Вы находитесь на странице: 1из 11

A simple model for assessing the energy performance of windows

J. Karlsson
*
, B. Karlsson, A. Roos
Department of Materials Science, The A

ngstrom laboratory, Uppsala University, P.O. Box 534, S-751 21 Uppsala, Sweden
Received 21 December 1999; accepted 6 November 2000
Abstract
A simple model for the annual energy balance of the window taking solar radiation and heat losses into consideration has been further
developed and analysed. Hourly meteorological data for the solar irradiation and the outside temperature are used together with the optical
and thermal performance of the window to evaluate the net energy heat ow through a window. The model renders a very simple way to
compare different advanced windows in different geographical locations, orientations and buildings using basically only the balance
temperature as building input. The energy balance and the cost efciency for several glazing combinations are evaluated for buildings with
different balance temperatures in a typical mid-Swedish climate. This model has a potential to be used for energy rating of windows.
# 2001 Elsevier Science B.V. All rights reserved.
Keywords: Windows; Energy efciency; Energy rating; Residential buildings; Climate; Simulation; Solar transmittance
1. Introduction
Several types of advanced windows exist in the market; in
northern climates, mostly ``low-emitting'' (low-e) windows
that are used in order to reduce the thermal losses but also
``solar control'' windows that are used in order to reduce the
cooling load. Alow-e windowis a windowcoated with a thin
lm layer that exhibits low thermal emittance and high solar
transmittance. A solar control window is a window coated
with a thin lm that exhibits low solar transmittance [1,2].
Opinions about which window to use in which place differs,
which is reected in the market of coated insulated glass
units, IGUs (a sealed unit of glass sheets without frame). In
some countries it is basically standard to install coated IGUs
in new production, and in some other countries these
advanced glazings are only a small fraction of the total
amount of installed windows in new production.
In order to assess the energy performance and cost
efciency of windows, it is necessary to have reliable and
handy models. Various computer simulation programs are
available, where the energy balance of a building is eval-
uated, e.g. DEROB-LTH [3], BRIS [4], DOE-2.1E, CHE-
NATH, HELIOS, TRNSYS and TSBI3 [5]. Another
approach is the optimising models [6], where the life cycle
cost of the building energy system is minimised. The
problem with these advanced programs is that one has to
have access to detailed information about the building, such
as size, internal power consumption and occupancy, etc.
Within the International Energy Agency, Solar Heating and
Cooling Programme, Task 18 [5], a need for models that can
be standardised and used for an international window rating
system was expressed. One way of doing this is to make a
simplied front-end for an existing building simulation
model where different glazing options can be evaluated,
e.g. RESFEN 3.1 based on DOE-2.1E [7]. Some models
based on simplied building data have been proposed (e.g.
[8,9]). Roos and Karlsson presented a model 1994 [10] that
uses only one parameter as input from the building, the
balance temperature. The balance temperature determines if
the solar irradiation and thermal heat ow is useful or not for
the energy system of the building. This model is sometimes
referred to as ``Karlsson's model'' [11] in Sweden after the
second co-author of this paper.
The purpose of this work is to further develop and analyse
the Karlsson model and use it to assess which windows are
energy efcient and/or cost effective to install in a typical
mid-Swedish climate. Furthermore, we want to present this
model as an alternative model for energy rating of windows,
taking into account different types of buildings as opposed to
some previously proposed models [12,13]. Apart from the
details of the model, selected simulation results from a
number of different windows are presented. In Section 2,
some previous features of the presented model are
addressed. Section 3 describes the energy balance model
Energy and Buildings 33 (2001) 641651
*
Corresponding author. Present address: Vattenfall Utveckling AB,
A

lvkarleby, Sweden. Tel.: 46-18-4713134; fax: 46-18-500131.


E-mail address: joakim.karlsson@angstrom.uu.se (J. Karlsson).
0378-7788/01/$ see front matter # 2001 Elsevier Science B.V. All rights reserved.
PII: S0 3 7 8 - 7 7 8 8 ( 0 0 ) 0 0 1 3 1 - 6
in detail and briey discusses the angle-dependent optical
properties of windows. Section 4 presents the results of
energy performance calculations and also a simple econom-
ical analysis. Furthermore, a sensitivity analysis is per-
formed in this section. Finally, Section 5 concludes the
paper with a discussion. A parametric study on windows
in an ofce module in a Swedish climate has previously been
presented by Bulow-Hube [14].
2. Previous model in brief
The previous (Karlsson) model uses the accumulated
irradiation data on vertical surfaces, S, and degree-hours,
G, versus temperature interval as plotted in Fig. 1a and b.
This is also used in the Norwegian SOLVOR-model [15]. A
reference year, TRY, for Stockholm (latitude: 59.11; long-
itude: 17.59) based on SMHI (Swedish Meteorological and
Hydrological Institute) measurements 19831992 was used
in the simulations. It should be pointed out that the Swedish
climate is quite unique with high temperatures considering
the high latitude and the low annual solar radiation. The
horizontal irradiation data are converted to data for a surface
randomly oriented in space using the Hay and Davies model
[16], with a mean albedo set to 25%. It is seen that for this
climate, the south-oriented surface gets the highest solar
irradiation of about 850 kW h/m
2
per year. Fig. 1b illustrates
the accumulated number of degree-hours, G, up to a tem-
perature T
0
, where G is dened as
G(T
0
) =
X
T
0
T=T
min
H(T)(T
eq
T) (1)
T is the temperature, H the number of hours that the
temperature stays at T, and T
min
the minimum temperature
during the year. T
eq
is the indoor temperature, assumed
constant at 208C. It is seen that the total annual number
of degree-hours is about 1158C h for this year.
When these two curves are produced from the weather
data, the procedure in the previous model [10,11] is to take
the balance temperature of the building, T
b
, (see Section 3)
and perform two simple equations to get the total energy
transport through the window. One of these equations, the
heating energy balance is given as
E = gS(T
b
) UG(T
b
) (2)
where g is the annual mean value of the total solar energy
transmittance and U, the U-value of the window. E, g and S
are bearing dependent. A similar equation for the cooling
energy balance can be found in [10]. This is a very simple
way of getting the energy balance of different windows in
different buildings, including solar energy. In a cold climate
as in Sweden, energy consultants and windowmanufacturers
often compare the energy efficiency of the windows by
calculating the UG-value of the windows. The Karlsson
formalism improves this by taking solar irradiation into
account which can be very important, especially in buildings
with high cooling needs. The drawback with the Karlsson
approach is that the S- and G-curves and g, has to be
prepared for each location and orientation of the window.
Furthermore, this limits the possibilities to account for a
thermal inertia of the building. These drawbacks can be
addressed by making a similar model but on an hourly basis
as outlined in Section 3.
3. Present model description
3.1. Window properties
The thermal ow through the window is still described by
the U-value of the window. When it comes to the total solar
energy transmittance, g, it is necessary to account for its
angle dependence. In this model, we use an empirical model
as proposed in [17]. In this model, the windows need to be
categorised in order to get the correct angle dependence. By
knowing the g-value at normal incidence, g(0), the category
of the window (coating category), q, and the number of
panes in the window, p, the angle dependence of g is known
within 3% maximum error at every angle. Fig. 2 illustrates
the extremes in angle dependence according to this model.
Fig. 1. (a) Accumulated total solar irradiation, S, on vertical surfaces in
four different orientations vs. temperature interval. (b) Accumulated
number of degree-hours vs. temperature interval.
642 J. Karlsson et al. / Energy and Buildings 33 (2001) 641651
There is as much as about 30% difference in transmittance
between the extreme cases (in percentage points on normal-
ised g-value curves) at high angles of incidence. By coin-
cidence, the maximum difference occurs for common angles
of incidence of solar irradiation against the window. Except
from the well-known parameters U, g(0), and p, the coating
category, q, of the window is needed when using this model.
3.2. Energy balance equation
The energy ow through the window can be divided into
thermal heat ow from the hot side to the cool side of the
window, and solar irradiation from the outside to the inside.
What needs to be known is when these heat ows are useful
or not for the purpose of the energy system of the building.
The parameter that is used to assess this is the balance
temperature, T
b
. The balance temperature is, in this model,
dened as an annual average outside temperature above
which the building does not need to be actively heated. This
characteristic parameter for the building differs depending
on the insulation of the building, activity inside, lighting, etc.
When the outside temperature equals the balance tempera-
ture, the internal sources are balanced by the heat losses
through the walls and windows. Active cooling will nor-
mally not take place as soon as the temperature exceeds T
b
,
but at a temperature of T
b
T
dyn
, where the dynamic
temperature swing, T
dyn
, may be about 58C. With these
denitions, we can now describe when solar irradiation is
positive or negative for the energy system, meaning reduce
or increase heating and cooling load, respectively (Table 1,
column 2). The ``usefulness'' of the thermal ow can in the
same way be assessed by making the assumption that the
indoor temperature, T
eq
, is constant at 208C. The thermal
ow will now affect the building energy system as described
in Table 1, column 3. If there is no active cooling system in
the building, only the conditions in the rst row of Table 1
(where T < T
b
) are relevant for energy or money saving
assessment.
In the previous model, the accumulated curves of irradia-
tion were plotted versus the temperature for every hour
(Fig. 1). Allowing the time interval above which the outside
temperature is averaged to change, we introduce some
dynamic features to the building. If the outside temperature
is averaged over a period of, for instance, 12 h it will give the
building some inertia in the form of a virtual memory of 1
day. This means that if the sun shines a lot during the middle
of the day and the outdoor temperature during some hours is
above T
b
(the grey part in Fig. 3), this solar energy will still
be useful for the building since the daily mean temperature
(T
t
in Fig. 3) is below T
b
.
The temperature that is compared with the balance tem-
perature will thus be a mean temperature over a time
interval, t, that in some way is related to the time constant
of the building. This mean temperature will be referred to as
T
t
from now on, and t will be referred to as the time interval.
This time interval is not the same as the time constant, as it is
normally dened in building simulations. The denitions of
T
b
and T
t
are further discussed in Section 4.2.
The direct and the diffuse horizontal solar radiation are
converted to direct, isotropic diffuse, ground reected and
circumsolar radiation impinging on the window [16]. The
direct radiation hits the window with a specied angle of
incidence, y
h
, dependingongeographical location, orientation
Fig. 2. Normalised total solar energy transmittance, g, for different
category parameters, q and number of panes, p. The q spans between 1 and
10 as described in [17]. The q = 1 corresponds to a strongly absorbing
window or coating, q = 10 corresponds to a window with high refractive
index coating and q = 4 represents normal float glass.
Table 1
Solar irradiation and thermal flow influence on the building energy system depending on the outside temperature as assessed by the balance temperature
Outside temperature Solar irradiation influence on energy system Heat flow influence on energy system
T < T
b
Positive, reduce heating load Negative, increase heating load
T
b
< T < T
b
T
dyn
Indifferent Indifferent
T
b
T
dyn
< T < T
eq
Negative, increase cooling load Positive, reduce cooling load
T > T
eq
Negative, increase cooling load Negative, increase cooling load
Fig. 3. Solar energy inflow during 59 h in this example is still considered
useful for the building when T is averaged over t hours, even though the
temperature at these hours exceeds the balance temperature (grey part).
J. Karlsson et al. / Energy and Buildings 33 (2001) 641651 643
of the window normal and the hour of day, h. For the diffuse
radiation, it is more difcult to dene a certain angle of
incidence, since it consists of multiply scattered and
reected radiation. However, an effective incidence angle
can be dened after integration over all angles, one for
ground reected radiation, y
er
, and one for isotropic diffuse
radiation, y
ei
[16]. By translating the assumptions in Table 1,
the annual energy transport through the window for tem-
peratures up to T
b
can now be expressed mathematically as
where g(y
h
), g(y
ei
) and g(y
er
) represent the total solar energy
transmittance of the window at angle y
h
, y
ei
and y
er
, respec-
tively. I
dir
(h), I
diff
(h) and I
refl
(h) are the direct, isotropic
diffuse and ground reflected solar radiation impinging on the
window at hour h. I
dir
(h) also contains the circumsolar part
of the solar radiation. E will be referred to as the heating
energy balance. E
conv
is a conversion factor depending on
what kind of energy system the building has. If, for instance,
a gas furnace is used, E
conv
should be equal to the annual fuel
utilisation efficiency (AFUE) of that furnace, which is in the
order of 0.51 [7]. This gives how much energy that needs to
be bought to replace the lost heat. If E becomes positive,
which is often the case for south-facing windows at Nordic
latitudes, this means that the window gains more energy in
terms of solar irradiation than it loses thermally over the
year. If a window has a high solar transmittance giving a
contribution to the heating during the cold season, this can
also lead to overheating during the warm season. The model
can also be used to assess how much different windows can
reduce the cooling load of the building. At temperatures
above T
b
T
dyn
, all solar irradiation is negative for the
energy system as seen in Table 1, and this energy has to be
cooled away or it will give rise to an increased indoor
temperature. The same procedure as above with the energy
flow for temperatures above T
b
T
dyn
gives the total annual
cooling load for the window
where P
conv
is a conversion factor similar to E
conv
. If an air
conditioner is used, P
conv
will be the coefficient of perfor-
mance (COP) of this refrigeration appliance, normally of the
order of 23. If the purpose is to compare windows, as for
example in an energy rating system, the factors E
conv
and
P
conv
should equal to 1. P will be referred to as the cooling
energy balance. Consequently, E represents the amount of
heat we gain or lose because of the windows and P represents
how much we have to cool because of the windows, all in
kW h/m
2
per year. If the building utilises an active cooling
system, the total annual energy balance for the windows can
thus be assessed by
E
tot
= E P (5)
Apart from the dependence of T
b
, T
dyn
, T
eq
and t, the
parameters E, P and E
tot
will of course also depend strongly
on weather data, i.e. geographical location and orientation.
In order to be able to compare the results, a base-case is
defined to be a double-glazed unit without coatings and gas
fillings. E, P and E
tot
for the base-case are now subtracted
from the tested glazings to see how much energy is saved
upon a change of window. The subtraction yields the para-
meters E
saved
, P
saved
and E
totsaved
, which thus give number
of saved kilowatt-hours per square meter window area and
year compared to the base-case. The whole procedure is
summarised in Fig. 4.
4. Results and related remarks
4.1. Energy balance and saved energy
A test set of 17 different windows with or without gas
lling, 2 ordinary-, 9 low-e- and 6 solar-control windows
were identied. Counting the gas lled units, we thus have
E =
P
T
t
(h)<T
b
[g(y
h
)I
dir
(h) g(y
ei
)I
diff
(h) g(y
er
)I
refl
(h) U(T
eq
T
t
(h))[
E
conv
(3)
P =
P
T
t
(h)>T
b
T
dyn
[g(y
h
)I
dir
(h) g(y
ei
)I
diff
(h) g(y
er
)I
refl
(h) U(T
eq
T
t
(h))[
P
conv
(4)
Fig. 4. Flow chart of the basic steps in the model. I
dirh
and I
diffh
are the direct- and diffuse-horizontal radiation, respectively.
644 J. Karlsson et al. / Energy and Buildings 33 (2001) 641651
34 different double- and triple-glazed units (from now on
called DGU and TGU) to investigate. The provided U- and
g-values are calculated according to the standards EN 673
and ISO 9050, and taken for the glazing part of the windows,
not including the frame. This means that this becomes a
comparison between different IGUs rather than between
whole windows. However, this can easily be generalised
to whole windows, since the U- and g-values are inputs to the
model. The frame is undoubtedly a part of the window from
a construction point of view, but from a physical point of
view it can equally be regarded as part of the wall (but
having much higher U-value) since it is not directly trans-
parent. All calculations are performed per square meter
glazed area. Glazing size effects can be accounted for by
changing T
b
and t, since a different glazed area changes
these parameters (see Section 4.2.1). In Figs. 5 and 6, the
total energy balance, E
tot
, for some of the investigated low-e
alternatives are displayed for eight different bearings scan-
ning in 458 steps: north, north-east, east, etc. with T
b
= 13

C
and t = 12 h. T
dyn
and T
eq
are always set to 5 and 208C,
respectively, throughout this section and all the results are
produced with E
conv
= P
conv
= 1 and the Stockholm, TRY
weather data as mentioned in Section 2. It should be pointed
out that the large number of curves in the graphs has been
chosen to clearly illustrate the trends. It is in some cases hard
to identify a specic windowamong all the graphs, but this is
not the issue here.
Obviously, in this climate the south-facing windows have
the best energy performance and many of the south-facing
windows gain more energy than they loose on a yearly basis.
It is seen that the uncoated DGUs have considerably lower
energy balance than their coated counterparts. Furthermore,
All the coated low-e DGUs have a better energy perfor-
mance than the uncoated TGU. The total energy gained from
changing from uncoated to low-e windows (E
totsaved
) in a
building with T
b
= 13

Cand t = 12 h is illustrated in Fig. 7.


The alternatives in Fig. 7 have considerably lower
U-values than the base-case, which lead to a high value
of E
saved
for northerly-oriented windows. However, these
alternatives also have lower transmittance than the base-case
which means that southerly-oriented windows will not save
as much as the northern ones. For a building with this
balance temperature and for this climate, E
totsaved
will
basically follow the trends of E
saved
since the cooling
Fig. 5. Total energy balance for some low-e windows in a Stockholm climate at T
b
= 13

C and t = 12 h. Results for uncoated DGU and TGU are displayed


for reference (bold letters in the legend). The y-axis gives the total energy balance for the windows and the x-axis gives the orientation of the window. The
column to the right gives the U- and g-values (in W/m
2
K and %, respectively) for the different glazings.
Fig. 6. Total energy balance for solar control windows in a Stockholm climate at T
b
= 13

C and t = 12 h. The results for uncoated DGU are displayed for


reference (bold).
J. Karlsson et al. / Energy and Buildings 33 (2001) 641651 645
demand is negligible. The window that saves the most
energy, in this case, is a TGU with two tin-oxide coated
panes and argon gas lling (U = 0:9 W/m
2
K and g = 56%)
for basically all window orientations. This alternative will
save about 200 kW h/m
2
(glazed area) per year in a northerly
orientation, and about 150 kW h/m
2
per year in a southerly
orientation compared to the base-case (not counting the
cooling it will save about 180 and 140 kW h/m
2
per year,
respectively). None of the solar control windows outper-
formed their low-e alternatives at T
b
= 13

C.
For a building with a low balance temperature, the cooling
effects will play a more important role than the above case,
as seen in Figs. 8 and 9 where P
saved
and E
totsaved
, respectively,
are plotted for a building with T
b
= 0

C and t = 12 h. The
Fig. 7. E
totsaved
for some low-e windows. T
b
= 13

C, t = 12 h. The y-axis gives saved kW h/m


2
per year compared to the base-case. Base-case is an uncoated
DGU. The uncoated TGU is displayed (bold) for reference.
Fig. 8. P
saved
for some low-e windows. T
b
= 0

C, t = 12 h.
Fig. 9. E
totsaved
for some low-e windows. T
b
= 0

C, t = 12 h.
646 J. Karlsson et al. / Energy and Buildings 33 (2001) 641651
shape of the curves in Fig. 8 comes from the fact that the
lower levels of transmittance of these windows reduce the
cooling load compared to an uncoated window, especially for
southerly orientations. The peaks of the curves are shifted to
the west because the outside temperature is normally higher
in the afternoon than in the morning, so the solar energy is
less useful these hours and at these window orientations.
At this balance temperature (08C), the solar control
windows have passed their low-e counterparts in terms of
energy efciency as seen in Fig. 10. The winning alternative
for this balance temperature is a TGU with one pane coated
with a solar control coating and one pane coated with a low-e
coating and a gas lling (U = 0:8 W/m
2
K and g = 24%).
This window saves about 200 kW h/m
2
in a south-facing
orientation and about 120 kW h/m
2
in a north-facing orien-
tation under the given conditions. These results are all
prepared with E
conv
and P
conv
equal to one. If air-condition-
ing is used and P
conv
is set to about three in order to calculate
the amount of bought electricity, then P would be scaled
down by a factor of about three so that the relation between
E
saved
and P
saved
would change. What values to use for E
conv
and P
conv
depends on if the window energy analyses are
made froma national point of viewor from a building-owner
point of view and how the energy is produced.
4.2. Sensitivity analysis
Since the presented model is based on the concept of the
balance temperature, it is important that it is not extremely
sensitive to changes in T
b
. A small change in T
b
should not
give rise to totally different answers in the energy balance,
since the balance temperatures can only be assessed within
an approximate interval of about 28C. Another very
important aspect to be aware of is that when we change
the windows of the building envelop, the balance tempera-
ture of the building may also change. In other words, the
window that is the best at a certain balance temperature may
not be the best in a comparison after the change is made,
since the balance temperature is different. This has to be
analysed, and is especially important to consider if the
building has a lot of window area or if a radical change
is made, e.g. from single panes to coated DGUs. Sections
4.2.1 and 4.2.2 study the effect of a balance temperature and
a time interval change, respectively.
4.2.1. Balance temperature
Fig. 11a and b illustrates how E
tot
depends on T
b
for
three very different glazing examples, clear oat DGU
Fig. 10. E
totsaved
for solar control windows at T
b
= 0

C and t = 12 h.
Fig. 11. (a) E
tot
for an uncoated DGU (U = 2:9 W/m
2
K and g = 76%), a
low-e (U = 0:9 W/m
2
K and g = 56%) and a solar control window
(U = 1:6 W/m
2
K and g = 27%) vs. balance temperature for a south-
facing window. (b) E
tot
for clear, low-e and a solar control window vs.
balance temperature for a north-facing window.
J. Karlsson et al. / Energy and Buildings 33 (2001) 641651 647
(U = 2:9 W/m
2
K and g = 76%), low-e TGU (U = 0:9 W/
m
2
K and g = 56%) and solar control DGU (U = 1:6 W/
m
2
K and g = 27%), and for a south- and a north-facing
window, respectively. In Fig. 11a, it is seen that for a south-
facing window, in this climate, a low-e window is the most
efcient in residential type buildings with balance tempera-
tures higher than 58C. For commercial type of buildings with
lower balance temperatures, it is seen that a solar control
window is the most efcient choice. If chosen correctly
(meaning the proper window for the building in question),
they clearly outperform the uncoated DGU. Basically, the
same conclusion can be drawn for a north-facing window
(Fig. 11b). For convenience, T
b
at the interception between
the curves of the low-e and the solar control window will
now be referred to as T
x
. A surprising result may be the high
E
tot
for the uncoated DGU in the north orientation at the
lowest T
b
. However, it is necessary to bear in mind that a
building with a balance temperature of 108C means it
needs to be cooled even at sub-zero outside temperatures.
Thus, this building has an extremely high internal heat
production (and/or high insulation) and, therefore, a severe
cooling need. In this case a high U-value would help to
reduce the heat excess. Buildings with extremely low bal-
ance temperatures are generally not realistic, and it must be
remembered that a simple mathematical model like this may
give misleading results when the parameters reach ``unphy-
sical'' limits. In order to visualise the trends and to under-
stand the model, it is nevertheless useful to include these
extreme values in the graphs.When analysing the sensitivity
of the balance temperature, it is more convenient to look at
E
totsaved
for the windows as plotted in Fig. 12a and b. For a
south-facing window, it is seen that when a correct choice is
made (low-e T
b
> T
x
, and solar control for T
b
< T
x
) the
inclination of the curves is low. This means that the balance
temperature does not need to be known exactly to get the
correct E
totsaved
. To be on the safe side T
x
should be shifted
some degrees to higher temperatures, since an installation of
a more advanced window will mostly lower the balance
temperature of the building. Consequently, when a building
has a balance temperature that exceeds T
x
by a small value
(that depends on the specic case but it should be of the
order of 158C), the low-e alternative should be used and a
solar control window otherwise. From Fig. 12, we can thus
deduce that low-e windows should be used for buildings with
balance temperatures above about 788C and solar control
windows for buildings with lower balance temperatures.
For a north-facing window, the inclination of the E
totsaved
curve is somewhat higher below T
x
than for a south-facing
window, indicating that for a window which receives less
radiation it is more important to have an accurate T
b
in order
to get the right E
totsaved
. It must be emphasised once more
that all this is valid for this specic climate. As an example
of this, E
totsaved
for the same windows as above but in a
Miami climate are illustrated in Fig. 13. Here, there is no
need for low-e windows in any type of building, but a solar
control window saves a lot of energy compared to the other
glazings. From a pure heating and cooling point of view, the
lower the g-value the better the solar control window per-
forms compared to an uncoated DGU. T
x
in this climate is
above 208C, which again means that only solar control
windows are of interest for this location. The sensitivity
of the balance temperature may well be much higher for
other climates, so a simple sensitivity analysis (varying T
b
and t) is recommended for whatever climate the model is
used in.
Fig. 12. (a) E
totsaved
vs. balance temperature for low-e and solar control
window in a south-facing window. (b) E
totsaved
vs. balance temperature for
low-e and solar control window in a north-facing window.
Fig. 13. E
totsaved
vs. balance temperature for a Miami TMY2 climate [18],
for north (N) and south (S) orientations.
648 J. Karlsson et al. / Energy and Buildings 33 (2001) 641651
The concept of the balance temperature can be explicitly
dened in terms of other building parameters, for the
purpose of this simplistic approach, however, we choose
to dene the balance temperature in words as the average
annual temperature above which the building does not need
to be actively heated by its installed heating system. It is our
experience that the persons who maintain a building have a
feeling for the balance temperature. Realistic values of T
b
should be somewhere in the range from about 58C up to
less than T
eq
. A future report will describe how the balance
temperature can be deduced and what are the typical balance
temperatures of some buildings.
4.2.2. Time interval
An increased time interval, t, leads to a higher energy
balance for a window because a higher t means that more
energy can be stored in the building. However, an increased
time interval reduces E
totsaved
for a low-e TGU (Fig. 14)
because the lower transmittance of the better window
reduces the throughput of useful energy.
The time interval, t, could possibly be dened as the time
it takes for the indoor temperature to decrease 238C.
However, the time interval or the time constant of a building
is generally not known, so the choices could more readily be
implemented as categories. These could, for instance, be
buildings with light, medium or heavy thermal mass corre-
sponding to a time constant of, for instance, 6, 15 and 100 h,
respectively (see arrows in Fig. 14).
4.3. Cost analysis
It is obviously not always true that the window saving
most energy also is the most cost efcient. To analyse this, a
simple economical calculation can be performed based on
the output from the presented model. Since, it is rarely cost
effective to replace ``fresh'' and functional windows just to
save energy, the analysis is based on a case where the
windows, of base-case type, should be replaced because
they are worn out (or similar reason). The installation cost
for the windows can then be discarded, since that would
come anyway. The same condition is valid if a new building
is going to be built and windows are to be chosen. Adding
the extra cost for the IGU and the extra cost for the frame (if
any) thus gives the additional cost, X, compared to the base-
case. The annuity, C
inv
, of this investment is described by
C
inv
=
Xr
1 (1 r)
L
(6)
where r is the appraised annual interest rate during the
appraised lifetime, L, of the window. In these calculations, r
was set to 6% and L to 30 years. C
inv
will now represent the
extra annual cost/m
2
, equally spread over L-years to invest in
a more advanced window compared to the base-case. Multi-
plying E
saved
or P
saved
with an appraised heating or cooling
energy price per kilowatt-hour will provide what it would
cost per year, C
y
to buy that amount of energy instead. The
yearly profit or loss per square meter will thus be
C = C
y
C
inv
. The outcome of this will of course depend
Fig. 14. E
totsaved
for a low-e window (g(0) = 56, U = 0:9) in a building
with T
b
= 13

C and time intervals, t = 1, 6, 10, 15, 24, 300 and 1000 h,


respectively. The arrows indicate how the building could be categorised as
a light, medium or heavy construction.
Fig. 15. Economy for low-e windows considering heating only. T
b
= 13

C, r = 6%, L = 30 years, extra cost for a triple glazing frame = 100 SEK,
E
conv
= 1 and energy cost = 0:4 SEK/kW h. The y-axis gives saved (SEK/m
2
per year) compared to base-case during the lifetime of the window. The x-axis
gives the orientation of the window.
J. Karlsson et al. / Energy and Buildings 33 (2001) 641651 649
on the economical factors that are hard to estimate, such as
the energy price during the lifetime of the window. Conse-
quently, the economical analysis should be adopted with
caution. An example of how it may look is given in Fig. 15,
where saved SEK per square meter and year, considering the
heating energy balance only, C
E
, is given for different
window orientations.
As soon as C
E
is positive here, it means that the invest-
ment is paying off with an amount of C
E
in SEK/m
2
per year
during the whole lifetime of the window. A bit surprising is
perhaps the fact that the most cost effective alternatives with
the given prerequisites here are the DGU alternatives, with
one low-e coating and gas lling. The ``winning'' alternative
saves about 4050 SEK/m
2
per year in heating energy cost,
during the 30 years that the windows presumably will be
intact. For every conguration it is more cost effective to
install them in the northerly orientations than in the south-
erly because of the accompanying lower transmittance with
the lower u-value. When performing a similar calculation for
the cooling load, with CP
energy
= 0:8 SEK/kW h electricity,
and P
conv
= 2:9, it turns out not to be important for this
climate and balance temperature. The higher the energy cost,
the more cost effective it is to install energy efcient
windows.
5. Conclusion and discussion
An hourly-based simple simulation tool for the energy
efciency of windows has been further developed and
analysed. The model takes into account the solar radiation,
thermal leakage, angle-dependence of the windows and to
some extent the thermal mass of a building. The balance
temperature and the time interval represent the building. The
balance temperature is used as a measure to assess whether
the energy ows are positive or negative for the building
energy system. The balance temperature is lowif the internal
load is high and/or the building is well insulated. The time
interval gives the building its dynamic properties making it
possible to store energy from one hour to another.
For a residential house with T
b
= 13

C in a mid-Swedish
climate, it is seen that changing from an uncoated DGU to
the best alternative can save about 200 kW h/m
2
per year
(glazed area) in a north orientation and about 150 kW h/m
2
per year in a south orientation. The best alternative for this
case was a TGU with two tin-oxide low-e coatings and gas
ll. All the coated DGU alternatives outperformed the
uncoated TGU. The economical analysis indicated that
gas-lled and coated DGUs may be the most cost effective
choice, even when compared with high performing TGUs.
For a commercial building with T
b
= 0

C, a solar control
window can save about 120 kW h/m
2
per year in a north
orientation and 200 kW h/m
2
per year in a south orientation
in the same location.
The model does not consider the fact that the balance
temperature will change when the windows are replaced.
However, using a sensitivity analysis as in Section 4.2, it can
be assessed how different T
b
will affect the choice of the
optimal window. Visual quality, lighting energy and comfort
issues are not accounted for in this model. The reduced
visual transmission can be interpreted as a problem for the
extreme solar control windows, but hardly for the common
low-e windows, since they have high visual transmittance.
Lighting energy is difcult to assess with a simple algorithm
like this, since it depends a lot on the interior of the building
and the window to wall ratio. The thermal comfort is
improved upon a change of window for virtually all win-
dows investigated here, since they will reduce the heat sink
close to the window. In a practical case, a change of windows
may also lead to lower power needs and thus simplied
heating and cooling systems. These and other types of
secondary effects that follow (e.g. new thermostat and
ventilation settings) from a change of windows are not
accounted for. The transmittance of solar energy is over-
estimated in the presented calculations, since no shading is
accounted for. This can, however, be added as an option.
Size effects can be accounted for by changing T
b
or t.
It is understood from the results that below a certain T
b
,
the solar control windows become energy/cost effective. If
T
b
is held constant, the solar control windows become
energy/cost effective below certain latitude, and the triple
glazed low-e windows become cost effective above certain
latitude. An hourly model like this, which requires limited
building data and detailed window data, makes it possible to
compare different windows under different conditions in a
simple way and also, with some extension, to do simple
studies of advanced windows like, e.g. variable and angular
selective windows. The formulation and presumed accuracy
of this model makes it suitable for a window energy rating
system, where for example the total energy balance can be
labelled for four different bearings and for two different
types of buildings, e.g. T
b
= 15 and 08C in each climate
zone. An example of how an energy rating system may look
like is given in Table 2. It might be preferable to separate E
and P, since majority of the residential buildings have no
active cooling system.
To summarise, we propose that the model should be used:
v as an energy rating (labelling) tool for windows;
v as a window selection tool in existing buildings where a
change of windows is about to be performed;
v to assist in the choice of windows in new buildings;
Table 2
Exemplified energy rating label of an uncoated DGU in a mid-Swedish
climate for two different types of buildings
a
E
tot
(E P) N E/W S
Residential (T
b
= 15) 210 120 17
Commercial (T
b
= 0) 180 310 320
a
Climate zone: mid-Swedish; window: uncoated, DGU; thermal mass:
medium.
650 J. Karlsson et al. / Energy and Buildings 33 (2001) 641651
v to make simple theoretical comparisons of the energy
efficiency between existing and/or non-existing advanced
windows in different climates, orientations and types of
buildings.
Caution should be taken in buildings with large windowto
wall ratios, which is presumably not possible to describe
with a balance temperature. In the case of newconstructions,
the balance temperature has to be assessed based on the
category of building that is going to be built. Even when it
comes to an existing building, the balance temperature is
only known within certain limits. This and the fact that t has
to be roughly categorised, limits the accuracy of the model.
Future studies will be concentrated on how T
b
is related to
building parameters and validation against detailed building
simulation programs. An implementation of the model has
been presented [19,20].
Acknowledgements
Lennart Alenrot at Pilkington Floatglass AB kindly sup-
ported the authors with samples, physical data and prices.
Bengt Perers, Vattenfall Utveckling ABprovided very useful
information about solar irradiation. Sture Palm at SP-fonster
AB kindly supported the authors with information about
prices of frames. Prof. John Bell, Queensland University of
Technology is acknowledged for his ideas and suggestions to
improve the simulation model. This project is nanced by
the Swedish Foundation for Strategic Research, SSF,
through the graduate school Energy Systems.
References
[1] C.G. Granqvist, Materials Science for Solar Energy Conversion
Systems, Pergamon Press, Oxford, 1991.
[2] T.E. Johnson, Low-e Glazing Design Guide, Butterworth-Heinemann,
USA, 1991.
[3] K. Kallblad, Thermal Models of Buildings, Determination of
Temperatures, Heating and Cooling Loads, Theories, Models and
Computer Programs, Report TABK-98/1015, Building Science,
Sweden, 1998.
[4] G. Brown, The BRIS computer program for simulating building
thermal behaviour, Royal Institute of Technology, Stockholm,
Building Services Engineering, Bulletin No. 14, 1989.
[5] R. Sullivan, P. Lyons, P.C. Thomas, I. Heimanen, I. Andresen, O.
Aschehoug, H. Simmler, P. Eggimann, T. Frank, Energy simulation
studies in IEA/SHC task 18 advanced glazings and associated
materials for solar and building applications, in: Proceedings of the
Window Innovations '95, Toronto, CA, 1995.
[6] K. Peippo, P.D. Lund, E. Vartiainen, Multivariate optimisation of
design trade-offs for solar low energy buildings, Energy and
Buildings 29 (2) (1999) 189205.
[7] J. Huang, R. Mitchell, D. Arasteh, S. Selkowtitz, Residential
Fenestration Performance Analysis Using RESFEN 3.1, LBNL-
42871, Lawrence Berkeley National Laboratory, US, 1997.
[8] I. Moore, J. Bell, H. Willrath, Simplified modeling of window
glazing performance in conditioned buildings, in: Proceedings of the
Solar 98, ANZSES, Australia, 1998.
[9] J.M. Schultz, S. Svendsen, WinSim: a simple simulation program for
evaluating the influence of windows on heating demand and risk of
overheating, Solar Energy 63 (4) (1998) 251258.
[10] A. Roos, B. Karlsson, Optical and thermal characterisation of multiple
glazedwindows withlowU-values, Solar Energy52(4) (1998) 315325.
[11] B. Karlsson, C. Brunstrom, Characterization of transparent insula-
tion, in: Proceedings of the Symposium on Building Physics, Lund,
Sweden, 2427 August 1987, pp. 96103, ISBN 91-540-4905-9.
[12] S.C. Carpenter, P.E. Alexander, G. McGowan, P.E. Steven, R. Miller,
Window Annual Energy Rating Systems: What They Tell Us About
Residential Window Design and Selection, ASHRAE Transactions,
Toronto, CA, 1998.
[13] T.R. Nielsen, S. Svendsen, Determination of net energy gain from
glazings and windows, in: Proceedings of the Eurosun 2000, 1922
June 2000, Denmark.
[14] H. Bulow-Hube, The effect of glazing type and size on annual
heating and cooling demand for Swedish offices, in: Proceedings of
Renewable Energy Technologies in Cold Climates '98, Montreal,
CA, 46 May 1998, pp. 188193.
[15] B.A. Brresen, T. Jacobsen, SOLVOR A

rlig energi og effektbereg-


ning, SINTEF, Norway, 1982, ISBN 82-595 3151-8.
[16] J.A. Duffie, W.A. Beckman, Solar Energy of Thermal Processes, 2nd
Edition, Wiley, New York, 1991.
[17] J. Karlsson, A. Roos, Modelling the angular behaviour of the total
solar energy transmittance of windows, Solar Energy 69 (4) (2000)
321329.
[18] NREL: National Renewable Energy Laboratory, USA, User's Manual
for TMY2s, Typical Meteorological Years, http://rredc.nrel.gov/solar/
old_data/nsrdb/tmy2/, 1995.
[19] J. Karlsson, WinSel: a general window selection and energy rating
tool, in: Proceedings of the World Renewable Energy Congress VI
(WREC2000), England, 2000, IV, pp. 27082715.
[20] www.angstrom.uu.se/ftf/joakim/winsel.zip.
J. Karlsson et al. / Energy and Buildings 33 (2001) 641651 651

Вам также может понравиться