Testing Duct Leakage Assumptions
In Midwest houses where little or no ductwork runs though unconditioned spaces, normal assumptions about duct leakage to the outside may not apply.
Take the test and earn 0.5 CE unit per article.
Many homes in Wisconsin and throughout the Midwest have little or no ductwork that runs through unconditioned spaces. Because of this, national standards for testing and required default assumptions for duct leakage can impose an unnecessary burden on midwestern housing stock. The standards are primarily based on leakage data from homes in other parts of the country where basements are less common and where many homes have substantial amounts of ductwork in the attic or crawlspace.
The goals of a study we performed in 2007 and 2008 for the Wisconsin Focus on Energy program were to quantify duct leakage to the outside in new homes with exterior duct runs and to compare several methods of measuring this leakage. We wondered how actual duct leakage to the outside compares to current national default values (as defined by RESNET), and whether Wisconsin homes can be reliably screened for duct leakage using less-expensive methods than duct pressurization. Note that while interior leakage and pressure imbalances can also be significant issues with ducted heating systems, these were not a particular focus for the study.
The homes for the study were identified through Focus on Energy staff and the network of Wisconsin energy consultants associated with the Wisconsin Energy Star Homes program. The 19 new homes (all unoccupied at the time of testing) were assessed between February 2007 and February 2008, using a daylong protocol involving four methods of duct leakage assessment: duct pressurization testing (Duct Blaster); pressure pan testing; blower-door-based duct leakage assessment using the Delta-Q method; and nulling testing.
The vast majority of Wisconsin homes have forced-air furnaces that are located in a full basement with exposed
sheet-metal ductwork. Most new Wisconsin homes have no ductwork outside the conditioned envelope, and duct runs to upper floors typically pass through interior wall cavities. Because current program rules require duct leakage testing if any part of the duct system lies outside the conditioned envelope of the building, all of the homes in the study had at least some exterior ductwork, However, the study homes varied in the extent to which the ductwork was exposed to conditions outside the thermal envelope of the home:
LOW-EXPOSURE HOMES. Eight homes had most of ductwork inside the conditioned envelope; these were typically homes with only the supply runs for a bonus room over a garage outside the conditioned envelope.
MODERATE-EXPOSURE HOMES. Seven homes had a mix of inside and outside ductwork; these were typically two-story homes with second-floor supplies through the attic.
HIGH-EXPOSURE HOMES. Four homes had most of ductwork outside the conditioned envelope; these were typically slab-on-grade homes with duct runs through the attic and the heating system located in a mechanical closet in the garage. The three single-story homes were all in this group (the other homes in the study were all multistory).
The homes in the study had an average of about 17 supply registers and 7 return registers, or about 130 ft2 and 330 ft2 of floor area per register for supply and return registers, respectively.
STATE OF DELTA-Q
The Delta-Q approach to measuring duct leakage is attractive, because it can be done using a computer-controlled blower door and the ability to flip the air handler on and off repeatedly, thus eliminating the time and expense of sealing registers in order to set up and run a direct duct pressurization test. But the calculations are considerably more complex than standard blower door testing, and researchers are still debating the finer points of how best to implement the Delta-Q algorithms. Moreover, the duct leakage “signal” that Delta-Q is trying to measure is often small and can be difficult to distinguish from wind-induced pressure changes.
The Energy Conservatory recently enlisted a group of volunteers to conduct repeated Delta-Q tests in a number of homes. As Figure A shows, the duct leakage estimates sometimes clustered tightly together (Homes A and B) and sometimes showed variations of several hundred CFM (Homes F and G) as a result of variations in windiness during testing.
A recent study by Darryl Dickerhoff and Iain Walker at Lawrence Berkeley National Laboratory examines the issue of wind-induced pressure changes and provides guidance on recognizing anomalies in Delta-Q test data.
Until experts settle on a method for quantifying the reliability of Delta-Q tests, practitioners in the field should follow a few guidelines to help ensure reliable test results:
Avoid windy test conditions. Test early in the morning or late in the day, when winds tend to be lighter, and try to avoid days when high winds are forecast.
Learn to recognize wind-induced anomalies in the Delta-Q data and repeat test sequences that show these anomalies.
For the house pressure measurement, carefully place the outdoor pressure reference tube in a protected location, and be sure none of the open manometer taps or open ends of tubing are ever directly exposed to the air stream from the blower door. This outdoor pressure reference tube can be placed on a side of the house other than the one where the blower door is installed. Note, however, that the outdoor pressure tube for referencing the blower door fan during pressurization mode should be on the blower door side of the house.
The protocol we used employed four methods to assess duct leakage in each home.
Duct pressurization is used to measure total duct system leakage (and leakage to the outside) only at artificially induced duct pressures. Measuring total duct system leakage involves masking all registers, pressurizing the duct system with a calibrated fan (typically to 25 Pa), and then recording air flow through the fan at the known level of pressurization —much as a standard blower door test measures house leakage at a known pressure.
To measure leakage to the outside, a blower door is used to pressurize the house. The duct pressurization fan is then used to zero out any pressure differences between the ducts and the house; the flow through the duct pressurization fan with the duct pressure zeroed with respect to the (pressurized) house represents the level of duct leakage to the outside. Because this testing is conducted at artificially induced pressures, the results do not necessarily indicate the amount of leakage that occurs under actual operating conditions.
For the study, total duct system leakage was measured at 25 Pa, and leakage to the outside was measured at 25 Pa, 50 Pa, and 100 Pa. In addition, to gauge the sensitivity of the leakage-to-outside results to measurement of the pressure difference between the ducts and the house, flow readings were taken at 0 Pa, 0.2 Pa, 0.5 Pa, and 1 Pa of pressure difference between the ducts and the house.
All of this testing was repeated twice—once for the entire duct system (supply and return), and once for the supply side of the duct system only. The latter test was done by installing an airtight block in the filter slot to isolate the supply ductwork from the return ductwork. Return leakage was then estimated by subtracting supply leakage from the supply-and-return measurements.
The Delta-Q test involves measuring blower door flow across a wide range of house-to-outside pressures, both with the air handler operating and without it operating, and then statistically extracting the estimated leakage from the resulting data. The process has been greatly facilitated with software developed by the Energy Conservatory that largely automates the process.
Unlike duct pressurization, Delta-Q measurements of duct leakage to the outside are meant to reflect leakage under actual operating conditions. Also Delta-Q testing does not require equipment beyond a blower door that can be controlled by a laptop computer. At present there are two test methods: ramping, in which the blower door is quickly operated through a range of house pressures while recording data on air flow and house pressure; and stations, which involves collecting data at each of a number of discrete pressure levels.
There are also various approaches to analyzing the data collected under a Delta-Q test to estimate duct leakage. We generally used the scanning technique, which assumes that a single pressure on each of the supply and return sides can characterize the leakage.
For this study, a protocol of multiple Delta-Q tests was implemented at each site. These tests involved both ramping and stations testing, as well as repeat tests and control tests to help assess the reliability of the Delta-Q method (a control test is a Delta-Q test in which the air handler is never turned on, and for which the actual leakage from air handler operation is therefore known to be zero).
The idea behind the nulling approach is that duct leakage to the outside induces a pressure difference across the building shell. An equal air flow in the opposite direction will nullify that pressure difference; if the opposing air flow is created with a calibrated fan, then the air flow through the fan can be measured and the amount of duct leakage thus quantified.
Though simple in theory, nulling is difficult to implement in practice. This is because the house-to-outside pressure differences induced by duct leakage are generally very small (often less than 1 Pa), making the duct leakage signal difficult to distinguish among pressure readings that vary from moment to moment due to wind. This makes it necessary to average readings over time to average out wind effects reliably.
With the assistance of Collin Olson of the Energy Conservatory, we developed a semiautomated, computer-driven nulling procedure that cycled the air handler on and off (at 15- to 20-second intervals) over 20 to 30 cycles, while recording house-to-outside pressure. This process was repeated for several nulling fan flow levels that were chosen to attempt to bracket the flow needed to nullify the duct leakage. We then used linear regression to estimate the nullling flow (and a confidence interval for the regression-based estimate).
As with duct pressurization, the entire process was repeated twice—once for the entire duct system (which estimates
the amount of unbalanced duct leakage, that is, which of the supply or return leakage is greater and by how much); and once with the return side of the duct system isolated. (In the latter case, a second Duct Blaster was needed to provide air flow to the air handler at a pressure mimicking the return side of the duct system.) The estimated return leakage was the difference between the entire duct system nulling test (that is, the unbalanced nulling test) and the supply-only nulling test.
Pressure Pan Testing
Pressure pan testing is not a duct leakage quantification approach, but rather a leakage diagnostic tool. It involves depressurizing the house with a blower door (typically to -50 Pa) with the air handler off, and then sequentially placing a gasketed pan (or other occlusion device, such as tape or a pillow) over each supply or return register and measuring the pressure difference across the pan. The size of the pressure reading indicates the degree to which the duct in question is connected to the outdoors: Large readings indicate significant leaks. This is similar to zone pressure testing of unconditioned zones in a house, but specially crafted for ducts.
Interpreting pressure pan readings can be tricky. The closer a given hole in the ducts is to the register being measured, the larger the pressure pan reading for that register. Therefore, small readings at a number of registers may indicate a large leak that is far away from the reading locations or small leaks at the boots; the energy implications for these two types of leak are very different. Also, registers that are close together (that is back-to-back through an interior wall) can lead to false low readings.
For this study, pressure pan readings were taken for all supply and return registers while the house was depressurized to 50 Pa.
Testing the Test Methods
We found that duct pressurization generally yielded higher estimates of duct leakage than either nulling or Delta-Q, suggesting that duct leakage under actual operating pressure is usually less than the 25 Pa level used for standard duct pressurization testing (see Figure 1). Just as one should not consider a blower door test to be a measurement of actual infiltration rates, so one should not assume that duct leakage from pressurization testing reflects leakage under actual operating conditions.
We also found that the reliability of measured leakage to the outside from duct pressurization tests depends on the level of leakage to the interior: High interior leakage makes for less-precise measurements of exterior leakage. This happens for two reasons. First, a substantial amount of air from the pressurization fan can flow through interior leaks if the duct-to-house pressure deviates even slightly from zero, making measurements of the flow at zero duct-to-house pressure inherently less precise. Second, large leaks to the interior can create nonuniformities of pressure in the duct system. These nonuniformities can create flows to the interior that add to—or subtract from—leakage to the outside. For both of these reasons, it is very important to seal registers properly before conducting leakage-to-outside tests using duct pressurization. And for both of these reasons, exterior leakage measurements in homes that use building cavities or panned joists as duct runs are subject to substantial error.
The test data indicate that the Delta-Q method of assessing duct leakage has promise, but the results are sensitive to wind, as evidenced by the fact that we sometimes obtained Delta-Q estimates that deviated substantially from zero during control tests when the air handler was never operated. Working with Collin Olson at the Energy Conservatory, we developed an approximate method to assess the reliability of Delta-Q estimates of duct leakage, but additional methodological work is under way in this area (see “State of Delta-Q”).
Finally, the pressure pan data generally did a good job of flagging homes where duct leakage was a problem. This suggests that pressure pan readings can probably be used to screen for significant duct leakage for homes that have low or moderate exterior duct exposure and have the air handler in a conditioned basement. If the high-pressure portions of the duct system are inside the conditioned space, then pressure pan readings are likely to detect any but the most trivial leakage. For all types of homes, high pressure pan readings indicate sizable holes somewhere in the duct system.
Assessing Wisconsin’s Ducts
In the 15 low- and moderate-exposure homes, the test results suggested duct leakage under actual operating conditions that was generally less than 5%. The four sites with most or all ductwork outside the conditioned envelope, however, showed considerably higher leakage rates, as Figure 1 shows. (Note that the analysis methods used for nulling and Delta-Q tests allow for negative estimates of duct leakage, even though this is physically nonsensical. This is done to avoid a positive bias in estimated leakage when measured leakage is close to zero.)
We used methods specified in ASHRAE Standard 152 to estimate overall duct system losses, based on the characteristics of the duct system, and on the measured leakage for each site. This method takes into account duct location and insulation level as well as measured leakage to the outside to estimate seasonal losses due to conduction and leakage. (We generally used the nulling estimates of duct leakage for this analysis, but we used the median of the Delta-Q estimates in cases where the nulling results were not available or were not reliable.)
The results suggest seasonal duct losses of less than 5% for the sites with minimal exterior duct exposure; of 5%–15% for the sites with a mix of interior and exterior ductwork; and of up to 30% for the four sites with mostly exterior ductwork (see Figure 2, p. 32, which shows heating losses). Note, though, that for all but the four high-exposure sites, the estimates suggest that duct leakage plays very little role in overall duct system losses; rather, conduction losses dominate the estimated losses.
As a point of reference, for a typical new Wisconsin home with about 1,000 annual hours of heating system operation and 300 annual hours of cooling, 10% distribution system losses would add about $95 per year to heating costs (at
$1.25 per therm) and about $7 per year to cooling costs (at $.10/kWh)—assuming that the system air handler is not operated continuously.
Current RESNET standards require a default assumption of 12% duct system distribution efficiency losses where ducts are located in conditioned space, and 20% losses for ducts located outside of conditioned space. Based on this standard, untested low-exposure homes would be subject to default estimated duct losses that were close to 12%, moderate-exposure homes would have defaults somewhere between 12% and 20%, and high-exposure homes would have default distribution losses of about 20%.
When these defaults are compared to the estimates based on the test data, only the high-exposure homes appear to be comparable; the default assumption for duct system losses in the low- and moderate-exposure homes are significantly higher than the field data would suggest.
The results strongly suggest that the best use of valuable on-site testing time may be to use simple pressure pan screening for low- and moderate-exposure homes, where the air handler and the high-pressure parts of the duct system are inside conditioned space. However, current standards require home performance professionals to choose between a full-fledged duct pressurization test or relatively high default duct loss estimates, and thus discourage simpler diagnostics and screens for homes that have little exterior duct exposure. We encourage the refinement of duct leakage test standards to accommodate simpler duct leakage diagnostics in these types of home.
Scott Pigg is a senior project manager at the Energy Center of Wisconsin, a private nonprofit based in Madison, Wisconsin, that conducts objective research and provides information and continuing education on energy issues to businesses, professionals, and policymakers nationwide.
Paul Francisco is a research specialist at the Building Research Council of the University of Illinois, which focuses on researching building energy and indoor air quality issues.
For more information:
The full technical report for this study can be downloaded from the Energy Center of Wisconsin at www.ecw.org.
You can also download the study by Dickerhoff and Walker. Go to http://epb.lbl.gov/publications/residential_ted.html
Enter your comments in the box below:
(Please note that all comments are subject to review prior to posting.)
While we will do our best to monitor all comments and blog posts for accuracy and relevancy, Home Energy is not responsible for content posted by our readers or third parties. Home Energy reserves the right to edit or remove comments or blog posts that do not meet our community guidelines.