Prediction of Renewable Energy Needs for Off-Grid Greenhouse Tomato Production

Abstract

Food insecurity is a widespread issue within communities in remote areas such as the Navajo Nation, and a leading cause of chronic health issues such as heart disease and diabetes. Development of broad, resilient food production networks can be possible by utilizing Controlled Environment Agriculture (CEA) to achieve year-round, high-yielding, local crop production. As a CEA system, greenhouses can provide these much needed yields by creating optimal growing conditions, through harnessing sunlight for heat and light, and through utilizing hydroponics to sustain crops using only a fraction of the water needed for field agriculture. However, greenhouse production in remote areas can be limited by access to water and energy. In particular, one third of homes on the Navajo Nation lack access to a grid connection, and 30% without running water (de Sam Lazaro, 2018; Navajo Safe Water, 2021). For CEA systems to be successful in remote areas such as the Navajo Nation, energy demand must be predicted to accurately size renewable energy systems for to maintain environmental conditions for crop production needs. However, there is a lack of tools that provide outputs that are readily integrated with software to size renewable energy systems for off-grid greenhouse crop production. The goal of this study is to develop a steady-state model to predict greenhouse resource consumption using minimal user inputs with an excel based workbench and to size renewable energy system components for an off-grid greenhouse using HOMER software. Inputs include hourly weather data, production season dates, environmental setpoints required for crop needs, as well as greenhouse dimensions and structural materials. The model was calibrated against energy and water consumption data from two greenhouses located at the Controlled Environment Agriculture Center at the University of Arizona, Tucson, AZ. Normalized RMSE values for the acrylic greenhouse validation were 26.7% for electricity, 50.5% for wet pad water consumption, 17.8% for transpiration, and 15.8% for total water consumption. For the less sealed, minimally controlled PE greenhouse, normalized RMSE values were 54.7% for electricity, 52.8% for wet pad water, 15.7% for irrigation water, and 38.5% for total water consumption. These results show the model can predict energy, transpiration, and total water consumption of a controlled greenhouse, and that prediction errors increase with decreased control over the greenhouse environment (ie. less sealed, roll-up side vents). The calibrated model was used to predict resource consumption of greenhouses considering three greenhouse sizes with single (100 m2), double (200 m2) and triple bays (300 m2) intended for use in the Navajo Nation. For single, double, and triple bay greenhouse cases, total energy and water consumption per unit floor area decreased proportionally with the addition of bays. Water consumption per unit area decreased by 4.5% from 20.84 to 19.90 L m−2 day−1with the addition of the second bay, and decreased 2.2% to 19.45 L m−2 day−1 with the addition of the third bay. Meanwhile, total energy consumption per unit area decreased by 17.2% from 0.072 to 0.060 kWh m−2 day−1with the addition of the second bay and decreased 9.0% to 0.054 kWh m−2 day−1with the addition of the third bay. HOMER Pro microgrid software was used to size energy systems for each greenhouse layout. The major components considered were PV solar panels, a battery storage system, and a 3.5 kW propane generator. For the single, double, and triple bay cases, the optimal PV capacities were found to be 2.16, 2.88, and 3.96 kW, respectively. Meanwhile, the optimal storage system for each respective case required 14, 18, and 22 12-Volt batteries. The cost of energy is comparable for each system, at 0.463, 0.464, and 0.461 $ kWh−1 for the single, double, and triple bay greenhouse cases. The net present cost of the energy systems per floor area decreases with the addition of each bay, indicating only a 66.2% increase in the net present cost with the addition of the second bay, and an additional 35.3% increase in net present cost with the addition of the third bay. For each layout, the 3.5 kW backup generator was selected in the optimal system, with total renewable energy fractions of 96.9% for the single, 82.7% for the double, and 81.7% for the triple bay case. Excluding the backup generation from the analysis leads to an NPC increase of 4.5% for the single bay, 11.9% for the double bay, and 21.3% for the triple bay cases. The cost savings from incorporating backup generation in off-grid PV systems increases with the scale of greenhouse production. Thus, back-up generation can be used to promote viability of off-grid greenhouse production in remote communities.