Develop Semi-Automated, Component-Level Diagnostic Procedures

Goal Statement:
Implement and assess component-level functional testing and performance-monitoring procedures for heating, ventilation, and air conditioning (HVAC) systems.

We use models to predict the performance of particular HVAC equipment. A significant difference between predicted and observed performance indicates the a fault. Relevant operating data are then displayed, so the operator can confirm the fault and take steps to diagnose its cause so that it can be remedied.

Results:
Models have been implemented for the following air-handling unit components and chillers:

• Air handler mixing box
• Heating and cooling coils, including control valves
• Variable air volume (VAV) fan system
• Gordon-Ng chiller (428.3 KB, 10 pp)

We have developed a toolbox of software procedures to support component-level, functional testing and performance monitoring. These include a:

• Transient analyzer
• Comparator
• Parameter estimator

We tested and refined the models and supporting procedures using off-line commercial building data and the Iowa Energy Center test facility. Currently, on-line implementation and testing are under way at two locations:

450 Golden Gate
This building is a large federal office in San Francisco CA, managed by the Government Services Administration (GSA). After collecting design information and building the off-line models, we coordinated with site management to reconfigure the energy management system so that it monitored the data points we need for the Fault Detection Diagnostics (FDD) model. Sufficient data have now been collected, and testing has begun.

925 L Street
FDD tools are also being implemented at 925 L Street, Sacramento CA. We have developed a plan for off-line and on-line testing of both component-level and whole-building fault detection methods that will capitalize on the installation of a high-quality monitoring system [the Information and Monitoring Diagnostics System (IMDS)] from a previous project.

Downloads and Links:

• "Field Testing of Component-Level Model-Based Fault Detection Methods for Mixing Boxes and VAV Fan Systems", P. Xu and P. Haves. Presented at the ACEEE 2002 Summer Study on Energy Efficiency in Buildings and published in the Proceedings. May 2002. (1.62 MB, 13 pp)

Contact: Philip Haves, Lawrence Berkeley National Laboratory (LBNL), (510) 486-6512