On error controlled numerical model reduction in FE2-analysis of transient heat flow
Artikel i vetenskaplig tidskrift, 2019

Numerical model reduction is exploited for solving the two-scale problem that arises from the computational homogenization of a model problem of transient heat flow. Since the problem is linear, an orthogonal basis is obtained via the method of spectral decomposition. A key feature is the construction of a symmetrized version of the space-time variational format, by which it is possible to estimate the error from the model reduction in (i) energy norm and in (ii) a given quantity of interest. In previous work by the authors, this strategy has been applied to the solution of an individual representative volume element problem, whereas the error transport to the macroscale problem in the finite element squared (FE2) approach was ignored. In this paper, such transport of error is included in the error estimate. It is remarkable that it is (still) possible to obtain guaranteed bounds on the error, as compared to the fully resolved discrete finite element problem, while using only the reduced basis and with minor extra computational effort. The performance of the error estimates is demonstrated via numerical results, whereby the subscale is modeled in three dimensions, whereas the macroscale problem is either one or two dimensional.

error control

model reduction

computational homogenization


Fredrik Ekre

Chalmers, Industri- och materialvetenskap, Material- och beräkningsmekanik

Fredrik Larsson

Chalmers, Industri- och materialvetenskap, Material- och beräkningsmekanik

Kenneth Runesson

Chalmers, Industri- och materialvetenskap, Material- och beräkningsmekanik

International Journal for Numerical Methods in Engineering

0029-5981 (ISSN) 1097-0207 (eISSN)

Vol. 119 1 38-73

Numerisk modellreduktion vid beräkningsbaserad homogenisering av deformation och strömning i porösa medier

Vetenskapsrådet (VR) (2015-05422), 2016-01-01 -- 2019-12-31.



Sannolikhetsteori och statistik




Mer information

Senast uppdaterat