Yokoyama Ryota, Matsuzaki Ryosuke, Kobara Tadahiro, Takahashi Kentaro
Department of Mechanical Engineering, Tokyo University of Science, Japan.
Heliyon. 2020 Oct 5;6(10):e05147. doi: 10.1016/j.heliyon.2020.e05147. eCollection 2020 Oct.
The models and parameters related to the generated curing heat in the molding simulation of composite materials are dependent on the type of resin used and the experimental conditions. Therefore, in this study, we estimated the generated curing heat that changes with time by a data assimilation method, which combines the observation values with simulation values, so that the heat curing simulation of carbon fiber reinforced polymers (CFRPs) becomes closer to the experimental conditions. In the data assimilation method, the temperature distribution on the surface of the composite material was used as an observation value, and the generated curing heat was estimated using an ensemble Kalman filter. By optimizing the data assimilation parameters in advance using the response surface method and estimating the generated curing heat by numerical experiments, the generated curing heat could be estimated with an accuracy represented by the time mean error of less than 6%.
在复合材料成型模拟中，与固化热生成相关的模型和参数取决于所用树脂的类型和实验条件。因此，在本研究中，我们通过一种数据同化方法来估计随时间变化的固化热生成量，该方法将观测值与模拟值相结合，以使碳纤维增强聚合物（CFRP）的热固化模拟更接近实验条件。在数据同化方法中，将复合材料表面的温度分布用作观测值，并使用集合卡尔曼滤波器估计固化热生成量。通过预先使用响应面法优化数据同化参数，并通过数值实验估计固化热生成量，可以以小于6%的时间平均误差所表示的精度来估计固化热生成量。
