Baker
Content
Energy and Buildings 36 (2004) 435–441
IQ-test—improving quality in testing and evaluation of solar
and thermal characteristics of building components
Paul Baker∗
BRE Scotland, East Kilbride G75 ORZ, UK
Abstract
IQ-test is a Thematic Network supported by the European Community under the EESD Programme. The objective of IQ-test is to
further the development of common quality procedures at the PASLINK test cell facilities in 12 European countries, for the assessment
of the thermal characteristics of building components. This should consolidate the network, integrate the newer test sites and strengthen
its common approach of support for new product developments in the field of innovative building components. Round robin tests are
underway to assess both the inter-site quality of testing and analytical procedures of the participants. Two components were designed:
(1) an opaque, well insulated, homogeneous panel and (2) a window, which is used to replace the central section of the first component.
Common test and quality procedures have been implemented at each test site. The data sets generated by each team have been made
available for cross-analysis by another team. The results available so far on the first component indicate good agreement between sites.
This paper summarises the progress to date. Results are also presented from a training exercise which asked participants to identify the
performance characteristics of an unknown component without providing any physical description of the component.
© 2004 Published by Elsevier B.V.
Keywords: IQ-test; PASLINK; Test cells; Thermal performance
1. Introduction
IQ-test is a Thematic Network supported by the European Community under the EESD Programme, which aims
to consolidate the work of the network of the PASLINK
outdoor test cell facilities in 12 European countries, involved in the energy performance evaluation of the thermal and solar properties of building component. More
information may be found on the PASLINK web-site:
WWW.PASLINK.ORG. The objective of IQ-test is to
further the development of common quality procedures
for:
•
•
•
•
testing;
calibration;
data gathering, processing and analysis;
interpretation of test results and scaling/replication to real
building; and
• maintenance of the test infrastructure at the test sites.
This should consolidate the network, by integrating the
newer test sites and strengthening its common approach of
support for new product developments in the field of inno∗ Current address: Centre for Research on Indoor Climate & Health,
Glasgow Caledonian University, G4 OBA, UK.
E-mail address: [email protected] (P. Baker).
0378-7788/$ – see front matter © 2004 Published by Elsevier B.V.
doi:10.1016/j.enbuild.2004.01.046
vative building components through semi-standardised tests
and pragmatic, practicable and affordable but accurate procedures.
As part of the work of the Network, round robin tests
are being performed as part of a feasibility study for standardisation activities. The objective is to assess both the
inter-site quality of testing and analytical procedures of the
members, with a view to developing standards for outdoor
testing. High quality data for model calibration will also be
generated.
Two components were designed, incorporating flexibility,
in order that the first component could be used as a platform
for the second component. The first component is an opaque,
homogeneous with a removable central section. The thermal
properties of the panel are very well defined. The second
component is a window, which is used to replace the central
section of the first component.
The objective of testing the first component is for each
participant to determine the thermal transmission coefficient
of the panel by both 1-D heat flux measurements and an
energy balance on the test cell. For the second component:
the whole wall U-value and solar aperture estimated from
the test cell energy balance and the window U-value and
solar aperture.
Common test procedures have been designed and quality
procedures have been implemented at each test site. The
436
P. Baker / Energy and Buildings 36 (2004) 435–441
test work is completed and the data sets generated by each
team are being made available for cross-analysis by another
team.
This paper describes the components and procedures and
summarises the results to date, including cross-analysis by
other teams on the opaque component.
The results of a training exercise are also presented. Data
were provided to participants from a test carried out at one
of the test sites with another component in the aperture of
the opaque wall. The aim was to determine the thermal characteristics of the component, without providing the participants with a physical description or any clues to the type of
the “unknown” component.
2. The test components
Given the number of organisations involved in the Thematic Network it was decided at the beginning that it was
impractical to circulate one component for testing for the
following reasons:
• variation in the test aperture size of test cells between
sites;
• high transportation costs; and
• likely difficulties in keeping to a strict timetable, given
the use of the test cells for other tests.
The approach adopted was for each organisation to construct its own component(s) according to strict instructions
regarding the selection of materials, manufacture and instrumentation.
2.1. Component 1—the opaque wall
The first component is an opaque, homogeneous panel
consisting of a sandwich of insulation between plywood,
with a replacable central section 1500 mm (h) by 1250 mm
(w). The thermal properties of the panel are well defined
and the required materials are available in the locality of
each participant. Expanded polystyrene (PS30), with a density of 30 kg/m3 and a nominal thermal conductivity of
0.033 W/mK, is used to form an insulating panel of thickness
200 mm. A white exterior finish is used on the plywood.
It was agreed that each team should measure the heat flux
and temperatures in two profiles through the wall (Fig. 1),
with one in the centre of the removable section (A) and
the other mid-way between the edge of the wall and the
removable panel (B).
The objective of testing the first component is for each
participant to determine the thermal transmission coefficient
(U-value) of the panel by both:
1. 1-D heat flux measurements through profiles A and B;
and
2. an energy balance on the test cell.
Heat flux sensors
flush mounted in
plywood
B
A
Removable
central panel
inside
Fig. 1. Measurement profiles through the opaque wall.
The latter value will include edge effects, which will vary
depending on the test cell and installation of the test wall at
each site.
2.2. Component 2—the window
The objective of the second component is to introduce a
greater degree of complexity by using a window to replace
the central section of the first component. The window design incorporates double glazing using ordinary clear float
glass in a timber frame. The frames for each participant were
produced centrally. Each participant obtained glass from a
local supplier and samples were tested centrally to ensure
consistency. The spectrophotometric tests have shown that
there are differences between the samples in their infrared
transmittance, which may give variations in the g-value (solar energy transmittance) of the double glazing between 73.2
and 77.3%. Fig. 2 shows the window installed in the opaque
surround.
The objectives of the test are to determine:
1. the whole wall U-value and g-value estimated from the
test cell energy balance; and
2. the window U-value and g-value (frame and glazing).
Fig. 2. The window installed in the opaque surround at BRE Scotland.
P. Baker / Energy and Buildings 36 (2004) 435–441
437
300
HIGH
POWER
Power Level [W]
250
ROLBS
200
150
100
50
LOW POWER
0
0
1
2
3
4
5
6
7
8
9
Day
Fig. 3. A typical heating power sequence with low power = 50 W air circulation fan, high power = 50 W fan power + 200 W resistance heater.
3. The test procedures
The procedures are based on the COMPASS Measurement and Data Analysis Procedures [1]. A test sequence was
devised to reduce the overall test duration, whilst maximising information. It includes a heating/cooling regime using
a Randomly Ordered Logarithmically distributed Binary Sequence (ROLBS). The maximum power level was calculated
to ensure that the mean test room temperature difference
between the low and high power parts of the test sequence
should be at least 10 K, but preferably 20 K, without exceeding the safe operational limits of the test cell. The choice of
whether to use heating or cooling depends on the local climatic conditions. In a “heating” climate a maximum power
level of 250 W is satisfactory for the window test: the heating sequence is shown in Fig. 3.
The air leakage of the test room is also determined by
pressurisation testing before and after the test to ensure that
the test cell meets a requirement of 0.5 air changes per hour
at 50 Pa. It is also recommended that the air leakage is monitored continuously during the test by tracer gas measurements.
Formats for reporting and data set descriptions have been
developed, including a statement of errors.
Heating
Power
Test Room
Temperature
Solar Radiation
H1-2
H2-3
External
Temperature
H3-4
C3
C2
Heat Flux
H4-5
C4
C5
Fig. 4. Lumped parameter model of test cell and component showing data
inputs and the RC network.
Testing is now complete and the data have been checked,
documented and circulated for the cross-analysis exercise.
4.1. Opaque component
The available results for the opaque component are given
in Table 1.
Table 1
Available opaque wall results
4. Results
The choice of analysis method is open to each team, however identification software such as LORD, developed for the
PASLINK EEIG, is widely used and some teams have used
MATLAB. Such techniques are required to obtain the steady
state performance characteristics from the dynamic climate
and test cell data. LORD, for example, solves a user-defined
network of conductances and capacitances (analogous to
electrical RC-networks) with the measured data as input. An
example for an opaque wall and test cell is shown in Fig. 4.
Team providing data
A
B
C
D
E
F
G
H
I
Opaque wall U-values (W/m2 K)
Whole wall
Profile A
0.20
0.19
0.19
0.17
0.19
0.17
0.33
0.18
0.23
0.18
0.18
0.17
0.16
0.17
Not available
0.19
0.18
Not available
438
P. Baker / Energy and Buildings 36 (2004) 435–441
Table 2
Whole wall U-value results from cross-validation exercise on opaque wall
re-analyse the data, and a check will be made by a third
party.
Team
providing
data
Whole wall
U-value
(W/m2 K)
Team
performing
cross-validation
analysis
Whole wall
U-value
(W/m2 K)
4.2. The window component
A
B
C
E
G
0.20
0.19
0.19
0.19
0.33
G
H
B
A
F
0.24
0.19
0.19
0.19
0.34
Table 3
Centre of panel (profile A) U-value results from cross-validation exercise
on opaque wall
Team
providing
data
Profile A
U-value
(W/m2 K)
Team
performing
cross-validation
analysis
Profile A
U-value
(W/m2 K)
A
B
C
E
G
0.18
0.18
0.17
0.17
0.19
G
H
B
A
F
0.17
0.18
0.17
0.18
0.19
The results so far are under evaluation. The indications
are that there is more inter-site variability. A thorough
analysis of the environmental boundary conditions under
which the tests were carried out is in progress as part of
the cross-analysis exercise. For example, Fig. 5 shows the
theoretical variation of the U-value of double glazing with
wind speed due to the change in the external heat transfer
coefficient.
4.3. Training exercise with data from “unknown”
component
Data were provided by team B from tests carried out on:
• the whole opaque wall; and
• the opaque wall with the unknown component in the
1250 mm × 1500 mm aperture.
The results indicate that satisfactory agreement on the 1-D
centre of panel U-values have been achieved. The difference
between the whole wall and the centre of panel U-values
indicates the magnitude of the edge effects of the wall. The
error estimates for the whole wall and 1-D centre of panel
U-values are 13 and 6%, respectively.
The cross-analysis was performed as a blind exercise:
each cross-analysis team was provided with a full test report
and data set, but without the test results or the model used
for analysis, by the test team. The results are summarised
in Tables 2 and 3. Generally there is good agreement between the results of the data provider and the cross-analysis
team. Where the results differ significantly, both teams will
The aim of the exercise was to determine the thermal characteristics of the unknown component after first estimating
the properties of the opaque part of the wall from the first
part of the test with the whole opaque wall with the removable panel in place. Information regarding measurement errors was also given to the participants.
Figs. 6–10 show the result of each participating
team.
Good agreement between the teams was achieved for
the thermal transmission (UA-value) and solar aperture
(gA-value) of the unknown component:
• The UA range = 2.9–3.1 W/K; and
• The gA range = 0.65–0.68 m2 .
Excellent agreement was achieved for the UA-value of
the opaque wall with the unknown component, with a range
of values 3.97–4.04 W/K.
2.9
2.7
U-value [W/m2K]
2.5
2.3
2.1
1.9
1.7
1.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Wind speed [m/s]
Fig. 5. The theoretical variation of double glazing U-value with wind speed.
5
P. Baker / Energy and Buildings 36 (2004) 435–441
439
1.7
1.6
1.5
W/K
1.4
1.3
1.2
1.1
1
A
B
C
D
E
F
G
H
J
K
TEAM
Fig. 6. The identified thermal transmission (UA-value) of the whole opaque wall.
0.2
0.19
2
W/m K
0.18
0.17
0.16
0.15
A
B
C
D
E
F
G
H
J
K
TEAM
Fig. 7. The centre of panel U-value of the removable part of the opaque wall.
4.3
4.2
4.1
W/K
4
3.9
3.8
3.7
3.6
A
B
C
D
E
F
G
H
TEAM
Fig. 8. UA-value of opaque wall with component in aperture.
J
K
440
P. Baker / Energy and Buildings 36 (2004) 435–441
3.4
3.3
3.2
W/K
3.1
3
2.9
2.8
2.7
2.6
A
B
C
D
E
F
G
H
J
K
H
J
K
TEAM
Fig. 9. The UA-value of the unknown component.
0.72
0.7
m
2
0.68
0.66
0.64
0.62
0.6
A
B
C
D
E
F
G
TEAM
Fig. 10. The solar aperture (gA-value) of the unknown component.
Overall, the largest contribution to variation in the estimates of the UA-value of the component is due to the identification of the whole wall UA-value of the opaque wall
(range 1.24–1.41 W/K).
Table 4 summarises the results on the unknown component. The data were supplied by BRE from tests on
roof-light components typically used in industrial roofing
systems, carried out as part of a UK Government research
programme. Although the U-values of such systems can
be calculated using KOBRA, TRISCO, etc. little is known
about the solar transmittance. With changes in UK Building
Regulations the g-value has become important, in terms
of restriction in the allowable area of roof-lights to prevent summertime overheating. The tests were performed
to provide validation information for U-value calculations
and provide real data on solar apertures. The component
is a triple skin GRP roof-light for a composite cladding
system (Fig. 11). The calculated U-value is ∼1.6 W/m2 K,
i.e. the UA-value for a 1250 mm × 1500 mm sample is
∼3.0 W/K.
Table 4
Summary of results on unknown component
UA-value opaque wall
U-value Profile A
UA-value opaque wall
with component in
aperture
UA-value of component
gA-value of component
Average value
Error
Error
(%)
1.32 (W/K)
0.18 (W/m2 K)
3.99 (W/K)
0.18 (W/K)
0.01 (W/m2 K)
0.25 (W/K)
14
8
6
3.01 (W/K)
0.66 (m2 )
0.28 (W/K)
0.04 (m2 )
9
6
P. Baker / Energy and Buildings 36 (2004) 435–441
441
Fig. 11. The triple skin GRP rooflight component mounted in the aperture of the opaque wall at BRE Scotland.
5. Conclusions
Acknowledgements
• The results of the round robin tests and the cross-analysis
exercises show that a high standard of testing and analysis
has been carried out across the Thematic Network.
• Workable standard procedures are in place.
• Feedback from the participants indicates that improvements are needed with respect to:
The author acknowledges the support of the European
Commission in funding the IQ-test Project.
• Choice of identification models;
• Error analysis procedures; and
• Test procedures to de-correlate gA and UA-values, for
example using a shading screen to give initial estimates
of UA.
[1] H.A.L. van Dijk, F.M. Tellez, Measurement and data analysis
procedures, Final Report of the JOULE II COMPASS Project
(JOU2-CT92-0216), 1999.
Reference
IQ-test—improving quality in testing and evaluation of solar
and thermal characteristics of building components
Paul Baker∗
BRE Scotland, East Kilbride G75 ORZ, UK
Abstract
IQ-test is a Thematic Network supported by the European Community under the EESD Programme. The objective of IQ-test is to
further the development of common quality procedures at the PASLINK test cell facilities in 12 European countries, for the assessment
of the thermal characteristics of building components. This should consolidate the network, integrate the newer test sites and strengthen
its common approach of support for new product developments in the field of innovative building components. Round robin tests are
underway to assess both the inter-site quality of testing and analytical procedures of the participants. Two components were designed:
(1) an opaque, well insulated, homogeneous panel and (2) a window, which is used to replace the central section of the first component.
Common test and quality procedures have been implemented at each test site. The data sets generated by each team have been made
available for cross-analysis by another team. The results available so far on the first component indicate good agreement between sites.
This paper summarises the progress to date. Results are also presented from a training exercise which asked participants to identify the
performance characteristics of an unknown component without providing any physical description of the component.
© 2004 Published by Elsevier B.V.
Keywords: IQ-test; PASLINK; Test cells; Thermal performance
1. Introduction
IQ-test is a Thematic Network supported by the European Community under the EESD Programme, which aims
to consolidate the work of the network of the PASLINK
outdoor test cell facilities in 12 European countries, involved in the energy performance evaluation of the thermal and solar properties of building component. More
information may be found on the PASLINK web-site:
WWW.PASLINK.ORG. The objective of IQ-test is to
further the development of common quality procedures
for:
•
•
•
•
testing;
calibration;
data gathering, processing and analysis;
interpretation of test results and scaling/replication to real
building; and
• maintenance of the test infrastructure at the test sites.
This should consolidate the network, by integrating the
newer test sites and strengthening its common approach of
support for new product developments in the field of inno∗ Current address: Centre for Research on Indoor Climate & Health,
Glasgow Caledonian University, G4 OBA, UK.
E-mail address: [email protected] (P. Baker).
0378-7788/$ – see front matter © 2004 Published by Elsevier B.V.
doi:10.1016/j.enbuild.2004.01.046
vative building components through semi-standardised tests
and pragmatic, practicable and affordable but accurate procedures.
As part of the work of the Network, round robin tests
are being performed as part of a feasibility study for standardisation activities. The objective is to assess both the
inter-site quality of testing and analytical procedures of the
members, with a view to developing standards for outdoor
testing. High quality data for model calibration will also be
generated.
Two components were designed, incorporating flexibility,
in order that the first component could be used as a platform
for the second component. The first component is an opaque,
homogeneous with a removable central section. The thermal
properties of the panel are very well defined. The second
component is a window, which is used to replace the central
section of the first component.
The objective of testing the first component is for each
participant to determine the thermal transmission coefficient
of the panel by both 1-D heat flux measurements and an
energy balance on the test cell. For the second component:
the whole wall U-value and solar aperture estimated from
the test cell energy balance and the window U-value and
solar aperture.
Common test procedures have been designed and quality
procedures have been implemented at each test site. The
436
P. Baker / Energy and Buildings 36 (2004) 435–441
test work is completed and the data sets generated by each
team are being made available for cross-analysis by another
team.
This paper describes the components and procedures and
summarises the results to date, including cross-analysis by
other teams on the opaque component.
The results of a training exercise are also presented. Data
were provided to participants from a test carried out at one
of the test sites with another component in the aperture of
the opaque wall. The aim was to determine the thermal characteristics of the component, without providing the participants with a physical description or any clues to the type of
the “unknown” component.
2. The test components
Given the number of organisations involved in the Thematic Network it was decided at the beginning that it was
impractical to circulate one component for testing for the
following reasons:
• variation in the test aperture size of test cells between
sites;
• high transportation costs; and
• likely difficulties in keeping to a strict timetable, given
the use of the test cells for other tests.
The approach adopted was for each organisation to construct its own component(s) according to strict instructions
regarding the selection of materials, manufacture and instrumentation.
2.1. Component 1—the opaque wall
The first component is an opaque, homogeneous panel
consisting of a sandwich of insulation between plywood,
with a replacable central section 1500 mm (h) by 1250 mm
(w). The thermal properties of the panel are well defined
and the required materials are available in the locality of
each participant. Expanded polystyrene (PS30), with a density of 30 kg/m3 and a nominal thermal conductivity of
0.033 W/mK, is used to form an insulating panel of thickness
200 mm. A white exterior finish is used on the plywood.
It was agreed that each team should measure the heat flux
and temperatures in two profiles through the wall (Fig. 1),
with one in the centre of the removable section (A) and
the other mid-way between the edge of the wall and the
removable panel (B).
The objective of testing the first component is for each
participant to determine the thermal transmission coefficient
(U-value) of the panel by both:
1. 1-D heat flux measurements through profiles A and B;
and
2. an energy balance on the test cell.
Heat flux sensors
flush mounted in
plywood
B
A
Removable
central panel
inside
Fig. 1. Measurement profiles through the opaque wall.
The latter value will include edge effects, which will vary
depending on the test cell and installation of the test wall at
each site.
2.2. Component 2—the window
The objective of the second component is to introduce a
greater degree of complexity by using a window to replace
the central section of the first component. The window design incorporates double glazing using ordinary clear float
glass in a timber frame. The frames for each participant were
produced centrally. Each participant obtained glass from a
local supplier and samples were tested centrally to ensure
consistency. The spectrophotometric tests have shown that
there are differences between the samples in their infrared
transmittance, which may give variations in the g-value (solar energy transmittance) of the double glazing between 73.2
and 77.3%. Fig. 2 shows the window installed in the opaque
surround.
The objectives of the test are to determine:
1. the whole wall U-value and g-value estimated from the
test cell energy balance; and
2. the window U-value and g-value (frame and glazing).
Fig. 2. The window installed in the opaque surround at BRE Scotland.
P. Baker / Energy and Buildings 36 (2004) 435–441
437
300
HIGH
POWER
Power Level [W]
250
ROLBS
200
150
100
50
LOW POWER
0
0
1
2
3
4
5
6
7
8
9
Day
Fig. 3. A typical heating power sequence with low power = 50 W air circulation fan, high power = 50 W fan power + 200 W resistance heater.
3. The test procedures
The procedures are based on the COMPASS Measurement and Data Analysis Procedures [1]. A test sequence was
devised to reduce the overall test duration, whilst maximising information. It includes a heating/cooling regime using
a Randomly Ordered Logarithmically distributed Binary Sequence (ROLBS). The maximum power level was calculated
to ensure that the mean test room temperature difference
between the low and high power parts of the test sequence
should be at least 10 K, but preferably 20 K, without exceeding the safe operational limits of the test cell. The choice of
whether to use heating or cooling depends on the local climatic conditions. In a “heating” climate a maximum power
level of 250 W is satisfactory for the window test: the heating sequence is shown in Fig. 3.
The air leakage of the test room is also determined by
pressurisation testing before and after the test to ensure that
the test cell meets a requirement of 0.5 air changes per hour
at 50 Pa. It is also recommended that the air leakage is monitored continuously during the test by tracer gas measurements.
Formats for reporting and data set descriptions have been
developed, including a statement of errors.
Heating
Power
Test Room
Temperature
Solar Radiation
H1-2
H2-3
External
Temperature
H3-4
C3
C2
Heat Flux
H4-5
C4
C5
Fig. 4. Lumped parameter model of test cell and component showing data
inputs and the RC network.
Testing is now complete and the data have been checked,
documented and circulated for the cross-analysis exercise.
4.1. Opaque component
The available results for the opaque component are given
in Table 1.
Table 1
Available opaque wall results
4. Results
The choice of analysis method is open to each team, however identification software such as LORD, developed for the
PASLINK EEIG, is widely used and some teams have used
MATLAB. Such techniques are required to obtain the steady
state performance characteristics from the dynamic climate
and test cell data. LORD, for example, solves a user-defined
network of conductances and capacitances (analogous to
electrical RC-networks) with the measured data as input. An
example for an opaque wall and test cell is shown in Fig. 4.
Team providing data
A
B
C
D
E
F
G
H
I
Opaque wall U-values (W/m2 K)
Whole wall
Profile A
0.20
0.19
0.19
0.17
0.19
0.17
0.33
0.18
0.23
0.18
0.18
0.17
0.16
0.17
Not available
0.19
0.18
Not available
438
P. Baker / Energy and Buildings 36 (2004) 435–441
Table 2
Whole wall U-value results from cross-validation exercise on opaque wall
re-analyse the data, and a check will be made by a third
party.
Team
providing
data
Whole wall
U-value
(W/m2 K)
Team
performing
cross-validation
analysis
Whole wall
U-value
(W/m2 K)
4.2. The window component
A
B
C
E
G
0.20
0.19
0.19
0.19
0.33
G
H
B
A
F
0.24
0.19
0.19
0.19
0.34
Table 3
Centre of panel (profile A) U-value results from cross-validation exercise
on opaque wall
Team
providing
data
Profile A
U-value
(W/m2 K)
Team
performing
cross-validation
analysis
Profile A
U-value
(W/m2 K)
A
B
C
E
G
0.18
0.18
0.17
0.17
0.19
G
H
B
A
F
0.17
0.18
0.17
0.18
0.19
The results so far are under evaluation. The indications
are that there is more inter-site variability. A thorough
analysis of the environmental boundary conditions under
which the tests were carried out is in progress as part of
the cross-analysis exercise. For example, Fig. 5 shows the
theoretical variation of the U-value of double glazing with
wind speed due to the change in the external heat transfer
coefficient.
4.3. Training exercise with data from “unknown”
component
Data were provided by team B from tests carried out on:
• the whole opaque wall; and
• the opaque wall with the unknown component in the
1250 mm × 1500 mm aperture.
The results indicate that satisfactory agreement on the 1-D
centre of panel U-values have been achieved. The difference
between the whole wall and the centre of panel U-values
indicates the magnitude of the edge effects of the wall. The
error estimates for the whole wall and 1-D centre of panel
U-values are 13 and 6%, respectively.
The cross-analysis was performed as a blind exercise:
each cross-analysis team was provided with a full test report
and data set, but without the test results or the model used
for analysis, by the test team. The results are summarised
in Tables 2 and 3. Generally there is good agreement between the results of the data provider and the cross-analysis
team. Where the results differ significantly, both teams will
The aim of the exercise was to determine the thermal characteristics of the unknown component after first estimating
the properties of the opaque part of the wall from the first
part of the test with the whole opaque wall with the removable panel in place. Information regarding measurement errors was also given to the participants.
Figs. 6–10 show the result of each participating
team.
Good agreement between the teams was achieved for
the thermal transmission (UA-value) and solar aperture
(gA-value) of the unknown component:
• The UA range = 2.9–3.1 W/K; and
• The gA range = 0.65–0.68 m2 .
Excellent agreement was achieved for the UA-value of
the opaque wall with the unknown component, with a range
of values 3.97–4.04 W/K.
2.9
2.7
U-value [W/m2K]
2.5
2.3
2.1
1.9
1.7
1.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Wind speed [m/s]
Fig. 5. The theoretical variation of double glazing U-value with wind speed.
5
P. Baker / Energy and Buildings 36 (2004) 435–441
439
1.7
1.6
1.5
W/K
1.4
1.3
1.2
1.1
1
A
B
C
D
E
F
G
H
J
K
TEAM
Fig. 6. The identified thermal transmission (UA-value) of the whole opaque wall.
0.2
0.19
2
W/m K
0.18
0.17
0.16
0.15
A
B
C
D
E
F
G
H
J
K
TEAM
Fig. 7. The centre of panel U-value of the removable part of the opaque wall.
4.3
4.2
4.1
W/K
4
3.9
3.8
3.7
3.6
A
B
C
D
E
F
G
H
TEAM
Fig. 8. UA-value of opaque wall with component in aperture.
J
K
440
P. Baker / Energy and Buildings 36 (2004) 435–441
3.4
3.3
3.2
W/K
3.1
3
2.9
2.8
2.7
2.6
A
B
C
D
E
F
G
H
J
K
H
J
K
TEAM
Fig. 9. The UA-value of the unknown component.
0.72
0.7
m
2
0.68
0.66
0.64
0.62
0.6
A
B
C
D
E
F
G
TEAM
Fig. 10. The solar aperture (gA-value) of the unknown component.
Overall, the largest contribution to variation in the estimates of the UA-value of the component is due to the identification of the whole wall UA-value of the opaque wall
(range 1.24–1.41 W/K).
Table 4 summarises the results on the unknown component. The data were supplied by BRE from tests on
roof-light components typically used in industrial roofing
systems, carried out as part of a UK Government research
programme. Although the U-values of such systems can
be calculated using KOBRA, TRISCO, etc. little is known
about the solar transmittance. With changes in UK Building
Regulations the g-value has become important, in terms
of restriction in the allowable area of roof-lights to prevent summertime overheating. The tests were performed
to provide validation information for U-value calculations
and provide real data on solar apertures. The component
is a triple skin GRP roof-light for a composite cladding
system (Fig. 11). The calculated U-value is ∼1.6 W/m2 K,
i.e. the UA-value for a 1250 mm × 1500 mm sample is
∼3.0 W/K.
Table 4
Summary of results on unknown component
UA-value opaque wall
U-value Profile A
UA-value opaque wall
with component in
aperture
UA-value of component
gA-value of component
Average value
Error
Error
(%)
1.32 (W/K)
0.18 (W/m2 K)
3.99 (W/K)
0.18 (W/K)
0.01 (W/m2 K)
0.25 (W/K)
14
8
6
3.01 (W/K)
0.66 (m2 )
0.28 (W/K)
0.04 (m2 )
9
6
P. Baker / Energy and Buildings 36 (2004) 435–441
441
Fig. 11. The triple skin GRP rooflight component mounted in the aperture of the opaque wall at BRE Scotland.
5. Conclusions
Acknowledgements
• The results of the round robin tests and the cross-analysis
exercises show that a high standard of testing and analysis
has been carried out across the Thematic Network.
• Workable standard procedures are in place.
• Feedback from the participants indicates that improvements are needed with respect to:
The author acknowledges the support of the European
Commission in funding the IQ-test Project.
• Choice of identification models;
• Error analysis procedures; and
• Test procedures to de-correlate gA and UA-values, for
example using a shading screen to give initial estimates
of UA.
[1] H.A.L. van Dijk, F.M. Tellez, Measurement and data analysis
procedures, Final Report of the JOULE II COMPASS Project
(JOU2-CT92-0216), 1999.
Reference
