Files

36 : Computational mechanics for engineers

Incontournable, la mécanique numérique s’insère aujourd’hui dans l’ensemble de la chaîne de conception rapide des produits fabriqués par l’industrie. S’appuyant sur les outils de modélisation géométrique et de visualisation, et intégrant les outils de simulation et d’optimisation, elle réduit les délais de conception, limite les erreurs et s’insère dans l’esprit du développement durable en aidant à concevoir des produits de plus en plus respectueux de l’environnement.

36 : Computational mechanics for engineers

Between democracy of uses and extreme demands in industry

Computational mechanics, at the base of new products product perspectives or training tools, is now widely used in industrial sectors. Nevertheless, for uses likes these, the underlying challenges and the results expected do not necessarily imply use of the same modelling tools, nor the same skills for those who implement the tools.

From a simple ball-point pen to the various latest Airbus A380 tat fly the world over, industrial design always begins with a modelling phase of the future product, on a CPU display screen. Industrialists have a means here not only to conceive and shape new artefacts - thereby limiting the need for test rig prototyping - more rapidly and at lower costs to meet an increasingly demanding specification as best as possible.

The object is drawn, it mechanical properties modelled with its environment and use constraints. From that point on, "it becomes possible to test large numbers of possibilities so to optimize the design of the object with respect to the given specification and to the usage envisaged", explains Professor Francisco Chinesta, Ecole Centrale de Nantes, a specialist in computational mechanics.

Different expectations, depending on the objectives

For engineers, the difficulty in the exercise is to identify the level of modelling best adapted and with an objective to keeping design time within reasonable bounds. "Necessarily, we must adapt the model to the objectives", underscores Francisco Chinesta, adding that the constraints of reliability, standards and risks are not of the same order for the ball-pin pen as for the double-decker wide-body aircraft. The difficulties meet to implement models, to think through and correctly plan for optimization and the computer time needed for the calculations ... all depend on choices and decisions made at the very start.

Today, certain models are so complex that even powerful computers take months to calculate and come up with a result. "Whatever the outcome ", adds Francisco Chinesta, "optimization tasks may lead to a result that does not eliminate all the risks". The work load, as he sees it, must be proportionate to the complexity of then system to be modelled, to the challenges and to the clients' expectations. For example, "an error in a weather forecast (say, less than a week ahead), is still something that we find acceptable, because we all know that the system is highly unpredictable", explains our research scientist.

Challenges and expectations are very different, whether it is an aircraft or model to train surgeons, for specific operations. In the latter example, we are not modelling reality but we do give the surgeons a 'hands-on' feeling that is as close as possible to real life scalpel work ... The perception here accepts a relatively large margin of inaccuracy, and so it is not really necessary to design a very accurate model for this case.

Increasingly accessible modelling tools

Whilst we can observe the application of more and more stringent constraints, in terms of safety factors in specifications, the trend now is to democratize the tools and workshop or design room equipment. Using well tested and certified models for standard optimization protocols needs less and less skilled operatives to work with the tools. "Today, some modelling is carried out by specialized technicians and there is no need to have support from qualified engineers", explains Frédéric Mercier, a research engineer who works with the Renault Automobile Group and is an expert in computational mechanics.

If as Francisco Chinesta imagines, the most complex problems still require the attention of highly qualified experts, lots of small scale applications will appear on the market, relatively easy to use and possibly even downloadable to smartphones®. The objective here is to make modelling a commonplace concept, with rapid and easy operational modes, that need less and less means to be implemented. For the qualified engineers, the aims should be to learn as early as possible to work on concrete problems found in industrial sectors.

The demands for better quality, less -or zero) risks, as well as new standards applicable , are forcing the actors involved to make the demand and then challenges more explicit and then to design an adapted model to the case to hand. It then becomes the responsibility of the research engineers and lecturers to ensure that there is a close connection with the realities of the engineers' professional world and, consequently, to adapt the courses accordingly, basing the exposes on industrial reality.

Perhaps this increased awareness where industrial reality is taken into account served as a driving force that made UTC one of the pioneer institutions and a current French leader in computational mechanics.