UC Davis

Global population growth and economic development continue to intensify the demand for food production, placing immense pressure on irrigated agriculture, an essential source of sustenance for billions. Controlled environment agriculture (CEA) presents a promising alternative by enabling precise control over environmental conditions, thus improving crop yield and quality. Despite their promise, the practical feasibility and adaptability of closed-loop hydroponic systems remain constrained by various technical and operational challenges, particularly in achieving their full water and nutrient conservation potential. We hypothesize that the transition to closed-loop hydroponics in CEA could be strategically facilitated by aligning system design and precision management with conventional practices currently employed in open-field production. This dissertation explores CEA's potential to enhance the sustainability of food systems, addressing key pain points through a comprehensive engineering lens—from global-scale assessments to farm-level technological innovations.The first study (Chapter 2) employs a water–nutrient–salinity–energy nexus framework to evaluate the sustainability of transitioning from open-field cultivation to closed-loop hydroponic greenhouse systems. Closed-loop hydroponic systems significantly reduce water use per kg of lettuce by 6% in the UK and over 90% in Bangladesh and India, but this is offset by higher energy demands, ranging from 48 MJ/kg (UK) to 66 MJ/kg (India). Focusing on global lettuce production, this analysis reveals significant improvements in resource efficiency and reductions in environmental impact with greenhouse hydroponic production compared with open fields, especially in water-scarce regions. However, the elevated energy demands of greenhouses remain a critical trade-off, necessitating integrated solutions and policy considerations to support this transition.Although hydroponic systems significantly improve resource use efficiency, especially water and nutrients, many commercial operations only partially recirculate nutrient solutions or periodically discharge them, undermining the system's sustainability potential. Achieving fully closed-loop hydroponics is obstructed by two primary barriers: i) sensor inaccuracy (Chapter 3) and ii) nutrient imbalance (Chapter 4). The second study (Chapter 3) investigates operational challenges at the farm scale, particularly the reliability of electrical conductivity (EC) and pH sensors used in hydroponic nutrient solution management. Faulty sensor readings can compromise the precision of fertigation strategies. Leveraging advancements in artificial intelligence, this chapter evaluates the performance of five machine learning models (ANN, RF, SVM, LSTM, and KNN) for fault detection and diagnosis in EC and pH sensors. The five types of synthetic faults (bias, precision, drift, stuck, and physical) were induced into the experimental data from an indoor hydroponic system. RF performs better with accuracy in fault detection, achieving 93.7% for EC and 96.5% for pH sensors. In fault diagnosis, RF also consistently performs better than other models for the EC sensor, with accuracies up to 100% for spike and 97.1% for bias. For the pH sensor, performance varies by fault type: SVM excelled in bias with 100% accuracy, while RF shows the highest accuracy for normal, drift, and precision degradation, reaching up to 92.1%. This chapter also focused on sensor-independent fault detection for EC and pH sensors using environmental data, water volume, and dosing events. The results underscore the effectiveness of sensor-based and sensor-independent approaches to enhance the sustainability of hydroponic production through precision nutrient monitoring and management.The third study (Chapter 4) advances toward closed-loop nutrient management by shifting from traditional EC-based control to individualized nutrient dosing. This chapter introduces a novel Model Predictive Control (MPC) framework designed to regulate six essential macronutrients: Nitrogen (N), Phosphorus (P), Potassium (K), Calcium (Ca), Magnesium (Mg), and Sulfur (S). This novel implementation of MPC enables precise, dynamic regulation of six selected fertilizers and water inputs, ensuring that all macronutrient concentrations are consistently maintained within agronomically recommended thresholds. Simulations across five scenarios showed that individualized dosing improved water use efficiency (5.9 to 34 g/L) and fertilizer use efficiency (6.2 to 12.5 g/g) while reducing input costs—water (0.32 to 0.08 USD/kg) and fertilizer (1.78 to 0.88 USD/kg).The outcomes from this research provide critical information on the water-food-energy-sanity nexus for sustainable lettuce production. The outcomes will offer critical insights to various stakeholders about sustainability dynamics and challenges globally. The developed algorithms for fault detection and diagnosis, as well as precision dosing, will provide new insight for improving the sustainability of hydroponic production with precision monitoring and management. Although this research is limited by the absence of an economic feasibility assessment, the findings provide a foundation for the future integration of AI-based decision support tools and adaptive control technologies in commercial hydroponic operations.