Development of an Artificial Intelligence Solution for Residential Heating and Cooling Equipment Sizing to Reduce Equipment Oversizing

The time and effort required to accurately size heating and cooling equipment inhibits HVAC contractors from aKempting load calculations. Instead, HVAC contractors oVen match the size of the equipment being replaced or use inaccurate rules of thumb. This results in many systems being oversized, increasing homeowner cost, and adding stress to the electrical grid. To simplify the process and increase the accuracy of residential heating and cooling equipment sizing RookStack proposes an artificial intelligence solution. Using publicly available data, the application will select the proper size heating and cooling equipment while requiring only a few simple inputs from the homeowner. In phase I RookStack will validate this approach by using machine learning on a housing stock dataset of nearly 7,000 homes and the energy code prescriptive requirements at the time of construction. The accuracy of the resulting models will then be assessed by Purdue University using data from smart thermostats installed in typical homes, which will support comparison of empirically measured heating and cooling loads to: (1) the current business-as-usual method of Manual J calculations as performed by HVAC contractors, and (2) the loads predicted by the models generated by RookStack. The hypothesis to be tested: are the heating and cooling loads predicted from RookStackĺs models using publicly available data and a small number of inputs from the homeowner more accurate than the current business-as-usual method of estimating these loads with a Manual J calculation by an HVAC contractor? The ground truth for this comparison will be measured heating and cooling loads. Electric utilities will derive significant value from technology that increases equipment sizing accuracy, because oversized HVAC equipment creates unnecessary incremental peak electrical demand. An equipment sizing process that homeowners can self-implement also simplifies the HVAC sales process. In-home HVAC sales consultations can be eliminated and e-commerce solutions for heating and cooling will be enabled, supporting the broad adoption of heat pumps.