Composite Structures Laboratory


Textile Composites
Size Effects
3-D Compsites
Manufacturing Effects
Blast Response

Adhesive Joints
Reliability Prediction
Thin-Film Laminates
Sandwich Fracture
Cohesive Zone Models

Multi-Scale Theories
Progressive Failure
Cohesive Method
C/SiC Composites

Biological Structures
Insect Wing

University of Michigan
Aerospace Engineering
Lab CTools


Modeling of the Curing Process of Triaxially Braided Composites and Measuring In-situ Matrix Material Properties
PI: Anthony M. Waas
Collaborators: Alan Wineman (ME)-coPI and John Kieffer (MSE)-coPI

Particpants: Christian Heinrich, Michael Aldridge, Lang Sui


In previous research it has been shown through macroscopic experiments, that the properties of the matrix in a textile composite shows different constitutive behavior when cured with (composite) and without (virgin) carbon fibers. In order to make accurate predictions about the response of a textile composite, it is necessary to understand the physico-chemical processes that the matrix undergoes in the process of curing and be able to accurately model these processes in order to predict the deformation response of the matrix for subsequent use. 

To address these issues,  two approaches are pursued: The curing processes is simulated including the heat created by the exothermic reaction during the curing process, the heat conduction within the material, the creation of stresses, the change in (visco elastic) material properties and the forming of new networks.

The second approach to obtain the material properties is through experiments. For that purpose micro hardness tests are performed with a Nanoindenter. It is a very small pyramid, that when pushed into the material creates a reaction force. The data obtained from the experiment represents the force-time-displacement response. Since we only know the integrated force from the experiment it is necessary to solve the corresponding boundary value problem to be able to relate the reaction force to all the stresses, strains and therefore material properties under the indenter. The solution can be fully analytical (such as in the Hertz contact problem), approximate or it can be obtained through a finite element simulation.

For the problem at hand the complex geometry and material behavior doesn’t allow for any closed form solution to be implemented. In order to create an appropriate analysis of the experiment one has to use a numerical method. The substrate is modeled as a visco elastic- plastic, non-linear hardening material using the commercial finite element (FE) package Abaqus. The problem is that in the FE analysis,  material properties are the known inputs and the reaction force and displacement is the output, whereas in the experiment, these roles are reversed.

It is therefore necessary to adjust the material properties in the FE analysis to the point that the output matches the experiment. Since every FE simulation is very time intensive, a meta-model is implemented that predicts the output of the FE simulation according to the inputs. The optimization problem of fitting computer and real experiment to each other will then be performed with that meta-model.

    Copyright © 2008 Composite Structures Laboratory | Webmaster: Scott Stapleton