Date of Award

Spring 4-23-2015

Document Type


Degree Name

Master of Science (MS)

First Advisor

Gregor Henze

Second Advisor

Adam Hirsch

Third Advisor

Balaji Rajagopalan

Fourth Advisor

Moncef Krarti


There is a growing amount of data being collected from commercial buildings that quantifies their energy use. Very few facilities managers have the means to transform this data to determine how the building is performing compared to expectations. Performance benchmarking of buildings can be used to draw this comparison, but first a benchmark must be established. Many tools exist for equipment fault detection diagnostics (FDD), but FDD alarms do not indicate fault severity. Other tools exist for rapid peer-benchmarking of energy performance, but statistically meaningful comparison groups are small. An Energy Signal Tool is proposed as a way to self-benchmark performance and quantify fault severity. Being able to quantify performance of buildings in a portfolio can help an energy manager prioritize operational changes or maintenance based on which facilities need attention most.

In this work, detailed building energy simulation modeling of a retail store is carried out with the OpenStudio / EnergyPlus software platform. Uncertainty analysis is used to enhance decision support with a probabilistic approach to energy consumption risk management. Expected model parameter distributions are characterized, and global sensitivity analysis is performed to quantify parameter significance. Latin Hypercube Sampling (LHS) batch simulation is used to sample from significant parameter distributions and generate expected ranges of energy consumption for four major building energy end-uses. The results of batch sampling form a probabilistic range that is used as the performance benchmark.

This work then goes on to demonstrate how sub-metered data can be put into the context of these expected ranges and transformed into plain output for self-benchmarking and energy management decision support with utility theory. The refined concepts of the Energy Signal Tool were tested synthetically in ten different fault scenarios across three climate zones. The results of the testing were processed to illustrate the fault sensitivity of the tool, and to demonstrate how such a tool could be applied to prioritizing actions across a portfolio of buildings. This work builds upon the concepts for an “Energy Signal Tool” originally proposed by Henze et al., (2015), with the goal of making the tool suitable for industry application.