Characterising material models for silicone-rubber using an inverse finite element model updating method

Du Toit, Viljoen (2018-03)

Thesis (MEng)--Stellenbosch University, 2018.

Thesis

ENGLISH ABSTRACT: Silicone-rubber was investigated and its mechanical behaviour was characterised. Uni-axial tensile tests were conducted on two different sample geometries: rectangular flat strip and dumbbell shaped. Bi-axial bubble inflation tests were done on membranes and unconstrained uni-axial compression tests were conducted on cylindrical samples. Two identification methods were incorporated to determine three constitutive hyper-elastic material models from every experimental test: the direct and inverse. The direct method is the more traditional approach, where experimental data is used with a least squares fit to determine the constants that govern the material model. The inverse method is fundamentally different, it requires a finite element (FE) model and experimental results. The experimental results are used as boundary conditions in the FE model. Numerical optimisation is then used to obtain the material model constants that minimise the error between the FE model and the experimental results. The material models investigated in this thesis include the Mooney-Rivlin two- and three parameter models along with the Ogden three parameter model. Finally, an independent validation test was done, with a complex stress state. The validation test along with the extrapolation of each material model into all three stress states (uni-axial tension, -compression and bi-axial tension), served as the criteria to determine the best material model and identification method. It was found that the Mooney-Rivlin three parameter model obtained from uni-axial tensile tests (both sample geometries) using both the direct and inverse FE model updating method delivered the best results. However, additional user-input constraints were needed for the direct method (inverse method required no constraints) to obtain a material model that predicted feasible material behaviour.

AFRIKAANSE OPSOMMING: Silikoon-rubber is ondersoek ten einde sy meganiese gedrag te karakteriseer. Eenassige trektoetse was uitgevoer op twee verskillende steekproef geometrieë: reghoekige plat strook en ’n hondebeen vorm. Verder was tweeassige borrel-inflasietoetse op membrane uitgevoer, asook onbeperkte eenassige druktoetse op silindriese monsters. Twee identifikasiemetodes is gebruik om drie hiperelastiese-materiaalmodelle vanaf elke eksperimentele toets te bepaal: die direkte en die inverse metodes. Die direkte metode is die meer tradisionele benadering waar eksperimentele data gebruik word met ’n minste vierkante passing om die konstantes wat die materiaalmodel beheer, te bepaal. Die inverse metode verskil fundamenteel hiervan deurdat dit ’n eindige element model tesame met die eksperimentele resultate vereis. Die eksperimentele resultate word as grensvoorwaardes vir die eindige element model gebruik. Numeriese optimering word dan gebruik om die materiaalmodel se konstantes te bepaal wat die fout tussen die eindige element model en die eksperimentele resultate minimeer. Die materiaalmodelle wat ondersoek is sluit in, die Mooney- Rivlin twee- en drie parametermodelle, sowel as die Ogden drie-parameter model. Tenslotte was ’n onafhanklike trektoets met ’n komplekse spanningstoestand uitgevoer om die modelle te valideer. Die valideringstoets saam met die ekstrapolasie van elke materiaalmodel in al drie spanningstoestande (eenassige spanning, -druk en tweeassige spanning) dien as die kriteria om die beste materiaalmodel en identifikasiemetode te bepaal. Daar was bevind dat die drie-parametermodel van Mooney-Rivlin, verkry vanaf eenasige trektoetse (vir beide steekproef geometrieë) met behulp van beide die direkte metode en die inverse eindige element model opdateringsmetode, die beste resultate gelewer het. Daar word egter addisionele beperkings benodig vir die gebruik van die direkte metode (teenoor die inverse metode wat geen beperkings vereis nie) ten einde ’n wesenlike model te verkry, wat logiese materiaalgedrag voorspel.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/103750
This item appears in the following collections: