1. ABSTRACT

Testing for air leakage in air distribution systems was done in 160 central Florida homes. Tracer gas tests found that infiltration rates were three times greater when the air handler was operating that when it was off, indicating that there are large leaks in the air distribution system. Infiltration averaged 0.91 air changes per hour (ach) with the air handler (AH) operating continuously and 0.28 ach with the AH off. Return leaks were measured by tracer gas and found to average 10.7% of AH total flow. House airtightness, in 99 of these homes, determined by blower door testing, averaged 12.7 air changes per hour at 50 Pascals (ACH50). When the duct registers were sealed, ACH50 decreased to 11.1, indication that 12.7% of the house leaks were in the air distribution system.

Duct leaks were repaired in 50 of the 160 homes. Blower door tests were done on these houses before and after repair. Before repair airtightness was 12.5 ACH50. After repair house ACH50 decreased to 11.2, indicating that 63.7% of the duct leaks were repaired. Infiltration tests were done before and after repair on 25 of these homes. Infiltration rates with the AH on decreased from 16.0% to 4.5% of total air handler flow.

Cooling energy use decreased as a result of duct repairs. Data was available for 46 of the 50 homes. Air conditioner energy use decreased by an average 17.2%, yielding estimated space conditioning energy savings of $110 per year. Duct repairs are a very cost-effective retrofit. At an average cost of $200 per home, duct repairs have a simple payback of less than two years.

Duct leaks have a dramatic impact upon peak electrical demand. While no peak demand data has yet been measured, theoretical analysis indicates that a 15% return leak from the attic can increase cooling electrical demand by about 90%. Detailed theoretical analysis of a winter Florida morning indicates that duct repairs in a typical, electrically heated Florida home reduce winter peak demand by about 1.6 kW per house at about one-sixth the cost of building new electrical generation capacity. Repair of ducts in 3 million Florida homes could reduce winter peak demand by 5000 megawatts, of 13% of the state’s generating capacity. This effort would be very cost effective, since the generation capacity made available by duct repair would cost only abut one-third to one-eighth what new capacity would cost, depending upon type of generation facility.

2. INTRODUCTION

Duct leakage has been observed in many homes by the authors of this paper and others. Significant duct leakage was observed in the majority of the 370 homes they have tested with blower doors (Tooley and Moyer, 1989). They have also reported on duct leakage in about 25 homes, indicating that significant duct leakage has been found (Cummings, 1988; Cummings and Tooley, 1989; Cummings and Tooley, 1990). Other authors across the nation have detected air distribution systems (AD) leakage, as well. Infiltration rates n 31 Tennessee homes were found to be 77% higher when the air handler was operating (Gammage, et al., 1986). Tracer-gas-measured infiltration rates in 6 new Pacific Northwest homes were found to be 70% higher in those with forced-air systems during a four month winter period compared to those without forced-air heating systems. This indicates that leaking air distribution systems increased infiltration, since blower-door testing indicated they should only be 12% leakier (Parker, 1989). Blower door tests done on 20 homes n the Pacific Northwest found that about 10% of the house leak area was in the air distribution system (Robinson and Lambert, 1989).

3. PROJECT DESCRIPTION

A study was begun in the spring of 1989 to investigate air distribution system (ADS) leakage in central Florida homes. Testing was done using tracer gas to determine how much infiltration is caused by ADS leaks. Blower door tests were done to measure house and ADS airtightness. Duct repairs and cooling energy use monitoring were also done. (Note that the use of the terms “duct system” and “air distribution system” (ADS) are energy used interchangeably in this report. “Air distribution system” is the more accurate term since it includes the air handler (or furnace), return plenums, and supply plenums, which are not usually referred to as ducts).

Tracer gas infiltration testing has been done in a sample of 160 homes in order to detect ADS leakage. The housing sample was randomly selected. The only screening criteria used was that the house have a forced-air distribution system. It was also decided that the sample should fall into five groups, based primarily on air handler (AH) location. Therefore, 30 houses were selected in each of the following categories: (1) AH in the attic, (2) AH in a closet or utility room, (3) AH in the garage, (4) AH outdoor, and (5) HUD-code mobile homes. (Mobile homes normally have package units – air handler outdoors. Category (4) includes only site-constructed homes with package units).

ADS leakage was elevated by means of tracer gas testing, blower door testing, and visual inspection. Tracer gas tests were done once with the AH operating continuously (interior doors open) and once with the AH off. If the infiltration rate was higher when the AH was operating, then a high probability of duct leakage was indicated. If the infiltration rate was much higher (say three to five times higher) because of AH operation, the ADS leakage was strongly indicated.

Another measure of ADS leakage is the return leak fraction (RLF). This is the proportion of air returning to the AH which originates from outside the conditioned space. RLF is determined by measurement of tracer gas dilution from the return (in the room at the return register or registers) to a supply register. Tracer gas concentration is also measured in the attic (or other buffer zone) at the return leak site, in order to quantify the concentration of tracer gas in the dilution air. This method has proved to be accurate and repeatable method for quantifying leaks on the return side of the air distribution system. Supply leaks cannot be so quantified.

Infiltration tests were done using the tracer gas decay method and a portable infrared specific vapor analyzer. A 20 minute period was used to mix the tracer gas throughout the house using the AH as the mixer. In the test with the air handler operating, sulfur hexafluoride samples were taken every five minutes for 30 to 40 minutes (data collected at a minimum of 7 time increments) at the intake to the return register or registers (typically Florida homes have only one or two return registers). Mixing was maintained by the continuous operation of the AH.

In the tests with the air handler turned off, samples were taken every 10 minutes for a minimum of 50 minutes (data collected at a minimum of 6 time increments). Tracer gas measurements were again taken at the return registers. Mixing of the tracer gas was maintained by turning the air handler on for 1 or 2 minutes during each 10-minute period. A change in this test protocol was made after 110 homes were tested because duct leakage during the minute(s) of AH operation caused error in the result. As a result, natural infiltration is slightly overestimated in these 110 homes. On the last 50 homes, sampling was done at four distributed locations throughout the house, and the AH was not turned on at all during this test. These testing procedures are described in more detail in Cummings (1989).

Because of funding limitations, blower door tests were done on only 100 of these homes. These tests were done in the depressurization mode only. It is the belief of the authors that pressurization artificially opens up “holes” in the house, such as awning windows, exhaust fan dampers, etc., while depressurization generally pulls them closed. Fan air flow was measured at 5 to 8 house-to-outdoors delta-pressure points, generally across the range from 10 to 60 pascals (Pa). These tests were repeated with all the supply and return registers covered by paper and tape. These tests permit determination of house air leakage at 50 Pa (ACH50), and by subtraction, the proportion of the house leak (at 50 Pa) which is in the air distribution system.

Duct repairs were done on 50 of the 160 homes. Blower door and tracer as tests were repeated after repair. Air conditioner energy use was monitored for a period of approximately four weeks before repair and four weeks after repair. Monitoring was accomplished in the following manner. Watt-hour and run-time meters were installed on the air conditioner (AC). The house occupant was asked to record the run-time meter, the AC meter, and the whole-house watt-hour meter daily. In order to obtain comparable data, they were asked to maintain their thermostats at the same setting throughout the testing and take meter readings at about the same time each day. A weather station was installed to collect temperature, dew point temperature, solar radiation, and wind speed. Cooling energy use was normalized to weather by plotting “daily AC kWh” versus “daily average outdoor temperature”.

4. PROJECT TEST RESULTS

Three different tests were performed to detect duct leakage: (1) tracer gas tests, (2) blower door tests, and (3) smoke test inspection. In each phase of testing, extensive evidence of duct leakage was found.

4.1 Tracer Gas Test Results

Infiltration tests revealed that duct leaks are large and widespread in central Florida homes. The average infiltration rate of the 160 homes tested was more than three times higher with the AH operating than when it was off. Figures 1 and 2 shows these results. The house infiltration rates averaged 0.91 ach when the AH was on, while only 0.28 ach when it was off.

4.1.1 Infiltration with air handler off

When the air handler is off, more than 50% of the homes have infiltration less than 0.25 ach while only 4% have infiltration greater than 0.75 ach. From this data it can be stated that natural infiltration is very low in Florida homes. Several considerations arise from this conclusion.

First, efforts to make the house envelope more airtight may not be cost-effective. Computer simulation of air conditioning and heating loads in Orlando, Florida using Thermal Analysis Research Program found that reduction of infiltration from 0.28 ach to 0.18 ach results in about $30/year energy use savings (Cummings and Tooley, 1990). This assumes that infiltration air comes from outdoors, as opposed to attic, garage, or crawl space. Even if much of the airtightening were in the attic and annual energy savings were $50, the added effort of airtightening would probably be marginally cost-effective.

Second, further tightening may make indoor air quality worse. ASHRAE 62-1989 calls for 0.35 ach (or greater depending upon occupancy) for residences. Therefore, making homes tighter would not be recommended from the occupant health point of view.

It should also be noted that our measurements overestimate natural infiltration. There are two reasons for this. First, all tests were performed during the summer (April through October) hours of 9 A.M. to 5 P.M.. Because daytime winds in Florida are typically stronger than nighttime winds, measured natural infiltration probably over predicts actual annual average infiltration. Second, the air handler was used one or two minutes of each 10-minute sample period to provide mixing in 110 of the 160 homes. Air distribution system leaks cause elevated infiltration during these one to two minute periods, and they therefore overestimate natural infiltration.

Since our measurements overestimate natural infiltration, the question can be raised, “What is the average annual natural infiltration rate for Florida homes?” Is it 0.25 ach? Or 0.22 ach? The answer will of course be speculative, but there may be some value in exploring this question. Comparison of wind speeds during the day and night is instructive. Wind speed from the Lakeland weather station averaged 5.3 MPH during the hours of 9 A.M. to 5 P.M., compared to 4.3 MPH for the entire day. Wind speed for the entire day is therefore 20% lower than for the testing hours of 9 A.M. to 5 P.M.. If a relation between wind speed and infiltration can be found, then an estimate may be made of daily infiltration rates.

Insight concerning wind speed impacts upon infiltration rates may be found in a study of 50 new homes (less than 5 years old) in Central Florida (Cummings, Moyer, and Tooley, 1990). Even though these homes are much more airtight (ACH50 = 7.3) and have lower infiltration rates (achon = 0.21) than the 99 home sample from this study, they show a pattern of increased infiltration with increased wind speed from which we may approximate a relation between test-period infiltration and daily infiltration (Figure 3). The plot contains a best-fit line for wind speed and infiltration. The average wind speed (measured on site using a 7 feet tower) during the tests was 3.3 MPH. If we look at the infiltration rate corresponding to a wind speed 20% lower, namely 2.6 MPH, we find that infiltration is 0.19 ach, or 10% lower. If we extrapolate this to the test results from our 160 house study, the measured natural infiltration of 0.28 ach would be reduced to 0.25 ach. If some adjustment is made for AH operation during the tests (which increased infiltration because of duct leaks), we might assume annual, natural infiltration of around 0.22 ach.

An additional piece of evidence concerning overall infiltration also comes from the same 50-house study. A 17-house subset of these 50 occupied homes had 7-day PFT (perfluorocarbon tracer) tests performed (Cummings, Moyer, and Tooley, 1990). The PFT tests provide a view of infiltration over the entire day-to-night spectrum. Seven-day infiltration averaged 0.17 ach, compared to 0.21 from the 80-minute tracer gas tests. PFT infiltration is 20% lower than that measured during the short tracer gas tests. When it is considered that the air handlers operated 27% of the time and caused average 0.44 ach infiltration when operating, then the PFT results actually are 36% lower. If we extrapolate this to the 160 sample in this study, overall infiltration would drop from 0.28 ach to 0.18 ach.

The preceding discussion does not resolve the question of what average annual infiltration rates. It does, however, suggest that overall natural infiltration rates in Florida homes may be in the range of 0.20 to 0.25 ach. In any case, Florida homes are already quite tight, and further tightening may lead to occupant health problems.

It also suggests that if all ADS leaks are repaired, some homes may have increased indoor air quality problems. This does not, however, constitute a strong argument for not repairing ADS leaks. There are two major reasons why duct leaks should be repaired. First, while they provide ventilation, they also draw pollutants into the house; from the garage, crawl space, attic, and soil. While repair of duct leaks may decrease house ventilation, they may improve indoor air quality, on balance, because they reduce pollution entry rates.

Second, the energy penalties resulting from duct leaks are generally much greater than infiltration from the outdoors, which is the most desirable source for ventilation air in terms of air quality. Supply leaks greatly increase energy use because they leak air that is highly conditioned (hotter than room air in the winter and colder than room air in the summer). In addition, in many cases they depressurize the house, causing air to be drawn in from the attic (among other places) and thereby increasing the cooling energy penalties. Return leaks generally cause higher energy use than ventilation air from outdoors, because a large fraction of return leaks are drawn from the attic of the garage, which locations have hotter air than outdoors.

For both indoor air quality and energy use reasons, duct leaks should be repaired. If additional air is required to maintain good indoor air quality (after some attempt has been made to remove pollution sources), then alternative means should be implemented to bring in air from outdoors, perhaps on a daily schedule which would minimize energy and peak demand impacts.

4.1.2 Infiltration with air handler on

When the air handler is operating, infiltration rates are much higher (see Figures 1 and 2). On the average, the infiltration rate with the AH on was 0.91 ach, or 225% greater than the natural infiltration rate. While more than 80 homes had infiltration rates less than 0.25 when the AH was off only 1 of 160 homes had infiltration less than 0.25 ach when the AH was on. Only 6 homes had achoff greater than 0.75 ach, while 92 had achon greater than 0.75 ach.

As can be seen from these numbers, duct leakage dominates infiltration in Florida homes. If we assume 40% air handler operation time (based on run-time from 50 monitored homes), then the daily average infiltration rate would be 0.54 ach (see calculations below). Duct leaks increase overall infiltration by 90%. If we assume that average natural infiltration is actually lower than measured (because nighttime winds in Florida are lower), say 0.22 ach, then overall infiltration would be 0.50 ach, and ADS leaks increase infiltration by 125%.

Return leaks were found to be dominant in the majority of homes based on visual inspection. Tracer gas tests were used to determine the size of return leaks. The return leak fraction (RLF) was measured in each home with the AH operating (Figure 4). Sixty-seven prevent were found to have return leaks equal to or greater than 5% of the AH total air flow. Thirty-six percent have greater than 10% RLF. The average for 160 homes was 10.7% RLF.

4.1.3 Low natural infiltration rates are reasonable

To some readers, 0.28 ach natural infiltration may seem lower than expected. The following observations are offered as an explanation for low infiltration in Florida homes. “Natural” infiltration is driven by pressure differentials caused by wind and thermal stack. Florida wind speeds are lower than in other parts of the country. Stack effect is small because: (1) temperature differences (indoors minus outdoors) are less than 15ºF most hours of the year, (2) homes are typically short (one story), and (3) concrete slabs (greater than 90% of homes) and block walls (greater than 80% of homes) do not offer many inlets for stack infiltration.

Infiltration testing from other projects indicate that these low natural infiltration rates are reasonable. Natural infiltration in 50 new (less than five years old) Florida homes was 0.21 ach (Cummings, Moyer, and Tooley, 1990). A sample of 9 Florida homes averaged 0.22 ach (Cummings, 1988). Another sample of 12 Florida homes averaged 0.26 ach (Cummings and Tooley, 1989). A sample of 5 Florida homes averaged 0.14 ach (Cummings and Tooley, 1990). Infiltration measured in other states is similar as well. In Tennessee, 31 homes averaged 0.44 ach with the air handler off and 0.78 ach with the air handler on (Gammage, et al., 1986). In 55 new homes in the Pacific Northwest, infiltration measured over a four-month winter period averaged 0.24 ach, in homes which did not have forced air space conditioning.

4.2 Blower Door Test Results

The significance of leaks in the duct system becomes more important when it is considered that most of the ADS system is under an order of magnitude greater pressure differential than the remainder of the house. The air distribution system is commonly under pressure most of the time. It is interesting to note that the 12.7% of the house leaks located in the duct system account for 70% of the infiltration when the AH blower is on.

The 100 homes which had blower door tests have slightly greater duct leakage than the larger sample of 160 homes, based on comparison of tracer gas test results. Infiltration with the AH on was 1.04 ach in these 100 homes compared to 0.91 ach for the 160 homes. RLF is 12.9% compared to 10.7%. Infiltration with AH off is 0.31 ach compared to 0.28 ach.

Blower door tests indicate that house airtightness is a function of age (Figure 7). Older homes are leakier. Homes built in the last 10 years have ACH50 of about 8 while those over 20 years old have an average ACH50 of about 16.

4.3 Air Handler Location

Variations in house airtightness, duct airtightness, infiltration rates, and return leakage can be seen for various air handler locations. The five air handler categories are attic, closet, garage, package, and mobile home. Table 1 table summarizes the results for the 98 homes which had complete blower door tests.

Blower door tests reveal that all air handler locations have duct leakage. The proportion of the house leak area which is in the air distribution system is similar among the five categories, all in the range of 12% to 14%. Tracer gas test results indicate that mobile homes and homes with package air handlers have greater air distribution system leakage, shown by higher infiltration rates with the air handler on. Attic and package units have greater return leakage, on average. Since the samples are only in the range of 17 to 22 houses, the variations among types are not statistically significant.

Tracer gas tests for the larger sample of 160 homes show a similar pattern to that in the 98 house sample. Air handler induced infiltration is again highest in the package and mobile home samples. Return leaks are substantially higher only in the package unit homes.

4.4 Smoke Test Inspection

Visual inspection of duct leaks was done in two ways. First, with the blower door in place, the house was pressurized to about 15 Pa. Using a smoke stick (titanium tetrachloride) each supply and return register was checked (with the AH off) to see if smoke would pass into it, and at what velocity. If the smoke did not enter the grill, or entered it slowly, then little or no leak was located in nearby ducts. However, if the smoke “raced” into the register, then leakage existed in the ducts nearby. Second, the duct system was inspected with the AH operating (blower door off). All connections, ducts, plenums, and air handlers were inspected for leaks, either with the smoke stick or by feeling by hand.

Return leaks are estimated to be about twice as large as supply leaks in site-built homes, based on the homes we have inspected. The reason this is an estimate is because we have no direct method for measuring supply leakage. Return leaks can be measured fairly precisely with tracer gas. The estimate that supply leaks are about one-half that of return leaks is based upon visual inspections alone.

Return leaks in mobile homes are large, but not as large as supply leaks. It is estimated that supply leaks are larger than return leaks in more than two-thirds of mobile homes.

4.4.1 Description of return leaks in site-built homes

Return leaks are found in the following locations, listed in order of magnitude, based on visual inspection (remember that return leak here refers to the fraction of air coming from outside the house):

4.4.2 Description of supply leaks in site-built homes

Supply leaks are found in the following locations, listed in order of magnitude, based on visual inspection:

4.4.3 Description of return duct leaks in HUD-code mobile homes

Return leaks are generally smaller than supply leaks in mobile homes. The most dominant return leaks are listed below, in order of estimated leak magnitude.

4.4.4 Description of supply duct leaks in HUD-code mobile homes

Supply leaks are found in mobile homes in the following locations in order estimated leak magnitude.

4.5 Duct Repairs Impacts in 50 Homes

Duct leaks were repaired in 50 homes. The impact of these repairs upon the following variables was measured: infiltration rate with the air handler operating, the return leak fraction, house airtightness, the proportion of the ADS leak area which was repaired, and cooling energy use.

4.5.1 Impact of duct repair upon infiltration

Infiltration rates were measured before and after ADS repairs. The results are summarized in Table 3 for the first 25 homes which were repaired in 1989. Results for the second 25 homes (repaired in 1990) are available only for “before repair”. Infiltration rates “after repair” are unavailable because of equipment failure. Since post-repair tests are unavailable for the second 25 homes, the following discussion of infiltration and return leaks relate only to the first 25 homes.

Because of duct repair, infiltration with the AH on decreased from 1.10 ach to 0.54 ach. It is interesting to note that duct repairs, which decreased house leak area by 10.7% (ACH50 decreased from 12.3 to 10.9), caused a 53% reduction in infiltration with the air handler operating. Perhaps more significantly, infiltration with the air handler on was reduced 66% of the way to the natural infiltration rate of 0.25 ach. RLF was reduced 73%, from 15.8% to 4.4%.

4.5.2 Impact of duct repairs upon house and ADS airtightness

Blower door tests results are summarized in Tables 4 and 5 for before and after repair. In the 48 which had complete blower door tests, ACH50 averaged 12.5, which is similar to 12.7 for the larger sample of 98. While overall house airtightness is similar to the larger sample, duct leakage is larger in these 48 homes than in the larger sample. The proportion of the whole-house leakage at 50 Pa which is in the air distribution system is 15.6%, compared to 12.7% in the sample of 98. Tracer-gas-measured infiltration with the air handler on averaged 1.07 ach (compared to 1.04 ach for the 98 and 0.91 ach for the sample of 160), and the RLF averaged 15.7% (compared to 12.9% for the 98 and 10.7% in the 160 house sample).

Repair of duct leaks reduced ACH50 in these 48 houses by 9.9%, from 12.5 to 11.2. This reduction in ACH50 of 1.23 indicates that 64% of the duct leak area was sealed.

4.5.3 Impact of duct repair upon cooling energy use

Air conditioner energy use was analyzed for these 50 homes. In one home the air conditioner failed during the test period, making analysis impossible. In three others, the quality of meter reading was too low to make the data useable. In the remaining 46, cooling energy use was reduced from an average 40.6 kWh/day to 33.6 kWh/day, or an average 17.2% reduction per home. Savings ranged from a low –4.0% (energy use rose by 4 percent) to a high of 44.6%. The distribution of cooling energy savings is presented in Figure 8.

In eight homes, energy savings were less than five percent. These cases were examined to see why savings were low. In two instances, energy savings were small because the initial duct leakage was small. Therefore, small energy savings were expected. In one, repairs reduced 7% return leak to 3.3%, and energy use was reduced 4.2%. In another, which had almost no duct leakage (achon = 0.35 and RLF = 4%), energy savings was 2.0%.

In other cases it was found that some of the duct leaks were inaccessible or unrepairable, so complete repair was not possible. In one case, which experienced only 3.5% energy savings, the majority of the large supply leaks were in inaccessible portions of an attic addition. In another case, a package air handler was badly rusted, and no repair of the air handler leaks could be made. A change-out of that air conditioner (about $1500) would solve the majority of those large duct leaks.

A number of homes were not included in the sample for repair because their ducts were inaccessible. The majority of these were new mobile homes with supply ducts in the ceiling. Because there is no attic space, access to duct leaks in this ceiling cavity is impossible, short of removing the ceiling. Since these ducts cannot be accessed and the duct leaks therefore cannot be located and repaired, this type of construction, in effect, makes the duct leaks permanent. Careful assessment should be made of all building practices which restrict access to ducts, plenums, and air handlers.

It is our recommendation that building energy codes require access and clearance around these components of the air distribution system. Precedence for this is in the Southern Building Code Congress International, Inc. building code, which Florida has adopted. It requires a minimum of one access to the attic so that various mechanical and utility systems may be serviced. The air distribution system clearly is a system which requires servicing and repair in a large majority of homes.

In two cases “short-circuiting” of supply to return leaks may have been occurring. By “short-circuiting” we mean the supply and return leaks are in close proximity, so much of the supply leak air is drawn back into the system through the return leak. In one mobile home which showed only 1% energy savings, supply leaks were discharging into the pan/sub-floor area from which the return leak was also drawing. In one attic-air-handler home which experienced 4% increase in energy use, supply leaks were dumping into a space below the air handler where the return leak was also drawing much of its air. This “short-circuiting” may partially account for that fact that energy savings were near zero.

In one other case where energy use actually increased by 2% and in another case where energy use decreased by 4.5%, no explanation could be found for the lack of savings, since significant leaks were repaired. It is possible that there were lifestyle, occupancy, or equipment changes in the home which could explain the lack of savings. In general, however, there seemed to be reasonable factors explaining the lack of energy savings in six of the eight homes which had less than 5% energy use reduction.

Five homes experienced greater than 30% energy savings (Figure 8). The greatest was 44.6% reduction. In this case there was a 26.2% return leak, nearly all of which was from the attic. If the air conditioner had been sized larger, the savings would have been even greater since the air conditioner, before repair, could not meet the cooling load during most afternoons. While the house thermostat was set at 79ºF, the house temperature often rose to 82ºF and above. It is suspected that energy savings were smaller than would have otherwise occurred in at least 10 of these homes because the air conditioner was unable to meet cooling loads caused by large duct leaks.

Comparison of energy use before and after repair was done in the following manner. A least-squares first-order best fit was made for the daily AC kWh verses average daily temperature. From these best fit lines, energy use before and after repair was determined at the average summer temperatures of 81.3ºF and 81.8ºF in Brevard and Polk counties, respectively. Plots for eighteen houses are presented in Figures 9, 10, and 11.

It is estimated that the average cost of duct repair in these 50 homes was about $200. With average energy savings of 7.0 kWh/day, cooling season savings of $85 could be expected. Including heating season savings of perhaps $25, duct repairs would have a simple payback of less than 2 years.

4.5.4 Energy savings by air handler location

Energy savings by air handler location is presented in Table 6. Some observations can be made. Energy use appears to be significantly higher in homes with attic air handlers both before and after repair. Attic AH houses used 30% more cooling energy (before repair) than the other four groups collectively. Following repair, attic AH houses used 35% more cooling energy than the others.

Garage AH houses used 32% less cooling energy than attic AH houses. This observation is even more significant when it is considered that garage AH houses are 20% larger than attic AH houses. Thermostat settings were essentially equal in the two groups of houses, 78.9ºF for the garage AH houses and 79.0ºF for the attic-AH houses. After adjusting for house size, homes with AH in the attic used 75% more cooling energy than homes with AH in the garage. While the sample sizes are too small to say that the findings are statistically conclusive, this data does suggest that attic AH houses are substantially more inefficient. The cause may be greater conductive and infiltration heat gains, since more of the air distribution system is located in the hot attic. Further study on a larger sample should be useful in verifying these findings, and determining the cause.

It can also be noted that all categories experienced about 65% duct repair based on blower door tests, expect package units, which had only 50% repair. In spite of this, package systems experienced greater than average energy use reduction. A possible explanation for this anomaly may be found in the fact that the leaks for these package systems were closer to the air handler fan and therefore were under greater pressure differential. Because of the greater delta-pressure, the amount of leakage would be greater.

5. DUCT LEAK IMPACTS UPON PEAK ELECTRICAL DEMAND

The impact of duct leakage upon peak electrical demand is greater than upon total energy use. The reason for this is that two factors are at a maximum during the utilities peak demand period – the air handler operation time and the temperature of the air brought into the house. While an air conditioner may operate 30% to 40% over an average day, during the hottest hours of the hottest day, the AH may be run 80% or more of the time. In addition, the outdoor temperature and especially the attic temperature are at their greatest extreme from the house interior temperature.

Figure 12 shows the impact of return duct leaks from the attic upon AC energy efficiency ratio (EER). Since the utility’s summer peak occurs at about 5 P.M., when the attic is likely to be 120ºF and have a dewpoint temperature of about 85ºF, a 14% return leak can reduce the effective capacity and EER of the system by 45%. A 30% return leak can completely overwhelm the capacity of the system, causing the temperature in the house to rise. The total increased electrical demand in many homes is limited by the capacity of the air conditioner. If air conditioner capacity were unlimited, the peak demand impact of duct leaks on the summer peak would be much larger.

Duct leak impacts on demand are illustrated in Figure 13. Measured infiltration and air conditioner performance are shown before and after repair of a 30% return leak. The infiltration rate with the air handler operating decreased nearly 75% from 1.15 ach to 0.30 ach as a result of the repairs. Temperature drop from return (measured in the room at the register) to supply increased from 7ºF to 16ºF, indicating a 130% increase in net, sensible cooling capacity. Put another way, air conditioner capacity and efficiency were reduced by about 55% as a result of drawing 104ºF attic air into the return air stream. It is interesting to note that the room temperature was 87ºF before repair, even though the air conditioner was insufficient to maintain the setpoint. Afterwards, the house could be easily cooled with the air conditioner cycling normally.