By deploying a systematic approach for coupling of different damage models, the project aim to develop predictive tools to describe damage accumulation in composites that are concurrently exposed to (i) infrequent ultrashort overheating, (ii) long-term isothermal aging and (3) and mechanical decay due to cyclic deformation. In view of the complexity of the mapping space, no model exists that can even consider two of those phenomena concurrently and thatÂ’s the main advantage of the proposed technology. Due to the modular design of the proposed platform, other essential damage behaviors such as photo-oxidation and hydrolysis can be later integrated into this platform, and used concurrently in simulation of the composite behavior under thermomechanical boundary conditions. Model used in the platform can be updated/modified or replaced for any compound, and only onetime training is needed once the platform is assembled. To show the feasibility of such platform, in Phase I we develop compatible models of the abovementioned three phenomena and implement them in the platform to validate them against experimental data. Isothermal oxidation phenomena is described with respect to oxygen reaction-diffusion, chemical shrinkage strain/stress, and multistep degradation. Considering oxygen absorption as a global damage precursor, Overheating is modeled using a hybrid physics-informed machine-learned hybrid approach that will be specifically trained to capture oxygen diffusion, gas transport, anisotropic heat pattern, mass loss and consequent mechanical damage in composite structures. Using the concept of network evolution modular platform, this work provides new prediction capabilities that can consider and couples the effects of continuous environmental/mechanical damage with those of thermal shocks in an arbitrary compound. The proposed hybrid framework is a more rigorous, history-dependent, extensible based on location and environment, and implementable (based on continuum mechanics) approach and portends a transformational change in damage predictive approaches. The final models will be coupled into the platform and two predictive tools will be developed based on that to describe damage accumulation and failure in composite systems. The model will be implemented and commercialized as Ansys User-programmable Feature.
