanufacturing whereas in the CFD model they merely increasethe computational requirement without influencing results.

Long having recognized the issues of CAD with simulation, CD-adapco has gone about resolving them through softwaredevelopments like their CAD-embedded solutions (STAR-CADSeries) and surface wrapping. However, significant improvementsto simulation efficiency can be achieved through intelligent use ofthe CAD tool’s own functionality. Bring on the Best PracticeGuidelines.

The Guidelines are aimed both at engineers who are new to CFDand want to know more about its CAD requirements, and atsimulation experts who want to know more about CAD. Througha series of simple examples the user is guided through a numberof techniques to optimize CAD for simulation. The guides are available for all four members of the STAR-CAD Series and theircorresponding CAD/PLM tool: STAR-CAT5 (for CATIA V5), STAR-NX

(for Unigraphics NX), STAR-Works (for SolidWorks) and STAR-Pro/E (for Pro/ENGINEER).

The STAR-CAD Series Product Manager, Jean-Claude Ercolanelli,recognizes the importance of these documents.

“CD-adapco is committed to sharing their expertisewith their partners, to ensure that engineers fullyleverage flow and thermal simulation. A few simpleguidelines on how best to use the CAD tool to focus

the simulation on important features can make a significantimpact on model turnaround time and productivity.” He alsorecognizes their synergy with features in the latest release of theSTAR-CAD Series, “In combination with the new automatic modelchecker in STAR-CAD Series V4.10, CD-adapco now offers themost robust CAD-embedded process on the market.”

The CAD or PLM tool is at the backbone of product design process and thestarting point for (almost all) simulation. Utilizing its full power puts parameterized geometry control into the hands of the simulation engineer:easing quick and automatic “what-if” design studies to enable simulation todirectly influence design evolution, and opening up techniques such asnumerical optimization. But, as countless upfront-CFD-wielding ProductDesigners and simulation specialists who work with CAD will testify, the blessingsof CAD for CFD can be mixed.

4Dynamics #27

..::PRODUCT NEWS Best Practices

BEST PRACTICESGuidelines on CAD for Simulation

! Fig:01 Example model from Best Practices

! DOWNLOAD www.cd-adapco.com/starcadseries/bestpractices

”“

STAR-CD V4.02 provides direct access to the latest automaticsurface cleaning and mesh generation technology from withinthe pro-STAR environment. This technology, pioneered in sistercode STAR-CCM+, has proved exceptionally popular with users

and in many cases reduced the time-to-mesh from days to hours.

“Our polyhedral mesh technology is now even more powerful”, saysDirector of STAR-CD Development Dr Riaz Sanatian. “The combination ofhigh quality automatic polyhedral mesh generation and efficient solvertechnology will provide you with improved ease-of-use, faster turnaroundtimes and increased solution accuracy, for all classes of flow, in allapplication areas.”

STAR-CD V4.02 also pushes back the boundaries of CFD by introducing anumber of key multi-physics capabilities, available for the first time in ageneral purpose CFD code. STAR-CD 4.02 is able to simulate problemsinvolving solid stress and melting or solidification.

“These capabilities will rapidly mature and open the door to routineanalysis of a range of applications, such as strongly-coupled fluid/solidinteraction problems and mould-filling, that were either very difficult orimpossible to solve before”, says Dr Sanatian.

STAR-CD V4.02 is available on request from CD-adapco’s User Servicessite or your local CD-adapco office.

Since it’s first release 20 years ago, STAR-CD has established a reputation forversatility and is unrivalled in its ability to tackle problems involving multiphysicsand complex geometries. In the latest release, STAR-CD V4.02 enhances thisreputation further with an improved meshing process and the ability to simulatephysics till now beyond the reach of mainstream CFD technology.

STAR-CD V4.02:Flow, thermal & stress simulation in a single solver

! Fig:01 Fluid-structure interaction in STAR-CD V4.02

! MORE INFORMATION http://www.cd-adapco.com/

Dynamics #273

..::PRODUCT NEWS STAR-CD 4.02

STAR-CD V4.02 hasproved exceptionallypopular with usersand in many casesreduced thetime-to-mesh fromdays to hours.

“”

Alex Read, CD-adapcoStephen Ferguson, CD-adapco

While Fluid and Structural Mechanics are both branches ofthe wider discipline of Continuum Mechanics, in practicalapplication they have traditionally been treated entirelyseparately, addressed by different groups of engineers,

using a different sets of tools for which different numerical simulationmethodologies have evolved. For the solution of stress analysisproblems the Finite Element Method (FEM) dominates, whereas forfluid dynamics the majority of commercial Computational FluidDynamics software is based around the Finite Volume Method (FVM).

Although most practical problems in fluid dynamics are in some wayrelated to the interaction between one or more fluids and a solidbody, direct coupling between CFD and structural analysis softwarehas, until now, remained relatively rare, largely due to the effortinvolved in creating models compatible with two disparate softwarepackages (although as the extensive review of coupling techniques onpage 10 demonstrates, CD-adapco have developed a range ofmethodologies to make this coupling as simple and as effective aspossible).

Most often, the interaction is either neglected all together (if thedegree of interaction is relatively small), or accounted for using amixture of mapping, interpolation and file-transfer, with the resultsfrom one type of analysis used to provide the boundary conditions forthe other.

CD-adapco is about to challenge that paradigm with the introductionof a capability to conduct linear and non-linear structural analysisproblems in STAR-CD V4, using similar FVM solver technology tothat which has led the CFD market for the past 20 years. Althoughmost people automatically associate structural analysis with theFEM, the FVM that underlies most CFD software is equallyapplicable to structural analysis and – as we shall explore in thisarticle - holds some significant advantage over traditionaltechniques.

CD-adapco does not expect that this approach will ever displace theFEM as the principle means of conducting stress analysis, howeverearly experience suggests that the methodology is very attractive forapplications that require: conjugate heat transfer and stressanalysis; fluid structure interaction; and melting and solidification(where the continuum changes state from solid to liquid or viceversa).

Coupled simulation without the couplingUsing STAR-CD V4, both fluid and solid calculations are performedsimultaneously on a single computational mesh, created automat-ically with CD-adapco’s advanced meshing technology. The mesh(which can be constructed from hexahedra, tetrahedra or arbitrarypolyhedra) automatically represents the interface between differentmaterial regions (whether fluid-solid or solid-solid) using a conformal

Stephen Ferguson, CD-adapco

STAR-CD V4.02 includes the capability to perform structural analysis calculationsusing a methodology based upon its industry-leading CFD solver technology, the firsttime that a comprehensive solution for flow, thermal and stress simulation has beenavailable in a single general-purpose commercial finite-volume code.

In this article we explore some of the benefits that integrated flow, thermal and stresssimulation brings, for both the fluid dynamics and structural mechanics communities.

! Fig:01 3-D temperature distribution for cooled gas turbine blade.

Dynamics #2765Dynamics #27

Simulation in STAR-CD V4! Fig:02

Thermal stresses predicted in engine cylinder head.! Fig:03

Temperature contours in engine cylinder head (note the polyhedral mesh).

..::FEATURE ARTICLE Flow, Thermal, Stress

interface, which means that solution domains are connected implicitlywithout mapping or interpolation.

As the coupling is performed in the memory of the computer, and notusing files passed via the hard-drive, the degree of interaction is muchhigher than through external coupling; information is interchanged at aninner iteration level, rather than at each external iteration (whichtypically takes place once per time step), resulting in both increasedefficiency and robustness, and the ability to post-process fluxes, forcesand displacements at the interface, as the solution progresses. TheFVM solver ensures that at each time step all the coupled, non-linearequations are satisfied within the prescribed tolerance; the choice oftime step size is guided by accuracy requirements only, since the fully-implicit formulation of the solution algorithm allows large time steps tobe used.

All of this is provided as part of the standard STAR-CD installation, usingthe regular STAR-CD interface and without requiring the purchase ofadditional licenses or software.

Problem-free automatic meshing with polyhedraA further advantage of the FVM is that it gives structural analystsaccess to the improvements in meshing technology recently pioneeredby CD-adapco.

Automatic mesh generation is typically more problematic for structuralsimulation than fluid dynamics. Unlike fluid dynamics problems (whichare typically convection dominated), structural mechanics problems are“diffusion” (i.e. stress) dominated, which makes their numericalsimulation very sensitive to poor mesh quality. While the FVM can, inprincipal, be applied to grids of arbitrary polyhedra, when using the FEManalysts are limited to the set of computational elements for which ashape-function has been pre-defined, which in practice meanstetrahedra if automatic meshing is required. A single distorted element,among the several thousand used to mesh a typical structural part, canbe enough to ensure that a solution cannot be calculated, or worse thatthe calculated solution is wrong.

CD-adapco leads the CAE industry in automatic mesh generation, andhas invested heavily in creating a methodology to mesh and solve upongrids of arbitrary polyhedral cells. These cells (which typically havebetween 12 and 14 neighbors / faces) have already delivered significantbenefits to the CFD community, and look certain to do the same forstructural analysis, as those relatively low quality tetrahedral elementsthat would typically cause a problem in stress analysis simulation canbe avoided by creating polyhedral cells.

Conformal polyhedral meshes for both the fluid and solid domains canbe created at the click of a button from a CAD surface and, wherenecessary, geometries can be automatically de-featured or repairedusing CD-adapco’s unique surface wrapping capability.

Bigger models with memory-efficient, scalable solver FVM solvers require less memory and are more inherently scalable thanthose that use the FEM. This is a major advantage for engineersperforming stress-analysis calculations who, due to the relatively poorparallel scalability of FEM solvers, have often been forced to generate anumber of highly refined local models in order to resolve detail in criticalareas while keeping the element count low enough to solve; anapproach which introduces a considerable degree of uncertainty intothe calculation, due to difficulty in accurately prescribing boundaryconditions for the local-models.

By virtue of its parallel scalability, STAR-CD can solve non-linear stressanalysis problems using the level of mesh density that has becomecommon practice for CFD simulations, which now routinely use 100million cells or more, thus allowing engineers to simulate entirestructures rather than use a multitude of local models.

The additional work required to set up a parallel calculation for flow,thermal and stress problems in STAR-CD V4 is minimal; the user needonly specify the names of the computers to be used for the simulationand the number of processors required and the software will take careof domain decomposition and job execution automatically.!

Benefits of coupled flow, thermal & stress modeling in STAR-CD:

1 Single-mesh solution for structural and fluid mechanics problems, with no interpolation, no mapping and no external coupling

2 Implicit coupling enhances efficiency and robustness of FSI calculations

3 Significantly reduced memory requirements over FEM, allowing engineers to tackle bigger structural problems using fewer computational resources

4 CG and AMG solvers at the heart of FVM solver are inherently scalable, allowing engineers to harness the power of parallel computing for structural analysis

5 High quality, fully automatic meshing using polyhedra, significantly reduces time for preparation of structural analysis

6 No additional licensing or software required to perform stress analysis

FACTS ❐

" MORE INFORMATION [email protected]

Image courtesy of Pavel Krukovsky

Thermo-mechanical analysesOn completion of the CFD analysis of the water circuit andthe surrounding metal cast, the temperature distribution inthe iron cast was evaluated. Since the choice of boundarycondition is responsible for the accuracy of the metal casttemperature calculation, the heat flux distribution was derivedboth from experimental measurements and numericalpredictions. Experimental measurements were used to set thecoolant temperature at both the gasket and the circuit outlet.For the heat flux, data from a 1-D GT-Power simulation of thewhole engine at a given operating condition was imposed,while the coolant/metal heat transfer was directly calculatedin STAR-CD. Although the 1D model is unable to accuratelyaccount for three-dimensional effects and non-uniformcylinder-to-cylinder distributions, the decision to derive theboundary data from the 1D model was considered to be agood trade-off between accuracy and computational effort.

ResultsCFDTwo main issues appear from the coolant/metal analysis(Figure 3):• the highest temperatures in the solid domain are located at

the junction between the pre-chamber and the combustion dome, towards the side opposite to the injector location;

• the high-temperature layer within the iron cast is quite thin,and obviously located towards the walls facing the combustion chamber, confirming the validity of the application of a uniform temperature at the solid domain outer walls.

CFD - FEMFigure 4 shows a direct comparison between computationaland experimental results in the pre-chamber 3 crack region; itis possible to observe a very convincing match in terms ofmaximum equivalent von Mises stress location and crackinitiation, thus confirming the validity of the simulationmethodology.

OptimizationAs a final validation, a comparison of the stress distributionbetween the BASE and EVO circuit designs was made. Figure5 clearly indicates in the EVO design peak tension valuesthat cause the cracking has been significantly reduced: itsvalue reduces from 240 MPa to less than 200 MPa (a 20%reduction). This clearly indicates the benefits resulting fromthe improved heat transfer coefficient in the critical area asa result of the optimized flow design.

ConclusionAn optimization study involving both fluid-dynamic andthermostructural aspects of a turbocharged diesel enginehead was carried out. A cost-effective methodology wasevaluated to correctly represent the fatigue-failure criticalregions without excessive computational costs. Since theaim of the work was to trade-off solution accuracy andcomputational, the following conclusions were drawn:

• a proper choice of both fluid-dynamic and mechanical boundary conditions is required in order to deliver the required accuracy;

• comparisons with experimental data confirmed that the methodology adopted was able to accurately predict locations prone to cracking;

• the modeling procedure allowed a sensitivity study to be carried out of the engine head to variations of the gasket plate design;

• the modifications of the gasket passages, although very simple, allowed the cooling performance of the circuit to dramatically improve (Figure 6), almost eliminating criticalstress concentrations at the cylinder 3 pre-chamber, which was experimentally detected to crack when operating a full engine load.!

An optimization study involving both fluid-dynamic andthermo-structural aspects was carried out. Using a cross-disciplinary approach, the structural and thermodynamicproblems were decoupled using an ad hoc methodology to

trade-off computational effort with accuracy. This procedure allows asensitivity study to be carried out, varying geometric parameters ofthe engine to obtain an optimized component.

MethodologyThe adopted methodology (shown in Figure 1) decoupled thestructural and thermodynamic simulations. In order to evaluate thetemperature distribution of the metal cast, a CFD analysis of both thewater circuit and the surrounding metal was performed. Boundaryconditions from a 1-D simulation of the whole engine were imposed,while coolant/metal heat transfer was calculated using STAR-CD.

The temperature field was then passed to the FEM code, and structuralanalyses were carried out in order to assess the fatigue strength of thecomponent. Finally, this methodology was applied to a comparison ofthe current circuit configuration and an improved design (where thewater jacket flow has been optimized) in order to estimate theeffectiveness of the design optimization on the fatigue strength of thecomponent.

Fluid-dynamic preliminary analysisCFD analyses were carried out to focus the flow distribution in thecritical regions, i.e. the valve bridges and the pre-chamber areas. Initiallyan isothermal analysis was performed on the whole engine water jacket.

In order to evaluate the effect of simple geometric modifications to thecircuit layout on the cooling effectiveness, the original configuration(BASE) was compared to a modified one (EVO). The flow in the EVOconfiguration is forced to cross the whole engine block before enteringthe head jacket and only reaches the jacket exit after crossing the wholeengine head (cross-flow).

A critical velocity Vcrit was defined, below which the local heat transferis considered to be ineffective. The percentage of the coolant volume inwhich the velocity fell below Vcrit was compared for the two solutions(Figure 2).



! Fig:02Isosurface of flow volume where velocities are below Vcrit for BASE and EVO designs

A detailed understanding of the flow and thermal behavior in the water jacket of aturbocharged diesel engine can result in significant design improvements. If thevelocities of the coolant flow drop too low then heat is not convected away from themetal engine block. Over time, the resulting high temperatures produce cracks inthe structure, ultimately causing its catastrophic failure. The analysis reported hereenabled the coolant flow path to be optimized, reducing the peak temperatures inkey locations and resulting in a 20 % reduction in the peak thermal stresses.

" Fig05Tensional peak improvement

# Fig:04Experimental numerical comparisonof crack locations with maximumvalues of Von Mises stresses.

! Fig:01 - Simulation methodology

Thermal CFD Analysis: Fluid-solid heat transfer

Preliminary cold flow CFD Analysis:gasket optimization

• Constraining• Structural loading (press fit of

valve seats or valve guides,gas pressure, bolt tightening)

• Contact problems• Material proprieties

CAD Design

Thermal Analysis

Structural Analysis

• Temperature• Thermal gradients

• Stresses• Strains

$ Fig03Temperature distribution

Water jacket optimizationusing CFD & FEMAuthors: Stefano Fontanesi, Vincenzo Gagliardi, Matteo Giacopini, Simone Malaguti and Reggio Emilia – University of Modena, Italy

$ Fig06Flow field improvement

! Fig:05 Computed pressure distribution

! Fig:06 Computed displacements

" Fig:03 (far right)On the left we see no deformation at the beginning. On the right we see the deformated membrane by the pressure load of the flow.

Membrane "

Train, bridge, buildings "

In order to fully simulate the membrane behaviourin the design process, it is usually necessary tocouple a CFD code with a structural analysis code.In this article we describe how this can be achieved

by coupling CD-adapco’s CFD software with the membranecode SCOOP (which is based on the Finite Element Method).

Fluid-structure coupling methodThe coupling is achieved on a node-to-node basis, using anautomated process that constructs the structural meshdirectly from the boundary of the CFD mesh, obviating theneed to map or interpolate data between the solvers.

The membrane is represented as a baffle in the CFD codeand as a shell element in the FEM solver. At each time-stepthe FEM solver calculates the membrane displacements fromthe pressure loading calculated by the CFD model.Displacements are calculated for each node of the membrane

and, within the CFD solver, used to distort the membrane andsurrounding mesh to the new position. In order to maintainthe quality of the hexahedral cells adjacent to the membranea number of smoothing cycles are performed. The CFD solverthen calculates new pressure loadings, which are used by theFEM solver to calculate new displacements in the next time-step. The modularity of the SCOOP framework, its flexible datastructures and intelligent interface methods allowed a verytight coupling between the membrane solver and CD-adapco’sCFD software.

The membrane parameters for the structural model weretaken from the results of material tests performed by LaborBlum (Figure 2) so that the model could take account notonly of the actual membrane stresses, but also of warp andweft orientation, surface curvatures, load ratios and loadhistory.

FSI for membrane structuresAaron Kneer, Tinnits Technologies GmbH

Dynamics #2710


9Dynamics #27


Buildings with membrane structures are often used to cover large areas, andare often subjected to large loads, both thermal and structural. Wind loadingis by nature unsteady, fluctuating in direction and amplitude in response toboth discrete gusts and longer-term diurnal variation. Even in an apparentlysteady wind, the membrane loading will be influenced by the unsteady wakeof nearby buildings.

Figure 3 shows a simple test case that represents the flow through a channelcontaining a membrane with a hole in the middle. A fixed air velocity of 1 m/s isprescribed at the inlet boundary, and the simulation shows how flow isaccelerated through the orifice and how the calculated pressure loading distortsthe membrane

FSI model of Fröttmaning stationIn order to validate this approach using a realistic scenario, a coupled analysiswas performed for the 5,054m2 roof membrane of Fröttmaning station inGermany (Figure 1), for which extensive model-scale wind tunnel data has beengathered. A computational model was constructed that included not only thestation and its roof, but also the surrounding buildings and topography.

The station roof is constructed from fifteen identical membrane segments that arestabilized using a steel construction with a wall protecting the rear of the station.To represent this a computational mesh of 550,000 hexahedra was constructed,with some 10,000 baffle cells representing the membrane. In addition to thestation itself, the mesh also includes solid obstructions that represent surroundingbuildings, trains and bridges (Figure 4).

For the preliminary simulation an oblique wind direction was prescribed andpressure loadings and displacements are shown in Figures 5 and 6 and predict amaximum membrane displacement of around 10mm.

ConclusionThe methods and results presented within this article show that large membranestructures can be modelled in using fully coupled fluid-structure interactionmodelling. Using this FSI-technology realistic displacements of membraneconstructions can be computed under different wind loadings, providing valuabledesign data to the membrane community.!

! Velocities

! Fig:01 (above)Large membrane construction of the Fröttmaning train station

" Fig:02 (right)Simulation results - Labor Blum


! Fig:04 10,000 baffle cells representing the membrane of Fröttmaning station

Phot

ogra

phs

and

roof

ligh

ting

tech

nolo

gy b

y O

sram

Gm

bH

A faithful representation is achieved by using STAR-CD’s free surfacecapability together with dynamic motion / sliding mesh capability togetherwith an efficient transient solution methodology, then communicating thefluid forces to a rubber deformation model of the tyre, which in turn returnsthe deformation to the fluid mesh.

Finite Volume combined fluid and structural approachComputational Continuum Mechanics (CCM) is the termed coined for acombined approach to multi-physics such as fluid and structural mechanics.As described in [3] the conservation equations apply to both and may besolved using a finite volume methodology whereas the constitutive relationsand material properties differ between the two systems. CCM for FSI,enabled in STAR-CD V4, offers some advantages compared to traditionalFEA methodologies. It is applicable to a wider range of applicationsincluding casting, solidification, melting and floating bodies, but becomespotentially unstable for structural analysis of thin structures (plates/shells).Our CCM implementation offers the added benefits of fully automatedpolyhedral meshing and full HPC-parallel scalability.

The manoeuvring effectiveness and structural integrity of rudders in cross-flow, typical in marine engineering, in the wake of the ships hull and invicinity of the propeller(s) may be studies through this approach.

The figures illustrate fluid velocities near/behind the rudder and also the pressurisation on the inclined faces, which result in structural stresses anddisplacement.

Slamming of ship structures in rough sea-weather conditions imposessevere structural loading on the hull. The two drop-tests described below, fora hollow cylinder and a solid inclined wedge, predict the motion due togravity and interaction with the free-surface flow, leading to walldeformations and 6DOF motion.

Full 6DOF motion results from the inclined wedge drop-test – here thecomputed linear and angular acceleration, together with the heel angle andimpact velocity, are compared with experiment [4].

! Fig:03 Final deformation (x10) relative to circular profile. The wheel is rotating counter-clockwise.

Dynamics #2712


11Dynamics #27

! Fig:09 140o wedge free-fall test � 5o inclination (from Azcueta [4])

! Fig:07 - Schematic (left) and vertical free-body motion compared with measurement (right)

! Fig:06Stresses in lateral (left) and longitudinal (right) rudder cross sections

! Fig:05Displacement in the longitudinal (left) and lateral (right) directionsthrough the rudder

" Fig:08 Fluid pressurization and cylinder hollow-wall displacement prediction at and shortly after free-surface impact.

! Fig:04Fluid pressure contours and flow cross-sections through (left) and behind (right) the rudder

Time [s]

Tube #2, #3 & #4 (Quadratic model)

Tube #1 & #5 (Quadratic model)

.05

.025

.0

-.05

-.025.0 .5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time [s]

.05

.025

.0

-.05

-.025.0 .5 1.0 1.5 2.0 2.5 3.0 3.5

13Dynamics #27


Substructure matrix approachThis is a unique approach in which any well-established structures code may be used togenerate the mass, stiffness and dampingmatrices, then exported to files. CD-adapco’sexpert system, es-fsi, reads in the matrix fileand solves the dynamics equations for thestructural displacements within the CFDsolutions system. Substructuring is used tocondense out many degrees of freedom withrespect to the fluid mesh. The fluid forcescomputed by STAR-CD are interpolated ontothe substructure nodes and the matrix invertedin a special-purpose API to compute thedisplacements at each time-step. The couplingmethod is explicit, but benefits by eliminatingall communication overheads.

Vortex-induced vibration (VIV) is common inheat-exchanger applications in the powergeneration, oil and gas, process and nuclearindustries. In this example, the flow past atube array creates a vortex shedding pattern,which perturbs the tubes. The perturbationsresult eventually in self-sustaining oscillations.The limiting amplitude of oscillation increaseswith the oncoming flow velocity, until at somecritical flow velocity, in this case 1.2m/s, theamplitude exceeds the separation betweentubes, causing them to collide.!

! Fig:11 - : Linear tube array in cross-flow

! Fig:12 - Flow patterns showing wake flow through one complete oscillation cycle

! Fig:10 140o wedge free-fall 5o

inclination test � linear (top, left) and angular (right) acceleration; heel angle and impact velocity (bottom) (from Azcueta [4])

Time = 3,5 s

Time = 3,65 s

Time = 3,8 s

Time = 3,55 s

Time = 3,7 s

Time = 3,85 s

Time = 3,6 s

Time = 3,75 s

Time = 3,9 s

! Fig:13 - Stable oscillations (top) and unstable (below) oscillations beyond critical velocity.

Ugap = 1.18m/s

15

10

5

0

0.35 0.37 0.39 0.41 0.43 0.45Time [s]

Acce

lera

tion

(g)

Experiment (Port)Experiment (Stbd)Experiment (Mean)Simulation

Experiment Simulation

Experiment Simulation

5

2.5

0

-2.5

-50.35 0.37 0.39 0.41 0.43 0.45

Angu

lar

Acce

lera

tion

(x10

3(d

eg/s

ec2)

Time [s]

15

10

5

00.35 0.37 0.39 0.41 0.43 0.45

Time [s]

Time [s]

Tube #2, #3 & #4 (Quadratic model)

Tube #1 & #5 (Quadratic model)

.1

.05

.0

-.05

-.15.0 .5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

.15

-.1

Time [s]

.1

.05

.0

-.05

-.15.0 .5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

.15

-.1

Ugap = 1.20m/s

[1] Mendonça F, “Approaches to Industrial FSI - An overview with case studies”, 7th MpCCI User Forum Proceedings, 21-22 Feb, 2006, Sankt Augustin, Germany

[2] Demerdzic I., Muzaferija S. and Peric M., “Benchmark solutions of some structural analysis problems using finite-volume method and multgrid acceleration”, Int. J. Num. Meth. Eng. 40, p.1893-1908, 1997

[3] Demerdzic I., Muzaferija S.,“Numerical method for coupled fluid flow, heat transfer & stress analysis using unstructured moving meshes with cells of arbitrary topology”, Comp. Meth. Appl. Mech Engrg., Vol 125, pp. 235-255, 1995

[4] Azcueta R., “Computation of turbulent free-surface flows around ships & floating bodies”, PhD Thesis (in German), TU Hamburg-Harburg, 2001

❐ REFERENCES

CD-adapco WorkshopsFlow, Thermal & Stress simulation workshops

Dynamics #2714

CD-adapco invites you to attend a one-day seminar in your local areato explore the business advantage that can be gained through theapplication of flow, thermal and stress simulation technology within theOil and Gas industry. The seminar will focus on the application ofsimulation techniques to safety and hazard management, anddemonstrate how flow and thermal simulation can reduce costs andincrease efficiency of existing processes.

Computational Fluid Dynamics (or CFD) is a technique that realistically simulates thebehavior of fluids using computer technology. CD-adapco’s flow, thermal and stresssimulation technology can provide insight into any problem that involves fluid flow (liquidor gas or combinations of both, and structural stress) and has been applied at everystage of the oil and gas production process - from exploration to extraction, fromtransport to processing.

Join us to learn about CFD, see a demonstration of our software and find out how it canbe applied for maximum success in your organization. Also hear from users of CD-adapcoCFD software in your industry, and learn how they have benefited from applying CFD.

Seminars are free of charge, but spaces are limited.For more information email [email protected]

Oil & Gas Industry

Aerospace & Defense IndustryThroughout 2007, CD-adapco are holding a series of exclusiveseminars that will explore the business advantage that can be gainedthrough the application of flow, thermal and stress simulationtechnology within the Aerospace and Defense industries.

Each of the seminars will focus on the application of a variety of simulation techniquesto a wide-range of aerospace applications and demonstrate how flow, thermal and stresssimulation can reduce costs and increase efficiency of the design process, and can helpto engineer products that will meet future technological and economic challenges.

The seminars will cover topics such as external aerodynamic simulation; aeroacoustics;combustion; heat transfer; multi-phase flows and fluid-structure-interaction.The seminars are ideally suited for:• Engineering Managers looking to streamline their flow and thermal simulation process• Lead engineers looking for solutions to their CFD bottlenecks• Program Managers and Project Engineers interested in learning how CFD can be

applied to their program• Engineers and Researchers interested in how others are attacking flow and thermal

simulation problems• Those interested in solving complex flow and thermal problems such as combustion,

heat transfer, aeroacoustics, fluid-structure interaction, etc....• Those interested in COTS flow and thermal simulation capability that compliments

existing government and research codes

Future seminars in this series are being planned for Seattle, WA, Wichita, KS, and otherkey industry sites. The seminars are free of charge, but spaces will be strictly limited.

For details on further seminars please email [email protected]

The scale of these losses combined with pressurefrom insurers, has led to a rapid re-evaluation tothe techniques used to design offshore platforms.Many operators are turning towards Computational

Fluid Dynamics in order to provide additional insight into howtheir platforms perform under the most extreme operatingconditions.

Computational Fluid Dynamics (or CFD) is a technique thatsimulates fluid flow phenomena using super-computertechnology. Although its origin is in the aerospace andautomotive industries, CFD is increasingly finding applicationin many areas of the oil and gas industry. CFD can be used to

simultaneously simulate the aerodynamic effect of strongwinds on the platform with the hydrodynamic influence ofwaves impacting upon it.

Although CFD technology has been routinely applied in manyindustries since the early eighties, it has only recently begunto be seriously used in offshore platform design. Most currentoffshore platforms were designed using extensive experi-mental model testing. Although experimental analysis providesconsiderable insight into the performance of a particulardesign, physical prototypes are expensive and time consumingto construct.

Dr Dennis Nagy, CD-adapco’s Director for the Oil and Gassector explains: “It isn’t that CFD technology wasn’t availablewhen the current generation of platforms was designed; CFDtechnology has been routinely applied in many industriessince the early eighties. It is just that the cost of performingthe analysis would, until very recently, have been tooprohibitive.”

Nagy feels that the biggest advantage of CFD is that its rapidturn-around time helps to break the dependence of offshoredesign on pre-existing design codes. Although design wind

and wave conditions are a useful starting condition foroffshore platform analysis, CFD simulation allowsdesigners to more easily pursue multiple “what if?”scenarios. Once a CFD model for a platform is set up,it is relatively simple to repeat the calculation formultiple loading scenarios. “Instead of becoming stuckby the fact that the design codes don’t deal with waveheights above 70 feet, using CFD designers are free toconsider the impact of wave heights of 80, 90, or even100 feet,” says Nagy. “All they need to do is input thenew condition and sit back while the computer doesthe number crunching. It is a very effective way ofassessing the limit of your design.”

Unlike testing of physical prototypes, CFD simulationsare typically carried out at full scale (the computermodel has the same dimensions as the actualproduction platform rather than those of a smallerexperimental model). This has the considerableadvantage that results can be interpreted directly anddo not have to undergo scaling, a process that canintroduce a significant uncertainty, especially fortransient phenomena such as the impact of a wave.

A further advantage is that, instead of being restrictedto retrieving data from a few experimental monitoring

probes, data is available at every point on the platform,at every discrete time interval for which the simulationis performed. The wave impact on a platform can beviewed from any angle, and the instantaneous forcesacting on any part of the structure can be calculated.

Data from CFD calculations can also be used to assistother types of analysis, for example, the forces actingon a platform can be exported to a stress-analysissoftware package. In extreme cases, where fluid forcescause large deflections of components, the CFDsimulation can be coupled directly with the stressanalysis tool and both stress and fluid simulations canbe performed simultaneously, each simulation feedingnew boundary conditions to the other.

In Nagy’s view, the adoption of CFD technology as aroutine part of offshore design is inevitable. “In theautomotive industry almost every component isdesigned with the aid of CFD technology, to bring anew product to market without it would beunthinkable,” he says. “The financial and environ-mental impact of the recent hurricanes means that theoil and gas industry has no choice but to follow suit.”!

Dynamics #2716

..::FEATURE ARTICLE Oil & Gas

15Dynamics #27


! Fig:01CD-adapco software has been used to evaluate the wind and wave loading onplatforms in storm conditions. These analyses allowed platform designers and operators to evaluate many different platform loading scenarios without the excessive cost of creating physical prototypes.

Current design standards require that platforms be built to survive so-called100-year storms, which generate wave heights of up to about 70 feet.However, during Hurricane Ivan peak wave heights of over 90 ft weremeasured (including one that severely damaged the Chevron Petroniusplatform) consistent with a once in 2500-year storm. The problem iscompounded by the fact that many of the 4000 platforms operating in theGulf of Mexico were designed before 1988, when the current 100-yeardesign standards came into operation (although some of the destroyedplatforms were of recent design).

Hurricane resistantoffshore platform designStephen Ferguson, CD-adapco

! Fig:02CD-adapco�s CFD solutions are routinely used in the design of marine applications to understand how a unit will react upon impact bya wave. This technology allows engineers to optimize the hydrodynamic performance of the ship, FPSO or platform and understand the range of conditions under which safe operation can be assured.


" Fig:03 Wave impact study of offshore platform

BENEFITS

� Rapid turnaround time.

� Easy investigation of �what-if� scenarios.

� Simulations are carried out at full scale.

� Data available at every part on the platformat every interval

� Can be exported to a stress analysis package

FACTS ❐

During the overhaul of a 6.4MW industrial gasturbine, Wood Group Light Industrial Turbines Ltd(WGLIT) adopt a standard practice of restoringcompressor labyrinth seal fins and abradable liners

as new design clearances. Repair of the rotating knife-edgeseals requires the compressor rotor to be disassembled whichitself incurs considerable further rework and repair cycle timethat would otherwise not necessarily be required. It thereforefollows that if the relationship between seal clearance, bearingchamber sealing efficiency and compressor delivery pressurelosses could be quantified, then the setting of acceptableoverhaul limits for the labyrinth seal clearances would result inreduced overhaul cost and turn times whilst delivering therequired quality and performance for the intended serviceduration.

ObjectivesThe objectives of this study were twofold:1) To quantify the effects of compressor inlet bearing

abradable labyrinth seal clearances on;• Bearing chamber to compressor annulus sealing• Compressor delivery pressure2) To examine acceptable overhaul limits for the above

mentioned labyrinth seals.

MethodologyComputational Fluid Dynamics:Analysis with Air OnlyOne eighth (45 degrees) of the complete labyrinth seal wasmodeled with axi-symmetric boundaries, Figure 1. Four modelswere built comprising 755,000 hexahedral computational cells.The labyrinth seal was modeled as a solid wall and the rotating

fins did not cut into the seal. The oil drain in the seal wasneglected. The rotor speed was taken to be 11,000 rev/min.The clearance between the fin the seal varied from 0.05 mm to0.250 mm in steps of 0.050 mm.

The flow was assumed to be steady state, compressible andturbulent. Turbulence was modeled using the twoequation k - ε turbulence model with hybrid wall functions. Thisapproach was chosen as the value of non-dimensionalwall distance, Y+, can range from about 1 to 250 and the walltreatment function method would still be valid. Second orderdifferencing was used for momentum to ensure accuracy andfirst order differencing on turbulence and enthalpy.

The pressure drop across the seal was unknown therefore twopressure drops were chosen to be representative: 2.28 bar – nothrottling had taken place after the sealing air had been bled offfrom the 8th compressor stage and 1 bar – some degree ofthrottling had taken place.

At a later stage one of the seal fins was removed, thedownstream fin closest to the inlet pipe, and an analysis wascarried out at clearances of 0.05 mm, 0.10 mm and 0.15 mm.

A run was also completed, assuming the fin had cut into theabradable seal with a clearance of 0.05 mm between the finand the non-abradable surface.

Extended Model (droplets - two phase flow):The initial model was extended (1,038,000 computational cells)to include the chamber with the inlet journal bearing, Figure 2.

Dynamics #2718


17 Dynamics #27


Authors: Allan Thomson & David Anderton, APM Team, Wood Group Light Industrial Turbine Ltd.

Repair sensitivity study for compressor inlet labyrinth seal usingComputational Fluid Dynamics

! Fig:01 (right)Inlet Labyrinth Seal

This study examines the potential to relax the repair limits for an industrial gasturbine compressor inlet labyrinth seal through numerical analysis, in order tominimize unnecessary, costly and timely rework. The study demonstrates that forsteady-state running conditions there is considerable potential to relax repairlimits without sacrificing seal performance. It recommends that these resultsshould be discussed for implementation and highlights areas for further work.

" Fig:04 - Model 1 Velocity Vectors

" Fig:05 - Model 1 Oil Particles

" Fig:06 - Model 2 Velocity Vectors

" Fig:07 - Model 2 Particles

" Fig:03 - Flow Rate Across the Inlet Labyrinth Seal

! Fig:02 (Far right)Inlet Labyrinth Seal Including Bearing Chamber

Morgan (1) proposed that an efficient way to simulate oil flow through an inletlabyrinth seal was to use a transient Lagrangian droplet model coupled with a liquidfilm analysis. Similar types of analyses have been carried out on crankshaft bearingsin internal combustion engines. Due to time constraints and computing resource itwas decided not to include the liquid film analysis, as the end result would be still berelatively correct. The rotational speed was kept at 11,000 rev/min and the time step0.5 degrees.

Inlet Bearing Leakage:A generic spreadsheet was adapted to determine the static load on inlet bearing andsince the geometry and oil type (Shell Turbo Oils Type T46) were known, an analysiswas carried out by Harrison (2) to determine the oil leakage from the bearing. Hefound it was 0.4 m3/s for the bearing. In order that a fine mist of oil was produced80 injection points were defined, and the particle diameter was chosen to be 0.05mm. The injection rate was then found to be 4.33 x 106 droplets/s at a temperatureof 50°C. Using the above calculations at a maximum bearing clearance of 0.124mm, the velocity of the oil droplets was found to be 13.5 m/s.

Results and Discussion:A plot of air leakage against clearance is shown in Figure 3. It can be seen that withall the seal fins present, a worst case pressure drop of 2.18 bar across the seal anda clearance of 0.25 mm, the leakage across the seal was just over 0.07 kg/s (lessthan 0.5% of the total air mass flow rate through the engine), compared to 0.014kg/s when the clearance was 0.05 mm. At 1 bar pressure drop across the seal, at aclearance of 0.25 mm the airflow was just over 0.04 kg/s compared to 0.008 kg/swhen the clearance was 0.05 mm. The peak Mach number across the seal fins wasapproximately 1 when the pressure drop was 2.18 bar, i.e. the flow was almostchoked at all seal clearances. At 1 bar the peak Mach number was approximately0.7. When one fin was removed (first fin downstream of the inlet pipe in the directionof the compressor exit) then the increase in airflow across the seal remained roughlyconstant at about 7% at both pressure drops, at each fin clearance, Figure 3.

Comparison of the straight through loss at a fin clearance of 0.05 mm and the lossincurred when the fins had cut into the seal shows that a reduction in loss of about5% at both pressure drops when the fins had cut into the seal, Figure 3. Due to timeconstraints and computer limitations only two models were run with oil injection fromthe inlet bearing:1). At a clearance of 0.25 mm and a pressure drop of 2.18 Bar.2). At a clearance of 0.25 mm and the inlet flow pipe blocked.

Two phase Lagrangian modeling can only be simulated when the second fluid phaseoccupies less than 40% by volume of the computational cell. If this was exceededthe model would become unstable, and fail. Due to the positioning of the oil injectionpoints, the flow boundaries and its nature, only about 4 revolutions of the enginewerepossible before the instability mentioned above occurred. However this was enoughtime to show that in model 1 (Figure 5) no oil passed across the seal even thoughoil had passed the lip on the rotor. Model 2 (Figure 7) showed oil passing throughthe seal, which would very quickly enter the compressor. The models showprogression of oil particles through the seal. Figures 4 and 6 show the centerlinevelocities, to the same scale 0 m/s to 60 m/s. In model 1 the maximum velocity was60 m/s whilst in model 2 the maximum velocity was found to be 310 m/s.

ConclusionIt has been shown that for the clearances considered here (Figure 3) under steadystate conditions there will be negligible reduction in compressor delivery pressure andflow rate. For the worst case (0.25mm clearance at a pressure drop of 2.18 Bar) 70g/s seal air leaked across the seal (less than 0.5% of the total air mass flow rate).For the largest clearance considered there was sufficient mass flow across the sealto prevent oil ingress to the seal and hence compressor. Oil is most likely to enterthe compressor via the lab seal when the seal airflow rate is at a minimum, i.e. if theair pipe is blocked.

RecommendationsBased on the analysis conducted in this study the compressor inlet labyrinth sealrepair tolerance should be reviewed with the intent of implementing relaxedclearances.

Further studiesTransient effects, which were considered to be secondary, such as thermal growthand start up / shut down effects, were not considered in this study.!

[1] Morgan, J. Computational Dynamics, Shepherds Bush Road, London. Personal communication.

[2] Harrison, J. Daido Industrial Bearings Europe Ltd, Winterhay Lane, ILMINSTER, England. Personal communication.

❐ REFERENCES

Dynamics #2720


19Dynamics #27

Safer flare designwith CFDStephen Ferguson, CD-adapco


Flaring systems were originally developed to dispose of thewaste gas produced as a side effect of the oil productionprocess, although continuous flaring has been effectivelyoutlawed by strict legislation and economic and environmental

concerns. Today, flaring systems are principally deployed as safetysystems, protecting the production system from over-pressurisationduring the extraction process. During surges in production, gas, andoccasionally liquids, are routed by a pressure-relief valve (or by anemergency safety valve) up through the flare-header towards theflare tip.

An obvious consequence of combusting large amounts of natural gas isthe considerable amount of radiation emitted by an operational flare.Modern flare systems are specifically designed to reduce the radiation,pollution and acoustic impact of a flare, by using the energy associatedwith the high-pressure gas to entrain large amounts of air (typicallyusing Coanda effects, or sonic or super-sonic nozzles). These aeratedflames are small and have a relatively low radiation signature – howeverthe amount of radiation that they generate is still enough to causesignificant damage to personnel and equipment on the installation.

Although the American Petroleum Institute API 521 effectively limits theamount of incident radiation on production surfaces to 1390 Wm-2

during normal operation -- a level at which “continuous exposure isallowed without causing permanent injury” –- the consequences ofexceeding those levels can be serious. At 1580 Wm-2, exposure ofmore than a minute will cause symptoms similar to mild sunburn. At1890 Wm-2, bare skin will begin to feel pain after 50 seconds ofexposure; by 2840 Wm-2 this time is reduced to 30 seconds; at 4730Wm-2 bare skin will begin to feel pain after 18 seconds, and personnelwithout protective clothing have just 23 seconds to escape to a safe area.

In practice, safety conscious operators try to limit radiation to wellbelow the minimum API standard. This is partly due to economicnecessity as, during unexpected surges in gas production, should theradiation levels rise to unacceptable levels, operators have no choicebut to reduce production or to stop it all together, effectively limiting theoverall profitability of the installation.

Traditionally, the radiation signature of a flare on a specific installationhas been calculated using a combination of empirical calculations andad-hoc post-installation modification Increasingly manufacturers andoperators are turning towards Computational Fluid Dynamics as a wayof predicting how flares will perform under realistic operating conditions,before even the first prototype is built.

Computational Fluid Dynamics (or CFD) is a powerful technique thatsimulates fluid flow phenomena using computer technology. Although itsorigin is in the aerospace and automotive industries, CFD is increasinglyfinding application in many areas of the oil and gas industry.

Unlike testing of physical prototypes, CFD simulations are typicallycarried out at full scale (the computer model has the same dimensionsas the actual production platform rather than those of a smaller experi-mental model). This has the considerable advantage that results can beinterpreted directly and do not have to undergo scaling, a process thatcan introduce a significant uncertainty, especially for problems involvingcombustion and radiation.

One of the biggest advantages of CFD is that its rapid turn-around timehelps to break the dependence of design on pre-existing design codes.Although design conditions are a useful starting condition for offshore

design analysis, CFD simulation allows designers to more easily pursuemultiple “what if?” scenarios.

CD-adapco has recently performed a number of radiation studiesdeployed on both fixed and floating units, including a projectundertaken on behalf of DPS, a leading engineering design companywith a vast amount of experience in the design, supply and support ofprocess equipment for the Oil and Gas industry, which involved theanalysis of a flare deployed on a Floating Production Unit (FPU).

Unlike fixed installations, FPUs are limited by stability considerations inthe length of boom that they can deploy in order to reduce the incidentradiation on the deck. During the DPS study the impact of variousmitigation scenarios was considered, including variations in boom angle,installation of physical shielding and the deployment of a protectivewater curtain. The simulation results allowed the design team toaccurately assess which areas of the deck would be exposed to highlevels of radiation and to adjust their protection strategy accordingly.

“Using CFD analysis we were able to predict the performance of theflare installation with confidence, which allowed us to carefully select alevel of protection that would ensure the safety of personnel andequipment aboard the FPU” said Jasbir Landa, Project Manager for theFPU project at DPS.

The CFD model was also used to examine the influence of wind speedand direction on the flame combustion and the shape of flame that itgenerates, something that is generally not possible with less sophisticated flare modelling packages. “In this case flame shape hada significant impact on the radiation footprint of the flare, whichwas something we had to address carefully when choosing amitigation strategy.”!

Using CFD analysis wewere able to predict theperformance of the flareinstallation withconfidence, whichallowed us to carefullyselect a level ofprotection that wouldensure the safety ofpersonnel and equipmentaboard the FPU.

“

”

The flaring of natural gas plays a critical role in the global oil and gas industry.According to the World Bank over 150 billion cubic meters (or 5.3 trillion cubicfeet) of natural gas are flared and vented annually, mostly as part of the oil andgas production process.


Dynamics #2722

..::FEATURE ARTICLE Aerospace

21Dynamics #27


The fast turn-around times required in the early design stagesof large cabin models are often not feasible due to the highamount of cells required to resolve small features like airinlets. In the past, the hardware resources and softwarecapabilities were not always available to run extensive cabinmodels, which resulted in the problem of “boundary conditionclosure”. Simulating just a slice of the cabin leaves two majorboundaries (front & back) in the open. This problem can beaddressed by using no-slip walls, symmetry planes,prescribed in- and outflow or cyclic boundary conditions.However, each solution comes with certain compromises oneither the accuracy and/or comparability of results. This isespecially true if the area of interest is close to a cabinfeature such as the galley, lavatory or, in the special case ofthe A380, stairhouses. The availability of low-cost CPU powerand the release of STAR-CCM+ allowed ICON to approachthe problem by creating a complete cabin model for a doubledeck aircraft seating 484 passengers in three classes.

The intention of this work was to:• investigate the practicality of handling models for very

large cabin ventilation simulations • develop an efficient work process including testing the

suitability of CD-adapco’s new STAR-CCM+ solver• investigate the cabin flow field in a complete “closed”

domain by erasing the problems associated with additionalunknown boundary conditions

• investigate the flow in the stairhouses and its effect on theadjacent cabin areas

Due to asymmetric geometry features in some galley areasand most notably the rear stairhouse, it was not possible tomodel just half of the aircraft. Only hex-dominant mesheswere considered suitable to meet the required refinementlevels near inlets and outlets and keep the overall mesh sizewithin reasonable limits. The final mesh consisted of 75million cells with over 90% being hexahedral cells (with aspect ratio equal to 1).

Using CD-adapco’s next generation software STAR-CCM+and exploiting its new approach to parallel processing, ICONwas able to use modest hardware to perform a detailed CFDstudy of the flow field in the whole cabin. This includedassessment of comfort related parameters such as PMV(Predicted Mean Vote), PPD (Percentage People Dissatisfied),Age of Air etc. These results helped ICON to understand thespecial flow phenomena existing in complex cabin geometriessuch as the A380. It can also be seen as a “worst” casescenario in terms of domain size, since the only similar sizedaircraft in the foreseeable future will be -900 stretch versionof the A380. Finally, as CPU power becomes cheaper andmore readily available, models of complete aircraft cabins willbecome even more detailed and/or quicker to run.!

In the case of the A380, Airbus’ next generation fulldouble-deck passenger aircraft, simulationengineers faced many new challenges. In itsstandard version the A380 holds 555 passengers in

three classes, on two decks. In theory, a layout holding morethan 800 economy class passengers is also possible. It is welldocumented that Airbus wants to offer airlines and passengersan improved level of cabin comfort and as a result significantdemands are placed on the cabin ventilation system.

The ventilation system has to fulfil not only the requirementsimposed by the authorities, like minimum air exchange rates ormaximum pollutant concentrations, but also Airbus’ selfimposed limits like maximum velocity, minimum humidity, etc.In addition, the nature of the cabin flow field has to bedesigned to ensure the risk for spreading airborne transmitteddiseases is kept to a minimum. Public concern over the spreadof Severe Acute Respiratory Syndrome (SARS) during 2003,ensured that aircraft cabin ventilation systems became thefocus of public interest.

Article by ICON - Simon Weston, Thomas Schumacher.ICON would like to thank Airbus-Deutschland for kindly allowing this work to be presented.

A long-standing co-operation with Airbus has seen ICON working on the designand improvement of ventilation systems for a number of different aircraftprogrammes such as the A380, A380-F, A340/330, A400M, A350XWB.

" MORE INFORMATION VISIT www.iconCFD.com

! Fig:01Cabin Ventilation for the all new Airbus A350XWB

! Fig:02Fully parametric, ready-to-mesh CAD model of the Airbus A380

! Fig:03STAR-CCM+ result for 75million cell mesh

" Fig:04Stairhouse flows can now be investigated

! Fig:05Flowfield of early A380 Economy Class design

Fig:01-04Copyright © symbol ICON CGLtd, Airbus S.A.S., Fixion-HCSGM

Fig:05Copyright © symbol ICON CGLtd, Airbus Deutschland GmbH

❐ IMAGE REFERENCES

STAR-CCM+ enables ICON to develop acomplete-cabin CFD simulationprocess for Airbus

ICON provides independent andconfidential technology and processconsulting. Since 1992, the companyhas been delivering tailor-madesolutions and advice to OEMs, SMEsand research organisations worldwide.CD-adapco software has enabled ICONto deliver solutions for our customersmost demanding requests.www.iconCFD.com

a = 1/4c

KEYα = 300

Re = 1000Xv = 2c

Xa = 2c

Dynamics #2724


23Dynamics #27


During the beginning of 2nd and 4th regions of the flappingmotion, the airfoil enters the trace of the leading edge vortex,translational vortex and rotational vortex that were generatedbefore the return. At angles of attack higher than 30°, theeffects of the downwash due to these vortices are stronger. Forsmall angles, the lift coefficient has both negative and positivevalues, which give average force coefficients close to zero. As theangle of attack is increased to 30°, these negative peaksdisappear and at 45° positive peaks are observable.

Vortex identification by considering Q-criteria and first and secondeigenvalues are visualized. The demonstrated results are obtainedby using STAR-CD for a 2-D, unsteady, laminar flapping motion.High positive values of Q imply vortex regions where the rotationrate is dominant compared to the strain rate. The vorticitycontours and pressure values are also represented forcomparison to the different techniques. The streamlines arevisualized in both inertial reference frame and body fixedreference frame.

The whole computational domain is moving, with the motionimposed by user subroutines, defining the flapping motion angularand translational velocities. The motion of the grid domain closeto the airfoil is shown in Fig.2. The whole domain is a 20c radiusO-type computational grid. Macros were developed for thecalculation of the second invariant of velocity gradient (also calledas second invariant of the mean rate-of-displacement tensor)Eq.1 and eigenvalue of the sum (called theµ criteria):

(Eq.1)

The results are also compared with the experimental visualizationtechniques as Particle Image Velocimetry (PIV) and laser sheetvisualizations (Figure 5). The numerical solutions are verysatisfactory compared to experimental results although theproblem is highly unsteady.

ConclusionIn this study, the flapping motion aerodynamics was consideredfor a symmetrical hovering case for use in future Micro Air Vehicleapplications. MAV’s resembling insects and small birds withflapping wings. The complexity of the problem raises the necessityof a simplified model, so a two-dimensional model wasinvestigated with a symmetrical airfoil with variable velocity andangular velocity laws.

The analysis tools used for the description of the phenomena arethe numerical simulations and the experimental investigations.The numerical simulations were performed with STAR-CD usingthe moving grid capability. STAR-CD was used for the parametricalstudy to get a first idea of the parameters which influence theflapping motion study. It was concluded that the most influencingparameter is the starting angle of attack. The second importantparameter is the center of rotation. The other parameters as thechange of position of the velocity and angle of attack and Re (orReynolds number) were found to be less important.!

D.Funda Kurtulus 1,2, Alain Farcy 1, Laurent David, Nafiz AlemdarogluEcole National Superieur de Mechanique et d’Aérotechnique, LEA, Poitiers, FranceMiddle East Technical University, Ankara Turkey

Flapping Airfoil Analysis of MicroAir Vehiclesusing STAR-CD

! Fig:02Flapping motion definition.

" Fig:01A hummingbird in hover.

" Fig:03 - Vorticity contours and aerodynamic force coefficients forthree different center of rotations at half amplitude location.

" Fig:04: Vorticity contours with velocity vectors related with lift coefficient at different time instants for different α.

! Fig:05 Laser sheet visualization(far left) and vorticity contours of PIV measurement(middle) compared withSTAR-CD results (right).


Flapping Motion Description Inspired from Birds and InsectsHummingbirds and several insects use normal hovering where thewings are moving through a large angle in an approximatelyhorizontal plane making a figure-of-eight motion with asymmetrical half-strokes.

The flapping motion is divided into 4 regions with the first regioncorresponding to half of the downstroke where the leading edge ispointing in positive direction and second one to the half-upstroke.While the third and fourth regions, are the mirror images of thesetwo regions, corresponding to the second half of upstroke anddownstroke respectively. Figure 2 shows a detailed description ofthe flapping motion in one period. In order to use a simplifiedmodel, the symmetrical NACA 0012 airfoil is chosen for thepresent study. The nomenclature of the parameters is such that α represents the angle of attack, c is the airfoil chord, V is the

translational velocity, xa is the position of change of angle ofattack, xv is the position of change of velocity. In the figure, thecenter of rotation (denoted by a) is at chord location of the airfoil.The motion is implemented into the program using the movinggrid option and user defined subroutines of STAR-CD.

Parametrical Study with STAR-CD and Comparison withExperimentsA total of 216 cases were investigated numerically, by changingthe parameters described in Figure 2. The vorticity contours at thehalf-amplitude location for different center of rotations are given inFigure 3 for α=30°, Re=1000, xv=2c, xa=2c. During thisparametric study it is found that the most influential parameter forthe aerodynamic force coefficients is the angle of attack and thesecond is the center of rotation.

The aim of the flapping airfoil analysis is to understand the aerodynamicphenomena and the vortex topology of this highly unsteady motion. Instead of the use of real insect/bird wing geometries and motions, which arehighly complex and difficult to imitate by a exact modeling, a simplified model isused to understand the unsteady aerodynamics and vortex formation during thedifferent phases of the flapping motion.

PIV: t*=0.9257, V/V0= - 0.8728, aoa=47.40û NUM.: t*=0.9329, V/V0= - 0.8391, aoa=48.40û

a = 1/2c a = 3/4c


Founded in 2000, Felt Racing is an Americanmanufacturer of high-end racing bicycles, partic-ularly aimed at the demanding Triathlon and Time-Trial markets. Under the leadership of Felt, an

internationally renowned frame-building guru with a host ofworld-championship winning designs under his belt, FeltRacing have quickly established a reputation for technologicalinnovation and aerodynamic design, with a stated mission “todesign, develop, and deliver the best bicycles in the world.Period.”

The DA is a significant step in that direction, featuring aremarkably narrow (25mm) frame, with aerodynamicallyoptimised tubing shape and innovative wind-defying featuressuch as a revolutionary brake-mounting that sits inside theseat tube, and a unique bayonet steering system.

According to Felt Racing Frame Designer Tim Lane, who wasresponsible for most of the CFD simulation, aerodynamicsplay a crucial role in Time Trial racing: “With no team-matesto pull you through and no wheel to draft, Triathlon and TimeTrial require not only a strong engine, but also a vehicle thatis ergonomically and aerodynamically advantaged. Racersmust convert every last ounce of energy into raw speed, andslice through the wind like a razor.”

In order to make sure that wind-tunnel resources are exploitedto their full potential Felt adopt a complimentary approach,using CFD simulation to determine which designs are the mostaerodynamically efficient, and only testing the best in the wind-tunnel. “As a company we’ve invest heavily in wind-tunneltesting”, says Lane, “but we recognize that wind-tunnel testingis both expensive and time consuming. By using CFDsimulation right from the start of the design process, we canensure that by the time we get to the tunnel, we are fine-tuning an already aerodynamically efficient design.”

Tim Lane and his colleagues at Felt Racing have establishedan impressive process for CFD modeling so that “right fromthe start of the design process”, literally means from themoment that first CAD models are generated, usually manymonths before prototypes are built. Through a process calledCAD-embedding, Lane and his team can access CD-adapco’sCFD software directly from their Pro/ENGINEER CAD package.This enables designers to perform CFD simulations of theircurrent design by expending just a few minutes of effort, withall the CFD functionality available from a small number ofadditional menus in the CAD tool.

Results of the CFD simulation (which typically take less thanan hour to compute using a standard desktop computer) are

! Fig:01The Felt DA is the fastest UCI legal Tri / TT bike built today. Developed over two years with careful use of NACA aerofoil profiles, CFD flow modelling and wind tunnel experimentation, the DA frame system delivers unsurpassed aerodynamic advantage with incredible drive-train and steering stiffness.

Felt Racing recently unveiled their new DA Carbon Fibre racing bicycle that,according to company founder Jim Felt, is designed to be “the mostaerodynamic, UCI-legal frame ever created.” Although Felt’s claim is a boldone, he has a sheaf of wind-tunnel data to prove it, and can point to a two-year design process for the bicycle that involved extensive CFD simulationright from the start.

automatically presented to the designer in terms of drag-coefficients, forthe whole bicycle.

The results are not always what the designer originally expected:“Bicycle aerodynamics is about the interaction between all the differentcomponents that make up the complete bicycle”, says Lane. “Justbecause a component or concept looks good on the CAD-screen orseems viable in theory doesn’t mean that it will work out on the road.Unless you are very careful, an aerodynamically optimized componentcan sit in the dirty air generated by someone else’s beautifullydesigned, yet aerodynamically inconsiderate component, thus stillgenerating a whole heap of drag.”

Lane and his team investigate any unusually good, or unusually badresults by examining a predefined set of flow-visualization plots that areautomatically generated and stored for each design simulated. “Thebeauty of CFD is that if we want to, we can investigate every singlecomponent, and look in detail at the flow-features that it generates”,says Lane.

This thorough investigation of the design envelope is warranted becausecompetitive cycling, like Formula 1 motor racing, is bound by a verystrict set of regulations, which are defined by the sport’s governing body(the Union Cycliste Internationale or UCI). The regulations are specif-ically designed to maintain the traditional shape of a bicycle and to limitthe scope for manufacturers such as Felt Racing to gain significantcompetitive advantage for their riders. As the DA proves, this doesn’tmean that there’s nothing that can be done: “In designing theframeset we took advantage of every possible lenience permitted withinthe UCI rules”, says Lane, “it’s not just a frame – but a completelythought out frameset comprised of a frame, fork and stem, blendedtogether as a single unit”.

Because the wind-tunnel mock-ups were unable to support a ridersweight; when the basic bicycle design had been decided upon, Laneused additional CFD modeling to see check that the bicycle performedwith a rider in a number of aerodynamic riding positions. Rider andbicycle were combined in CD-adapco’s STAR-CCM+ and joined togetherusing advanced surface meshing, that creates a single contiguoussurface suitable for CFD modeling, while respecting the complexgeometry of the bicycle – right down to every gear-tooth on thegroupset.

Importantly, using CFD Felt Racing were able to speed up their designprocess: “Of all the CFD technology we tried, only CD-adapco’scombination of CAD-embedding and surface wrapping provided a robustand efficient process by which we could optimise our designs withoutdelay to our demanding production schedule”, says Lane. “Ultimately,using CFD, we were able to build a more aerodynamically optimisedbicycle at less expense, because of the cost and time saved in reducingthe number of wind-tunnel prototypes.”

Although it is difficult to say whether Felt Racing have achieved theiraim “to design, develop, and deliver the best bicycles in the world”,every triathlete and time-triallist that manages to race faster, using less-energy, because of Felt Racing’s investment in CFD technology, willprobably agree that the DA is a significant step towards it.!

" Fig:02 (overleaf)Felt Racing Design Engineer Tim Lane aboard the revolutionary DA carbon-fibre bicycle.

" MORE INFORMATION ON FELT RACING AND THE DA http://www.feltracing.com/

CD-adapco help Felt Racing to design “the most aerodynamic,UCI-legal bicycle frame ever created” Stephen Ferguson, CD-adapco

Business Benefits

Estimated tripling of potential sales volume due to lower feed pressurerequirement: from 400,000 units/year to around 1.2 million units, apotential top-line revenue increase of approximately $16 million.

Dynamics #2728

..::FEATURE ARTICLE Biomedical

27Dynamics #27

..::FEATURE ARTICLE Biomedical

Expectations were high that incorporating STAR-Worksinto VIASYS’s already efficient design process woulddeliver significant savings in design cost and result ina faster time to market. Even so, Steve’s

management were shocked when, just a few days afterinstalling STAR-Works, he delivered with guidance from SteveDuquette, Director of Research and Development, a revolu-tionary new ventilator device design; something that wouldnormally take upwards of a year in design effort.

In Steve’s own words: “Within three days STAR-Works had paidfor itself!”

VIASYS Healthcare, is a global market leader in health caretechnology. The Respiratory Care division delivers a compre-hensive line of respiratory products serving the smallest neonateto the largest adult patient.

Steve and his colleagues were working on the redesign of aninfant CPAP ventilator device. Continuous positive airwaypressure (or CPAP) is a method of respiratory ventilation, usedin the treatment of apnea, a condition in which a patient’snatural external breathing suddenly stops due to a self imposed(but involuntary) obstruction to their airway. In adult patients themost common manifestation is sleep apnea, an extreme form

of snoring, during which the soft tissue at the back of thepatient’s throat becomes relaxed, collapsing the airway andblocking the passage of air, resulting in a temporary interruptionto the patient’s breathing.

The CPAP unit in question is used in the care of small infantsthrough to premature babies (although not for those in a criticalcondition). Apnea is relatively frequent occurrence in newborninfants due to immaturity of the respiratory system. While inadults the most likely symptom of apnea is serious fatigue, theconsequences in small infants are potentially lethal - anybreathing interruption over 20 seconds has the potential tocause serious damage or even death.

The CPAP machine works by delivering a constant stream ofcompressed air via a mask that fits over the patients nose (ornose and mouth). The slight positive pressure delivered to themask acts to splint the patient’s airway, preventing obstructionand allowing the patient to breathe freely. Maintaining theconstant pressure prescribed for each patient is critical; if thepressure at the mask is too high then it opposes the patient’snatural exhalation and increases the so called “work ofbreathing”.

Steve’s analysis concentrated on the redesign of the jet-venturi, the critical component in regulating the patientpressure. The orifice of the venturi measures just 0.04 of an

Like most engineers, pneumatic specialist Steve Han is accustomed to thehard slog that accompanies the design of any new product. Realizing that smallimprovements in product performance are usually only accomplished throughmany weeks or even months of detailed engineering analysis, Steve’semployers, VIASYS Healthcare, recently purchased STAR-Works.

inch, making physical testing of prototypes very difficult due to the tightmanufacturing tolerances involved and the possibility that small defects couldsignificantly influence the results of the test program. Using STAR-Works, Steveand his colleagues were rapidly able to evaluate multiple design changes fromtheir desktop computers. All of the analysis was performed from within theSolidWorks environment. As all of the CFD functionality is accessible throughadditional menus in the SolidWorks GUI, Steve was able to begin using STAR-Works for serious design analysis within a few hours of first installing. “Thewhole project took about three days, with about a day to get familiar with thesoftware”, says Steve.

The improvements were both immediate and impressive. “It was difficult tobelieve at first”, says Steve, “almost every design change resulted in asignificant improvement. Incredibly, within three days we had reduced thesupply pressure required to the unit by 48%.”

Part of the reason that the analysis was able to proceed so rapidly is that theSTAR-Works CFD results are linked directly to the CAD geometry (a processcalled associativity). After any modification in the CAD model the simulationresults can be updated almost instantly by clicking the “update solution”button, allowing the rapid and thorough investigation of the design space.

Excited by the results, VIASYS management immediately approved thefunding for the construction of a prototype, the testing of which immediatelyconfirmed the improvements predicted by the CFD analysis. “We were prettyamazed,” says Steve, “the results of the first prototype were within a fewpercent of those predicted by STAR-Works.”

This significant reduction in supply pressure has a number of positive ramifi-cations, not least of which is a significant increase in the saleability of theproduct. “Saving 48% of the supply pressure allowed us to use the CPAP unitwith the majority of our patient ventilators, rather than just two ‘critical care’ventilators as had been the case previously”, says Steve. “The reduction in costachieved in doing this alone, has the potential to treble sales of the product.”

Because a much lower supply pressure is required the new design is alsoaround 30% more efficient than previous designs: “Because the new CPAPdesign uses less energy than its predecessor it is both cost effective to run andmore environmentally friendly.”

By reducing the development time to just a few days from a year, VIASYSHealthcare has managed to save several hundred thousand dollars indevelopment costs.

At the time of writing, the new ventilator device is undergoing patient testingand will be on the market shortly.!

STAR-Works savedus $250,000 inphysical testingalone, and paid foritself in three days!Steve Han, VIASYS

“”❐ FACTS Technical Benefits

� Reduction in feed pressure by 48%� Improved the flexibility of the product and reduced power requirement� Improved patient comfort� Design phase reduced from 1 year to just 3 days

STAR-WORKShelps VIASYS healthcarebreathe more easilyStephen Ferguson talks to VIASYS Healthcare’s Steve Han

! Fig:01VIASYS CPAP ventilation mask

! Fig:02Flow path through CPAP unit

! Fig:02Contours of total pressure in CPAP unit

Dynamics #2730

..::FEATURE ARTICLE Pharmaceutical

29Dynamics #27

..::FEATURE ARTICLE Pharmaceutical

BeardingA common problem with spray guns is an effect known as“bearding”. This is when deposits of coating fluid form on thenozzle surface. In severe cases, these deposits can beenough to block the nozzle.

Early experimental tests on one of Manesty’s new spray gunmodels, revealed that bearding had occurred. Rather thanundertake a potentially lengthy and costly trial-and-errorexercise to resolve the problem, Manesty deployed CFD inorder to understand and resolve the issue.

STAR-Pro/E GatewayAlthough Manesty possessed significant in-house expertisewith CAD tools (primarily Pro/E), they were new to CFD.Consequently, an incremental approach to the simulation wasadopted using CD-adapco’s STAR-Pro/E Gateway.

The benefits of STAR-Pro/E Gateway are numerous. First,the user operates within the Pro/ENGINEER environment, sothe learning curve for engineers new to CFD is significantlyreduced. Second, 3D-CAD tools are significantly better athandling/creating/modifying geometry than dedicated CFDtools: that is after all, what they’re designed for. With thiskind of CAD-embedded solution, the CFD model is directlyassociated with the CAD model, so when the design ismodified in Pro/E the CFD model is also automaticallyupdated. Finally, STAR-Pro/E Gateway is based on thetechnology of STAR-CD and so the user has full access to thecapabilities of STAR-CD. An issue commonly found with lesscomprehensive “upfront” solutions is that the user quicklyoutgrows the limited capabilities of these tools. With STAR-Pro/E Gateway, models are entirely compatible with STAR-CD,so after Manesty had completed the cold flow analyses, theywere able to progress to running compressible flow,multiphase spray analyses in STAR-CD.

The analysisUsing STAR-Pro/E Gateway the geometry was created andmeshed, and boundary conditions were applied withinPro/ENGINEER. Calculations on an initial design were run andthe results analyzed.

Figure 2 shows that there are large areas of low pressure(seen in blue) close to the nozzle surface: the scales havebeen limited to accentuate these regions. Droplets of coatingfluid are entrained into these “dead zone” or areas of re-circulation, and deposit on the nozzle. Manesty’s expertengineers were then able to make some quick designalterations within Pro/E in an attempt to resolve this issue.Once the geometry had been modified, they simply had toclick one button to obtain updated flow results and to identifya new design, where the low-pressure regions had beenmoved away from nozzle surface.

Buoyed by their success, Manesty’s engineers quicklyprogressed onto more advanced simulations. These fall intotwo categories: multiphase and thermal.

One of the key design parameters for the sprays guns is theshape of the spray. As mentioned earlier, side-jets (fan air)are used to alter the shape of the jet. Without these the jetwould be a hollow cone, with them it forms an elliptical shape.Manesty wanted to understand whether they could reproducethis effect using CFD – thereby enabling them to run detailedstudies on how to obtain an optimized spray shape. The nextstep for their multiphase project is to analyze the film as itforms.

The thermal analyses concentrated on the cabinets inwhich the spray guns sit. These rotate the tablets, movingthem into position to be sprayed and then rotating them whilethey dry. Air is blown through the cabinet to dry the tablets.The analysis involved understanding the flow through thecabinet, and to see whether it would be sufficient to influencethe flow in the vicinity of the spray guns.

ConclusionThe pharmaceutical industry has identified that by adoptingtechnology commonly used in other industries it can enhanceprocess understanding, and improve manufacturing efficiency.An example is Manesty, who have successfully adopted theuse of CFD, a tool commonly used in other industries, withintheir design cycle. As a result of this investment, designproblems discovered early in the development cycle wereinvestigated, understood and resolved: reducing overalldevelopment costs while improving product quality.!

One such technique is Computational FluidDynamics (CFD). CFD is the simulation of flow andthermal behavior. It can be used to analyzeanything from flow around or through objects,

through to chemical reacting flows, phase change and heattransfer.

In other industries (such as automotive and food) CFD hasnow become an integral part of the product design andmanufacturing process; almost every component on a modernautomobile is designed with the aid of CFD. Manestytraditionally has had a strong ethos of technical excellence,which means that they are always seeking to leveragetechnology to improve the performance of their products forend user companies. This article details how they haveadopted CFD technology and successfully deployed it withintheir design process. This has enabled their expert engineersto understand and correct problems early in the design cycle,thereby considerably reducing product development costs atthe same time as improving product performance.

Tablet coatingManesty supply the pharmaceutical industry with tabletcompression and coating equipment of which spray guns arean integral part of the process. Coating improves theappearance of tablets, makes them easier to swallow,modifies drug release, masks taste, and gives environmentalprotection.

The coating process consists of a number of complex fluiddynamic phenomena. Coating fluid and air pass throughchambers within the spray gun; the coating fluid is thenatomized at the spray gun nozzle; a further two side-jets(known as “fan air”) are directed into the side of the main jetto control its shape. As the droplets are transmitted betweenthe nozzle and the tablets, further droplet break-up andcoalescence occurs and moisture is lost from the droplets byevaporation due to the airflow in the coating chamber. Oncontact with the tablet surface the droplets stick, splash andspread before drying to form the film coating. This continuouscoating process is carried out over a moving bed of tabletsuntil a uniform coating is applied to each of the tablets.

Any problems during this process can have a detrimentaleffect on the end product: “logo bridging”, edge chipping andsplitting, color variation, orange peel effect, picking, twinningand surface roughness are common descriptions of coatingproblems.

When they occur, fixing these problems can be a lengthyand expensive process. Traditionally a trial-and-error approachis adopted, varying the spray gun settings or design. Manestydecided that it would be more cost-effective to invest ingaining a deeper insight into this process early in the designcycle, thereby improving product performance and significantlyreducing total costs.

! Fig:01Shows velocity magnitude plots through two sections close to the spray gun nozzle.

In recent years, questions have been raised about how current manufacturingmethods in the pharmaceutical industry can be improved. One of the catalystsfor this discussion was a directive from the US Government’s Food and DrugAdministration on innovative manufacturing and quality assurance. The directivechallenged the pharmaceutical industry to enhance the understanding of theirprocess, to improve manufacturing efficiency, and to investigate the adoption oftechniques that are routinely used in other industries.

ManestyMedicine Manufacturewith STAR-Pro/EAlex Read - CD-adapco.Barry Lyon, George Smith - Manesty


! Fig:02Shows pressure contours close to the spray gun nozzle.

shown in Figure 3 show the flow patterns inside theincinerator. Temperature contours on the incinerators’centerline are shown in figure 4. Other simulations were carried out with reduced excess airand pre-heated combustion air in order to determine the levelof reduction of CO emissions that could be achieved. Thesimulations showed that these options would reduce COemissions. However, the comparison of predicted emissionswith the field data for current operational set points andreduced excess air showed lower predicted CO emission levelsthan those measured when averaged over an entire day’soperation.

Following the presentation of these results to the engineersand operators responsible for the incinerator, the chiefoperator remarked that the current automated control loopwas based on assuring a steady steam generation rate ratherthan steady state conditions for air flow and the quantity ofwaste being burned. The chief operator then proceeded torun his own tests by manually controlling the air flow anddisposition of waste on the combustion grates. The COemissions measured during manual control closely matchedthe predicted CO emission levels to within a few ppm.Furthermore, the fluctuation of CO emission levels was alsogreatly reduced.

This study has shown that with the intimate knowledgeafforded by STAR-CD of the combustion phenomena insidethese industrial units, operators and engineers can easilydevise ways of improving the process. Ongoing studies are

currently being carried out to determine if further emissionreductions can be achieved by further modification of theoperational set-points. Other physical modifications to theincinerator are also being investigated. These include modifi-cations to the position and dimensions of the over fire airnozzles as well as the pre-heating of the humidified flue gasesthat were used to dry the sewage sludge prior to being re-injected into the incinerator (F).

“STAR-CD has become an essential tool for BMA engineering.It enables us to qualify quantify ways of enhancing theenvironmental performance and efficiency of industrial sizecombustion equipment”, says Francois McKenty.!

This incinerator in question (shown in Figure 1) isused to burn municipal waste composed of organicand inorganic matter on the primary (A) and finishing(B) grates. The incinerator has been modified over

the years in order to dispose of additional waste such as driedsewage sludge, which is injected as pulverized particles (C).Over fire air is injected on the front and rear walls (D).Additionally, 8% of the hot flue gases are extracted at port (E)and used to dry the sewage sludge. The humidified flue gasesare then re-injected into the incinerator (F) in order to destroythe odors resulting from the VOC’s (Volatile OrganicCompounds) emanating from the drying of the sewage sludge.Current operational data showed highly fluctuating levels ofcarbon monoxide, which sometimes exceeded new emissionstandards. BMA’s goal was to find solutions to this problem.

In order to tackle this task, a new 7-stream combustion modelwas developed by BMA to take into account for multiplestreams of fuels of varying composition and humidity. Five ofthe streams represent different fuel and FGR (Flu GasRecirculation) compositions, while the two remaining streamsrepresent dry air and water vapor.

Four of the five fuel streams are for waste fuels; one extra fuelstream was included in the model to account for an auxiliary

natural gas burner, which seemed likely to be needed. Thestreams were modeled by the transport of conserved scalarsrepresenting the mass fraction of each of the fuels, dry air andwater vapor. Water vapor is accounted for separately becausethe concentration of water vapor varies greatly from stream tostream.

The local mixture fraction for each of the fuels was obtained bydividing the local mass of available air and water vapor betweeneach of the fuel streams. Each fuel was assumed to reactindependently.

The reactions were modeled using a chemical equilibriumhypothesis. Local chemical equilibrium was modeled byminimization of Gibbs free energy. The combustion productswere then recombined to yield the local concentration ofchemical species, which were passed back to STAR-CD via theUser subroutines. The burning process on the grates (A and B)was modeled externally using a simple pyrolysis model yieldingcombustion hot product streams composed of CO, CO2, SO,SO2, H2, H2O and N2. The distribution of the waste and by-pass air on the combustion grates was modeled as air andpyrolysis product inlets disposed in a checkerboard fashion. Thedried sludge stream (C) was modeled using Lagrangian particletracking.

An initial steady-state simulation was carried out using flow ratesobtained from the operational data logs. The flame contoursshown in Figure 2 closely resembled those observed through theincinerator’s viewports by BMA’s field engineers. The streamlines

Operational Optimization ofa Municipal WasteIncineratorF. McKenty, L. Gravel, L. CharestBMA – Brais Malouin & Associates Inc, Montreal, Canada

Dynamics #2732

..::FEATURE ARTICLE Chemical Process

31Dynamics #27


As part of an ongoing modernization program, Brais, Malouin and Associates(BMA) recently undertook the simulation of a municipal waste incinerator, withthe ultimate goal of reducing pollutant emissions and improving the overallefficiency of the process.

! Fig:01 - (left)Schematic of incinerator

! Fig:02 - (right)Flame contours

! Fig:03 - (left)Flow patterns inside incinerator

! Fig:04 - (right)Temperature contours on the incinerator


❐ FACTS Gibbs free energy

Gibbs free energy is a thermodynamic potential, which measures the�useful� work obtainable from an isothermal, isobaric thermodyamicsystem - Wikepedia

STAR-CD has become an essential tool forBMA engineering. It enables us to qualifyquantify ways of enhancing the environmental performance and efficiencyof industrial size combustion equipmentFrancois McKenty, BMA

“”

Dynamics #2734


33Dynamics #27


“We quickly realized, that to accurately handle thediverse issues that arise at each particular location, itis more effective to implement CFD at every plantthat has analytical needs”, says Hiroaki Nagai, an

Engineer in Mitsubishi’s Computer Aided Modeling andSimulation Lab in Kurosaki. “In the beginning, CFD was usedlargely for trouble-shooting within the Kurosaki Plant, butawareness in the field eventually grew to the point whereimplementation expanded to such areas as preliminaryinspections prior to plant construction, and our license usagerapidly increased.”

More than half of the analysis conducted at the Kurosaki Plantis focused on tank stirring, with some 70-80% concerningmultiphase analysis. In simulating these problems Mitsubishiengineers employ Lagrangian, Eulerian or free-surface modeling,each method being used in about a third of cases.Since no ‘off-the-shelf’ empirical data is available for many ofMitsubishi’s chemical reactions (particularly those to do withpolymerization), the Kurosaki group maintains a process anddesign lab that is equipped with the reaction analysis technologyrequired for determining reaction speed from experiments andoperation data. In order to derive reaction parameters for theCFD model, testing samples are actually mixed in a beaker andthe test data processed analytically, often reducing the manyreactions down to the few most important. Reaction parametersare then integrated into STAR-CD using user subroutines.

Mitsubishi originally adopted STAR-CD in Kurosaki because ofthe strength of its meshing generation process and its ability toconstruct unstructured meshes quickly and easily. “I like theease with which models can be created and, should any occur,it is also easy to correct problems with the mesh quality”, saysMitsubishi Engineer Masaru Futagawa. Through many years ofusage they have also come to depend on the strength androbustness of its solver: “Compared to other software,convergence is superior”, says Nagai. “In a word, the robustnessof the software is amazing and with the powerful user

subroutines, we are able to perform a wide range of complexanalyses.”

In addition to tank stirring simulations, Mitsubishi’s CFDEngineers are regularly called upon to conduct other types ofCFD simulation in support of plant activities. Recently they havebeen involved in modeling smoke dispersion in the plant, and ananalysis of a wastewater purification tank.

The team at Kurosaki has also successfully completed extensiveCFD analyses of a granulation tower at the plant, a device thatforms powder from molten droplets of liquid as they fall into aplume of cold air. “The Lagrangian two-phase flow function wasused to the fullest extent in this analysis, providing extremelyinteresting results”, says Nagai.!

Reproduced in English from an original Japanese interview conducted by Shuichi Ogawa for CDaJ

The first in a series of articles that explores how CD-adapco’s software is used in some of Japan’smost successful companies.

Mitsubishi Chemical Corporation is one of Japan’sleading diversified chemicals companies and is a longtime investor in CFD research, particularly in thecomplex fields of two-phase flow and chemicalreaction.

The use of CFD at Mitsubishi has spread rapidly sincethe early nineties. Starting from a single installationat the Mitsubishi Research Institute in Yokohama,CFD is now used in Mitsubishi plants across thewhole of Japan, including Kajima, Yokkaichi,Mizushima and Kurosaki.

In a word, therobustness of thesoftware is amazing andwith the powerful usersubroutines, we areable to perform a widerange of complexanalysis.

“

”


! Fig:02Flow paths in a mixer

Kajima

Kurosaki

Yokohama

YokkaichiMizushima

City

scap

eWal

lpap

ers.

Com

Big in Japan

Dynamics #2736

..::FEATURE ARTICLE Building & Environment

35Dynamics #27

[1] Rodi,w. Ferziger,J.H, Breurer,M. Pourquie,M. “Status of large Eddy Simulation: Results of Workshop, Transactions of ASME, vol 119, June1997 pp 248-262

[2] Lee,B.E, 1975,“The effect of turbulence on the surface pressure field of a square prism”, Journal of Fluid Mechanics vol 69, part 2pp 263-282.

[2] Bearman, P.W & Obasaju E.D,1982 “An Experimental study of pressure fluctuations on fixed and oscillating square section cylinders”, Journal of Fluid Mechanics, Vol 119pp 297-321.

To date LES and more recently DES methods areregarded as being the only methods with a practicalchance of resolving the large scale instabilities thatdominate flows in the built environment. For wind

loading purposes, the large Reynolds numbers, geometricalextent and complexity of groups of buildings, imply that thecomputing power required to calculate these flows to anaccuracy comparable with wind tunnel experiments, makestheir resolution currently uneconomical by around two orders ofmagnitude. Several benchmark problems are used to test newdevelopments, of which vortex shedding from a square cylinderis probably the most universal.

It is usually considered that correctly modeling the velocitygradients perpendicular to the wall is a necessary condition forthe prediction of flow separation. However sharp edged bluffbodies have well defined flow separation points at their corners,creating shear layers that are highly unstable and drive vortexshedding. The upside to this situation is that resolution of thenear wall perpendicular velocity gradients is less important. Thedownside is the resulting instability requires time dependentsimulation to resolve the unstable flow close to the body andvortex shedding in the wake. The surface pressure distribution isdominated by these large scale structures. Tensys Dynamicshave been investigating the combined use of embeddedpolyhedral meshes and time dependent laminar simulations inSTAR-CCM+ to resolve these flows. The level of meshrefinement used in this study implies the size of eddiesresolvable does not qualify the method to be called DirectNumerical Simulation (DNS) it is rather a LES model with noexplicit sub grid scale model, with implicit sub-grid viscosityarising from the advection scheme upwinding.

Simulation DetailsThe classic case of vortex shedding due to cross flow over aninfinite square cylinder was simulated at Re = 22,400. Thecomputational domain was arranged with a cylinder of side D in

a rectangular domain such that the front of the cylinder was4.5D from the constant velocity inlet, while the constantpressure outlet was 20D from the back edge of the cylinder. Thedomain was 15D in height and 10D in span. The span wise andtop and bottom walls were defined as slip walls.

The simulations were run on a series of three progressivelyrefined meshes consisting of 40k, 260k and 620k polyhedralcells, each employing refined embedded regions within them.The height of the first cell from the surface was kept, constantat 0.025D for all cases. The embedded meshes were created inSTAR-Design utilizing a recursive method of imprinting nestedsub domains which captured the cylinder and its wake. Theimprinting method has the flexibility to create sub domains ofany shape with arbitrary levels of refinement. This hasadvantages over traditional hexahedral trimmed cell embeddingwhere only rectangular blocks of meshes refined in multiples oftwo are possible.

The unusual choice of not using a sub-grid scale model wasmotivated by a desire to firstly assess the extent to whichnumerical diffusion contributed to the solution of conventionalLES simulations. Secondly to simultaneously assess theinherent numerical diffusion resulting from the use ofpolyhederal mesh and STAR-CCM+’s 2nd order spatialdiscretization, as a implicit sub-grid scale model.

The time step used for all the simulations was 0.1 units ofnon-dimensional time (D/V) giving a maximum Courant numberon the fine grid of 4. Second order spatial and temporaldiscretization schemes were used for all calculations. Eachsimulation was run for at least 50 non-dimensional time unitsbefore vortex shedding became sufficiently regular. After thistime mean Cp profiles and rms Cp profiles were extracted onthe intersection of the centre plane with the cylinder using fourline probes at 500 time steps corresponding to about 7 vortexshedding periods. The laminar time dependent method wascompared with simulations using unsteady k-w, DES and inviscidmodels on the coarsest grid only.

The requirement to successfully simulate mean and fluctuating surfacepressure distributions on the surface of buildings in the atmospheric boundarylayer is a challenging problem, which has occupied wind engineers for at leasttwo decades.

Large Eddy Simulationswith STAR-CCM+Peter Arnold, Tensys Dynamics, UK

! Fig:01Shows the centre plane velocity vectors. Figures 2 and 3show the distribution of themean and rms pressure coeffi-cients around the mid plane ofthe cylinder starting at thefacing wall. The experimentalmeasurements of this flow bydifferent researchers [1] showmore spread in the values thanthe difference between thethree meshes used. Both themean and rms Cp profiles areknown to be sensitive to theinlet turbulence intensity (T.I)with an increase in T.Isuppressing the magnitude ofboth [2]. Consequently a zeroT.I inlet condition can be viewedas the worst-case scenario froma wind loading perspective. Theexperimental data shown has aT.I of 0.04 % [3] and representsthe closest case to thenumerical simulations where noturbulent fluctuations wereimposed on the inlet velocityprofile.

Distance around cylinder

! Fig 02: Mean Cp distribution on surface of cylinder with mesh refinement.Distance around cylinder

! Fig 03: Cp rms distribution on surface of cylinder with mesh refinement.

Distance along cylinder Distance along cylinder

! Fig 05: Mean Cp distribution on surface of cylinder with turbulencemodel (Coarse mesh)

ResultsTables 1 and 2 compare the integrated parameters calculatedfrom the laminar simulations with mesh refinement and withturbulence modeling alongside the experimental data of severalauthors [1]. As expected the drag is the most consistentlymeasured in the experiments and most accurately predicted in thesimulations. Different levels of turbulence intensity and lengthscale at inlet were present in the experimental results.

On the coarse mesh the laminar predictions are closer to theexperimental values than the other models. In particular the k-wURANS and DES models add additional unnecessary dampingwhich reduce all parameters. It is quite possible that the sevenvortex shedding cycles over which the integrated parameters wereextracted over, is an insufficient period. However time constraintson the finest mesh made this necessary.

The effect of grid refinement is seen to steadily increase themagnitude of both the rms and mean Cp values. The increase inthe rms values is consistent with the reduction of numericaldiffusion originating from the spatial and temporal discretization.The mechanism which correlates an increase in rms Cp valueswith an increase in mean Cp values in the experimental results isunclear but appears to be mimicked in the numerical results.

The mean and particularly the rms Cp profile from the medium260K mesh matches the experimental values well, indicating thatthis amount of numerical dissipation acts as an effective sub gridscale model for this level of inlet turbulence. The coarser meshsolution clearly adds more numerical diffusion and suppresses therms and mean Cp values further while the finer mesh adds lessresulting in increased wind loadings.

The effect of various turbulence models on the 40k coarse meshis shown in figures 4 and 5. Clearly the addition of more turbulentviscosity worsens the result, which is especially true of the k-wURANS model. This implies there is already too much dampingpresent in the solution from the inherent numerical diffusion.

ConclusionsSimulations of sharp edged bluff bodies such as buildings in thebuilt environment are dominated by large scale unstablestructures originating from the flow separations at the buildingscorners. The satisfactory resolution of these structures can beachieved using an unsteady laminar simulations in STAR-CCM+which are equivalent to an LES model with no explicit sub gridscale model, provided the inherent numerical diffusion, arisingfrom the discretization process, is monitored and reduced by aprocess of mesh refinement. In comparison the type of sub gridmodel used in conventional LES simulations is far less important.

For this benchmark flow we have found that LES simulationswith no explicit sub grid model using embedded polyhedralrefinement gives results accurate enough for wind loadingpurposes with a simulation time equivalent to 50 hrs on a 3.00MHz PC (260k mesh). Without imposing turbulent fluctuations onthe velocity inlet profile equivalent to the experimental values,further mesh refinement may lead to an over estimation ofmagnitudes of mean and rms pressure coefficients, however inthe wind loading context this is no bad thing and can beinterpreted as a worst case scenario.!

! Fig 04: Mean Cp distribution on surface of cylinder with turbulencemodel (Coarse mesh)

❐ REFERENCES

! Table 1: Mesh Refinement Laminar simulations ! Table 2: Turbulence model comparisons on coarse mesh

" MORE INFORMATION LES capabilities will be released with STAR-CCM+ V2.06

growing need to be able to simulate flows in urban areas forapplications ranging from pedestrian comfort and windloading of buildings to dispersion of traffic pollutants andterrorist releases. This new partnership promises the tools weneed to tackle these issues.” Fred Mendonça, CD-adapco’sdirector of Vertical Applications and Expert Systems adds, “Weare particularly pleased with NCAS’s decision and look forwardto productive collaborations. Urban Weather Modelingcontinues to be an area of keen interest to many researchgroups, environmental agencies and building services organi-zations worldwide; our commitment to this partnershipconfirmsour serious intentions towards environmental flowpredictions”.

Dr Xie has produced a set of demonstration predictions usingSTAR-CD v4.0. Results to date point clearly at the viability ofLES with polyhedra for building applications, validated againstexperimental measurements and the University’s own DNScode, used by Dr Glyn Thomas (its originator) and Dr OmduthCoceal (at Reading) to produce genuine DNS data for similarsituations. The first predictions were performed on laboratorymodels comprising simple box-shaped buildings. Results

pointed clearly to the viability of polyhedral meshes forbuilding environmental flows. CD-adapco software has nowbeen applied to more complicated models – for exampleLondon’s Marylebone district - see the Google Earth imageand a view of the polyhedral mesh at ground level in Figure 2.Mean flow and turbulence statistics, agree well with windtunnel data obtained using a 1:200 scale model of the regionshown in Figure 1.

In the next phases of the work, NCAS teams at theUniversities of Southampton and Reading will concentrate onexploiting CD-adapco’s modeling tools, centered around therecently released STAR-CD version 4.0, linking it with the UKMet Office’s codes and exploring the issues that arise inscalar (pollutant) dispersion and heat transfer effects.Licensing arrangements between CD-adapco and NCAS – freefor all approved NCAS PI’s working on urban meteorologyunder the Weather Directorate – were finally approved by allparties in August 2006. Further information is available fromProfessor Ian Castro at the University of Southampton.!

A detailed assessment of three major commercialCFD codes has led to indications that proper useof Large Eddy Simulation (LES) capabilities leadsto essentially identical results from all three codes

and is a feasible approach. NCAS teams, initially at theUniversities of Southampton and Reading led by ProfessorsIan Castro and Stephen Belcher, respectively, will concentratein the next phases on further exploitation of CD-adapco’sadvanced modeling techniques, centered around the recentlyreleased STAR-CD version 4.0.

Professor Castro, who with Dr Zhengtong Xie led theevaluation of commercial software at Southampton notes,“Partly for technical reasons but also because they have beenthe most engaged in the collaboration, CD-adapco has beenidentified as the vendor most appropriate for NCAS’s needs”.Then referring to CD-adapco’s own ParaSolids basedgeometry modeler and CAD integrated product, he adds that,“STAR-Design is very user-friendly for complex geometries,generating polyhedral meshes which we consider essential forgeneral urban topologies”. Professor Belcher said, “There is a

Fred Mendonça, CD-adapco

Just before his retirement in April 2006, NCAS director of weather research,Professor Paul Mason, and CD-adapco’s President, Steve MacDonald, agreedexclusive Partnership terms, which will lead to an urban scale weathermodeling code. The prediction methodology, using the UK Met Office’soperational Universal Modeling code to drive localized STAR-CD models atbuilding and street-level scales, is aimed eventually at 5-day forecasting ofurban air quality, dispersion and heat transfer.

Urban WeatherModeling continues tobe an area of keeninterest to manyresearch groups.

“”


Urban ScaleWeather Modeling NCAS and CD-adapco Partnership

! Fig:02Ariel photograph of Marylebone in London (Google Earth image)

! Fig:01Polyhedral mesh at ground level

Dynamics #2738

..::FEATURE ARTICLE Building & Environment

37Dynamics #27

! Fig:02Prof Paul Mason (right) with Steve MacDonald (centre) andProf David Gosman (Imperial College & CD-adapco, left) at the signing of the Memorandum of Understanding.

Dynamics #2740

..::REGULARS Training

39Dynamics #27

..::REGULARS Events

CRAY's IMEM April 15, 2007

Detroit, MI

SAE World Congress 2007 April 16-19, 2007

Booth 1901, Cobo Center

Detroit, MI

2007 PLM World ConferenceApril 23-27, 2007

Long Beach Convention Center

Long Beach, CA

2007 COE Spring ConferenceApril 29 - May 02, 2007

The Rio All-Suites Hotel & Casino

Las Vegas, NV

OTC(Offshore Technology Conference)April 30 - May 03, 2007

Booth 508, Reliant Center

Houston, TX

Electric PowerMay 01-03, 2007

Booth 1334,

Donald E. Stephens Convention Center

Rosemont, IL

AUVSI Unmanned Systems Europe 2007May 08-10, 2007Cologne, Germany

Engine Expo 2007 May 08-10, 2007Stuttgart, Germany

ASME Turbo ExpoMay 15-17, 2007Booth 412, Palais des Congres de MontrealMontreal, Canada

VTMS VIIIMay 20-24, 2007Stand 8, East Midlands Conference CentreNottingham, UK

NAFEMS World Congress 2007May 20-25, 2007The Westin Bayshore HotelVancouver, Canada

ABAQUS AUC 2007May 22-24, 2007Paris, France

All Energy 2007May 23-24, 2007Stand A36, Aberdeen Exhibition and ConferenceCentre (AECC)Scotland, UK

PTC/USER World Event 2007June 03-06, 2007Tampa Convention CenterTampa, FL

GO-EXPO:Gas and Oil Exposition 2007June 12-14, 2007Hall D, Stampede ParkCalgary, Canada

GIFA 2007June 12-16, 2007Duesseldorf, Germany

AIAA Fluid DynamicsJune 25-28, 2007Hyatt Regency MiamiMiami, FL

Power Gen Europe 2007June 26-28, 2007Madrid, Spain

HOLD THE DATE!

STAR AMERICAN CONFERENCES June 25-26, 2007Detroit, MI

June 28, 2007Seattle, WA

Upcoming eventsTrade Shows and Workshops

CD-adapco regularly participates in many global trade shows. To get the chance to talk inperson with our experienced and friendly representatives, please make a note of the datesbelow. For more information please contact our events staff: North America: Tara Firenze [email protected]: Maeve O’Brien [email protected]

Training at CD-adapco

We regularly hold CD-adapco product training sessions at our offices in: London,Detroit, Seattle, Nürnberg, Paris and Turin. Other courses as listed on our websitecan be scheduled to suit your requirements and information can be requested fromour training administrators (see below).

To register for a course:Use the online form, or request a faxable form from your training administrator:

USA: [email protected] (+1) 631 549 2300 x129 UK: [email protected] (+44) 020 7471 6200Germany: [email protected] (+49) 911 946433 France: [email protected] (+33) 141 837560Italy: [email protected] (+39) 011 562 2194

New CourseA new foundation level Introduction to CFD course is now being offered at CD-adapco for those with little or no training in FluidMechanics. The course is aimed at introducing new users, or non-users who encounter CFD, to the fundamentals of the subject. Pleasecontact your local office for more information.

Note: In most situations it will be possible to register trainees on the course of their choice. However, if requests for places on courses arereceived too close to the course date, this may not be possible. Availability of places can be obtained by contacting your local office.Please see below for our upcoming schedule of training courses. See our website for most up to date schedules.

Choose from courses including:• STAR-CD Basic• STAR-CCM+• pro-STAR advanced meshing• STAR-Design • Advanced STAR-CCM+• Advanced STAR-CD

• Advanced Meshing • Moving Mesh• Advanced Modeling• User Subroutines• Spray & Combustion• E2P

41Dynamics #27

..::REGULARS Dr Mesh

If there’s one tool that over the years has saved me more time thanany other, it has to be the surface wrapper. Gone is the time when Ihad to spend days if not weeks, closing gaps, fixing non-manifoldsurfaces, removing double surfaces, extracting the fluid volume and allthe other tedious operations that needed to be done before volumemeshing. A bit of clever surface wrapping and I can go from any oldCAD to volume mesh in no time at all. So this edition of Dr Mesh isdedicated to a couple of hints on getting the most out of the surfacewrapper. It also provides the perfect opportunity for me to share myrecently discovered flair for a little known modern art form: self-portraits through the medium of CAD!

Dynamics #2742

..::REGULARS Dr Mesh

Figure 1: Starting CAD portrait.A common desire when meshing is the ability to treat two areas of themodel in different ways: coarsen and remove geometric features in one,while retraining all detail and sharp corners in the other.

My starting point is some non-ideal CAD (obviously ignoring its aestheticvalue). My portrait is made of lots of different closed bodies (shown indifferent colors in figure 1) that overlap and intersect. My objective is toautomatically produce a high quality surface mesh, demonstrating how wecan use contract prevention, volume sources and lines to improve featurecapture and to focus mesh density in the specific areas.

Contact PreventionThis feature simply requires the user to specify which types should notintersect. The wrapper will then do its level best to prevent these twofrom touching. Figure 2 shows two wraps: on the right I specified contactprevention for all the parts, and on the left I have removed head, noseand mustache. With the contact prevention on all parts, two differencesare seen. First, the nose and mustache no longer join, and second thecapture of the sharp corner between the mustache and the head is muchimproved. Using contact prevention, even in places where we know thesurfaces touch, is a neat way of getting the wrapper to focus its attentionon intersections and thereby retain the sharp corners.

Figure 2: The power of contact prevention.LinesGenerally it is not necessary to put lines on the model, as the wrapperautomatically identifies sharp angles and will attempt to resolve them.Where lines are useful is if you need to resolve a feature on a flatsurface, or that is defined by a shallow angle. For example, in figure 3we can see two wraps. In the first wrap (upper) a line and new regiontype have been defined on the central section of the lamp (where thebulb would be). These have been removed in the second (lower) wrap.As you can see, the interface between the two regions has been wellresolved in the wrapped mesh, despite it being on a flat surface.

Figure 3: Resolving features with lines.Volume SourcesVolume sources are a method of specifying areas of local refinement inthe mesh. In figure 4 you can see two pink boxes (or sources) aroundthe left eye and eyebrow. These sources can be a number of differentshapes – bricks, spheres, cylinders and cones – and I can either specify arelative mesh size or an absolute value. Here, it’s set to 5 % of the basesize. By using volume sources I can get highly detailed capture of specificfeatures, without significantly increasing my total mesh count.

Figure 4: Focusing the mesh with volume sources.Having used the volume sources in the surface wrapping, they can alsobe applied in the volume meshing (see figure 5) – another of STAR-CCM+’s time saving devices!

Figure 4: Dr “Poly” Mesh.Whether the CAE world will ever catch onto to the exciting possibilities ofCAD based art, I can only hope (I’m currently working on a portrait of HerMajesty the Queen), but it’s certainly been sold on the benefits andtimesavings of surface wrapping. Yours,

Dr. MeshDr Mesh (Ph.D. CFD)

! COMMENTS & SUGGESTIONS [email protected]

global offices - cd-adapcomdx2.plm.automation.siemens.com/.../files/magazine/pdf/dynamics2… ·...

Documents