Electro-Mechanical Analysis Involving Stochastic Contact Hypothesis in Engineering - EeMAISCHen

Coordinating Institution: CRP Henri Tudor
Other Partner(s): IEE S.A. , University of Liège
From: 01/01/2009
To: 30/06/2011
Budget: 278,000.00€
Contact(s): Rauchs Gaston

Progress Summary 2009

The main objective of this project is the development of a computational method based on numerical optimization and finite element modelling for improving the performance of a pressure sensor relying on the electric current transfer between two contacting coated polymer membranes. To this avail, electro-static current flow will be implemented in an in-house solid mechanics finite element code.

Contact constitutive laws based on existing stochastic micromechanical models, which derive the true microscopic contact surface from surface roughness parameters, material parameters and mechanical pressure, like for example Greenwood-Williams or Cooper-Mikic-Yovanovich models, will be implemented, both for the mechanical pressure transfer between surfaces as well as for the flow of electric current between surfaces.

For efficient use with gradient-based numerical optimization algorithms, the resulting coupled electric-mechanical finite element code will be extended to compute derivatives with respect to statistical parameters required in the rough surface contact laws. Sensitivity analysis based on direct differentiation will be implemented in order to obtain the sensitivities of the governing field variables with significantly less computational effort and better accuracy than common finite difference algorithms. For the design of the pressure sensor, the finite element code and the sensitivity analysis will include the ability to treat locally varying roughness and conductivity parameters of the surface During the first year of the project, the in-house finite element code has been extended for handling coupled electro-static and mechanical analysis of boundary value problems. In a second step, a rough surface contact law has been implemented into the finite element code.

Because of the generally stiff nature of such contact laws, appropriate algorithms have to be used for dealing with this kind of contact analysis. Some efficient solution algorithms use specifically adapted augmented Lagrangian algorithms for treating rough surface contact laws. However, in the framework of fast sensitivity analysis, this algorithm has some drawbacks regarding variable storage, computational effort and accuracy. For this reason, an algorithm using successive approximations, with increasing stiffness of the rough contact law until reaching the true physical value, has been developed and successfully implemented into the finite element code.