247
Chapter 18
Industrial
Applications Journey
to Supercomputing
at Renault
Yves Tourbier and Marc Pariente
18.1 INTRODUCTION
France’s Renault Group has been an automotive industry pioneer in the
usage of high-performance computing (HPC) since the 1980s. A yearly
process is utilized to measure the needs and schedule the evolution of
internal HPC capacity. Optimization studies, especially those on the
upstream projects, are used to size the breaks on HPC capacity because
they are representative of anticipated needs of projects in 5 years.
CONTENTS
18.1 Introduction 247
18.2 Case Study 248
18.2.1 Project Background 248
18.2.2 Critical Role of the HPC Community 248
18.2.3 Business Opportunity When Using HPC 249
18.2.4 Data Processing and Visualization 249
18.2.5 Advanced Manufacturing, Big Data, and the
Industrial Internet 250
18.2.6 Lessons Learned 250
18.3 Whats Next and Conclusion 250
248 Industrial Applications of High-Performance Computing
Concerning the structure of the vehicle body, innovations are
systematically designed and tested with the help of numerical simulation
augmented by validation with a physical prototype. at gives a lot of opti-
mization studies with data-heavy numerical models such as crash, acoustics,
and durability—a great number of design parameters (>100 in our study)
many real-world industrial constraints like assembly process or carryover
maximization, and very complex objectives because all digital design prob-
lems requires compromise between dozens of contradictory specications.
is case study is a review of car body optimization and the trade-os asso-
ciated with mass and specications with combinatorial constraints.
Renault has no intention of directly developing simulation tools, using
commercial tools and implementing collaborative projects with soware
vendors and French laboratories to improve the performance of these
tools. Renault uses national or European (through PRACE) HPC facili-
ties in collaborative research projects, while numerical simulations for
the vehicle or powertrain projects use in-house HPC only. As a result of
these collaborations, Renault funds new PhDs every year in the eld of
numerical simulation. Renault also participates in an IRT (Institut de
Recherche Technologique) SYSTEM X project dealing with the use of
Model Reduction and Multiphysics Optimization.
18.2 CASE STUDY
18.2.1 Project Background
is is a world rst-ever in this domain: neither Renault, simulation ven-
dor ESI, nor their competitors, had ever used a head-up crash numerical
simulation of this size in a combinatorial optimization study with such
levels of parameters.
18.2.2 Critical Role of the HPC Community
Renault has been one of the rst automotive companies to use HPC for
crash or CFD simulations; this allowed Renault to stay in the forefront
of the European EuroNCAP rules for safety certications. By developing
more and more realistic crash models, including ne-resolution descrip-
tion of all the parts of a car and detailed crash test dummy modeling,
Renault has been able to embed a wide range of soware for developing
appropriate models and tools to reduce the number of physical tests and, by
consequence, the cost of development and the overall time to market. is
role of Renault and other car suppliers has been mandatory for developing

Get Industrial Applications of High-Performance Computing now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.