The main source of the time-dependent behavior of fiber-reinforced composites is their polymeric matrix, which causes concerns about their long term durability. Although for composites where organic fibers such as Graphite are used, the fibers are also a contributing factor. A composite material may exhibit an appreciable amount of creep, depending on the state of stress and temperature. Viscoelastic flow in the matrix and internal flaw formation and growth are the main sources of this creep. Thus a study was made on the viscoelastic behavior of GI/BMI fiber reinforced composite. An experimental method for testing a large number of composite materials in compression was developed. The samples were tested according to the test matrix consisting of combinations of static and cyclic loads and temperatures. The fixtures were calibrated to check the validity of measurements and reproducibility of results. Stress gradients were caused by frictional effects between the fixture and samples. The modulus change of samples over a period of time were studied. Bending parameters in samples were measured and analyzed for different stresses, clamping forces, temperatures and time. Mechanical models were used to explain the basic principles behind creep of a viscoelastic material followed by a theoretical explanation and study of creep. The linear and non-linear viscoelastic constants were studied and a methodology to analyze these results was presented. The linear and non-linear constants were used in a prediction model and predictions of a composite creep strain with time were made. Creep data obtained tor [45/0/-45/90]������ for a period of three months were compared to the prediction model. / Graduation date: 1994

A comparison of the creep/duration of load (DOL) performance of a new structural wood composite material called Parallam® parallel strand lumber (PSL) to two grades of machine-stress-rated (HSR) Douglas-fir lumber is presented in this thesis. Evaluation of the creep/DOL performance was made on nominal 2x4 members under constant bending stress at three stress levels. A total of 306 test specimens were evaluated for a 15-1/2 month time period. The analysis suggests that the duration of load effect for Parallam PSL was consistent with the Madison curve for the time period studied while the MSR Douglas-fir lumber was consistent with recent duration of load models developed for structural lumber. The analysis also indicates that the current duration of load adjustment factors can be applied to develop working stresses for Parallam. The creep behaviour of the Parallam PSL was found to be equivalent or better than the two MSR lumber grades under dry-service conditions. Furthermore, evidence of linear viscoelastic behaviour was found for all test materials within the range of applied stresses evaluated. Two mathematical models of creep were fitted to the creep data and compared. A '4-parameter linear viscoelastic' model fitted the creep data better than an empirical 'power curve' model. The model parameters developed provide a basis for estimating the mean creep behaviour and variability in creep response for these materials under in-service load conditions for dry-service environments. / Forestry, Faculty of / Graduate

Bibliography: pages 92-98. / In this thesis a simplified analysis procedure is developed, in which creep laws are decoupled from damage laws, for the purposb of constructing methods of use in the early stages of high temperature design. The procedure is based on the creep and damage laws proposed by Kachanov and Rabotnov. The creep laws are normalised. with respect to a convenient normalising stress. As a consequence of this normalisation, the dependence of the creep law on the stress constant, the time and temperature functions, and the actual load level is removed. In addition, if the reference stress of the component is chosen as the normalising stress, the creep law becomes insensitive to the stress exponent. The non-dimensional creep laws are then implemented in a standard finite element scheme, from which the results of a stationary state creep analysis are then in non-dimensional form. In order to estimate rupture times, the maximum stationary stresses in a component are used together with the damage laws. Conservative failure criteria are derived from the creep and damage laws to extend the method to residual life assessment and damage monitoring. The procedure is illustrated and tested against simple examples and case studies.

This study introduces a new three-dimensional (3D) multi-scale constitutive framework for the nonlinear viscoelastic analysis of laminated and pultruded composites. Two previously developed nonlinear micromechanical models for unidirectional and in-plane random composite layers are modified to include time-dependent and nonlinear behavior. A new recursive-iterative numerical integration method is introduced for the Schapery nonlinear viscoelastic model and is used to model the isotropic matrix subcells in the two micromodels. In addition, a sublaminate model is used to provide for a through-thickness 3D nonlinear equivalent continuum of a layered medium. The fiber medium is considered as transversely isotropic and linear elastic. Incremental micromechanical formulations of the above three micromodels are geared towards the time integration scheme in the matrix phase. New iterative numerical algorithms with predictor-corrector type steps are derived and implemented for each micromodel to satisfy both the constitutive and homogenization equations. Experimental creep tests are performed for off-axis pultruded specimens in order to calibrate and examine the predictions of the constitutive framework for the multi-axial nonlinear viscoelastic response. Experimental creep data, available in the literature, is also used to validate the micromodel formulation for laminated composite materials. Nonlinear viscoelastic effects at the matrix level, such as aging, temperature, and moisture effects can be easily incorporated in the constitutive framework. The multi-scale constitutive framework is implemented in a displacement-based finite element (FE) code for the analysis of laminated and pultruded structures. Several examples are presented to demonstrate the coupled multi-scale material and structural analysis.