If we wish to use a material for engineering purposes, we must answer three basic questions: how strong is the material, how stiff is the material, and how long will it last? Durability is defined by the answer to the third of these questions, and frequently takes the form of life prediction. The characteristics displayed by such a material under long-term conditions, such as resistance to creep, stress rupture, and fatigue, are quite different from, for example, metals. The life of composite systems is often determined by the accumulation of defects and damage, rather than by the occurrence or growth of such flaws. And the changes in stiffness and strength during that accumulation process may be quite large, of the order of 50 percent or more, before fracture occurs. Hence, it is necessary to consider large changes in the constitutive behavior of the materials if a representation or predictive model is to be constructed. Perhaps the greatest challenge associated with such an enterprise is the determination of the constitutive information that is needed to uniquely and completely define the long-term behavior of composite materials, especially under complex applied conditions, i.e., the determination of what to measure and how to measure it.
This paper will address the question of how to construct mechanistic models and related experiments that provide an interpretative link between the fundamental mechanical, chemical, kinetic, and thermodynamic processes that control the long-term behavior of composite materials and the remaining strength and life that defines durability and damage tolerance of those materials. The paper begins with the premise that any coherent philosophy must include a systematic and consistent analytical representation of all processes that define the evolution of properties and performance and of the resulting local stress states and material states that determine remaining strength and life. For the present case, kinetic theory, in a generalized form, is used to make such a construct. Then, we add the premise that the analytical representation must be cast in terms of constitutive quantities that are independent variables, i.e., that can be measured in the laboratory with experiments that produce unique and clearly defined physical constants, and that a canonical set of such constants can be defined. Finally, we address the question of how to actually measure such constants. At the heart of this discussion is the question of feasibility. A philosophy and attending model that requires the measurement of vast arrays of physical constants is not likely to be of interest to the applied community. And experiments that require some significant fraction of a lifetime to conduct should be minimized, or avoided if possible. “Accelerated testing” is the quixotic answer to this constraint, but the acceleration of the controlling processes must be done with a complete knowledge of how they work and what an appropriate accelerating parameter is. The present paper will discuss these questions and offer examples of defining experiments, a general life prediction philosophy, and some accelerated test methods that enable the application of composite materials in this context.