The simulation of battery systems needs to consider structural heterogeneity in battery electrodes as the electrodes are highly irregular in shape and size while containing pores or even cracks at different length scales, which may result in non-uniform transport and kinetics throughout the electrodes. A more realistic battery model necessitates considering realistic microstructure heterogeneity as well as incorporating ongoing microscale phenomena (e.g., the predictability of diffusion, interfacial side reactions, and stress evolution) into numerical problems governing macroscale behaviors. However, developing such model with detailed 3D microstructure can be very computationally expensive for direct simulations, as shown in Figure 1. Here, we propose to reduce the computational cost by developing a more realistic battery model via the variational multiscale method (VMM). The VMM couples micro- and macro- scales by decomposing the problem into micro- and macro- scale problem. In detail, we apply the VMM to decompose the response (e.g., ) into coarse-scale (e.g., 1D macro-scale) and fine-scale (e.g., 3D micro-scale), as shown in Figure 1. The associated governing PDEs will accordingly be decomposed into coarse-scale components and fine-scale components. Without introducing the assumption of scale separation, one of the key features of VMM lies in that the relevant governing PDEs will be decomposed into coarse-scale and fine-scale components. Therefore, two sets of solution will be obtained on different domains, i.e., is the macro solution at, and is the micro solution at. The developed model aims at obtaining accurate micro-solution at micro-scale in a subdomain where it is known a priori that important physicochemical phenomena are likely to occur. The effect of micro-scale is then considered in the solution of the problem. Due to consideration of integrals or the weak formulation as distribution (i.e., average of functions over certain areas of the computational domain), we can perform integration operation on the function that represents the solution to the problem, regardless of existence of discontinuity points in the domain. Thus, one of the main advantageous of developed model is that it resolves the heterogeneities of micro-scale without considering the minor details (such as discontinuity points) but rather our focus in invested in the distribution of the function in the a priori areas of the subdomain, which reduces the computational costs of the numerical analysis. This is of significant importance, especially for cases that the effect of ongoing micro-scale phenomena is more significant than the minor details of the micro-scale. Figure 1
Read full abstract