scholarly journals A general hyper-reduction strategy for finite element structures with nonlinear surface loads based on the calculus of variations and stress modes

2021 ◽  
Vol 379 ◽  
pp. 113744
Author(s):  
Lukas Koller ◽  
Wolfgang Witteveen ◽  
Florian Pichler ◽  
Peter Fischer
Author(s):  
Lukas Koller ◽  
Wolfgang Witteveen ◽  
Florian Pichler ◽  
Peter Fischer

Abstract Model reduction via projection is a common method to accelerate time integration of finite element (FE) structures by reducing the number of degrees-of-freedom (DOFs). However, nonlinear state-dependent surface loads are usually computed based on the nonreduced DOFs of the FE model. When a considerably high number of DOFs are involved in the nonlinear surface loads, their computation becomes a bottleneck. This paper presents a general approach for reduced time integration and reduced force computation for FE models. The required force trial vectors can be computed easily and systematically out of deformation trial vectors, commonly called “modes.” Those force trial vectors, which we call “stress modes,” can be determined a priori so that a nonlinear computation of the full system is not necessary. The new idea in this contribution is that stress recovery is used to decrease the number of equations for the force computation. A general framework for semihyper-reduction (SHR) is developed and its practical implementation is discussed. The term SHR is introduced because it is an intermediate approach between the straight-forward method of using the FE DOFs and pure hyper-reduction (HR) where the FE DOFs are omitted for computing state-depended surface loads. In order to demonstrate the proposed SHR approach practically, a numerical example of a planar crank drive is given, where a hydrodynamic lubrication film separates piston and cylinder. Thereby, very good result quality has been observed in comparison to a finite difference reference solution.


1990 ◽  
Vol 57 (3) ◽  
pp. 758-761 ◽  
Author(s):  
Michel P. Robert

The gap profile of a two-dimensional self-acting gas bearing is determined such that the static stiffness it can achieve is maximum. Three fundamental profiles are obtained according to the stiffness mode to be considered: normal, pitch, or roll. The optimization process takes place within the framework of the compressible lubrication theory among all the profiles having a given minimum film thickness. The method proposed here is based on the calculus of variations and uses a finite element technique coupled with an iterative mapping to converge to the final solution. As an example, the case of a square bearing is treated and the three fundamental gap profiles, along with their optimum characteristics, are plotted to illustrate the solutions.


2014 ◽  
Vol 989-994 ◽  
pp. 751-754
Author(s):  
Wei Zhang ◽  
Zhou De Qu ◽  
Xiao Hu Deng ◽  
Xing Wang Duan

The excessive residual stress induced by quenching in steels will easily result in distortion and failure of parts. In order to obtain the more suitable quenchant, quenching process of Cr12MoV steel with different mediums involving water and oil are simulated, respectively. In present paper, the influence of nonlinear surface heat transfer coefficient, thermodynamic parameters and latent heat are considered comprehensively. The distribution of temperature, microstructure, hardness and residual stress after quenching for Cr12MoV steel are simulated by DEFORM finite element software. According to the results mentioned above, the variations of each field of the steel are analyzed.


2010 ◽  
Vol 10 (2) ◽  
pp. 137-163 ◽  
Author(s):  
C. Carstensen ◽  
C. Ortner

AbstractAmongst the more exciting phenomena in the field of nonlinear partial differential equations is the Lavrentiev phenomenon which occurs in the calculus of variations. We prove that a conforming finite element method fails if and only if the Lavrentiev phenomenon is present. Consequently, nonstandard finite element methods have to be designed for the detection of the Lavrentiev phenomenon in the computational calculus of variations. We formulate and analyze a general strategy for solving variational problems in the presence of the Lavrentiev phenomenon based on a splitting and penalization strategy. We establish convergence results under mild conditions on the stored energy function. Moreover, we present practical strategies for the solution of the discretized problems and for the choice of the penalty parameter.


2012 ◽  
Vol 16 (2) ◽  
pp. 473-480 ◽  
Author(s):  
Adriana Amaro Diacenco ◽  
Antônio Marcos Gonçalves de Lima ◽  
Edmilson Otoni Corrêa

Sign in / Sign up

Export Citation Format

Share Document