ICF10B

ORAL REFERENCE: ICF 1003060R EFFECTS OF HETEROGENEITY ON THE STRENGTH OF 3D COMPOSITES Sivasambu Mahesh 1 , Irene J. Beyerlein 2 , and S. Leigh Phoenix 1 1 Theoretical & Applied Mechanics, Cornell University, Ithaca NY 14853. USA. 2 Materials Science & Technology, Los Alamos National Laboratory, Los Alamos, NM 87545. USA. ABSTRACT Monte Carlo simulation interpreted with theoretical modeling is used to study the statistical failuremodesinunidirectionalcompositesconsistingofahexagonalarrayofelastic bersembedded in an elastic matrix. Composite structure is idealized using the chain-of-bundles model in terms of bundles of length Æ arrangedalongthe berdirection.Fiberselementstrengthsin Æ-bundles are taken to be Weibull distributed and Hedgepeth and Van Dyke load sharing is assumed for transverse berbreakarrays. Simulations of Æ-bundlefailurerevealtworegimes.When berstrengthvariabilityislow,the dominantfailuremodeisbygrowingclustersof berbreaksuptoinstability.Whenthisvariability ishigh,clusterformationissuppressedbyamoredispersed berfailuremode.Correspondingto these two cases, we construct simple models that predict the strength distribution of a Æ-bundle. Their predictions compare very favorably with simulations in the two cases. KEYWORDS unidirectional bercomposites,brittle/ductiletransition,extremevalueproblem,chainofbundles model, shear-lag model. INTRODUCTION Quasistatic failure of unidirectional composite materials, which consist of long aligned reinforcing bers embedded in a matrix is a complex stochastic process. While complexity stems from the occurrence of various damage events preceding formation of a catastrophic crack, statistical variationinstrengthprimarilyarisesduetothevariabilityin berstrength.Consequentlycomposite tensile strength is itself a statistical quantity and methods to determine its distribution are of considerablesigni canceinassuringcompositereliability. Idealization of composite structure and material properties are found to be inevitable before furtheranalysiscanbeattempted.Inthisstudyweassumesti linearelastic bersarrangedin

RkJQdWJsaXNoZXIy MjM0NDE=