P2-5. Pattern Recognition-Based FDD > Outcomes

Technical Outcomes:

• Chillers, boilers, and pumps (circulation and cooling towers) were selected for development of FDD methods. At the beginning of the project, the research team decided to limit the systems and components investigated to those not covered in other PIER FDD projects.
  
  
• Analysis of how an expert analyzes the plots of data for chillers and boilers led to adoption of rule-based methods for detecting faults associated with chillers and boilers. Research on general pattern recognition techniques was documented in an excellent summary report on methods that could be applied to building energy fault detection and diagnostics.
  
  
• A spreadsheet with a VBA graphical user interface was developed to implement a small set of FDD algorithms to test and illustrate the concept of automation. The interface was not developed to the level originally anticipated because of an unexpected match-funding shortfall in the second year of the project.
  
• A prototype diagnostic software tool was developed for chiller diagnostics. The full set of diagnostic algorithms was not implemented to the level originally anticipated because of an unexpected match-funding shortfall in the second year of the project and greater than expected level of effort required to capture and document the underlying diagnostics.
  
  
• Testing was performed using real data collected from buildings in pre-program projects. The match-funding shortfall also resulted in eliminating the field testing of the FDD tool.
  
  
• A fully documented software specification was produced that documented the diagnostic algorithms for all selected building components (chillers, boilers, pumps, and cooling towers) as well as the software actually implemented. This report, along with an economic impact letter report, will guide future efforts to more completely develop an FDD tool. The match-funding shortfall also curtailed full development of the software.

Market Outcomes:

Automation of diagnostic techniques that are currently applied manually will increase the number of facilities that use diagnostics as part of routine maintenance. Large office buildings, which tend to have building automation controls in place, represent over 1,000,000 SF of floor area in California's commercial building stock and are a primary target for diagnostic software. Development of automated diagnostic tools for a spectrum of systems and equipment would respond to one of the needs expressed by building managers and operators participating in Project 2.4.

Incorporating Project 2.5 diagnostic methods into a web based diagnostic module is the goal of a follow-on project. Trial testing of this module should be started by the fall of 2004.

Back to Previous Page