ON STORED GRAIN MONITORING AND MANAGEMENT SYSTEMS

Stored grain monitoring technologies have proven to be a good management tool for grain storage. Despite the importance of stored grain monitoring technologies, there is still a low adoption of these technologies among grain farmers and elevators. More so, understanding data measured and displayed on the dashboard of stored grain monitoring technologies with little or no actionable recommendations remains a challenge in the use of these systems by grain farmers and elevators. There is a need to improve the diagnostics of stored grain conditions using measured data from sensors in grain storage bins to enable better management decisions by grain farmers and elevators.

The review of the state and emerging stored grain monitoring technologies and management systems were investigated based on a systematic literature review of five academic databases, and the identification of common agricultural technology (agtech) companies producing stored grain monitoring and management technologies on the market today. Results showed a gradual rise in agtech companies focusing on stored grain monitoring technologies. Twenty-one (21) agtech companies were identified, with all not monitoring grain quality in relation to loss in financial value.

Monitoring the quality of stored grain using carbon dioxide (CO₂) sensors at the headspace in a bin has been used for early warning when grain is going out of condition. However, monitoring CO₂ as a long-term stored grain management tool has not been thoroughly investigated. Additionally, the location of the CO₂ sensor in other regions of the grain bin, apart from the bin headspace has also not been thoroughly investigated. CO₂ levels were monitored in the headspace and plenum of eight (8) 500 bu (12.7 t) pilot bin with 350 bu (8.89 t) of shelled yellow corn for one year. There were similar patterns and trends in CO₂ concentrations in the headspace and plenum. While analysis of variance (ANOVA) showed a significant difference (p < 0.05) in CO₂ concentrations in the headspace and plenum in fall, spring, summer, and winter, had mean value differences in the range of 9 to 126 ppm, which is not of consequence in practical terms. Based on the small seasonal differences and similar patterns of CO₂ concentrations in the headspace and plenum, both headspace and plenum (bins with a full plenum) are two options for monitoring the stored grain condition in a bin. Additionally, insect population from probe traps, flight traps and corn samples increased mostly in the warm summer to fall storage period and corresponded with increased grain temperatures and headspace and plenum CO₂ levels.

An analytical approach was developed to estimate the financial value of stored corn in a bin. The financial value model employed a machine learning model using corn dry matter loss (DML) according to Steele’s 0.5% DML maximum threshold, and elevator buyer’s discount for corn. Gradient Boost gave the best predictive model to predict the stored corn DML, with a coefficient of determination (R²) value of 0.9770. The developed approach to estimate the financial value of stored grain in a bin was demonstrated by a sensitivity test, which showed that losses were incurred after about one year of storage due to DML and buyer’s discount based on corn quality at point time of sale. Tracking grain quality based on buyer’s discount of stored grain could incentivize better management of stored grain than either the grain value which accounts for only losses due to DML or the total financial value which accounts for DML and buyer’s discounts, but can be masked by market price appreciation.

In the survey on adoption of stored grain monitoring technologies among grain farmers and elevators, age was a significant factor among other factors affecting the adoption of these technologies, as shown from the binary logistic regression model. Younger farmers aged 25-34 years were more likely to adopt stored grain monitoring technologies than those not in the age range. The respondents (51%) adopted at least one form of stored grain monitoring technologies. Some challenges and limitations facing the adoption of stored grain monitoring technologies were the availability of capital, knowledge on how to use the technology, cost of technology, user experiences, and value derivation from collected data.

A framework for a decision support system (DSS) with actionable decision support (ADS) for stored grain management was developed to show how measured data outcomes of stored grain monitoring technologies can be improved. Three major components of the framework include the knowledge base, inference engine and communication. The developed framework was demonstrated, and the results showed that the application of the ADS framework in addressing a challenge identified in the case study was achieved. Artificial intelligence models are the most feasible approach to the development of a robust ADS systems for stored grain management. The use and integration of machine learning empowered by predictive AI models was not explored in this dissertation research. This is the most feasible approach to the development of robust ADS systems for stored grain management and should be the focus of future research efforts. The outcomes resulting from expert advice generated from data of the stored grain condition could be used to improve the ADS system using machine learning and predictive AI.

The experimental results from this dissertation work, analytical approach, and model for the estimation of the financial value of stored corn in a bin will be beneficial for agtech companies to further improve the development of stored grain monitoring technologies. The overall outcome of this study will help to drive the improvement in the management of stored grain by grain farmers, elevators, and operation managers. To increase the adoption of stored grain monitoring technologies, there is a need to educate and train farmers on stored grain monitoring technologies and the benefits of their use, while agtech companies need to tackle the challenges and limitations of current stored grain management technologies revealed from this study.