© Geodise Project 2003 Grid Enabled Optimisation and Design Search for Engineering (G EODISE ) Expo May 12 th Southampton Prof Simon Cox Southampton University
Thanks to … Nicola Reader for organisation … everyone for coming!
Grid Enabled Optimisation and Design Search for Engineering (G EODISE ) Southampton, Oxford and Manchester Simon Cox- Technical Director Southampton e-Science Centre. Grid/ W3C Technologies and High Performance Computing Andy Keane- Director of Rolls Royce/ BAE Systems University Technology Partnership in Design Search and Optimisation Mike Giles- Director of Rolls Royce University Technology Centre for Computational Fluid Dynamics Carole Goble- Ontologies and DARPA Agent Markup Language (DAML) / Ontology Inference Language (OIL) Nigel Shadbolt- Director of Advanced Knowledge Technologies (AKT) IRC BAE Systems- Engineering Rolls-Royce- Engineering Fluent- Computational Fluid Dynamics Microsoft- Software/ Web Services Intel- Hardware Compusys- Systems Integration Epistemics- Knowledge Technologies Condor- Grid Middleware
The GEODISE Team... Richard Boardman Sergio Campobasso Liming Chen Mike Chrystall Trevor Cooper-Chadwick Simon Cox Mihai Duta Clive Emberey Hakki Eres Matt Fairman Mike Giles Carole Goble Ian Hartney Tracey Hunt Zhuoan Jiao Andy Keane Marc Molinari Graeme Pound Colin Puleston Nicola Reader Angus Roberts Mark Scott Nigel Shadbolt Wenbin Song Paul Smart Barry Tao Jasmin Wason Fenglian Xu Gang “Luke” Xue
Expo Objectives Demonstrate 18 month deliverables Technology talks by RAs Demos & Posters ‘deskside’ over lunch Talks by industrial partners Future plans
Design
Modern engineering firms are global and distributed “Not just a problem of using HPC” CAD and analysis tools, user interfaces, PSEs, and Visualization Optimisation methods Data archives (e.g. design/ system usage) Knowledge repositories & knowledge capture and reuse tools. Management of distributed compute and data resources How to … ? … improve design environments … cope with legacy code / systems … integrate large-scale systems in a flexible way … produce optimized designs … archive and re-use design history … capture and re-use knowledge Design Challenges
Base GeometrySecondary Kinetic Energy Gas Turbine Engine: Initial Design Collaboration with Rolls-Royce 23/7/2001
RSM Construct RSM Evaluate Search Using RSM Best Design Adequate ? RSM Tuning Build Data-Base CFD DoE Initial Geometry CFD … … … … Cluster Parallel Analysis Design of Experiment & Response Surface Modelling 23/7/2001
Optimised Design GeometrySecondary Kinetic Energy 23/7/2001
Distributed Systems 2003 Network IP HTTP (HTML) Compute/ Data Moore’s Law Grid Services Drivers = Building Blocks + Protocols Software (HTML) XML Web Services (Proprietary, Open, Shared)
The Grid Problem “Flexible and secure sharing of resources among dynamic collections of individuals within and across organisations” Resources = assets, capabilities, and knowledge v Capabilities (e.g. application codes, analysis tools) v Compute Grids (PC cycles, commodity clusters, HPC) v Data Grids v Experimental Instruments v Knowledge Services v Virtual Organisations v Utility Services Grid middleware mediates between these resources
Geodise will provide grid-based seamless access to an intelligent knowledge repository, a state-of-the-art collection of optimisation and search tools, industrial strength analysis codes, and distributed computing & data resources G EODISE
Technologies (i) Grid Middleware (To coordinate and authenticate use of components of Geodise) Globus (and GGF grid-computing protocols) v Security Infrastructure (GSI) v Resource Allocation Mechanism (GRAM) v Resource Information System (GRIS) v Index Information Service (GIIS) v Grid-FTP v Metadirectory service (MDS 2.0+) coupled to LDAP server Condor (distributed high performance throughput system) v Condor-G allows us to handle dispatching jobs to our Globus system v Active collaboration from with the Condor development team at University of Wisconsin (Miron Livny) 23/7/2001
(ii) Data & Open W3C Standards (To access and interchange data) SRB (Storage Resource Broker) XML and XML Schema v Representing data in a portable format WSDL (Web Service Description Language) UDDI (Universal Description, Discovery and Integration) v Publish and discover information about web services 23/7/2001
(iii) Ontologies & Semantic Web (conceptualisation of a community’s knowledge of a domain) DAML - OIL (DARPA Agent Markup Language/ Ontology Inference Language) Genetics / v Virtual Enterprises v Product Specifications v Medicine v Encyclopaedic Knowledge 23/7/2001
(iv) Knowledge Technologies 23/7/2001
What we said… 28/06/2001 “Consider the CFD based design optimisation of a typical aero-engine or wing component or system. For a single loop of the design process it is necessary to (1) specify the geometry in a parametric form which defines the permitted operations and constraints for the optimisation process (this goes beyond the STEP/ IGES interchange standards), (2) decide which code to use for the analysis, (3) generate a mesh for the problem (though this may be provided by the analysis code), (4) decide the optimisation schedule, (5) execute the optimisation run coupled to the analysis code, and finally (6) monitor and steer the search as it takes place, possibly stopping it mid run to modify or rework the design process. Such loops are typically passed through several times. In our Grid environment these operations are large-scale and physically distributed computational steps: this will stretch the computation dimension in our Grid and will be delivered out first by a wizard-based Grid demonstrator: “Geodise-W” after 18 months.” “Whilst this is being developed we will be working on the knowledge-based demonstrator: “Geodise-K”, which we will deliver at 36 months. Here we seek to enhance each of the components of our system by using databases, ontologies and knowledge capture tools to provide intelligent guidance and assistance to the engineer using our system. We will develop, refine and deploy knowledge bases for CAD, commercial CFD code (Fluent), user supplied/ source-available CFD code (Oxford’s Hydra), optimisation and computation services. The knowledge bases will be physically distributed: integrating these large-scale distributed data sources will stretch the data interchange dimension in our Grid beyond that already required to execute and visualise our results. A conceptual architecture of our system is shown below.”
Geodise-W “18 month” 23/7/2001
© Geodise Project mth deliverable Geodise “Build complex things from lots of simple things”
Geodise Workflow/ Integration Requirements Flexibility v Customise the workflow and its components initially v Compose a work flow via drag & drop activity node component into an editor panel v Link to knowledge and other services Monitoring v Interact with the workflow or its components during simulation v Job status v Resource usage Maintainable v Modify & re-use the workflow either in a GUI or in a human readable file Usability v Easy to use by engineering users
Integration & Scripting Building Blocks Knowledge Services Matlab (or Jython) Java / C# Web Service Grid Service Java / C#/.NET.EXE/ Fortran/ Matlab Code Intelligent Support Interface Geodise Architecture
CFD-based shape optimisation using Geodise toolkits Nacelle Optimisation Problem – problem definition The aim is to understand the effect of various geometry parameters on the aerodynamic performance of engine nacelle, there is no attempt at this stage to calculate the radiated noise from fan, it is simply assumed that the bigger the scarf angle, more reduction in noise will be achieved. Two parameters were first chosen: scarf angle and axial offset Performance is measured using Total Pressure Recovery Conventional Inlet 0 12 Total Pressure Recovery (TPR) = Negative Scarf Inlet
Parallel Grid-enabled evaluations of multiple design jobs Within the Matlab hosting environment: Define the problem; Generate a proxy using user’s credentials; Retrieve the CAD definition file from repository; Retrieve the Gambit Journal file from repository; Retrieve the Fluent Journal file from repository; Submit CAD-Gambit-Fluent jobs sequentially or in parallel; (gd_cfdone, gd_cfdanalysis) a)Submit ProEngineer jobs to Windows Condor Pool via Webservice interface; (grid_submit, grid_status) b)Submit Gambit jobs to Grid-enabled Computing Servers; (gd_jobsubmit, gd_job_status) c)Submit Fluent jobs to Grid-enabled Computing Servers; 7.Postprocess results and archiving data files. (gd_archive, gd_datagroupadd) CFD-based shape optimisation using Geodise toolkits
Scarf angle Axial offset DoE using OPTIONS CFD-based shape optimisation using Geodise toolkits Optimisation using Design of Experiment/ Response Surface Modelling Problem definition Design of Experiment Response surface modelling Optimisation on Response surface Validation
1.Generate a proxy using user’s credentials; 2.Load in DoE data; 3.Create the RSM model; 4.Genetic Algorithms search on the RSM; 5.A further gradient-based search on GA result CFD-based shape optimisation using Geodise toolkits Optimisation using DoE/RSM models and two-stage approach
© Geodise Project 2003 Geodise Demo 1
Arcadia-Options Demo Matlab Geodise file archive Globus server gd_archive.m gd_objsubmit.m gd_jobpoll.m gd_objvalue.m Matlab optionsmatlab.dll projectstruct.xml arcadiaobjfun.m x5
© Geodise Project 2003 Geodise Demo 2
Deliverables Summary Application v CFD (Adjoint & Hydra, Fluent & RSF) v CAD (ProE) Optimisation v Options toolkit Computation/ Middleware v Compute Toolkit u Globus & Condor u Using UK Level 2 Grid u SMS v Workflow in Matlab Database v XML Toolkit v Database Toolkit u Archive for Files u Archive for Metadata u Query & Retrieve Knowledge v Acquisition v Ontology Services v Workflow Construction u Advice Services u Ontology driven editing “Build complex things from lots of simple things”
The future of design optimisation Design Optimisation needs integrated services Design improvements driven by CAD tools coupled to advanced analysis codes (CFD, FEA, CEM etc.) On demand heterogeneous distributed computing and data spread across companies and time zones. Optimization “for the masses” alongside manual search as part of a problem solving environment. Knowledge based tools for advice and control of process as well as product. Geodise will provide grid-based seamless access to an intelligent knowledge repository, a state-of-the-art collection of optimisation and search tools, industrial strength analysis codes, and distributed computing and data resources