Professional Documents
Culture Documents
Dynamic Daylight Simulations: Impact of Weather File's Choice
Dynamic Daylight Simulations: Impact of Weather File's Choice
Dynamic Daylight Simulations: Impact of Weather File's Choice
com
ScienceDirect
Solar Energy 117 (2015) 224–235
www.elsevier.com/locate/solener
Received 16 March 2015; received in revised form 27 April 2015; accepted 1 May 2015
Available online 19 May 2015
Abstract
This paper is the second step of a research project aimed at investigating the impact of the use of different weather data files on day-
light simulations’ results. Simulations were carried out for a simple standalone office using three weather files (IWEC, Meteonorm and
Satel-Light) for two European locations (Copenhagen and Rome); moreover the office’s exposure was changed according to the four
main orientations (North, East, South and West).
Results were analyzed both in terms of Annual and Monthly Light Exposures, dynamic daylight performance metrics (DA, DAcon,
UDI) and sunlight’s incidence.
It was demonstrated that differences between the results obtained with the different weather files are more significant considering
Annual and Monthly Light Exposures (highest value 20%), whereas they decrease when analyzing illuminances with a statistical
approach (DA, DAcon, UDI). The analysis of sunlight’s incidence also determined similar results using the three weather data files
and the maximum difference is 5% independently from the orientation.
Ó 2015 Elsevier Ltd. All rights reserved.
Keywords: Dynamic daylight simulations; Indoor environment quality; Weather data files; Dynamic daylight performance metrics
http://dx.doi.org/10.1016/j.solener.2015.05.002
0038-092X/Ó 2015 Elsevier Ltd. All rights reserved.
L. Bellia et al. / Solar Energy 117 (2015) 224–235 225
dry bulb temperature, dew point temperature, wind speed, use of TRY determines indoor daylight levels which are
relative humidity, irradiance, illuminance, sky cover. always lower than those obtained with the other weather
Weather files can be uploaded in dynamic daylight sim- data files. The analysis of data obtained using four different
ulations software, which import data referred to global and weather data files for each one of the five cities determined
diffuse irradiances and convert them into illuminances a great amount of results; consequently the choice of a
thanks to a luminous efficacy model (Reinhart, 2006; North-oriented office simplified the data processing, since
Jakubiec and Reinhart, 2011). Then illuminances values it allowed to neglect sunlight’s contribution.
are employed to generate a luminance distribution which However the analysis of the impact of the use of a
is necessary to model the sky dome and finally simulate weather data or another on direct sunlight modeling is fun-
indoor daylight levels (Mardaljevic, 2000). damental for orientations different from North. Indeed, the
There are different types of weather data files that can be prediction of sunlight-related illuminances affects the eval-
used to perform dynamic daylight simulations: IWEC uation of disability glare risk due to daylight and of an
(International Weather for International Calculation), environment’s performances in terms of comfort. For
Meteonorm, TRY (Test Reference Year), TMY (Typical example, the incorrect prediction of sunlight’s incidence
Meteorological Year), DRY (Design Reference Year). may determine an imprecise evaluation of the number of
Meteorological data reported in each weather file are hours in a year for which it is necessary to use shading
obtained by processing historical sets of annual measure- devices and consequently it may influence their design.
ments through a statistical calculation model. For example, Considering that the use of shadings reduces indoor day-
an USA TRY weather file is obtained by selecting one real light levels, thus determining the need to use electric light,
year between those included in a historical set. The identi- the incorrect modeling of sunlight may also affect the eval-
fication of the typical year is based on an excluding process uation of energy savings.
which aims at eliminating years characterized by months Given these premises, the goal of this paper is to repeat
with very high or low average temperatures; this process the study reported in Bellia et al. (2015), varying the office’s
is repeated until only one year remains (Crawley, 1998). exposure according to the four main orientations (North,
On the other hand the procedure to develop an IWEC file East, South and West), in order to analyze the effect of
is different: it consists in selecting twelve typical months the use of different weather data files also on sunlight’s
from the different years available; then these months are modeling.
combined together to create a yearlong weather file (U.S. Given that previous researches (Crawley, 1998; Bellia
Department of Energy, Energy Efficiency & Renewable et al., 2015) did not recommend the use of TRY, only three
Energy). weather data files are considered: IWEC, Meteonorm and
Considering that each type of weather file is developed Satel-Light. In order to reduce the amount of results to
starting from different historical sets and by using a specific process, only two cities, characterized by very different
statistical process, the values contained in these files are weather conditions, are selected: Rome and Copenhagen.
likely to diverge if compared (http://www.satel-light.com/
; http://meteonorm.com/; U.S. Department of Energy, 2. Method
Energy Efficiency & Renewable Energy; Crawley, 1998).
Consequently the use of a weather data file or another The study compares dynamic daylight simulations’
could affect the results obtained with dynamic simulations. results referred to a simple office located in two
Iversen et al. (2012) investigated the effect of using three European cities, Copenhagen and Rome, and exposed
different weather data file (DRY, IWEC and Meteonorm) according to the four main orientations.
on the output of dynamic daylight simulations carried out As already mentioned in the introduction, Copenhagen
for a South-oriented office located in Copenhagen. They (55°410 N 12°350 E) and Rome (41°530 3500 N 12°280 5800 E)
analyzed the data in terms of lighting dependencies and were chosen because they are characterized by very differ-
found that differences in results showed variations up to 2%. ent weather conditions. According to the Köppen climate
Bellia et al. (2015) compared results obtained by per- classification, Copenhagen belongs to the warm summer
forming dynamic daylight simulations of a simple continental or hemiboreal climatic zone and Rome to the
North-oriented office located in five European cities dry-summer subtropical-mediterranean one. In more
(Copenhagen, Milan, Nancy, London, Rome), each one detail, in Copenhagen the temperature varies between an
characterized by different weather conditions, using four average low equal to 12.3 °C and an average high of
weather data files (IWEC, Meteonorm, Satel-Light and 20 °C during summer and between an average low of
TRY). They found that the use of IWEC, Meteonorm 1.6 °C and an average high temperature of 2.5 °C during
and Satel-Light weather data files produces similar results winter. The Danish city is characterized by yearly precipi-
in terms of dynamic daylight performance metrics tation equal to 613 mm and by yearly mean sunshine hours
(Daylight Autonomy, Continuous Daylight Autonomy equal to 1539 (Danmarks Meteorologiske Institut). On the
and Useful Daylight illuminance), whereas significant dif- other hand, in Rome the temperature varies between an
ferences are obtained for Annual and Monthly Light average low of 16.2 °C and an average high of 30.4 °C dur-
Exposures. Furthermore the study demonstrated that the ing summer and between an average low temperature of
226 L. Bellia et al. / Solar Energy 117 (2015) 224–235
Table 1
DIVA simulation parameters.
Ambient bounces Ambient divisions Ambient super samples Ambient resolution Ambient accuracy
0 or 7 1500 100 300 0.05
to Rome (Table 4) assume completely different trends. and West exposures, Satel-Light determines the lowest val-
When the window is East or South-facing the highest val- ues for North orientation and Meteonorm for the East one.
ues are calculated with Satel-Light, when it is This means that it is not possible to identify a weather
North-facing they are obtained with IWEC and when it data file that always determines results higher than the
is West-facing they are achieved with Meteonorm. On the others and vice versa, indeed in Fig. 2 I–S and M–S per-
other hand, IWEC provides the lowest values for South centage differences assume positive or negative values
depending on the city taken into account, the orientation
Table 4 and the calculation point.
Annual Light Exposure values referred to Rome. Furthermore, I–S and M–S differences are characterized
Rome Annual Light Exposure (klx h/year) by a dissimilar order of magnitude. Considering percentage
values, their variation can be quite significant, for example
IWEC Meteonorm Satel-Light
they can range between 20% (I–S, Copenhagen, West ori-
North entation, S5) and 22% (M–S, Rome, S5); or they can also
N 4870 (max) 4492 4018 (min)
C 1347 (max) 1298 1252 (min) be very small (i.e. Copenhagen, North orientation, S4).
F 735 (max) 717 712 (min) I–S and M–S values change depending on the calcula-
tion point. Only for North orientation, I–S and M–S values
East
N 14,128 13,140 (min) 16,256 (max) are always positive for all sensors and they decrease when
C 3134 2917 (min) 3747 (max) the distance from the window increases. This does not
F 1343 1232 (min) 1440 (max) occur for the other orientations.
South To provide an in depth analysis of the results obtained
N 21,376 (min) 23,708 24,860 (max) for each city, orientation and weather data file Montly
C 4599 (min) 4903 5398 (max) Light Exposure were also calculated. Figs. 3 and 4 show
F 1643 (min) 1691 1744 (max) the trend assumed by Monthly Light Exposures [klx h/-
West month] (MLE) for each city and each orientation. For
N 11,162 (min) 12,950 (max) 12,392 readability reasons the maximum value of the y axis is dif-
C 2141 (min) 2377 (max) 2231
ferent depending on orientation, but a 500 klx h/month
F 1038 (min) 1137 (max) 1098
step is always assumed between tick marks.
Fig. 2. Percentage differences for Annual Light Exposure referred to Copenhagen and Rome.
L. Bellia et al. / Solar Energy 117 (2015) 224–235 229
Figs. 3 and 4 demonstrate that the curves tend to be Consequently, considering that sunlight’s modeling
more similar in winter than in other seasons for North, depends on the direct irradiance data uploaded in the cal-
East and West orientations, whereas for South the contrary culation software, results related to indoor daylight avail-
applies. ability may show higher differences for orientations
Regardless of the window’s orientation it is interesting different from North.
to notice that also the slope of the curves can be very dif- For this reason the following paragraph reports an anal-
ferent depending on the weather data. ysis of the direct component of daylight entering in the con-
Differences in results observed for the four orientations sidered office.
are likely to be determined by sunlight’s incidence. Data
reported in the different weather files and related to direct 3.2. Analysis of the direct component
and diffuse component of irradiance can indeed diverge if
compared. Bellia et al. (2015) analyzed cumulative fre- An analysis of the direct component of daylight was per-
quency curves related to global and diffuse horizontal irra- formed. To do so, simulations were repeated setting the
diances, calculated starting from the data reported in ambient bounces (ab) parameter to zero, as previously
different weather files (IWEC, Meteonorm, TRY and reported, to exclude any reflection of light by surfaces.
Satel-Light) and concerning five cities (Copenhagen, This analysis was only performed for East, West and
Milan, London, Nancy, Rome). They demonstrated that South orientations.
global irradiance curves show very similar trends except Fig. 5a and b reports for Copenhagen and Rome the
for TRY, whereas those referred to the diffuse component annual percentages of occupied hours during which
turn out to be slightly different. For example, considering sunlight-related illuminances (Esun), calculated using the
the specific case of Rome, they found that Satel-Light’s dif- different weather data files for each orientation, are com-
fuse irradiance curve assumes lower values compared to prised between 0 and 1000 lux or they are greater than
those representing the other files, whereas its global irradi- 1000 lux. This 1000 lux threshold was assumed since it is
ance curve is very similar to those related to IWEC and the limit value considered in ASE calculation (IES
Meteonorm. This means that values referred to the direct LM-83-12 “Approved Method: IES Spatial Daylight
component of irradiance, reported in different files, diverge. Autonomy (sDA) and Annual Sunlight Exposure (ASE)”,
230 L. Bellia et al. / Solar Energy 117 (2015) 224–235
2013). For readability reasons, the order of magnitude of 3.3. DA and UDI
the graphs related to the East and West exposures in both
cities is different. Fig. 6a and b reports DA and DAcon values for each
It can be noted that the total percentage of occupied city, orientation and weather data file.
hours during which sunlight is calculated is similar for From the graphs reported in Fig. 6a and b it can be
both cities. On the contrary, the percentage of occupied noted that the differences in DA and DAcon values
hours for which Esun is greater than 1000 lux is signif- between the chosen weather data files remain small for each
icantly higher for Rome. In the South orientation in orientation and each city.
Rome, using all weather data files, sunlight is not calcu- It is interesting to highlight that, for Rome, DA and
lated for the sensor farthest from the window, whereas DAcon values obtained using the Meteonorm file are gen-
the contrary applies to Copenhagen; the explanation erally a little lower than those calculated with the other
of this result resides in the fact that in Copenhagen weather data files. This is also true for Copenhagen’s
sun’s altitude is lower and therefore it enters deep into DAcon results, whereas DA values obtained with
the room. Meteonorm weather data file for this city are lower than
The greatest differences among weather data files are the others only in a few cases.
found for the sensor near the window in the East orienta- Another interesting finding is that, for both cities, DA
tion. In more detail, for both cities, using the Meteonorm values calculated for the central sensor (C) with the
file, the percentage of occupied hours for which Esun is IWEC weather data file are always higher than the others
greater than 1000 lux is higher than the one calculated with for each orientation. Moreover the use of Satel-Light
IWEC and Satel-Light weather data files (for always determines, for Copenhagen, the highest DA values
Copenhagen); whereas in Rome’s case it is only higher than for the sensor closest to the window, whereas in Rome’s
the one calculated with the IWEC file. case this only occurs in South and West orientations; in
Usually differences in the total percentage of occupied the other orientations IWEC’s and Satel-Light’s DA values
hours during which sunlight is calculated are small (not are almost coincident.
greater than 5%) and more evident for the sensor near Fig. 7a and b reports UDI values for each city and each
the window. orientation.
L. Bellia et al. / Solar Energy 117 (2015) 224–235 231
Fig. 5. (a) and (b): Annual percentage of occupied hours during which 0 < Esun 6 1000 lux or Esun > 1000.
232 L. Bellia et al. / Solar Energy 117 (2015) 224–235
Fig. 7. (a) and (b): UDI values for Copenhagen and Rome.
It can be observed from Fig. 7a and b that the use of for the sensors near the window in both cities and for all
Meteonorm files always determines slightly greater orientations. In more detail, in Rome such differences are
UDI100 values. more significant.
Moreover it is interesting to highlight that the greatest Furthermore the use of IWEC files generally determines
differences are found between UDI100–2000 and UDI2000 the highest UDI2000 values.
234 L. Bellia et al. / Solar Energy 117 (2015) 224–235
Mardaljevic, J., 2000. Simulation of annual daylighting profiles for Rogers, Z., 2006. Daylighting metric development using daylight auton-
internal illuminance. Light. Res. Technol. 32 (3), 111–118. omy calculations in the sensor placement optimization tool. <http://
Mardaljevich, J., Nabil, A., 2005. Useful daylight illuminance: a new www.archenergy.com/SPOT/download.html>.
paradigm for assessing daylight in buildings. Light. Res. Technol. 37 Servizio Meteorologico Aeronautica Militare, n.d. <http://www.meteoam.
(1), 41–59. it/> (retrieved 28.10.14)
Meteonorm, Handbook Part II – Theory. (2014, September). <http:// Stevens, R., Blask, D., Brainard, G., Hansen, J., Lockley, S., Provencio,
meteonorm.com/images/uploads/downloads/mn71_theory.pdf> I., Reinlib, L., 2007. Meeting report: the role of environmental lighting
(retrieved 27.10.14). and circadian disruption in cancer and other diseases. Environ. Health
Partonen, T., Lonnqvist, J., 2000. Bright light improves vitality and Perspect. 115 (9), 1357–1362.
alleviates distress in healthy people. J. Affect. Disord. (57), 1–3. U.S. Department of Energy, Energy Efficiency & Renewable Energy, n.d.
Reinhart, C.F., 2006. Tutorial on the use of daysim simulations for <http://apps1.eere.energy.gov/buildings/energyplus/weatherdata_
sustainable design. Ottawa (Ont.): National Research Council Canada. about.cfm> (retrieved 22.5.14).
Reinhart, C., Mardaljevic, J., Rogers, Z., 2006. Dynamic daylight
performance metrics for sustainable building design. Leukos 3 (1), 7–31.