Figure 3: Three Levels of the Diagnostic System

The rationale for the selection of these systems is as follows. First, the selection of whole-building diagnostics is the starting point of the proposed diagnostic system. Whole-building data contain the basic yard-sticks by which a building operator can get an overall set of metrics to evaluate building performance. The rationale for the selection of the cooling system is related to the benefits of working with this system, plus the difficulties related with the competing systems of lighting or ventilation systems. First, we found that there can be a major improvement in cooling plant analysis with the addition of improved metering equipment, such as thermistors and magnetic flow meters. With the addition of these systems, we will characterize the energy efficiency of the cooling plant. Chillers are the largest single energy-using components in large office buildings, and are thus one of the most logical items to evaluate. Evaluating the entire cooling plant will allow us to understand the overall system performance, which is more important than examining a component in isolation from the system.

For comparison, the measurement issues associated with ventilation and lighting are more distributed; literally distributed throughout a building. Measuring air flows is particularly problematic. A similar confounding issue with ventilation systems is that ventilation requirements in individual zones vary because of duct configurations and thermal variations. The result is that ventilation systems are difficult to evaluate, and are complex because they are coupled to tenant comfort and indoor air quality issues. These were determined not to be good candidates for the initial demonstration, but are suitable for future research. Our initial diagnostic system, by contrast, is restricted to monitoring cooling plant equipment that is located either in the central plant or on the rooftop.

The components selected for the analysis are chillers and cooling towers. Both of these components were targets of complaints from building managers as subjects of poor sizing. Chillers are often oversized, thus they require more power per ton than optimal because they are less efficient at low partial load. Cooling towers are often undersized. Larger towers allow the chiller to operate at cooler condensing temperatures. The diagnostic system will explore major failure modes for these components.

The IMDS will be a permanently installed and continuously active system. This is necessary because buildings continuously change. For example, some problems reoccur, such as those from modifications to schedules to handle special events. These modifications often lead to equipment being left on when not needed. The diagnostic system is designed to operate in parallel with any existing EMCS, rather than expanding or modifying the EMCS. This decision was made because current EMCSs do not have the necessary capabilities in sensing, network capacity, frequency of sensing, data archives, data processing and visualization.

Knowledge Base and Failure Modes
The Phase 1 study included a detailed analysis of performance metrics and benchmark data to characterize the fundamental principles of how building, system, and component performance can be characterized. We developed a series of standard graphics that will allow the metrics to be displayed in a manner that assists in the diagnosis. These graphs were analyzed to determine benchmark signatures for good performance, such as where measured values should fall on a given analysis plot, or what the curve shape should look like if the system or component is performing properly. We developed a series of measurements and sensing requirements to evaluate the systems and components. We also listed common modes of failure that one can diagnose with the given metrics and graphics based on case study data and related literature. The discussion of failure modes is not an entirely exhaustive list of failures, but covers common and critical modes of failure. In most of these categories, we have used the following relationship to search for efficiency increases (i.e. detect problems). There are three categories of deficiencies: efficiency degradation, scheduling, and load issues. For example, in a chiller system, these three categories can be summarized by:

For each level in the hierarchy, a step by step diagnostic procedure was developed. The proposed knowledge base is designed to be modular, with a set of standard graphs and standard information. These graphs also serve as a tutorial that is designed to orient the building operator on how best to understand the system or component's energy performance.

As mentioned, whole-building data are critical elements of a top-down system, allowing an operator or building owner to evaluate the annual energy use and annual energy costs. For comparison with other buildings, these data are typically normalized by floor area, to produce an energy-use intensity (EUI). As examples, we present two sample diagnostic plots at the whole building level. In Figure 4, benchmark whole building EUIs are plotted for comparison with a given actual building. Data sources for this plot include: EIA, 1995, Akbari et al, 1993 and 1989, Piette et al, 1986, and Energy User News, 1995. One compares these values with those of the current building to determine whether large differences exist. Independently, or as an adjunct to results from Figure 4, one can detect and diagnose time of day usage problems with three-dimensional, hourly whole-building graphics. Excessive off hour use can be easily detected with such plots.

Figure 4: Office Building Energy Use Intensities for Comparison

Examples of key failures that can commonly be identified with whole building hourly data include:

• Start-up peaks - such as high peak demands from six AM to nine AM that are substantially greater than average daytime peak demands. Start-up peaks are common because equipment are scheduled to cool down or heat up building thermal systems without careful sequencing.
• High afternoon peaks - similar to start-up peaks, peak demands may creep throughout the afternoon. Careful tracking of these peaks can help reduce peak demand charges.
• High loads at night - one of the most important energy savings opportunities in buildings is to identify equipment that is on at night when it is not needed.

The entire cooling system efficiency can be evaluated using the efficiency versus load analysis (kW/ton vs. cooling tons) as an important element of the Knowledge Based System (KBS). The total cooling system performance in kW/ton is affected by the kW/ton for each component. It is convenient to view the entire system performance as a series of plots for each component, each showing the kW/ton vs. percent load. The shape of the efficiency versus percent load curve is dominated by the chiller, so the entire cooling system kW/ton curve tends to look like the chiller curve. Examples of cooling plant failure modes are the following:

Efficiency Degradation

Deviation from Manufacturers Specifications because of Design Flaws

Deviations From Historical Performance (e.g., cooling tower down time)

Poor Water Flow (resulting in need for multiple chillers)

Chiller Component Malfunctions (e.g., condenser fan cycling)

Fouling

Inappropriate Refrigerant Charge

Schedule and Control Deficiencies

Excessive On Time

Short Cycling

Load Issues

Poor Full-Load or Part-Load Performance (e.g., weather correlation)

Component Oversized or Undersized

Failure Modes Detected with Condenser and Chilled Water Flow Data

Low Flow

Flow Not Varying.

Failure Modes Detected with Cooling Plant Temperature Data

Improper Setpoints

Sub-optimal Chiller Controls

Poorly Calibrated or Poorly Located Thermostats

One common example of chiller diagnostics is the evaluation of efficiency (kW/ton) versus load (tons). Chillers should ideally operate near their rated efficiency (purchase point). Various problems (oversizing, improper scheduling, control problems etc.) exhibit signatures on this type of plot. The failure modes for chillers are similar to those listed for cooling systems. Cooling tower failure modes were also developed in Phase 1.

The chiller monitoring will capture key parameters in the chiller operation such as water flows and temperatures, pressure drop, and power. These data will allow determination of chiller efficiency and loads. We will also measure the pressure drop across the chiller heat exchangers to determine the extent of fouling. The cooling tower monitoring will also include water temperatures and flows, plus local outdoor air data and cooling tower fan power. The local outdoor air data are an important factor in assessing the performance of cooling towers. A temperature measurement station including an aspirated psychrometer will be installed on the top of the building as far away from the cooling towers as possible. Data from this psychrometer will be used to evaluate "nano-climate" effects. Cooling tower intake conditions will be compared with outdoor air conditions to evaluate recirculation of cooling tower exhaust.

Database and Visualization
The data base is needed to correctly archive and retrieve desired information from the large amount of data. The advanced graphics and powerful computing environment are needed in a prototype system like the one proposed to implement desirable data computation and visualization without constraints imposed by the computing platform. This is both a hardware and a software issue. We intend to use UNIX graphics workstations⁴ and a modified form of the Electric Eye building data visualization software.

Continue to Phase 2 Activities

Return to the Table of Contents.

Return to Diagnostic Technology and System Design Criteria.