Dynamic integrated modelling for controlled environment agriculture: methodology and experiment

The continuously changing climate poses risks to traditional agriculture production and food distribution. Controlled environment agriculture (CEA) grows crops year-round under carefully regulated conditions that include light (spectrum and intensity), temperature, and humidity. Greenhouses, indoor vertical farms, and shipping containers are the most important CEA types, with greenhouses dominating the current market. Many indoor farming companies have financially failed. The energy-intensive nature and high carbon footprints of the industry make it difficult for CEA, especially indoor vertical farms, to be sustainable.

Currently, there is a lack of integrated modeling tools to support the design and operation of CEA facilities. In this study, we present a CEA modeling method that dynamically integrates such components as energy, evapotranspiration (ET), and crop growth. Interactive coupling between ET, crop growth, and energy-component models enables comprehensive evaluations of how indoor microclimate conditions influence CEA performance including energy, water, and crop yield. Importantly, the integrated modeling tool can be employed to understand the impacts of decisions on CEA performance at the early design stage.

Experimental data from growth chamber experiments were used to test and select crop growth and ET models. The experimental setup involves cultivating lettuce in a growth chamber under controlled environmental conditions. Data collected during these experiments serve as the benchmark for evaluating the predictive capabilities of the crop growth and ET models. We monitor key performance parameters, including plant dry and fresh weights, leaf area, and evapotranspiration.

In collaboration with industry partners, we conduct case studies comparing performance prediction from CEA models and measurement data trended by control systems in industry CEA facilities for greenhouses, indoor vertical farms, and shipping containers. We also present the agreement between measurement data and modeling results. Through these case studies, we share lessons learned for modeling CEA spaces and the opportunities to further improve the capabilities of dynamic modeling of CEA.