PUBLISHABLE REPORT

 

 

Contract no: BRPR CT96 0282

Project no: BE96 – 3046

Title: (CE)2 : Computer Experiments for Concurrent Engineering

Project coordinator: CRF (IT)

Partners: SNECMA ( FR), INTRASOFT (GR), UNIVERSITY OF WARWICK (GB),POLITECNICO DI TORINO (IT), CNRS (FR).

 

Reference period: from the 1st of November 1996 to the 31st of October 1999

Starting date: the 1st of November 1996 Duration: 36 months

Date of issue of this report: 31st of Genuary 2000

 

Project funded by the European Community

Under the Industrial and Materials Technologies

Programme (Brite-EuRam III)

 

 

PREFACE

This report has been prepared in html format.

 

 

 

MOVIE

 

Computer Experiments for Concurrent Engineering

The reduction of time to market is a declared objective of the European industry. Concurrent Engineering philosophy is a way to reduce the duration of all the product development process. Several aspects of the product, such as reliability, environmental impact, performance, safety are taken into account at the same time. Here numerical simulation is playing a major role. It now has at least equal status to experimentation on prototypes in terms of cost and accuracy.

But numerical simulation is not sufficient to solve multidisciplinary and multiobjective problems. The couple "simulation+multiobjective optimisation" ought to provide the solutions but is not yet readily available. This is because simulators usually requires high computational time to compute one single run (as much as a day) and, at the same time, multiobjective algorithms need many iterations to converge. Only an integrated approach based on a multiobjective optimisation of product design can lead to useful and usable solutions.

In 1996 the (CE)2 "Computer Experiments for Concurrent Engineering" project started. The main objective of the project is the development of a tool and a methodology to achieve multiobjective optimisation through the use of accelerated numerical simulation.

This 3-year project has been partially funded by the EC under the Brite Euram programme. It is co-ordinated by CRF. The partners are SNECMA, INTRASOFT, University of WARWICK, CNRS and POLITECNICO of Turin.

The key idea of the project is to replace simulators with emulators. These are simple but highly adaptive mathematical functions that approximate simulators but run thousands of times faster. Emulators are identified using the design and analysis of computer experiments and, because of their speed, allow the use of the most advanced optimisation algorithms (including global and multiobjective).

The main modules of the tool are:

A critical validation role has been played by successful and in some cases highly complex case studies.

 

CRF worked to optimise the front part of the car body of the FIAT Multipla. Crash behaviour, acoustic and static behaviour were taken into account through the implementation of ad-hoc simulators. The goal of the study was to design the front member so that the internal energy absorbed during a crash and the stiffness were maximised while the global mass and the noise were minimised.

The SNECMA case study covers the optimization of a rotating disc and its air cooling within an aircraft engine. Optimal efficiency of the engine requires low disc mass, low mass cooling flow and low clearance between the shroud and the disc. The disc cannot support more than a critical stress and the cooling flow temperature must not increase too much. These objectives and constraints are incompatible. Thus the goal of this case study is to fit the cooling flow and the shape of the disc that maximize the efficiency of the engine during a specific flight cycle while respecting the geometric and physical constraints.

In both cases the software was able to find optimal solutions close to those found with the standard process but with a reduction in the computational time of almost 90%.

Engine mapping to reduce emission and fuel consumption of a truck engine was the third case study and was carried out by IVECO in co-operation with CRF. This project reflects an early, and in the event wise, decision to include at least one physical experiment case study to widen the market for (CE)2. This case study was conducted on real experiments and the results found were very encouraging, pointing out the power of the approach.

Using the (CE)2 methodology the computational time requested for numerical simulation based optimisation can be reduced up to 10000 times with a considerable reduction of the time-to-market.

The trio of numerical simulation experiments, emulation and optimisation provide a platform to conduct concurrent engineering philosophy.

Statistics + Optimisation = (CE)2 = Concurrent Engineering

 

 

DISSEMINATION

 

 

Several conferences were attended.

Here is the list of the papers that have been published.

H P Wynn and R A Bates, Design Reuse: Engineering Design Conference '98, Brunel University, Professional Engineering Publishing, Emulator technology in complex engineering environments, 67-76, 1998

R A Bates and R Fontana and L Pronzato and H P Wynn, I Parmee, Adaptive Computing in Design and Manufacture, Springer, Multi-domain optimisation using computer experiments for concurrent engineering, 354-364,1998

R A Bates B Giglio E M Riccomagno and H P Wynn, Compstat 98, Grobner basis methods in polynomial modelling, 1998

R A Bates and L Pronzato and H P Wynn, Optimisation'98 - University of Coimbra, Portugal, Non-parametric modelling of Pareto surfaces for Multi-objective optimisation, 1998

R Bates, R Fontana, C Randazzo, E Vaccarino, H P Wynn, "Empirical Modelling of Diesel Engine Performance for Robust Engineering Design", "Statistics for Engine Optimisation", Professional Engineering Publishing Limited, Bury St Edmunds, Suffolk IP32 6BW, UK, 1998

R A Bates and R Fontana, Optimisation based on Computer Experiments in the Automotive Field, Proceedings of SIS XXXIX, 193-200, Sorrento, Aprile 1998

R A Bates and R Fontana and H P Wynn, Multi-Domain Optimisation in Engineering Design, Proceedings of ASMO99, Bradford, UK, 1999

R A Bates and H P Wynn, I Parmee, Proceedings of Adaptive Computing in Design and Manufacture, ACDM2000, Plymouth, UK, Adaptive radial basis function emulators for robust design, submitted

T Bosio and R Fontana, Ottimizzazione a calcolo dell'ossatura anteriore di un'autovettura, Giornata di Studio "Valutazione della Qualità e Customer Satisfaction: il ruolo della statistica", Bologna, 24 Sep 1999

R Bates and H P Wynn and R Fontana and L Pronzato, 6th International Conference "The new role of experimentation in the modern automotive product development process", "Computer Experiments for Concurrent Engineering", Florence, Nov 1999.