Local Exhaust Ventilation is a primary method used by industrial hygiene and engineering practitioners to eliminate or reduce exposures to airborne contaminants. Capture hoods are an important component of many ventilation systems. Currently, the exhaust airflow required for these hoods is based on exceeding a specified minimum velocity at a point just beyond the greatest distance of the source from the front of the hood. Estimating the expected velocity at any point along the midline of the hood (V_x) is currently done using one of many predictive equations developed since the 1930’s. There are two problems with these equations: (1) they are based on unrealistically ideal conditions, and (2) they disagree substantially with each other.

This study measured the values of V_x along the midline of the hood face at distances of 1–14 inches in front of a capture hood. This was done for the ideal conditions used to develop the previously published equations. In addition, velocities were measured for realistic conditions, including all combinations of 3 0° cross-draft velocities and the presence or absence of both a manikin and work table. All tests were done in a large wind tunnel, measuring V_x with a particle image velocimeter. The objective was to compare V_x between ideal and non-ideal conditions and to published models.

For most test conditions, V_x declined nearly exponentially with increasing distance (x), from the hood face. Log-transformation revealed two different slopes at x=1” to 8” and x=8” to 14”, for ideal conditions. The different slopes suggest that factor(s) other than x become important as V_x decreases with distance.

The experimental results found substantial and statistically significant effects (p<0.001) for 3 levels of cross draft velocities (V_cross), the presence of a manikin, the presence of a work table, and all interactions of those factors. The effects were most pronounced at distances greater than 10” from the hood face. At that range, and V_cross=4 fpm, the presence of a table increased V_x by an average of 60% while the presence of a manikin decreased V_x by an average of 12%. At V_cross=50 fpm, the presence of a table increased V_x on average by 2.3 times V_x for ideal conditions. However, V _cross made little difference in V_x measured in the presence of a manikin.

None of the predictive models evaluated here accurately predicted V _x under the realistic conditions tested in this study. A satisfactory model will have to include terms for V_cross and the presence or absence of a work table and a worker.