Radiant Barrier Impact on Selected Building Performance
Measurements
Model Home Case Study
Centex Homes – Charlotte, NC
Appalachian State University
Energy Center
Summer Study 2008
Bruce Eugene Davis
Jeffery Tiller
March 31, 2009
Table of Contents
Executive Summary
1
Project Objective
3
Background
4
Methodology
5
Results
30
Discussion
42
Conclusions
45
Appendix A – Building Performance Measurements, Centex Model Homes
46
Appendix B – Installation Dates and Labor Efforts
49
Appendix C – Sensor Code Identification Information
50
Acknowledgements
Funding for this effort was in part provided by a Building America Grant sourced through the
North Carolina Energy Office.
The following individuals each contributed to the completion of this case study.
Jeff Tiller, Interim Chair, ASU Technology Dept. and ASU Energy Center, originated the
general idea, secured the funding, provided the Centex contact, and asked important questions
Edwin Woods, Area Construction Manager, Centex Homes, Charlotte, NC, provided guidance
and approval for the model home summer study
Kristin Bryan, Spicewood Field Manager, Centex Homes, Charlotte, NC, appointed to be
Centex direct contact for ASU summer study staff and facilitated the necessary interactions with
Centex staff
Woodbury Sales Center Staff, Centex Homes, Charlotte, NC, always helpful as ASU staff
checked in and checked out of the summer study model homes
Andrew Palmer, Woodbury Field Manager, Centex Homes, Charlotte, NC, appointed to
maintain ongoing Centex direct contact for ASU staff
Lee Ball, Instructor, ASU Technology Dept. and ASU Energy Center, installation partner for all
the sensors and in taking the initial building performance measurements
Laurel Elam, ASU Energy Center, multiple general support, operational guidelines, and keeps
Omnisense and internet cable accounts open to maintain data flow and access.
Sharon Yates, ASU Energy Center, provided invaluable guidance on travel and purchasing and
ASU project guidelines
Andrew Sams, ASU Technology Student and Building Performance Engineering staff, was the
main individual helping to air seal the attic ceiling plane
Adam Milt, Building Performance Engineering, provided support and background information
with regard to Centex Homes, and provided the critical extra assistance for the attic ceiling plane
air sealing
Andrew Windham, ASU Technology Dept. Graduate Student, assisted with taking follow up
building performance measurements, and was one of the three person team that installed the attic
radiant barrier
Scott Critcher, ASU Technology Dept. Graduate Student, helped prepare some of the materials
prior to the attic air sealing, and was one of the three person team that installed the attic radiant
barrier
Wes Stewart, ASU Technology Dept. Graduate Student, one of the three person team that
installed the attic radiant barrier
Ben Hannas, Advanced Energy, provided initial consulting support with regard to managing the
data created through the Omnisense data logging system
Jason Hoyle, ASU Energy Center, managed the first request for and the securing of the initial,
massive, raw data down load onto the ASU Energy Center computer
Ross Gosky, ASU Math Dept., has provided an opportunity for discussions about exploring the
data set
Ross Dillon, ASU Technology Dept. Graduate Student, assisted with the site visit to sample the
attic insulation depth across the attic of the two model homes, and is analyzing a portion of the
Omnisense data for a paper
Executive Summary
In the home construction industry one common solution to a customer cooling discomfort
complaint is to dispatch the HVAC contractor to replace the existing system with a larger one.
This case study explored one of several alternative strategies to solve a summer cooling
complaint. It is generally accepted that attic radiant barriers can reduce attic temperatures during
daily peak temperatures in the cooling season. It is also generally accepted that for homes with
R-30 ceiling insulation and no HVAC equipment or ducts in the attic that the addition of an attic
radiant barrier does not provide an improvement in the home energy efficiency. However this
case study was designed to measure the impact of the addition of an attic radiant barrier in an
attic which contained both HVAC equipment and ducts. Beyond documenting the impact on roof
deck and top of attic insulation temperatures, this case study explored heat gain in an attic supply
air duct between the HVAC supply plenum and the ceiling supply boot. It also documented the
subsequent impact on the HVAC air handler run time.
Benefiting from a long standing working relationship between the ASU Technology Department
and Centex Homes – Charlotte, North Carolina, initial discussions were held to determine if
Centex management would be open to an unobtrusive summer study being set up in one or more
of their model homes at one of their active subdivision sales centers. Once operational ground
rules were established and the design outlines of the summer study were provided Centex agreed
to the study. A site supervisor was assigned to be the study staff’s direct contact at Centex and a
subdivision was recommended for our review.
The Belmont was chosen to be the intervention home and the Parkwood was held as the
nonintervention home. This approach allowed for staged modifications to the intervention home
to investigate their impact. If measured conditions in the home altered following the modification
but did not change for the nonintervention home, then the measured difference in the
intervention home is more reliably to have been caused by the building modification.
Building performance measurements were recorded for both homes. As appropriate, follow on
measurements were also recorded. Omnisense data logging systems [www.omnisense.com] were
installed in both homes. Individual sensors were placed inside the home, inside the HVAC
system, in the attic, and outside. The sensors send a radio signal to an Omnisense Gateway in
each home. Each Gateway is connected to a router which is in turn connected to a cable modem
and through the local cable into the internet. Data is stored on the Omnisense server for viewing
online or for down loading. Data for each sensor location includes sensor ID, location
description, date and time stamp, temperature, relative humidity, grains per pound, dew point,
and wood moisture number [if attached to wood with the required stainless steel screws].
Two staged modifications were planned for the Belmont. Air sealing of the ceiling plane was
included to remove a confounding variable. The second modification was the installation of the
radiant barrier to the bottom of the top chord of the roof truss. A summary of the study sequence
includes the characterization of each home, instrumentation, two week baseline data, one week
set aside for air sealing, two week data period, one week set aside to apply radiant barrier, and
two week data period.
1
61 sensors were installed. There are 30 sensors in the Belmont and 29 in the Parkwood, plus 2
outside [1 is on the Belmont front porch, 1 is in a tree shelter tube at the tree line between the
two back yards]. Sensors are placed in comparible locations in each home.
The installation of an attic radiant barrier attached to the bottom of the top chord of the roof truss
was shown to have several air conditioning season beneficial impacts with regard to the energy
efficiency of the Belmont model home.
1. The peak temperature measured at the top of the attic insulation was reduced by
approximately 23 degrees F at peak temperature for comparible pre and post dates. This
reduction in temperature provided a positive benefit across a nine hour period, though to
a lesser and lesser amount as the data is examined further from the peak.
2. The run time of the HVAC system for the second floor of the Belmont was reduced by
14% over the total 24 hour period. During the 7 peak hours the run time was reduced by
20%.
3. Heat gain through the R-6 flex duct into the cooling supply air between the attic air
conditioning unit and the ceiling supply boot was reduced from 7 degrees F to a heat gain
of 3 degrees F.
4. Heat gain of 7 degrees F equaled a capacity loss of 33.8%. Heat gain of 3 degrees F
equaled a capacity loss of 14.4%. Thus the radiant barrier improved the efficiency of the
cooling air delivered through this duct during the peak attic temperatures by 57%.
5. Air sealing the ceiling plane produced a substantial disruption to the physics of building
natural air change for both winter and summer seasons. As measured by a blower door,
this 3205 square foot home started with a whole building air leakage of 1700 CFM50 and
was air sealed to 1250 CFM50 which is a 26.5% reduction. This included a notable
reduction in the duct air leakage to Outside by sealing the boot penetration of the ceiling
sheetrock.
On the not beneficial side of the energy efficiency effort, the HVAC system cycled on and off
more during the peak period and the radiant barrier kept the top of attic insulation slightly hotter,
longer over the cool down period in the evening.
The Omnisense sensor system, once the initial wrinkles were ironed out, has proved to be very
flexible, reliable and accessible building measurement equipment. There are still multiple
additional data sets that could be analyzed. Potentially it may be possible to solve the mystery of
the differences in attic heating patterns between the intervention home and the nonintervention
home attics.
2
Project Objective
In the home construction industry one common solution to a customer cooling discomfort
complaint is to dispatch the HVAC contractor to replace the existing system with a larger one.
This case study explored one of several alternative strategies to solve a summer cooling
complaint. It is generally accepted that attic radiant barriers can reduce attic temperatures during
daily peak temperatures in the cooling season. It is also generally accepted that for homes with
R-30 ceiling insulation and no HVAC equipment or ducts in the attic that the addition of an attic
radiant barrier does not provide an improvement in the home energy efficiency. However this
case study was designed to measure the impact of the addition of an attic radiant barrier in an
attic which contained both HVAC equipment and ducts. Beyond documenting the impact on roof
deck and top of attic insulation temperatures, this case study explored heat gain in an attic supply
air duct between the HVAC supply plenum and the ceiling supply boot. It also documented the
subsequent impact on the HVAC air handler run time. One desirable response with a positive
benefit outcome documented would be the conclusion that the addition of an attic radiant barrier
could be one more option, when appropriate, in a solutions tool kit for solving customer summer
cooling discomfort related to peak outdoor temperatures.
Three secondary objectives were also envisioned. To achieve the primary objective, data logging
equipment would need to be installed. The decision was made to install equipment that could
extend access to the building measurements beyond the required summer period. The Omnisense
sensor system once installed to provide the primary period data could continue to provide that
secondary objective. Another objective was to provide an opportunity for ASU students to see
the range of building measurements over time and experiment with manipulating real world data
to extract useful patterns of information with regard to how building operate. The final objective
was to set up a two building opportunity to learn how building operate over time and the
potential opportunity to introduce additional building modifications and then be able to measure
and report impacts.
3
Background
During the year prior to the beginning of this case study in May 2008 a student lead laboratory
experiment was completed which used a testing module to examine in that setting the impact of a
radiant barrier on attic temperature. The module provided for simulated solar heat gain and attic
temperature measurement either with or without a radiant barrier installed. The experiment
recorded an approximate 22 degrees F temperature difference in attic temperature at the peak of
simulated solar heat gain.
Since it is nearly impossible [and would be extremely expensive] to completely simulate all the
interactions present in a home over time [hours, days, weeks, seasons], this case study was the
obvious extension of that experiment.
4
Methodology
Benefiting from a long standing working relationship between the ASU Technology Department
and Centex Homes – Charlotte, North Carolina, initial discussions were held to determine if
Centex management would be open to an unobtrusive summer study being set up in one or more
of their model homes at one of their active subdivision sales centers. Once operational ground
rules were established and the design outlines of the summer study were provided Centex agreed
to the study. A site supervisor was assigned to be the study staff’s direct contact at Centex and a
subdivision was recommended for our review.
An initial site visit, June 17, 2008, was scheduled with the site supervisor to the Woodbury
Subdivision Sales Center Model Homes. There were four models available at the site. Two were
chosen for the study, the Belmont and the Parkwood, which are side by side, two story homes
built on slabs and have tuck under garages. They both have 4 bedrooms and 2.5 bathrooms. The
orientation of the front of the homes is toward the west. The homes were completed by
approximately January 2008.
The Belmont is on the left and the Parkwood is on the right. The Belmont garage is currently
used as a lighting center and has a separate mini split HVAC system.
5
The Belmont was chosen to be the intervention home and the Parkwood was held as the
nonintervention home. This approach allowed for staged modifications to the intervention home
to investigate their impact. If measured conditions in the home altered following the modification
but did not change for the nonintervention home, then the measured difference in the
intervention home is more reliably to have been caused by the building modification.
Building performance measurements were recorded for both homes. As appropriate, follow on
measurements were also recorded. Omnisense data logging systems [www.omnisense.com] were
installed in both homes. Individual sensors were placed inside the home, inside the HVAC
system, in the attic, and outside. The sensors send a radio signal to an Omnisense Gateway in
each home. Each Gateway is connected to a router which is in turn connected to a cable modem
and through the local cable into the internet. Data is stored on the Omnisense server for viewing
online or for down loading. Data for each sensor location includes sensor ID, location
description, date and time stamp, temperature, relative humidity, grains per pound, dew point,
and wood moisture number [if attached to wood with the required stainless steel screws].
Two staged modifications were planned for the Belmont. Air sealing of the ceiling plane was
included to remove a confounding variable. The second modification was the installation of the
radiant barrier to the bottom of the top chord of the roof truss. A summary of the study sequence
includes the characterization of each home, instrumentation, two week baseline data, one week
set aside for air sealing, two week data period, one week set aside to apply radiant barrier, and
two week data period.
Summer Study General Time Line:
June 17
Site visit to review and confirm the application of the study to the model homes and choose the
two model homes for the study and document each with photographs
June 19
Receive final approval to proceed from Centex
June 20
Begin sensor system and cable access purchases
June 30 – July 11
Install sensor monitoring systems and establish that they are running smoothly in both the
intervention and the nonintervention model homes
Perform and record all building performance measurements and equipment information
July 14 – 27
Record and establish baseline conditions in both homes
July 28 – August 3
Air seal attic and duct boot penetrations, and redistribute attic insulation evenly
6
August 4 – 17
Record and establish condition changes from first modification to the intervention model home
and continue recording conditions in nonintervention model home
August 18 – 24
Install Radiant Barrier system in intervention model home
August 25 – September 7
Record and establish condition changes from second modification to the intervention model
home and continue recording conditions in nonintervention model home
September 21
End of summer study and transition to continue recording the data for a full 12 month period
Omnisense Data Logging Dates and Sampling Intervals:
Sensors installed:
2008 – 07 – 7 and 8 and 9
Data Start:
2008 – 07 – 09
1:00pm [13:00]
Baseline Period
5 minute sampling [11 to 12 measurements per hour]
Attic Air Sealing – Belmont
2008 – 07 – 28 and 29
Sampling time change
2008 – 08 – 12
1.27 minute sampling [47 to 48 measurements per hour]
10:35am [10:35]
Radiant Barrier Install – Belmont 2008 – 08 – 18 and 19
Sampling time change
1 hour sampling [24 per day]
2008 – 09 – 22
1:25pm [13:25]
Sensor Installation Locations:
61 sensors were installed. There are 30 sensors in the Belmont and 29 in the Parkwood, plus 2
outside [1 is on the Belmont front porch, 1 is in a tree shelter tube at the tree line between the
two back yards]. Sensors are placed in comparible locations in each home.
First floor:
2 room sensors
1 at thermostat
2 in supply air boots [in rooms with room sensors]
1 in return air box
7
Second floor:
4 room sensors
1 at thermostat
4 in supply air boots [in rooms with room sensors]
1 in return air box
Attic:
2 under attic insulation [1 north end, 1 south end]
2 on top of attic insulation [directly over under insulation sensors]
2 on string, 5 feet above insulation [directly above the previous sensors]
2 in return air ducts at air handler [for first floor and for second floor]
1 in each supply air plenum at air handler [2 for Belmont, 1 for Parkwood-zoned]
4 screwed into top chord of roof truss touching roof decking [nE, sE, sW, nW]
Main codes used for sensor naming:
Two homes are:
B = Belmont – intervention [has two furnaces with A/C in attic, fiberglass attic insulation]
P = Parkwood – nonintervention [has one furnace with A/C with zone control in attic, cellulose
attic insulation]
F = first floor
S = second floor
A = attic
wb = wood block base [allows wood moisture number reading]
wt = wood truss [allows wood moisture number reading]
All data being generated from the two model homes is stored on the Omnisense server.
Additionally a copy of all the data is downloaded periodically and stored on the ASU server.
Data from selected sensors were compiled for targeted analysis related to the project objectives.
Photo Review
Following are a selection of photographs covering the Omnisense sensor placements, ceiling
plane air sealing, radiant barrier installation, recording attic insulation depth pattern, and
window, equipment, and insulation information.
8
Occupied space conditions [six per home, temperature, relative humidity, wood moisture
content, two on first floor, four on second floor]
Return air conditions from occupied space [one per floor]
9
Supply air conditions [six per home, two on first floor, and four on second floor]
Conditions at thermostat [two per home, one on each floor]
10
Attic air conditions [two per attic]
Roof deck conditions [four per attic, temp, RH, wood moisture content]
11
Top of insulation and bottom of insulation at ceiling sheet rock [two sets per attic]
Exterior conditions at tree between Belmont and Parkwood homes [weather protected]
12
Return air at air handler [two systems, temp, RH, wood moisture content]
Placing sensor on Douglass Fir wood block inside HVAC supply plenum
13
Supply air at air handler [two systems, temp, RH, wood moisture content]
Omnisense Gateway, router, and cable modem [equipment pushed back under low furniture to
reduce the potential for curiosity tampering]
14
Ceiling plane air sealing at top plate
Foam air sealing in progress
15
Attic ceiling insulation was moved aside to allow detailed air sealing protocol
16
Safety first, Ground Fault Circuit Interrupter for all electricity for attic lighting
Bathroom exhaust fan prior to sealing exhaust fan and sheet rock penetrations with caulk
17
Top plate and abandoned plumbing hole prepared for air sealing with foam
Attic insulation moved aside for air sealing at duct chase and top plates
18
Lid on duct chase prepared for sealing with foam
Close up of ceiling plane duct chase lid needing air sealing [duct penetration sealed by HVAC]
19
Cutting radiant barrier material [cutting was quickly moved out of the hot attic]
Fitting radiant barrier around truss members
20
Fitting radiant barrier around plumbing and duct strapping
Attaching radiant barrier to the bottom of the top chord of the roof truss
21
Completing the first side and it is still hot
Sensor now between radiant barrier and roof deck with flagging tape marking location
22
Tough work starting at the edge of the roof with low head space
Ridge vent left open to both the attic space and the air path behind radiant barrier
23
Reflecting attic light at the moment
Redistributing the attic insulation that was disturbed during radiant barrier installation
24
Finishing touches
Close up detail
25
Looking pretty and flex duct to Belmont nW ceiling supply boot reported in this case study
Leaning over to measure attic insulation depth, sampling taken in pattern across entire attic
26
Comparing our ruler with the installed attic ruler, it was often not this deep
Fiberglass attic insulation card from Belmont
27
Yes it is low-E glass
Model information was gathered for all equipment
28
Cellulose attic insulation card from Parkwood
29
Results
Following the completion of the initial two week data logging period to establish Omnisense
baseline measurements, air sealing protocols were completed. As outlined in the methodology
most potential air leakage paths through the ceiling plane, between the home and the attic were
sealed from the attic. In addition, all ceiling boot penetrations for both the first and second floors
were sealed after temporarily removing their supply grille. These efforts changed the whole
building air leakage for the 3205 square foot home from 1700 CFM50 to 1250 CFM50. This
elimination of 450 CFM50 equals a 26.5% reduction in whole building air leakage which by
itself enhanced the efficiency of the building shell. More importantly, during the heating season a
substantial barrier has been installed to retard the continual, passive leakage of warm air out of
the top of the home thus substantially disrupting the natural air change physics of the home. In
some homes during the summer the action of hot air exiting the attic can create suction on the
home, if the attic inlet venting is undersized, and draw conditioned air from the home. This
summer action was not examined but would also be disrupted.
Duct air leakage to Outside for the home was reduced from 4.9% to 3.8% per square foot of
conditioned space, which is a 22% reduction. The improvement was noticeably from the boot
leakage air sealing of the second floor system. This system air leakage dropped from 70 CFM25
to 31 CFM25 to Outside. This reduction is part of the successful overall ceiling plane air sealing
efforts. The absence of notable duct air leakage also removed a confounding variable from the
case study. The air sealing protocols required 27 hours of onsite labor, mostly in a very hot attic,
with difficult gymnastics, and moving insulation aside to access the air leakage sites. The
insulation was redistributed following completion of the air sealing work. More details of the
building performance measurements are located in Appendix A. Appendix B contains the
installation dates and labor effort details.
This air sealing effort removed one source of potential complicating building physics setting the
stage for more reliability should there be measureable changes recorded associated with the
installation of the radiant barrier. With the completion of the ceiling plane air sealing, Omnisense
data was then continued for an additional two week period. Following this second two weeks of
Omnisense data collection was the installation of the radiant barrier and a third two weeks of
Omnisense data collection.
With this accumulation of data, attention was turned to addressing what building and equipment
impacts could be measured and associated with the radiant barrier installation. In particular, what
were the identifiable impacts during the daily peak temperature periods when the radiant barrier
provides it primary benefit? The first effort was to identify a 24 hour day prior to and post to the
radiant barrier installation that had similar, though not identical, solar intensity and peak
temperatures. As a beginning point, roof deck sensor data was used to indicate solar intensity and
pattern. In both the Belmont [intervention home] and the Parkwood [nonintervention home] the
two roof deck sensors attached to the west facing roof at the south and north ends were reviewed.
The south end sensors were designated as “sW” and at the north end as “nW”. As a reminder,
during this time period the sensors were recording temperature measurements every 1.27
minutes. Detailed sensor code identification information is located in Appendix C.
30
Roof deck temperatures for the Parkwood [nonintervention home] were reviewed for the two
week periods pre and post to the radiant barrier installation. August 15, 2008 was chosen as the
pre date and August 22, 2008 was chosen as the post. The roof deck temperatures were plotted
for visual comparison. These two dates had the most similar, though not identical, roof deck
sensor data suggesting similar solar intensity and temperature over their 24 hour periods. On the
15th the sW and the nW roof deck temperatures display a slight difference at the peak
temperature of the day of 132 versus 129 degrees F. Otherwise both ends of the roof deck curves
closely track each other. For the post data on the 22nd both sensors follow identical temperature
tracks and both reach a peak temperature of 129 degrees F. Comparison of the data for the 15th
and the 22nd show more variation. The most obvious is that the peak is reached earlier on the 15th
and displays a less uniform heating and cool curve than does the curve for the 22nd. Two other
elements are that the data for the 22nd shows that during the morning hours the roof deck is 5
degrees F cooler and during the evening it is 7 degrees F warmer. For clarification, the reference
to “pre/post radiant barrier” in the title of this Parkwood graph is only a reference to the time
period involved. A radiant barrier was not installed in the Parkwood since it is the
nonintervention home.
31
Roof deck temperatures for the nW sensors were plotted for both the Belmont [intervention
home] and the Parkwood [nonintervention home]. At the peak temperature of the day on the 15th
the Belmont roof deck sensor nW compared with that of the Parkwood reported 135 versus 132
degrees F. The nW sensor data for both homes on the 15th followed similar patterns. However on
the 22nd the roof deck nW sensor data for the Belmont reported that it was 14 degrees F hotter
than the Parkwood. This is expected since the roof deck sensor for the Belmont following the
radiant barrier installation is located behind the radiant barrier which is restricting the roof deck
heat flow into the attic. During the non peak periods on the 22nd the roof deck sensors reported
similar temperatures for both homes. With these two examinations of the roof deck sensor data it
was decided to continue with the comparison of radiant barrier impacts from the August 15th and
22nd data.
Prior to focusing solely on the impact of the radiant barrier on the temperature at the top of the
attic insulation, two combination plots were created to visually display the roof deck data with
the top of insulation data. Graphs were created, first with the nW roof deck data and then with
the sW roof deck data. For some, the graphical display may be viewed with an appreciation for
its esthetic artistry. Others will notice the similarity of both. Still others will be encouraged by
the significant drop in the top of the attic insulation temperatures during the peak period
following the installation of the radiant barrier. That review immediately follows these two
graphs.
32
Before passing over these two 8 sensor data curves, a couple of comparative observations are of
interest. Prior to the radiant barrier installation the Belmont top of attic insulation temperature
was 11 degrees F cooler than its roof deck. For this same time period the Parkwood top of attic
insulation temperature was 23 degrees F cooler than its roof deck.
33
Heat gain into the home from the attic begins at the top of the attic insulation. The sensor data for
the north end of the attic at the top of the insulation for both homes on the 15th and 22nd were
plotted. What is immediately noticeable is that although the previously reviewed roof deck
temperatures for both homes prior to the radiant barrier were similar at their peak and over their
curve the top of the attic insulation temperatures for the two homes are different. At the 3:00pm
peak the degree F temperature for the Belmont is 123 and for the Parkwood it is 109. This is a 14
degree F difference. Even though both homes were shown to have been exposed to the same
environmental and orientation conditions, the resulting temperature at top of the attic insulation
was different. Neither sensor had a duct within close proximity and the duct air leakages for both
homes were similar and acceptable. However, this study was not designed to explore the
difference in performance of the two different homes. The focus was to determine what impacts
could be discerned from installing an attic radiant barrier in the intervention home.
In the Parkwood the top of the attic insulation data for the peak depicted that during the post
period the peak occurred 45 minutes later at 3:45pm and reached a slightly cooler 106 degrees F.
With the radiant barrier installed the Belmont peak did not occur until 4:00pm. Its peak was 100
degrees F which was 23 degrees F cooler than prior to the radiant barrier. Additionally, the
separation of the temperature curves in the morning and their convergence in the evening are
interesting. Prior to the installation of the radiant barrier the Belmont top of insulation sensor
starts to report hotter temperatures than the Parkwood sensor at 11:10am, at 85 degrees F. In the
evening the Belmont remains hotter over the cool down period. At 8:45pm the Belmont is 3
degrees F hotter at 79 versus the Parkwood 76. On August 22 these temperature curves separate
earlier at 9:00am, at 67 degrees F. The Belmont retains the advantage until around 6:15pm, at 95
degrees F, when their positions flip and the Parkwood is cooler. Across the evening the radiant
barrier appears to influence a hotter top of attic insulation temperature. Again, at 8:45pm the
Belmont is 4 degrees F hotter at 84 versus the Parkwood 80.
Given that the environmental conditions were similar for both August 15th and the 22nd, and that
the Parkwood peaks were close and the temperature curves similar, it can be reliably inferred that
the radiant barrier did provide a substantial reduction in the temperature at the attic air and top of
attic insulation interface. As hypothesized, the benefit of the radiant barrier accrues during the
daily peak temperatures. Adversely the radiant barrier retains additional heat in the attic during
the evening cool off period.
34
To clearly see the impact of the radiant barrier on the top of attic insulation temperatures just the
pre and post radiant barrier top of insulation curves for the Belmont are compared below.
35
Omnisense sensors were placed in several locations within the HVAC system. With the
approximate one minute data reporting it was possible to review the ceiling boot temperature
data and accumulate the approximate periods of time that an HVAC system was operating and
cooling the home for each 24 hour period. Data from the second floor, nW ceiling supply boot
was used for this analysis. This information was developed into a bar graph for both the Belmont
[intervention home] and the Parkwood [nonintervention home] for August 15th and 22nd.
Reviewing the Parkwood data, the indicated HVAC run time was 9.2 hours pre and 9.4 hours
post. This supports the position that the two dates, August 15th and 22nd were comparible
environmental days with regard to air conditioning needs. The Belmont indicated HVAC run
time was 13.7 hours pre and 11.8 hours post. For the Belmont second floor cooling this change in
run time is equal to a 14% reduction. The reduction during the top 7 peak hours is 20% which
emphasizes the radiant barrier benefit during the hottest period of the day. Although run time is
not a direct measurement of energy used for cooling it does indicate a beneficial efficiency.
To additionally support that the reduction in HVAC run time was related to the radiant barrier
performance the Omnisense data from the four sensors attached to the four thermostats was
plotted. Each time that the ASU staff was on-site the thermostats were checked to be set at
approximately 75 degrees F. For the Parkwood the pre and post graphs for each the first and
second floors follow similar patterns. Comparing the first with the second floor graphs also
depicts the whole home with comparible cooling patterns. The Belmont thermostat patterns have
more variability but are not wildly different. While the second floor thermostat shifts noticeably
at the end of the 24 hour period, after 7:00pm, the difference is only about 1 degree F. Please
note that the vertical axes are at .1 and .2 degree F which visually exaggerates the differences.
36
37
38
Heat gain into attic ducts from the time the conditioned air leaves the supply plenum of the
HVAC system located in the attic until it reaches the ceiling supply boot was examined. In the
Parkwood data the stress of the peak attic temperatures can be seen in the change of pattern in
the afternoon of each day. But the afternoon patterns on both days are comparible. Heat gain
during the peak is 4 degrees F in the pre period for the Parkwood.
39
Examination of the duct heat gain pre radiant barrier installation in the Belmont during the
3:00pm peak period measured the supply plenum and the ceiling supply boot at 55 and 62
degrees F respectively. This 7 degree F heat gain occurred through R-6 flex duct exiting the
supply plenum 5 feet high in the attic air and then traveling across the top of the attic insulation
[noted in earlier photo] to the ceiling boot covered in insulation. The HVAC system was
continually cooling without stopping during this period from 1:56pm until 3:26pm, a 1.5 hour
run time period. Following the installation of the radiant barrier the supply plenum was still
producing 55 degree F cooling air. However the ceiling boot air temperature was 58 degrees F.
The duct heat gain changed to a 3 degree F rise during the period from 4:37pm until 5:10pm, a
33 minute run time period. That brings up another impact of the radiant barrier. Following the
installation, the HVAC system did have less total run time but the pattern of operation shifted to
more cycling on and off during the peak temperatures.
Data from the sensors at the inlet and outlet of the HVAC equipment for the Belmont second
floor system can be used in calculations for several performance elements. On August 15th during
the peak at 2:50pm following cooling for 54 minutes the degree F measurements were inlet 76.1
dry bulb / 63 wet bulb and outlet 55.4 dry bulb. The SystemVision HVAC Refrigerant Charging
Quick Reference, Chart 2 – Target Temperature Split for Cooling Systems provided a desired
temperature spread of 19.2 degrees F. The measured temperature spread was 20.7 degrees F
which was 1.5 degrees F above the desired temperature spread. The following calculation was
used for determining the air flow for this 2.5 ton cooling system.
25 cfm/ton x 1.5 degrees above desired = 37.5 cfm x 2.5 tons = 93.75 cfm low
1000 desired cfm – 93.75 cfm low = 906.25 cfm actual system air flow
The duct heat gain listed above for August 15th was 7 degrees F. The capacity loss for this duct
between the air handler supply plenum and the boot was 33.8%.
7 degrees heat gain -divided by- 20.7 degrees spread = .338 loss
On August 22nd during the peak at 5:00pm following cooling for 23 minutes the degree F
measurements were inlet 75.3 dry bulb / 62 wet bulb and outlet 54.5 dry bulb. The desired
temperature spread was 19.3 degrees F. The measured temperature spread was 20.8 degrees F
which was 1.5 degrees F above the desired temperature spread. Again, the calculation produced a
system air flow of 906.25 cfm giving the same result prior to and post to radiant barrier
installation.
The duct heat gain listed above for August 22nd was 3 degrees F. The capacity loss for this duct
between the air handler supply plenum and the boot was 14.4%.
3 degrees heat gain -divided by- 20.8 degrees spread = .144 loss
Thus the radiant barrier improved the efficiency of the cooling air delivered through this duct
during the peak attic temperatures by 57%.
40
41
Discussion
The primary objective of the case study was to document some of the impacts of the installation
of an attic radiant barrier on the performance of a home. With the beneficial outcomes
documented in this case study there are now economic and construction practice issues that will
need to be addressed with regard to costs. What are the comparative installation costs, for
example, if a builder decides to consider choosing between, a radiant barrier, increasing HVAC
size, or moving the HVAC system and ducts out of the attic? If an attic radiant barrier is chosen
as a potential solution, will it be used as standard practice or will it be used as one available
solution for the occasional home that ends up with a peak cooling discomfort problem?
With regard to applying an attic radiant barrier as we did in this study as an aftermarket repair,
will there be an interest in providing the other half of the attic work which is the ceiling plane air
sealing protocol? To provide that air sealing protocol the case study staff used a decidedly very
low-tech approach to move aside and then redistribute the attic insulation. A cheap, plastic dust
pan with a flexible lip proved most effective at moving the blown fiberglass insulation and a
plastic leaf rake worked well to redistribute it. Hopefully a more labor efficient process would be
utilized by a crew that provided this work as an ongoing service. However, the case study work
does show that a Mr or Ms everyone can achieve the same result without the requirement of
paying someone with high tech equipment to perform these tasks.
If the decision was made to include a radiant barrier as a standard construction protocol then
another existing solution would be the OSB roof decking with the radiant barrier already
attached. However this case study does not shed any light on the comparability of these two
different approaches to applying an attic radiant barrier. Another question that has been asked
with regard to the OSB radiant barrier and the under truss top chord application of this case study
is that when the attic insulation is applied following the radiant barrier is there a surface dust
impact on the performance of the radiant barrier performance?
Turning to the Omnisense sensors, when they are installed in the HVAC system there will be
some time delay in the measurement in the change in duct air temperature due to the mass of the
sensor. How much of a time delay was not methodically investigated but a sample review of the
detailed sensor temperature data revealed measurement fluctuations as small as one hundredth of
a degree F and as large as 5 degrees F at the 1.27 minute sensor sampling rate. Whatever the
potential delay, it is not a problem for this case study. The study is comparing the same duct
location temperatures pre and post the radiant barrier installation. To the degree that there is a
measureable time delay [at a minimum the data could be off up to 1.27 minutes on each end of
the on cycle], it is the same for calculations of HVAC run time for both occasions. Also in the
study the run time is referred to as indicated run time and not exact run time.
The data does show that the radiant barrier installation reduced the top of attic insulation
temperature and the amount of duct heat gain. It also measures a reduction in the total run time
and the run time during the peak temperature hours. These suggest an improvement in home
energy efficiency. However the increase in HVAC cycling pattern during the peak introduces a
penalty into the efficiency gains that this case study cannot quantify. Another energy penalty that
was associated with the radiant barrier was that it supported heat retention in the attic throughout
42
the evening cool down period. This occurred even though the air path from the soffit air baffles
behind radiant barrier up to the ridge vent was maintained.
One complicating factor in the Belmont model home attic that was not corrected was the R-value
of the installed blown fiberglass insulation. During the initial inspections and work the depth of
the insulation near the attic depth cards near the center of the attic was approximately 12 inches.
However, it appeared that out in the perimeter of the attic the insulation was under blown.
Complicating that was the probable loss of some loft in the blown insulation as a result of
moving it and then redistributing it to perform the ceiling plane air sealing. The combined result
was an average attic insulation depth of 8.73 inches. The depth ranged from 12 inches down to 7
inches with the lesser depths around the perimeter as previously mentioned. The average thermal
value across the attic was calculated to be R-21.5 during the study. The depth was not measured
during the first two week baseline period. To directly quantify the efficiency improvement from
the case study work would require energy use sub metering which was not included.
The main surprise at this point in reviewing the data for the two model homes remains
unexplained. Why was there a difference for the top of attic insulation temperatures for the two
homes which were so similar and were exposed to the same solar intensity and temperature? As
stated earlier the design of this study was not to directly compare the intervention and the
nonintervention home. But when the data popped up that the Parkwood [nonintervention home]
without the radiant barrier had a top of attic insulation temperature that almost equaled the
performance of the Belmont’s [intervention home] that had the radiant barrier, one wonders
why? Why was the Belmont top of attic insulation only 11 degrees F cooler than its roof deck
temperature and the Parkwood’s was 23 degrees F cooler? Maybe upcoming work with the
model homes or more analysis from student involvement will discern the cause or causes.
The combustion appliances in the Belmont will operate safely within this home. The gas log set
is the direct vent type which is aerodynamically uncoupled from home negative pressures. The
two furnaces are in the vented attic and are mechanically exhausted, 90 plus condensing types.
Given the positive aspect that the home is substantially air tight at 1250 CFM50 for its 3205
square feet, it is recommended that there be a mechanical ventilation strategy added. One
strategy that has been used by Centex on other homes is the upgraded Panasonic bathroom
exhaust fan which runs constantly at a selectable low speed when the fan switch is in the off
position. It then runs at full speed when the switch is moved to the on position. It has the
additional benefit of very quiet operation.
43
The discussion section would not be complete without revisiting the graphs of the thermostat
temperatures. At first view as presented in the results section the separation between the pre and
post thermostat temperature graphs in the Belmont home for both the first and second floors is
eye catching. Even though the written commentary stated that the conditions were similar for
both August 15th and 22nd and that the reader should notice the fine gradient of the vertical axis it
seemed appropriate to show the data again with the 50 to 90 degree F scale that are common to
thermostats. The rescaled graphs below may offer a more comforting view and a confirmation
that both homes and both floors of each home were indeed being maintained at comparible
temperatures. As another reference point the older mercury bulb thermostats generally allowed
around a 3 degree F temperature swing. Digital thermostats will often, as in these homes, keep
the temperature swing to less than 1 to 1.5 degrees F.
Yes the data loggers were capable of measuring differences in temperatures at the thermostats
but for the general purposes of this study those temperatures were within allowable variation.
44
Conclusions
The installation of an attic radiant barrier attached to the bottom of the top chord of the roof truss
was shown to have several air conditioning season beneficial impacts with regard to the energy
efficiency of the Belmont model home.
1. The peak temperature measured at the top of the attic insulation was reduced by
approximately 23 degrees F at peak temperature for comparible pre and post dates. This
reduction in temperature provided a positive benefit across a nine hour period, though to
a lesser and lesser amount as the data is examined further from the peak.
2. The run time of the HVAC system for the second floor of the Belmont was reduced by
14% over the total 24 hour period. During the 7 peak hours the run time was reduced by
20%.
3. Heat gain through the R-6 flex duct into the cooling supply air between the attic air
conditioning unit and the ceiling supply boot was reduced from 7 degrees F to a heat gain
of 3 degrees F.
4. Heat gain of 7 degrees F equaled a capacity loss of 33.8%. Heat gain of 3 degrees F
equaled a capacity loss of 14.4%. Thus the radiant barrier improved the efficiency of the
cooling air delivered through this duct during the peak attic temperatures by 57%.
5. Air sealing the ceiling plane produced a substantial disruption to the physics of building
natural air change for both winter and summer seasons. As measured by a blower door,
this 3205 square foot home started with a whole building air leakage of 1700 CFM50 and
was air sealed to 1250 CFM50 which is a 26.5% reduction. This included a notable
reduction in the duct air leakage to Outside by sealing the boot penetration of the ceiling
sheetrock.
On the not beneficial side of the energy efficiency effort, the HVAC system cycled on and off
more during the peak period and the radiant barrier kept the top of attic insulation slightly hotter,
longer over the cool down period in the evening.
The Omnisense sensor system, once the initial wrinkles were ironed out, has proved to be very
flexible, reliable and accessible building measurement equipment. There are still multiple
additional data sets that could be analyzed. Potentially it may be possible to solve the mystery of
the differences in attic heating patterns between the intervention home and the nonintervention
home attics.
45
Appendix A
Building Performance Measurements
Centex Model Homes – Woodbury
Charlotte, NC
Summer 2008
Belmont
3205 square feet, two stories, tuck under garage, slab floor
HVAC: two systems in attic, up flow, single pipe 90+ condensing furnaces [they each have
second pipe taking combustion air from attic] with SEER 13 air conditioning
Equipment Information:
1st Floor Gas Furnace Lennox - G51MP-36B-070-08
Coil Lennox - C33-30B-2F-3
2nd Floor Gas Furnace Lennox - G51MP-36B-070-08
Coil Lennox - C33-30B-2F-3
st
1 Floor Outdoor Condenser- Lennox - 13ACD-030-230-03
2nd Floor Outdoor Condenser Lennox - 13ACD-030-230-03
Ducts in attic and between floors for first floor and in attic for second floor – two returns for
each system
Building air leakage:
Pre ceiling plane and all boot penetrations air sealing =
Post ceiling plane and all boot penetrations air sealing =
1700 CFM50
1250 CFM50
[Reduction = 450 CFM50 or 26.5%]
Duct air leakage:
First floor
Pre = 180 CFM25 Total
Post = 161 CFM25 Total
86 CFM25 Outside
91 CFM25 Outside
Second floor
Pre = 131 CFM25 Total
Post = 82 CFM25 Total
70 CFM25 Outside
31 CFM25 Outside
91 + 31 = 122 CFM 25 Outside divided by 3205 SF = 3.8% leakage/SF
Exhaust fan air flow
First floor
½ bath
Second floor hall bath
master bath area
master bath shower
broken
41 cfm
41 cfm
46 cfm
46
Building pressures
[Both HVAC blowers on, all interior doors closed]
Main body WRT Outside =
-1.8 Pascals
First floor rooms WRT Outside
Utility room
+3.8 Pascals
Half bath
+4.3
Second floor rooms WRT Outside
NW bedroom
+16.0 Pascals
NE bedroom
+6.8
Hall bath
+3.8
SE bedroom
+5.3
SW bedroom
-2.5 [has return]
Floor cavity WRT Home
Pre = +17.0 Pascals
Garage WRT Home
Pre = +25.0 Pascals
Post = +16.3 Pascals
Pressure Pan Readings in Pascals
First floor
Liv
Din
Den
Kit
Util
½ bath
Din Ret
Den Ret
Pre
+0.5
+0.5
+0.5
+0.8
+1.0
+0.5
+1.1
+1.3
Post
+0.3
+0.3
+0.5
+0.5
+0.5
+0.3
+1.1
+0.9
Attic Insulation:
White fiberglass blown
Second Floor
NW bd cl
NW bd
Loft
NE bd
H bath
SE bd
Mb bath
Mb cl
Mb 1
Mb 2
Hall
Hall Ret
Mb Ret
Pre
+0.9
+0.4
+0.5
+0.5
+0.4
+0.4
+0.3
+0.7
+0.3
+0.3
+0.4
+0.4
+0.4
Post
+0.2
+0.2
+0.1
+0.1
+0.2
+0.1
+0.2
+0.1
+0.1
+0.1
+0.1
+0.2
+0.5
R-2.46/inch
Sampling pattern of attic insulation depth measured an average depth of 8.73 inches.
Attic R value: R-2.46/inch x 8.73 inches = R-21.5 attic insulation value
Windows: Vinyl, insulated glass, Low-E coating on interior surface of exterior pane
Radiant Barrier installed to bottom edge of top chord of roof truss
47
Parkwood
2525 square feet, two stories, tuck under garage, slab floor
HVAC: one system in attic, horizontal, zoned for 2 floors, single pipe 90+ condensing furnace
[has second pipe taking combustion air from attic] with SEER 13 air conditioning
Equipment Information:
Gas Furnace, Zoned Coil Outdoor Condenser -
Lennox – G51MP-48C-090-07
Lennox – CH33-48C-2F-2
Lennox – 13ACD-042-230-02
Ducts in attic and between floors for first floor and in attic for second floor – one return first
floor and two returns second floor
Building air leakage:
Pre ceiling plane and all boot penetrations air sealing =
Post ceiling plane and all boot penetrations air sealing =
1323 CFM50
N/A
[Reductions = N/A]
Duct air leakage:
Pre = 192 CFM25 Total
Post = N/A CFM25 Total
108 CFM25 Outside
N/A CFM25 Outside
108 CFM 25 Outside divided by 2525 SF = 4.3% leakage/SF
Floor cavity WRT Home
Pre = +15.7 Pascals
Garage WRT Home
Pre = +23.5 Pascals
Attic Insulation:
Cellulose blown
Post = N/A Pascals
R-3.75/inch
Sampling pattern of attic insulation depth measured an average depth of 8.6 inches.
Attic R value: R-3.75/inch x 8.6 inches = R-32 attic insulation value
Windows: Vinyl, insulated glass, Low-E coating on interior surface of exterior pane
48
Appendix B
Installation Dates and Labor Effort
Placement of Omnisense Sensors, Gateway, Router, and Activation of Internet Access
July 7 – 9, 2008
Ceiling Plane Air Sealing: Belmont – July 28 & 29, 2008
Manually removed blown fiberglass insulation from top plates, exhausts fans, and chases
Air sealed with foam - top plate penetrations and both edges [excluding low attic at roof edge]
Air sealed with foam - penetrations in capped chases
Capped one open chase and air sealed with foam
Air sealed with caulk - penetrations at bathroom exhaust fans, including fan housing
Air sealed with caulk - the boot to sheetrock crack [included ceiling boots of first floor]
Air sealed with caulk - the return air filter grille box to sheetrock crack
Weatherstripped and added Thermax foam board [layered to R-20] to attic access panel
Manually redistributed blown fiberglass insulation across ceiling plane
Labor
July 28:
7:00am – 1:00pm
6:00pm – 9:00pm
6 hours x 2 persons = 12
3 hours x 3 persons = 9
July 29:
7:00am – 10:00am
3 hours x 2 persons = 6
Total on-site labor = 27 hours
[travel and other hours not included]
Radiant Barrier, attic: Belmont – August 18, 2008
7:00am – 1:00pm
6:00pm – 9:00pm
6 hours x 3 persons = 18
3 hours x 3 persons = 9
Total on-site labor = 27 hours
[travel and other hours not included]
49
Appendix C
Sensor Code Identification Information
B = Belmont – Intervention home
P = Parkwood – nonintervention home
F = first floor
S = second floor
A = attic
15 = August 15, 2008 – pre radiant barrier installation
22 = August 22, 2008 – post radiant barrier installation
Sensor numbers
0CD9 = BA top of insul N
0610 = PA top of insul N
10E7 = BA-wt roof deck sW
orientation]
0C9B = PA-wt roof deck sW
orientation]
112F = BA-wt roof deck nW
1377 = PA-wt roof deck nW
orientation]
[Belmont attic top of insulation North end of attic]
[Parkwood attic top of insulation North end of attic]
[Belmont attic-wood truss at roof deck south end, West
[Parkwood attic-wood truss at roof deck south end, West
[Belmont attic-wood truss at roof deck north end, West orientation]
[Parkwood attic-wood truss at roof deck north end, West
Run time B-15 = 1328
Run time B-22 = 1328
1328 = BS bdrm supply nW
[Belmont second floor bedroom ceiling supply boot north end, West orientation]
Run time P-15 = 1C7A
Run time P-22 = 1C7A
1C7A = PS supply frt bdrm nW
[Parkwood second floor ceiling supply boot bedroom north end, West orientation]
0F5E = BF thermostat 53” central [Belmont first floor thermostat 53” high, central in home]
15D2 = BS thermostat 54” [Belmont second floor thermostat 54” high, central in home]
2051 = PF thermostat
[Parkwood first floor thermostat, central in home]
1B4C = PS thermostat mst bdrm nE
Parkwood second floor thermostat in master bedroom, north end, East orientation, at door
0F96 = BA-wb supply A/H [upstairs]
[Belmont attic-wood block in supply plenum of air handler for upstairs HVAC unit]
114D = PA-wb supply A/H [up & dn]
[Parkwood attic-wood block in supply plenum of single air handler zone dampered to
condition upstairs and downstairs]
50