Graph-based simulated annealing: a hybrid approach to stochastic modeling of complex microstructures

A stochastic model is proposed for the efficient simulation of complex three-dimensional microstructures consisting of two different phases. The model is based on a hybrid approach, where in a first step a graph model is developed using ideas from stochastic geometry. Subsequently, the microstructure model is built by applying simulated annealing to the graph model. As an example of application, the model is fitted to a tomographic image describing the microstructure of electrodes in Li-ion batteries. The goodness of model fit is validated by comparing morphological characteristics of experimental and simulated data.