A reverse updated Lagrangian finite element formulation for determining material properties from measured force and displacement data

2014 ◽  
Vol 54 (6) ◽  
pp. 1375-1394 ◽  
Author(s):  
M. Tartibi ◽  
D. J. Steigmann ◽  
K. Komvopoulos
Author(s):  
Nicolas Aquelet ◽  
Benjamin Tutt

The modelling of parachutes at Irvin Aerospace Inc. was based on the penalty Euler-Lagrange coupling method to compute the interaction between an Arbitrary Lagrange Euler formulation for the air flow and an updated Lagrangian finite element formulation for the canopy dynamics. This approach did not permit the effect of fabric porosity to be accounted for. In this paper, a new porosity Euler-Lagrange coupling models the fabric permeability by assessing the interaction forces based on the Ergun porous flow model. This paper provides validations for the technique when considering parachute applications and discusses the interest of this development to the parachute designer.


Author(s):  
Enrico Babilio ◽  
Stefano Lenci

The present contribution reports some preliminary results obtained applying a simple finite element formulation, developed for discretizing the partial differential equations of motion of a novel beam model. The theoretical model we are dealing with is geometrically exact, with some peculiarities in comparison with other existing models. In order to study its behavior, some numerical investigations have already been performed through finite difference schemes and other methods and are reported in previous contributions. Those computations have enlightened that the model under analysis turns out to be quite hard to handle numerically, especially in dynamics. Hence, we developed ad hoc the total-lagrangian finite-element formulation we report here. The main differences between the theoretical model and its numerical formulation rely on the fact that in the latter the absolute value of the shear angle is assumed to remain much smaller than unity, and strains are piecewise constant along the beam. The first assumption, which actually simplifies equations, has been taken on the basis of results from previous integrations, mainly through finite difference schemes, which clearly showed that, while other strains can achieve large values in their range of admissibility, shear angle actually remains small. The second assumption led us to define a two-nodes constant-strain finite element, with a fast convergence, in terms of number of elements versus solution accuracy. Although, at the present stage of this ongoing research, we have only early results from finite elements, they appear encouraging and start to shed new light on the behavior of the beam model under analysis.


Author(s):  
K. S. Surana ◽  
H. Ngyun

Abstract This paper presents a new theoretical and computational framework for computing solutions of right classes for laminated composites using 2D p-version least squares finite element formulation incorporating the correct physics of interlamina behavior. At the interface between two laminas of dissimilar materials we have continuity of displacements u, v, stresses σyy, τxy, and strain εxx, while the stress σxx and the strains εyy and γxy are discontinuous. Thus, a finite element formulation, incorporating the physics of laminate behavior, would require interpolation of u, v, εxx, σyy and τxy instead of u, v, σxx, σyy and τxy which is generally the case in most mixed formulations. In the p-version LSFEF presented here, we interpolate u, v and σyy, τxy (εxx = ∂u/∂x is used to eliminate εxx as a variable) using appropriate p-version interpolations which would ensure correct interlamina behavior of these components. When the mating lamina properties are different, interlamina discontinuity of σxx, εyy and γxy is automatically generated due to dissimilar material properties of the laminas. In this formulation interlamina jumps in σxx, εyy and γxy do not constitute singularities, hence mesh refinements and higher p-levels are not needed in the vicinity of inter-lamina boundaries. The major thrust of this paper is to construct interpolations for the dependant variables that are of right classes in appropriate spaces so that a sequence of converged solutions in these spaces may be computed which, when converged, would yield a numerical solution that has exactly the same characteristics (in terms of continuity and differentiability) as the analytical or theoretical (strong) solution.


Sign in / Sign up

Export Citation Format

Share Document