PhD Oral Exam - Danlin Hou, Building Engineering
A New Bayesian Inference Calibration Platform for Building Energy and Environment Predictions
This event is free
School of Graduate Studies
When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.
Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.
Buildings consume nearly 40% of total global energy consumption. It is predicted that the combined energy consumptions from the residential and commercial sections worldwide will increase to 22% in 2050 of the world's total delivered energy. Meanwhile, requirements for indoor health, safety, thermal comfort, and air quality have been demanding due to more intensive and frequent extreme climate events, such as heatwaves and cold waves. The issues become pretty challenging for the building energy and environment field, especially during the COVID-19 pandemic.
Computer simulations play a crucial role in achieving a safe, healthy, comfortable, and sustainable indoor environment. As an integral step of the building model development, model calibration can significantly impact simulation results, model accuracy, and model-based decision-making. However, conventional calibration methods are often deterministic. As a result, the uncertainties which have been investigated for a building computer model, and those from the inputs have not been given adequate attention and thus are worth in-depth studies.
Bayesian Inference is one of the best approaches to calibrate computer models with uncertainties. Several studies explored its application to building energy modeling, but a comprehensive application to the general field of building energy and environment has not been adequate. This thesis first started with a comprehensive literature review about the Bayesian Inference calibration focusing on building energy modeling. Then a systematic Bayesian calibration workflow and a new platform were developed. Besides the general studies of its application to building energy models prediction, the thesis investigated how to use the platform to calibrate building thermal models and indoor air quality models. To solve the issue of the computing cost of Bayesian Inference, another calibration and prediction method, Ensemble Kalman Filter (EnKF), was proposed and applied to ventilation performance estimation and free cooling load predictions. Finally, the integration of EnKF and Bayesian Inference calibration to combine their benefits and avoid the weaknesses is proposed for future study.