Mechanical behavior analysis of flexible parts in a real-time virtual environment for aided design

The interaction between a designer and a virtual prototype is a promising way to optimise the design of parts. Indeed mechanical industries of automotive and aeronautics already use real-time interactive simulators to evaluate virtual prototypes composed of rigid parts. Thus, there exists an industrial need to solve the problem of real-time deformation of flexible parts. Our method is composed of two phases: a campaign is calculated during the first phase of training, and then these results are used during the second phase of real-time immersion. In this paper, we focus on the phase of training. We present an a posteriori method and an a priori method. The a posteriori method need to complete calculation campaign to apply the Karhunen-Loeve expansion and keep only representative data. The a priori method is an adaptive strategy. As a linear combination of shape functions defines the displacement, these functions are enriched during the campaign. The size of the shape functions basis increases with time, so we use a model reduction approach. These two methods allow to calculate a surface response of possible displacements using a model reduction technique. These methods are applied on an automotive hose, and should allow to simulate complex mechanical behaviour of flexible parts representative of many industrial applications.