Validation of simulated mean air temperature and mean radiant temperature using monitored data from an experimental test cell.

Abstract

Building performance simulation provides a technique for the prediction of the performance of buildings at the design stage, and retrofit strategies for existing buildings. No matter which software is used, a Validation of the simulation model is necessary to reduce discrepancy between predicted and actual performance. Adopting simulation and experimental approaches, this paper presents the procedures involved in the Validation of a simulated model. An experimental test cell was designed; its construction was based on carefully selected parameters (site location, climatic conditions, thermo-physical properties, geometrical descriptions and air flow systems of the optimized model); and monitored to obtain data to calibrate the simulated model. The simulated hourly mean air temperature and mean radiant temperature values were compared with their respective monitored data using graphical presentations of time-series plots.

Moreover, statistical analysis involving Coefficient of Variation for the Root Mean Squared Error (CV (RMSE)); Root Mean Square Difference (r2) values were computed; and checked to fall within allowable tolerances. Obtained values included CV (RMSE) of 14.75% (indoor air temperature); 16.80%

(mean radiant temperature); r2 of 0.8% (indoor air temperature); and 0.83% (mean radiant temperature). The results confirm that the simulation model is calibrated, and can be used within the context of the

experimental test cells to extrapolate performance to other contexts by means of scaling and replication procedures.