Development of Data Quality Control Limits for Data Screening through the "Energy Balance" Method

Abstract

Energy management processes such as accounting energy cost, finding overuse in energy, and determining savings from energy conservation programs largely depends on measured energy use data. Identifying and correcting faulty data properly would avoid over/underestimation in energy use and increase accuracy in further analysis. It allows engineers and administrators to make more confident and low-risk decisions. This paper proposes a methodology to construct statistical control limits for data screening using “Energy Balance” methodology (Shao and Claridge, 2005). Energy Balance (EBL) parameter represents quasi-steady state thermal energy storage in a building and indicates a predominant linear behavior when it is plotted versus the outside air temperature. A regression model of EBL parameter developed as a function of the outside air temperature from a long-term data can be used as a data screening tool for newly measured energy use in the building. However, EBL model is known to have functional discontinuities called “change points” and non-uniform residuals. To construct control limits that fit the EBL data uniformly over a wide range of outside air temperature, a new technique was introduced which estimate mean square error (MSE) for a change point model as a function of outside air temperature by using Bin-MSE data. This methodology has successfully constructed data quality control limits of EBL parameter for example buildings. The numerical criteria developed by this methodology would help to have uniform results in quality control for energy use data, and it would be used as a rule for an automated data screening process.