The capture of rare cells from complex fluids, such as circulating tumor cells (CTCs) from a peripheral blood sample, has the potential to significantly advance our understanding and treatment of disease. Here, we consider microfluidic devices designed to isolate rare cells by bringing them into contact with, and binding the cells to, an antibody-functionalized obstacle array geometry. Each downstream biomedical assay, such as single-cell genetic analyses or enumeration for the monitoring of disease progression, requires a different balance of capture efficiency and sample purity in isolating the rare cells; this work addresses that need for application-specific microfluidic device geometries by presenting a series of numerical simulations for design optimization. We have developed coupled computational fluid dynamics, particle advection, and cell adhesion Monte Carlo simulations that predicts the probability of capturing target and contaminating cells in a given device geometry, and have applied these simulations to the study the capture of prostate and pancreatic cancer cells. We expand these simulations to consider the effect of dielectrophoresis (DEP), and show that it is possible to apply DEP forcing within the obstacle array to simultaneously increase the capture of target pancreatic cancer cells (using positive DEP) and decrease the capture of contaminating cells (using negative DEP). Finally, we present a transfer function approximation of cell transport in obstacle arrays, and apply that approximation to study the effects of reversing arrays and off-design boundary conditions. This work advances our understanding of rare cell immunocapture in microfluidic obstacle arrays, lays the groundwork for the experimental study of DEP-immunocapture devices, and presents an engineering framework to identify optimized geometries for each unique rare cell capture application.