Experimental verification of an energy consumption signal tool for operational decision support in an office building

Gregory S. Pavlak, Gregor P. Henze, Adam I. Hirsch, Anthony R. Florita, Robert H. Dodier

Research output: Contribution to journalArticle

Abstract

This paper demonstrates an energy signal tool to assess the system-level and whole-building energy use of an office building in downtown Denver, Colorado. The energy signal tool uses a traffic light visualization to alert a building operator to energy use which is substantially different from expected. The tool selects which light to display for a given energy end-use by comparing measured energy use to expected energy use, accounting for uncertainty. A red light is only displayed when a fault is likely enough, and abnormal operation costly enough, that taking action will yield the lowest cost result. While the theoretical advances and tool development were reported previously, it has only been tested using a basic building model and has not, until now, been experimentally verified. Expected energy use for the field demonstration is provided by a compact reduced-order representation of the Alliance Center, generated from a detailed DOE-2.2 energy model. Actual building energy consumption data is taken from the summer of 2014 for the office building immediately after a significant renovation project. The purpose of this paper is to demonstrate a first look at the building following its major renovation compared to the design intent. The tool indicated strong under-consumption in lighting and plug loads and strong over-consumption in HVAC energy consumption, which prompted several focused actions for follow-up investigation. In addition, this paper illustrates the application of Bayesian inference to the estimation of posterior parameter probability distributions to measured data. Practical discussion of the application is provided, along with additional findings from further investigating the significant difference between expected and actual energy consumption.

Original languageEnglish (US)
Pages (from-to)75-92
Number of pages18
JournalAutomation in Construction
Volume72
DOIs
StatePublished - Dec 1 2016

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Civil and Structural Engineering
  • Building and Construction

Cite this