UC Berkeley

Statistical Process control is a technique often used in manufacturing in order to determine whether a process is changing based upon most recent measurements. In this thesis, I will cover how this is applicable to energy efficiency in buildings, and construct the framework that allows for this viewpoint to be effective in understanding how buildings consume energy. I will discuss how to effectively baseline a buildings energy performance, predict that performance in basic and advanced ways, and then use that prediction to determine when building parameters are out of line. I outline a methodology for determining when building operation parameters have changed, attempt to identify them, and suggest potential problems, given a top-down monitoring methodology combined with a bottom-up fault diagnostics approach. The approach explored in this talk will combine Shewhart and other control charts with modern monitoring techniques in order to explore the efficacy of this approach. I conclude by developing new modifications on the idea of control charts and by using those modifications to produce an analysis of the campus energy use at the University of California, Berkeley.