The long-term goal of this project was to develop a CFD model for assessing respiratory exposure to airborne particulate matter on the basis of aerosol size and density. The discrete vortex method (DVM) was chosen for modeling the airflow in this project because it is well-suited for capturing the time-dependent flow which occurs when a person stands in front of a local exhaust ventilation hood. The original specific aims of this research project were: 1. Write a program which models flow of monodisperse aerosols in the near wake for a bluff body flow, and test the program for solution convergence; 2. Run numerical simulations of air flow and particle motion matching initial conditions of the experiments performed by Helgeson (1992) with mondisperse aerosols having various Stokes numbers flowing past a cylinder, and compare the results of these numerical simulations with data from Helgeson's experiments to identify limitations in the models. The first specific aim was implemented in a straight-forward manner. Before integrating the DVIVI with the particle-tracking algorithm, benchmark testing of the particle tracking code was performed for a single particle moving in a uniform velocity field. This testing demonstrated good agreement between simulation results and an analytic solution when particle size was at or below 0 (10 1). For larger particle sizes, an error maximum occurred around time 0(r) then vanished over long times. Here, r = particle relaxation time. During the course of model development, it became obvious that a rigorous test of the convergence properties of the discrete vortex method was required. The DVM was tested at Reynolds numbers of 5,232 and 140,000. For the higher Reynolds number tested here, convergence with respect to the time-averaged characteristics of the flow and validation with experimental data were found easily. Convergence was not demonstrated at the lower Reynolds number. The second specific aim was not fully accomplished because testing of the airflow model 'required much effort. Furthermore, it was found that the original dataset to be used was inadequate for validation purposes. For this reason, laboratory experiments were performed to obtain particle velocities and concentrations downstream of a cylinder. Although polydisperse particles were used in the experiments, the most reliable of these data was for particles with a Stokes number of 0(1 0- 7). Preliminary testing of the particle-tracking module in conjunction with the DVM revealed sensitivity to particle size and location of particle release into the domain. Furthermore, a more detailed preliminary simulation matching the experimental conditions showed some agreement with the particle concentration field, although this agreement was not in entirety. Comparison of the time-averaged concentration field for this small particle simulation and a tracer gas simulation revealed large differences. The tracer gas had a uniform concentration everywhere in the domain. In contrast, the small particles collected at the edge of the recirculation zone; this high concentration region extends into the far wake of the cylinder. Unfortunately, definitive results were not garnered with the particle tracking code. For this reason, no conclusive statements can be made with respect to the implications of these results for worker exposure to particulate matter.