Publication Type

Conference Proceeding Article

Version

publishedVersion

Publication Date

4-2015

Abstract

Simulation model calibration has been long identified as a key means of reconciling the consumption and efficiency characteristics of buildings. A key step in this process is the creation of the actual diversity factor profiles for occupancy and various energy end uses such as lighting, plug-loads, and HVAC. Creation of these model inputs is conventionally a tedious process of site surveys, interviews or temporary sensor installation. Sometimes measured energy data can be used to create these schedules, however there are many challenges, especially when the sensor network available is large or unorganized. This paper describes a process applying a series of knowledge discovery filters to screen data quality, weather sensitivity, and temporal breakouts from large nonresidential building performance datasets collected by building management and energy information systems (BMS/EIS). These screening techniques are used to qualify the desirability for calibrated model diversity schedule creation from a forensic perspective. A diurnal pattern filtering technique is then applied that automatically extracts frequent daily performance profiles, which can then be normalized and used as model inputs according to conventional industry techniques. The process is applied on a raw dataset of 389 power meter data streams collected for eight years from the EIS of a campus of 32 higher education buildings. The results are discussed in the context of time and effort savings for creating urban and building scale simulation model inputs.

Keywords

knowledge discovery, measured building performance, simulation feedback, diversity schedules, temporal data mining, visual analytics

Discipline

Environmental Design | Urban Studies

Research Areas

Integrative Research Areas

Publication

SimAUD '15: Proceedings of the Symposium on Simulation for Architecture & Urban Design, Alexandria, Virginia, April 12-15

First Page

136

Last Page

143

ISBN

9781510801042

Identifier

10.13140/RG.2.1.2286.0964

Publisher

ACM

City or Country

New York

Additional URL

https://doi.org/10.13140/RG.2.1.2286.0964

Download

Share

COinS