Computational & Technology Resources
an online resource for computational,
engineering & technology publications |
|
Civil-Comp Proceedings
ISSN 1759-3433 CCP: 65
FINITE ELEMENTS: TECHNIQUES AND DEVELOPMENTS Edited by: B.H.V. Topping
Paper III.2
Discretisation Error Estimator for Transient Dynamic Simulations J.Ph. Combe, P. Ladevèze and J.P. Pelle
Laboratoire de Mécanique et Technologie, Université Paris 6, Cachan, France , "Discretisation Error Estimator for Transient Dynamic Simulations", in B.H.V. Topping, (Editor), "Finite Elements: Techniques and Developments", Civil-Comp Press, Edinburgh, UK, pp 71-85, 2000. doi:10.4203/ccp.65.3.2
Abstract
In this paper, we demonstrate how the concept of error in the
constitutive relation enables proposing a response to the problem
of error estimation in transient analysis.
The construction of our error measurement is based on the reformulation of a transient dynamic problem. From the solution to the discretized model, we have built a set of fields which exactly satisfy the kinematic constraints, the initial conditions as well as the equilibrium equation. The quality of this numerical solution depends on the manner in which the constitutive relations are satisfied. Our error estimator can be used with both explicit and implicit time integration schemes. It is first computed on a simple linear dynamic problem with one degree of freedom, and different tests demonstrate the satisfactory behaviour of this estimator. Moreover, it has been compared with different indicators in the literature. Next, we explain how to apply the error measurement to problems involving both time and space. The initial unidimensional test results obtained from a cantilever beam, are then presented and discussed. Lastly, we introduce a new error indicator, which appears as a time integration scheme error indicator for the initial reference problem. This indicator allows to extract from the global error estimation, the main part relative to the time integration scheme chosen. It is then computed in order to evaluate the amount of error relative to the mass matrix lumping hypothesis for problems solved using the the explicit central differences method. purchase the full-text of this paper (price £20)
go to the previous paper |
|