Miller Magazine Issue: 113 May 2019

70 ARTICLE MAY 2019 • the degree to which the grain already has been inva- ded by fungi before it arrives at a given site • presence of insects and mites All these factors interact with one another to some ex- tent, but the major determinants are moisture content, re- lative humidity, temperature and time [3]. Among the fun- gal species that may contaminate grain, some of them will be able to produce mycotoxins. Some of the molds and their associated mycotoxins that are currently considered to be of worldwide importance are shown in Table 1. THE CHALLENGE The scientific literature [5] suggests that a management strategy of prevention of mold spoilage and mycotoxin contamination of sto- red grain should consists of the following steps: • Identification of critical storage situations enabling mold growth • Monitoring early signs of fungal activity • Preventive measures and anticipation of mycotoxigenic mold growth • Practical solutions for the reduction of exis- ting contamination (by mold and/or mycotoxins) PREDICTING MOLD DEVELOPMENT In view of the above, in-depth knowledge and un- derstanding of silo microcli- mate is crucial. An efficient method for tackling this is through the combination of field measurements and com- puter simulation based on Computational Fluid Dynamics (CFD) models. CFD is a bran- ch of fluid mechanics that uses numerical analysis and data structures to solve and analy- ze problems that involve fluid flows and heat transfer. Fast computers (typically on the cloud) are used to perform the calculations required to simu- late the interaction of liquids and gases with surfaces defi- ned by boundary conditions. To evaluate accurately the storage structure interaction with its surroundings, the computational model should integrate weather forecast for the specific location and time period. Additionally, the algorithm process temperature and Figure 2: Time variation of weather conditions at the silo site: solar radiation, temperature, relative humidity, atmospheric pressure and wind velocity Figure 1: Example of wireless sensor installation inside a silo feeding the predictive algorithm with real-ti- me data