Energy Conservatory Articles
Measured Duct Leakage, Mechanical System Induced Pressures and Infiltration in Eight Randomly Selected New Minnesota Houses.
By Gary Nelson, Robert Nevitt, John Tooley and
Neil Moyer (prepared for the 1993 EEBA/NESEA Conference)
Introduction
Previous investigations have found that residential forced air heating and cooling systems often dramatically increase infiltration rates of houses when the air handler fan is operating. In addition, residential forced air distribution systems have been found to cause negative pressures capable of backdrafting natural draft combustion appliances. One study of a group of 50 Florida houses (Cummings, et. al., 1990) found that infiltration rates more than doubled from 0.21 to 0.46 air changes per hour when the air handler was turned on due to duct leaks connected to the outside. Infiltration rates in these house increased further to 0.61 ACH when interior doors were closed while the air handler was running as a result of pressures associated with distribution system imbalances. These observations have been made in regions where much of the ductwork for forced air systems runs outside of the thermal envelope and it is standard practice to use a single (or double) return grill in a central location.
In Minnesota and many other parts of the US and Canada, most ductwork runs inside of the thermal boundary and return grills are installed in almost every room. While it has been thought that this style of duct system eliminates the majority of pressure imbalance and duct leakage problems found in single return systems, little research and monitoring has been performed to support this conclusion. The purpose of this study was to measure the effect of the forced air distribution system on infiltration rates and house pressures in typical new Minnesota houses.
House Selection
The 8 houses in this study were selected at random from new house permits granted in 1990 by three municipalities in the Minneapolis-St. Paul metropolitan area. The 8 study houses are a subset of a larger group of houses currently being studied by the University of Minnesota Cold Climate Housing Center. All 8 houses have natural gas fired forced air heating systems located in conditioned basements. Every house has return grills installed in all habitable rooms, except for bathrooms and some of the basements. All ductwork for first floors are located in the basements, typically in basement ceiling cavities. In two story houses, most ducts go from the basement through interior partition walls to the floor cavity between the first and second floors. One or two houses had some ducts running through exterior wall cavities to the second floor. No ducts for these houses are located in attics or crawl spaces (none of the houses have crawl spaces). All of the house visits were completed during the week of June 1, 1992.
Airtightness Measurements of Houses
Blower door depressurization tests were conducted on each house at a house to outside reference pressure of 50 Pascals. All interior doors, including the door between the basement and the rest of the house were open for the test. All intentional openings were left open during the blower door tests, except the combustion air inlets which were sealed. The airtightness test results are shown in Table 1.
Table 1
| House ID | House CFM50 | Duct CFM50 | House Volume | House ACH50 | House ACH (Nat) |
| 1 | 1745 | n.a. | 32000 | 3.27 | 0.21 |
| 2 | 2330 | 2680 | 34500 | 4.05 | 0.26 |
| 3 | 1820 | 3065 | 27000 | 4.04 | 0.26 |
| 4 | 1300 | 1365 | 15400 | 5.06 | 0.33 |
| 5 | 1375 | 1926 | 16200 | 5.09 | 0.33 |
| 6 | 1475 | 2085 | 24600 | 3.60 | 0.24 |
| 7 | 1650 | 1916 | 20500 | 4.83 | 0.32 |
| 8 | 2375 | 4695 | 41000 | 3.48 | 0.23 |
| Avg. | 1759 | 2217 | 26400 | 4.18 | 0.27 |
The average blower door CFM50 measurement was 1,759 with a corresponding air changes per hour (ACH50) of 4.18. These results are slightly higher than a 1984 study of 64 new Minnesota houses which found average CFM50 and ACH50 readings of 1,390 and 3.70 respectively. The average annual natural infiltration rate for the 8 houses is estimated to be 0.27 air changes per hour, which is below the ASHRAE recommended ventilation guideline of 0.35 air changes per hour. (ASHRAE, 1989)
Duct Leakage Measurements
A prototype duct tester was used to measure the airtightness of each of the duct systems. All supply registers and return grills were taped shut and the blower compartment access panel on the furnace was removed. A piece of cardboard was fabricated to allow the duct tester's calibrated fan to be sealed to this opening so that the duct tester would pressurize the duct system. The duct system pressure was measured at the supply plenum. Using this testing method, the measured duct leakage includes both leakage to the inside of the house, as well as leakage to the outside.
Even though the duct tester was capable of moving 1,000 cfm of air, it wasn't possible to get any of the duct systems up to the target pressure of 50 Pascals. The duct tester was turned to full power and the maximum achievable pressure and the corresponding airflow were measured. In order to estimate the flow required to pressurize the ducts to 50 Pa, an extrapolation was done assuming a flow exponent of 0.65. These results are also found in Table 1.
The ducts were found to be extremely leaky, with an average leakage of 2,217 CFM50. In every case, the flow required to pressurize the ducts by 50 Pascals was greater than the flow necessary to depressurize the entire house by 50 Pa. We were not able to accurately differentiate between duct leaks to the inside of the house and duct leaks to the outside because of the larger than expected leakage between the ducts and the interior of the house. Measurements taken did show that leakage to the outside was very small compared to interior leakage. We estimate that at least 90-95% of the measured leakage was to the inside of the house. For future testing, a protocol has been developed which will allow accurate measurement of outside duct leakage in these type of houses.
In all of the houses, it appeared that the largest leaks were in the return side of the systems. All 8 houses made extensive use of building cavities for return ducts. Sheet metal panning was added to the underside of the floor joists for the first floor to serve as return ducts. Wood or sheet metal blocking at the ends of these ducts fit very poorly and weren't sealed. In two story houses partition walls and floor joist cavities between the first and second floor were often used to get returns to the second floor. Supply ducts were also very leaky.
Tracer Gas Tests
Three separate sulfur hexaflouride tracer gas decay tests were done at each of the eight houses in order to measure the infiltration rate under three different conditions. The first test was done with all interior doors open and the air handler off. For the second test, the air handler was turned on and interior doors were left open. For the third test, the air handler was left running and all interior doors were closed. Each room contained an operating oscillating fan to assist the internal air mixing. Tracer gas concentrations were measured at four to six locations in each house and averaged every 10 minutes. Each test was run for approximately one hour. Wind speed and temperature were monitored with a weather station that was located in the front yard of each house. Tracer gas test results are given in Table 2.
Table 2
| House ID | ACH (Nat) Air Handler Off | ACH (Nat) Air Handler On | Difference Air Handler (Off to On) | ACH (Nat) Air Handler On (D.C.) | Avg Wind (MPH) |
| 1 | 0.08 | 0.08 | 0.00 | 0.09 | 0.90 |
| 2 | 0.13 | 0.15 | 0.02 | 0.15 | 0.80 |
| 3 | 0.11 | 0.15 | 0.04 | 0.29 | 2.60 |
| 4 | 0.18 | 0.20 | 0.02 | 0.16 | 2.30 |
| 5 | 0.21 | 0.27 | 0.06 | 0.26 | 5.60 |
| 6 | 0.16 | 0.28 | 0.12 | 0.23 | 4.80 |
| 7 | 0.10 | 0.22 | 0.12 | 0.17 | 0.40 |
| 8 | 0.06 | 0.13 | 0.07 | 0.14 | 0.10 |
| Avg. | 0.13 | 0.19 | 0.06 | 0.19 | 2.19 |
Differences between the first two tracer gas tests (interior doors open - air handler off and on) should be a measurement of infiltration caused by duct leakage to or from the outside of the house. As shown in Table 2, the average difference between the first and second tests is only about .06 air changes per hour, suggesting that there is very little duct leakage to the outside. Four of the houses had combustion air inlets directly ducted to the return side of the forced air system. In order to measure only unintentional duct leaks, we sealed all combustion air inlets for the tracer gas tests.
Differences between the second and third tracer gas tests (air handler on - interior doors open and closed) should be a measurement of infiltration induced by pressures caused by imbalances in the supply and return flows to various rooms in the houses. The average measured increase in infiltration due to door closure was insignificant.
Measured Room Pressures
When the air handler is on and interior doors are closed, a room will be pressurized or depressurized if the airflow supplied to the room through supply registers is more or less than the airflow removed through return grills. The magnitude of the induced pressure depends on the tightness of the room and the imbalance in the flows. In each of the houses the interior doors were all closed, the air handler was turned on, and the induced room pressures were measured.
For the most part, room-to-room pressures found in the 8 Minnesota houses were small compared with measured results from other parts of the country where single or double return systems are commonly used. Few room pressures exceeded 1 Pascal relative to the outside when doors were closed. This is primarily a function of having return grills installed in all rooms of the house.
A notable exception to this finding is the fact that a number of basements were significantly depressurized with the air handler running and the basement door closed. In one house, the basement was depressurized by 6 Pascals and in another, the basement was depressurized almost 5 Pascals by the air handler alone. This was due to large leaks in return ducts located in the basement. Basement depressurization can be a severe combustion safety problem and is discussed in more detail below.
Combustion Safety
A set of tests was performed on each house to determine if, under worst case conditions, spillage or backdrafting of combustion products into the house is likely. Research done for the Canadian Mortgage and Housing Corporation has shown that natural draft furnaces and water heater heaters, with properly designed and installed vents, begin having problems venting properly if they are in spaces that are depressurized by more than about 5 Pascals. Fireplaces start having problems maintaining adequate draft at about 3 Pascals according to the Canadian research. (It has been our experience that water heaters often have backdrafting problems in the summer at pressures as low as 2 or 3 Pascals.)
For this study, we determined the largest negative pressure that could be induced in all rooms containing either a naturally vented furnace, water heater, or fireplace; due to the operation of installed mechanical equipment. All exhaust fans, including dryers and central vacuum cleaners, were turned on and the position of interior doors was determined that created the largest negative pressure in the combustion appliance room. This test was repeated with the air handler fan running. Four of the eight houses tested had measured worst case depressurization in the combustion appliance zone (C.A.Z.) which exceeded the Canadian limits. All six fireplaces in the test houses experienced worst case depressurization in the fireplace zone (F.P.Z.) in excess of the 3 Pascal limit for fireplaces.
With the house at the determined worst case condition, natural draft furnaces and water heaters were turned on and checked for spillage. Any spillage after one minute of burner run time was considered a failure. Three of the five natural draft water heaters in the study houses were found to spill beyond the one minute limit under worst case conditions. In one of the houses (#3), the water heater spilled whenever the basement door was shut and the air handler was operating, due solely to basement depressurization from leaky return ducts.
Of the eight furnaces in the test houses, only one was a natural draft appliance (#4). This furnace passed the spillage test. However, one of the induced draft furnaces was found to be spilling through the water heater draft diverter for approximately 30 seconds.
We did not light fires in fireplaces while measuring negative pressures and checking for spillage at furnaces and water heaters, although this would have been a much stricter worst-case depressurization test. Similarly, we did not turn furnace or water heater burners on when measuring negative pressures in the fireplace zone. The one house that had a fireplace and a natural draft appliance that passed the worst case spillage test, house #6, probably would have failed if there had been a fire in the fireplace.
It may seem that the above worst case testing is too severe a test since there is only a very slight chance that all fans are on for a long enough time to spill large quantities of combustion gasses into the house. However, it has been observed by us and many others that in cold weather once a vent backdrafts and gets cold it will often continue backdrafting (sometimes for days) even after the source of negative pressure that started the backdraft has been removed.
In five houses, basement depressurization was caused by closing only the basement door with the air handler running. Basements were depressurized from 1 up to 6 Pascals, making combustion product spillage more likely. The measured depressurization was caused by the air handler drawing more air from the basement than it supplies back to the basement. This is a function of severe return duct leaks in all of the houses, and at least three of the houses had return registers located in the basement. Section 706.f of the 1988 Uniform Mechanical Code, adopted with amendments as the Minnesota Mechanical Code, appears to prohibit the drawing of return air from all the basements in this study.
Our interpretation of the 1988 Uniform Mechanical Code also suggests that all the houses experiencing spillage under worst case are in violation of the Minnesota code. Section 606 of the 1988 UMC says that "Operation of exhaust fans, kitchen ventilation systems, clothes dryers or fireplaces shall be considered in determining combustion air requirements to avoid unsatisfactory operation of installed gas appliances." We saw no indications that there was any such consideration in any of the houses in this study.
Measured Airflow Through Combustion Air Intakes
All 8 houses are equipped with combustion air inlets. Four houses have ducts that run from the outside and are hard ducted to a connection point somewhere on the return air side of the forced air distribution system. This connection point varies from right at the return plenum to the very end of the longest return duct. When the air handler is running, the negative pressure in the return system is designed to pull in the combustion air, which gets distributed to the furnace room via a non-closeable register in the supply plenum. The other four houses have passive combustion air inlets. This consists of a duct running from outside to a spot near the floor next to the furnace.
A flow measuring station was mounted in each combustion air duct. In the houses with hard ducted inlets, the airflow was measured with the air handler running. In all the houses the airflow was measured with the house depressurized to 50 Pascals with a blower door. From this we calculated the flow that would occur at a house depressurization pressure of 5 Pascals, the house depressurization limit (or HDL) suggested in much of the Canadian literature as the point at which pressure induced spillage and backdrafting start becoming a problem. The results of these measurements and calculations appear in Table 3. We also give in Table 3 a calculation of the total exhaust flow that would be needed to depressurize the entire house to the HDL of 5 Pascals, with the combustion air inlet open.
Table 3
| House ID | Type of Air Inlet | Size of Air Inlet | Airflow at 5 Pa (CFM) | Airflow at 50 Pa (CFM) | Airflow w/ A.H. On (CFM) | Exhaust Flow to Depress by 5 Pa (CFM) |
| 1 | Passive | 6" | 26 | 83 | n.a. | 415 |
| 2 | Passive | 6" | 25 | 78 | n.a. | 545 |
| 3 | Hard Duct | 4" | 8 | 27 | 11 | 415 |
| 4 | Hard Duct | 4" | 21 | 66 | 20 | 310 |
| 5 | Hard Duct | 4" | 19 | 61 | 29 | 325 |
| 6 | Passive | 6" | 24 | 75 | n.a. | 355 |
| 7 | Passive | 4" | 16 | 52 | n.a. | 385 |
| 8 | Hard Duct | 6" | 22 | 71 | 63 | 555 |
It can be seen from Table 3 that very little air is coming into the house through these inlets under normal operation (see 5 Pa flow column). When compared to the large total installed exhaust capacity in a typical house, often two to four hundred cubic feet per minute plus another 200 or so when there is a fireplace, it becomes clear that the presence of a combustion air inlet does little to reduce the chances that pressure induced venting problems will occur. Combustion air inlets as typically installed in Minnesota simply do not provide sufficient airflow to relieve negative pressures caused by large exhaust appliances or forced air distribution imbalances.
Carbon Monoxide Testing
Carbon monoxide (CO) levels in flue products were measured in 5 of the furnaces and 6 of the water heaters found in the test houses. CO levels were recorded after approximately 5 minutes of appliance run time. All tested furnaces and water heaters had very low CO readings with the highest recorded reading of 13 PPM. Many weatherization programs use 100 PPM as an action threshold, while AGA standards allow as much as 400 PPM in the flue products of furnaces and water heaters.
Three gas cooking stoves were also tested for CO and both were found to have between 40 and 90 PPM directly in the oven vent. When the self-cleaning oven was turned on for one stove, the oven CO level increased to 183 PPM after 10 minutes of operation.
Observations
1. House #2 has a single forced air furnace with automatically operated electric zone dampers in the supply plenum. There are no zone dampers on the return side of the system, so return air is always drawn from all areas of the house. If the basement thermostat is calling for conditioned air and the first floor is not, all of the supply air is directed to the basement zone, which contains bedrooms, a children's play room and a family room with a wood burning fireplace. Under this condition (and with the door between the basement and the rest of the house closed) the first floor of the house was found to be depressurized by 7 Pascals, enough to cause serious backdrafting problems with the fireplace in the living room. When the first floor thermostat is calling for conditioned air and the basement zone is not, all the supply air goes to the first and second floors and the basement is depressurized by 6 Pascals, potentially backdrafting the fireplace in the family room.
If there had been a natural draft water heater in the furnace room in the basement, it would have failed our combustion safety tests. It seems to us that serious safety problems may be common with zoned systems of this design.
2. A common complaint in two story houses in Minnesota is that basements get too cold and second floors get too hot in the summer when the central air conditioner is running. A common recommendation by contractors is to run the air handler fan continuously to even out the temperatures. In light of the tremendous amount of duct leakage in basements, it's not surprising that basements get overcooled in the summer. A large percentage of the cooling is delivered to a space (through supply leaks) that has little, if any, cooling load. House #3 was found to have this problem and the occupants were running their air handler fan continuously, as recommended by their builder. When the basement door was closed, the basement was depressurized by 4.7 Pa and the water heater backdrafted continuously, with 100% of its combustion products dumping into the house.
We also noticed in House #3 that when the basement door was closed, air was leaving the house through the combustion air inlet, even though the inlet was hard ducted to the end of a long panned under return air duct. To our surprise, we found the end of this return duct to be positive in pressure with respect to outside, even when the air handler fan was operating. The reason for the pressurized return duct appears to be that the duct is very well connected to the upstairs (through floor leaks in the panned under joist) and the upstairs is positively pressurized by the air handler.
3. House #7 has a natural draft water heater and an induced draft furnace that are vented into the same class B vent. Under worst case depressurization, the furnace combustion products spilled out of the water heater's draft diverter for about 30 seconds. Although the furnace passed this test because it wasn't spilling at 1 minute, we are concerned about this situation. It is our understanding that induced draft furnaces are required to have devices that turn them off in the event that the vent becomes clogged. It appears that these safety devices won't work if the flue gasses can get out through the draft diverter of another appliance. All of the authors of this report have noticed that significant spillage of this type is common.
4. In two houses with baseboard supply registers on the first floor, we measured several Pascals of positive pressure in the exterior wall cavities directly above several registers. Whenever the furnace fan is running, air behind these registers is being blown under the gypsum board, between the gypsum board and the bottom plate, and into the exterior wall. This could result in significant condensation problems in these walls in the winter.
Conclusions
The increase in infiltration due to running the forced air distribution fans in typical new Minnesota houses was found to be much smaller than has been measured in warmer climates where much of the duct work runs outside of the thermal envelope and the use of a single return grill is standard practice. Average infiltration rates in the eight Minnesota houses increased only .06 air changes per hour as a result of air handler induced duct leakage. Measured pressure differentials due to door closure were also much lower than those measured in single return systems, with few room pressures exceeding 1 Pascal. It appears that the typical Minnesota construction practice of putting the ductwork inside the building envelope and reducing room-to-room pressures imbalances by installing return grills in all rooms, effectively reduces the impact of the distribution system on measured air infiltration.
Even though typical air handler induced room pressures were found to be very low, serious combustion safety problems were found in all but one of the eight houses. In two houses, we found that closing the basement door and running the furnace fan either completely backdrafted a natural draft water heater or created sufficient negative pressure to likely backdraft a fireplace. In the other 5 houses that failed our safety tests, it took a combination of the furnace fan running, exhaust fans running, and/or interior door closure to cause the depressurization problem. The one house that passed our test could still have problems in the summer when the air conditioner is running since it may take even less than 5 Pascals of negative pressure to backdraft a water heater in the summer.
Measured infiltration rates and estimated annual infiltration rates were all low. The average estimated annual infiltration rate for the eight houses was 0.27 air changes per hour. It appears that during much of the year the amount of ventilation provided by natural infiltration will be substantially below recommended ventilation rates in the ASHRAE ventilation standard. None of the eight houses tested contained a whole house mechanical ventilation system, however every house had at least one or two spot ventilation fans in the bathrooms or kitchen.
Combustion air inlets in these 8 houses were found to have very low airflow rates. It appears that code approved combustion air inlets in Minnesota houses can not be relied upon to keep pressures induced by exhaust fans low enough to assure proper venting of natural draft appliances.
Houses with zone dampers that shut off the supply air to zones containing fireplaces or other natural draft appliances, while still drawing return air from these zones, may be particularly susceptible of combustion gas spillage and backdrafting. This is based on observations in only one house and should be further investigated.
Acknowledgements
This study was funded by Natural Florida Retrofit of Montverde, Florida and The Energy Conservatory of Minneapolis, Minnesota. Thanks to Pat Huelman from the University of Minnesota Cold Climate Housing Center for providing the houses. Many thanks to Joe Kuonen of Energy Rated Homes of America in Little Rock, Arkansas and Steve Klossner of Advanced Certified Thermography in St. Paul, Minnesota for donating an entire week of their time to help with the field investigations. Also a special thanks goes to the homeowners who put up with a large group of researchers doing strange things in their new houses.
References
Cummings, J.B., Moyer, N., and Tooley, J.J. 1990. "Radon Pressure Differential Project, Phase II: Infiltration." Report number FSEC-CR-370-90. Florida Solar Energy Center.
Scanada-Sheltair Consortium Inc. 1988 "Chimney Safety Tests Users' Manual: Procedures for Determining the Safety of Residential Chimneys." Report prepared for Research Division, Canada Mortgage and Housing Corporation.
ASHRAE 1989. ASHRAE Standard 62-1989, "Ventilation for Acceptable Indoor Air Quality." American Society of Heating
