Development of CFD models and an automatic monitoring and decision support system for precision structural fumigation

Watcharapol Chayaprasert, Purdue University


Over many years, structural fumigation (i.e., flour mill, food processing or warehouse structures) has remained more of an art rather than a science. Its success depends largely on the knowledge and experience of the applicator. Even though it is generally known that the success of a fumigation is affected by a multitude of structural, environmental, and fumigation method factors, there is no data in the literature or used by industry that accurately describes the relationship of these factors and the dosage application rate. Because it is not practical to perfectly seal the structure, the fumigation process can be better optimized only if the dynamics of gas movement in the fumigated space and the effects of environmental conditions on the fumigation process are well understood. In this dissertation, the computational fluid dynamics (CFD) method was adopted to develop two structural fumigation models based on two reference flour mills. The models were validated with real world data and utilized for analyses and predictions of structural fumigation events. Several fumigation simulations were conducted. The simulations allowed evaluation of various "what if" scenarios and possible fumigation strategies such that fumigation applications can be custom-applied based on the prevailing site-specific conditions and target pests to control. The effect of multi-year weather conditions on the gas leakage rate (HLT) and the concentration*time (Ct) product during structural fumigation were evaluated. Eleven fumigation simulations were performed using historical weather data of the same time period between 1996 and 2006, assuming the fumigation practices including temporary sealing quality were maintained the same. Due to year-to-year variations in the weather conditions, the HLT varied more than 100% (from 10.7 to 23.3 hours), yielding a difference in the Ct product by more than 70% (from 476 to 840 g-hr/m3). This implied that prediction of HLT and thus fumigation performance should incorporate quantifiable sealing effectiveness and weather information for the planned fumigation period. Several fumigations with different circulation fan configurations were simulated and the uniformity of fumigant distribution was quantitatively evaluated. The simulation results showed that the number, flow capacity, location and orientation of circulation fans as well as the location of fumigant introduction had an effect on fumigant distribution. Based on these results, general guideline recommendations for proper sizing and placement of circulation fans were established. The application of the pressurization test and quadratic superposition method for prediction of structural fumigation performance were evaluated. Predictions of HLT and Ct product by the superposition method were compared to the respective values produced by fumigation simulations. For HLT less than 20 hours, the predicted HLT and Ct product were within ± 20 and ± 10% of the simulated values, respectively. This indicated that the pressurization test and superposition method have potential application benefits for optimizing structural fumigation. Another accomplishment of this dissertation was the development of an automatic monitoring and decision support system for precision fumigation. The automatic monitoring hardware was constructed based on technologies presently available to the fumigation industry. The recommendations provided by the decision support computer program were designed based on the findings from this dissertation to optimize fumigant gas usage. Although the regulation of dosage rate to maintain the desired gas concentration still had to be manually done by the fumigator, the monitoring and decision support system helped prevent over dosing, reduce error and risk from human mistakes, and increase the success rate of fumigation.




Ileleji, Purdue University.

Subject Area

Agricultural engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server