doi: 10.1098/rsta.2009.0035, 2741-2751367 2009 Phil. Trans. R. Soc. A

 Tom Crick, Peter Dunning, Hyunsun Kim and Julian Padget and workflowsEngineering design optimization using services  

Referencesl.html#ref-list-1http://rsta.royalsocietypublishing.org/content/367/1898/2741.ful

This article cites 9 articles

Subject collections

(34 articles)computer modelling and simulation   � collectionsArticles on similar topics can be found in the following

Email alerting service herein the box at the top right-hand corner of the article or click

Receive free email alerts when new articles cite this article - sign up

http://rsta.royalsocietypublishing.org/subscriptions go to: Phil. Trans. R. Soc. ATo subscribe to

This journal is © 2009 The Royal Society

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

Engineering design optimization using servicesand workflows

BY TOM CRICK1, PETER DUNNING

2, HYUNSUN KIM2

AND JULIAN PADGET1,*

1Department of Computer Science, and 2Department of Mechanical Engineering,University of Bath, Bath BA2 7AY, UK

Multi-disciplinary design optimization (MDO) is the process whereby the often conflict-ing requirements of the different disciplines to the engineering design process attempts toconverge upon a description that represents an acceptable compromise in the designspace. We present a simple demonstrator of a flexible workflow framework forengineering design optimization using an e-Science tool. This paper provides a conciseintroduction to MDO, complemented by a summary of the related tools and techniquesdeveloped under the umbrella of the UK e-Science programme that we have explored insupport of the engineering process. The main contributions of this paper are: (i) adescription of the optimization workflow that has been developed in the TAVERNA

workbench, (ii) a demonstrator of a structural optimization process with a range of tooloptions using common benchmark problems, (iii) some reflections on the experience ofsoftware engineering meeting mechanical engineering, and (iv) an indicative discussionon the feasibility of a ‘plug-and-play’ engineering environment for analysis and design.

Keywords: multi-disciplinary design optimization; Web services; workflows;semantic web

Onand

*A

1. Introduction

(a ) The engineering perspective

Complex engineering design is realized through the combined efforts of a numberof specialist design teams with discipline-specific skills, tools and knowledge. Thecompetitive environment of the engineering industry demands an optimumdesign through continuous improvement in performance and economy of design.However, due to the interdependent nature of disciplines, achieving an optimaldesign can be difficult and slow as each specialist team often lacks a directunderstanding of the global consequences.

Multi-disciplinary design optimization (MDO) has been defined as ‘method-ology for the design of systems in which strong interaction between disciplinesmotivates designers to simultaneously manipulate variables in several disciplines’(Sobieszczanski-Sobieski & Haftka 1997). Research in MDO has received

Phil. Trans. R. Soc. A (2009) 367, 2741–2751

doi:10.1098/rsta.2009.0035

e contribution of 16 to a Theme Issue ‘Crossing boundaries: computational science, e-Scienceglobal e-Infrastructure II. Selected papers from the UK e-Science All Hands Meeting 2008’.

uthor for correspondence (jap@cs.bath.ac.uk).

2741 This journal is q 2009 The Royal Society

T. Crick et al.2742

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

increasing interest as it aims to reconcile potential conflicts by treating the designas a whole, taking account of the interdependency of design disciplines. Twocategories of methods have emerged: single- and multi-level formulations. Theformer typically uses a single optimizer which is applied to the entire MDOsystem.While this approach may be considered the more intuitive, the typical sizeand highly complex nature of the problems are usually unsuitable for a singleoptimizer. Conversely, the multi-level approach decomposes a design problem.Both analysis and optimization are carried out for each subsystem and theinteractions between subsystems are considered by a system-level optimizer todetermine the solution (Martins & Marriage 2007; Yi et al. 2008).

In the modern engineering environment, the multi-level formulation is oftenthe preferred approach. There are several reasons for this, but two key factors are:(i) the decomposition of a large problem into a number of subsystem optimizationsis naturally suited to parallel and distributed computing, and this can significantlyreduce computational costs to realistic levels; and (ii) as the structure ofengineering organizations typically reflects disciplines and each discipline teamworks largely independently, the multi-level approach is somewhat akin to thecurrent design industry, and hence more suited to industrial practice.

The multi-level formulation considers the highly coupled interdependency ofmulti-disciplinary design criteria at system level. Therefore, it follows that theoptimum solution can be sensitive to the system architecture and implemen-tation (Alexandrov & Kodiyalam 1998; Brown & Olds 2006). There have beenlimited studies that provide a generalized understanding of various methods andtheir performance, and the MDO implementations have been specific to designproblems and applications. As the design evolves and the tools are modified, itrequires a significant modification to the MDO framework and the suitability ofthe modified architecture for the given problem cannot be guaranteed.

Additional difficulties arise in the changing environment of engineeringindustry, where designs are increasingly carried out in various geographicallocations on heterogeneous platforms. This calls for a more flexible and easy-to-understand MDO system for integrating both legacy codes and proprietarysoftware with a range of model representations and fidelity (Giesing &Barthelemy 1998).

(b ) From e-Science to e-Engineering

A key aim of e-Science activity has been to ease accessibility to data andcomputational resources, wherever they might be, and however they might bedescribed. In technological terms this translates to workflow design andenactment on the one hand, and service description and discovery on theother. Active research initiatives on service discovery can be seen particularly inthe chemistry and bio-informatics communities, e.g. BIOMOBY at www.biomoby.orgoffers interoperability between biological data hosts and analytical services,BIOCATALOGUE at www.biocatalogue.org provides a curated catalogue of LifeScience Web Services. The MYGRID project at www.mygrid.org.uk hasdemonstrated how bio-informatics research can be assisted (Oinn et al. 2004)through the automated identification of Web services and a visual programmingenvironment—the TAVERNA toolkit (Oinn et al. 2006)—for the construction andexecution of workflows. Association of semantic information with services

Phil. Trans. R. Soc. A (2009)

2743Services for optimization

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

and its subsequent discovery is being enabled by tools such as the semantics-enabled extension of UDDI, GRIMOIRES (Fang et al. 2008), and the generalpurpose, extensible brokerage framework, KNOOGLE (Chapman et al. 2007)—bothOMII projects (see www.omii.ac.uk)—while Seekda at www.seekda.com offers atext-based searcher for Web services and the FETA (Lord et al. 2005) componentof TAVERNA is developing a semantic search interface. In previous work, we madesome steps towards the semantic description of mathematical services (Caprottiet al. 2004), which is particularly relevant for engineering, but in many fieldseffective domain-specific description is an emerging topic.

The objective of the work reported here is to investigate the applicability of theTAVERNA workbench to a classical engineering optimization problem. We chose touse TAVERNA because it has been used for a diverse range of domains includingmedicine, astronomy, social science and music, as well as bio-informatics. It isclear from the degree of up-take that the use of tools to support the authoring andenactment of workflows has been beneficial to many different communities by(i) enabling access at a distance to resources, and (ii) allowing the researcher tofocus on how to combine those resources. There are many parallels to be found inthe engineering design process: (i) workflows capture common, even standardizedprocesses, (ii) resources are not necessarily co-located, and (iii) much tedious effortis spent copying and transforming data output by one analysis as input for anotheranalysis. It is notable that commercial software packages are starting to exploitthe workflow concept in their interfaces but, by being essentially closed systems,have the effect of locking users into particular products.

By building a system based on workflow, we are able to delegate filemanipulation and house-keeping details to the enactment engine. Furthermore,from an engineering design perspective, it is now feasible to deploy many moreanalysis codes and with equal ease integrate them into workflows and visualizethe results. We now proceed to discuss the proof-of-concept demonstrator.

2. Services and workflows

(a ) Structural optimization algorithm

The design problem considered for the demonstrator study is a material distributionproblem in a continuum such as figure 1. The continuum domain is represented byUenclosed by a boundary G, with body forces f, the boundary traction t on Gt2G andsupport onGu2G. The design variable x represents the existence ofmaterial which iseither present or absent, x2{0,1}, as shown in figure 1.

The formulation of the optimization problem is to minimize the totalcompliance subject to the equilibrium and volume constraints. The implemen-tation typically employs the existence of finite elements as design variables.A common approach to this discrete problem is to relax the design variables to0!x%1, but to penalize the intermediate values by power-law (Bensøe 1995).The optimization problem can therefore be written as

min CðxÞZX

i

xpi uTi kui subject toKU ZF and V ðxÞ%V0;

where C denotes total compliance; ui and k are the elemental displacement andstiffness, respectively; p is the penalization power; K is the global stiffness

Phil. Trans. R. Soc. A (2009)

Figure 1. Generalized two-dimensional continuum.

T. Crick et al.2744

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

matrix; U is the global displacement vector; F is the force vector; V is the totalvolume of the design; and V0 is the specified volume limit for the solution.

The optimization problem can be characterized by several parameters:(i) structural, i.e. the material properties, boundary conditions, loading anddesign domain size, and (ii) optimization, i.e. the penalization power, volumefraction and convergence criteria.

Optimization begins by discretizing the design domain continuum using finiteelements and the design variables x are the continuous variation of the existenceof an element, sometimes referred to as artificial density of an element. A finite-element analysis is usually employed to compute the nodal displacements, whichin turn are used to determine the sensitivities required for optimization. Theoptimization uses the method of moving asymptotes (MMA) optimizer(Svanberg 1987), which works by updating the elemental artificial densities.This process is repeated until the convergence criterion is met. This process isformalized by algorithm 1.

(b ) Workflow construction

TAVERNA can discover and use both local programs and deployed Webservices as components in workflows. All the components reported here werepublished as Web services and are thus potentially re-usable by others.A TAVERNA workflow that implements algorithm 1 is shown in figure 2. Thealgorithm and the workflow evolved iteratively through a process of collaborativeauthoring, each functioning as a boundary object between the domains ofthe participants, and resulting in the identification of five major componentsfor the workflow: (i) design initialization, (ii) analysis of the current design,(iii) sensitivity analysis, (iv) optimization and design update, and (v) theconvergence test. For the analysis step, three finite-element analysis programswere available:

(i) Computational finite element (CFE), an ‘in-house’ legacy C code which wasdesigned specifically to undertake the analysis step of the optimizationprocess. This is deployed as a Web service using the GSOAP toolkit(van Engelen & Gallivan 2002) and demonstrates the capacity to publishC/CCC/Fortran codes as Web services.

Phil. Trans. R. Soc. A (2009)

Algorithm 1. The structural optimization process

input: a problem independent component PI

input: a problem dependent component PD

input: optimization process control parameters PO

output: an optimized problem dependent component

var: the value of PD at step i, denoted PDi

var: the value of PD at step iK1, denoted PDiK1

var: the value of PD at step iK2, denoted PDiK2

var: the strain energy of the model, denoted SE

1 PDi)Setup (PD,PO)

2 PDiK1)PDi

3 PDiK2)PDi

4 repeat

5 SE)Analyse (PI,PDi ,PO)

6 dSEdx ; d

2SEdx2

)Sensitivity-analysis (PO,SE,PDi)

7 PDiK2)PDiK1

8 PDiK1)PDi

9 PDi)Optimize (PDi, PDiK1, PDiK2, PO, PI,dSEdx ; d

2SEdx2

Þ10 until Coverged? (PDi ,PDiK1,PO)

11 return PDi

2745Services for optimization

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

(ii) A commercial package, ANSYS, as an example of proprietary softwarewith a license requirement (ANSYS 11.0SP1). To incorporate ANSYSinto theworkflowamacrowaswritten in theANSYSparametric developmentlanguage that executes the required analysis. Additionally, a pre-processingstep—implemented as a TAVERNA shim (see §4.4 of Oinn et al. (2006)), i.e. aservice whose purpose to carry out some minor operation to establishcompatibility between two other services—converts the input data fileinto the format required by ANSYS. The interface is generated bythe SOAPLAB2 (Senger et al. 2003, 2008) Web service deployment tool anddemonstrates the creation of services from command-line driven engineeringanalysis tools.

(iii) A Matlab script (Matlab r2003a). The script was specifically written toexecute the required analysis. This service is also deployed using SOAPLAB2and demonstrates the means to publish services built on widely usedengineering scripting software.

All the analysis programs accepted the same input files with the same formatand produced output files in a consistent format, making them completelyinterchangeable. Each analysis program was limited to a two-dimensional staticlinear elastic analysis of a rectangular domain of square elements of varyingdensity. The inputs and outputs for all workflow components are summarized intable 1. All data are in plain text file format except for the convergence testoutput, which is a binary number.

Phil. Trans. R. Soc. A (2009)

preprocess inputs

workflow inputs

workflow outputs

select analysis

FE_MATLAB_ws CFE_ws

SIMP_Sens_ws

SIMP_Opt_ws

SIMP_Conv_ws

converged

optimize

FE_ANSYS_ws

Initialization of problem variables. In this case theinitial value for each element density is set to thespecified volume limit as a fraction of the totaldesign space volume. This volume fraction isdefined by the optimization control variables file.

Sensitivity analysis.This computes thefirst- and second-orderderivatives forcompliance withrespect to elementdensity.

Density update.The optimizer updatesthe design variables todetermine theoptimal materialdistribution update.

Convergence test. Convergence is achieved when the changein the design variables between two subsequent iterations isless than a prescribed value or convergence criterion, asdefined by the optimization parameter file.

Figure 2. The structural optimization workflow in TAVERNA.

T. Crick et al.2746

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

3. Results

In this section, we use two popular structural benchmark problems and solvethem using the optimization workflow described in §2b. The two problems are ashort cantilever beam and a Messerschmidt–Bolkow–Blohm (MBB) aircraft floorbeam (Bensøe 1995). Each structure was discretized with square elements of unitarea. Both optimization problems were run using all available analysis programsand the optimization parameters were the same for both problems: pZ3;V0Z0.4; convergence criteriaZ0.01.

Figure 3a depicts the structural design environment for the popular cantileverbeam of aspect ratio 1.6, with one edge clamped and a central vertical loadapplied on the other side. The optimum solution was obtained after 42 iterationsas shown in figure 3b–e. This is typical of the solutions obtained in existingliterature, thus validating the optimization algorithm implemented as aworkflow. The second test case is the MBB beam, which is a simply supportedbeam of aspect ratio 6 with a central vertical load. The results obtained were ineach case the well-known optimum solution.

Phil. Trans. R. Soc. A (2009)

64

0 0.2 0.4 0.6element density

0.8 1.0

40

20

(b)

(a)

(c)

(d) (e)

Figure 3. Initial design domain of short cantilever beam (a) and its optimization at iteration 10 (b),iteration 20 (c), iteration 30 (d ) and converged solution at iteration 42 (e).

Table 1. Workflow component inputs and output.

component input output

design variableinitialization

structural parameters initial element densitiesoptimization parameters

FE analysis structural parameters element complianceoptimization parameterselement densities

sensitivity analysis element densities first-order derivativeelement compliance second-order derivativeoptimization parameters

optimizer (MMA) structural parameters current MMA variablesupdated element densitieselement densities

densities from previous two iterationselement compliancefirst- and second-order derivativesoptimization parametersMMA variables from previous iteration

convergence test updated element densities convergence result(1, converged; 0, continue)previous element densities

optimization parameters

2747Services for optimization

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

4. Discussion and related work

We have built and validated—using some well-known benchmark structures—aproof-of-concept workflow demonstrator that addresses or facilitates the pointsraised in Giesing & Barthelemy (1998)—specifically flexibility, provenance,multiple (consistent) models, distribution and resource brokerage—by

Phil. Trans. R. Soc. A (2009)

T. Crick et al.2748

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

redeploying tools conceived for e-Science, to enable more flexible and intuitiveMDO processes that allow for: (i) the continuing use of favoured legacy code andprototype scripts as well as commercial software in a common framework,and (ii) the necessary variation in model representation and precision. It isinteresting to observe that 10 years on from the above paper, Moore et al. (2008)make similar observations and call for the development of an open-sourceframework for MDO.

A notable difference between our work and several examples in the currentliterature on workflow in MDO is our use of an off-the-shelf framework, allowingfor concentration on developing programs and wrappers, instead of developingthe framework itself, as is the focus of Wang et al. (2003), Hao et al. (2004), Kimet al. (2006) and Lahr & Bletzinger (2007). Working with an existing communityof (bio-informatics) users clearly underpins claims for usability and evenlongevity, as well as the capacity for more rapid growth, given an attractiveset of services and workflows, in new domains.

Our planned next steps include evaluation of alternative workflow descriptionlanguages and engines, the development of semantic descriptions ofthe components, and undertaking a wider range of case studies.

(a ) Technical issues

Commercial software packages, such as ENGINEOUS, NOESIS and PHOENIX offersimilar facilities for workflow creation, visualization and management althoughin each case the user is effectively limited to using the components provided withthe package. The real benefit of using a workflow engine such as TAVERNA is therelative ease of using Web services, potentially leading to: (i) management anduse of proprietary analysis tools by contractors, integrating them into the MDOworkflow, minimizing risk of loss of intellectual property, reducing design timeand improving quality control over (subcontracted) components, (ii) sharing ‘in-house’ codes with external users instead having to act as a software provider andmaintainer, as well as allowing straightforward version control, and (iii) capacityfor the automatic capture of component provenance information and potentialfor full life-cycle knowledge management.

We foresee a particular benefit arising from the use of Web services forthe engineering research community. There has been significant effort into thedevelopment of alternative MDO methods, but few attempts at comprehensivecomparison between methods. One reason is the amount of work required todevelop and implement the sizeable range of available methods on a commonplatform, as most MDO methods for practical problems tend to be designed onlyfor specific applications (Alexandrov & Kodiyalam 1998; Martins & Marriage2007). A possible solution is the concurrent publication of the article and itsimplementation as a Web service—an approach along these lines has beenimplemented by the London Mathematical Society’s Journal of Computationand Mathematics for several years (e.g. http://www.lms.ac.uk/jcm/11/lms2007-056/). Some recent publications in the MDO literature have reached similarconclusions to ourselves about the desirability of workflow approaches (e.g. Shiet al. 2005; Bereneds et al. 2008;Moore et al. 2008), butwe observe that the publisheddescriptions appear to use bespoke software rather than workflow tools.

Phil. Trans. R. Soc. A (2009)

2749Services for optimization

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

A notable drawback of accessing commercial analysis tools as services is that itbypasses much of their value-added functionality, such as post-processing andvisualization. This factor is also identified by Oinn et al. (2006) in their extensiveanalysis of TAVERNA for life sciences applications and holds true for theengineering sector as well.

(b ) Reflections on process

Bringing together engineering codes and workflow software has been a learningprocess for both parties. Apart from the initial challenge of appreciating andunderstanding each other’s vocabulary, there have been deeper-rooted issuesaround the advantages (or otherwise) of bringing in another layer of softwaretechnology and the adaptation of legacy code for the new environment.

There have been several situations during this work that might be thought ofas ‘cultural’ issues, but, not being sociologists, these observations should be seenas purely anecdotal and without significant foundation. One aspect thatsurprised the computer scientists was the apparently low importance given tomaking software re-usable: modifications were proposed to make componentswork in the particular context of use, with relatively little consideration of newenvironments. In other words: aspects of engineering practice are now common-place in computer science (software engineering), but these principles have notnecessarily made their way back to software development in engineering.

Much of the literature on cultural issues in engineering has addressed ethnicityor organizational factors rather than domain discipline. However, Bond & Ricci(1992) explored the ways in which different disciplines work together in thecontext of aircraft design. Interestingly, the conclusions they reached resonateequally well with our experience of computer scientists working with mechanicalengineers. We paraphrase and summarize their conclusions here and commentupon them: (i) an inter-disciplinary project proceeds through the cooperation ofspecialists, (ii) each specialist has its own model (or models) of the design forvarious purposes—we used a shared model (the algorithm) to communicate withone another, (iii) specialists have limited ability to understand each other’smodels—we now each have a limited understanding of each other’s domain,(iv) design proceeds by successive refinement of the models, which arecoordinated and updated together—indeed, we prototyped the workflow andthe algorithm and revised and updated them together, and (v) the designdecisions, which are acts of commitment and model refinement, are negotiatedby the specialists among themselves.

Finally, over and above the practical issues identified in §4a, there is aqualitative aspect that is enabled by the adoption of Web services and workflow.Much of design optimization has been based on the parametric representationdefined at the initial design stage. This restricts the solution space and preventsoptimization methods from exploring all potential solutions. As the designmatures and the scale increases to higher levels of detail, more refined analysis andoptimization methods are required and results from the previous stages and legacysystems do not always translate well, both requiring many hours of manualprocess and potential loss of information. Previous design decisions in onediscipline may be challenged as the problem is better understood, but theconsequences of change for other disciplines are less understood, thus it is simpler

Phil. Trans. R. Soc. A (2009)

T. Crick et al.2750

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

to remain at the local optimum. Furthermore, the selection of codes and numericaltools leading to local attractors may as much be a function of economic and socialfactors as technical suitability. However, the accessibility of a wide range of codes,capture of provenance information and the ease of trying out alternative designavenues would ease the exploration of multiple design spaces, and offers the chanceto make a notable step forward in MDO.

The authors would like to thank: (i) Prof. K. Svanberg for the implementation of the MMAalgorithm, (ii) the Numerical Analysis Group at the Rutherford Appleton Laboratory for theFortran 77 HSL package MA57, and (iii) Ian Dunlop and Stian Soiland-Reyes of the MYGRID teamfor their help with TAVERNA via the TAVERNA-users mailing list. This work was partially funded bythe EPSRC Bridging the Gaps programme, through the project ‘Creativity in Systems Design: theBath Interactive Ideas Factory’ (EP/E018122/1).

References

Alexandrov, N. M. & Kodiyalam, S. 1998 Initial results of an MDO evaluation survey. In Proc. 7thAIAA/USAF/NASA/ISSMO Symp. on Multidisciplinary Analysis and Optimization.Reston, VA: American Institute of Aeronautics and Astronautics.

Bensøe, M. P. 1995 Optimization of structural topology, shape and material. Berlin, Germany:Springer.

Bereneds, J. P. T. J., van Tooren, M. J. L. & Schut, E. J. 2008 Design and implementation of a newgeneration multi-agent task environment framework. In Proc. 49th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference. Reston, VA: AmericanInstitute of Aeronautics and Astronautics.

Bond, A. H. & Ricci, R. J. 1992 Cooperation in aircraft design. Res. Eng. Des. 40, 115–130. (doi:10.1007/BF01580149)

Brown, N. F. & Olds, J. R. 2006 Evaluation of multi-disciplinary optimization techniques appliedto a reusable launch vehicle. J. Spacecraft Rockets 43, 1289–1300. (doi:10.2514/1.16577)

Caprotti, O., Dewar, M., Davenport, J. H. & Padget, J. 2004 Mathematics on the (Semantic) net.In Proc. European Symposium on the Semantic Web, vol. 3053 (eds C. Bussler, J. Davies,D. Fensel & R. Studer). Lecture Notes in Computer Science, pp. 213–224. Berlin, Germany:Springer.

Chapman, N., Ludwig, S., Naylor, W., Padget, J. & Rana, O. 2007 Matchmaking support fordynamic workflow composition. In Proc. 3rd IEEE Int. Conf. on eScience and Grid Computing,IEEE, December 2007, pp. 371–378. (doi:10.1109/E-SCIENCE.2007.48)

Fang, W., Miles, S. & Moreau, L. 2008 Performance analysis of a semantics-enabled serviceregistry. Concurr. Comput. Pract. Exp. 200, 207–223. (doi:10.1002/cpe.1204)

Giesing, J. P. & Barthelemy, J. M. 1998 A summary of industry MDO applications and needs. InProc. 7th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis andOptimization. Reston, VA: American Institute of Aeronautics and Astronautics.

Hao, Q., Shen, W., Park, S.-W., Lee, J.-K., Zhang, Z. & Shin, B.-C. 2004 An agent-basede-engineering services framework for engineering design and optimization. In IEA/AIE’2004:Proc. 17th Int. Conf. on Innovations in Applied Artificial Intelligence, pp. 1016–1022. Berlin,Germany: Springer.

Kim, H., Ragon, S., Mullins, J. & Sobieszczanski-Sobieski, J. 2006 A web-based collaborativeenvironment for bi-level integrated system synthesis (bliss). In Proc. 47th AIAA/ASME/AS-CE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, 1–4 May 2006,pp. 1–10. Reston, VA: American Institute of Aeronautics and Astronautics.

Lahr, A. & Bletzinger, K.-U. 2007 Agent-based distributed and interdisciplinary collaboration andanalysis in heterogeneous engineering environments. In Proc. 7th World Congress on Structuraland Multidisciplinary Optimization, Seoul, Korea, 21–25 May 2007, pp. 637–646.

Phil. Trans. R. Soc. A (2009)

2751Services for optimization

on 1 June 2009rsta.royalsocietypublishing.orgDownloaded from

Lord, P. W., Alper, P., Wroe, C. & Goble, C. A. 2005 Feta: a light-weight architecture foruser oriented semantic service discovery. In Proc. ESWC, vol. 3532 (eds A. Gomez-Perez &J. Euzenat). Lecture Notes in Computer Science, pp. 17–31. Berlin, Germany: Springer.

Martins, J. R. R. A. & Marriage, C. 2007 An object-oriented framework for multi-disciplinarydesign optimization. In Proc. 48th AIAA/ASME/ASCE/AHS/ASC Structures, StructuralDynamics, and Materials Conference, Honolulu, Hawaii. Reston, VA: American Institute ofAeronautics and Astronautics.

Moore, K. T., Naylor, B. A. & Gray, J. S. 2008 The development of an open source framework formultidisciplinary analysis and optimization. In Proc. 12th AIAA/ISSMO MultidisciplinaryAnalysis and Optimization Conference, Victoria, Canada, 10–12 September 2008, pp. 1–13.

Oinn, T. et al. 2004 TAVERNA: a tool for the composition and enactment of bioinformaticsworkflows. Bioinformatics 200, 3045–3054. (doi:10.1093/bioinformatics/bth361)

Oinn, T. et al. 2006 TAVERNA: lessons in creating a workflow environment for the life sciences.Concurr. Comput. Pract. Exp. 18, 1067–1100. (doi:10.1002/cpe.993)

Senger, M., Rice, P. & Oinn, T. 2003 Soaplab—a unified sesame door to analysis tools. In Proc. UKe-Science All Hands Meeting (AHM2003 ), pp. 509–513. See http://www.nesc.ac.uk/events/ahm2003/AHMCD/pdf/115.pdf.

Senger, M., Rice, P., Bleasby, A., Oinn, T. & Uludag, M. 2008 Soaplab2: more reliable sesamedoor to bioinformatics programs. In Proc. 9th Bioinformatics Open Source Conference(BOSC 2008 ). See http://www.open-bio.org/w/images/3/3c/Rice_Soaplab2_anstract.pdf.

Shi, D., Yin, J., Chen, G. & Dong, J. 2005 A workflow-based distributed multi-disciplinary designoptimization framework. In Proc. 9th Int. Conf. on Computer Supported Cooperative Work inDesign, vol. 1, pp. 497–502. Piscataway, NJ: IEEE.

Sobieszczanski-Sobieski, J. & Haftka, R. T. 1997 Multi-disciplinary aerospace design optimization:survey of recent developments. Struct. Optim. 14, 1–23. (doi:10.1007/BF01197554)

Svanberg, K. 1987 The method of moving asymptotes—a new method for structural optimization.Int. J. Numer. Methods Eng. 24, 359–373. (doi:10.1002/nme.1620240207)

van Engelen, R. & Gallivan, K. 2002 The GSOAP toolkit for web services and peer-to-peercomputing networks. In Proc. 2nd IEEE Int. Symp. on Cluster Computing and the Grid(CCGrid2002 ), pp. 128–135. Washington, DC: IEEE Computer Society. See http://www.cs.fsu.edu/engelen/ccgrid.pdf.

Wang, Y. D., Shen, W. M. & Ghenniwa, H. 2003 Webblow: a web/agent-based multidisciplinarydesign optimization environment. Comput. Ind. 52, 17–28. (doi:10.1016/S0166-3615(03)00066-6)

Yi, S. I., Shin, J. K. & Park, G. J. 2008 Comparison of MDO methods with mathematicalexamples. Struct. Multidisciplinary Optim. 35, 391–402. (doi:10.1007/s00158-007-0150-2)

Phil. Trans. R. Soc. A (2009)