Simulation And Optimization For The Pultrusion Process Of Carbon Fiber Reinforced Polymer

Carbon fiber reinforced polymer (CFRP) is known as advanced composite materials (ACM). And pultrusion, one of the major techniques to manufacture composites, can produce products of arbitrary length with high efficiency and high material utilization ratio. Due to the high specific strength and rigidity of pultruded products, they are widely used in petroleum, aerospace, construction, chemical and other industries. However, the technological parameters making plutruded products are determined by experiences and"try and error methods", which usually lead to low efficiency, high energy consumption and unreliable quality for final products. Therefore, mathematic simulation, optimization and experiments for pultrusion process were carried on in our research.Pultruded products are in motion during the whole pultrusion process. The two dimensional unsteady heat conduction and curing models were established on the basis of heat conduction and reaction kinetics of polymer, which included coupling relationship of temperature and curing degree. And the kinetic parameters of epoxy resin were calculated through regression analysis and DSC experiments. In the process of computation, finite element, finite difference and indirect decoupling methods were combined to calculate temperature and curing degree of pultruded products under difference technological parameters, and laid the foundation for further optimization of pultrusionTemperature and curing degree tests were processed to verify the reliability of simulation program. Fiber Bragg grating temperature sensors, encapsulated by metal capillaries to prevent load effect and curing shrinkage (non-temperature effect), were applied to monitor the temperature of products during pultrusion. The test results agreed well with the simulated results. At the same time, the curing degree of pultruded products was measured by Soxhlet extraction, and they were also in substantial agreement with the simulated ones. Therefore, it was conluded that the simulation program could predict temperature and curing degree of pultruded products effectively and reliably.According to the simulated results, artificial neural network was trained to form the relationship between technological parameters (die temperatures, pull speed) and quality of products (curing degree). It saved cost and time for experiments, and made the optimization of technological parameters more feasible.Finally, the mutil-objective optimization problem, including die temperatures and pull speed goals, was solved by improved non-dominated sorting genetic algorithm (NSGA-â…¡), and found out the optimal parameters for pultrusion process. On the other hand, this optimization method is also suitable for other multi-objective optimization problems.