Humidity-Aware Model Predictive Control for Residential Air Conditioning: A Field Study

TL;DR

The paper addresses the challenge of incorporating indoor humidity dynamics into model predictive control for residential air conditioning, demonstrating a humidity-aware MPC in a real house with two humidity modeling approaches: latent time-varying SHR and a simpler sensible model.A low-order 2R1C thermal circuit together with data-driven humidity predictors (including SHR as a function of indoor wet-bulb temperature) enables tractable open-loop MPC formulations solved with CVX for either cost-minimization or power-limiting objectives.Field results over 38 MPC days show similar energy savings for both humidity models, with the latent model providing markedly better adherence to a 4 PM–8 PM power constraint and reduced peak violations, while comfort remains acceptable (PPD ~7.2%).Estimated annual cooling/heating cost savings reach about 419 USD (27%), supporting the practical value of humidity-aware MPC, though generalization across climates and housing types requires further study.

Abstract

Model predictive control of residential air conditioning could reduce energy costs and greenhouse gas emissions while maintaining or improving occupants' thermal comfort. However, most approaches to predictive air conditioning control either do not model indoor humidity or treat it as constant. This simplification stems from challenges with modeling indoor humidity dynamics, particularly the high-order, nonlinear equations that govern heat and mass transfer between the air conditioner's evaporator coil and the indoor air. This paper develops a machine-learning approach to modeling indoor humidity dynamics that is suitable for real-world deployment at scale. This study then investigates the value of humidity modeling in four field tests of predictive control in an occupied house. The four field tests evaluate two different building models: One with constant humidity and one with time-varying humidity. Each modeling approach is tested in two different predictive controllers: One that focuses on reducing energy costs and one that focuses on constraining electric power below a utility-specified threshold. The two models lead to similar performance for reducing energy costs. Combining the results of this study and a prior heating study of the same house, the estimated year-round energy cost savings were $340-497 or 22-31% (95% confidence intervals); these savings were consistent across both humidity models. However, in the demand response tests, the simplifying assumption of constant humidity led to far more frequent and severe violations of the power constraint. These results suggest that accurate building models are important for nonlinear objectives, such as reducing or constraining peak demand, while for linear objectives such as reducing energy costs or emissions, model accuracy is less important.

Figures (17)

• Figure 1: Flow-chart of the supervisory control system. Weather information (temperature, humidity, solar irradiance, wind speed) is obtained from a weather service and sent to the controller alongside disturbance predictions and real-time building measurements. The selected indoor air temperature set-point is sent to the smart thermostat and implemented in the house.
• Figure 2: The DC Nanogrid House is a 208 m$^2$, 1920s-era house with all-electric appliances in West Lafayette, USA.
• Figure 3: The workflow used in this paper to model and control the building and equipment.
• Figure 4: A 2R1C thermal circuit model.
• Figure 5: Indoor temperature prediction errors (left) and air conditioner cooling rate predictions and measurements (right). Predictions match measurements reasonably well in both training (first month of data) and validation (last ten days).