Buildings generally do not perform as well in practice as anticipated during the design stage. There are many reasons for this, including improper equipment selection and installation errors, the lack of rigorous commissioning and proper maintenance, and poor feedback on ongoing performance, including energy performance. These problems are prevalent in most building systems, and frequently found in dynamic systems such as heating, ventilation, cooling, and lighting controls.

This project was conceived to develop and introduce state-of-the-art information technology in buildings in order to substantially enhance building energy performance by continuously improving operation and maintenance (O&M). The project is being conducted by an interdisciplinary team to assess the current state of technology, develop an appropriate information and diagnosis capability, and test it in real buildings.

This paper summarizes the key results from the first phase of project (Sebald and Piette, 1997). The scope of the Phase 1 effort included:

• Assessing common O&M problems and the needs of Class A office building owners and property managers to address these problems.
• Evaluating state-of-the-art building performance monitoring and diagnostic information technologies, including coordination with research involved in other promising R&D efforts.
• Assessing the capabilities of building diagnostic technologies and related information support services, including identifying current and future suppliers of these technologies and services.
• Designing an engineering prototype of a building diagnostic system that will meet the needs of building owners and property managers.

Early in the project we decided to focus on large Class A commercial office buildings. These buildings are managed by large property management companies and we think there are potential "early adopters" among these companies. The Phase 1 market assessment activities included in-depth interviews with six technical managers who had been identified as the most sophisticated in California. These interviews included a review of their perceptions of operations and maintenance problems with all major building systems, including controls. We found it difficult to identify a single system or component that was most problematic. Rather, there are systemic problems associated with the lack of feedback available from current Energy Management and Control Systems (EMCS). Today's EMCSs are designed for control, with extremely limited capabilities in sensing, archiving, data analysis, diagnostics, and data visualization.

In the first phase of the project we concluded that significant benefits could be derived by large commercial and institutional building owners¹ and operators from developing and demonstrating the use of a continuous performance tracking and diagnostic system. We produced a conceptual design for a diagnostic system that would provide continual feedback on building performance and assist operators in diagnosing operational problems. The performance data and data visualization techniques developed were designed to report on common failure modes for specific building systems discussed below. The data visualization techniques were designed to be effective computer-based graphics that display the most important metrics to describe system performance. These graphics were developed from basic engineering principles for the whole building and the water side of the cooling system. The system also includes automated data collection and analysis software that prepare the data for the visualization system. Together, the data collection, analysis, and visualization systems should serve as a robust and reliable platform for an integrated and automated diagnostic system. The evolution toward automating the diagnoses will be include in the Phase 2 research efforts, which began in mid 1997. The diagnostic system should be useful allocating operations and maintenance (O&M) resources over the short run and capital investments over the long run.

The proposed diagnostic system differs from previously developed systems in several important ways. First, it is specifically targeted toward sophisticated building operators and engineers. Most related research efforts or related techniques are targeted toward a remote expert user. Second, the proposed system is designed to be permanently installed. Related approaches that are known for being easy to use are built around short-term rather than continuous monitoring systems. Third, the monitoring system is based on laboratory-quality sensors that are far more reliable than sensors found in most commercial building systems. Fourth, the proposed system continuously archives data on approximately 80 points once each minute. Most current systems do so every 15 minutes or longer. Fifth, the diagnostic system has a top-down design that logically flows from the general whole-building analysis to system and component diagnostics. This is in contrast to bottom-up approaches that attempt to detect performance failures associated with specific individual devices.

These five features, (1) sophisticated building operators and engineers as users, (2) permanent installation, (3) laboratory-quality sensing, (4) short-interval data archives, and (5) top-down design, are critical to our proposed system. Much of this report is devoted to explaining how these departures from previous approaches increase the probability that the proposed system will perform well. As an initial test of the response to our ideas, we visited additional buildings and held two technical seminars with the managers to assess their reaction to the proposed diagnostic system. Approximately one dozen building managers have formed the core of our contact base; they have all expressed strong support for our approach and expressed an interest in participating in demonstration projects.

There are difficult trade-offs between advancing the automation of the diagnostic systems versus designing the system for optimal human-based diagnostics. The development of automated diagnostics can be justified by the recognition that building systems are becoming more complex over time and are difficult for the average operator to understand (Hyvarinen, and Karki, 1996). We have chosen to work with the most sophisticated operators we can find. But, we will explore how to automate some of the "expert diagnosis" so that the system could be developed for a broader set of users. Automated diagnostics consist of methods to detect faults and may include identification of fault sources. Most diagnostic systems include a general user interface to aid the human user in getting information about the detection of a fault. Automated diagnostic systems generally include model-based (e.g., simple functions, physical, or black-box) fault detection and classifiers (knowledge or association based).

The demonstration in Phase 2 is oriented toward ensuring that the diagnostic system is cost-effective. Average reductions in whole-building energy savings of 15% have been demonstrated by compiling and assessing whole-building and major energy end-use data (Herzog and Lavine, 1992). Savings from 20 to 40% have been demonstrated in some cases (Claridge et al, 1994). With average whole-building energy costs (gas and electricity) of about $2/ft²-year, a 15% reduction translates to a savings of $0.30/ft²-year. This amounts to $150,000/year for a 0.5 million ft² building, which is the size building the demonstration projects are oriented toward. We anticipate that mature market costs for the diagnostic system will be under $75,000 in hardware and software. Thus, the proposed system will be designed to be cost-effective for large buildings. It is worthwhile mentioning that by many accounts, these savings estimates are conservative.

Looking statewide, the target technical potential energy savings is a fifteen percent reduction in large office building energy use for California. Large office buildings (those greater than 50,000 sqft) in California consume about 24 BkWh (site electricity) and 32 B kBtu (gas). Fifteen percent savings in office energy use translates into 41.8 *10¹² Btu (source)². There are about 11,000 large offices in California, which at about 100,000 sqft each, total 1.1 Bsqft of office space. With average energy intensities of about 22 kWh/sqft-year for electricity and 31 kBtu/sqft-year for gas requiring about $240,000 per year, a fifteen percent savings translates into about $35,000/year (Akbari et al, 1993 and Akbari et al, 1991).

Additional peak demand savings are likely, and not included here. Furthermore, it is likely that with the use of a diagnostic system like that described below, additional HVAC savings are possible. These additional savings are likely to be achieved from improved equipment sizing given reliable data on actual cooling loads. Some examples of this are presented. Finally, greater statewide savings are also possible from extending these technologies to other building types.

One of the project sponsors, the California Institute for Energy Efficiency, asked the project team to focus on large Class A office buildings. Class A office buildings represent the most prestigious buildings in the office market³. They have above average rents, high quality standard finishes, state-of-the-art building systems, exceptional accessibility, and a strong market presence. Class A building owners and operators are among the most sophisticated in the commercial sector. They are considered innovative, and early adopters of emerging technologies. Several building owners and operators have expressed a strong interest in collaborating with the research team on a field demonstration of the prototype diagnostics system. One benefit of performing the field test demonstrations with Class A building operators is that the property management companies who operate the buildings tend to have large real-estate portfolios. They indicate that they are eager to utilize lessons learned during the demonstrations in their other buildings, thus the approaches we explore could propagate rapidly throughout similar buildings. Additional factors regarding the selection and advantages of collaborating with this market segment are described in the next section.

² 0.15 * 24 BkWh elec. * 10240 source Btu/kWh + 0.15 * 32 BkWh gas = (36.9 + 4.8) * 1012 Btu (source) = 42 * 1012 Btu (source).

³ The Building Owners and Managers Association classifies office buildings as follows:
Class A buildings : The most prestigious buildings in the particular market that have above average rents, high quality standard finishes, state of the art systems, exceptional accessibility, and a definite market presence.
Class B buildings : Buildings that compete for a wide range of tenants with rents in the average range. Building finishes are fair to good and systems are adequate, but cannot compete with Class A at the same rent level.
Class C buildings : Buildings seeking tenants requiring functional spaces at rents below the average for the market area.

Continue to Identifying Major O&M Problems

Return to the Table of Contents.