Development and validation of computational fluid dynamics models for precision structural fumigation

Three-dimensional Computational Fluid Dynamics models for structural fumigation were developed and validated using Fluent((R)) based upon comprehensive data sets collected during the fumigation of a commercial flour mill. The external flow model, which included the flour mill and surrounding structures, was used to predict stagnation pressures on the mill's walls as a function of the wind speed and direction data. The pressure differences due to density differences between the gas inside and the air outside the mill (stack effect) were estimated using the environmental temperature and relative humidity data. The combined effect of the stagnation pressure and the stack effect was used as the boundary conditions of the internal flow model. The internal flow model incorporated interior details of the mill such as building plans, locations of major equipment, partitions and ducting. Because it was not possible to obtain the actual number and sizes of the cracks in the structure envelope, the idea of representing the cracks as effective leakage zones (ELZ) was adopted. The flow resistance coefficient, k(L), of the ELZs determines the gas tightness of the mill. Nine simulations were conducted with different k(L) values. Both experimental and simulation concentration data indicated that the fumigant was uniformly distributed within the entire mill building. Using a manual optimization approach, one specific kL value was determined for which the models were able to yield a half-loss time (HLT) value identical to the experimental HLT (17h) and minimize the prediction of the concentration x time (Ct) product to within 10.5% of the observed value. Therefore, it was concluded that the models were validated and the assumption of ELZ was reasonable. The modeling methodology established in this paper could be utilized for the prediction of fumigation performance in any type of structure. (c) 2007 Elsevier Ltd. All rights reserved.