Performance of deterministic workplace exposure assessment models for various contaminant source, air inlet, and...

Contaminant concentration estimates from simple models were compared with concentration fields obtained by computational fluid dynamic (CFD) simulations for various room and source configurations under steady-state conditions. Airflow and contaminant distributions in a 10 x 3 x 7-m room with a single contaminant source on a 1-m high table were simulated using CFD for steady, isothermal conditions. For a high wall jet inlet, simulations were performed for nine room air exhaust locations and eight source locations. For a ceiling diffuser inlet the impact of two exhaust locations and eight source locations were investigated. Because CFD treats determinants of contaminant transport explicitly and agreed well with experimental results, it was used as the standard for comparison. Parameters of the one- and two-zone completely mixed models (CM-1 and CM-2) and the uniform turbulent diffusivity model (UD) were determined from CFD simulation results. Concentration estimates from these were compared with CFD results in the breathing zone (BZ) plane (1.5 m above the floor) for the entire BZ, the source "near field," and the source "far field." In the near field the mean percentage difference between the model concentration estimates and the CFD results for all room configurations were -21.9, 32.3, and 126% for the CM-1, CM-2, and UD models, respectively, with standard deviations of 26.8, 111, and 103%. In the far field the mean percentage difference between the model estimates and CFD results were -4.8, -2.3, and -36.3%. The CM-1 model had generally the best performance for applications such as occupational epidemiology for the conditions and configurations studied. However, CM-1 tended to underestimate the near field concentration; thus, CM-2 was judged to be better in the near field when underestimation is undesirable, such as when determining compliance with occupational exposure limits. The agreement of CM-2 estimates with CFD results in the near field was more variable than that of the CM-1. The UD model performed poorly on average in both near and far fields, and the difficulty in accurately estimating the turbulent diffusivity presents a significant impediment to UD model use for exposure estimation.