The Main UT campus has been modeled with the help of the open street map and UT architectural drawings in the Rhino Modeling tool. Buildings have been categorized based on their construction year, and they are color-coded. Material properties have been decided based on the construction year. Also, we calculated window to wall ratio for each building by architectural drawings. When we did not have access to the drawings, we decided on the WWR with the help of google maps/earth. We have also included some surrounding buildings as shading surfaces.
We can run the model and obtain cooling and heating consumption analysis for the UT campus; however, this model must be calibrated using real-time consumption data. The 113 buildings have been included in the 3D modeling, but the hourly measured data is available only for 85 buildings. Prior to the Bayesian calibration, our research included sensitivity analysis in deciding the most influential input parameters on heating and cooling demand. We performed Sobol Sensitivity Analysis on a cubic building with 60m *60m Floor area and 24 m height by positioning in the center of the ground surface. As a result of our sensitivity analysis performed by the Python script and external library- SALib13, we identified the key parameters as infiltration, minimum temperature setpoint, and maximum temperature setpoint variables to perform Bayesian calibration.
Python programming basics and machine learning (scikit-learn) is a must
Knowledge on Rhino/Grasshopper is advantageous
Knowledge on building energy performance is not necessary but useful/helpful
immediate start, summer and fall semester
We need to calibrate the CitySIM model by looking at the real-time data for discussed key parameters. For heating consumption, we need to calibrate the buildings for Tmin (minimum setpoint temperature) and infiltration, whereas, for cooling consumption, we need to calibrate buildings for infiltration and Tmax (maximum setpoint temperature). The sensitivity results show that window to wall ratio also has one of the critical parameters for heating consumption; however, it is not required to calibrate it since we already have provided realistic proportions from drawings.
