Acknowledgements

Funding for this project was provided by the Florida Energy Office, Department of Community Affairs. We would like to thank Ed Cobham and the other members of FEO staff for their continued support of research into uncontrolled air flows in buildings.

We would also like to thank John Toth of FSEC who contributed considerable time, energy, and talent to doing field testing, organizing our files, and organizing our slides; and to Tom Hansen of Future Controls, our Test and Balance expert, for his considerable help in deciphering complex control systems and assisting us with various aspects of Test and Balance; and to John Tooley of Natural Florida Retrofit for help in field testing several buildings. Thanks to members of the advisory committee for giving of their time and energy to participate in this project. Thanks to the three paid consultant members of the advisory committee, Joe Lstiburek, Terry Brennen, and Bill Turner, for contributing their expertise and encouragement to our effort. Thanks to the following volunteer members of the advisory committee for their considerable contributions to this effort.

1. Executive Summary

The objective of this study was to develop the capability to substantially reduce energy use, building degradation and indoor air pollution caused by uncontrolled air flow in nonresidential buildings. For purposes of the study, “uncontrolled air flow” was defined as air moving across the building envelope or between zones or components of a building, where the pathways of flow, the direction of flow, and the origin of the air are unknown, unspecified, or unintended.

The study comes to three major conclusions as follows:

A major goal of this study has been to characterize the nature and extent of uncontrolled air flow through testing, measurement and monitoring in 70 small commercial buildings. Results showed that the nature of uncontrolled air flows in these buildings varies widely, and is strongly dependent on a large variety of complex building system interactions. As to extent, the study found that uncontrolled air flow is ubiquitous. Out of the 70 buildings studied, only one was deemed to be a “good” building. Repairs were made on 20 of these buildings. Before and after monitoring showed average energy savings of 15%.

Perhaps the most profound and compelling finding of the study is that, given the present state of practice, whether a building will avoid serious, or even catastrophic problems due to uncontrolled air flow, is primarily a matter of ijk. Building practitioners do not have access to information about uncontrolled air flow and its potential consequences in buildings. Even the research community has only recently “discovered” uncontrolled air flow, as evidenced by the fact that this study is the first major research effort of its kind. This study has found uncontrolled air flow to be surprisingly complex, and the impacts and consequences to be quite well camouflaged. Under these circumstances, it is not surprising that a good building depends almost entirely on chance.

The salient characteristic of uncontrolled air flow that appears most elusive is the fact that our buildings are functioning as pressure vessels. Mechanical air distribution systems force air through not only “leaky” ducts, but also through the various zones of this complex pressure vessel. Because we cannot see the uncontrolled air flow that results, we normally are not aware of it. As a result, forced air systems are typically treated as discrete, separate systems that begin and end with the fans, blowers, and duct systems. In actuality, the building is a complex series of pressure vessels that connect together the two ends of the air handler and duct systems, and are thus an integral part of them.

It appears that this lack of information on building air flows has forced practitioners to accept things on faith. They are condemned to the smart air syndrome -- the belief that the air is “smart’ enough to flow where it is supposed to flow, which is only within the designated ducts of the mechanical systems.

There is scant evidence from this study that the impacts of uncontrolled air flows in buildings are understood or appreciated. Buildings are designed and tested and balanced considering only the air flows at designated air inlets and outlets of the duct systems. Even though duct systems have been shown to be quite leaky, they are practically never tested for tightness. The field evidence points to two ill-fated assumptions: 1) the ducts will not leak, and 2) forced air flow begins and ends at the designated termination points of the air handler’s duct systems. Numerous examples that show the widespread existence of the smart air syndrome are contained in the body of this report.

This study conclusively shows that duct systems normally leak, often extensively. Given this fact, and the fact that the building acts as a pressure vessel, it is clear that building pressure measurements are the only accurate means of determining air flow balance in buildings. The study concludes that building pressures must be measured under a variety of building and mechanical system operating conditions in order to accurately evaluate and understand the impacts of the forced air flow in buildings. Air flow measurements alone are simply insufficient.

The extensive test and measurement data from the 70 small commercial buildings studied here show that these buildings are considerably more “leaky” than residences. The data also show that uncontrolled air flow is more pervasive and more complex in these buildings than in residences. The data also disclose a labyrinth of complex building system interactions that make diagnostic generalizations virtually impossible. Almost any given diagnostic result (for instance, the degree of supply duct leakage) can have an enormous variety of impacts, depending quite literally on everything else in the building. Simple generalizations and rules-of-thumb inevitably have proven insufficient. Nonetheless, this study provides answers to many previous questions. It also provides much needed insight, and discloses an extensive array of additional gaps in our knowledge base. (See especially Sections 4, 7 and 8 of this report.)

It has become clear from the study that more research and extensive education and training are required before practitioners will be able to successfully evaluate the impacts of these complex building system interactions. One recommendation of this study is to transfer this “new” knowledge to the building industry -- its researchers, its practitioners, and its regulators. Section 7 of this report details the specific needs in this area and recommends implementation strategies for proceeding with this important task.

It has also become clear that, given the proper attention, most uncontrolled airfiows and the problems they engender can be avoided. Section 6 of this report provides extensive discussion on the characteristics of a good building. It is now clear that a much improved set of “best practices” for building design, construction, and commissioning can be achieved. Extensive effort on the part of many individuals and organizations will be necessary to widely effect such a standard in our diverse building industry.

This study is important because uncontrolled air flows in buildings often have serious and sometimes catastrophic consequences. Increases in energy and demand costs may be the least critical of these. Often more critical and costly are the decreases in health and safety, building material durability, indoor air quality, indoor comfort, building moisture control, worker productivity, and business revenues that are experienced in problem buildings. Much too often, uncontrolled air flows cause massive building failures that result in dramatic remediation and litigation costs. In some cases, these costs have exceeded the original cost of constructing the building. Indoor air quality problems alone are estimated to result in tens of billions of dollars in unwanted costs annually in U.S. buildings.

The study presents strong evidence that many serious indoor air quality problems either arise directly from, or are exacerbated by, uncontrolled air flows in buildings. One of the important overall conclusions of this study can be concisely summarized as follows:

It is highly unlikely that indoor air quality can be ensured in any building
unless the potential for uncontrolled air flow in that buildings is eliminated.

In other words, if we can’t control the air flow, how can we expect to control the air quality?

In summary, virtually all uncontrolled air flows in buildings stem from one or more of the following three straightforward, building-system characteristics:

The results of this study show that buildings do not have to be plagued by these unwanted air flow problems. Good buildings can be achieved.

2. Project Description

2.1 Introduction

“The overall objective of this project is to develop the capability to substantially reduce energy use, building degradation, and indoor air pollution caused by uncontrolled air flow in non-residential buildings.” (Workplan: Uncontrolled Air Flow in Non-Residential Buildings, February 4, 1993). The workplan goes on to say, “This involves gaining knowledge of air flow and pressure differentials in non-residential buildings, identifving tools and testing techniques for research and diagnostic work, and developing recommendations for dealing with uncontrolled air flow in non-residential buildings.”

Before proceeding, a definition of uncontrolled air flow is helpful. Our working definition has been:

“Uncontrolled air flow - air moving across a building envelope or between
zones or components of a building, where the pathways of flow, the direction
of flow, and the origin of the air is unknown, unspecified, or unintended.”

Seventy small commercial buildings in central Florida were tested for various types of uncontrolled air flow, including duct leakage, return air imbalance problems, and exhaust air/intake air imbalance. These buildings ranged in size from 704 square feet to 22,461 square feet, and averaged 5030 square feet. Typical tests included building airtightness, duct system airtightness, infiltration rates, and pressure differentials; and air flow rates of supply, return, air handlers, exhaust fans, outdoor air, and make-up air. This study concludes that uncontrolled air flow is a serious and wide-spread problem that contributes to energy waste, elevated peak electrical demand, high relative humidity, building materials moisture degradation problems, mold and mildew growth, combustion safety concerns, and indoor air quality complaints.

Uncontrolled air flow is a function of the intensity of drivers (primarily mechanical air moving systems) and building complexity and tightness (see section 2.2 for discussion of uncontrolled air flow potential). Compared to residences, small commercial buildings have larger air flow drivers because they have larger HVAC systems which run a greater proportion of the time. Cooling systems are considerably larger, averaging 3.38 tons per 1000 square feet in these 70 buildings compared to 1.8 tons per 1000 square feet in typical residences, and in some small commercial buildings the air handlers run continuously. Exhaust fans are larger and operate longer periods of time in commercial buildings. Additionally, commercial buildings have outdoor air and make-up air which do not normally exist in Florida residences.

Duct leakage, as measured by duct depressurization test, is three times greater in commercial buildings than in Florida residences (area normalized). Even considering that the cooling systems of commercial buildings are nearly twice as large (in terms of tons of cooling capacity per 1000 square feet), this duct leakage is considerably in excess of duct leakage in Florida residences. According to SMACNA, these commercial duct systems are approximately 70 times more leaky than the SMACNA duct tightness standard. In Florida residences, duct leaks are almost always to and from unconditioned spaces (attics, garages, reducing exhaust air, and airtightening of leaky t-bar ceilings).

Cooling energy use was monitored for four to six summer months in each building. Repair of uncontrolled air flow was done in the middle of the summer. On average, cooling energy consumption decreased from 87.4 kWh/day to 75.1 kWh/day, or 12.4 kWh/day. On average, cooling energy use declined by 14.7% from repair of uncontrolled air flow. Based on the assumed $0.075/kWh electricity cost, projected annual cooling energy savings are $182. Given that the average projected retrofit cost is $454, simple payback is 2.5 years. This indicates that UAF repairs can be very cost-effective retrofit measures.

This report concludes that uncontrolled air flow is widespread in commercial buildings. Energy waste is one important consequence. However, other consequences of occupant discomfort, cooling/heating equipment being improperly sized, high humidity, moisture problems, indoor air quality complaints, and sick buildings are of equal or greater importance. Since uncontrolled air flow problems are widespread and often severe, there is a strong need to bring about changes to the way buildings are created and used.

A plan was developed to implement the findings of this project (chapter 7). The plan suggests three areas of further work. 1) More research needs to be done to understand the extent, consequences, and solutions to uncontrolled air flow in commercial buildings. 2) Based on findings from this research, various standards relating to building design, construction, commissioning, and maintenance need to be modified. 3) To bring about resolution of uncontrolled air flow problems, training programs need to be established to bring improved design, construction, commissioning, and maintenance skills to those responsible for buildings.

Uncontrolled air flow may be caused by duct leaks, return design problems, and exhaust/intake imbalance. Another and more inclusive way of stating this is that uncontrolled air flow occurs:

In order to achieve project objectives, we have tested 70 commercial buildings and made repairs of uncontrolled air flow (UAF) in 20 commercial buildings. (Note: by commercial buildings, we mean all non-residential buildings except industrial.)

2.2 Building Categorization

Commercial buildings fall into a wide variety of building uses and types. By comparison, there is much less diversity in single family homes. Most homes have two, three, or four bedrooms, a living room, a dining room, a kitchen, one or two bathrooms, and a garage. There are, of course, regional variations regarding foundation style (slab on grade, crawlspace, or basement), number of stories, and attic configurations. In terms of air flow, etc. In small commercial buildings, however, duct leaks commonly exist in four different ambient environments. 1) In some cases, the duct leakage occurs inside both the building air barrier and thermal barrier (insulation). In these cases there is not much air exchange or heat exchange with outdoors. 2) In other cases, the duct leaks occur inside the building air barrier but outside the thermal barrier (such as when the ducts are in the space between the ceiling and the roof deck, but the insulation is on top of the ceiling tiles) and therefore have significant energy penalties. 3) In yet other instances, the duct leaks exist outside both air and thermal barriers (such as in vented ceiling spaces or vented attic spaces), and the energy penalties of duct leakage can be severe. 4) Ducts may also be located on the roof. In many cases, a small portion of the ductwork is located on top of the roof. In a small number of cases, the entire duct system is on top of the roof. The energy penalties of duct leakage are generally worst in case 3, but are also substantial in cases 2 and 4.

Small commercial buildings are more leaky and often have greater complexity than residences. Testing in 70 small commercial buildings found them to be 30% more leaky than residences, primarily because suspended t-bar ceilings are quite leaky. ACH5O, the air exchange rate of the building when depressurized to -50 pascals by a blower door, averages 16.7 compared to 12.7 in Florida residences. The difference is even more pronounced in new construction. New commercial buildings are more than twice as leaky as new Florida residences.

Small commercial buildings are very leaky primarily because suspended t-bar ceilings are on the order of 10 times more leaky than gypsum board ceilings. Because the ceilings of most commercial buildings are so leaky, overall building leakiness depends primarily upon whether the ceiling space or attic space above the ceiling is well ventilated to outdoors. In those which have tight ceiling spaces, building airtightness may be 5 ACH5O or less. Several of the 70 buildings were considerably tighter than any Florida residence we have tested. In the majority of cases, where the ceiling space or attic space is vented, building airtightness is 15 ACH5O or greater. Also of note is that attached units, such as those in strip malls, are more than twice as leaky as stand-alone commercial buildings. Because much of the leakage of these attached units is to adjacent units, the energy impacts of this excessive building leakage may be relatively less severe.

Commercial buildings are often more complex than residences -- that is, they have a greater number of partitions and compartments. A building can be thought of as a matrix of barriers to air flow and pathways for air flow. Interior walls, closed interior doors, firewalls, and multiple stories create potential barriers to air flow which can interrupt the flow of air throughout the building. This disruption of air flow can create pressure imbalances which can cause elevated infiltration rates, accumulation of moisture in building cavities, and backdrafting of combustion equipment. Filters and coils, as they become dirty, become barriers to air flow and can create pressure imbalance in the air distribution system.

Repair of uncontrolled air flows was done on 20 of the 70 buildings. Repair candidates were selected based on the perceived potential for energy savings and whether the repair was financially feasible within the project budget. Repairs included sealing of duct leakage, provision of return air pathways, reducing outdoor air flow, airtightening the building envelope, and turning off attic exhaust fans. The majority of repairs were duct repair. Three major types of repairs that were not attempted were provision of make-up air, homes often have air distribution systems, but no make-up air, no outdoor air, and small and infrequently used exhaust fans. Variations in homes is much smaller than in commercial buildings.

Commercial buildings have a much wider range of sizes, uses, construction styles, height, and mechanical systems. A conceptual framework was developed for thinking about the diversity of building types and configurations found in commercial buildings based on the potential for and consequences of uncontrolled air flow.

2.2.1 Uncontrolled air flow potential

A two-dimensional matrix has been developed that attempts to describe the potential for uncontrolled air flow in a building. The two axis are “drivers” and “building” (Figure 2.1).

“Drivers” are the forces which move air within a building and across the building envelope. They include air handlers, exhaust fans, outdoor air, make-up air fans, and operation of combustion equipment (Figure 2.2). Drivers can also include wind driven air flow and pressure differentials, and stack effect pressures (especially in tall buildings in cold weather). Drivers range from mild to intense. The more intense or larger the drivers, the greater the potential for uncontrolled air flow to occur.

“Building” refers to building size and complexity, but complexity is most important. A building is an interwoven fabric of barriers and pathways. A wide-open retail space, for example, has few compartments, and is therefore “simple”. A building which has many partitions and subdivided spaces is complex. Partitions provide the potential to restrict air flow and therefore have the potential to create substantial pressure imbalances and UAF (Figure 2.3).

Simple buildings with mild drivers fall into the “low potential” portion of the matrix and have less potential for UAF and large pressure differentials. Examples include restaurants (large exhaust, make-up air, and outdoor air) and some recreation and light industrial/warehouse facilities. Complex buildings with mild drivers (only air handlers) also fall into the "moderate potential" portion of the matrix and have substantial potential for UAF because of closed doors when returns are centrally located.

Complex buildings with intense drivers fall into the "high potential" portion of the matrix and have the greatest potential for UAF; these would include hotels, hospitals, larger restaurants, and some sports facilities.

2.2.2 Uncontrolled air flow consequences

While uncontrolled air flow potential can be defined by a two-dimensional matrix consisting of drivers and building complexity. The potential for uncontrolled air flow can be thought of as the area defined by the two matrices (Figure 2.4). Uncontrolled air flow consequences can be conceived as a three-dimensional matrix defined by drivers, building complexity, and sources, and the extent of consequences can be thought of a volume defined by the three parameters of drivers, building complexity, and sources (Figure 2.5). Sources include heat, cold, moisture, and air contaminants. When transported by UAF, these sources often have negative consequences. Sources fall into three primary categories; 1) cold/heat, 2) moisture, and 3) pollutants.

2.2.2.1 Cold and Heat

When drawn into the building by UAF, cold and heat can cause the following consequences.

2.2.2.2 Moisture or dryness

When drawn into the building by UAF, moisture or dryness (air with low moisture content can cause the following consequences.

2.2.2.3 Pollutants

Pollutants can be drawn itno buildings or not properly diluted (diminished ventilation) as a result of uncontrolled air flow. UAF can cause indoor air quality problems by four mechanisms; 1) mining of pollutants, 2) transporting of pollutants, 3) generation of pollutants, especially microbial growth, and 4) diminishing of ventilation.

Mining of pollutants occurs when depressurization caused by UAF draws pollutants from the soil (radon, methane), sewer pipes (methane, sulfur dioxide), or combustion equipment (carbon monoxide, moisture, particulates, and nitrous oxides). Space depressurization draws pollutants through cracks or penetrations in the slab (soil gases such as radon or methane), from sewer lines when plumbing fixtures are not seated properly or when traps are dry, and from combustion appliances when backdrafting or spillage is induced.

Transport of pollutants occurs when UAF carries pollutants from zones that have air contaminants. Following are examples. Return leaks in ducts or air handlers located in crawl spaces, garages, attics, etc. may transport radon, volatile organic compounds (oil and gasoline in garage), and insulation particles from these zones to the occupied space. In at least two reported incidents in Duval and Brevard counties in Florida, a total of seven persons died in their homes from carbon monoxide poisoning when cars were left running in closed garages. Leaks in the air handlers and ducts in the garage transport carbon monoxide-laden air into the house, resulting in lethal carbon monoxide poisoning. Return leaks in a roof-top package air conditioner may be drawing air from contaminated sources such as cooling towers, plumbing stacks, or exhaust fan discharges.

Generation of pollutants occurs when UAF actually creates or increases the pollution source. Four examples are presented.

Diminished ventilation may occur if outdoor air, exhaust air, or. make-up air ducts have leaks. Consider some examples. Outdoor air ductwork is located in a ceiling space. If the ductwork leaks, then building air is drawn into the outdoor air ducts and thus reduces building ventilation air. If exhaust fan ductwork leaks (blower at or near the grill), so that air is discharging from the ducts into the building, then polluted air may be re-entrained into the building and the overall ventilation rate may be diminished.

2.2.3 Types of air quality consequences

Uncontrolled air flow generates very substantial energy consequences. • However, these consequences are often’ dwarfed by the indoor air quality and materials damage consequences. Air quality problems produced by the four UAF mechanisms listed in the preceding section can result in various types of consequences.

1. Odor problems
2. Health consequences
   a) chronic
   b) acute
3. Reduced worker productivity
4. Sick buildings
   a) headaches
   b) respiratory stress
   c) increased illness
   d) anxiety. developed among workers
5. Evacuation of buildings, temporarily or long-term
6. Law suits

UAF may lead to increased sick leave and reduced worker productivity. Add to this the possibilities of having to evacuate the building, renovation costs, and law suits, the air quality:nationwide consequences of UAF may run into the billions of dollars each year.

2.2.4 Consequences are somewhat random

Just because there are intense drivers, a complex building, or both, does not ensure that uncontrolled air flow will result. These factors simply increase the potential for uncontrolled air flows.

Just because there are uncontrolled air flows does not mean there will be consequences. The presence of uncontrolled air flow simply increases the potential for significant consequences. Whether there are consequences or not depends in large part upon luck (often the randomness of how buildings are put together) and of course whether there are sources.

Just a matter of luck. Whether uncontrolled air flow leads to serious consequences is often just a matter of luck. Consider the case of one manufactured office building (#16 on Master Table in Appendix B) tested in this project. Seven package air handlers are mounted on the exterior wall of the building, through-the-wall returns pull air from office, return transfer grills in the office doors were greatly undersized (allowing only about 5% of the needed air flow), and office doors are closed most of the time (Figure 2.6 -- note the numbers are pressure differentials and the arrows indicate the direction of pressure gradient and air flow). The closed doors then acted as barriers to air flow, creating strong depressurization in the closed offices, and thus pulling a majority of the return air from the space above the ceiling, which had 34 passive roof vents to outdoors (Figure 2.7).

Two fortunate circumstances prevented this major form of uncontrolled air flow from being a serious problem. First, the roof vents had dampers and these dampers were quite tight. Second, the ceiling insulation (batts) was attached to the. bottom of the roof deck. Therefore, this ceiling space, which was acting like a return plenum, was located inside both the air barrier and thermal barrier of the building. If the insulation had been located on top of the ceiling tiles or the roof vents actually ventilated the ceiling space, then there could have been substantial energy and humidity consequences.

Importance of sources. The importance of sources can be illustrated by considering the following example. If we have a building with considerable UAF, and it is drawing considerable air from outdoors, but the air being drawn into the building is virtually identical to the desired indoor conditions (75F and 50% RH) and has no pollutants, then there may be no energy, comfort, humidity, building material damage, or indoor air quality consequences from UAF. This could occur, for example, in a city like San Diego, California when cool and clean breezes blow from the ocean much of the time. In real life, however, the air brought into buildings by UAF is generally too cold, too hot, too humid, too dry, or polluted. Even in San Diego on a perfect day, air can be drawn from an attic space (which may be hot) or a loading dock area (which may be polluted), and there may be negative consequences.

3. Diagnostic and Testing Protocols

One of the primary objectives of this project was to develop diagnostic and testing methodologies which can be used to diagnose UAF problems in buildings. Since this project is research, the testing actually done on these buildings was more comprehensive than we expect will be used in typical “real world” diagnosis. Nevertheless, many of the research procedures and testing methodologies developed in this project will be incorporated into real world diagnostics. A number of new measurement techniques were developed and refined in this project, and they allow mçasurement of air flows and pressure differentials in buildings more accurately and quickly than is commonly available in the industry.

Following is a discussion of the various diagnostic and testing protocols used to characterize airtightness, pressure differentials, and air flow rates in this project.

3.1. Not All Tests Were Done In All 70 Buildings

Not all tests were done in all 70 buildings. In terms of building airtightness, one business owner would not allow us to do a blower door test because he felt it would disrupt his business activities. Building airtightness tests exist for the other 69 buildings.

In 24 of the 70 buildings, duct system airtightness was not measured. Three factors determined whether the duct airtightness test was done.

In 13 of the 70 buildings, the building infiltration rate with the air handlers (and normally operating exhaust fans) turned off was not measured. This occurred almost exclusively in cases where the occupants were very reluctant to allow us to turn off the air conditioning systems for a one-hour-plus period. Business owners are generally very sensitive to things that affect worker productivity or customer comfort. In many buildings, turning off the cooling system on summer days results in rapidly rising temperature and humidity. In restaurants, it is virtually impossible to turn off the exhaust fans because the cooking appliances require continuous exhaust.

In one small real estate office building, for example, the air handler-off infiltration test was done over the lunch hour when most of the employees were out to lunch. During the 70- minute test, the temperature in the space rose by about 5F. Because the air conditioner was undersized (this was a thermally inefficient building because of a very leaky building shell and poorly located insulation), the temperature in the space never recovered and the occupants were hot (and not particularly happy with us) throughout the remainder of the work day.

Typical protocol. A typical protocol includes visual inspection, building and duct airtightness testing, pressure differential measurement, infiltration/ventilation tests, air flow measurement, and visual inspection of building mechanical components which may contribute to uncontrolled air flow and pressure differentials. The objective of the testing is to characterize air flows and pressure differentials within the building, characterize the air flow balance across the building envelope, identify the cause of air flow and pressure imbalances, and understand the interacting relationships between building airtightness, air flows, pressure differentials, the operation of building equipment, indoor air quality, ventilation, and energy consumption.

3.2 Visual Inspection

The first step in the diagnostic process is visual inspection and obtaining information from persons familiar with the building. Following are steps typically taken.

3.3 Important Definitions

Following are definitions that are important to understanding some of the following discussions.

Attic space is a space above the ceiling that has trusses or joists which can be walked on and often has sloped roof, wood decking, intentional ventilation, and insulation at the attic floor level.

Backdrafting is reversal of flow of gases down the chimney or vent of a combustion appliance.

Ceiling space is the space above a ceiling that has no structural members that could support a person’s weight, may have insulation at the ceiling or roof deck level, and generally is not intentionally vented.

Pressure pan is a pan that can be placed over supply and return registers to measure pressure difference. It is made of a cake pan (or other type of pan). Gasketing is put on the pan rim to facilitate an airtight fit and a tap penetrates the pan to allow measurement of pressure in the ductwork.

Primary air barrier is that portion of the building envelope which provides the greatest resistance to air flow and the greatest pressure drop when the building is exposed to a significant pressure differential compared to outdoors. In the ceiling/roof plane, the roof deck may be the primary air barrier or the ceiling may be the primary air barrier. Note that this means only that the primary air barrier is only relatively tight; that is, it is tight only by comparison to any other air barriers in series with the primary air barrier. A suspended t-bar ceiling may be the primary air barrier, but it nevertheless may be very leaky.

Primary thermal barrier is that portion of the building envelope which provides the greatest resistance to heat flow. In the ceiling/roof plane of the building, insulation may be located at the ceiling, at in the roof deck, or suspended in between. Often the primary insulation barrier is located on top of the ceiling tiles, but the (flat) roof assembly may also have some significant R-value, so the ceiling space is actually sandwiched between two thermal barriers.

Spillage occurs when only a portion of the combustion gases leave the building through the vent pipe. The remainder spill into the space. Spillage often occurs during start-up of a combustion appliance when the chimney or vent is cold. Ad the flue gases warm, draft strength increases and complete drafting of combustion gases results. Continuation of spillage beyond one minute indicates a draft problem.

3.4 Building Airtightness Testing

Typically, the second step in the diagnostic process is building airtightness testing. The building is prepared by turning off vented combustion equipment and air moving equipment including air handlers, exhaust fans, make-up air fans, and clothes dryers. Outdoor air, exhaust fans, and make-up air openings are sealed off, since these holes in the building do not respond passively to building air flow and pressure dynamics.

A multi-point airtightness test is performed (we follow the ASTM E 779-87, “Standard Test Method for Determining Air Leakage Rate by Fan Pressurization”, except we only depressurize the building), using from one to six calibrated fans (blower doors), depending upon the building airtightness and size, and generally obtaining air flow at five to eight building pressures in the range from -10 pascals to -60 pascals depressurization. (Note: all pressures expressed in this paper are “with respect to outdoors” unless otherwise indicated). Knowledge of building airtightness assists in interpretation of other field testing, especially pressure differential measurements, and in developing recommendations for air flow and pressure balancing.

3.5 Identification of Building Air Barriers

With the building depressurized to -50 pascals by the calibrated fan(s), pressures in various zones of the building are measured in ordet to identify the primary building air barrier; in other words to determine which portions of the building are “indoors” and which are “outdoors”. Pressure may be measured in the ceiling space, attic space, wall cavities, chases, soffits, mechanical rooms, ducts, space between floors, etc. Consider an example; if the ceiling space of an office buildings is at -5 pascals when the occupied space is at -50 pascals, this indicates that the ceiling is the primary air barrier. It may also indicate that the ceiling space is reasonably well ventilated to outdoors and that ducts located in the ceiling space are in a zone that is effectively “outdoors”. If the ducts are “outdoors”, then it will be more important to test for duct leakage.

If, on the other hand, the ceiling space is at -49 pascals when the occupied space is at -50 pascals, then the ceiling space and the ducts are located “indoors”. (Note that being inside the building air barrier does not ensure, however, that the ceiling space and the ducts are inside the thermal barrier; for example, the insulation may be on top of the ceiling tiles while the roof deck is the air barrier.)

Consider another example; a mechanical room containing a gas water heater is at zero pressure when the occupied zones are at -50 pascals. This indicates that the mechanical room is well connected to outdoors arid poorly connected to indoors, and that the combustion equipment located in that room will not be significantly affected by any pressures which may be created in the occupied zones. On the other hand, if the mechanical room is well ventilated to the occupied space and is at -48 pascals when the occupied space is at -50 pascals, then the combustion equipment located in that room could be significantly affected by any pressures which may be created in the occupied zones. This could lead to spillage or backdrafting of combustion equipment.

3.6 Duct System Airtightness Testing

Airtightness of the duct system can be measured by means of a depressurization test. Airtightness of the duct system can also be indicated by means of a pressure pan test.

3.6.1 Duct system depressurization test

Duct system airtightness may be measured using calibrated fans (duct test rigs or duct testers). All registers except one supply and one return (in proximity to the air handler) are masked off. Outdoor air inlets, if any, are masked off. Calibrated fans are attached to the open registers. An air flow barrier is placed in the air handler (at the filter, coil, or blower) to divide the system into supply and return. Air is drawn from the duct system by the calibrated fans and a multi-point (multiple pressures) airtightness test is done, with each side of the system at the same pressure (duct pressure is measured near the air handler and referenced to the zone in which the ducts are located). CFM25 (air flow through leaks in the duct system when the ducts are at -25 pascals) is determined for both the supply and the return side of the system. The combined CFM25 (add supply and return sides together) represents the combined leakage to outdoors, unconditioned building space, and conditioned building space, and can be expressed as CFM25total.

The duct system airtightness test can be repeated to determine what portion of the duct leakage is to outdoors (or to buffer zones which are well ventilated to outdoors). Using a calibrated fan, the building is depressurized to the same pressure as the duct system, usually at just one pressure of -25 pascals, and the duct system airtightness test is repeated. Since the occupied zone and the ducts are at the same pressure, the duct test rig is measuring only duct leaks to outdoors. The resulting CFM25 can be expressed as CFM25out.

If the ducts, plenums, and air handler are within the air barrier of the building, then the second test (with the building depressurized) is not needed. Note that in some cases, the ducts may be inside the building air barrier but there are leak pathways between the “designated” duct system and interstitial cavities that lead to outdoors.

In other cases, the ducts are located within the air barrier of the building, but outside the thermal barrier, such as when the roof deck is the building air barrier, the ducts are located in a ceiling space, and the insulation is on top of a t-bar ceiling. (T-bar ceilings are used in the vast majority of all commercial buildings. They are composed of t-shaped metal framework suspended from above with ceiling tiles supported within the metal framework). Duct leakage to and from these ducts causes considerable heat gain from the ceiling space during the cooling season and heat loss during the heating season. Even though this leakage occurs within the air barrier of the building, it occurs outside the thermal barrier of the building and consequently causes significant energy penalties (the extent of the energy penalties largely depends upon the thermal resistance inherent in the roof construction, the color of the roof surface, whether the roof surface is covered by gravel, and the extent of roof shading).

3.6.2 Pressure pan test

Duct system airtightness can be indicated (though not measured) by means of a pressure pan test. In this test, the building is depressurized to -50 pascals by the blower door. The air handler is turned off. A pan, similar to a cake pan, is attached to a pole and placed over each supply and return register, one register at a time. A micromanometer is attached to a pressure port on the pan so that the pressure in the ductwork (with respect to the room) can be measured. If the duct pressure is the same as the room pressure, then there is no duct leakage (to outside the building envelope). If the duct pressure is considerably different from room pressure, this indicates substantial duct leakage near that register. The following table provides a means of interpreting the pressure pan results.

Table 3.1
Interpretation of pressure pan results when house is depressurized
to -50 pascals. (Cummings, Tooley, and Moyer, 1993.

The interpretation of pressure pan test results shown in Table 3.1 assumes that the duct system is located in a zone which is well ventilated to outside, so that when the occupied space is at -50 pascals, the duct zone is near neutral with respect to outdoors. In single family residences, this assumption is most often correct. In commercial buildings, however, the zone containing the ductwork is frequently located inside the primary air barrier and may experience pressure much closer to indoors than outdoors.

Consider the following example. In Realty 2 (building #37), the attic space is at -43 pascals with respect to outdoors when the occupied space was at -50 pascals with respect to outdoors (the attic is at +7 pascals with respect to the occupied space). Pressure pan tests were done on the 10 supplies and 1 return of the duct system. The indicated pressure ranged from 0.7 pascals to 5.0 pascals and the average was 2.0. Since the maximum pressure that could occur was 7 pascals (50 pascals - 43 pascals), these pressures indicate large duct leakage.

In order to convert these number to values which can be interpreted on Table 3.1, multiply the pressure pan pressures by 50 (the assumed pressure difference between the occupied zone and the duct zone) and divide by the actual pressure difference between the occupied zone and the duct zone. In this example, multiply by 50 and divide by 7. The converted pressure pan pressures would range from 5 to 35.7 pascals. In many commercial buildings, pressure in the duct zone is the same or virtually the same as in the occupied space (blower door operating); consequently the pressure pan test will not work in these buildings. In conclusion, while the pressure pan test is widely useful in residences, it is often not an effective diagnostic tool in commercial buildings.

3.7 Building Pressure Differentials

Pressure differentials were measured in the building with the building and HVAC systems in various modes of operation. Pressure in the building is measured with respect to (wrt) outdoors once with the air handlers turned to continuous “on” and the exhaust fans operated in normal operation. Pressure in the building is measured a second time with all mechanical systems turned off. Pressure in various rooms and zones of the building are measured with doors open and closed and various HVAC equipment turned on and off in order to characterize pressure differentials between various zones of the building and between those zones and outdoors. A primary objective is to characterize the effect of the air moving equipment on building and zone pressures, especially negative pressure.

Pressure differential measurements are made with 2-channel and 8-channel digital micromanometers with resolution to 0.1 pascals. The hand held units have time-averaging capabilities which allow discriminating small pressure differentials even when significant fluctuations exist because of wind. The 8-channel micromanometer with interface to computer display and memory was used to sample at up to eight locations simultaneously throughout the building and mechanical systems, plot continuously on a computer screen, and store data for later analysis.

3.8 Infiltration/Ventilation Rates

Using tracer gas decay methodology (ASTM E 741, “Standard Test Method for Determining Air Leakage Rate by Tracer Dilution”), the building infiltration/ventilation rate was measured, once with the HVAC equipment operating and then again with the HVAC equipment turned off (if possible or practical). In 13 of the 70 commercial buildings, the “equipment off ‘ test was not done because it was difficult to find a period when the occupants would allow turning off the mechanical systems.

Foxboro Miran 101 Specific Gas Analyzers and a Bruel and Kjaer 1302 multi-gas analyzer were used in various types of infiltration testing. Sulfur hexafluoride and nitrous oxide were the two gases that were commonly used as tracers. The typical test method using the Miran 101 was as follows:

3.9 Air Flow Rates

In virtually all buildings, HVAC system air flow rates were measured. A number of different measurement techniques were used.

Various methods were used because 1) some methods work better than others because of intake or discharge configuration, 2) some methods take more time, and 3) some methods take duct leakage into account and others do not. Air flow hoods are useful for getting a “quick and dirty” picture of overall flows of the HVAC systems. They are not the best choice, in many cases, for measuring exhaust air flows, make-up air flows, or outdoor air flows. They also may overestimate air flows, especially at discharge grills/registers (more discussion in 3.8.1).

Tracer gas injection in conjunction with a gas analyzer can measure total air flow through ducts, air handlers, exhaust fans, and make-up air fans. Calibrated blowers can be used to measure air flows through air handlers, exhaust fans, make-up air fans, and even outdoor air. Both methods can deal with most intake or discharge configurations and accurately take duct leakage into account, but they are both more time consuming compared to using a flow hood.

Since an important aspect of diagnosing uncontrolled air flow is accurately measuring net air flows across the building envelope, the more time consuming methods are often called for. Following are detailed descriptions of the various air flow measurement methods.

3.9.1 Flow hood

Air flow at supply registers and return grills is measured by air flow hood. Outdoor air is typically measured with a flow hood, by placing the air flow hood over the outdoor air intake opening. Both Shortridge and Alnor hoods were used in this project.

These air flow hoods provide the fastest means for determining flow from registers and grills. They are generally quite accurate measuring flow into return grills and exhaust grills. They are not so accurate measuring air coming from some supply registers, especially those that discharge mostly to one side or have jetting” of air into the hood (these occur in small commercial and are most common in residential systems). With some register configurations and air discharge configurations, we have found measurement of air flow from supply registers often are 20% too high and can be as much as 50% to 80% too high. The larger 24”x24” grills with diffuse holes or discharging equally to all four sides can usually be measured with very little error.

3.9.2 Tracer gas

While tracer gas is most commonly used to determine the building infiltration or ventilation rate, it can also be used to determine return leak or outdoor air flow rates. Tracer gas is distributed into the building (as in the tracer gas decay infiltration test) and well mixed (about 15 minutes with the air handlers operating). Tracer gas concentration is then sampled at three locations for each air handler; A) in the room near the return grill, B) at the discharge of a supply grill, and C) at the return leak or outdoor intake location (since some tracer gas may be re-entrained into the outdoor air intake). To get return leak fraction (RLF), run the test once with the outdoor air intakes masked off. Then rnn the test a second time with the outdoor air grills open.

RLFt, the proportion of return air flow that enters the return air distribution system through all leaks, whether from inside the building shell or not, is calculated by the following equation (Cummings and Tooley, 1989).

RLFt = (A-B)/(A-C)

where

A is the tracer gas concentration of air entering the return grill(s)
B is the tracer gas concentration of air coming from a supply grill
C is the tracer gas concentration of air at the return leak site

Note that as the concentration of tracer gas at C approaches the value of that at A, the accuracy of the test diminishes rapidly.

RLFo, the proportion of return air flow that enters the return air distribution system through leaks from outside the building, is calculated by the following equation:

RLFo = (A-B)/A

Note that this test is done with outdoor air intake sealed off.

To obtain outdoor air fraction, repeat the return leak fraction test but this time with the outdoor air intake open. OAFt, the proportion of return air entering through both the outdoor air opening and return leaks from outdoors, is calculated by:

OAFt = (A-B)/(A-C)

where

C is the concentration of tracer gas enetering the outdoor air opening.

Outdoor air fraction (OAF; air entering only through the outdoor air vent) is calculated as follows:

OAF = OAFt - RLFo

Total return leak flow is calculated by:

RLFflowt = RLFt * air handler flow rate

Return leak flow from outdoors is calculated by:

RLflowo = RLFo * air handler flow rate

Outdoor air flow rate (OAflow) is calculated by multiplying OAF times total air handler flow rate.

OAflow = OAF * air handler flow rate

3.9.3 Measuring air flow rates using tracer gas injection

Flow rate of exhaust fans and make-up air fans (and air flow through most duct systems) can be measured by means of tracer gas injection and sampling with a gas analyzer downstream. Tracer gas is continuously injected into the air stream such that the gas is well distributed (tubing with holes is often used), and the gas is sampled downstream in a distributed manner (a loop of tubing attached to a sampling pump). The injection rate of the tracer gas must be accurately measured. Tracer gas concentrations before the injection point (Cs) and downstream (Cb) must be accurately measured. Flow rate is calculated by means of the following formula (Grieve, 1991):

q = dose/(Cs-Cb)

where

q is the air flow rate in cfm
dose is tracer injectin rate in cfm
Cs is the tracer concentration at the sample point
Cb is the tracer concentration upstream of the injection location

Gas flow meters.
Two sizes were purchased to allow a wide range of gas flow rates. Computer software was purchased which allows accurate determination of flow rates through these two meters for a wide range of gases.

A water-displacement procedure was developed to calibrate the flow meters. An airtight, five-gallon container of water was placed on an accurate weight measurement scale. Tracer gas then flowed into the container and displaced the water. The change in water volume was determined by means of change in weight. In order to avoid compression of the tracer gas within the five gallon container, pressure (wit the room) in the container was measured. The flow rate of water out of the tank was modulated so that pressure inside the container remained within about ±10 pascals of the room environment. The calibration results indicate that the computer software/flow meter was accurate to within 3%, and the data was extremely linear (r2 = 0.9915 fit to straight line indicates little scatter in the experimental data).

Using the calibrated flow meter and a tracer gas analyzer (either our Bruel and Kjaer 1302 or Miran 101), the air flow rate through a duct, air handler, or exhaust system could be measured quickly. On a test building, for example, we were able to measure the flow rate through six roof-top package units in less than one hour. By comparison, measuring that same rate using a duct blaster or blower door would have required considerably more time. Depending up the air handler configuration and outdoor air flow rate, an air flow hood could provide a fast and relatively accurate means of measurement. Using a capture tent with a calibrated blower might have been almost as fast, depending upon how much time is involved in assembling and moving the tent (see section 3.8.4 for further discussion of the capture tent method).

3.9.4 Calibrated fans

Calibrated fans, such as those used with blower doors and duct test rigs, can be used to measure air flows. They can be installed in capture tents, attached directly to the HVAC system, or used with the building as a capture tent.

Capture tent. Air flows, especially the discharge of exhaust fans or the intake of make-up air fans, can also be measured by means of calibrated fans and a capture tent (polyethylene sheeting on a PVC frame will do). A tent is placed over the discharge of the exhaust fan, for example, and a calibrated fan is mounted into the side of the tent (a flow conditioner may be needed to reduce turbulence at the intake side pressure sensor). A micromanometer measures the pressure in the tent wrt outdoors. The calibrated fan is turned on to draw air out of the tent and reduce the pressure in the tent (recall that the exhaust fan is blowing air into the tent) until the pressure in the tent is neutral wrt outdoors. Air flow through the calibrated fan is then equal to the flow through the exhaust system. Make-up air may be measured in an analogous manner, with the calibrated fan blowing air into the tent. Figure 3.1 illustrates use of a capture tent and calibrated fan.

Calibrated fans mounted directly to the HVAC system. The calibrated fan may be mounted directly to a portion of the HVAC system. flow of an air handler, the following steps the supply plenum as the system normally operates. Then remove a panel from the air handler, attach the calibrated blower to the opening in the air handler, seal off any remaining openings around the calibrated fan, place a barrier at the bottom of the air handler to isolate the return side of the system, turn on the air handler, then turn on the calibrated fan and increase its speed till static pressure in the supply plenum is equal to that measured previously. The air flow through the calibrated fan is equal to the air handler flow rate (Figure 3.2).

Building as a capture tent. The building can also be used as a capture tent. Look at an example of measuring exhaust fan flow in a restaurant. With the exhaust fan(s) turned on (all other air moving equipment turned off), the pressure in the building is measured (this could be -60 pascals or greater in many restaurants).

The calibrated fan is turned on to blow air into the building until the pressure in the building is neutral wrt outdoors. The air flow through the calibrated fan is equal to the exhaust fan flow rate. (Note this method works well if the building is tight or the exhaust fans are large. Restaurants may be tight and often have large exhaust fans).

Note, however, that this is the flow rate that occurs when the building is at neutral pressure. If the normal operating pressure (NOP) of the building is much different
from neutral, -10 pascals for example, then the measurement of exhaust or makeup air may be different than that which actually occurs, because building depressurization will reduce exhaust fan flow and increase make-up air flow. To correct for this, the calibrated fan is turned on to move air into the buildinguntil the NOP is reached (-10 pascals); record the flow rate through the calibrated fan (say 3600 cfm). Now calculate the air flow rate into the building due to the -10 pascals depressurization using the airtightness curve developed from the building airtightness test. (Example: say C 300, n .65, then q would be equal to 1340 cfm at -10 pascals). Exhaust fan flow is equal to flow through the calibrated fan plus the building leakage; 3600 cfm plus 1340 cfm = 4940 cfm.

The flow of make-up air can be measured in an analogous manner. Turn on only the makeup air fan and use a calibrated fan to pull air out of the building until reaching NOP. Makeup air flow is then equal to the air flow through the calibrated fan minus the air flow through the building envelope (this assumes the NOP is negative). Alternatively, the make-up air flow can be measured with the make-up air and the exhaust fans operating simultaneously (note that the make-up air flow is usually less than the exhaust air flow). If the building is at its NOP, then make-up air is equal to exhaust fan flow minus building leakage at NOP. If the building is depressurized beyond its NOP, then turn on the calibrated fan to blow air into the building until it reaches its NOP. The make-up air flow is equal to the exhaust fan flow minus building leakage minus calibrated fan flow. Note also that in this situation, the calibrated fan flow is equal to OAflowT (the sum of outdoor air flow and return leaks that draw air from outdoors).

4. Project Findings

Air flow across a building envelope is a function of hole size and pressure differential across that hole. Without a hole there is no air flow. With no pressure differential, there is (almost) no air flow. It is the combination of holes and driving forces which create most uncontrolled air flows in buildings.

Mechanically induced pressure differentials are a function of the net air flow into a space and the airtightness of that space. This applies to occupied and unoccupied building spaces, building spaces used as ducts, and ducts themselves.

In commercial buildings, we expected mechanically induced pressure differentials to be the big story. And certainly they are. However, a big concern, and a surprise to most of us, emerged -- that many small commercial buildings have big holes, especially in the ceiling plane, and that these big holes may be the source of substantial energy problems. A significant number of commercial buildings are tighter than most Florida homes (12 of 70 buildings have ACH5O less than 5). On the other hand, a large number of commercial buildings are much leakier than almost all Florida homes built in the last 20 years (26 of 70 buildings have ACH5O greater than 20). On the whole, central Florida commercial buildings are considerably more leaky than central Florida homes.

Also of considerable concern is the magnitude of problems associated with ceiling insulation systems. Many ceiling insulation systems are functioning poorly, because the insulation is missing, has been moved around, or is being by-passed by uncontrolled air flow. Following is a discussion of the major findings of this project.

4.1 General Building Information

A large portion of the project findings are contained in a single table contained in Appendix B. The following general summary can be presented for our sample of 70 central Florida small commercial buildings. Average floor area was 5030 square feet. Five buildings were manufactured (non-metal), either an office trailer or a modular office space. Four buildings were metal. Twenty buildings were frame or predominantly frame. The remaining 41 buildings were masonry or predominantly masonry (Figure 4.1).

Six buildings are on crawl spaces. The remaining 64 are slab-on-grade construction.

Five are two story. The remaining 65 are one story buildings.

Nine buildings have no attic or ceiling space. The bottom of the roof deck, in most of these cases, is the ceiling of the occupied space. Tn two cases, the space above the ceiling is a warehouse. In other words, this is an office space in an unconditioned warehouse. Fifteen buildings have attics. The remaining 44 buildings have a ceiling space above the ceiling.

There seems to be confusion among contractors about whether ceiling spaces should be vented and about construction details at the eaves. It is common, for example, to install perforated soffit facia while attempting to block off the soffit space from the ceiling space by insulation batts. The insulation batts, of course, do not stop air flow so the ceiling space is unintentionally vented. In general, it appears that small commercial buildings in Florida contain a hybrid of residential and commercial construction materials and methods which often do not make sense from the perspective of heat flow and air flow control.

4.2 General HVAC Information

Cooling capacity. Cooling capacity in commercial buildings is greater than in residential buildings. Total cooling capacity per building ranged from 2 tons to 129 tons, with an average of 16.9 tons. Number of air conditioning systems ranged from one to eight per building, with an average of 3.1 units per building. While Florida residences typically have 1.5 to 2.0 tons of air conditioning capacity per 1000 square feet floor area, the commercial buildings in our sample averaged 3.38 tons per 1000 square feet. They ranged from 1.54 tons per 1000 square feet to 11.1 tons per 1000 square feet. Figure 4.2 presents a breakdown of cooling capacity for the 66 buildings for which cooling capacity is known.

The three buildings with the highest capacity per floor area are fast-food restaurants. Cooling capacity in these three averages 9.8 tons per 1000 square feet. Cooling capacity in all eight buildings which are restaurants or contain restaurants (and for which we know cooling capacity) is 6.1 tons per 1000 square feet, or twice the capacity of the non-restaurant commercial buildings in this study. Interestingly, the fourth largest capacity per floor area is in a 1512 square foot dentist office with a 9 ton air conditioner.

Air handler location. Air handlers are located in mechanical rooms in 13 buildings and in mechanical closets in 16
buildings. We define a mechanical room as a space containing mechanical equipment (air handlers, etc.) in which persons can move around. A mechanical closet is smaller, too small to allow persons to move around or in some cases even enter. Air handlers are in attics in four buildings, in ceiling spaces in two cases, and in the occupied space in six instances. Air handlers are outdoors in five cases and on the roof in 19 cases. Air handlers were in unconditioned warehouses in three buildings (Figure 4.3). (In eight buildings air handlers are located in two locations. The numbers reported in this paragraph consider only the dominant location for each building.)

Ductwork location. Duct systems are generally located in the ceiling space or attic space. Some are located outdoors and some are located in a warehouse space. The attic space and outdoors are always unconditioned. Some ceiling spaces are conditioned and some are unconditioned. Some warehouses are conditioned and some are unconditioned. In two buildings, the ducts are primarily outdoors. In 48 buildings, ducts were located in unconditioned building space, either in an unconditioned ceiling, an attic, or an unconditioned warehouse. In 17 buildings, ducts were located in conditioned space. Conditioned space is defined as either the occupied space or another building space that is inside both the air barrier and thermal barrier of the building. When ducts are located in the conditioned space, penalties from duct leakage are greatly minimized.

Types of ducts. Three types of duct materials dominate air distribution system construction. Nine systems were entirely metal. Twenty-one systems were entirely ductboard. None were entirely flex duct. Hybrid systems were the norm. (In the following, the dominant duct type is listed first.) Nine systems were metal with ductboard. Three were metal with flex duct. Twenty-three systems were ductboard with flex. One system was flex with ductboard. Two buildings had no ductwork (Figure 4.4).

Building cavities as ducts. Building cavities are also used as portions of air distribution systems. In Florida single family residences, it is common for air handler support platforms to be used as part or all of the return air ductwork. Wall cavities, panned floor joists, mechanical closets, dropped ceiling cavities, and spaces below staircases are also used as ducts in homes. Since these building cavities are almost always very leaky, they represent a large portion of all duct leaks in existing Florida residences. In total, it is estimated that 60% of all duct leak air flow in Florida residences occurs in building cavities used as ducts.

It is also common to use building cavities as ducts in small commercial buildings. In 34 of 70 buildings, one or more types of building cavities are used as ducts (Figure 4.5). In eight buildings, the mechanical room is used as a return plenum. In six buildings, a mechanical closet is used as a return plenum. In most of these 14 cases, the ceiling of the room or closet is t-bar, and in many of those cases the space above the mechanical room is unconditioned attic or ceiling space. Since t-bar ceilings are very leaky, it is common for these mechanical rooms or mechanical closets to experience 20% to 40% return leakage (20% to 40% of the systems return air is originating from leakage through the ceiling or other leak pathsways from unconditioned space).

In five buildings, wall cavities are used as ducts or plenums. In eight buildings, the ceiling space is used as a return plenum. In 11 buildings, air handler support platforms are enclosed to form a return plenum, similar to what is found in many residences. In two buildings, chases are used as return ducts. In another building, the space between two dropped ceilings (one above the other) was enclosed to form a return plenum. In this case, one of the ceiling tiles in the upper ceiling had been removed and as a result there was an enormous return leak drawing from an unconditioned ceiling space (CFM25 for that leak site alone was about 4000). Even if all the tiles had been in place, there would still be considerable leakage from the unconditioned ceiling space because t-bar panel construction of the top portion of this plenum was very leaky. (Note that while 34 buildings use building cavities as ducts, the total of the various building cavity types comes to 41 because more than one building cavity type exists in 7 buildings.)

In general, use of building cavities as ductwork is poor practice. Building cavities virtually always leak, and in many cases the leakage can be very severe. Ceiling spaces used as return plenums can be a significant exception to this general rule. If the negative pressure in the ceiling space is fairly small (which is what we observed in most builings) and the ceiling space is reasonably tight to unconditioned spaces (which we found in some cases), then leakage associated with use of the ceiling space as a plenum can be relatively small. It is possible to design the ceiling return plenum to operate at neutral pressure with respect to outdoors. If outdoor air pressurizes the building to +2 pascals, for example, and the plenum runs at -2 pascals with respect to the occupied space, then the plenum will be at neutral pressure with respect to outdoors. Consequently, the plenum can experience next to no duct leakage. (Note that we have observed ceiling return plenums in several buildings operating at positive pressure with respect to outdoors, so in effect these return plenums are experiencing “supply leaks”.)

On the other hand, if ceiling space return plenums are rather leaky to outdoors and are significantly depressurized wrt outdoors, then use of the ceiling space as a return plenum will produce substantial air distribution system leakage. Use of the ceiling space as a return plenum can also be poor practice if the roof has little insulation R-value.

Outdoor air. Twenty-seven buildings have outdoor air. “Outdoor air” is air intentionally drawn into (or pushed into, if there is a dedicated outdoor air fan) the return side of the air distribution system and is generally designed to provide ventilation for occupants in the space. The other 43 buildings do not have outdoor air. They rely on naturally occuring infiltration, leakage in the distribution system, or operation of exhaust fans to bring in ventilation air.

Outdoor air only occurs when the air handler is on. Therefore, when the air handler is off, either because the system has been turned off, or the air handler has cycled off with the compressor, outdoor air ceases, building pressure will decrease, and the building ventilation rate may decrease.

Make-up air. Five buildings have make-up air. “Make-up” air is mechanically induced air blown into the building simultaneously with the operation of the exhaust fans. It is filtered, but in most cases (and in all five cases found in this study) it is unconditioned. Typically the same control switch activates both the make-up air and the exhaust air. Make-up air is designed to reduce the air flow imbalance on the building and thereby reduce space depressurization and space conditioning energy use. Typically make-up air is designed to be 75% to 80% of the flow rate of the exhaust fans (Gaylord Industries, Inc. Smoke Pollution Control Ventilator Model CG-AB-SPC technical bulletin).

As can be seen in Table 4.1, make-up air is substantially undersized in the five restaurants that have make-up air, averaging only 50% of exhaust air flow. Outdoor air is insufficient to make up the balance, averaging only 12% of exhaust air flow. Combined, make-up air and outdoor air represent only 62% of total exhaust air. As a consequence, these buildings operate at negative pressure.

TABLE 4.1.
Building air flow and pressure imbalance in five commercial buildings which have make-
up air and which contain restaurants. Note that while make-up air operates when the exhaust fans operate,
outdoor air may or may not occur when the exhaust fans
operate. Therefore, the net air flow balance may actually be worse than indicated.

In four other buildings containing restaurants, there is no mechanically induced make-up air. In one case, the kitchen exhaust fans are not operated at all (building #19). In another, an exterior kitchen door is left open throughout the day thereby reducing building depressurization from -26 pascals to -4.6 pascals (building #62). In the third, the building is depressurized to -1.8 pascals and passive make-up air comes primarily through the leaky t-bar ceiling from a residential-type attic (building #69). In the fourth, the exhaust fan operates 8 to 10 hours a day. After the duct leaks were repaired, the exhaust fan caused depressurization of -1.8 pascals and most of the passive make-up air comes through a leaky interior knee-wall facade above the bar (building #3 1). (See section 5.5.9 for additional discussion of building #31.)

Exhaust fans. Forty-three buildings have exhaust fans that discharge air from the occupied space to outdoors. There are a few additional buildings that have exhaust fans that discharge air into ceiling spaces that are within the air barrier of the building. Operation of these latter fans does little to affect building ventilation rates or pressure differentials. The size of exhaust fans (building total) in these 43 structures varies from as little as 44 cfm to as much as 10,606 cfm and averages 1767 cfm. In some buildings the exhaust fans operate virtually whenever the building is occupied, including restaurants, sports facilities, and some other buildings. In others, the exhaust fans are manually operated, either wired into the light switch or on a separate switch, and their operation time is then a function of user interaction.

4.3 Ducts Are Leaky

Testing of duct system airtightness was done in 46 of the 70 buildings. In a majority of these buildings, ducts were found to be very leaky. On average, CFM25 (air flow through leaks when the ducts are depressurized to -25 pascals) was 1209, or 341 CFM25 per 1000 square feet of floor area. By comparison, leakage in 100 central Florida residences (old and new together) was only 116 CFM25 per 1000 square feet, or about one-third as much (Cummings, Tooley, and Moyer, 1991). Note that the ducts measured for airtightness excluded ceiling return plenums in all but one case (building 17).

One factor that affects the amount of duct leakage is that commercial buildings have about twice the installed cooling capacity per 1000 square feet, so ducts in commercial buildings should have about 45% greater surface area. Nevertheless, even considering that difference, duct leakage in these small commercial building is disproportionately much greater than in residences. Note that SMACNA (Sheet Metal and Air Conditioning Contractors Natiohal Association) has set a leakage standard for ductwork that is much tighter than what is found in residences or commercial buildings. For Class 6 ductwork (ductboard, flex duct, and other sealed ducts), the standard expects 1.2 CFM25 per 100 square feet of duct surface area. For a typical residence, this might be 3 CFM25 per 1000 square feet of floor area. In commercial buildings, this might be 5 CFM25 per 1000 square feet of floor area. Therefore, duct systems in small commercial buildings are about 70 times more leaky than the SMACNA standard.

While the excessive leakiness of commercial duct systems is evident, the impact of duct leakage on energy consumption is not as clear. In some buildings, we expect that duct leaks have little impact on energy consumption when the ducts are fully inside the building. When they are outside the building or in an attic, the energy penalties should be quite large. However, when the ducts are partly inside the building, that is, they are inside either the primary air barrier or the primary thermal barrier, the energy impacts are less certain. We anticipate that a significant amount of energy lost from the duct leaks will be recovered in this scenario.

Energy savings from repair of UAF was examined by repairing UAF in 20 of the 70 buildings. Energy consumption in these 20 buildings decreased by an average 15%. There are still many questions to be answered about the savings that are likely to result from repair when the ducts are in. various environments. The candidates selected for energy monitoring generally had ducts that were outside the conditioned space and therefore would be likely to experiencc significant energy savings.

A number of factors dictate what, the energy costs of duct leakage will be. Not all duct leaks are created equal, even those that are the same size. The impact of duct leakage depends upon whether the ducts are completely “inside” the building, partially “inside” the building, outdoors, or -- in the worst case --‘ in a ceiling or attic space that is outside the air and thermal barrier ‘of the building. By “inside the building” we mean inside the primary air barrier, inside the primary thermal barrier, or inside both the air barrier and the thermal barrier. If the duct leaks occur in a space that is outside the building air and thermal barriers, then only a small portion of the energy lost by duct leaks may be recovered. If the ceiling space is inside the air and thermal barrier, then we expect that the majority of lost energy is covered to the occupied space.

Note that being located in an attic is even worse than being located outdoors. Attics are generally hotter than outdoors, especially during the portions of the day when air conditioning system operate the most. Therefore, any air drawn into duct leaks will be hotter (and generally have a higher dewpoint temperature) than outdoors.

Another factor to consider is the direction of mechanically driven air flow. Consider this example. Building 35; a fast food restaurant serving chicken, had a completely disconnected supply duct discharging approximately 500 cfm of air into the ceiling space. Insulation batts were located on top of the ceiling tiles. Kitchen exhaust fans were drawing about 9200 cfm of air from the restaurant’ while, on average, make-up air fans and outdoor air were discharging about 5800 cfm into the restaurant. The net air flow imbalance caused the entire restaurant to be depressurized to -43 pascals. This negative pressure was sucking air into the building from all directions, including from the ceiling space. Therefore, the approximately 500 cfm of supply air that was being dumped into the ceiling space would be almost immediately drawn into the occupied space. What portion of the lost energy was being recovered? 25%? 50%? 75%? 100%? We do not know. What would be the energy savings that would result from repairing that large duct leak? Again we do not know. Before retrofit programs can effectively determine which duct systems should be repaired, more investigation needs to. be done to understand the energy losses ‘associated with duct leakage in various types of ceilings spaces.

Consider another example, building 69. This is a convenience store with a small kitchen. A kitchen exhaust fan draws 1546 cfm from the building and depressurizes the space to -1.8 pascals. Since nearly all of the building leakage is in the ceiling, nearly all of the air drawn into the building comes from the vented, residential style attic. Since this air may be .11 OF or hotter in the summer, the’ energy impacts will be great. Consider, however, if make-up and outdoor air were added to the building so that it ran at a slight positive pressure, say +0.5 pascal. Air would then be passing from the occupied space into the attic. Under this new condition, cooling energy use would be much reduced. Savings are
greatest from make-up air, because this air does not need to be conditioned before being injected into the building and when properly designed, a large proportion of the make-up air is drawn directly into the exhaust air stream and is not distributed more generally into the occupied zones. Savings are also significant with outdoor air (compared to drawing air from the attic). Both outdoor air and air drawn in from the attic must be conditioned, but the air from outdoors is cooler and generally drier than air from the attic.

4.4 Ceilings Are Leaky

Commercial buildings, on average, are leaky, and the primary location for building leakage is the ceiling. ACH5O, the air exchange rate of the building when depressurized to -50 pascals by a blower door, averages 16.7, about 50% greater than Florida residences, and about 150% leakier than new Florida residences. Not only are these commercial buildings more leaky than residences, the range of leakiness is much greater. As can be seen in Figure 4.6, there is a bi-polar distribution of building airtightness, with peaks at the 5 to 10 ACH5O bin and at the 20 to 25 ACH5O bin.

This twin peak occurs, we believe, because of t-bar suspended ceilings which are used in the greatest majority of commercial buildings. (In our sample, 54 of 70 buildings have suspended, t-bar ceilings. Six of the remaining 16 have no ceiling other than the bottom of the roof deck. Six have gypsum board. Four have some other form of ceiling tiles but are not suspended on t-bar framework.) These suspended ceilings are very leaky, because of cracks, joints, penetrations, openings, vented light fixtures, and duct leaks.

Since the ceilings are almost always leaky, it is what is happening above the ceiling level that determines in most buildings whether they are very tight or very leaky. In almost all cases, a very leaky building is leaky because the space above the ceiling has large leak pathways to outdoors, sometimes intentionally and sometimes unintentionally. If the building is very airtight, it is because this space above the ceiling is quite tight to outdoors.

There appears to be some confusion on the part of builders and architects about whether ceiling spaces should be vented or not. It is common, for example, for small commercial buildings to have vented soffits under the eaves (with a flat or sloped roof), openings from the eaves into the ceiling space, and insulation batts filling those openings. Since the batts do not create a tight air barrier, air can flow through and around the batts. It is not clear whether the builder intends for ambient air to flow through the ceiling space or not. Based on our understanding of energy and moisture control issues, we suggest that the air and thermal barriers of the building be located at the roof deck and that the ceiling space be made airtight with respect to outdoors.

Single family homes in Florida have become increasingly more airtight over time, as can be seen in Figure 4.7 (Cummings, Tooley, Moyer, 1991). By contrast, commercial buildings in Florida do not show a significant trend toward airtightness (Figure 4.8). Those built in the past 10 years are only slightly more airtight than the entire sample; 14.5 ACH50 in the past 10 years versus 17.7 ACH50 for buildings more than 10 years old. Commercial buildings built in the past 10 years are twice as leaky as home built in the past 10 years. Homes built since 1985 have an average ACH50 of about 7 while commercial buildings have an average ACH50 of 14.5.

Some attempts were made to measure ceiling airtightness. ifi one facility, an office space is located inside a warehouse. Three exterior walls face outdoors, one wall faces into the warehouse, and the ceiling is exposed to the warehouse. Blower door tests were done, once with the office at -50 pascals and the warehouse at neutral (both with repect to outdoors) and a second time with the warehouse also depressurized to -50 pascals. By sealing obvious leak sites in the one wall facing the warehouse and assuming that remaining leakage in that wall was negligible, we calculate that ceiling leakage is 5.49 CFM5O/ft2. For the entire 360 square feet of ceiling area, CFM5O equals 1976, or 86% of total CFM5O. Measurements found that the ceiling surface area breaks down to 76.7% suspended ceiling (t-bar and tiles), 17.8% fluorescent light fixtures, 4.4% supply grills, and 1.1% return grill. Ceiling airtightness was measured with the duct registers and grill sealed off, so duct leakage is not included in this number. Results from testing in several other buildings indicate similar ceiling leakage.

4.5 Ventilation and Infiltration Rates

In most residences, ventilation is not provided intentionally. It results from natural infiltration (wind and stack driven infiltration), duct leakage, operation of exhaust fans and exhaust equipment, and to a small extent the operation of combustion equipment. Tests in 160 central Florida homes found natural infiltration (air handler and exhaust systems turned off) of 0.28 ach (air changes per hour). When the air handlers were turned on, infiltration increased to 0.91 ach (Cummings, Tooley, and Moyer, 1991), indicating that homes have substantial duct leakage and that mechanically driven infiltration dominates.

As indicated in the previous section, commercial buildings are quite leaky. This shows in passive infiltration rates. In the 55 buildings in which an “air handler off’ test was done, passive ventilation averaged 0.43 ach. For these same 55 buildings, ACH5O averaged 17.66. The ratio of ACH5O to natural ach is 41. In a sample of 99 central Florida homes, the ACH5O to passive ach ratio was 39 (Cummings, Tooley, and Moyer, 1991). In a sample of 70 new Florida homes, this ratio was 35. The general rule for central Florida buildings, then, is that passive infiltration can be predicted (on average for a large sample) by dividing ACH5O by 40, for both residences and small commercial buildings. (Note: it must be recalled that passive infiltration varies considerably depending upon wind and temperature conditions at the time of the test.)

In commercial buildings, ventilation is often provided intentionally. Twenty-seven buildings have outdoor air, which intentionally ventilates the space when the air handlers operate. Forty-three buildings have exhaust fans that discharge air to outdoors. In some buildings the exhaust fans operate virtually all the time that the building is occupied, including restaurants, sports facilities, and some other buildings. In others, the exhaust fans are manually operated, either wired into the light switch or on a separate switch, and there operation time is then a function of user interaction.

When the air handlers are operating and exhaust fans are in their normal operating mode, the ventilation rate averaged 1.24 ach in 68 buildings (unavailable in two buildings). Note, however, that air handlers do not operate continuously in a majority of these buildings, so the actual “as operated” ventilation rates will be somewhat smaller.

Note that “AH on” ventilation rates are much lower in the 55 buildings for which a passive infiltration test was done. The 13 buildings for which a passive infiltration test was not done are, to a large extent, restaurants and other buildings with large exhaust fans, outdoor air, and make-up air. It is more difficult to obtain the “AH off’ test in these buildings because the business owners want
the equipment on all the time, especially the kitchen exhaust fans. AHon ventilation in the 55 buildings was 0.85 ach. Passive and mechanically induced ventilation varies substantially from one building to another. Figure 4.9 graphically displays the distribution of ventilation for the sample of 55 buildings for which AR off infiltration tests were done. Figure 4.10 graphically displays the distribution of ventilation for the sample of 68 buildings for which AH on ventilation tests were done.

Figure 4.11 illustrates passive infiltration rates versus building airtightness and shows that in general tighter buildings have lower passive infiltration rates.

4.6 Air and Thermal Barriers Are Often Separate

In small commercial buildings, the primary air barrier and primary thermal barrier in the ceiling/roof plane are often separated. The primary air barrier may be located at the roof deck (38 of 68 cases) or at the ceiling plane (30 of 68 cases). The primary thermal barrier may be located at the roof deck, on top of the ceiling tiles, or suspended from the trusses. (By ‘primary -- referring to air or thermal barriers -- we mean that this air or thermal barrier provides greater resistance to air or heat flow compared to any other barriers that may be in series.)

By contrast, the ceiling air and thermal barriers in residences are almost always in the same plane. The ceiling gypsum is the air barrier and the thermal barrier (insulation) lies on top of the gypsum.

In small commercial buildings we have identified seven different ceiling/attic space configurations. These configurations are defined by the location of the air and thermal barriers. These configurations also define the type of conditions that likely exist in the space above the ceiling. These configurations are illustrated in Figure 4.12.

Table 4.2 illustrates that buildings with leaky or ventilated ceiling spaces are generally considerably more leaky than those with no ceiling space or those where the roof deck is the primary air barrier. Figure 4.12 illustrates the seven identified ceiling barrier configurations. As indicated earlier, commercial buildings are generally quite leaky, and the most important determinant is leakiness of the ceiling space to outdoors.

Table 4.2
Types of ceiling spaces as defined by the location
of the air and thermal barriers in 68 buildings.

In ceiling barrier configurations 1, 2, 3, and 6 (total of 38 buildings), the roof deck is the primary air barrier. That does not mean the roof deck and exterior walls above the ceiling level are completely airtight or even very airtight. It just means that they are tighter than the suspended ceiling. Those buildings where the roof deck is the primary air barrier have an average airtightness of 11.0 ACH50.

In ceiling barrier configurations 4, 5, and 7 (total of 30 buildings), the suspended ceiling (or in a few cases the gypsum board ceiling) is the primary air barrier. Those buildings where the ceiling is the primary air barrier have an average airtightness of 25.9 ACH50. As indicated earlier, this greater leakiness is caused by the leakiness of the ceiling space (or attic space) to outdoors.

4.7 Problems That Result From Separate Air and Thermal Barriers

Ideally, the thermal and air barriers should be located in the same plane, and those barriers should be sufficiently effective to stop most heat and air flow into the building. The heat flow resistance of the primary thermal barrier and the air flow resistance of the primary air barrier should be great enough to minimize space conditioning energy loads. A substantial proportion, however, of the small commercial buildings examined in this study have inadequate thermal and air flow resistance in the ceiling/roof plane of the building. Following is a discussion of the pros and cons of the various categories related to the location of the ceiling air and thermal barriers. (Note: categories in the following discussions are as shown in Figure 4.12.)

Categories 1 and 2. The 10 buildings that fall into category 1 and the 16 buildings that fall into category 2 have air and thermal barriers at the roof deck. In almost all of these cases, the air barrier is effective. Average airtightness for these buildings is 9.2 ACH5O. The thermal barrier is also likely to be effective, because it is located at the primary air barrier, so air transported heat does not have an opportunity to pass through the insulation. In this study, we have almost no information about the insulation levels that exist in the roof construction; all we could see was the bottom of the roof deck and the top of the roof membrane. Nevertheless, we feel that in most of these cases the building envelope (in terms of both air and thermal barriers) is working quite well, especially in comparison to the following categories.

Categories 3 and 6. The six buildings that fall into category 3 and the six buildings that fall into category 6 may have significant problems. The roof deck is the primary air barrier. This means, however, only that leakage from the ceiling or attic space to outdoors is smaller than the leakage across the ceiling. Note that four of the six category 6 buildings have vented attics! Even though the attic spaces are vented to outdoors, the attic is still more airtight (to outdoors) than the suspended ceiling. In other cases, the ceiling space may be very tight to outdoors. Two examples are a video productions office building (building #6) and a fast food restuarant (building #22) which have ACH5Os of 3.7 and 3.9, respectively.

Since the primary thermal barrier is located on top of the ceiling tiles, the space above the insulation may become quite hot on summer days. The extent of the heat penetration into the ceiling space (or attic space) depends upon the thermal resistance of the roof construction. (While the roof may not have insulation materials, nevertheless flat roofs often have significant thermal and mass characteristics related to materials used in its construction.) Heat in this ceiling space can be easily transported into the occupied space by means of duct leakage or air flow through the permeable t-bar ceiling by several types of driving forces, including wind, return duct leaks, supply duct leaks, closed interior doors, and exhaust fans.

Category 4. Like categories 1 and 2, category .4 buildings have the air and thermal barriers located in the same plane, but both barriers are at the ceiling. This is important because in nearly all cases (18 of the 20 cases), the primary air barrier is a suspended t-bar ceiling, and t-bar ceilings are very leaky. Therefore, even though the ceiling is the primary air barrier (i.e., it is tighter than the ceiling space to outdoors interface), it is nevertheless quite leaky.

Therefore, air can easily be transported from the hot ceiling space above the insulation into the occupied space by various driving forces such as wind, duct leakage, closed interior doors, and exhaust equipment. By contrast, if the air and thermal barriers are located at the roof deck, then the ceiling space will stay cool and mechanically driven air flow across the suspended ceiling will have very little impact on building energy use.

Category 5. In this category, which is rather rare, the ceiling space or attic space is fairly well ventilated (such that the t-bar ceiling is the primary air barrier), but the insulation is at the roof deck. The insulation may stop most of the heat flow through the roof into the ceiling space, but air from outdoors is free to flow through this space. The ceiling space is likely to be warm and humid. This may result in elevated cooling energy use and possible moisture problems as humid air comes into contact with cool building surfaces, especially the ceiling tiles.

Category 7. This is very similar to category 4, except the insulation is floating” above the ceiling tiles (attached to the bottom of the trusses). This configuration has all the problems of category 4, but is even worse because in many cases some of the insulation batts have fallen (partly or completely) from the bottom of the trusses onto the ceiling tiles below. This leaves large gaps in the insulation barrier and thus allows air to flow freely through the insulation barrier.

4.8 Insulation Problems

The location of insulation in the ceiling/roof plane of the building has important implications. In 29 buildings, insulation was located at the roof deck, either integral to the roof construction or attached to the bottom of the decking. In 28 buildings, insulation was located on top of the ceiling. In 13 cases, insulation was attached to the bottom of attic trusses and therefore was ‘floating” between the ceiling and roof.

In general, locating insulation at the roof deck is the best choice. In all but one building, this means the air and thermal barriers .are at the same location. As a consequence air flow and heat flow (in the plane of the ceiling) is well controlled. Additionally, this keeps the ceiling space cool and dry which in turn minimizes the energy impacts of other types of uncontrolled air flow which may exist in the building.

Insulation located on top of the ceiling tiles can be a problem, for two reasons.

Insulation attached to the bottom of attic trusses generally has problems. Often insulation batts partially or completely detach from the trusses and fall to the ceiling below (often 6 inches to 36 inches), leaving significant gaps in the insulation system. Even if the batts remain attached to the trusses, air can flow through this thermal barrier. With a ventilated attic above and a leaky t-bar ceiling below, there is great potential for hot attic air to move into the occupied zone and conditioned air to flow out of the building.

Insulation may be located at the roof deck, but the ceiling space is intentionally or unintentionally vented. The insulation may be effective in rejecting the solar load coming through the roof deck, but air transported heat from outdoors bypasses the insulation, so its overall effectiveness is greatly reduced.

4.9 Exhaust Air/Make-up Air Imbalance

Forty-five of the 70 buildings have exhaust fans that discharge air from the building. There are a few additional buildings that have exhaust fans that discharge air into ceiling spaces that are within the air barrier of the building. The exhaust systems flow rates range from a low of 44 cfm to 12,907 cfm per building. Normalized to floor area, flow rates range from 5 cfm per 1000 square feet to 3355 cfm per 1000 square feet. On average, exhaust fans in these 45 buildings move 362 cfm per 1000 square feet out of the building.

Restaurants generally are at the high end of the exhaust fan air flow rate spectrum. In fact, in this sample of 70 buildings, the eight highest exhaust “cfm per 1000 square feet” are restaurants or buildings that contain restaurants (hotels, golf club house, convenience store). These eight range from 411 to 3355 cfm per 1000 square feet, with an average of 1521 cfm per 1000 square feet of floor area. One other restaurant, actually a stop-and-go gas station/convenience store with a kitchen area, has exhaust flow of 375 cfm per 1000 square feet. The remaining 35 building having exhaust fans have an average exhaust air flow of 100 cfhi per 1000 square feet.

Exhaust fans tend to depressurize buildings. The greater the exhaust flow rate and the tighter the building, the greater the depressurization. In some buildings the exhaust fans operate only intermittently and are typically occupant controlled. In other buildings the exhaust fans operate continuously or almost continuously.

Space depressurization may be off-set by outdoor air or make-up air. Outdoor air is generally an intentional return leak drawing air from outdoors when the air handler is operating, and is generally, intended to meet ventilation requirements. In some cases outdoor air is induced by its own blower but still operates only when the air handler operates. Makeup air, by contrast, operates only when the exhaust fans are operating and is intended to make-up for air drawn out of the building by the exhaust fans. It pushes (generally unconditioned) air into the building in proximity to the exhaust fan intakes so that less conditioned building air is exhausted from the building. Its purpose is to reduce space depressurization, allow downsizing of space conditioning equipment, improve indoor humidity and thermal comfort, and reduce space conditioning energy use.

Twenty-five of the 45 buildings with exhaust fans have outdoor air. In most small commercial buildings, outdoor air is intermittent because the air handlers cycle according to load. In other cases, the air handlers may run continuously but have variable speed, and as a consequence outdoor air varies with load as well, In buildings where exhaust fans operate continuously for extended periods, outdoor air is therefore not a reliable means to avoid excessive space depressurization.

Make-up air can be quite useful in avoiding excessive space depressurization. Since it operates whenever the exhaust fans are operating, the building air flow balance can be controlled and predictable. Two problems exist, however. 1) Most buildings do not have make-up air. Of the 45 buildings that have exhaust fans that discharge to outdoors, only five have make-up air. These five are all restaurants. Four other restaurants do not have mechanically induced make-up air. 2) Make-up air flow is substantially less than exhaust air flow. By design, make-up air is set at only 60% to 80% of the flow rate of the exhaust fans, so that the kitëhen area will be depressurized with respect to the dining area. This will cause heat, humidity, and odors from cooking operations to move from the dining area toward the kitchen, and not vice versa.

For example, if the kitchen exhaust fans move 10,000 cfm, the make-up air fan may move only 7000 cfim Thus there is a 3000 cfm net deficit, so the building will be depressurized. In some cases, outdoor air can make-up some or all of the remaining imbalance. As indicated, since the air handlers may cycle with load, outdoor air may therefore operate only a portion of the time and net air flow and building pressure may fluctuate up and down.

In the case study of the chicken restaurant 1 (building #33), air flow and pressure imbalance exists because the four air handlers cycle on and off according to load. As a consequence, building pressure fluctuates up and down in. a stair-step fashion throughout the day as air handlers cycle on and off, ranging from -3 pascals when all are on to -18 pascals when all are off (Figures 13 - 15). During the afternoon hours of hot summer days, most of the outdoor air is operational, and building pressure is in the range of -3 pascals to -8 pascals. During cool morning hours in the spring and fall, three or even all four of the air handlers may be off and the pressure in the building will be in the range of -13 pascals to -18 pascals. Late at night when the thermostats are set to higher levels in order to save energy overnight, all four air handlers (and the outdoor air) shut off and building pressure goes to a consistent -18 pascals. As reported in the case study, this space depressurization has lead to considerable mold and mildew growth on the interior walls, backdrafting of the water heater, and flame roll-out from the water heater on two occasions.

4.10 Variable Air Volume Systems

Variable air volume (VAV) systems impact internal air flows, ventilation rates, and building pressure differentials. Instead of cycling on and off to meet load, these systems vary air flow through individual air terminal boxes and through the main air handler in response to changes in cooling or heating load. In VAV systems, total air flow through the air handlers varies, either because of variable speed drive or inlet vane damper control. In many systems, as a consequence, outdoor air flow rates also vary because depressurization on the return side of the system varies as a function of the air handler flow rate. In other cases, outdoor air may be fairly constant because it has its own blower to induce flow. When outdoor air fluctuates, primarily as a function of load, the building ventilation rate can change and the net air flow into the building can change. In many buildings, exhaust fans operate continuously and at a constant rate. Therefore, it is possible that building pressure can flip- flop back and forth between positive and negative pressure as a function of VAV operation. Our observations are based on testing in three buildings which have VAV systems, including one which was a large 145,000 square foot central Florida office building not tested as part of this project. Additional research into VAV systems and the impacts of their operation upon building air flow balance and pressure differential is needed.

4.11 Failure of Design

In many small commercial buildings, it appears that HVAC systems are not designed in any detail. Rather, they are simply installed by the HVAC contractors. In many cases, layout of ductwork, sizing of exhaust fans, settings of outdoor dampers, and even whether outdoor air is provided is left up to the contractor and/or the installing technicians. As commercial buildings get progressively larger, the proportion with detailed HVAC design and specifications increases.

In other cases, HVAC designs are flawed. Provision of return air, fOr example, is often left to chance. Returns are often located in central zones of buildings with little or no provision for return air to get from closed rooms to those returns. Significant pressure imbalances may result. In many commercial buildings, the ceiling space is used as a return plenum. If the ceiling space is significantly depressurized with respect to outdoors, substantial return leakage may result if the ceiling space is leaky. In other cases, firewalls may separate various sections of the building and restrict the return air from moving from one zone to another and back to the air handler. In some cases, no return pathways are provided through the firewalls. In other cases, return transfer windows are provided (with fire dampers) but the windows are undersized. In other cases, the dampers (which are supposed to be closed only in case of fire) are closed and thus restrict return air flow. Consequently, significant pressure imbalances may result.

In one building one air handler served two zones, and because the return transfer windows were only about 1/3 the required size, pressure in the remote zone went to as high as +17 pascals while pressure in the zone housing the air handler went to as low as -8 pascals.In other buildings, no provision was made for make-up air. Consequently, exhaust fans cause serious depressurization of the building which can result in humidity control problems, microbial growth problems, moisture damage to building materials, and backdrafting of combustion appliances. In one restaurant, no mechanical make-up air was provided and the building was depressurized to -26 pascals. Since this created excessive infiltration and made it difficult to maintain acceptable room temperature and humidity, the business owner responded by keeping the exterior door to the kitchen open throughout the entire day. Even with the door open, the building was depressurized to -4.6 pascals. In another restaurant, a passive make-up vent was located in the ceiling above the pizza assembly counter. Because the make-up air would blast down onto the workers (and a considerable portion of this air was coming from the attic space above), the kitchen exhaust fan was never operated in spite of the fact that it was always very hot because of the large pizza oven.

In some buildings, little or no mechanical ventilation was installed in spite of considerable sources of indoor air contamination. One print shop had neither outdoor air or exhaust air, and measurement of total volatile organic compounds (TVOCs) was measured at 700 ppm (measured by Bruel and Kjaer 1302 multi-gas monitor). By contrast, TVOCs in most other buildings we have measured were less than 4 ppm. In another print shop, there was no outdoor air and only 23 cfm/l000 square feet of exhaust. TVOC levels were also very high, at over 150 ppm. Also in a plastics fabrication facility, no mechanical ventilation of any kind was provided. Since these three buildings have considerable air contamination sources, it is important that adequate ventilation (either spot ventilation or distributed ventilation) be provided in order to ensure acceptable indoor air quality.

In other cases, HVAC components may be specified but important information is left out. The size of the cooling system may be specified, but the amount of outdoor air or whether there should be any outdoor air is omitted. In some cases, return transfer windows may be indicated but the size of the windows may not be specified. Exhaust fans may be indicated, but not how they are to be controlled. Cooling system capacity may be specified, but whether it will be two-stage or whether the air handler blower cycles with the compressor is left to the discression of the contractor. Therefore, these operation and control decisions then go by default to the HVAC contractor and/or his technician.

4.12 Test and Balance Issues

In this study, a number of building failure modes have been detected that relate to test and balance (TAB). The first point is that in a majority of small commercial buildings, it appears that no test and balance is done. The system is installed by the HVAC contractor, who may adjust flows to within reasonable specifications, but no independent TAB is done. Consequently, air distribution air flows and exhaust/intake air flows may be considerably out of balance.

The second point is that outdoor air settings seem to be somewhat random, though it is not clear from our research that the blame rests with the TAB firms. In some buildings, the outdoor air may be sealed off or in other cases it may be wide open and providing excessive amounts of outdoor air. In one dentist office, for example, we found that the cooling system was greatly oversized (a 10-ton unit serving a 1512 square foot building and outdoor air was wide open and providing 911 cfhi, or 32% of the total air handler air flow). Since average building occupancy was 9 persons, this means there was over 100 cfm of ventilation per person. As a consequence, cooling energy use and indoor relative humidity were high.

Several months later we came back to this dentist office in order to monitor the building for energy savings that would result from reducing outdoor air. To our surprise, we found that the large roof-top package unit had been moved about 15 feet from its earlier position, the old sheet metal duct system (located on the roof) had been replaced by a completely new duct system (ductboard ducts completely exposed on the roof with only a mastic coating), and the outdoor air was completely sealed off. This clearly shows that after-the-fact modifications can result in significant system changes that are, of course, beyond the control of both designer and TAB personnel.

The third point is that TAB firms may, in some cases, fail to balance building air flows or may even create unbalanced air flows because there may be little room or inadequate provisions made for proper air flow measurements. The TAB firm may achieve the wrong building pressure because they often rely too greatly on measurement of air flows to determine if the building is at positive or negative pressure. Relying upon air flow measurements to determine building net air flow (whether it is positive or negative) can lead to significant error because of measurement inaccuracies, duct leakage (duct leak air flows are not accounted. for), or air flow imbalances caused .by restricted return air.

Air flow hoods, the primary tool of the TAB industry, often overestimate air flows being discharged into the room from supply registers. This is especially true for smaller registers and those that discharge the air to one side of the hood or the other. Hoods more accurately measure air flow from larger registers (2 foot x 2 foot) and those that discharge in all directions or have a diffuse discharge (e.g., perforated screen with multiple holes). Hoods also tend to more accurately measure air entering return grills and exhaust grills. Since they sometimes overestimate air flow entering the conditioned space and generally accurately measure air leaving the conditioned space, relying upon the balance between the supply and return measurements can lead one to believe that the building air flow balance is positive when in fact in may be negative This bias can cause TAB personnel to sometimes determine that spaces and buildings are at positive pressure when in fact they are depressurized. It may also cause them to adjust the speed of exhaust fans, make-up air fans, and the size of outdoor air openings in order to decrease net air flow into the building under the mistaken belief that the building is at positive pressure.

The fourth point is that the air flow measurement approach to building air flow balance also produces error because duct systems leak. Supply ducts, return ducts, exhaust ducts, makeup air, and outdoor air ducts can all have air leakage. When this leakage occurs within the primary building air barrier, measurements done at grills and registers do not provide an accurate picture of the air flows across the building envelope. Figure 4.16 illustrates potential air flows within a building, including supply, return, outdoor air, exhaust air, and make-up air flows, and duct leakage which may occur in each of these.

It is important to understand how duct leakage can distort the picture obtained from measurements of air flows. Lets look at Figure 4.16 in some detail to understand how duct leakage affects building net air flow. Looking at the exhaust system in specific, note that there are seven different air flows indicated; EAin is air going into the exhaust grill, EA out is the air flow discharging, three possible leak sites on the suction side, and two possible leak sites on the discharge side If one measures airflow entering the exhaust, grill, this measurement will not truly reflect air flow out of the building if EAleaki, EAleak2, EAleak3, or EAleak4 are occurring. If one measures air flow discharging at the roof level, EAout, this is likely to accurately reflect the exhaust air flow leaving the building, unless leak EAleak5 is occurring (duct leak on the discharge side going to outdoors).

These types of leaks occur frequently in commercial buildings. Consider hotels, for example. It is common for a building shaft to be used as part of the exhaust duct system, with a large exhaust fan attached at the top of the shaft. Grills or short ducts go from the rooms to this shaft. Air is drawn from the building through exhaust grills and through leakage from this shaft to the building Test and balance personnel may measure the exhaust flow rate into the grill (typically in the bathroom), but not account for the leakage. Therefore, the actual exhaust rate from the building may be much greater than indicated from the exhaust grill measurements. In one hotel tested in this project (building #68), air flow into the exhaust grills in 40 guest rooms which composed the entire building totalled 1324 cfm. Measurement of air flow from the exhaust fans at the roof totalled 2799, or more than twice as much. In other words, there is 1475 cfm of air flow that is not accounted for. The majority (we believe) falls under the category of EAleaki (air leaking from conditioned space) but there is also an EAleak4 component as well (air leaking from outdoors into the exhaust shaft).

The importance of not accounting for this leakage, from a TAB perspective, is that TAB personnel may conclude that the building air flow balance is positive (keeping the building at positive pressure in hot and humid climates is important for moisture control) when in actual fact the building is operating at negative pressure. This is truly important because the authors have had first-hand experience with TAB contractors who believed they had achieved positively pressurized buildings when in fact the building was operating at negative pressure and the subject building subsequently had serious moisture problems related to uncontrolled air flow. It is also possible, that misinterpretation of true building air flow balance can lead TAB personnel to improperly adjust the flow rate of exhaust fans and outdoor air and cause a building to go to negative pressure when prior to their testing it was in fact operating at positive pressure. The same misinterpretation can occur with make-up air, supply and return ducts, and with outdoor air.

Consider two examples of duct leaks that were occurring in the outdoor air ductwork in one recently tested commercial building. In one case, an outdoor air duct on the first floor stopped short of the grill at the exterior wall of the building leaving a 2 inch gap. As a consequence, about 75% of the “outdoor air” was actually being pulled from the building. In the other case, outdoor air ducts went from two second floor air handlers to panels in exterior walls where there were supposed to be exterior outdoor air grills. However, there were no grills -- just a solid brick wall! Nevertheless, the TAB contractor had done a traverse on the ducts and found flows of 1037 cfm and 897 cfm! In actual fact, this air flow did exjst, but virtually all the air was being drawn from the mechanical room from leaks in the ducts and around the wall panels. (Note that the TAB contractor could not see that there was no grill because metal panels covered the access to where the exterior grills were supposed to be.) Therefore, in this building, outdoor air was being overestimated by over 2000 cfm. As a consequence, the building was operating with less ventilation and more depressurization than was thought based on the TAB report.

Pitot tube traverse measurements are subject to error under two conditions. 1-Measurements will be in error where insufficient straight duct section is available. 2-Measurements may be in error when they do not account for duct leakage just as was true for air flow hoods.

We recommend, therefore, that determination of whether a space is at positive or negative pressure with respect to outdoors or to another zone is best made by direct measurement of pressure differentials using a sensitive manometer. The alternative -- which is much more difficult and often impossible -- would be to carefully measure air flows at the point where they cross the building envelope, thereby accounting for any duct leakage which exists.

The fifth point is that building air flow imbalance can also occur because of return air design problems, such as centrally located returns in conjunction with closed interior doors and restricted return air through fire walls. The resulting pressure imbalance can cause unbalanced air flow across the building envelope and thus cause even greater air flow imbalance. (Stated another way, the high pressure in some zones pushes air out of the building while the negative pressure in other zones draws air into the building, and these flows are likely to be unbalanced.)

The sixth point is that building air flow imbalance can occur when building cavities are used as portions of air distribution systems. Pressure differentials in these cavities, especially negative pressure, can move considerable amounts of air across the building envelope and thus change or even reverse net building air flow and pressure. TAB contractors should examine the level of depressurization that exists in building cavities used as portions of the air distribution systems. For example, if the ceiling space is used as a return plenum, sufficient return grills should be provided in the ceiling so that the ceiling space is not at serious depressurization with respect to outdoors. In most cases in this study, the ceiling plenum was only about 1 pascal negative with respect to the occupied space. In one case, however, the ceiling space was at -6 pascals with respect to outdoors. Simply adding 10 drop-in ceiling grill panels in place of ceiling tiles reduced the ceiling to -0.8 pascals with respect to the occupied space.

Reiterating, it is important for TAB personnel to not only measure and adjust air flows, but to also measure pressure differentials between indoors and outdoors and between various zones/building cavities and outdoors.

4.13 Strip Malls Have Special Problems

Strip malls, where building units are attached to each other in series, represent a special case and have special problems because of connectivity. While adjacent units are separated from each other by interior walls, which are generally fairly airtight, above the ceiling level these adjacent units are often well connected, either by having common attic spaces or significant leak paths through fire walls. Unbalanced air flows caused by mechanical systems can move air between adjacent units and cause energy and air quality problems. In one case, a government office (building #42) located next to a cocktail lounge experienced air quality problems, including tobacco smoke and smells of cleaners resulting from air transport between units. Above the suspended tbar ceilings, the ceiling spaces are well connected to each other by openings in the fire wall that totalled over 30 square feet in size. In addition, the bathroom exhaust fans from the bar discharged into the ceiling space of the office. Once in the ceiling space, air contaminants were transported into the office space by means of large return leaks.

In two other cases, investigated outside of this project, substantial air quality problems were experienced because of uncontrolled air flow in strip malls. In one case, a book store unit located next to a print shop smelled of printing chemicals. Upon investigation, it was found that the return air duct for the book store passed through the print shop and it had considerable duct leakage. In the other case, a national chain pizza parlor was located next to a dry cleaning/laundromat facility. The pizza restuarant moved out within one year because of complaints of smells from the cleaners. The cause was depressurization of the pizza store by the kitchen exhaust fans.

Wind driven pressure differences can also induce substantial pressure differentials and uncontrolled air flows, though these uncontrolled air flows are not unique to strip malls. When wind blows against a building it generates positive pressure on the windward side and substantial depressurization on top of the flat roof. In one strip mall, one-foot round roof vents were scattered intermittently across the flat roof venting the ceiling space to outdoors. On one particularly windy day, opening of the front door to a retail store unit caused five or six ceiling tiles to lift and float several inches above the ceiling level until the door closed (this event was captured on video tape). In another case, a unit with a residential style attic and eve vents located only on one side, pressure in the entire office space went to +40 pascals with respect to outdoors during the approach of a gusty thunderstorm. When strong wind blew toward the vented side of the attic, pressure in the attic went to a very large positive pressure, and since the suspended t-bar ceiling was very leaky, that positive pressure extended down into the occupied space as well.

As seen by the blower door, strip mall units tend to be considerably more leaky than stand-alone buildings. Seventeen of the 70 buildings were attached (i.e., connected to other buildings units), primarily in strip malls. ACH5O averaged 31.2 in 16 attached units (no blower door test was done in one unit) and 12.6 in 53 stand-alone buildings. This means strip mall units are 2.5 times more leaky than stand-alone units, based on the limited sample examined in this study. These 17 units are located in five different strip malls. It is telling that the tightest measurement from all these units was 21.2 ACH5O. Figure 4.17 shows distribution of building airtightness for stand-alone and attached units. The difference between stand-alone and attached units is striking.

An important reason why the blower door “sees” so much leakage in attached units is that they are often well or fairly well connected to each other above the ceiling level. Since the ceilings are almost always quite leaky, the blower door then “sees” leaks that are in ceiling spaces of adjacent units, which in turn are well connected to other ceiling spaces and all the occupied spaces below those ceiling spaces. Since much of the leakage seen by the blower door is “far away”, located in units that may be one, two, or even more units away, the infiltration and energy impacts of this leakage may be somewhat smaller than that which occurs in stand-alone buildings.

5. Energy and Demand Savings from Repair of UAF

Repair of duct leakage in residences has been identified as an extremely cost-effective retrofit measure. In 46 central Florida homes, duct repair reduced cooling energy consumption by 17.2%, or 7.0 kWh per day. The average cost of repair was estimated to be $200. Given estimated annual space conditioning energy use savings of $110, this retrofit pays for itself in less than two years (Cummings, Tooley, and Moyer, 1991).

We expected that repair of duct leakage and other forms of uncontrolled air flow in small commercial buildings would be cost-effective as well. From the sample of 70 small commercial buildings tested, 20 were chosen to be monitored for energy savings from repair of uncontrolled air flow. Cooling energy savings analysis was successfully done in 19 of the 20 buildings. In the remaining building (#12), analysis was complicated by the facts that cooling energy use could not be directly metered (cooling energy was provided to the building by a university-wide chilled water distribution system) and electric reheat was used to control building humidity.

5.1 Selection of Buildings for Repair

Twenty buildings were monitored for energy savings from repair of uncontrolled air flow over single cooling seasons. Two were monitored during the summer of 1993. Seven were monitored during the summer of 1994. Eleven were monitored during the summer of 1995. The plan was to monitor cooling energy use for a period of four to six weeks, make repairs during the middle of the summer, then monitor for another four to six weeks.

Only a portion of the 70 buildings tested were good candidates for repair. By “good candidate” we mean that it appeared that repair could be readily done and would achieve reasonably large energy savings. Based on our assessment of potential savings, it appears that 45, or 65%, of the 70 buildings would make good repair candidates. By contrast, it was estimated from an earlier residential duct leakage study, that duct repair would be justified in about 85% of central Florida residences. Additional research is needed to determine the repair cost and the savings which can be achieved in a wider range of buildings and attempted repairs. Additional development work is needed to identify repair options for the wide range of uncontrolled air flows, such as airtightening t-bar suspended ceilings, adding make-up air, and down-sizing or reducing the run time of exhaust fans. When these additional tasks are completed, then a better estimate of what proportion of buildings are good candidates can be made.

An important task for future research, therefore, will be to develop screening criteria and procedures by which to determine the energy savings which could be expected from specified UAF retrofits and thereby determine which buildings should be retrofitted.

Table 5.1 summarizes some building and HVAC characteristics of the twenty buildings in which repairs were done. Compared to the larger sample of 70 buildings, the 20 monitored buildings were about two years older, 23% smaller, 24% more leaky, and had ducts that were 13% tighter.
When deciding whether considerable energy savings were likely to occur from duct repair, we considered the size of the duct leakage and where the ducts were located. If ducts were located in the occupied space or within a ceiling return plenum, we anticipated little or no savings. If they were located in a ceiling space inside both the building’s air and thermal barriers, then again, little or no savings were anticipated. However, if they were located outside the thermal barrier (insulation on the ceiling tiles), then significant savings would be anticipated from duct repair. [Note that the savings would still depend upon roof thermal factors -- how much R-value exists in the roof construction, the thermal mass of the roof (concrete roof construction, for example, has high mass), the reflectivity of the roof surface, the presence of gravel (which does not transfer heat readily downward), and any shading of the roof surface.]

In the selection process, it was important that the magnitude of the savings be sufficiently large that the signal (energy savings) to noise (occupant use and weather variables) ratio be large enough that clear indication of savings would result. Tn 17 of the selected buildings, we expected significant energy savings. In three others, only moderate energy savings were expected (buildings numbers 45, 59, and 60).

5.2 Repairs

Various types of repairs were done. The primary focus, however, was airtightening of the air distribution system (ADS), including ducts and building cavities used as ducts. The breakdown of repairs was; duct repair (16 buildings), repair of building cavities used as ducts (9 buildings), improved return air pathways (4 buildings), reduction of outdoor air (3 buildings), and tightening of the building shell (3 buildings). These are indicated by “x” in Table 5.2. Note that an attic exhaust fan was turned off in one building, though not shown in the table. Note that two major types of uncontrolled air flow repairs were not addressed in the repair side of this project even though considerable savings could be expected from them. These were 1) airtightening of the ceiling/roof assembly of the building when that portion of the building is excessively leaky and 2) balancing exhaust air flows. The latter could be achieved by some combination of reducing exhaust air flows and increasing intake air flows (make-up air and outdoor air). A significant number of the 50 buildings not included in the repair sample had excessive building leakage or exhaust/intake imbalance which should be remedied. These were not dealt with to any significant extent because of project budget limitations and lack of knowledge about the best methods to implement solutions.

In most buildings, more than one type of retrofit was used. The breakdown of repair packages implemented in this project is as follows.

Among the 20 repair candidates, some repairs that should have been done were not done because of diagnostic error (we did not see that the measure should be done) or because there were insufficient flTnds to accomplish the repair. Repairs not done that should have been done, indicated by !!o? in Table 5.2, include return air (2 buildings), tightening of building shell (5 buildings), adding make-up air (1 building), and reduction of exhaust (1 building).

After repairs were complete, airtightness, pressure differential, and air flow tests were repeated. Table 5.3 summarizes some building and HVAC characteristics of the twenty buildings after repairs were done. Three tables compare the buildings and duct systems before and after uncontrolled air flow retrofits. Table 5.4 compares pre-repair and post- repair building and duct system airtightness and shows percent tightening which occurred. On average, these buildings became 13.9% more airtight. For buildings in which duct system airtightening was done, the duct systems became 58% more airtight. Table 5.5 compares pre-repair and post-repair building and duct system airtightness, building ventilation rates, and return leak air flows. The average building ventilation rate with the mechanical systems operating (ach AHON) decreased by 10% from 1.03 ach to 0.93 ach. The building ventilation rate (infiltration rate) with all mechanical systems turned off increased substantially, from 0.35 ach to 0.46 ach in the 14 buildings for which both prerepair and post-repair numbers were available. In 11 of the buildings, infiltration was approximately equal or less after repair. In three cases, however, post-repair infiltration was dramatically higher (1.13 ach compared to 0.42 ach pre-repair). Substantially higher wind speed associated with hurricane Opal explained the disparity in one case (building #45). Investigation of wind speed and other variables provided no explanation for the substantial disparity in the other two instances. Table 5.6 compares pre-repair and post-repair outdoor air+make-up air, return leak flow, and normal building operating pressure.

Repair of uncontrolled air flow was planned for the middle of summer so that approximately comparable weather would occur during the pre-repair and post-repair periods. Schedule conflicts, however, and the initial lack in availability of candidates caused monitoring starts later than anticipated for some sites. Most repairs occurred in July and August, but three buildings were repaired in September and two were repaired in October. (Note that summer weather occurs May through October in central Florida.)

For each repair candidate, a repair plan was developed, including what would be repaired, what materials would be needed, and estimates of repair time. Repair typically occurred 5 to 6 weeks after monitoring began. A team of two to three persons with considerable experience in repair of residential duct leakage usually did repairs. In two cases, repairs were done by HVAC contractors with the guidance of the research team.

5.3 Monitoring

The following variables were monitored by datalogger on 15-minute intervals.

At many of the sites carbon dioxide levels and indoor pressure differentials were monitored, especially in tight buildings and those in which repair of uncontrolled air flow would significantly reduce ventilation. Outdoor environment conditions of dewpoint temperature and solar radiation were collected at several central locations, generally within 15 miles of the monitored buildings. The number of sensors and meter locations were planned based on size, airtightness, number of air conditioners, and location of thermostats, ducts, and air handlers. Monitoring equipment was assembled partly in the lab and partly in the field.

5.3.1 Data Transfer

All data was stored as 15 minute averages except energy use which was stored as total kilowatt hours every 15 minutes. Data transfer occurred through a modem and phone line. A central computer system called each datalogger daily and downloaded site data to disk storage. Data transfer was scanned for errors by comparing to prescribed value boundaries. If bad data was detected, a second attempt to download data from the datalogger occurred. Suspect data was marked.

Computer programs were created to call each sites datalogger, download data, and plot up to eight graphs containing up to 20 variables every twenty-four hours. These plots were automatically produced overnight and then reviewed daily to see that equipment was working well and to identif’ any unusual circumstances. Such circumstances could be unusual thermostat settings for a particular time of day, air conditioning turned off, faulty sensor, or in a real case, hurricane Erin’s damage to some sensors. When there was an indication of trouble, the site would be visited to repair or replace faulty equipment.

5.4 Energy Savings

Energy savings from the repair of uncontrolled air flow has been determined by comparing cooling system energy consumption for periods before and after repair. In order to filter out variations caused by weather and changes in thermostat settings, cooling energy consumption (kWh) was plotted against the temperature difference between indoors and ambient. Thermostats were controlled in three manners. 1) At seven sites, thermostat settings remained constant over 24 hours (building #s 29, 31, 40, 45, 46, 56, and 61). 2) At ten sites, thermostat settings were raised (typically to 80°F or higher) after hours. 3) At two sites, air conditioners were turned off at the end of business (building #s 2 and 3).

When comparing energy use before and after repair, all data was examined to make sure the most comparable days (in terms of weather and occupancy factors) were used for the analysis. For 15 of the buildings, the analysis includes only weekdays. Weekends were excluded in most cases because of increased variability in building use on weekends. The analyses for the remaining four buildings used all days (buildings 29, 31, 40, and 56). Days with unusual cooling system operation or weather conditions that were not common to both periods were excluded from the data analysis. Examples of unusual operation would be if the thermostat was not raised to its typical after-hours setting or if the air conditioner was turned off for a period during the day. With no control over weather and, in some cases, relatively limited time for monitoring, some sites have less data than desired.

As indicated, cooling energy use is plotted versus temperature difference between indoors and ambient. While ‘ambient” usually means outdoors, in some cases the dominant thermal environment was a warehouse or attic space. In the plastic fabrication office, for example, which is located inside an unconditioned warehouse, energy consumption correlates better to “warehouse minus office space” temperature difference. Attic temperature can also have an strong impact on building cooling load, especially when poor insulation or air transport brings heat into the building. Energy use in realty office 3 (building #3 9), for example, correlates better to attic temperature than outside temperature, in large part because this attic had very little insulation. In these cases, warehouse or attic temperature was used as the reference temperature to calculate the energy savings.

The best-fit lines on the plots are produced by least-squares linear regression. The graphs for 20 building-retrofits are shown in Figures 5.1 - 5.4. Note that the energy savings for realty office 2 (building #37) is presented for two UAF retrofits. “Realty 2 duct” is savings from duct repair only. “Realty 2 fan” reflects cooling energy savings of 3.1 kWh/day and reductions in attic fan energy use of 11.2 kWh/day when the attic fan was turned off. Also note that the sports complex building shows the north and south zones separately, because the two zones are separated by a common wall and a fire wall, the door between the zones was always closed, each zone had a separate cooling system, and the thermostats were controlled separately.

5.4.1 Percent savings and payback calculation

Seasonal energy savings are calculated in the following manner. The least-squares, best fit lines (energy use vs temperature differential) are used in conjunction with 10 year meteorological data from the FSEC weather station to calculate the expected cooling energy use for each day during a typical cooling season. Percent savings is based on a six month period May 1 through October 31. Daily energy use was summed over the entire six month period and divided by 184, the number of days in this period. Percent savings is calculated by dividing the difference between pre-repair and post-repair energy use divided by the prerepair energy use.

To examine the cost-effectiveness of UAF repair, energy consumption for an 8-month cooling season was computed (again based on the kWh vs dT best fit curve and the daily 10- year weather database) from mid-March to mid-November. Adjustments were made to weekend energy use because 10 of 19 commercial buildings were less utilized on the weekends. We found that cooling energy use was 41.7 percent lower in these 10 buildings on the weekends, on average. Therefore, weekend cooling energy use in these 10 buildings was reduced by 41.7 percent for purposes of these computations.

Some heating energy savings will result from these UAF retrofits. But since we did not monitor heating energy use, we do not know what energy savings would be expected. Several factors make this difficult to predict. First, commercial buildings are more fully conditioned during day-time hours when temperatures are warmer. Second, they tend to have greater internal heat generation. Therefore, we have based annual savings only on an 8-month cooling season.

Some businesses pay only for electricity consumption. Others pay demand charges as well.

In order to simplifi the analysis, we have assumed a cost of $0.075 per kWh for all buildings. Simple payback was calculated by dividing the estimated repair cost by the energy savings.

On average, cooling energy consumption decreased from 87.4 kWh/day to 75.1 kWh/day, or 12.4 kWh/day. On average, cooling energy use declined by 14.7% from repair of UAF (Table 5.7). Based on the assumed $0.075/kWh electricity cost, projected annual cooling energy savings are $182. Given that the average projected retrofit cost is $454, simple payback is 2.5 years. This indicates that UAF repairs can be very cost-effective retrofit measures.

5.4.2 Two types of uncontrolled air flow were not repaired

Note that there were two major types of uncontrolled air flow repairs that were not addressed in the repair side of this project. These were 1) airtightening of the ceiling/roof assembly of the building when that portion of the building is excessively leaky and 2) balancing exhaust air flows, by some combination of reducing exhaust air flows and increasing intake air flows (make-up air and outdoor air).

Substantial exhaust imbalance problems exist in perhaps 20% of the total 70 building sample. These imbalances cause excessive space ventilation and elevated cooling energy consumption. In cases where the building has a leaky ceiling and a vented attic space, the energy penalties can be very large and therefore the energy savings from repair can be very large as well. These were not dealt with because of project budget limitations and/or lack of knowledge about the best methods to implement solutions. In one of the buildings which we monitored and in which we made extensive duct repairs (building #31), we tried to get the business/building owner to install make-up air but he felt he could not afford the cost.

Consider another example. A seven-year old convenience store (building #69) has a kitchen exhaust fan which draws 1546 cfm from this store from 5 AM to 7 PM seven days per week, depressurizing the building to -1.8 pascals. Since the floor and walls are tight (all masonry) and the ceiling is very leaky (suspended, t-bar construction), most of the passive makeup air originates from the vented attic space. Since this space is very hot and humid, unbalanced exhaust air adds tremendously to the building cooling load. Some ‘back-of-theenvelop&’ calculations will provide some perspective on the potential savings that may result.

Assume the following:

5.5 Description of Uncontrolled Air Flow Retrofits

Following is a building-by-building discussion of the nature and impacts of the retrofits. Building numbers refer to the master data table which may be found in Appendix B.

5.5.1 Auditorium (#2)

One large 25 ton air conditioning system serves this 5000 square foot, 90000 cubic foot building. Because of the height of the ceiling (18 feet) and size of the supply grills (6 foot diameter) and the fact that the ceiling “stucco’ was asbestos, duct system airtightness was not measured. The supply ductwork was located on the roof in a protective enclosure which prevented access for inspection or repair. Large return leaks (26% return leak fraction) were drawing air from an exterior mechanical room which was vented to outdoors. Repair of the return leaks caused an 8% reduction in cooling energy use from 130.2 kWh/day to 119.6 kWh/day. Repairs were accomplished in five person-hours using $180 in materials. Energy savings pay for the estimated $430 repair cost in 2.1 years.

5.5.2 Dentist 1 (#3)

Two attic-mounted air handlers serve this 2754 square foot building. The ducts are located in the attic space. Substantial leakage existed in the ductwork. Repairs, however, were effective in sealing only 53% of the total leakage. Return leaks were reduced from an average of 5.2% to 2.7%. Many supply leaks remain. It is clear, in retrospect, that additional effort should have been put into duct repair. Ideally, duct repair would have been done with a duct test rig installed on the duct system (registers and grills masked off) so that airtightness could be periodically retested in order to gauge progress. However, in nearly all these buildings, retrofits were done during occupied hours, so the air conditioning system could not be readily turned off during the repairs. To optimize the repair process, it would be best to schedule these operations when the business is closed.

Relative humidity was largely unchanged as a result of the repairs (58% pre-repair and 60% post-repair). Carbon dioxide levels increased from an average 588 ppm to 709 ppm, since repair of the duct leaks diminished the building ventilation rate. Repairs were accomplished in eight person-hours using $50 in materials. Energy savings pay for the estimated $450 repair cost in 2.7 years.

5.5.3 Dentist 2 (#9)

A two-stage, 10-ton air conditioner serves this 1512 square foot building. Unlike most retrofits in this project, no duct repair was done. Rather, the oversized outdoor air was reduced. When originally tested, this roof-top package unit had no outside air damper and total outdoor air flow was 25% of the total 3623 cfh air handler air flow. This equals 911 cfm, which is about 100 cfm per person or 3.9 air changes per hour. With this high infiltration rate, the owner experienced difficulty in controlling humidity in the occupied space. Because the building was thermally very inefficient, the ductwork was located on the roof in the hot sun, and room temperature was typically kept at 72°F, the air conditioning system operated nearly continuously during business hours.

Upon being informed of our findings, the building owner/occupant requested that outdoor air flow be reduced. Therefore, the pre-retrofit monitoring occurred with the outdoor air reduced to 65% of original flow, or 590 cfm. The post-retrofit monitoring occurred with outside air completely sealed. Before retrofit, building ventilation was 835 cfm or 93 cfhi per person as measured by tracer gas decay. This ventilation consists of the 590 cfm of outdoor air plus approximately 232 cfm of return leakage, primarily from rooftop air handler panel leaks. After sealing off the outdoor air, ventilation decreased to 232 cfm (return leaks) or 26 cfh-i per person.

Both carbon dioxide levels and relative humidity were monitored. Carbon dioxide levels during business hours increased from an average 390 ppm with outside air 65% open to 570 ppm with outside air closed. (Note that 570 ppm is well below the 1000 ppm maximum recommended by ASHRAE 62-1989.) This indicates, as was also documented by tracer gas measurements, that ventilation exceeds 20 cfh-i/person. Relative humidity decreased from an average 59% to 54% because of the retrofit. One person-hour is figured for this retrofit and no materials. Cooling energy consumption decreased by 16.2% from 88.4 kWh/day to 74.0 kWh/day as a result of downsizing the outdoor air. Energy savings pay for the estimated $50 repair cost in 0.2 years.

5.5.4 City hall (#20)

Two five-ton roof top package air conditioners serve this 2952 square foot concrete block building. Most of the ductwork was located in the hot and dry ceiling space. It was hot and dry because the roof deck is the primary air barrier and the insulation is on top of the ceiling tiles. About 20% of the ductwork is located on the roof. Substantial duct leakage existed both below and above the roof.
All of the ductwork on the roof, both return and supply, was replaced by an HVAC contractor. Return and supply ducts within the building were sealed by research staff. Duct CFM25 decreased 51% from 1632 to 795. Much of the remaining duct leakage was in supply ducts located in the ceiling space and wrapped in exterior duct insulation. These leaks were not repaired because of the difficult access. Future research should focus on cost-effective methods to airtighten ductwork that is difficult to access and which has exterior insulation wrap. (Note that access to the ceiling space is through ceiling tiles which have insulation batts on top and that the ducts run through a maze of hangers (wires) which support the suspended ceiling.)

In addition to duct repair, the duct access hole (through the roof) was tightened thereby decreasing building leakiness from 7.4 ACH5O to 5.2 ACH5O. Carbon dioxide levels increased slightly from an average 561 to 589 as a result of the repairs. Indoor relative humidity increased slightly from 47% pre-repair to 49% post-repair. Cooling energy consumption decreased by 20% from 137.4 kWhlday to 109.9 kWh/day. Total repair time was sixteen person-hours and $100 in materials. Energy savings pay for the estimated $900 repair cost in 2.6 years.

5.5.5 Health clinic 2 (#21)

Three air conditioning systems served this 2560 square foot strip mall space. Air handlers were located in mechanical closets which were used as return plenums and were depressurized to -12 pascals, -33 pascals, and -17 pascals. Illustration of the second system is shown in Figure 5.5. Because the closet ceilings were suspended t-bar construction and were very leaky, return leaks of 48%, 13%, and 48% were being drawn from the hot attic space above, which was as hot as 120°F. This means that, on average, these three systems were drawing 36% of their return air from the attic. Even with 10 tons of air conditioning, it is not surprising that this building was often uncomfortable. Repairs were made by “hardducting” the system, that is, installing continuous ducting from the air handlers to the return ductwork in the attic, so that the closets were no longer plenums. Significant supply leaks at register connections were also repaired.

Duct CFM25 decreased 91% from 2576 to 227 (most of that leakage was in the return plenum closets). Relative humidity decreased from an average 57% to 52% as a result of repairs. Carbon dioxide levels increased from an average 498 ppm to 539 ppm. Cooling energy consumption decreased by 25.6% from 96.7 kWh/day to 72.0 kWh/day. Twenty person-hours and $300 in materials were used in this repair. Energy savings pay for the estimated $1300 repair cost in 3.9 years.

5.5.6 Sports building (#12)

Four air conditioning systems served this 16,713 square foot university building. Air handlers were located in two mechanical rooms which are located inside the ceiling space. Air handler #1 serves zone #1. Air handlers #s 2-4 serve zone #2. These two zones are separated by a fire wall and doors between the two zones are closed most of the time.

Mechanical room #1 acts as a return plenum for air handler #1. The other three air handlers have “hard-ducted” returns. Air handler #1 draws air from the mechanical room and depressurizes the mechanical room to -10 pascals, and in turn the mechanical room draws air through “windows” from the ceiling space which is a return plenum. Exhaust fans in bathrooms and locker rooms depressurized zone 1 (that portion of the building served by air handler #1) to -1.7 pascals.

The ceiling plenum, which served only zone #1 had large leakage to outdoors. A three inch gap existed where the top of the exterior walls met the sloped roof deck. Considerable return leakage occurred through this gap in spite of the insulation batts which were stuffed into the gap. Not being an air barrier, these batts did not stop the air flow but rather filtered the air as it entered the building.

The uncontrolled air flow retrofit on this building consisted of sealing this gap, using foil- backed fibrous board and class 1 foam. A repeat of the building airtightness test found that 22% of the total building leakage was eliminated by this repair alone. Since cooling energy was delivered to this building by chilled water from a central plant, energy consumption was measured by monitoring air handler air flow rate, temperature drop, and humidity drop.

Energy consumption of the electric reheat was monitored by a power meter, and it was found that reheat was consuming about $10,000 per year on air handler #1 alone!

Cooling energy savings were less than expected. Savings may have been reduced by and analysis made more complicated by the fact that electric reheat was used to control building humidity. The reheat is controlled by relative humidity, such that on cooler summer days the reheat operated more of the time, and therefore the cooling load did not respond as expected to changes in external weather conditions. Savings were also compromised by diagnostic failure. After the previously described repairs (sealing the return plenum) were made and only 7% energy savings were identified, we reevaluated our diagnosis and concluded that this building had an “exhaust air vs intake air” imbalance problem. Exhaust fans operate continuously and have larger air flow than the outdoor air. While we had stopped leak pathways and made the building more airtight, the driving forces of the exhaust fans were still operating and depressurization of zone #1 increased from -1.7 pascals to -3.4 pascals as a result of the this tightening.

The following observations and recommendations were given to the building owner based on the reevaluated diagnosis.

“First, exhaust fans continue to draw 2230 cfm of air from the building, causing the average building pressure to be about -1.2 pascals (area weighted average of zones 1 and 2). This is consistent with the fact that exhaust fans are drawing about 750 cfm more air from the building than outdoor air ducts are bringing into the building (outdoor air = 1477 cfm and exhaust fan air = 2230 cfh).

Second, a significant portion of the supply air from air handler #1 is delivered to zone 2 (about 3700 cfm), and the return air path is inadequate and is returning only about 1200 cfm (this means there is a net pumping of 2500 cfm of air from zone 1 to zone 2). Therefore, zone 2 (about 60% of the building) operates at about +0.2 pascals (all pressures with respect to outdoors), while zone 1 operates at about -3.4 pascals. Therefore, while zone 1 was made much more airtight (about 22% of the total building leak area was eliminated by the repair of the ceiling return plenum), it is now under much greater negative pressure (it was -1.7 pascals), and so considerable ambient air is still being drawn into the building.

Third, air handler #1 receives very little “outdoor air”. While air handlers 2, 3, and 4 receive 457 cfm, 538 cfhi, and 420 cfm, respectively, air handler #1 receives only 62 cfi-n. Zone 1 has 500 cfm of exhaust air operating continuously, or 438 cfm net loss, and consequently, this further depressurizes zone 1.

Recommendations

Based on these observations, the following recommendations are made. The ultimate objective of these recommendations is to reduce humidity levels in the building so that the electric reheat may be turned off. Based on one week of reheat energy use data, we project that AH#1 reheat is using about $10,000 per year (assume $0.08/kwh), and the cooling energy required to meet the reheat cooling load may be about $3500. Therefore, if the humidity can be controlled without reheat, savings from AH#1 alone could be $13,500 per year.

First, we recommend creating a larger return window in mechanical room #1 to allow a greater amount of return air to get back to AH# 1. This window should be about three times larger than the presently existing window. This will reduce pressure imbalance between the two zones.

Second, we recommend that several exhaust fan flow rates be reduced in the building. Following are the current exhaust flow rates (as measured) and what we suggest for reduced flow:

This would reduce exhaust flow from 2230 cfm to 1450 cfm. Furthermore, we suggest that the middle two (locker room bathroom and coachers locker room) be placed on occupancy sensor control with an appropriate (15, 30, or 60 minute) time delay.

Third, we recommend that the outdoor air be increased to air handler #1, to about 500 cfrn. This would bring zone 1 to positive pressure. When these first three steps are taken, nearly all the air entering the building will enter through the outdoor air ducts and therefore, the air will be conditioned by the outdoor air cooling coil before it reaches the air conditioning system. This will produce much improved dehumidification control.

Fourth, we recommend that the flow rate on the air handlers be reduced -- to levels that are just adequate to meet the cooling load on the hottest afternoons. This might mean the flow rates would need to be reduced by 70%. This reduction in air flow would cause the average cooling coil temperature to be colder and therefore remove more moisture.

Alternatively, the control of the chilled water could be changed. Currently, the water temperature is modulated to adjust to the cooling load as sensed by the thermostats. The recommended change would be to cycle the chilled water. In this mode, the water would be sent through the coil at full coldness, and then cycled off when the thermostat senses the setpoint has been satisfied. If this is done, it will be important that the “on” cycles be of sufficient length to allow good dehumidification (say 15 minute on cycles, at the very least) and it will be important that the drain pan not hold much water, because this water will evaporate during the “off’ cycle, thus adding moisture back to the room air.”

By reducing the flow rate and run time of exhaust fans, and increasing the outdoor air so the building operates at a slight positive pressure, building ventilation rates would decrease and dehumidification would improve. Moisture removal from air coming into the building would be enhanced because it would all (or at least virtually all) pass through the cooling coils on the outdoor air intake ducts. This yields a superior latent removal performance compared to trying to remove the moisture after it has already mixed with the room air as it passes through the air handlers -- especially since the air handlers are considerably oversized.

These recommendations had not yet been implemented by the fall of 1995 when monitoring ended. When they are implemented, we expect that relative humidity, which typically runs around 57% with the aid of the electric reheat, could be kept under control without the use of reheat and annual HVAC savings of $15,000 to $25,000 per year could be anticipated.

5.5.7 Recreation center (#25)

Two air conditioners (4 tons and 5 tons) serve this 2708 square foot building. A suspended t-bar ceiling is below a residential-style ventilated attic space. Insulation balls are attached to the bottom members of the attic truss system; some batts are missing and some have fallen. The air handlers are situated on enclosed support platforms located in closets. Return leaks of 12.2% south zone) and 19.4% (north zone) exist in each return plenum because the plenums are built into the closet corners and the walls are partially open to the inside of the plenum. Air is drawn down closet walls from the attic space, from above the ceiling tiles but below the suspended insulation bails. Voids between insulation bails allowed this space to become as hot as 91°F when the attic space above the bails was 118°F. Return leaks created slight positive pressure in the building, pushing room air into the space above the ceiling and thus keeping it from being much hotter than outdoors. Return leak flow was a total of 555 cfm from the two systems. The inside of the support platform return plenums and through-the-wall grill penetrations were sealed using fiberglass mesh and mastic. Supply leaks were sealed at some grill to duct connections.

Duct CFM25 was reduced by 81% from 788 to 154. Relative humidity was largely unchanged; 53% pre-repair and 52% post-repair. Cooling energy use decreased by 17.4% from an average 77.3 kWh/day to 63.9 kWh/day. Repair time was 8 person-hours and materials cost was $50. Energy savings pay for the estimated $450 repair cost in 2.2 years.

5.5.8 University office 1 (#29)

Four roof-top package air conditioners serve this 5040 square foot manufactured office space. Ductwork is located in the ceiling space. Insulation bails are located at the bottom of the roof deck. The ceiling space is warm and humid because it is well ventilated to outdoors. Vented soffit is located at the eaves on three sides of this buildings. Only vertical insulation bails separated this ventilated soffit space (outside the building) from the ceiling space (inside the building), and these bails allowed substantial air flow. Building airtightness was 24.9 ACH5O largely because of the ceiling space ventilation. Two retrofits were employed; airtightening the ceiling space and reducing outdoor air.

After identifjing that the ceiling space was leaky to outdoors, it was decided that tightening the ceiling space would be preferable to sealing the rather leaky ductwork. Though the ducts are located inside the building thermal barrier (insulation attached to the bottom of the roof deck), they are effectively outside the building air barrier. The retrofit consisted of sealing the exterior walls above the ceiling level with fibrous ductboard, mastic, and foam. Airtightening of the ceiling space resulted in reduction of building ACH5O by 74% to 6.4. The second retrofit was reduction of outdoor air from 1296 cfm (86 cfln per person) to 588 cfm (39 cfm per person). In addition, panel leaks on the package air conditioners were sealed with metal tape.

As a result of tightening the ceiling space and reducing the outdoor air, relative humidity in the occupied space declined from an average 53% to 48%. Relative humidity in the ceiling space declined from 50% to 41%. Cooling energy use decreased from an average 211 kWh/day to 170 kWh/day, or 19.3%. Repair time was 24 person-hours and $80 in materials. Energy savings pay for the estimated $1280 repair cost in 2.2 years.

5.5.9 Bar and grill (#31)

Two four-ton air conditioners serve this rather leaky (17.5 ACH5O) 2400 square foot restaurant/bar. The air handlers and ductwork are located in the attic space. The ductwork was very leaky; 655 CFM25. Large return leaks (505 cfmn) were drawing hot attic air into the building. There were substantial supply leaks as well, but return leaks dominated. A kitchen exhaust fan pulled 987 cfm of air from the building throughout most of the day, and there was no make-up air. The building was depressurized to -0.8 pascals when both exhaust and air handlers were operating.

Based on the size of the duct leaks and the fact that they were drawing in hot attic air, cooling energy savings of 30% or more could have been expected. To our surprise, monitored savings were only 10.6%. In hind-sight two problems were identified. First, only 58% of the duct leaks were sealed. Second, we had mis-diagnosed the building UAF problems, or more accurately we only got it half right. We believed that sealing the large return leaks would reduce the amount of hot attic being brought into the building. It turned out that for the most part this was not true because the exhaust fan would continue to draw nearly 987 cfm of hot attic air into the building for 8 to 10 hours per day. (Most of the air would come from the attic since nearly all of the shell leakage of this slab-on-grade, concrete block building existed in the ceiling.) As a result of duct repair, building pressure went from -0.8 pascals to -2.0 pascals. After repair, less attic air was being drawn into return leaks, but more attic air was being pulled into the building by space depressurization.

We believe that substantially greater energy savings could have been realized if make-up air were installed in the kitchen. Alternatively, the large leak paths between the attic and occupied space (above the bar) could have been sealed. This would have reduced the amount of air drawn from the hot attic and increased the amount drawn from the relatively cooler outdoors. To make this second approach even more effective, a passive make-up air vent could be installed in the kitchen, in the proximity of the cooking area and the exhaust fan. This would minimize the impact of the make-up air by causing it to “short circuit” -- go almost directly from the make-up grill into the exhaust intake. This case illustrates how important correct diagnosis is and the importance of taking all uncontrolled air flows into account when specifying repairs.

As a result of these repairs, duct CFM25 decreased 59% from 655 to 272. Relative humidity in the occupied space declined from an average 63% to 57%. Cooling energy use decreased from an average 142.4 kWh/day to 127.3 kWhlday, or 10.6%. Repair time was 18 person- hours and materials totalled $75. Energy savings pay for the estimated $975 repair cost in 4.2 years. Calculations indicate that installation of make-up air could save over $1000 per year.

5.5.10 Realty office 2 (#37)

This was one of the most interesting retrofit cases. An old, inefficient (estimated 6.0 SEER) five-ton air conditioner served this 1845 square foot building. Major breaks and offsets of sections of supply duct located in the attic had occurred as a result of foil tape adhesive failure producing very large supply leaks. An attic exhaust fan was pulling 730 cfm from the attic space throughout the day. It was depressurizing the attic space, which has only two small eave vents, to -10.0 pascals and the occupied space to -10.6 pascals (Figure 5.6). (Note that these pressures were with the air handler also operating and that the supply leaks were depressurizing the occupied space relative to the attic space.) It is also interesting to note that a new roof was put on the building after our initial testing and before the duct repairs were done. A new attic exhaust fan was also installed and it depressurized the attic space to -16.0 pascals and the occupied space to -15.6 pascals. (Note that the attic space is now more depressurized than the occupied space because the supply leaks had been repaired at this stage.)

Retrofits to this building were completed in three phases. The first retrofit was duct repair. CFM25 in the ducts was reduced by 80% from 571 to 112. Cooling energy consumption declined 31% from 99.6 kWh/day to 69.1 kWh/day as a result of this repair. The second retrofit was replacement of the air conditioner by a 12 SEER unit. (This replacement was done entirely at the initiative and expense ($2800) of the owner.) Cooling energy consumption was reduced by 43% from 69.1 kWh/day to 39.7 kWh/day. The third retrofit was turning off the attic exhaust fan. Energy consumption (including the exhaust fan motor) was reduced by 36% from 39.7 kWh/day to 25.4 kWh/day. In total, the three retrofits cut cooling energy consumption 74% from 99.6 kWh/day to 25.4 kWh/day.

In addition to reducing cooling energy consumption by 36%, turning off the attic exhaust fan lowered building pressure from -16.0 pascals to -0.4 pascals and decreased the building ventilation rate from 0.79 ach to 0.24 ach. Carbon dioxide concentration increased from an average 620 ppm to 1150 ppm during weekday afternoon hours. Since concentrations above 1000 ppm is indicative of ventilation below 20 cfh/person, it would be recommended that additional ventilation be installed. One option would be to downsize the attic exhaust fan from its current 730 cfm to perhaps 200 cfm. This would draw hot air from the attic space, slightly depressurize the attic with respect to the occupied space, and ensure that air was flowing from the occupied space to the attic and not vice versa. Note that air entering the occupied space from outdoors has much less heat than air entering from the attic. Therefore, the direction of flow across the leaky ceiling plane is very important from a cooling load point-of-view.

Indoor relative humidity increased from 72% pre-duct repair to 76% post-duct repair. Relative humidity remained at 76% when the air conditioner was replaced. However, when the attic exhaust fan was turned off, indoor relative humidity decreased from 76% to 58%, as a result of the decrease in ventilation rate from 0.79 ach to 0.24 ach.

The energy savings in this building were dramatic. The total retrofit package -- duct repairs, AC change-out, and turning off the exhaust fan -- reduced cooling energy use by an estimated $1114 per year. Given the total retrofit cost of $3180, the simple payback period is 2.9 years. Looking at the individual measures, duct repair pays for itself in 0.7 years, AC change-out pays for itself in 6.5 years, and shutting off the attic exhaust fan pays for itself in 0.2 years (assume $50 service call). Duct repairs required six person-hours and $30 in materials.

5.5.11 Realty office 3 (#39)

This building is a 50-year-old converted residence served by two air conditioners. Ductwork is located in the attic. The air handlers are in one closet and are located on support platforms that act as return plenums. Considerable return leakage (414 cfm as operated) comes from the attic by way of walls. It is interesting to note that the attic insulation had been compressed and badly soiled by rodent infestation, and the air drawn from the attic was contributing to indoor air quality complaints. The attic was a very difficult environment in which to spend any time because of the pungent odor.

A number of flaws lead to considerable return leakage associated with the support platforms that were enclosed to form return plenums. One plenum had two through-the-wall return grills and another return grill drawing air from inside the closet, thus making the closet act as a return plenum. There was a return transfer in the closet door, but it was undersized. As a result, the mechanical closet was substantially depressurized and was drawing air from the attic. The second plenum had one transfer through a wall and some leakage through a hole in the wood floor over a small crawispace. Unsealed return transfers in the closet walls allowed hot attic air to be drawn down the walls from the attic.

Supply leakage was minor and due to time constraints was not repaired (repairs were begun after close of business, two hours were spent cleaning up a condensate leakage problem which interfered with the repair, and post-repair testing work was completed at 1 AM!). Leakage in the closet was also not repaired, due to oversight. Post repair testing showed that only 45% of the duct leakage had been sealed. Return leak fractions decreased from 10.5% to 5.2% on the west system and from 2 1.6% to 5.6% on the east system. Overall return leakage decreased by 66%.

Relative humidity in the occupied space did not change as a result of the repairs, remaining at 56%. Cooling energy use decreased by 7.5% from an average 117.7 kWh/day to 108.9 kWh/day. Repair time was 4 person-hours and $35 in materials. Energy savings pay for the estimated $235 repair cost in 2.0 years.

5.5.12 Safety class (#40)

Two roof-top air conditioners with combined capacity of 9 tons serve this 2460 square foot unit located in the same strip mall as the court office and school supply. This space had some of the worst uncontrolled air flows of the entire sample, yet repairs produced only modest savings. Following is a brief list of the failures.

1. 
   Transfer grills were located in the doors going to the adjacent classrooms, allowing some return air flow. However, since this closet has a suspended, t-bar ceiling and was depressurized to -13 pascals, it was pulling considerable air from the unconditioned ceiling space above. The return for the other air conditioner was located in the hall while the supplies for that system were located exclusively in the classrooms. Thus, the classrooms were operating at positive pressure and the closet and hallway where the return grills were located were operating at negative pressure. Therefore, room air was moving from the room, through the suspended t-bar ceilings and batt insulation, through the ceiling space where it picked up heat, and then down through the insulation and suspended ceilings in the closet and hallway.
2. Each of the air conditioners served both classrooms, causing virtually identical cooling energy use whether one or both classrooms was in use. The east thermostat was located in one of the two classrooms while the west thermostat was located in the hallway, a space that had no supply registers. Since the west system thermostat was not sensing the cooling that was occurring in the classrooms, the majority of the cooling for both classrooms was done by the west system.

Repairs included sealing 51% of the duct leaks, installing return transfer ducts (two 2’x4’ grills in each classroom), adding an additional return duct to the south classroom, installing an additional return transfer grill from the closet to the north classroom, airtightening the tbar ceiling in the depressurized closet, and moving the thermostat from the hall to a classroom. Surprisingly, energy savings were only 17% when 50% savings could have been imagined. Two factors may explain the savings shortfall. 1) Moving the west thermostat shifted much of the cooling systems run time from the west unit to the less efficient east unit. 2) Considerable recapture of lost cooling may be occurring because the uncontrolled air flows occur primarily within the ceiling space. Though it is outside the primary thermal barrier (insulation located on the ceiling), it is inside the primary air barrier (the roof deck) and the roof deck may have significant insulation value (thermal resistance).

Relative humidity in the occupied space declined from 55% to 51% as a result of the repairs.
Cooling energy use decreased by 17.0% from an average 50.8 kWh/day to 42.1 kWh/day.
Repair time was 8 person-hours and $60 in materials. Energy savings pay for the estimated
$460 repair cost in 2.5 years.

5.5.13 School supply (#45)

One 7.5-ton roof top package air conditioner serves this strip mall business. Supply leaks at panel connections and panel knockouts were sealed on the air handler. Difficulties were experienced in gaining access to the remaining duct leakage. Large return leaks were sealed by squeezing one person into the air handler on the return side and coating the interior of the plenum ductboard with mastic. Leakage in the remainder of the supply ductwork was not repaired because finding and repairing all the duct leaks would have been extremely time consuming, largely because of exterior insulation wrap. Finding the leak sites requires removing the insulation wrap and replacing it after sealing air leaks. Since access to the ceiling space requires standing on a ladder and moving the ladder from one ceiling tile location to another, such a process is difficult and time consuming, especially considering the office furniture. The fact that the ceiling tiles have insulation batts and dust on them makes the process even more difficult and makes it likely that fibers and dust will fall onto people and furnishings below. In general, the ceiling access issue represents a significant barrier to efficient duct repair in a significant proportion of commercial buildings.

Repairs to this system decreased CFM25 by 55% from 418 to 190. Relative humidity in the occupied space declined from 48% pre-repair to 46% post-repair. Cooling energy use decreased 6.9% from an average 69.8 kWh/day to 64.8 kWh/day. Repair time was 2.5 person-hours and $30 in materials. Energy savings pay for the estimated $155 repair cost in 2.4 years.

5.5.14 Court office (#46)

Two roof-top air conditioners with combined capacity of 11.5 tons serve this 3735 square foot strip mall unit located next to the school supply store. Sixty-five percent of total leakage in this leaky duct system was sealed. Repairs were made to the return ducts at the grill connections and the east system return was sealed by coating the interior of the ductboard with mastic. The suspended ceiling had not been hung very securely, and over time the panels began to sag. As a result the supply grills pulled away from the ducts in some locations. The only supply side repair was at the supply-register-to-ceiling tile connections, since the ducts were wrapped metal. Even though 65% of the duct leakage was repaired, cooling energy consumption actually increased by 6.6 percent! (This was the only retrofit which did not show energy savings.)

In retrospect we have tried to determine why energy use would increase. The one factor that may provide some explanation is leakiness of the building shell. This office is the leakiest of the seventy buildings tested and has an ACH5O of 52.8. It may be that because the very leaky ceiling and vented roof deck allows wind to move air across it, air exchange may not change significantly as a result of duct repair. Note that the plastics office, which showed an unexpectedly small savings of 4% also had a very leaky building shell (ACH5O=50.0). It may be that leaky buildings will often yield only small energy savings from duct repair. More research is required to answer that question.

Repairs decreased duct CFM25 by 65% from 830 to 292. Relative humidity in the occupied space declined slightly from 49% pre-repair to 47% post-repair. Cooling energy use increased by 6.6%from an average 137.8 kWhlday to 147.0 kWh/day. Repair time was 4 person-hours and $25 in materials.

5.5.15 Metal building company (#53)

Two cooling systems with total capacity of 9 tons serve this 3672 square foot metal building. Ceiling insulation is located both on top of the ceiling tiles and on the bottom of the sloped metal deck above. Some moderate sized supply leaks existed. The large duct leaks, however, occurred because the mechanical room for one of the larger of the two systems was used as a return plenum, and the ceiling of that room is suspended t-bar. Of the total 1453 CFM25 in the two duct systems, 49% existed in the ceiling of that mechanical room.

Repairs consisted of repairing one major supply leak, sealing return ducts, airtightening the t-bar ceiling in the mechanical room, and reducing the mechanical room pressure from -19.1 pascals to -4.2 pascals by installing a louvered door in place of the solid door. Return leak fraction for the mechanical room decreased from 28% to 4% as a result. Cooling energy use declined by 10.8% as a result of the retrofit. Larger savings might have been expected, except that the building has two thermal barriers -- insulation on top of the ceiling and insulation at the roof deck. All of the duct leakage occurs inside the building air barrier and inside one of the two thermal barriers. If the roof deck insulation had not been there, these duct leak repairs might have produced energy savings of 25% or more. Another way of saying this is that this building was saved from the full effects of uncontrolled air flow by having a second thermal barrier at the roof deck.

A general lesson can be stated here. If the ductwork and ceiling spaces are located inside the primary air barrier and thermal barrier, then the consequences of uncontrolled air flow are largely obviated.

Repairs to this system decreased CFM25 by 43% from 1453 to 833. Relative humidity in the occupied space was at 41% before and after repairs. Cooling energy use decreased from an average 87.2 kWh/day to 77.8 kWh/day, or 10.8%. Repair time was 5 person-hours and $90 in materials. Energy savings pay for the estimated $340 repair cost in 2.2 years.

5.5.16 Realty office 4 (#54)

Two air conditioning systems serve this 2635 square foot building. The air handlers are located in closets which act as return plenums. The east closet was depressurized to -19 pascals and the west closet to -16 pascals, but only when the closet doors were closed. The ceiling of the east closet was moderately leaky (one foot square tongue-in-groove ceiling tiles) and the ceiling of the west closet was gypsum board. The east closet door was closed at all times thereby causing substantial return leaks. The west closet door was open all the time, so there was no return leakage into that mechanical closet and no tightening of the closet was done. The east mechanical closet was repaired primarily by tightening the ceiling and walls. Some supply leaks were repaired. Repairs decreased duct CFM25 from 885 CFM25 to 289 CFM25. Relative humidity in the occupied space increased from 54% prerepair to 55% post-repair. Cooling energy use decreased 13.7% from an average 61.9 kWhlday to 53.4 kWh/day. Repair time was twelve person-hours and $75 in materials. Energy savings pay for the estimated $675 repair cost in 4.9 years.

5.5.17 Plastic fabrication office (#56)

This was the most simple yet most perplexing retrofit. While large duct leaks were repaired, energy savings was a meager 4%. A single two-ton air conditioner serves the small 360 square foot office space of this plastics manufacturing warehouse facility. The office itself, the air handler, and the ductwork are all located inside an unconditioned warehouse space. Significant leakage existed at the return support plenum and filter access causing 26% of the return air to originate in this hot warehouse. Supply leaks also exist at the supply register connections. Total duct leakage was reduced by 70% from 186 CFM25 to 55 CFM25. Return leaks declined from 26% to 2%. The small energy savings was a surprise.

Several factors were examined to determine why the savings were so small. The temperature drop from the return to supply increased from 10.0°F (5.6°C) to 12.2°F (6.8°C) when repaired. Based on this, the authors would expect at least an 18% reduction in cooling energy use. Variations in room temperature were examined; the thermostat setting remained constant throughout the monitoring period. A later experiment was done to look at creating a new return duct leak (19% return leak fraction); this increased energy used by 2.8%.

One variable which may offer explanation is the extreme leakiness of the space; ACH5O equals 50.0. This office was the second leakiest of all sites monitored, nearly three times as leaky as the average for the entire 70 building sample. It may be that shell leaks allow considerable air transport between the occupied space and the warehouse, even when the duct leaks are not operating. An additional factor may be that since most of the shell leakage is to the unconditioned warehouse, perhaps some of the energy lost from duct leaks is recovered by cooling and drying the warehouse space.

Relative humidity in the occupied space increased from 54% pre-repair to 55% post-repair.
Cooling energy use decreased 4.3% from an average 22.8 kWh/day to 21.8 kWh/day.
Repair time was four person-hours and $25 in materials. Energy savings pay for the
estimated $225 repair cost in 11.5 years.

5.5.18 Carpet store (#59)

A 3-ton air conditioner serves this 1584 square foot converted automotive service station. The air handler and return are completely within the conditioned space. Moderate supply leaks exist. This site was chosen even though the potential for savings was not expected to be great. Only supply grill connections and one elbow seam were repaired. These repairs reduced total duct leakage by 45.6%, from 158 CFM25 to 86 CFM25. Relative humidity decreased from 53% pre-repair to 51% post-repair. Cooling energy use decreased 11.9% from an average 21.7 kWh/day to 19.1 kWh/day. Repairs were completed in one person- hour using only $8 in materials. Energy savings pay for the estimated $58 repair cost in 1.5 years.

5.5.19 Manufactured office 3 (#60)

One ground-mounted 3-ton package air conditioner served this single-wide commercial trailer. Duct repairs were made to the main return duct at the air handler and the floor register, and to supply register connections in the floor. Some package air conditioner panel leaks could not be sealed. Duct CFM25 decreased 45% from 251 CFM25 to 138 CFM25. Relative humidity increased from 54% pre-repair to 57% post-repair. Cooling energy use decreased 14.3% from an average 31.4 kWh/day to 26.9 kWh/day. Repairs were completed in five person-hours using only $10 in materials. Energy savings pay for the estimated $260 repair cost in 4.7 years.

5.5.20 Manufactured office 4 (#61)

The two wall-mounted air conditioning package units serve this modular office. Part of the exterior wall cavity is used for return air for both systems. Return leakage was repaired in the wall cavity and at the duct board connections at the air handler. Supply ducts were located in the small space between the ceiling and roof deck, and were therefore largely inaccessible. Therefore, only supply leaks accessible through the registers were repaired. Only 30% of the total duct leakage (as seen by the duct test rig) was repaired. As a consequence, energy savings was only 4%. Greater savings would be expected if the ductwork was more accessible.

Duct CFM25 decreased 30% from 793 CFM25 to 554 CFM25. Relative humidity decreased from 46% pre-repair to 45% post-repair. Cooling energy use decreased 4.3% from an average 70.74 kWh/day to 67.6 kWh/day. Repairs were completed in four person-hours using only $25 in materials. Energy savings pay for the estimated $225 repair cost in 4.7 years.

5.6 Discussion of Energy Savings

Energy savings were found in 18 of 19 buildings in which uncontrolled air flow was repaired. In one building, energy use increased by 6.7%. In the 18 which showed energy use reduction, savings ranged from 4.3% to 36% (Figure 5.9). For 19 buildings (including the one which showed an increase in energy use), cooling energy consumption declined by
an averaged 14.7%, or 12.4 kWh/day during the hottest six months of the cooling season. Over an eight month period, savings would be $0.75/day savings or $182 per year (based on 7.5 cents/kWh). Since the average repair cost was $454, the average simple payback period was 2.5 years.

While 14.7% savings is substantial and the payback periods are attractive, it is the authors opinion that greater energy savings could be available. In retrospect, we see that only 59% of the duct system leaks were sealed (in buildings where the ducts were the target of repair). With added attention to achieving more complete repair, we estimate that 70% or more of the leakage could have been sealed.

There are several reasons why we believe that considerably more repair savings can be achieved:

If more of the duct leakage had been repaired and more complete addressing of the exhaust fan and building leakage had occurred, then savings would have been considerably greater. If it is assumed that 75% of duct leaks can be repaired and that repair technologies for providing make-up air and airtightening buildings is available, then we project that about 25% cooling energy savings could have been realized in these 19 buildings.

5.6.1 Commercial savings compared to residential savings

While the energy savings in these commercial buildings was less than that found in residences on a percentage basis (14.7% for commercial versus 17.2% for residential duct repair), the savings are larger in absolute terms, 12.4 kWh/day savings versus 7.2 kWh/day savings. Even normalizing to building size, the savings are larger in commercial buildings. In 48 residences, pre-repair cooling energy consumption was 24.6 kWh/day per 1000 square feet and savings were 4.24 kWh/day per 1000 square feet (Cummings, Tooley, Moyer, 1991). By comparison, in these 19 commercial buildings (average floor area 2541 square feet), pre-repair cooling energy consumption was 34.4 kWh/day per 1000 square feet (or 40% greater than in residences) and savings were 4.84 kWh/day per 1000 square feet. From an electrical demand point of view, commercial buildings cooling energy consumption is focussed more intensely on daytime periods of weekdays, which is also the time when most utilities experience peak demand. Therefore, it might be expected that commercial buildings UAF retrofit programs should provide disproportionate benefit to demand management. In the following section we will look at the monitored demand savings which resulted from repair of uncontrolled air flow.

5.7 Peak Electrical Demand Analysis

Compared to residences, commercial building cooling energy consumption is more concentrated during weekday daytime hours, because that is the period of maximum occupancy. In 63% of the 20-building sample, thermostats were raised or turned off during evening or weekend hours. In the remaining 37%, set-points remained constant throughout the entire week. Based on questionnaire results, the average business was occupied from 7:30 AM to 6 PM on weekdays and open about 4 hours per weekend, for a total of 56.5 hours per week. By contrast, average residential occupancy may be on the order of 130 hours per week (of a total 168 hours per week). Therefore, internally heat generation and cooling systems operation is much more focussed on weekday, daytime periods, which also coincides with periods when summer electrical demand is at a peak.

Reduction in peak electrical cooling demand was analyzed for these 19 buildings. Peak electrical demand was evaluated by selecting pre-repair and post-repair days which are approximately comparable, and then creating a 24-hour composite energy use plots. Following is a discussion of the criteria used for selecting days for analysis.

5.7.1 Criteria for selecting peak demand days

Analysis of power demand reduction was done by choosing the warmest sets of days for which comparable pre-repair and post-repair outside temperature, solar radiation, and “outdoor - indoor temperature difference” data were available. Typically the sample size was six to twelve days. These sets of days were then converted to composite days -- that is, the data is averaged for each hour of the day. This results in 24 values, for temperature, power, and solar radiation for both pre-repair and post-repair periods. Since the FSEC VAX computer, which collects all off-site data, operates using Eastern Standard Time, the results are plotted in Eastern Standard Time. A business which operates from 9 AM to 5 PM DST will be reflected on the peak demand reduction graphs as operating from 8 AM to 4 PM EST.

Florida utilities experience summer peak demand in the window of 2 PM to 6 PM EST, with a maximum around 3:30 EST. In the 19 buildings monitored for energy savings, business hours typically ran from 7 AM to 4 PM EST, though some deviated from that. For our analysis, we identified the peak demand period for each building, usually a two or three hour period, and used that period for comparison of electrical demand. In 10 of the 19 cases, the period 1 PM to 3 PM EST was the peak. The average peak demand period for these 19 buildings was 1:10 PMto 3:16 PMEST (which is 2:10 PM to 4:16 PMDST).

5.7.2 Peak demand results

The results of the peak demand analysis before and after repairs are shown in Table 5.8. The number of days used for comparison are shown as “# Days”. On average, 8.7 days of prerepair and 9.7 days of post-repair data were included in the peak demand analysis. The temperature and solar radiation data are 24-hour averages while peak demand is the cooling energy consumption rate for the indicated peak demand period. Solar radiation appears as a low number because it is a 24 hour average.

Composite 24 hour demand profiles were developed for the 18 buildings for which sufficient pre-repair and post-repair data was available. These are presented as Figures 5.10 - 5.14.

If more than one retrofit was employed at a building, they were all done at the same time and monitored as if one repair. The exception to this was Realty 2 (building #3 7). In this case, three retrofits were implemented in three phases and the energy savings due to each was monitored. The first was duct repair, the second was replacing an old inefficient air conditioner with a new high efficiency one (this measure was taken at the initiative and expense of the business owner), and turning off an attic ventilation fan. The demand reduction shown in Table 5.8 for this building represents only the duct repair and turning off the attic exhaust fan (measures related to uncontrolled air flow), and does not include the demand savings from installing the new air conditioner. Note that it also includes the fan power savings (0.468 kW) when the attic fan was turned off.

The demand savings from all three phases of retrofit in Realty 2 are shown in Table 5.9 and Figure 5.12. (Note that the second retrofit phase, replacement of the air conditioning system is not an uncontrolled air flow retrofit.) Repairing the substantial duct leaks reduces peak demand by 1.07 kW. Replacing the air conditioning system reduces peak demand by 1.84 kW. Turning off the attic exhaust fan, which was depressurizing the entire building to -15 pascals, lowered cooling energy demand by 0.28 kW and fan power by 0.47 kW. In total, peak electrical demand was reduced by 3.66 kW. Since the cost of the three retrofits was $3180, avoided demand cost was $870 per kW. Since the uncontrolled air flow retrofits (duct repair and turning off the attic exhaust fan) cost was $380, avoided demand cost was $104 per kW.

For the entire sample of 18 buildings (data for buildings 3 and 12 was insufficient for analysis), average demand savings were 0.71 kW, or 9.4% of total cooling demand. Since the average cost of these retrofits was $454, this means the cost of avoided electrical generating capacity was $767 per kW, approximately in line with the cost of building new generation capacity. When one considers the combined demand savings and energy savings, repair of uncontrolled air flow can be seen as a very cost-effective measure.

The average 24-hour outdoor temperature on these days was 80.7F, or about 3F cooler than the hottest summer days. Therefore, we could expect even greater demand reduction on the hottest summer days when the utilities are experiencing their system-wide peak demand.

Thirteen of the 18 buildings had 6% or greater peak cooling energy demand reduction, ranging from 6.5% to 28% and averaging 13.4%. Five of the 18 buildings had less than 2% peak demand reduction, ranging from +1.4% to -2.5% and averaging -0.9% savings. Essentially, we can say that five buildings showed no demand reduction while 13 buildings showed an average 13.4% demand reduction. Let’s look at the five that had no demand reduction to see what factors could account for the poor results.

One of them, the Bar and Grill (building #31) had a significant UAF problem which was not dealt with because of diagnostic oversight, as discussed earlier in this report in the energy savings section. That problem was a large exhaust fan, with no designed make up air, pulling attic air into the building for about eight hours per day. Three of the other four buildings had very little insulation in the ceiling/attic spaces. This insulation deficit tends to cause a sharp (solar radiation induced) spike in cooling load during the hot hours of the day. Transport of heat from the hot attic/ceiling space operates in parallel to the conductive heat transfer through the poor insulation. Therefore, addition of insulation in these buildings may have produced both substantial energy savings and demand savings. The fifth is the court office (building #46). We cannot identify reason why energy savings did not occur.

5.8 Energy Savings and Peak Demand Savings Conclusions

Repair of uncontrolled air flow in 19 buildings produced cooling energy savings of 14.7%. It is estimated that better diagnosis and more complete repair would yield 25% cooling energy savings. The average cost of repair was $454. Average daily cooling energy savings were 12.4 kWh per day. Projected annual savings, excluding the heating season, were $182. The energy savings pay for the repairs in 2.5 years.

Peak electrical demand savings from uncontrolled air flow repairs in 18 buildings produced peak demand savings of 9.4% or 0.71 kW. Based on an estimated $700 cost of bringing new generation capacity on line, this retrofit displaces the need for about $500 in new capacity.

When energy and demand savings are considered together, the cost benefits of uncontrolled air flow retrofits are extremely promising. 1f for example, a utility were willing to pay half the cost of the retrofit ($227), then the utility would in essence be getting new system capacity at a cost of $320/kW and the customer would be obtaining $182/year energy savings which provide a 1.2 year simple payback. Add to this possible comfort and indoor air quality benefits, such a program is a win-win situation for all parties.

Table 5.2
Types of Uncontrolled Air Flow Repair That Were Implemented (X) and Were Not Implemented (O)

Table 5.3
Building Airtightness, Duct Airtightness, Infiltration rates, Air Flows, and Pressures After Repair