Approximation of three-dimensional velocity characteristics of local exhaust inlets using the boundary element method

Local exhaust ventilation (LEV) is used to control airborne contaminants and reduce potential hazards in workplaces. LEV inlets (hood) are designed to capture and remove contaminants near sources of emission. Inlet airflow velocities are a primary factor affecting the capture of contaminants. Velocity characteristics of freestanding symmetrical inlet configurations have been studied and modeled empirically since the 1920's.;Three-dimensional airflow models have been proposed and developed for only the simplest of inlet configurations. They have not been developed for non-symmetrical inlet shapes or for the presence of nearby external plane barriers, which are common in actual workplace situations. Previous experimental studies of inlet velocities focused on measurements in one or two dimensions, and generally did not verify model predictions in three dimensions. The objectives of this study were to develop computer models to predict three-dimensional velocity characteristics for various inlet configurations; to obtain experimental three-dimensional velocity data; and to evaluate the predictive models by comparison with the experimental data.;Airflow velocity models were developed using the Boundary Element Method (BEM) in three dimensions, assuming airflow into the inlet to be potential flow. The BEM was selected because it allowed flexibility of inlet configurations, velocities could be calculated at specific points in the flow, and to minimize computational time.;Experimental studies were conducted for free-standing inlets and inlets with one and two external plane barriers. Velocities were measured in three-dimensions with two measurement systems: one using a three-dimensional differential-pressure velocity probe and the other using a non-directional thermal velocity probe. Experimental velocity data from these systems were in close agreement with the empirical data provided by other investigators.;The BEM model converged to the analytical solution for a square-flanged inlet when sufficient numbers of nodes and elements were employed. Model-predicted velocity contours were in generally close agreement with the experimental contours. The BEM models tended to slightly underpredict velocity magnitudes and directions in front of exhaust inlets. The discrepancies, typically about 8%, were attributed primarily to viscous effects and experimental error, and were acceptable for LEV inlet design with a conservative margin of error.