Content uploaded by Max Howard Sherman
Author content
All content in this area was uploaded by Max Howard Sherman on Aug 05, 2015
Content may be subject to copyright.
1
LBNL-43724
Distribution Effectiveness and Impacts on Equipment Sizing
for Residential Thermal Distribution Systems
Walker, I., Sherman, M., and Siegel, J.
Environmental Energy Technologies Division
Energy Performance of Buildings Group
Lawrence Berkeley National Laboratory
Berkeley, CA 94720
June 1999
This report was prepared as a result of work sponsored by the California Energy Commission
(Commission), through a contract with the Regents of the University of California, California Institute for
Energy Efficiency (CIEE). It does not necessarily represent the views of the Commission, its employees,
the State of California, The Regents, or CIEE. The Commission, the Regents, the State of California,
CIEE, their employees, contractors, and subcontractors, make no warranty, express of implied, and
assume no legal liability for the information in this report; nor does any party represent that the use of this
information will not infringe upon privately owned rights. This report has not been approved or
disapproved by the Commission or CIEE, nor has the Commission or CIEE passed upon the accuracy or
adequacy of the information in this report.
2
Table of Contents
TABLE OF CONTENTS.....................................................................................................................................2
EXECUTIVE SUMMARY..................................................................................................................................4
INTRODUCTION................................................................................................................................................5
1. DUCT LEAKAGE DIAGNOSTICS................................................................................................................5
Summary of duct leakage diagnostics in previous phases................................................................................5
A New Duct Leakage Test: DeltaQ.................................................................................................................6
ASTM Duct Leakage Standard (E1554)..........................................................................................................7
2. DUCT SEALANTS AND LONGEVITY TESTING.......................................................................................7
3. DUCT SYSTEM INTERACTIONS WITH SYSTEM SIZING......................................................................9
FIELD MEASUREMENTS ........................................................................................................................................ 9
Table 1. Diagnostic Test Results...............................................................................................................................10
Table 2. System Capacity Comparisons.....................................................................................................................11
Continuous Monitoring ................................................................................................................................11
Table 3. Performance Metrics....................................................................................................................................13
COMPUTER SIMULATIONS..................................................................................................................................14
Simulation Model Improvements ..................................................................................................................14
Figure 1. Simulations of Pulldowns from 3:00 p.m. on a Sacramento Design Day......................................................15
Extension of previous simulations................................................................................................................. 15
Table 4. List of REGCAP Simulation Cases..............................................................................................................16
Table 5. Start Time to Pulldown by 5:00 p.m. ...........................................................................................................16
Table 6. Relative Energy Consumed in Order to Pulldown by 5:00 p.m.....................................................................17
Table 7. Model Delivered Capacity (TAR) Comparison (system on for 1.75 hours) ...................................................17
COMPARISON OF FIELD MEASUREMENTS AND COMPUTER SIMULATIONS .............................................................. 18
Figure 2: Modeled and Measured Attic Temperatures at Site 4 on August 11, 1998....................................................19
Attic Temperature ........................................................................................................................................19
Figure 3: Modeled and Measured House Air Temperatures at Site 4 on.....................................................................20
August 11, 1998 ........................................................................................................................................................20
House Temperature......................................................................................................................................20
Figure 4: Modeled and Measured Return Duct Air Temperatures at Site 4 on August 11, 1998 ..................................21
Return Duct Air Temperature....................................................................................................................... 21
Figure 5: Modeled and Measured Supply Duct Air Temperatures at Site 4 on August 11, 1998 ..................................22
Supply Duct Air Temperature....................................................................................................................... 22
4. SUPPORT FOR TITLE 24 AND HERS........................................................................................................23
5. TECHNOLOGY TRANSFER.......................................................................................................................24
ASHRAE: RATING OF DISTRIBUTION SYSTEMS - ASHRAE 152P....................................................................... 25
ASTM: RATING OF DUCT SEALANTS AND REVISING DUCT LEAKAGE MEASUREMENT METHODS .............................. 25
OTHER THERMAL DISTRIBUTION SYSTEM EFFICIENCY SUPPORT ACTIVITIES........................................................ 25
Health and Safety Assessment of Aerosol Sealant (EPA)............................................................................... 25
Field Testing of Energy Star Equipment (EPA).......................................................................................... 25
Developing Energy Star Ratings for Duct Systems (EPA) ............................................................................. 25
Public Dissemination of Research Results....................................................................................................26
6. REFERENCES...............................................................................................................................................26
3
7. RECENT PUBLICATIONS ..........................................................................................................................28
8. APPENDICES................................................................................................................................................29
APPENDIX 1: ASHRAE SP152P DUCT LEAKAGE WORKSHOP SUBCOMMITTEE MEETING ..................................... 29
Table A1.1 Duct Leakage Workshop Attendance........................................................................................................29
Summary...................................................................................................................................................... 29
Chuck Gaston: "Inverse" Test....................................................................................................................... 30
Paul Franciso: Nulling Pressure Testing......................................................................................................30
John Andrews: Combined HPT and Pressurization (hybrid).......................................................................... 31
Gary Nelson: Improved Blower Door Subtraction ........................................................................................ 31
ASHRAE 152P Efficiency Limits Due to Extremes of Duct System Pressure Variation.................................. 32
Table A1.2 Sacramento, CA ASHRAE 152P Distribution System Efficiency, %.......................................................33
Table A1.3 Fargo, ND ASHRAE 152P Distribution System Efficiency, %................................................................34
APPENDIX 2. DELTA Q DUCT LEAKAGE TEST...................................................................................................... 35
Derivation of DeltaQ test ............................................................................................................................. 35
Uncertainty Estimate for exponent and duct pressure assumptions...............................................................36
Table A2.1 DeltaQ Sensitivity Test............................................................................................................................37
Flow Adjustments for Exact Pressure Matching............................................................................................ 37
Comparison to other measurements.............................................................................................................. 38
Table A2.2 Comparison of duct leakage measurement procedures............................................................................38
APPENDIX 3. SUMMARY OF FIELD MEASUREMENT PERFORMANCE METRICS..........................................................39
Table A3.1 Tons At the Register................................................................................................................................39
Table A3.2 Capacity at the indoor coil ......................................................................................................................40
Table A3.3 System Power consumption ....................................................................................................................41
Table A3.4 Key Temperatures and Enthalpies for calculating system performance .....................................................42
Table A3.5 Temperature at different locations in the house during pulldown tests......................................................43
Table A3.6 Delivery Effectiveness.............................................................................................................................44
Table A3.7 Equipment Coefficient of Performance (COP)..........................................................................................45
Table A3.8 Total System Coefficient of Performance (COP) ......................................................................................46
Table A3.9 Pulldown time and temperature variation in different locations in the house.............................................47
APPENDIX 4. FLOWCHART FOR REGCAP MODEL.............................................................................................. 48
APPENDIX 5. REGCAP SIMULATION SENSITIVITY TO INPUT DATA UNCERTAINTY ................................................ 49
Table A5.1. Comparison of measured and modeled temperatures illustrating problems with measured input data......49
APPENDIX 6. EVALUATION OF FLOW HOOD MEASUREMENTS OF RESIDENTIAL REGISTER FLOWS............................. 50
Table A6.1 Flowhood characteristics..........................................................................................................................50
Table A6.2 Comparison of flowhood measurements of supply registers (cfm)............................................................51
Static vs. total pressure balancing................................................................................................................ 51
Changing balancing pressure measurement location.................................................................................... 51
Comparing return measurements.................................................................................................................. 51
Table A 6.3 Comparison of flowhood measurements of return register flow (cfm) ......................................................52
4
Executive Summary
This report builds on and extends our previous efforts described in "Leakage Diagnostics, Sealant Longevity,
Sizing and Technology Transfer in Residential Thermal Distribution Systems- CIEE Residential Thermal
Distribution Systems Phase VI Final Report, December 1998". This report concentrates on a new area of work: the
interaction between distribution system effectiveness and equipment sizing. This issue focuses on the ability of
downsized equipment with a good distribution system to deliver the same cooling to conditioned space as a typical
Heating, Ventilating and Air Conditioning (HVAC) system. The cooling of the conditioned space is evaluated by
looking at the concept introduced in the previous phase of this study: “Tons At the Register" together with comfort
issues, such as how quickly a house is cooled (“pulldown time”), and the distribution of cooling throughout the
house.
The key outcomes of this study are:
• This investigation yielded a new duct leakage test called DeltaQ.
• The existing ASTM Standard (E1554) for measuring duct leakage has been rewritten and submitted to the
ASTM standards review process.
• A draft ASTM standard for longevity testing of duct sealants was developed. A draft was submitted to ASTM
subcommittee E06.41 for balloting and comment. The comments on the draft resulted in changes to the test
method and apparatus. A new test apparatus was constructed with funding from the Department of Energy
(DOE).
• Simulations of summer temperature pulldown time have shown that duct system improvements can be
combined with equipment downsizing to save first cost, energy consumption, and peak power and still provide
equivalent or superior comfort.
• Air conditioner name plate capacity ratings alone are a poor indicator of how much cooling will actually be
delivered to the conditioned space. Duct system efficiency can have as large an impact on performance as
variations in Seasonal Energy Efficiency Ratio (SEER).
• Installation of high SEER units can reduce energy consumption with no apparent drawbacks
• Duct efficiency calculations are included in the Low-Rise Residential Alternative Calculation Method
Approval Manual for 1998 Energy Efficiency Standards for Low-Rise Residential Buildings” (CEC (1999)).
• Procedures for HVAC System Design and Installation (for Home Energy Raters) have been updated.
• Field testing has shown that standard flowhoods can be poor for measuring residential register flows.
Results from this study were used by the California Energy Commission (CEC) in the formation of the current
Energy Efficiency Standards for Low-Rise Residential Buildings (CEC, (1998)), often referred to as Title 24.
Current information on ducts and thermal distribution research can be found at http://ducts.lbl.gov
5
Introduction
Previous studies (including earlier phases of this research project) have shown that losses from residential thermal
distribution systems have significant energy and comfort implications. This study looks at the potential for
improvement of thermal distribution systems and the possibility of reducing equipment size as a result. These
distribution system and equipment interactions were examined through field testing and computer simulation. In
addition, this report outlines our efforts to transfer the results of this research to the marketplace so as to reduce
energy losses and improve thermal comfort. This study describes the results of efforts made during the Transitional
Phase of this Residential Thermal Distribution Systems research. Results of earlier Phases were described in
Walker et al. (1997 and 1998).
1. Duct Leakage Diagnostics
The objectives of this task were:
• Improve duct leakage test methods.
• Update the American Society for Testing and Materials (ASTM) Standard E1554 – “Determining
External Air Leakage of Air Distribution Systems by Fan Pressurization”
Summary of duct leakage diagnostics in previous phases
In Phase V of this work we performed field evaluations of several diagnostic techniques for measuring duct
leakage:
• House Pressure Test (HPT).
• Nulling Pressure Test (NPT).
• Duct and house pressurization with separate supply and return leakage.
• Duct only pressurization with combined return and supply leakage.
• Tracer gas.
These tests were evaluated in terms of ease of use, time requirements and the bias and precision errors associated
with each test by using the tests in several houses. The results of the testing indicated that none of these methods
were ideal (hence our continuing work on improving duct leakage diagnostics). However, for screening of low
leakage levels for compliance testing the duct leakage diagnostic of choice is the fan pressurization test of total
duct leakage (test 4). The reasons for this are:
• Robustness. The fan pressurization test has almost no restrictions on the type of system it can be used on, or
the weather conditions during the test.
• Repeatability. Combining the results of both the phase V and VI reports together with the field experience of
other users showed that the repeatability of the pressurization testing was found to be very good.
• Precision. The uncertainty in leakage flow will be small if the allowable leakage is set to a low number because
the uncertainties for the pressurization test scale with the amount of leakage.
• Simplicity. It is easy to interpret the results of fan pressurization without having to perform many (or any –
with the appropriate hardware) calculations. This allows the work crew to evaluate the ducts during the test
and also allows the work crew to ensure that the test has been performed properly because they can see if the
results make any sense.
• Familiarity. Work crews that have performed envelope leakage tests are familiar with the test method for
ducts, because envelope testing uses a similar apparatus and calculation/interpretation methods.
The biggest drawback with this test is the requirement of covering all the registers which can be time consuming.
In addition, this precision of this test is reduced at higher leakage levels that might be found by home energy raters
in existing construction, rather than the low leakage levels required in compliance testing. Because this test
measures the total leakage and not just the leakage to outside it will overestimate the leakage required for energy
loss estimates, however, from a compliance testing point of view, this error is in the right direction because it
6
means that the true losses will be less than those indicated by the test. In other words, a system whose total leakage
passes a leakage specification is guaranteed to have the leakage to outside be less than the specification.
In Phase VI we extended the duct leakage measurements to include separate measurements of the boot and cabinet
leakage because these were thought to be two main leakage sites. The measurement results confirmed this idea:
combining these two leakage sites together accounted for about three quarters of all duct leakage. The average
leakage to the outside was about 25 cfm for the boots and about 34 cfm for the cabinets.
A New Duct Leakage Test: DeltaQ
In order to find a duct leakage test that is better than those discussed above, a duct leakage measurement workshop
was held as part of the ASHRAE Standard 152P (ASHRAE, 1999) committee meetings in January 1999. We have
prepared a summary of this workshop, and it is included as Appendix 1. In addition, we discussed potential
innovative measurement techniques with other researchers throughout the US and Canada.
The result of these discussions is a new technique for measuring duct leakage that we have evaluated using a pilot
study of local homes. This new technique is called the DeltaQ test because it measures changes in flow (Q) caused
by distribution system operation. This new test method has several features that give it the potential for success:
• It has simple equipment requirements. Only a blower door and some pressure sensors are required to perform
the test. The blower door is a common item that most building diagnosticians already have and are familiar
with its operating principles. Some existing tests require less common equipment, for example specialized
combined fan/flowmeters for pressurization tests, or tracer gas analysis equipment.
• It directly measures the value that we want from the test: the leakage to outside at operating conditions of the
supply and return separately. Other existing tests require conversion from measured pressures to operating
pressures, or they require complex balancing of house and duct pressures to obtain leakage to outside rather
than total duct leakage.
• It is quick. There is no requirement for blocking off all the registers or blocking between the supply and return
parts of the system.
• It is robust. Our field testing has shown that the DeltaQ test is not as sensitive to wind induced envelope
pressure fluctuations as the House Pressure Test, or Nulling Pressure Test.
• It does not have the detailed assumptions (that lead to additional uncertainties) about the house envelope that
the House Pressure Test requires.
The DeltaQ test works by using a blower door to maintain the same pressure across the building envelope with the
duct system fan on and off. The flow with the system on and off is measured over a range of envelope pressures.
This results in pairs of flow data (one with the system fan on and one with the system fan off) at several pressures.
As the blower door pressurizes (or depressurizes) the house relative to outside, the pressures in the ducts will also
change relative to outside by the same amount. Because the pressure across the leak changes, the flow through the
leaks changes and this change in leakage flow appears as a change in envelope flow through the blower door. In
addition the operating pressures in the ducts when under normal operating conditions are also measured. These
operating pressures are measured at the plenums because this gives the biggest and most repeatable pressure signal
and avoids the uncertainties of register pressure measurements. Combining the measured system pressures and
the pairs of blower door flow data together with the algebraic analysis of the changes in duct leakage flow allows
the calculation of the supply and return leakage coefficients and pressure exponents. Appendix 2 gives more
details of the derivation and application of the test method.
So far, only six houses have been DeltaQ tested and more houses will be tested in the near future. Of these six
houses, one test was at low wind conditions and gave results that closely matched other measurement techniques.
The second test was on a windy day, but still managed to give reasonable results based on visual observation of the
duct system, i.e., it showed that the ducts were not very leaky. This is a significant result because other tests that
use envelope pressures (HPT, NPT and duct and house pressurization) have not given reasonable results under
windy conditions. The third house was tested on a very windy day (wind speeds > 20 m.p.h. and highly variable)
and the DeltaQ test did not give satisfactory results under these extreme conditions. These three tests have shown
that the DeltaQ test is more robust than most of the existing tests but still fails at the very high wind speeds. Under
7
extremely windy conditions the only test that can be used on ducts is the duct pressurization test because it does not
require envelop pressure measurement or measured flow through a fan flowmeter between the house and outside.
Future work will apply the DeltaQ test to more houses and include repeatability studies.
ASTM Duct Leakage Standard (E1554)
The existing test procedure in E1554 is called the blower door subtraction method and is no longer used by many
researchers due to the poor results obtained from the test. This standard is currently due to be revised by ASTM so
we have prepared a revision of E1554 (ASTM (1999)) that incorporates the DeltaQ test together with the combined
house and duct fan pressurization test from proposed ASHRAE 152P. In addition to revising the standard, we have
also been performing administrative tasks such as attending ASTM meetings and collaborating with ASTM staff to
produce this revised standard. This revision of E1554 will be evaluated by an ASTM Task Group in October 1999.
After initial review by the Task Group, it will take a year or two for the revised draft to become a test method. This
time allows us and other potential users to evaluate the revised procedures in more homes. At the ASHRAE 152P
meetings in June 1999 the ASHRAE 152P committee members were given copies of the test procedure and asked
to use it and report back to us in order that we can build up a consensus of experience with this test method in as
broad a range of homes and test conditions as possible.
Duct leakage diagnostic outcomes:
• This investigation yielded a new duct leakage test called DeltaQ.
• The existing ASTM Standard (E1554) for measuring duct leakage has been rewritten and submitted to
the ASTM standards review process.
2. Duct Sealants and Longevity Testing
The objective of this task was to:
• Develop and introduce a draft ASTM standard for longevity testing of duct sealants
The development of the longevity test method and preliminary results have been discussed in previous phases
(Walker et al. 1997 and 1998). The final results and details of the experiment were given in “Can Duct Tape Take
the Heat” - LBNL report # 41434 and its companion Home Energy Article (Home Energy, Vol. 15, No.4, pp. 14-
19. http://www.homeenergy.org/898ductape.title.html).
The results of work in previous phases of this study have been included in California’s Residential Energy Code
(usually referred to as Title 24). In the Alternative Calculations Manual of Title 24 no cloth backed rubber
adhesive duct tape is allowed as a duct sealant on systems obtaining credit for energy efficient duct systems. This
has caused some consternation on the part of HVAC installers and duct tape distributors; however, we have been
able to show these concerned parties that the test results are real and that there are viable alternatives. In addition,
some of the duct systems that were tested for phase VI of this study, and other systems we have observed over the
last six months have been sealed in accordance with Title 24 requirements and our leakage measurements have
shown them to be well sealed systems.
The longevity test method (ASTM (1999b)) was prepared in ASTM standard format and submitted to ASTM
Subcommittee E06.41 for consideration. The ballot results had only one significant technical comment that the
high temperatures were too low because some attics can be at a higher temperature than those used in the test (150
°F surface temperatures). This comment says that in order for the longevity test to be “accelerated” the attic
temperatures should be at least as high as measured peak temperatures and possibly higher. However, the evidence
for extreme attic temperatures higher than 150°F is poor. A literature search of attic temperature studies was
undertaken to find evidence of higher measured temperatures. Much of the existing literature does not address
peak or extreme values because the studies were interested in estimating energy savings where longer time average
values are needed. However, a few studies were found that gave explicit attic peak temperature information, and
8
those with the highest reported peaks are discussed here. Some of the following studies tested several houses but
we discuss the results from the hottest attic only. Carlson et al. (1992) measured peak attic air temperatures of
155°F. Parker et al. (1997) measured attic temperatures of 134°F in a house in Florida, however, this was an
average of the hottest 2.5% of the summer hours, so peak temperatures would be expected to be higher. The tests
discussed in Phase VI of the current study had a peak attic air temperature of 151°F for a house in Sacramento.
An additional parameter that changes duct temperatures is the radiant exchange between the ducts and the roof
deck surfaces that are hotter than attic air under peak conditions. For example, Wu (1989) measured attic floor
temperatures 7 °F hotter than the attic air. The upper exterior surfaces of ducts in attics are heated by a similar
amount. It is important to consider this increase in surface temperature due to radiation because duct sealants are
generally applied from the outside and will experience these elevated exterior surface temperatures. Combining
the existing peak temperature field data with the radiation effect results in a temperature of about 160°F being a
reasonable target temperature.
Another point of view is to look at the duct temperatures experienced by heating systems which can be higher than
those for cooling systems. Field studies have found that many furnaces are operating on their high-limit switches –
usually set at about 200°F. The Uniform Mechanical Code (ICBO (1994)) has a limit of 250°F (121°C) for furnace
and duct heater controls. The Canadian Natural Gas Installation Code (CGA (1995)) gives the same limit of
250°F (121°C) for forced air systems, but includes a higher limit of 350°F (175°C) for gravity furnaces. This
indicates that the 160°F high temperature limit from the peak attic temperatures would be too conservative for
heating systems. However, the furnace high-limit temperatures will not be used because there is an additional high
temperature limit constraint imposed by duct tape manufacturers of 200°F for some of their products. A reasonable
compromise is to be half way between the upper limit for cooling (160°F) and the limit set for tape (200°F)
resulting in an upper temperature of 180°F. This compromise was chosen to be far enough away from the upper
limit for tape that we can be reasonably sure that the tape does not exceed this limit during testing because the
temperature control in the experimental apparatus has some uncertainty.
The major difference between the heating and cooling values is that the extreme temperatures for cooling ducts in
hot attics occur with the system off and the heating ducts have their extreme with the system running, so the
heating limit may be a more realistic scenario. In addition, the temperature gradient across the duct (hot inside air,
cool surroundings) is the correct situation for the heating duct case. However, it is the explicit duct surface
temperatures that are the temperatures experienced by the duct sealant and so not too much importance should be
placed on the direction of heat transfer through the duct walls.
Based on the above temperature limit changes and other feedback from ASTM members and other interested
parties, in addition to our own research, we have begun development work on a revised longevity test apparatus
funded by the Department of Energy. This new apparatus will allow us to test different procedures for evaluating
longevity. The procedures include:
• The existing procedure of alternating between hot (150°F) and cold (0°F) air flows with a pressure difference
across the seal.
• Changing the temperatures in the alternating temperature test to be more extreme (180°F).
• Splitting the test into a hot test and a cold test. The hot test will vary the temperatures in the range of 70°F
(close to room temperature) to 180°F. The cold test will have the range from 70°F down to 0°F. This
narrower temperature range will allow for more rapid cycling and greater temperature extremes (e.g., we could
possibly go lower than 0°F). For both the hot and cold test there will be a pressure difference maintained
across the leakage sites.
• Having no cycling of temperature and maintaining a steady hot (180°F) or steady cold (0°F) temperature.
Unlike the previous baking tests there will be airflow through these samples and pressure differences across
the leaks.
The leakage testing will be the same as in the previous apparatus. Periodically the samples will be removed for
individual leakage testing of leakage flows at 25 Pa. The leakage of the samples will be measured before any
sealing and immediately after sealing before installation in the test apparatus. The “failure” of a sample will be
9
determined the same as in the previous study, by evaluating the leakage flow as a fraction of the unsealed flow.
The failure level is fixed at 10% of the unsealed leakage. The 10% level was chosen by examining test results to
determine the point beyond which failure can be rapid and difficult to measure. .This 10% leakage also
corresponds to empirical estimates of “unacceptable” leakage for an individual connection.
Duct Sealant longevity testing outcome:
• A draft ASTM standard for longevity testing of duct sealants was developed. A draft was submitted to
ASTM subcommittee E06.41 for balloting and comment. The comments on the draft resulted in changes
to the test method and apparatus.
• A new test apparatus was constructed with funding from the Department of Energy (DOE) and will be
used to evaluate new sealants.
3. Duct System Interactions with System Sizing
The objectives of this task were:
• Measure the performance of residential cooling equipment and associated distribution systems.
• Compare the REGCAP simulation model to the measured field data.
In this study the duct system interactions with system sizing were examined using both computer simulations and
measured data. The measured data were used to examine field performance of cooling systems and to evaluate and
validate the computer simulations but were not used to tune any model coefficients so that the model retains its
general applicability. The following sections discuss the field measurements, computer simulations and
comparisons between the two.
Field measurements
The cooling system performance was measured in six test houses. Each house was tested in several configurations
in order to estimate the effect of duct systems on the capacity, energy performance and comfort. The previous
Phase (Walker et al. (1999)) reported the preliminary results considering sensible Tons At the Register (TAR)
(“delivered capacity”) and capacities only. The current study looks in more detail at pulldown tests and equipment
performance for both latent and sensible cases. The pulldown tests were evaluated by determining the pulldown
time (the amount of time to cool down a house) for different parts of the house: at the thermostat, the master
bedroom and the kitchen. In addition the temperature in each location at the end of pulldown as indicated by the
thermostat was investigated. The differences between these locations indicate the relative comfort for the
occupants. e.g., in the summer, a house where the temperature is much higher for the master bedroom when the
system turns off (end of pulldown at the thermostat) will not be comfortable when the occupants go to bed. This is
a common complaint about air conditioning systems and was specifically mentioned by the people who lived in the
occupied house used for this study.
The field measurements included diagnostics to determine building and system characteristics and continuous
monitoring over several days to determine pulldown system performance. Six houses were monitored for this
project: 2 houses in Palm Springs, CA (sites 1 and 2), one house in Mountain View, CA (site 3), two houses in
Sacramento, CA (sites 4 and 5), and a single house in Cedar Park, TX (site 6). All of the houses were new and
unoccupied, except for the Mountain View house that had been occupied for less than a month at the beginning of
our tests. The houses were tested in their “as found” configuration, then with the duct systems sealed. Houses that
did not have very much "as found" duct leakage had holes added.
In two houses, the cooling equipment was replaced with Energy Star equipment (greater than SEER 13.0). The
original cooling equipment in each house was rated at the federal minimum SEER 10. In Sacramento (site 4), just
the outside compressor unit and the control system were changed. In the Texas house (site 6), the indoor coil, fan
10
and cabinet (and electric heating system) were also replaced. Details of the HVAC systems and house construction
can be found in the report on the previous phase of this work in Walker et al. (1999).
Table 1 summarizes the most significant diagnostic test results for the thermal distribution system and equipment
for the six test houses. The air handler flows for these systems were higher than has been found in previous studies
(e.g., Blasnik at el. 1996) that suggested that most systems typically had about 15% less than the 400 cfm/ton
recommended by manufacturers. In several cases the air handler flow was considerably higher than the 400
cfm/ton benchmark, particularly at site 4 with almost 550 cfm/ton. This high flowrate will limit the ability of the
system to handle latent loads. However, site 4 is located in Sacramento, CA and does not have a high latent load,
so these high flows are probably acceptable. The leakage expressed as a fraction of fan flow is lower than has been
found in previous studies, indicating that these duct systems were better than average installations; in fact, they
were some of the least leaky systems we have tested. The exception to this was surprisingly Site 3, where most of
the duct system was in interior partition walls or dropped soffits between the first and second floor, with none of
the duct system in the attic. A detailed examination of the ducts at Site 3 showed that much of the leakage was at
the plenum to duct connections that were in the garage. In addition, the soffits and partition walls were not air
sealed with respect to the garage or the attic so that air leaking from the ducts did not leak into the conditioned
space but was allowed to escape to outside. This result reinforces the requirement of field testing duct systems for
leakage because this system that looks like it is inside conditioned space in engineering drawings and initial visual
inspection leaks considerably to outside.
The refrigerant charge was determined gravimetrically each system and compared to the correct system charge.
The correct charge was determined by performing superheat tests and tuning the quantity of refrigerant to produce
the required superheat. Table 1 also shows that these systems were close to having the correct system refrigerant
charge, except for sites 2 and 4, where the systems showed the undercharging that was typical of that found in
other studies. Site 2 was the only site where the system charge was an extreme concern because at only 70% of
required charge, this system is undercharged to the point where significant equipment damage could occur.
Table 1. Diagnostic Test Results
Site
Nominal AC
Capacity
[Tons]
Air Handler
Flow
[CFM/Ton]
Supply
Leakage
Fraction
[%]
Return
Leakage
Fraction
[%]
% of
Correct
Refrigerant
Charge
[%]
1
5 375 4% 2% 98%
2
5 379 4% 1% 70%
3
3.5 491 8% 19% 101%
4
2 547 5% 3% 85%
5
2.5 467 6% 4% 95%
6
3 501 4% 5% 91%
Part of this study examines the possibility of resizing systems in order to reduce HVAC system first cost and peak
energy consumption. To provide background information for answering this question, Table 2 contains a
comparison of system capacities. For each site, the ACCA Manual J (1986) sensible load was calculated using the
measured house dimensions and construction details. This was compared to data from the manufacturer
(nameplate capacity), from the ARI (1999) ratings and the measured sensible TAR. The measured TAR were the
quasi-steady-state values obtained after the equipment had been operating for some time so as not to include
transient effects that are not part of the other ratings. Table 2 shows that the nameplate capacities far exceed the
requirements of the ACCA Manual J calculations indicating significant oversizing. The ARI and maximum
sensible ratings diminish the oversizing effect and reinforce the overrating in the nameplate capacities. For Site 6
the maximum sensible capacity is actually very close to the Manual J load estimate and this is probably the correct
size air conditioner for this house. The measured TAR is even closer to the Manual J estimates and at Site 6 the
TAR is less than the Manual J load estimate. The variation in TAR illustrates the impact of the system
11
performance in converting from what is purchased by the homeowner or contractor (nameplate capacity) and is
actually delivered to the conditioned space. At sites 1 and 2 the nameplate capacity is the same but the delivered
TAR is a ton less for Site 2. Site 3, with a 1.5 ton less nameplate capacity system has almost the same TAR as Site
2. Sites 3 and 4 have TAR that are almost the same as the Maximum Sensible Capacity of the equipment, but the
other sites have considerable lower TAR than this maximum. Site 1 is the only site with a significantly higher
TAR than the Manual J estimate. Overall, the results shown in Table 2 illustrate that nameplate capacity is a poor
way to evaluate the capacity of the equipment (compared to Maximum Sensible Capacity) and the system as a
whole (TAR). In addition, the apparently gross oversizing of nameplate capacity compared to Manual J is offset by
lower actual equipment performance and thermal distribution system losses.
Table 2. System Capacity Comparisons
Site
Manual J
Sensible Load
1
[Tons]
Nameplate
Capacity
[Tons]
ARI
Capacity
[Tons]
Maximum
Sensible
Capacity
[Tons]
Tons at the
Register
[Tons]
1
2.25 5 4.4 3.98 3.6
2
2.18 5 4.4 3.61 2.7
3
1.63 3.5 3.3 2.56 2.5
4
1.02 2 1.9 1.56 1.5
5
1.45 2.5 2.4 2.14 1.7
6
2.28 3 2.9 2.34 1.8
1
What the capacity should be assuming no duct leakage, 400 CFM/ton airflow, perfect refrigerant charge, no
additional safety factor.
Continuous Monitoring
The continuous monitoring used computer based data acquisition systems to store data approximately every 10
seconds. The monitored parameters were:
• Temperatures: at each register, in each room, outdoors, attic, garage, return plenum and supply plenum. The
supply plenum temperatures were measured at four points in the plenum to account for spatial variation in
plenum temperatures.
• Weather: wind speed, wind direction, total solar radiation and diffuse solar radiation.
• Humidity: outside, supply air, return air and attic (or garage if system located in garage).
• Energy Consumption: Compressor unit (including fan) and distribution fan power.
The measured system temperatures and relative humidities were used to calculate the energy flow for each register
(and therefore the total for the system) and the energy change of the air stream at the heat exchanger at each time
step.
An overview of all the test data, averaged by a consistent set of test conditions (i.e. amount of duct leakage,
refrigerant charge, type of air conditioning unit) appears in Appendix 3. The performance metrics that were
calculated are listed in Table 3. Each metric has a sensible and a latent component (reported as a sensible and
total) and for all of the metrics except the pulldown time, the value is reported from an average of a minute of data
at 5, 30, and 60 minutes from when the pulldown test began. This range of times for evaluation purposes was used
due to the large transient changes in system performance between the beginning of a cycle and the quasi-steady-
state operation reached later in the pulldown test.
Pulldown time is often very different for each of the three reported locations: thermostat, kitchen, master bedroom.
For example, Site 3 was a two-story house with very poor distribution, particularly upstairs to the master bedroom.
The upper floor of this house had had a significantly increased load due to several skylights in the space as well as
an inadequate return system (there was no return from upstairs). At Site 3 the pulldown time at the thermostat was
12
less than half an hour, but the upstairs took another hour and a half to pulldown to the same temperature. When
the thermostat had reached the pulldown temperature it was 3°C (6°F) hotter upstairs than downstairs.
TAR is determined form the airflows and enthalpy at each register. The indivudual register values are combined
into a single value for the system. TAR is often negative when the air conditioner first comes on because the hot
air inside the duct system is blown into the house. This rapid initial change in temperature (a rapid initial increase
followed by a gradual cooling further into the cycle) made analysis of the initial 5 minutes of the cycle difficult
because of the response time of the sensors. The response time of the temperature sensors is rapid enough that any
time response errors are insignificant. However, the slower response of the relative humidity (RH) sensors
increases the uncertainty in the transient latent (and therefore total) TAR estimates. The time lag of the RH sensor
compared to the temperature sensor means that the measured RHs are higher than they should be (assuming a
reduction in moisture content of the air due to condensation on the coil) which leads to an underprediction of the
latent TAR. Alternatively, if there is no moisture removal by the air conditioner, the RH of the air at the register
should rise as the temperature drops. The longer time response of the RH sensors means that they read artificially
low and this overpredicts the TAR, i.e., it gives the appearance of moisture removal without there being any.
Because this time response issue can drive the results high or low depending on the particular operating conditions,
it is not possible to estimate a generalized effect that would apply to all the measurements (i.e. a bias), instead it
simply adds to the uncertainty of the latent TAR calculations during the start of each cycle. Because of these
uncertainties, most of the comparisons made in later sections of this report are based on the 30 minute value when
the air conditioner is operating much closer to steady state.
Air Conditioner Capacity is estimated by measuring enthalpy change and air flow through over the evaporator
coils. It is a useful way of examining the effect of low evaporator air flow or incorrect refrigerant charge without
confounding the impacts of a leaky duct system. It suffers the same sensor response limitations during the initial
transient at the start of a cycle as TAR.
Air conditioner COP is a measure of efficiency of an air conditioner, and it is typically around 2-3 for a
residential unit. It is estimated from the measured capacity and electricity consumption. Unlike the COPs
presented by the manufacturer, the COPs reported here include the energy (and heat generation) of the air handler
fan.
System COP is the most inclusive performance measure, and it is a simple ratio of the cooling energy delivered to
the conditioned space (TAR) divided by the power consumption of the air conditioner and fan. System COP is
affected by changes in the air conditioner capacity as well as any losses/gains in the distribution system.
Delivery efficiency is a simple ratio of the energy of the air that comes out of the registers divided by the energy of
the air in the supply plenum. It is a measure of losses that occur in the duct system. The five minute delivery
efficiency is almost always higher than the 30 and 60 minute delivery efficiency. This is because after five
minutes, the air in the ducts has not cooled down very much and conduction losses (which are proportional to the
temperature difference between the ducts and the air around the ducts) are low. By the time the system has reached
steady state, the conduction losses have increased and the delivery efficiency drops slightly.
Although these metrics are all intuitive and useful ways to understand cooling system performance, they have
limited utility for comparing houses or even for comparing different conditions (amounts of duct leakage,
refrigerant charge, etc.) at the same house. All are very strongly influenced by many external factors (in addition
to those already discussed such as duct leakage, evaporator airflow, refrigerant charge level, etc.), most notably
indoor temperature and humidity, outdoor temperature and conditions, and attic conditions. The tables in
Appendix 3 list the average of these parameters at each of the conditions. Because only a few days of
measurements were performed at each site condition, there are often large variations associated with each of the
mean values for the performance parameters. Also, in some cases only a single day of data was taken, resulting in
a single data point being used as the “average”. Because of these limitations, the measured results often do not
reveal the expected results and make the data difficult to interpret.
13
For these reasons, this report concentrates on comparing individual sites where conditions are similar. Because
weather conditions varied widely during the test period, there were few days that were identical to each other.
Limiting comparisons to similar weather conditions means that the comparisons discussed below are often based
on a single pulldown test at each condition and should be interpreted with some caution.
Table 3. Performance Metrics
Pulldown Time Time that it takes for a zone to reach 24°C. Three
pulldown times are reported: for the thermostat (how
the house would actually respond), kitchen, and master
bedroom. A wide disparity between these times
indicates an inadequate distribution system.
Tons At the Register (TAR) Amount of energy delivered to the space
Air Conditioner Capacity Capacity of the air conditioner calculated from
temperatures and relative humidities measured in the
supply and return plenums
Air Conditioner Coefficient of Performance (COP) Air conditioner capacity divided by power consumed by
air conditioner, including fan energy
System COP Tons at the register divided by power consumed by air
conditioner, including fan energy
Delivery Efficiency Tons at the register divided by air conditioner capacity
Site 1: At this site, there were two similar days of weather that allow us to examine the impact of sealing the ducts.
Both the sensible and the total delivery efficiency improved 8 percentage points from 81% to 89%, about a 10%
increase. Similarly the system COP improved, but only by 8% (total) and 3%(sensible). The latent improvement
may be an overstatement because of uncertainties in the RH measurements, even though the air conditioner had
been running for 30 minutes in each case and thus the relative humidity should not have been changing very
rapidly. The relatively small improvement in system performance is an indication that inefficiencies of this large
(5 ton) air conditioner tended to dominate the system losses, rather than the distribution system losses.
Site 2: Although there were four conditions at this site (as found duct leakage and low refrigerant charge, leaks
added and low charge, leaks added and correct charge, ducts sealed and correct charge), only two days had similar
enough weather conditions to compare them. Unfortunately, two things changed between the days that had
comparable weather conditions: a very low level of refrigerant charge was corrected, which would tend to improve
system performance and 212 cfm (12% of fan flow) of leakage was added to the duct system, which would tend to
diminish system performance. The added leakage was split almost evenly, with 97 cfm (5.5% of fan flow) added to
supply side and 115 cfm (6.5% of fan flow) added to the return side. Paradoxically, correcting the very low
refrigerant charge level appeared to very slightly diminish the system performance. However, a close examination
of the weather data indicated that, although the outdoor temperatures were very similar, the enthalpy of the
outdoor air was significantly greater when the charge was lower which made the unit appear to perform better.
Comparisons of delivery efficiency showed a reduction of about 10% due to the added leaks.
Site 3. Correcting a slightly low charge level resulted in an increase in air conditioner capacity of about 6% and an
improvement in COP of about 8%. Sealing 200 cfm (14% of fan flow) of duct leakage improved the delivery
efficiency by 11%. At this site, the ducts leaked into the garage (mostly outside) and into an interstitial space
between the first and second floor that was thermally inside, but outside of the pressure boundary. These effects
overall combined to improve the system COP by 17%.
Site 4. Many of the performance metrics are harder to interpret at this site because the new Energy Star units’
energy consumption varied with time (probably due to a variable speed compressor). The air handler capacity of
the Energy Star unit improved by 5% and the sensible COP improved by 25%. This result suggests that higher
14
efficiency units may improve sensible cooling at the expense of latent cooling; however the uncertainty of RH
measurements means that this result requires additional verification. The delivery efficiency dropped by about 4%
after changing the equipment. This drop might have been due to small variations in outdoor and attic
temperatures, however, it may have also been caused by the lower supply plenum temperatures (and hence higher
temperature differences for conduction losses) of the new unit. Adding 107 cfm (9% of fan flow) of leakage at Site
4 reduced the delivery efficiency by 6%. The extra leakage was greater on the supply side 75 cfm (6% of fan flow)
vs. 32 cfm (3% of fan flow) for the return side.
Site 5. Sealing 92 cfm of leakage (7% of fan flow) caused insignificant improvements of the delivery efficiency
(between 1 and 2 percent). The overall COP was improved by 3%, however (as with other sites) the uncertainty in
the RH and plenum temperature measurements means that this change is about the same as the uncertainty in the
measurements and cannot be interpreted as a significant change.
Site 6. The combination of highly variable weather and some data collection problems at site six means that there
were no results suitable for comparison. A more detailed analysis of the measured data will be required if any
conclusions are to be drawn from the Site 6 measurements.
Computer Simulations
The details of an earlier version of the simulation model (called REGCAP – short for REGister CAPacity) have
been given previously by Walker et al. (1998 and 1998b). A flowchart for the simulation program is shown in
Appendix 4. The changes made to improve the simulation model for this study are discussed below. This
improved model was compared to measured results for validation purposes. In the previous work, the simulations
were used to show how pulldown time changed with duct system performance, different weather conditions (a
typical design day and a peak day) and with system capacity. The simulations were able to show several key
results:
• A good duct system allowed the capacity of the equipment to be reduced by about one quarter: from four tons
to three tons nameplate capacity.
• If system nameplate capacity is unchanged, either improving duct systems (to have little leakage) and correctly
installing the equipment, or moving the ducts inside results in significant pulldown performance
improvements. In these cases pulldown times were reduced by more than an hour and initial tons at the
register were approximately doubled.
• The model results also showed the wide range of pulldown times for different duct systems.
Simulation Model Improvements
In our continuing efforts to develop models of HVAC system performance, the model used in the last Phase has
been upgraded:
• It includes additional air flow paths through duct leaks when the system is not operating.
• It has a simple moisture balance for use in latent load and equipment capacity calculations.
• Solar load and thermal mass calculations for building load have been improved.
• A simple thermostat model and the ability to make calculations at small timesteps allowed the model to be
used for examining cyclic effects.
• Improved equipment modeling accounts for changes in capacity and energy consumption with outdoor weather
conditions, fan flow, system charge and indoor air conditions.
The equipment model used to predict the capacity of the air conditioners for the REGCAP simulation is an
empirical model developed by John Proctor (Proctor (1999)). This model is the only available model that accounts
for refrigerant charge level and is sufficiently general for use in this project. Proctor has used this model in much
of his research (see Proctor (1997), (1998) and (1998b)) and continues to update it as he collects new data.
Currently, the portion of the model that accounts for deviation from recommended refrigerant charge is taken
directly from Rodriguez et al. (1995) and the rest of the model is based on Proctor Engineering Group fieldwork in
about one hundred houses.
15
The model requires the following inputs: nominal (nameplate), capacity, ARI capacity, air flow, outside
temperature, indoor (return plenum) enthalpy, refrigerant charge level, and expansion valve type (capillary
tube/orifice or TXV (thermostatic expansion valve)). The model predicts sensible capacity and, with the
assumption of a sensible heat ratio for the unit, latent and/or total capacity can also be predicted. The comparison
of the measured capacities at the six houses (8 air conditioners) in this study indicate that the model overpredicts
capacity by about 10%. There is no obvious reason for this consistent deviation from Proctor’s data, but a possible
reason is that most of the Proctor’s verification of the model occurred in very dry climates, rather than the more
humid weather that we encountered during the field testing.
Figure 1. Simulations of Pulldowns from 3:00 p.m. on a Sacramento Design Day.
Extension of previous simulations
The improved model was used to reexamine the pulldown simulations performed in the previous part of this study.
In these simulations, eight different thermal distribution systems are used in the same house for the same weather
conditions. Table 4 lists the simulation cases that were examined here. The BASE case is typical of new
construction in California. The POOR system represents what is often found at the worst end of the spectrum in
existing homes. The BEST system is what could reasonably be installed in new California houses using existing
technologies and careful duct and equipment installation to manufacturers’ specifications. The BEST RESIZED
Time of Day (hours)
14 15 16 17 18 19 20 21
Indoor Temperature, C
24
26
28
30
32
POOR
BASE
BEST RESIZED
INTERIOR RESIZED
BEST
IDEAL
INTERIOR
IDEAL OVERSIZED
16
system looks at the possibility of reducing the equipment capacity using the best duct system. The INTERIOR
system examines the gains to be had if duct systems are moved out of the attic and into conditioned space. The
INTERIOR RESIZED system examines the system performance if reduced capacity equipment is used together
with interior ducts. Lastly, the IDEAL system is an interior duct system that has been installed as well as possible.
The IDEAL OVERSIZED simulations were included to examine the difference in pulldown if the IDEAL system
were sized using current sizing methods (i.e., still 4 tons).
The difference between the simulation cases listed in Table 4 and those done previously is that the flow used for the
“correct flow” cases is 400 cfm instead of 425 cfm. This minor change was done because the equipment used in
the equipment model was rated at this flowrate. The large range for pulldown results are illustrated in Figure 1,
with each simulation starting at the same time. The better systems were able to pulldown the house in a reasonably
short time (under three hours) but the poor systems took over six hours. The longer pulldown times mean that the
house would not be comfortable for occupants returning in the afternoon. For example, the house with the POOR
system is still not pulled down at 8:00 p.m. For the occupants this would be unacceptable and a better question to
ask is: At what time would an occupant have to turn on the air conditioning in order to have the house comfortable
upon returning home in the afternoon at 5:00 p.m?
Table 4. List of REGCAP Simulation Cases
System
Charge
[%]
Air Handler
Flow
[CFM/Ton]
Duct Leakage
Fraction
[%]
Duct and
Equipment
Location
Rated
Capacity
[Tons]
BASE
85 345 11 Attic 4
POOR
70 345 30 Attic 4
BEST
100 400 3 Attic 4
BEST RESIZED
100 400 3 Attic 3
INTERIOR
85 345 0 House 4
INTERIOR RESIZED
85 345 0 House 3
IDEAL
100 400 0 House 3
IDEAL OVERSIZED
100 400 0 House 4
Table 5. Start Time to Pulldown by 5:00 p.m.
Rated Capacity [Tons] Start Time
BASE
4 10:30 a.m.
POOR
4 Not possible
1
BEST
4 2:30 p.m.
BEST RESIZED
3 11:45 a.m.
INTERIOR
4 1:45 p.m.
INTERIOR RESIZED
3 9:45 a.m.
2
IDEAL
3 12:30 p.m.
IDEAL OVERSIZED
4 2:30 p.m.
1- 37°C at 5:00 p.m., pulldown to 24°C at 9:00 p.m. (drawing in cool outdoor air through return leaks)
2- Although this system is basically on all day, this result is misleading because the indoor temperature never
gets above 25°C and a more lenient pulldown criteria drastically changes this result. For example, increasing
the setpoint temperature by 1°C (to 25°C) changes the system ontime to 11:00 a.m. and makes it better, not
worse, than the base case.
17
The results in Table 5 show that the time that the systems have to run covers a very wide range from two and a half
hours for the BEST and IDEAL OVERSIZED systems to all day for the POOR system. In effect, looking at
pulldown this way has further exaggerated the differences between the systems. This is mostly because the systems
are now operating more during the heat of day rather than the cooler evening and night time. As with the results
reported previously, this table shows that resized systems with good ducts can be as good or better than an existing
BASE system and that there are large gains to be had by improving the duct systems. Assuming that the energy
consumption scales with system capacity and ontime, and normalizing by the BASE case energy consumption it is
possible to calculate the relative energy consumption for each simulation, as shown in Table 6.
Table 6. Relative Energy Consumed in Order to Pulldown by 5:00 p.m.
Percent of BASE case
BASE
100
POOR
260
BEST
40
BEST RESIZED
60
INTERIOR
50
INTERIOR RESIZED
85
IDEAL
50
IDEAL OVERSIZED
40
Because the POOR system is on all day, the energy consumption is far greater than for the other systems. All the
other systems consume less energy than the base case while providing equal or superior comfort in terms of
pulldown time. In particular, the resized systems all consumed less energy than the BASE case for these
simulations.
Table 7. Model Delivered Capacity (TAR) Comparison (system on for 1.75 hours)
Nameplate
Capacity
[Tons]
Tons at the
Register
[Tons]
Tons at the Register
Nameplate Capacity
[%]
Ratio to
Base Case
[%]
BASE
4 1.66 42% 100%
POOR
4 1.51 38% 91%
BEST
4 2.21 55% 133%
BEST RESIZED
3 1.66 55% 133%
INTERIOR
4 1.84 46% 110%
INTERIOR RESIZED
3 1.36 45% 109%
IDEAL
3 1.68 56% 135%
IDEAL OVERSIZED
4 2.28 57% 137%
Table 7 compares the results of the calculated TAR between the simulations. Note that for these calculations the
systems have been running for almost two hours and are at quasi-steady-state and do not show the transient
capacity reductions at the start of the pulldown. This was done so that the results are as close as possible to the
manufacturers rating conditions, and we are not unfairly comparing the nameplate capacity to the transient system
performance. In other words, we are being as generous as possible in our comparisons by reporting close to the
highest system capacities. All but the POOR ducts are better than the BASE case in terms of delivered TAR and
also TAR as a fraction of the nominal (nameplate) capacity of the equipment. All of the resized systems have TAR
values closer to their nominal capacity than the BASE or POOR cases. However in all cases (even the ideal
situation with correct system charge and airflow and minimal duct losses) the equipment capacities are much less
than the nominal nameplate rating that a home owner has paid for.
18
Comparison of Field Measurements and Computer Simulations
The model was evaluated by comparing predicted temperatures to measured temperatures. Given the same
temperatures, other variables used to determine energy flows (e.g., register flowrates) and comfort parameters (e.g.,
pulldown times) are the same for both modeled and measured data. An essential part of simulation design and use
was verifying that the simulation makes accurate predictions. In this case, we were interested in predicting two
parameters: tons at the register (delivered capacity) and pulldown time (time to cool down the house). For this
purpose, we examined the temperatures of the four air nodes described above (attic, house, supply duct, return
duct).
Over 100 days of measured data at 5 sites were used to evaluate the simulation model (4 in California and 1 in
Texas). Overall there was very good agreement between the modeled and measured house and attic temperatures
and good agreement between the duct air temperatures when the air handler fan was on, but not very good
agreement when the air handler was off. In order to illustrate these and other strengths and weaknesses of
REGCAP, the modeled/measured comparison is shown for two sites and each of the four modeled temperatures
will be discussed individually. There was no attempt to show data that was either particularly favoring or
condemning of REGCAP: the following illustrations are included to demonstrate both the strengths and the
weaknesses of the model. Appendix 5 contains a preliminary analysis of some of the problems encountered when
comparing modeled and measured results due to the sensitivity of the model to measured weather data.
The results for two homes are described in this section (sites 4 and 5). Both homes have floor areas of
approximately 140 m
2
(1500 ft
2
) and are located in a subdivision in Sacramento, CA. The ducts, air handler,
furnace and indoor cooling coil were located in the attic in both homes. Site 4 had supply duct leakage fraction
that is 5% of air handler flow, return leakage is 3%. Site 5 had a very tight duct system (both leakage fractions are
less then 3%). Site 4 had a 2 ton system with a fixed orifice expansion valve and was found at 85% of
manufacturer’s refrigerant charge. Site 5 had a nominally 2.5 ton system with a thermal expansion valve (TXV)
and was fully charged. For brevity, graphs comparing modeled and measured data are shown for sites 4 and 5
only, and the generalized discussion applies to all the comparisons between measured and modeled data.
19
Site 4 on August 11, 1998
Time [Hours]
0 5 10 15 20
Attic Temperature [°C]
10
20
30
40
50
60
70
Measured
Predicted
|∆|=2.4°C
Figure 2: Modeled and Measured Attic Temperatures at Site 4 on August 11, 1998
Attic Temperature
These two houses show excellent agreement between the modeled and the measured attic temperature over the
whole day. The agreement at site 4 is near perfect for the first half of the day and then the predicted temperature
drops slightly below the measured temperature (Figure 2).
The average absolute difference in temperatures is 2.4°C (4.3°F). There are several hypotheses that explain this
small discrepancy: the most plausible is a problem with the measured solar radiation input data (the dip in the data
when the sun comes up is an indication of this) or, perhaps, the ducts are too strongly coupled with the house so
that when the air conditioner comes on the duct leakage cools the attic more in the modeled case than in the
measured case. Another possible problem is the fact that the radiative transfer involving the attic endwalls and the
combined mass of wood in the attic was neglected. The modeled data at site 5 overpredicts the temperature for the
first half of the day and then underpredicts it for the last half, but the overall average absolute temperature
difference is 1.9°C (3.4°F), smaller than the difference at site 4.
20
Site 4 on August 11, 1998
Time [Hours]
0 5 10 15 20
House Temperature [°C]
22
24
26
28
30
32
34
Measured
Predicted
|∆|=0.4°C
Figure 3: Modeled and Measured House Air Temperatures at Site 4 on
August 11, 1998
House Temperature
The comparison of house temperatures at site 4 is shown in Figure 3. The average absolute difference between the
modeled and the measured values is 0.4°C (0.7°F). The modeled house air responds very quickly to changes in
climatic conditions. This may be due to insufficient coupling between the house air and the house mass. The
agreement at site 5 is not as good, with an average absolute temperature difference of 0.6°C (1°F): examination of
the weather data collected on the day of test indicates very strong winds from about 11am until 6pm. This is a
failure of the model to deal with extreme conditions and is probably the cause of the wide temperature swings
evident in the measured data. Both modeled houses have a single spike in the temperature when the air
conditioner come on. This is an artifact of the ducts pushing hot air into the house that doesn’t seem to be evident
in the measured data (which was collected every 10 seconds, a finer resolution than the minute long timestep of the
simulation). Despite these discrepancies, both sets of simulated data seem to reflect the overall shape of the
temperature curve in each house. An improved house load model, such as Suncode,™ will probably increase the
accuracy.
One problem with the house model is that the thermal mass of the house seems to be very weakly coupled to the
house air. There are two most likely causes of this problem: the first is that the convection heat transfer coefficient
for the house mass is biased towards natural, rather than forced, convection. This is an issue when there are strong
winds (which lead to larger pressure differences and air velocities in the house), and when the air handler is on.
This is a good example of where reducing the input value (i.e., no average air velocity in the house) leads to a less
accurate predicted result. The second is that the surface area active in heat exchange between the thermal mass of
the house and house air is too small in the model. Future work will further investigate this thermal mass issue.
21
Site 4 on August 11, 1998
Time [Hours]
0 5 10 15 20
Return Duct Air Temperature [°C]
10
20
30
40
50
60
70
Measured
Predicted
|∆|
fan off
=5.1°C
|∆|
fan on
=0.3°C
Figure 4: Modeled and Measured Return Duct Air Temperatures at Site 4 on August 11, 1998
Return Duct Air Temperature
The return duct agreement is quite good at site 4 (Figure 4) when the air conditioner is on (absolute difference of
only 0.3°C). When it is off, the predicted duct temperature is much hotter than the measured temperature (absolute
difference of 5.1°C). A very similar pattern occurs at site 5, with the same average absolute difference between the
modeled and measured. The strong winds in the middle of day again affect the simulation quite strongly. Overall,
REGCAP does an adequate job of prediction the temperature plots at both sites when the air handler fan is on.
22
Time [Hours]
0 5 10 15 20
Supply Air Temperature [°C]
10
20
30
40
50
60
70
Measured
Predicted
Site 4 on August 11, 1998
|∆|
fan off
=8.0°C
|∆|
fan on
=0.2°C
Figure 5: Modeled and Measured Supply Duct Air Temperatures at Site 4 on August 11, 1998
Supply Duct Air Temperature
The supply duct air temperature has a very similar pattern at both sites (Site 4 is shown in Figure 5). Like the
return duct, the temperature shows good agreement when the air handler fan is on, but poor agreement when the
air handler fan is off. When the air handler is off, the modeled supply duct temperature is very strongly influenced
by the attic temperature and radiation exchange with the interior attic surfaces. The fact that the agreement is not
very good for the duct air temperatures when the air handler is off may seem surprising because the model
explicitly calculates the mass flow through these ducts when the air handler is off. However, there is a subtle
distinction: REGCAP calculates the mass flow of air passing from the attic to the house (or the house to the attic)
through the ducts, but does not calculate thermosiphon flows. Thermosiphon flows occur as air moves in one
register and out another when the air handler is off. These flows are very difficult to calculate because to do so
requires extensive information about the geometry of the duct system as well as being able to model flows between
and within different rooms in the house.
The lack of air-handler off agreement for the duct temperatures is not particularly significant for the objectives of
this study: predicting the pulldown time and the tons of cooling at the register. The only temperatures that are
directly needed for these calculations are the house air temperature and the supply duct air temperature when the
air conditioning fan is on. For this reason, REGCAP is well suited to calculating the performance parameters that
are the focus of this project.
Field Measurement and Computer Simulation Outcomes
• Improved ducts and system installation can allow the use of a smaller nameplate capacity air conditioner
(almost one ton less in the simulations presented here, and at least one ton in more demanding situations)
without any comfort penalty in terms of pulldown, and with large energy savings (roughly halving
energy consumption).
• If system nameplate capacity is unchanged, either improving duct systems and correctly installing the
equipment, or moving the ducts inside results in significant pulldown performance improvements.
23
• Simulations confirm field test results regarding delivered capacity and equipment and distribution
system performance.
• Comparisons of computer simulation results to measured field data show that the simulations predict the
equipment attic and house performance with sufficient accuracy to be a useful prediction tool.
• Field measurements of delivered cooling capacity are considerably less (20% to 50%) than nameplate &
ARI ratings.
• Nameplate and ARI capacity ratings of equipment installed in houses exceed those indicated by ACCA
Manual J load calculations.
• Thermal distribution system losses and poor equipment installation combine to reduce delivered
capacity. Measured delivered capacities are close to those indicated as necessary by ACCA Manual J
load calculations.
• Improving ducts by reducing leakage can lead to significant energy efficiency gains in addition to
cooling the house faster.
• Efficient systems can still have problems satisfying occupant comfort even though the total delivered
capacity for the system is correct due to room-to-room variations in delivered capacity for each room.
The room-to-room variations result in large temperature variations throughout the house.
• Using higher SEER units indicated significant peak energy savings of about 25% with no apparent
drawbacks in the houses measured.
4. Support for Title 24 and HERS
The objective of this task was to:
• Provide technical support to the California Energy Commission (CEC) for updating the “Low-Rise
Residential Alternative Calculation Method Approval Manual for 1998 Energy Efficiency Standards for
Low-Rise Residential Buildings” (CEC (1999)) and Procedures for HVAC System Design and
Installation (for HERS).
One of the most significant technology transfer activities in this project has been the inclusion of credits for energy
efficient ducts in the Low-Rise Residential Alternative Calculation Method Approval Manual for 1998 Energy
Efficiency Standards for Low-Rise Residential Buildings (ACM). The changes and additions were made to the
Alternative Calculations Manual based on our technical and editorial input. They allow an energy credit to be
claimed by having improved ducts that are field tested for leakage and do not use rubber adhesive cloth tape for
duct seals.
We have also provided technical support for research sponsored by the California Energy Commission (CEC) on
home diagnostics (for HERS). We have worked with Davis Energy Group (DEG) on the development of
residential commissioning test protocols for these home diagnostics. This has included measurement of register
flows, fan flows and duct leakage. For the register flow measurements, a combined study with DEG, LBNL, CEC
and The Energy Conservatory that used flowhoods to measure register flows was undertaken. Eight different
flowhoods were evaluated in a new house in Sacramento. The results of this testing (given in detail in Appendix 6)
showed that standard flowhoods can be poor at measuring the register flows. This is due to a combination of:
• Low flows result in a small pressure signal from the flowhoods that leads to low precision.
• Poor calibration. Some of the flowhoods had large bias errors for all measurements, indicating a calibration
problem.
• Sensitivity to flow asymmetry. The flowhoods are calibrated and designed to be used on registers with a
uniform face velocities, but the registers in residential buildings are rarely operated in this manner and have
strong flow variations across the face of the register.
• Flow restriction lowering the flow during the measurement. The restriction of flow due to inserting the
flowmeters can be significant.
All of these problems were reduced by using fan assisted flowhoods. The fan assist is used to balance the pressure
24
in the hood with the static pressure in the room. This was originally done to remove the effect of restricting the
flow, but the side benefits are of equal, or greater, importance because the fan assist tends to remove the flow
asymmetries and give better results with any remaining asymmetry. The calibrations for the flowmeters are well
known and easily checked, and the flowmeter can be adjusted to be sensitive to low flow rates, thus improving the
precision of the measurements. Unfortunately, the fan assisted flowhood is extra equipment to carry around a house
and is equipment that home testers would have to become familiar with. In addition, there is the added expense of
additional equipment purchase.
In addition to individual register flows, the CEC is also interested in requiring fan flow to be measured. The
proposed method in ASHRAE 152P (and as used in our field testing) that requires blocking of the return and
matching operating pressures is considered too time consuming and difficult. Some alternatives – such as
measuring return grille flows with a flowhood and adding an estimate of the return leakage – are insufficiently
accurate for use in a rating tool. Future possibilities for measuring fan flow may include the use of device under
development at ECOTOPE that attaches in place of the filters in the system and requires less time and effort. We
hope to evaluate the ECOTOPE system in the near future.
We are working to improve the ASHRAE 152P method by developing a new fan flow measurement device that
utilizes a large powerful (but still portable) fan with a built in flowmeter. This device replaces the small fan
flowmeters used in previous studies. It should allow us to replicate the fan flow in most residential systems without
having to extrapolate from the measurement point to the system operating point (as required with the small fan
flowmeters).
We worked with the staff of the CEC to evaluate an HVAC system performance tool developed by Federal Air
Conditioning Technologies (FACT). This tool evaluates both the duct air flow and thermal performance as well as
the refrigerant systems. We did comparison tests on a house in Sacramento and in LBNL’s Building 51 laboratory
with LBNL measurement equipment and the FACT equipment. We found some important measurement
differences between the two types of equipment. However, without additional testing, it was not possible to
pinpoint the exact reason for the discrepancies.
A collaborative of CIEE, CEC, the California Building Industry Association (CBIA) and the Natural Resources
Defense Council (NRDC) developed procedures for improved design, fabrication, installation and testing of HVAC
systems. We supported the updating of these procedures to ensure compatibility with the changes incorporated into
the ACM for duct energy efficiency credits by reviewing a draft of “Procedures for HVAC System Design and
Installation” (Hammon 1999). The draft procedures are in Appendix X. We are collaborating with CONSOL (the
contractor who is upgrading this document) through this review process.
The key outcomes of this task were:
• Duct efficiency calculations are included in the Low-Rise Residential Alternative Calculation Method
Approval Manual for 1998 Energy Efficiency Standards for Low-Rise Residential Buildings” (CEC
(1999)).
• Procedures for HVAC System Design and Installation (for Home Energy Raters) have been updated.
• Field testing has shown that standard flowhoods can be poor for measuring residential register flows.
5. Technology Transfer
The objective of this task was to:
• Support ASHRAE, ASTM and EPA duct leakage research and interface with realted projects funded by
other agencies.
25
ASHRAE: Rating of Distribution Systems - ASHRAE 152P
ASHRAE published and distributed a draft version of ASHRAE 152P for public review during May and June 1999.
It is expected that the final draft of this standard will be ready by January 2000. We have also developed a web-
based tool for performing 152P calculations. This tool can be accessed at http://ducts.lbl.gov. This web tool
includes many defaults as guides for the uninitiated user that are taken from the appendices of 152P. These
defaults are intended to make this web-tool easier to use.
ASTM: Rating of duct sealants and revising duct leakage measurement
methods
We have attended ASTM meetings and corresponded with ASTM to discuss the implementation of an ASTM
standard for longevity testing of duct sealants (ASTM (1999b)). A draft of the standard was prepared and voted on
by ASTM E6.41 subcommittee members. Several comments were made on this draft which was then revised and
will be reballotted later this year.
The current duct leakage test measurement in ASTM E1554 is obsolete. This standard has been rewritten based on
the results of duct leakage test evaluations performed for the last three phases of the current research sponsored by
CIEE/CEC, together with input from other ASTM members and the members of ASHRAE SSPC 152P. The new
draft of the standard has two leakage test methods: the DeltaQ test and duct pressurization. The standard also
includes the benefits and drawbacks of the two methods, so that the user can select the most appropriate test
method for the test they are performing. For example, the DeltaQ test is better for measuring leaky duct systems
for HERS testing, but the pressurization tests are more robust for low leakage compliance testing.
Other Thermal Distribution System Efficiency Support Activities
Several other tasks were performed under the scope of this study that relate directly to thermal distribution systems.
The following is a summary of these activities:
Health and Safety Assessment of Aerosol Sealant (EPA)
As with any new industrial material, concern exists over the potential health hazards related to human exposure.
Potential health and safety issues regarding the duct seal material were evaluated and discussed in Buchanan and
Sherman (1999). This report examines the characteristics of the sealants’ individual components as determined
from current literature. There are three primary means by which exposure could occur: ingestion, eye/dermal
contact, and inhalation. Each of these possibilities is examined. Exposure and safety risks were assessed with
regard to the currently known constituents that are believed to pose potential hazards: VAP, VAM, 2EH, and
acetaldehyde.
Field Testing of Energy Star Equipment (EPA)
This field testing was performed in conjunction with the field testing for Phase VI. In one of the Sacramento
houses and the Texas house the air conditioning equipment was replaced by higher efficiency Energy Star
equipment, but there was no resizing of the equipment. In both cases we replaced a standard SEER 10 unit for one
rated at SEER 13. These additional tests funded, by EPA, added an additional three “systems” (the Sacramento
house with SEER 13 plus the Texas house in two systems configurations) to the database for the Phase VI work.
Developing Energy Star Ratings for Duct Systems (EPA)
In addition to previous work on incorporating duct system efficiency in Energy Star Ratings for houses, we have
also worked with EPA on developing rating methods, baseline studies and possible duct efficiency improvements
for an Energy Star rating system. A preliminary report by Walker (1999) summarizes a sensitivity study
performed for EPA that examines variability of distribution system efficiency with geographic location (climate)
and duct system parameters (e.g., leakage). Additional ongoing work will determine baseline duct efficiencies
throughout the country and estimate how much of the energy losses could be saved. This program is currently
aimed at existing houses, but we plan to adapt it in the future for application to new construction.
26
Public Dissemination of Research Results
During this year we have been developing the Thermal Energy Distribution Web page - http://ducts.lbl.gov. This is
intended to be a central reference point for disseminating information about thermal distribution systems in
buildings, and the papers resulting from the work done for the current project will be “published” on this web site.
We have also assisted CEC by preparing information for their thermal distribution system web page. We have
further assisted CEC by participating in their triennial review process.
The results from work done for this phase of the Thermal Distribution Efficiency research program have been
presented (and published) mostly at ASHRAE meetings and at the ACEEE 1998 Summer Study. The following
presentations have been given in the last 12 months, some of which were based on work performed for the previous
phase of this work. Section 7 lists recent publications associated with this research program.
Walker, I.S., (1999), "Distribution System Leakage Impacts on Apartment Building Ventilation Rates", ASHRAE
Trans. Vol. No. (presented at ASHRAE TC 4.10 Symposium, January 1999), LBNL 42127.
Walker, I.S. and Sherman, M.H, (1999), “Assessing the Longevity of Residential Duct Sealants”, RILEM 3
rd
International Symposium: Durability of Building and Construction Sealants, February 2000. LBNL 43381.
Walker, I., Sherman, M., Siegel, J., and Modera, M., (1999), “Comfort Impacts of Duct Improvement and Energy-
Star Equipment”, EPA Contract Report, LBNL 43723.
Walker, I, (1999), CIEE report on Benefits Estimates for CIEE Residential Thermal Distribution projects
Walker, I.S., (1999), “Sensitivity of Forced Air Distribution System Efficiency to Climate, Duct Location and Duct
Leakage”, EPA Report, LBNL 43371.
Walker, I.S., and Sherman, M.H., (1999), “Can Duct Tape Take the Heat”, LBNL 41434.
The key outcomes of this task were:
• ASHRAE standard 152P for rating distribution systems has been prepared for and submitted to the ASHRAE
public review process.
• The ASTM standard for duct leakage testing has begun the review process and a new standard for longevity of
duct sealants has been proposed to ASTM.
• Support was provided for several thermal distribution system efficiency tasks sponsored by EPA.
• Several reports and papers have been published to allow public dissemination of research results.
6. References
ACCA. 1986. Manual J - Load Calculation for Residential Winter and Summer Air Conditioning - Seventh
Edition. Air Conditioning Contractors of America (ACCA), Washington, D.C.
ARI. 1999. Electronic Unitary Directory, ARI UD99s V1.2. Air Conditioning and Refrigeration Institute,
Arlington, VA.
ASHRAE. 1999. Standard 152P - Method of Test for Determining the Design and Seasonal Efficiencies of
Residential Thermal Distribution Systems. ASHRAE, Atlanta, GA. (Public review draft).
ASTM. 1999. Draft revision to Standard E-1554 – Determining External Air Leakage of Air Distribution Systems
by Fan Pressurization. American Society for Testing and Materials, West Conshohocken, PA.
ASTM. 1999b. Task group draft – Standard Test Method for Longevity Testing of Duct Sealant Methods.
American Society for Testing and Materials, West Conshohocken, PA.
27
Blasnik, M., Downey, T., Proctor, J. and Peterson, G .1996. Assessment of HVAC installations in New Homes in
APS Service Territory. Proctor Engineering Group Report for Arizona Public Service Company.
Carlson, J.D., Christian, J.E., and Smith, T.L. 1992. In Situ Thermal Performance of APP-Modified Bitumen Roof
Membranes Coated with reflective Coatings. Proc. ASHRAE/DOE/BTECC/CIBSE Thermal Performance of the
Exterior Envelopes of Buildings V, Clearwater Beach. Florida. December 1992. pp. 420-428.
CEC. 1998. Low-Rise Residential Alternative Calculation Method Approval Manual for 1998 Energy Efficiency
Standards for Low-Rise Residential Buildings, California Energy Commission, Sacramento, California.
Hammon, R., . 1999. Procedures for HVAC System Design and Installation. CONSOL
ICBO (International Conference of Building Officials). 1994. Uniform Mechanical Code. Section 306. ICBO,
Whittier, CA.
CGA (Canadian Gas Association). 1995. Natural Gas Installation Code. Canadian Naitonal Standard CAN/CGA-
B149.1-M95. Section 6.8.6. CGA, Etobicoke, ON., Canada.
Parker, D.S., Sherwin, J.R., and Gu, L. 1997. Monitored Peak Attic Air Temperatures in Florida Residences:
Analysis in Support of ASHRAE Standard 152P. FSEC-CR-944-97, Florida Solar Energy Center, Cocoa, Fl.
Proctor, J. 1999. Air conditioning Equipment Model. Personal Communication, March 1999.
Proctor, J. 1998. Monitored In-Situ Performance of residential Air-Conditioning Systems. ASHRAE Trans. Vol.
104. Part 1. ASHRAE, Atlanta, GA.
Proctor, J. 1998b. Performance of a Reduced Peak kW Air Conditioner at High Temperatures and Typical Field
Conditions. Proc. ACEEE Summer Study on Energy Efficiency in Buildings 1998. Vol. 1 pp.265-274. American
Council for an Energy Efficient Economy, Washington, D.C.
Proctor, J. 1997. Field Measurements of new residential air conditioners in Phoenix, Arizona. ASHRAE Trans. Vol.
103 Part 1. ASHRAE, Atlanta, GA.
Rodriguez, A.G., O’Neal, D.L., Bain, J.A., and Davis, M.A. 1995. The Effect of Refrigerant Charge, Duct
Leakage, and Evaporator Air Flow on the High Temperature Performance of Air Conditioners and Heat Pumps.
Energy Systems Laboratory report for EPRI, Texas A&M University.
Walker, I, Sherman, M., Modera, M., Siegel, J.Dickerhoff, D. 1997. Leakage Diagnostics, Sealant Longevity,
Sizing and Technology Transfer in Residential Thermal Distribution Systems. CIEE Residential Thermal
Distribution Systems Phase V Final Report, October 1997, LBNL Report 41118.
Walker, I, Sherman, M., Siegel, J., Wang, D., Buchanan, C., and Modera, M. 1998. Leakage Diagnostics, Sealant
Longevity, Sizing and Technology Transfer in Residential Thermal Distribution Systems, Part II. CIEE
Residential Thermal Distribution Systems Phase VI Final Report, December 1998, LBNL 42691.
Walker, I.S., Brown, K., Siegel, J. and Sherman, M.H. 1998b. Saving Tons at the Register. Proc. ACEEE Summer
Study, Vol. 1, pp. 367-383. LBNL 41957.
Walker, I.S. 1999. Sensitivity of Forced Air Distribution System Efficiency to Climate, Duct Locationa nd Duct
Leakage. LBNL Report #43371.
Wu, H. 1989. The Effect of Various Attic Venting Devices on the Performance of Radiant Barrier Systems in Hot
28
Arid Climates. Proc. ASHRAE/DOE/BTECC/CIBSE Thermal Performance of the Exterior Envelopes of Buildings
IV, Orlando, Florida, December 1989, pp. 261-270.
7. Recent Publications
Walker, I.S. 1999. Distribution System Leakage Impacts on Apartment Building Infiltration Rates. ASHRAE
Trans. Vol. 105 Part 1. (presented at ASHRAE TC 4.10 Symposium, January 1999), LBNL 42127.
Walker, I.S. and Sherman, M.H. 1999. Assessing the Longevity of Residential Duct Sealants. RILEM 3
rd
International Symposium: Durability of Building and Construction Sealants, February 2000. LBNL 43381.
Walker, I., Sherman, M., Siegel, J., and Modera, M. 1999. Comfort Impacts of Duct Improvement and Energy-Star
Equipment. EPA Contract Report, LBNL 43723
Walker, I. 1999. Benefits Estimates for CIEE Residential Thermal Distribution Projects. Engineering report for
CIEE.
Walker, I. 1999. Sensitivity of Forced Air Distribution System Efficiency to Climate, Duct Location and Duct
Leakage. Engineering report for EPA. LBNL 43371.
Walker, I.S., and Sherman, M.H. 1999. Can Duct Tape Take the Heat? LBNL 41434.
Sherman, M.H. and Walker, I.S. 1998. Can Duct Tape Take the Heat? Home Energy Magazine, Vol.15, No.4,
Berkeley, CA.
Walker, I.S. 1998. Technical Background for Default Values used for Forced Air Systems in Proposed ASHRAE
standard 152P. ASHRAE Trans. Vol.104 Part 1. (presented at ASHRAE TC 6.3 Symposium, January 1998 also as
LBNL 40588.
Walker, I.S. and Modera, M.P. 1998. Field Measurements of the Interactions between Furnaces and Forced Air
Distribution Systems. ASHRAE Trans. Vol. 104 Part 1. Presented at ASHRAE TC 6.3 Symposium, January 1998.
LBNL 40587.
29
8. Appendices
Appendix 1: ASHRAE SP152P Duct Leakage Workshop Subcommittee
Meeting
The workshop was held in January 1999 at the ASHRAE Winter Meeting in Chicago. Table A1.1 lists the
attendees and their affiliations.
Table A1.1 Duct Leakage Workshop Attendance
Name Affiliation
Iain Walker LBNL
Paul Francisco Ecotope, Inc.
Collin Olson Energy Conservatory
Bruce Wilcox BSG
Gary Nelson Energy Conservatory
John Andrews BNL
Chuck Gaston Penn State - York
Mark Modera LBNL
Michael Lubliner WSU Energy
Three main proposals for alternative duct leakage measurements were discussed at this meeting by Chuck Gaston,
Paul Francisco and John Andrews. In each case, the proposals were presented by each of the three proponents, and
discussed by the committee.
Summary
There are several test methods that require further investigation, and the use of any individual method may depend
on the application it is used for. For example, screening tests for compliance for tight duct systems require high
repeatability at the expense of precision, but Home Energy Rating of houses with leaky duct systems require a
higher precision so as to improve estimates of potential energy savings. Thus, the test method(s) used in proposed
ASHRAE 152P may not be the same as those used by Home Energy Raters, code authorities or utility programs. It
would be possible to include more than one test in proposed ASHRAE 152P and then have recommendations based
on end use for selecting the appropriate test method. However, in a rating tool (which is what 152P has become) it
is preferable to have a single test method so that different users of the test will achieve the same result. These are
all issues for Standard Project Committee 152P members to think about and discuss during the public review of
152P. In addition, the ASTM standard E1554 needs to be reviewed and changed where necessary based on the
additional experience gained in the past few years regarding duct leakage measurement.
A main conclusion of the discussion that is common to all the test methods (this includes the pressurization test if
leakage to outside is measured and the house pressurized to the same pressure as the duct system) is that automated
measurements offer a significant increase in the precision and usability (it broadens the range of acceptable
weather conditions for testing). The standard(s) could specify that multiple automated pressure and flow
measurements are a requirement to force users to improve the precision of their measurements. A caveat is that
any test requiring envelope pressure measurements can be very difficult to perform on houses with leaky envelopes
common to many older houses in the US housing stock.
The following four tests need to be evaluated over the coming months:
Inverse Test: This test was developed into the DeltaQ test that we are currently evaluating.
Nulling Pressure Testing: The applicability of this test is much improved if an automated measurement system is
used, without this the test is really too difficult to perform. This test still requires blocking inside the duct system
30
to separate supply and return leakage and determine total leakage.
Hybrid: This uses part of the HPT/Nulling test to determine the difference between supply and return leakage by
measuring the envelope pressures, then using pressurization to estimate the total leakage. This test really requires
automated envelope pressure measurements to minimize problems arising from weather induced pressure
fluctuations (and inherent in all low pressure measurements).
Blower Door Subtraction: The automation of this test method has the same benefits as for the other tests in that
the envelope pressure measurements are vastly improved. The other recommendation regarding multipoint testing
requires more detailed investigation.
Chuck Gaston: "Inverse" Test
In this test, a blower door is set up the same way as for an envelope leakage test. A blower door test is performed
over a range of pressures (going both positive and negative with respect to outside) for two cases:
1. with the distribution system fan off.
2. with the distribution system fan on.
Pressure (across the envelope) and flow data needs to be taken at as many pressure differences as possible. In
theory, only as many data points as unknowns are required, extra data points overspecify the system but allow for
fitting to remove the effects of noise in the signal caused primarily by wind pressure fluctuations. The multiple
data points contain all the information required to determine the supply and return leakage flows at operating
conditions (in the form of pressure exponents and flow coefficients). In addition, the envelope leakage is also
determined. For the general case we have what is termed an "inverse" problem. This method has a great appeal
because the test procedure and equipment is very simple. The analysis is very complex, but in use this complexity
would be hidden from the user because a computer will be required to perform the analysis. As with all the tests
that measure envelope pressure, it was recommended that the uncertainties can be reduced by automating the
pressure measurement procedure to take large numbers of pressure readings until a certain variance limit is met.
The use of curve fitted data that can be applicable in this test will also act to reduce the influence of pressure
fluctuations
Paul Franciso: Nulling Pressure Testing
In this test, a fan and flowmeter are used to return the house envelope to its distribution fan off pressure difference
when the system fan is on and therefore measure the leakage flow imbalance. Essentially, this acts to remove the
uncertainties about envelope leakage from the existing house pressure test (HPT). Past experiments with this test
procedure showed that it was difficult to adjust the fan precisely enough to get to exactly the same pressure as with
the system fan off, particularly with wind induced fluctuations of envelope pressure. Paul discussed how this
problem had been much reduced by using automated measurement software developed by Collin and Gary. Paul
also showed how the supply leakage could be measured separately from the return leakage also using a nulling
technique by using two fan/flowmeters. The combination of the measured supply leakage and the measured
difference can then be used to calculate the return leakage. In the supply leakage measurement, the return is
blocked off before the air handler and Fan/flowmeter (1) installed so as to blow air through the system.
Fan/flowmeter (1) is then set up to produce the same flow through the system as at operating conditions by
matching static pressure between the supply plenum and the conditioned space with that measured during normal
system operation (this is the same technique as is currently in SP152P for measuring air handler fan flow).
Fan/flowmeter (2) is connected across the building envelope and is then used to return the envelope pressure to the
same value as with fan/flowmeter (1) off. The flow through fan/flowmeter (2) is then the supply leakage at
operating conditions. As discussed above, using an automated system for pressure measurement (and possibly fan
control) is an important part of this test.
Paul pointed out that this supply leakage test works best when fan/flowmeter (1) can be connected to draw air from
inside the conditioned space. A second way this test can be done is with fan/flowmeter (1) drawing air from
outside and using a blower door to depressurize the house. The flow through the blower door when the pressure
across the envelope is the same as with the air handler off is then the supply register and supply leakage to inside.
The difference between the blower door and fan/flowmeter (1) is the supply leakage to outside. This second
method has some disadvantages that make the first method preferable:
1. You need to determine the difference between two large measured numbers whose uncertainty can be the same
31
as the number you are looking for - this leads to large uncertainties in supply leakage flow.
2. The air drawn into the house will be at a different temperature than indoor air and some temperature
corrections will have to be made to the measured volumetric flows.
John Andrews: Combined HPT and Pressurization (hybrid)
John has suggested that the first part of the HPT (House Pressure test) that is used to estimate the difference
between supply and return leakage should be used in conjunction with a duct pressurization tests of the whole duct
system. This procedure removes the time consuming and difficult task of physically separating the supply and
return portions of the duct system. Although most of those present felt that eliminating the need to block the
supply from the return is a significant time and effort savings, the sensitivity of the envelope pressure
measurements to wind induced pressure fluctuations may create large errors in estimating the imbalance flow.
Paul Franciso estimated that the uncertainty of the supply-return split using this method would be in the order of
+/- 100 cfm. As with all the other envelop measurements the uncertainty can be significantly reduced if suitable
time averaging of the pressures is performed using an automated data acquisition system.
Alternatively, if the separation were installed and the pressurization tests used to separate the supply and return
duct leakage, using the HPT result for supply/return imbalance could be used to provide a simple additional check.
This would be useful in systems where the estimate of operating pressure is poor (because all the leaks are
concentrated in a single location), and the resulting flow calculated from the pressurization tests is incorrect by a
large amount.
John also reported (and all those present agreed) that the current procedure in 152P of using pressure pans to
estimate register pressures can be biased too high. There was a brief discussion regarding balancing the reduction
in precision uncertainty against the increased biased errors inherent in the use of the pressure pans. This
discussion also included removal of pressure measuring requirements altogether (except for systems with simple
returns (and with filters at the return grille) where a single pressure measurement will be a good indicator of
pressure across any return leaks). The measured flow at 25Pa would then be used as the leakage flow at operating
conditions. In this case the test method is trading an increase in repeatability (everyone doing the test will use the
same pressure difference) and simplicity against a possible decrease in precision. However, given the uncertainty
in the measurement of pressures so that the resulting calculated flow is the flow out of the leaks (this requires a
leakage flow weighted pressure) it is difficult to estimate the reduction in precision using this single pressure
method. If the test is being used as a rating tool (particularly if a pass/fail criteria is set at a low leakage level) then
this single pressure approach has merit. This approach has been taken by the California Energy Commission in
the new Title 24 Residential Building Energy Code and has been used previously in utility and local government
programs, in which repeatability is a key factor because the credit obtained for having good ducts should depend on
the test personnel as little as possible. The opposite may be true for Energy Ratings of existing houses that have
leaky duct systems where the exact leakage at operating conditions is a major factor in determining the energy use
of the building and/or the energy benefits that would occur if the duct system were repaired/sealed.
The efficiency changes using a fixed pressure rather than attempting to use the measured register pressures (or
plenum pressures) has been calculated using proposed ASHRAE 152P. The leakage flows were calculated using
the limits of the register and plenum pressures as well as the fixed pressure (see details of these calculations
below). It was difficult to reach a single conclusion from these example calculations, but they show that in many
cases the selection of a single pressure may not introduce undue penalties. This is less true with leakier systems,
where the change efficiency becomes more sensitive to leak pressure selection.
Gary Nelson: Improved Blower Door Subtraction
The current ASTM standard (E1554) for measuring duct leakage has not proved to be very popular. Gary Nelson
has been working on improving this test procedure by focussing on automated data sampling to improve precision.
Gary also discussed the possibility of doing multipoint tests, then using regression results in a modified version of
E1554.
32
ASHRAE 152P Efficiency Limits Due to Extremes of Duct System Pressure Variation
At the recent ASHRAE 152P Duct Leakage Diagnostic Subcommittee Workshop, a question was raised regarding
the maximum uncertainty in distribution system efficiency that arises from changing the assumptions about
pressures across duct leaks. In order to investigate this, measured register and plenum pressures combined with
measured pressure exponents were used to determine the change in leakage flows for measurements made at 25 Pa
(as required in the standard). The pressures and exponents were measured in six test houses last summer. The
houses were all new construction, with only one house already occupied. Two houses were in Sacramento, CA.,
two in Palm Springs, CA., one in Mountain View, CA., and the last house was in Cedar Park, TX (near San
Antonio). The houses were tested in a total of 14 conditions that included sealing and adding leakage. The
register pressures averaged 5 Pa and the plenum pressures averaged about 60 Pa. There was significant house to
house variation (expressed as standard deviation) of these pressures of ±3 Pa for registers and ±30 Pa for plenums.
These variations depended on many duct installation factors and there were significant differences even between
houses with the same duct layout and equipment. It should be noted that the register pressures were not measured
using a pressure pan, but were measured with carefully located pitot-static pressure probes inserted at the edges of
the registers so as to be away from the main flow field. Based on our field experience (and that of other
researchers) we can assume that pressure pan measurements would have yielded higher register pressures, thus
leading to less extreme pressures and flows. Therefore the results of the calculations presented here are biased
towards extreme values and using pressure pan measurements would have lowered the spread of calculated
efficiencies. The leakage tests were performed at both 25 Pa and 50 Pa so that an estimate could be made of the
pressure exponent. For these houses the average pressure exponent was 0.6. This value has been measured before
in other multi-pressure duct leakage tests by LBNL.
Two base leakage levels at 25Pa were chosen (expressed as fractions of fan flow): the 22% total used as the default
value in California T24 energy Code and the 6% total used in T24 for duct efficiency credit. This is a reasonable
range of values for new construction. Note that the 22% and 6% are assumed to be evenly split between supplies
and returns, i.e. 11% supply and 11% return for the 22% total case. Using the 5 Pa and 60 Pa limits and the mean
measured pressure exponent of 0.6 gives the following leakage flow multipliers:
38.0
25
5
multiplier leakageregister @ All
6.0
=
=
7.1
25
60
multiplier leakage plenum @ All
6.0
=
=
Combining the two base leakage levels with the three pressures (25 Pa, register and plenum) yields the following
six leakage cases:
• 6% @ 25 Pa, 2% @ registers and 10% at plenums
• 22% @ 25 Pa, 8% @ registers and 38% at plenums
We can see from these simple calculations how the effect of the leak location assumption is much less at lower
leakage levels.
For the 152P distribution system efficiency calculations, the following house was modeled:
2000 ft
2
, 2 Story, Attic Ducts (10% regain), ACCA D design, single speed heating and cooling equipment, R4 flex
duct with surface area equal to 152P defaults (assuming two return registers). The fan flow and capacity were
determined using the defaults used in T24: 0.5 cfm/ft
2
for heating, 0.7 cfm/ft
2
cooling, heat exchanger temperature
change was 55°F for heating and 20°F for cooling. This results in fan flows of 1000 cfm for heating and 1400 cfm
for cooling. The capacities are 17.3 kW for heating and –8.7 kW (2.5 tons) for cooling. In Sacramento, the
cooling design temperature is 98°F and the seasonal temperature is 73°F. For heating, the design temperature is
32°F and the seasonal temperature is 48°F. Entering all this information (plus the cooling humidity information)
into the 152P calculations resulted in the following table:
33
Table A1.2 Sacramento, CA ASHRAE 152P Distribution System Efficiency, %
Efficient Ducts Typical Ducts
Total
Leakage
Base
6%
Registers
2%
Plenum
10%
Base
22%
Registers
8%
Plenum
38%
Heating
Design
84 86 81 74 82 66
Heating
Seasonal
85 87 83 76 84 68
Cooling
Design
72 78 67 51 69 33
Cooling
Seasonal
80 83 78 70 79 60
Mean 80 84 77 68 79 57
Table A1.2 shows that for the efficient ducts, the assumption about leak location changes the seasonal efficiencies
by only a couple of percentage points. The design efficiencies are more sensitive and changed relative to the Base
case by more than 5 percentage points for cooling. The acceptability of these results depends on what they are used
for. With annual energy calculations (energy codes and home energy ratings) the seasonal results appear
satisfactory, however, if one is interested in peak demand (design) calculations, then the five percentage point
cooling differences may be too large. Bear in mind that the above results are extreme values, i.e. as bad as it could
be, and so a five percentage point maximum error may still be acceptable for load and peak demand calculations.
The typical duct system showed much greater variation, with a 10 percentage point change relative to the Base case
even for seasonal calculations.
It is reasonable to ask if using extreme results is a good indicator of the range of performance of real duct systems.
To answer this we can look at recent tests performed by LBNL that separated the register boot and plenum/cabinet
leakage from the total. In the seven houses tested, the combined plenum/cabinet and boot leakage accounted for
25% to 75% of the total system leaks, with an average of about 74%. This result indicates that the above results
are extreme and would be unlikely to occur in a real house. Looking at the 25 Pa leakage results for registers and
plenums/cabinets separately: the registers averaged 53% of leakage to outside (range of 20% to 60%, standard
deviation of 22%) and plenums/cabinets averaged 21% of leakage to outside (range of 5% to 50%, standard
deviation of 18%). Summarizing these results, it looks like the extreme case is to have about half of the system
leaks at either the plenum/cabinet or the registers, while a typical system has about one half the 25 Pa leakage at
the registers and about 20% at the cabinet/plenum. To first order, this implies that you would expect about half the
variation indicated in Table A1.2. This means that all the results would be within a plus or minus five percentage
point band.
Another aspect of the results is that they are for ducts in an extreme location – the attic. For another type of system
– a basement heating system in Fargo, ND., Table A1.3 shows that the leakage induced variations in efficiency are
small because the total losses are small, even for very leaky ducts. In this case, I used the same house as above, but
placed the ducts in an “unconditioned” basement with insulated walls (regain of 75%), and a design temperature of
–18°F and a seasonal temperature of 14°F.
34
Table A1.3 Fargo, ND ASHRAE 152P Distribution System Efficiency, %
Efficient Ducts Typical Ducts
Total
Leakage
Base
6%
Registers
2%
Plenum
10%
Base
22%
Registers
8%
Plenum
38%
Heating
Design
93 94 91 87 92 81
Heating
Seasonal
93 94 92 87 92 82
35
Appendix 2. Delta Q duct Leakage test
Procedure:
1. Install blower door and envelope pressure difference tubing/sensor.
2. With blower door fan opening blocked, blower door off and system off measure pressure difference across
envelope with blower door off ∆P
zero
. ∆P
zero
is subtracted off all the envelope pressure measurements (or
remeasured at the end of the test and some average used).
3. Turn on the system and measure the pressure across the envelope, ∆P
env
(at Q=0, where Q is the flow through
the blower door).
4. Measure the plenum operating pressures - ∆P
s
for supply and ∆P
r
for return – relative to the conditioned space.
Note that both pressures are recorded as positive numbers for use in the analysis, i.e., the return pressure is
NOT negative.
5. Turn on the blower door until there is 5 Pa across the envelope. Record ∆P
env
, and Q
on
.
6. Turn off the system fan and adjust the blower door fan to obtain the same pressure ∆P
env
across the envelope.
When the pressures are matched, record Q
off
.
7. Repeat steps 5 and 6, but with the envelope pressure, ∆P
env
, incremented by about 5 Pa each time. At each
∆P
env
there will be a pair of flows Q
on
and Q
off
.
8. Subtract ∆P
zero
from each ∆P
env
to obtain ∆P.
9. Calculate ∆Q
i
at each P
i
by subtracting Q
off,i
from Q
on,i
.
10. Do a non-linear fit of the P and ∆Q pairs to:
( )
∆
+
∆
−−
∆
−
∆
+=∆
n
r
n
r
r
n
s
n
s
s
P
P
P
P
1Q
P
P
P
P
1QPQ
to find supply leakage: Q
s
, return leakage: Q
r
, and the pressure exponent for duct leaks: n.
Note that all envelope pressures are measured relative to outside – i.e. P
in
–P
out
, so that pressurization of the house
is a positive pressure. Similarly, flows into the house through the blower door are also positive.
Derivation of DeltaQ test
Nomenclature:
C
env
= flow coefficient for building envelope
C
r
= flow coefficient for return duct leaks
C
s
= flow coefficient for supply duct leaks
n
env
= envelope pressure coefficient
n
r
= return leak pressure coefficient
n
s
= supply pressure coefficient
Q
on
= measured flow through blower door with A/H fan on
Q
off
= measured flow through blower door with A/H fan off
Q
s
= supply leak flow at operating conditions to outside
Q
r
= return leak flow at operating conditions to outside
∆P = pressure difference across envelope (in-out)
∆P
s
= pressure difference across supply leaks at operating conditions.
∆P
r
= pressure difference across return leaks at operating conditions (note that this is a positive number for flow
into ducts, so Q
r
is positive)
With the A/H fan off we have:
(
)
(
)
(
)
(
)
srenv
n
s
n
r
n
envoff
PCPCPCPQ ∆+∆+∆=∆
36
With the A/H fan on we have:
(
)
(
)
(
)
(
)
srenv
n
ss
n
rr
n
envon
PPCPPCPCPQ ∆+∆+∆−∆+∆=∆
“DeltaQ” is the difference between these two:
(
)
(
)
(
)
(
)
(
)
[
]
(
)
(
)
[
]
rrss
nn
rr
nn
ssoffon
PPPCPPPCPQPQPQ ∆−∆−∆+∆−∆+∆=∆−∆=∆∆
Defining the supply and return leakage flows:
(
)
s
n
sss
PCQ ∆=
(
)
r
n
rrr
PCQ ∆=
and
( )
s
n
s
s
s
P
Q
C
∆
=
( )
r
n
r
r
r
P
Q
C
∆
=
Substituting C
s
and C
r
into the deltaQ equation, we get:
( )
∆
∆
−
∆
∆−∆
+
∆
∆
−
∆
∆+∆
=∆∆
rrss
n
r
n
r
r
r
n
s
n
s
s
s
P
P
P
PP
Q
P
P
P
PP
QPQ
This equation can be solved for Q
s
, Q
r
, n
s
and n
r
given the measured plenum pressures, ∆Q’s and ∆P’s. However,
it is easier (and more robust) if we fix the duct leakage pressure exponents. Experiments to characterize the
pressure exponent have shown that a value of 0.6 is suitable for most duct systems. The variability in this exponent
is between 0.5 and 0.7. If we fix the value of n, and do alittle algebraic manipulation we get a form that gives
DeltaQ in terms of a difference between the supply and return leaks and is a little clearer to interpret (e.g., it is
easier to see that when ∆P=0, then ∆Q is the difference between supply and return leaks).
( )
∆
∆
+
∆
∆
−−
∆
∆
−
∆
∆
+=∆∆
n
r
n
r
r
n
s
n
s
s
P
P
P
P
1Q
P
P
P
P
1QPQ
Uncertainty Estimate for exponent and duct pressure assumptions
Using plenum pressures assumes that these pressures characterize the pressure across the leaks. The uncertainty
associated with fixing the value of n and using plenum pressures has been investigated parametrically by using an
actual DeltaQ test and varying n and the supply and pressures. The following table contains the results of this
parametric study. In Table A2.1, the pressures were varied over a range that captures the variation we expect to
find in a duct system. If the leaks were all at the registers, then we need to use a low pressure: 5 Pa in this case,
and if we change the pressure measurement location (and orientation of the pressure probe) in a plenum we find
that pressures can change by a factor of two: as shown by the increased pressures in the table. Note that in this
table we have used the worst case values in order to bound the problem. An estimate of typical uncertainty would
be less than the variation shown here.
37
Table A2.1 DeltaQ Sensitivity Test
∆P
s
, Pa ∆P
r
, Pa
n Q
s
, cfm Q
r
, cfm
9.8 22.4 0.6 14 167 plenum pressures: actual measurements, fixed n
9.8 22.4 0.7 44 187 High value of n
9.8 22.4 0.5 -7 155 Low value of n
20 22.4 0.6 31 194 doubled supply pressure
5 22.4 0.6 -2 151 halved supply pressure
40 22.4 0.6 33 199 quadrupled supply pressure
9.8 10 0.6 1 177 halved return pressure
9.8 40 0.6 21 163 doubled return pressure
9.8 5 0.6 -18 178 quartered return pressure
These results show that this test method is not very sensitive to the assumed pressure exponent or the leak
pressures. Note that this result only applies to this particular test, so we will have to do similar sensitivity studies
on some more house results before we can say that the test method is insensitive for all situations.
Flow Adjustments for Exact Pressure Matching
The trickiest part of the test procedure is the matching of pressures with the distribution fan on and off. With an
automated system that monitors the envelope pressure and can adjust the fan this would be made much easier
(particularly if the envelope pressures have the typical fluctuations seen in field tests). Because the pressure and
flow pairs will not be exactly matched, we need to have a procedure for determining what the flow difference
should be with the pressures matched exactly. This procedure can also be automated as follows. If we take the
system fan off as the reference pressure and flow condition we need to match the fan on conditions. By doing a
power law fit to all the fan on data we can obtain the pressure exponent for the fan on data. Using this exponent
and the ratio of the reference pressure to the actual fan on pressure we can find the fan on flow at the reference
pressure:
Let the reference pressure for a given data point be ∆P
off
and the corresponding flow is Q
off
. We now take some
distribution system fan on data at ∆P
on
with a corresponding Q
on
. Although ∆P
on
and ∆P
off
are close they are not
exactly the same. If we fit to all of the fan on data we can obtain the pressure exponent n
on
. The on flow can now
be corrected to be at the same pressure as the off data:
on
n
on
off
oncorrected,on
P
P
QQ
=
This correction can be applied to all the fan on data so that we have flows at exactly matched pressures. Because
we aim to have the measured P
off
and P
on
close to begin with, any uncertainties in assuming that the pressure flow
relationship is a power law and in evaluating the pressure exponent are small. In other words, because the flow
corrections will be small anyway (probably less than 5%), the errors in this interpolation procedure will not be
significant.
38
Comparison to other measurements
The pilot test of the DeltaQ procedure was performed in a house that we have already made many duct leakage
measurements in. The following table summarizes the test results for comparison purposes.
Table A2.2 Comparison of duct leakage measurement procedures
DeltaQ Duct
Pressurization
1
Duct Pressurization
2
NPT
3
Tracer gas
Q
s
, cfm 14 51 30 17 n/a
Q
r
, cfm 167 116 95 151 160
1- Converted to operating pressures using pressure pans
2- Converted to operating pressures using plenum pressures
3- NPT = Nulling Pressure Test.
39
Appendix 3. Summary of field measurement performance metrics
In the following tables, there are some results that are counter intuitive. The main culprit in these cases is the
uncertainty in the relative humidity measurements. For example, in Table A3.1 there are some cases where the
“total” TAR is less than the “sensible” TAR. This is particularly evident in the 5 minute results for Site 2, where
the leaks sealed case gives the lowest TAR. Detailed examination of the measured data has shown that these
anomalies are due to poor RH measurements. Improved RH measurements (and plenum temperatures) are needed
to reduce the incidence of these results. For this reason, we are now performing improved RH calibrations on the
RH sensors and also performing period field recalibrations.
Table A3.1 Tons At the Register
5minutes 30 Minutes 60 minutes
Total Sensible Total Sensible Total Sensible
condition Mean Mean Mean Mean Mean Mean
Site 1 as found 1.8 1.6 3.0 3.0 2.8 2.9
Site 1 sealed 3.5 3.0 3.0 2.9 2.7 2.6
Site 2 as found 3.9 2.6 2.6 2.7 2.6 2.8
Site 2 leaks added 4.1 2.6 2.3 2.5 N/a N/a
Site 2 leaks added,
correct charge
4.8 2.6 3.1 2.4 2.2 2.4
Site 2 sealed,
correct charge
2.6 2.8 2.2 2.6 2.2 2.6
Site 3 as found 3.3 2.1 2.9 2.1 2.8 2.1
Site 3 as found
correct charge
3.4 2.2 3.1 2.3 3.1 2.3
Site 3 sealed,
correct charge
4.1 2.4 3.7 2.6 3.6 2.7
Site 4 as found 3.1 1.1 3.0 1.2 2.9 1.3
Site 4 as found,
new compressor
1.9 1.3 1.7 1.3 1.6 1.3
Site 4 Leaks added 1.4 1.0 1.3 1.0 1.4 1.2
Site 5 as found 1.8 1.3 1.6 1.2 1.6 1.4
Site 5 sealed 1.8 1.3 1.6 1.4 1.6 1.4
Site 6 as found 2.8 1.2 2.0 1.3 1.9 1.4
Site 6 as found,
new compressor
n/a n/a 2.2 1.4 2.0 1.4
Site 6 leaks added 2.8 1.4 2.1 1.4 2.0 1.4
40
Table A3.2 Capacity at the indoor coil
5 minutes 30 minutes 60 minutes
Total Sensible Total Sensible Total Sensible
Condition KW kW kW kW kW kW
Site 1 as found 7.4 6.7 6.2 5.8 12.5 12.7
Site 1 sealed 13.6 12.2 12.45 11.7 11.6 11.2
Site 2 as found 18.2 12.3 11.9 12.3 11.9 12.7
Site 2 leaks added 18.4 12.4 11 12
Site 2 leaks
added, correct
charge
20.9 12.3 14.4 11.8 10.9 12
Site 2 sealed,
correct charge
10.9 11.6 10.1 11.6 9.8 11.2
Site 3 as found 13.7 8.6 12.1 8.8 11.386 8.4
Site 3 as found
correct charge
13.2 8.2 12.2 8.56 12 8.7
Site 3 sealed,
correct charge
14.5 8.2 13.6 9.5 13.2 9.7
Site 4 as found 12.3 4.8 11.7 5 11.5 5.2
Site 4 as found,
new compressor
7.6 5.17 6.9 5.4 6.73 5.5
Site 4 Leaks
added
6.1 4.53 5.6 4.7 5.49 4.67
Site 5 as found 9.9 6.9 8 7.1 7.8 7.1
Site 5 sealed 7.96 6.1 7.1 6.3 5.6 5
Site 6 as found 12.1 6 9.5 6.9 9.2 7.1
Site 6 as found,
new compressor
9.94 7.17 9.7 7.4
Site 6 leaks added 11.8 6.6 9.3 6.7 8.97 6.8
41
Table A3.3 System Power consumption
5 minutes 30 minutes 60 minutes fan as fraction of
compressor
Condition kW kW kW
Site 1 as found 5.6 5.1 5.1 0.1
Site 1 sealed 5.4 5.3 6.1 0.1
Site 2 as found 5.6 5.7 5.4 0.1
Site 2 leaks added 5.3 5.2 0.1
Site 2 leaks
added, correct
charge
5.7 5.7 5.6 0.1
Site 2 sealed,
correct charge
6.1 6.6 6.5
Site 3 as found 4.6 4.5 4.7 0.2
Site 3 as found
correct charge
4.4 4.2 4.2 0.2
Site 3 sealed,
correct charge
4.2 4.2 4.2 0.2
Site 4 as found 2.9 2.9 2.8 0.2
Site 4 as found,
new compressor
2.4 2.4 2.3 0.3
1
Site 4 Leaks
added
2.1 2.1 2.1 0.3
Site 5 as found 3.1 3.1 3.1 0.2
Site 5 sealed 3.2 3.2 3.1 0.2
Site 6 as found 3.7 3.7 33.7 0.2
Site 6 as found,
new compressor
3.7 3.7 0.2
Site 6 leaks added 3.8 3.9 3.9 0.15
1 – Large variation indicating a variable speed compressor see above
42
Table A3.4 Key Temperatures and Enthalpies for calculating system performance
5 minutes 30 minutes 60 minutes
condition Tout
1
(°C)
hreturn
2
(kJ/kg)
Tattic
3
(°C)
Tout
(°C)
hreturn
(kJ/kg)
Tattic
(°C)
Tout
(°C)
hreturn
(kJ/kg)
Tattic
(°C)
Site 1 as found 29.3 44.4 35.4 26.9 36.4 30.3 26.7 35 37.3
Site 1 sealed 30.7 44.5 37.7 31.6 41.3 37.3 36.7 43.1 40.5
Site 2 as found 28.8 41.4 33.6 30.7 40.3 36.7 27.5 36.3 29.9
Site 2 leaks added 26 40.3 28.4 25.5 37.6 27.2
Site 2 leaks
added, correct
charge
30.3 41.9 32.6 29.8 38.7 31.5 29.3 37 30.3
Site 2 sealed,
correct charge
33.2 43.4 40.1 36.9 43.7 41.1 36.2 41.6 39.9
Site 3 as found 28.9 56.2 29.5 52.6 32.9 54.9
Site 3 as found
correct charge
26.3 55 25.6 50.7 24.4 48.5
Site 3 sealed,
correct charge
24.7 50.7 24.5 45.6 24.1 43.1
Site 4 as found 36.9 61.4 60.2 36.5 56.1 57.5 35.8 53.6 55
Site 4 as found,
new compressor
32.2 57.4 56 32.0
2
52.2 53.4 31.9 49.6 50.6
Site 4 Leaks
added
32.82 55.7 53.6 32.6 51.8 50.9 31.9 49.6 48.1
Site 5 as found 31.6 52.4 44.5 31.7 47.1 40.3 31.3 44.8 38.6
Site 5 sealed 34.7 52.6 47.3 33.4 46.9 42.4 33.1 44.8 40.6
Site 6 as found 33.7 54.5 53.8 33.9
6
48.9 54.4 34.3 46.7 53.1
Site 6 as found,
new compressor
38.8 33.4 48.8 48.9 33.8 46.3 50.6
Site 6 leaks added 27.4 41.8 37.1 27.1 46.4 37 27.2 43.6 35.8
1- outside air dry bulb temperature
2- enthalpy of air in return
3- attic air dry bulb temperature
43
Table A3.5 Temperature at different locations in the house during pulldown tests
5 minutes 30 minutes 60 minutes
condition
Thermostat
[°C]
Master
BR [°C]
Kitchen
[°C]
Thermostat
[°C]
Master
BR [°C]
Kitchen
[°C]
Thermostat
[°C]
Master
BR [°C]
Kitchen
[°C]
Site 1 as found
26.1 27.7 27.2 23.2 24.1 21.9 22.1 23.2 20.7
Site 1 sealed
25.8 27.1 25.1 24 25.3 23 23.1 25.2 22.7
Site 2 as found
25.8 25.7 23.8 23 24.7 22.3 22.5 22.9 19.9
Site 2 leaks
added
24.5 24.3 22.6 22.5 22.6 20.4
Site 2 leaks
added, correct
charge
24.7 25 23 22.9 23.3 20.9 22.1 22.8 20.3
Site 2 sealed,
correct charge
25.2 26.3 23.9 24.8 26 23.3 23.8 25.2 22.3
Site 3 as found
25.3 27.7 25.1 23.6 26.6 23.64 23.9 27.4 24.2
Site 3 as found
correct charge
25.7 28.2 25.2 23.8 26.7 23.2 23 26 22.6
Site 3 sealed,
correct charge
25 27.7 24.6 23.2 25.5 22.7 22.2 24.7 21.9
Site 4 as found
30 26.9 28.1 28.5 26.7 25.9 27.6 26.1 25
Site 4 as found,
new compressor
28.6 27.5 26.8 27 25.5 24.7 26.1 24.8 23.8
Site 4 Leaks
added
28.1 27 26.9 26.7 25.1 24.9 26 24.4 24.1
Site 5 as found
28.4 31.1 27.6 25.4 29 24.8 24.6 28.6 23.9
Site 5 sealed
29.2 30.8 27 25.9 28.5 23.8 24.9 28 22.96
Site 6 as found
26.6 30.1 28 27.8 28.6 25.5 27.3 28.2 24.7
Site 6 as found,
new compressor
27.6 28.4 25.2 27 27.7 24.4
Site 6 leaks
added
24.4 26 23.6 23.6 24.5 21.4 22.9 23.7 20.7
44
Table A3.6 Delivery Effectiveness
Sensible Total
5 minutes 30 minutes 60 minutes 5 minutes 30 minutes 60 minutes
Site 1 as found
0.95 0.81 0.80 0.95 0.81 0.80
Site 1 sealed
0.88 0.85 0.82 0.88 0.85 0.83
Site 2 as found
0.76 0.76 0.77 0.83 0.78 0.76
Site 2 leaks
added
0.73 0.74 0.79 0.73
Site 2 leaks
added, correct
charge
0.74 0.72 0.72 0.81 0.75 0.70
Site 2 sealed,
correct charge
0.85 0.79 0.80 0.85 0.78 0.78
Site 3 as found
0.87 0.86 0.89 0.85 0.85 0.87
Site 3 as found
correct charge
0.95 0.94 0.94 0.90 0.90 0.91
Site 3 sealed,
correct charge
1.05
1
0.98 0.98 0.99 0.96 0.97
Site 4 as found
0.85 0.85 0.84 0.88 0.89 0.88
Site 4 as found,
new compressor
0.86 0.83 0.85 0.88 0.85 0.87
Site 4 Leaks
added
0.78 0.77 0.90 0.80 0.79 0.90
Site 5 as found
0.64 0.68 0.70 0.70 0.71 0.72
Site 5 sealed
0.72 0.77 0.79 0.77 0.80 0.80
Site 6 as found
0.69 0.67 0.67 0.82 0.75 0.74
Site 6 as found,
new compressor
0.78 0.74
Site 6 leaks
added
0.77 0.75 0.75 0.85 0.80 0.80
1- Error in supply plenum temperature from spatial variation and response time
The Delivery Effectiveness is often higher at the start (5 minute values) due to lower conduction losses. A t the
beginning of the cycle the air in the ducts is not as cool as later in the cycle.
The sensible DE is a function of both conduction and leakage. The total DE contains the moisture losses that are
from leakage only.
45
Table A3.7 Equipment Coefficient of Performance (COP)
Sensible Total
5 minutes 30 minutes 60 minutes 5 minutes 30 minutes 60 minutes
Site 1 as found
0.7 2.6 2.5 0.7 2.6 2.5
Site 1 sealed
2.3 2.1 2.0 2.5 2.1 2.1
Site 2 as found
2.3 2.1 2.4 3.5 2.1 2.2
Site 2 leaks
added
2.4 2.3 3.5 2.1
Site 2 leaks
added, correct
charge
2.2 2.1 2.0 3.7 2.5 1.9
Site 2 sealed,
correct charge
1.9 1.7 1.7 1.8 1.5 1.5
Site 3 as found
1.9 2.0 1.8 3.0 2.7 2.4
Site 3 as found
correct charge
1.9 2.0 2.1 3.0 2.9 2.9
Site 3 sealed,
correct charge
1.9 2.3 2.3 3.4 3.2 0.3
Site 4 as found
1.7 1.8 1.9 4.4 4.3 4.3
Site 4 as found,
new compressor
2.2 2.4 2.4 3.3 3.0 2.9
Site 4 Leaks
added
2.2 2.2 2.2 2.9 2.7 2.6
Site 5 as found
2.2 2.3 2.3 2.9 2.6 2.5
Site 5 sealed
1.9 2.0 2.0 2.5 2.3 2.2
Site 6 as found
1.6 1.9 1.9 3.2 2.6 2.5
Site 6 as found,
new compressor
1.9 2.0 2.7 2.6
Site 6 leaks
added
1.8 1.7 1.8 3.1 2.4 2.3
46
Table A3.8 Total System Coefficient of Performance (COP)
Sensible Total
5 minutes 30 minutes 60 minutes 5 minutes 30 minutes 60 minutes
Site 1 as found
0.7 2.1 2.0 0.7 2.1 2.0
Site 1 sealed
2.0 1.8 1.6 2.2 1.7 1.7
Site 2 as found
1.6 1.6 1.8 2.5 1.6 1.7
Site 2 leaks
added
1.7 1.7 2.8 1.5
Site 2 leaks
added, correct
charge
1.6 1.5 1.5 3.0 1.9 1.4
Site 2 sealed,
correct charge
1.6 1.4 1.4 1.5 1.2 1.2
Site 3 as found
1.6 1.7 1.6 2.6 2.3 2.1
Site 3 as found
correct charge
1.8 1.9 2.0 2.7 2.6 2.6
Site 3 sealed,
correct charge
2.0 2.2 2.3 3.4 3.1 3.1
Site 4 as found
1.4 1.5 1.6 3.9 3.8 3.8
Site 4 as found,
new compressor
1.9 2.0 2.1 2.8 2.5 2.6
Site 4 Leaks
added
1.7 1.7 2.0 2.4 2.1 2.4
Site 5 as found
1.4 1.5 1.6 2.0 1.8 1.8
Site 5 sealed
1.4 1.5 1.6 1.9 1.8 1.8
Site 6 as found
1.1 1.2 1.3 2.7 1.9 1.8
Site 6 as found,
new compressor
1.4 1.4 2.1 2.0
Site 6 leaks
added
1.4 1.3 1.3 2.6 1.9 1.9
47
Table A3.9 Pulldown time and temperature variation in different locations in the house
Pulldown Time (minutes)
Temperatures (°C)
Thermostat Master BR Kitchen Thermostat Master BR Kitchen
Site 1 as found
11 14 22 24.3 25.2 26.8
Site 1 sealed
32 19 50 24.1 23.1 25.4
Site 2 as found
26 10 27 24.5 22.3 24.4
Site 2 leaks
added
8 2 7 24.0 21.9 23.8
Site 2 leaks
added, correct
charge
10 3 13 24.0 22.2 24.2
Site 2 sealed,
correct charge
30 10 55 23.9 22.8 25.3
Site 3 as found
29 45 120 24.0 24.1 27.0
Site 3 as found
correct charge
28 14 180 24.0 23.4 26.8
Site 3 sealed,
correct charge
15 10 95 24.0 23.6 26.6
Site 4 as found
239 122 198 24.1 21.9 23.2
Site 4 as found,
new compressor
159 64 107 24.0 21.9 23.1
Site 4 Leaks
added
170 75 92 24.0 22.1 22.7
Site 5 as found
87 63 204 23.9 23.5 27.7
Site 5 sealed
107 34 180 24.0 22.2 26.7
Site 6 as found
257 93 266 24.0 21.0 24.2
Site 6 as found,
new compressor
118 20 123 23.2 20.3 23.1
Site 6 leaks
added
94 8 75 23.6 22.8 25.2
48
Appendix 4. Flowchart for REGCAP Model
START
Output
Options
Initialize
Simulation
Input
Data
House Data
Attic Data
Duct Data
Equipment Data
First Time
Step
Equipment Model
Mass Flows
Temperatures
Humidity Ratios
Output
END
Write
Output
Weather Data
Initial Temps
(optional)
49
Appendix 5. REGCAP Simulation Sensitivity to input data
uncertainty
Additional model verification tests have been completed at sites 1 and 2. The results of these comparisons
are less encouraging, but are based on problems with the input (measured) data rather than problems with
the model. The average absolute temperature differences for each air node at each site are shown below in
Table A5.1. It is clear that the model to measured comparison is not very good at these sites, although it is
acceptable in the house and ducts when the air handler is on during the pulldown tests. The model
overpredicts the attic temperature difference by a very large margin both when the air handler is on and
when the air handler is off. This discrepancy is caused by errors in the solar input data. The solar sensors
were poorly calibrated at these two sites and the shading device was very rudimentary and didn’t always
work. This caused us to overestimate total horizontal radiation and to underestimate direct normal
radiation. This in turn causes the model to incorrectly overpredict the solar gain on both the house and
the attic (although more significantly on the attic). Other problems with the model that these two sites
revealed include the lack of coupling between the solar gain and the house mass and the inadequate heat
transfer coefficient between the house mass and the house air. These problems will be corrected when an
improved load calculation routine is implemented.
Table A5.1. Comparison of measured and modeled temperatures illustrating
problems with measured input data
Site Date |T
hse
| |T
attic
| |T
supply
| |T
return
|
Whole
day
Pulldown
only
Whole
day
Pulldown
only
Whole
day
Pulldown
only
Whole
day
Pulldown
only
1 June 6,
1998
4.1 0.4 21.4 14.7 8.7 1.1 9.4 1.7
2 June 10,
1998
5.0 0.4 16.7 12.8 6.2 0.7 5.4 1.2
50
Appendix 6. Evaluation of flow hood measurements of
residential register flows
Eight flowhoods were used in a new single story house in Sacramento CA. The house had 9 supply
registers and a single return. The following discussion will concentrate on the supply measurements
because none of these flowhoods was large enough to cover the large return grille and correctly measure
the return flow. The registers were either high on the wall or in the ceiling. All the ducts in this house
were in the attic.
Two of the flowhoods were powered flowhoods that used the flowhood as a flow capture device and
measured the flow with a separate flowmeter. The two powered flowhoods used a fan to compensate for
the insertion loss of the flow capture hood and the flow measurement device. The compensation is
accomplished by balancing the pressure between inside the flowhood and the room so that the flow out of
the register is not reduced by placing the flowhood over the register. To test the sensitivity of the powered
flowhoods to the setting of the balancing pressure, the tests were repeated with different measurements of
balancing pressure. The LBNL fan assisted flow hood was tested in two modes:
1. Balancing the room and flowhood static pressure with flowhood static pressure measured in the
corner of the flowhood up against the wall – i.e., as distant from the bulk flow field as possible.
2. Balancing the flowhood total pressure with room static pressure. The flowhood total pressure was
measured using the pressure sampling array normally used for direct flowhood measurements.
The Energy Conservatory flowhood was also tested on two modes:
1. Balancing the flowhood and room static pressures with static pressure measured in the corner of the
flow capture hood near the wall (the same as the first LBNL test)
2. With the flow capture hood static pressure measured at the other end of the flow capture hood near
the fan/flowmeter entrance.
The other six flowhoods were from four manufacturers. The following table summarizes the important
information about each flowhood tested.
Table A6.1 Flowhood characteristics
Code for flowhood Description/characteristics
Hood1 Hard glass fiber capture hood with propeller flow measurement (manufacturer A)
Hood2
Standard flow hood with ∆P flow measurement (manufacturer B)
Hood3
Small flow hood with ∆P flow measurement – specifically for low residential flows
(manufacturer B)
Hood4
Standard flow hood with ∆P flow measurement (manufacturer C)
Hood5
Standard flow hood with ∆P flow measurement (manufacturer D)
Hood6
Standard flow hood with ∆P flow measurement (manufacturer D – same model as
Hood5, but different serial number)
Hood7 LBNL fan assisted flow hood – total hood pressure balance
Hood8 LBNL fan assisted flow hood – static hood pressure balance
Hood9 Energy Conservatory fan assisted flowhood – corner hood pressure balance
Hood10 Energy Conservatory fan assisted flowhood – near fan entry hood pressure balance
51
The following table summarizes the individual register measurements for each of these flowhoods:
Table A6.2 Comparison of flowhood measurements of supply registers (cfm)
Register Hood 1 Hood 2 Hood 3 Hood 4 Hood 5 Hood 6 Hood7 Hood 8 Hood 9 Hood 10
1 184 194 182 205 240 226 197 194 197 203
2 167 181 163 173 232 238 169 186 173 175
3 234 243 275 265 370 341 239 246 244 250
4 156 162 158 164 233 248 151 154 149 164
5 68 69 80 86 110 122 75 75 72 72
6 56 52 55 60 91 66 58 65 53 53
7 103 84 93 102 153 157 98 96 90 95
8 54 43 53 43 83 4 49 53 45 46
9 51 43 52 60 87 82 44 52 47 46
sum 1073 1071 1111 1158 1599 1484 1080 1121 1070 1104
These results show that some flowhoods (4, 5 and 6) can give substantially different results from the
others. In particular, Hoods 5 and 6 that are from the same manufacturer give flows that are much too
high. There are also some significant differences on a register by register basis. If we take Hood 7 to be
our measurement standard for comparison purposes, all the flowhoods except Hood 9 have differences for
an individual register of 13 cfm or greater. This magnitude of difference may be a concern if the
flowhood measurements are used to verify ACCA designs, for example. Another key result is that register
9 was in an interior bathroom and the duct design called for only 5 cfm at this register. Measuring flows
this small is very difficult with existing portable flow measurement equipment and would be very difficult
to verify. However, in a real duct system it is almost impossible to install it to get this low a flow and, as
our results show, the actual flow out of the register is substantially higher than its design flow. This case
shows that some interpretation of required design flows is required (for field test verification and
compliance testing) because it is probably better to allow the higher than design flow in this case, rather
than attempt to restrict the flow to the design value.
Static vs. total pressure balancing.
Comparing the two LNBL tests (Hood 7 – total and Hood 8 -static) shows that the static pressure
balancing results in consistently (7 out of 9 registers) lower flow measurements. However, the differences
are quite small – less than the specified flow measurement uncertainty (±3% of flow) for 5 of the 9
registers and so this difference is not very significant.
Changing balancing pressure measurement location
Comparing the Hood 9 and Hood 10 results shows that measurement location is not critical. The
differences between the two tests are less than the flow meter measurement uncertainty (±3% of flow) for
all but two of the registers.
Comparing return measurements
Due to limited capacity not all of the flowhoods had the capability to measure the return flows. Table
A6.3 compares the results for the six flow hoods that were used to measure the return flow. These results
show good agreement between hoods 4 through 8, but hood 3 gave results that were too low. For hood 3,
the return flow measurement was at the upper limit of its measurement range. This flowhood added
significant flow restriction to the return, resulting in a lower flow through the register and flowhood.
52
Table A 6.3 Comparison of flowhood measurements of return register flow (cfm)
Hood 3 Hood 4 Hood 5 Hood 6 Hood 7 Hood 8
860 995 1028 1055 1037 1057
