Grinding is a complex material removal process with a large number of parameters influencing each other. In the process, the grinding wheel surface contacts the workpiece at high speed and under high pressure. The complexity of the process lies in the multiple microscopic interaction modes in the wheel-workpiece contact zone, including cutting, plowing, sliding, chip/workpiece friction, chip/bond friction, and bond/workpiece friction. Any subtle changes of the microscopic modes could result in a dramatic variation in the process. To capture the minute microscopic changes in the process and acquire better understanding of the mechanism, a physics-based model is necessary to quantify the microscopic interactions, through which the process output can be correlated with the input parameters. In the dissertation, the grinding process is regarded as an integration of all microscopic interactions, and a methodology is established for the physics based modeling. To determine the engagement condition for all micro-modes quantitatively, a virtual grinding wheel model is developed based on wheel fabrication procedure analysis and a kinematics simulation is conducted according to the operational parameters of the grinding process. A Finite Element Analysis (FEA) is carried out to study the single grain cutting under different conditions to characterize and quantify the grain-workpiece interface. Given the engagement condition on each individual grain with the workpiece from the physics-based simulation, the force, chip generation, and material plastic flow can be determined through the simulation results. Therefore, the microscopic output on each discrete point in the wheel-workpiece contact zone can be derived, and the grinding process technical output is the integrated product of all microscopic interaction output. From the perspective of process prediction and optimization, the simulation can provide the output value including the tangential force and surface texture. In terms of the microscopic analysis for mechanism study, the simulation is able to estimate the number of cutting and plowing grains, cutting and plowing force, probability of loading occurrence, which can be used as evidence for process diagnosis and improvement. A series of experiments are carried out to verify the simulation results. The simulation results are consistent with the experimental results in terms of the tangential force and surface roughness Ra for dry grinding of hardened D2 steel. The methodology enables the description of the 'inside story' in grinding processes from a microscopic point of view, which also helps explain and predict the time dependent behavior in grinding. Furthermore, the process model can be used for grinding force (or power) estimation for multiple-stage grinding cycles which includes rough, semi-finish, finish, and spark out. Therefore, the grinding process design can be carried out proactively while eliminating 'trial and error'. In addition, the grinding wheel model itself can be used to guide the recipe development and optimization of grinding wheels. While the single grain micro-cutting model can be used to study the mechanism of single grit cutting under various complex conditions, it can also be used to derive the optimal parameters for specific grains or process conditions.