Citation for published version:
Herrera, M, Natarajan, S, Coley, D, Kershaw, T, Ramallo Gonzalez, AP, Eames, M, Fosas, D & Wood, M 2017,
'A Review of Current and Future Weather Data for Building Simulation', Building Services Engineering Research
& Technology, vol. 38, no. 5, pp. 602-627. https://doi.org/10.1177/0143624417705937
DOI:
10.1177/0143624417705937
Publication date:
2017
Document Version
Peer reviewed version
Link to publication
University of Bath
Alternative formats
If you require this document in an alternative format, please contact:
openaccess@bath.ac.uk
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Download date: 10. Aug. 2022
A Review of Current and Future
Weather Data for Building
Simulation
Manuel Herrera
a*
, Sukumar Natarajan
a
, David A. Coley
a
, Tristan Kershaw
a
, Alfonso P. Ramallo-
González
b
, Matthew Eames
c
, Daniel Fosas
a
and Michael Wood
c
a
EDEn – Dept. of Architecture and Civil Engineering, University of Bath, Bath, UK
b
Faculty of Computer Science, University of Murcia, Murcia, Spain
c
College of Engineering, Mathematics and Physical Sciences, University of Exeter, Exeter, UK
*
Corresponding author (Manuel Herrera) e-mail: amhf20@bath.ac.uk
Abstract
This article provides the first comprehensive assessment of methods for the creation of weather
variables for use in building simulation. We undertake a critical analysis of the fundamental issues
and limitations of each methodology and discusses new challenges, such as how to deal with
uncertainty, the urban heat island, climate change and extreme events. Proposals for the next
generation of weather files for building simulation are made based on this analysis. A seven-point
list of requirements for weather files is introduced and the state-of-the-art compared to this via a
mapping exercise. It is found that there are various issues with all current and suggested
approaches, but the two areas most requiring attention are the production of weather files for the
urban landscape and files specifically designed to test buildings against the criteria of morbidity,
mortality and building services system failure.
Practical applications: Robust weather files are key to the design of sustainable, healthy and
comfortable buildings. This article provides the first comprehensive assessment of their technical
requirements to ensure buildings perform well in both current and future climates.
1. Introduction
Writing near the end of the first century B.C.E., the Roman architect Vitruvius, suggested that a
building should seek to offer firmitas, utilitas, and venustas (firmness, commodity, and delight).
Utilitas included the arrangement of spaces and the way the building provides both shelter from the
external environment and comfortable internal conditions in which to carry out everyday tasks.
Most buildings meet the basic requirements of shelter, so it is mainly the provision of comfort that
taxes environmental designers. We spend 80-90% of our time inside buildings [1, 2] and poor
internal conditions will not merely affect comfort but also impair occupant health and productivity.
Since local weather and climate greatly affect the construction and performance of the building,
good quality weather data is essential to simulate building performance. In this context, weather
data should allow designers to stress test building performance for atypical conditions such as heat
waves or cold snaps, since such conditions are more likely to cause performance failures. For
example, the European heat wave of 2003 is said to have resulted in 70,000 excess summer deaths,
primarily as a result of maladapted built environments [3, 4]. In the near future, the return period for
such a heat wave is likely to change from 1 in 250 years to 1 in 50, or even 1 in 35 [5]. Since typical
building lifetimes can be around 60 years or more, weather data needs to cover future changes, and
be local to the building.
Weather files ideally need to:
1. Contain examples of typical conditions;
2. Contain examples of extreme conditions;
3. Be at the temporal resolution required by simulation packages (typically a 1 hour or higher
resolution)
4. Be at a geographic resolution that matches changes in weather due to local topography in the
country of interest;
5. Express the effect of the urban micro-climate; and
6. Contain examples of possible future climates, ideally considering the effects of climate
change;
In addition to the six technical features above, weather files also need to be credible. This suggests a
seventh feature necessary for the success of any weather file:
7. Proven track record with industry.
In this paper, we summarise the approaches and methodologies used to produce files that aim to
match these requirements. This review is timely given that DEFRA and the UK Meteorological
Office are about to expend a very large amount of computer time on producing the next set of
climate predictions. However, the Directive on the Energy Performance of Buildings (EPBD) [6]
has required built environment professionals to take measures that adapt planning policies and new
building specifications to guarantee a minimum level of comfort and safety since 2002 [7]. This is
especially important when considering the issues of providing adequate ventilation and in particular
of limiting overheating [8] as this is associated with reduced worker productivity, morbidity and
even mortality in vulnerable groups such as infants and the elderly [9, 10]. There is not a directive
equivalent to EPBD in other world areas. However, the International Energy Agency (IEA) fosters
the implementation of building energy saving policies at a national level in its member countries,
but also in non-member countries such as China, India, and Russia.
This work is divided into the following sections. Section 2 reviews the state-of-the-art for creating
weather data for building simulation. The different methodologies are classified according to how
well the resultant weather meets the seven-point list of requirements. Section 3 focuses upon the
methodologies for creating weather data from synthetic time series from a weather generator.
Section 4 considers the creation of future weather data for building simulation both from synthetic
data sets and from the morphing of historical observations. Finally, a discussion section considers
the challenges that building engineers face when considering the future performance of buildings
and increasing the resilience of the built environment to climate change.
2. Weather !les for building simulation
The building industry mainly uses weather data to assess the design and performance of the built
environment at the planning stage. This is becoming more important because climate change is
likely to lead to an increase in the frequency of extreme weather events [11].
Dynamic building energy simulations were developed as early as the 1950s [12], but it was not until
the energy crisis of the 1970s that the scientific community started using them to help improve the
energy performance of buildings [13]. In early models, weather data were applied to the software
packages in a variety of formats, but were standardised into ‘weather files’ by the 3
rd
generation of
dynamic building simulators (following the classification of Clarke [14]). These generally take the
form of typical weather years, created from hourly historic observations at a specific location [15].
However, the need to adapt buildings to the impacts of likely future climate change has created a
requirement to incorporate climate change projections into these weather files, either by morphing
the weather data or synthetically generating it [16, 17]. Weather files representing ‘extreme’ years
(i.e. the selection of observed weather occurrences far from the norm) have also been introduced to
analyse a building design’s response in case of severe weather conditions [18].
2.1. Format of weather !les
According to the bi-annual publications of the International Building Performance Simulation
Association (IBPSA), the most commonly used building simulator is EnergyPlus. As well as being a
stand-along programme, EnergyPlus is also the basis for more sophisticated software (e.g.
DesignBuilder), which are capable of analysing not only energy consumption, but water use and
daylighting.
The popularity of EnergyPlus has driven the popularisation of its native weather file format - the
‘.epw’ file (Energy Plus Weather). These .epw files are text-based CSV files that contain a years-
worth of hourly weather variables for a given location. The file structure was developed by USDoE,
and is also used in other software such as ESP-r, IES, and TAS. The USDoE offers 2,590 weather
files for different worldwide locations in the current climate.
Figure 1 shows the header of a typical .epw weather file, the first line of the header shows the
location and the specific format of the weather data (in this case TMY2). The header also contains
the location, longitude, latitude, time zone, elevation, annual design conditions, monthly average
ground temperatures, typical/extreme periods and holidays/saving periods. It also has information
on which data periods were included.
Fig. 1. Sample of .epw weather file, showing header information and weather data form line 9.
2.2. Files for typical weather conditions
Files for typical weather conditions include hourly data on temperature, dew point, global
horizontal radiation, diffuse solar radiation, wind speed and wind direction. These files are used to
estimate the average building energy use and carbon emissions [19, 20]. A typical weather file is
created from historic data (usually around 20-30 years of data, depending on the data availability).
This data is compiled by comparing the cumulative and the empirical distribution functions of
different meteorological variables within the base data set. The number and weighting of different
meteorological variables considered is a feature of the weather file type (i.e. TMY, TRY, etc. – see
below). Table 1 shows a representative sample of typical weather files used around the world. It is
worth mentioning that despite having different sources or ways in which weather files are created to
form distinct file types, several of them use common file formats such as the EPW format.
Table 1. A short list of weather file types in various countries. The period depends on the data availability at the location
1
.
Acronym Complete name Region Sites Period
RMY Representative Meteorological Year Australia 69 locations 1967-04
CSWD Chinese Standard Weather Data China 270 locations 1982-97
ISHRAE Indian Typical Years from ISHRAE India 62 locations 1991-05
IGDG Italian ‘Gianni De Giorgio’ Italy 68 locations 1951-70
SWEC Spanish Weather for Energy Calculations Spain 52 locations 1961-90
UK TRY Test Reference Year (CIBSE) UK 14 locations 1984-13
TMY Typical Meteorological Year USA and others 1,020 locations 1991-05
WYEC Weather Year for Energy Calculations USA/Canada 77 locations 1953-01
IWEC International Weather for Energy Calculations Worldwide 3,012 locations 1991-05
There are two ways to construct a typical weather year. The first is by identifying a continuous 12-
month period as typical. The second is by applying the ranking criteria to individual months from
the basis set, which are then assembled into a composite 12-month year. The UK TRY and TMY
both use the latter approach and are computed using the Finkelstein-Schafer (FS) statistic [21]. This
means that each month in the file might be from a different year. Comparisons of these composite
years with the basis set indicate that both the UK TRY [20] and TMY (with the updated file formats
to TMY2 and TMY3) [22] have advantages over a single year approach.
In the following a summary of the characteristics of the most representative composite weather year
files (TRY, TMY, and IWEC) and some of their extensions and updates is given.
The Test Reference Year (TRY) was developed in 1976 for 60 locations in the United States
[23]. The baseline period was from 1948-1975. From this baseline, years with monthly
extreme values were filtered out until a single year remained containing the least severe (or
most average) weather conditions. TRY initially contained dry bulb, wet bulb, and dew point
temperatures, wind direction and speed, barometric pressure, relative humidity, cloud cover
and type. Later on, the TRY methodology was modified [24, 25] and its scope was expanded
to generate a complete weather data set for several locations worldwide. The exact
modification of the the TRY depended on the institution that created the files. The
differences included the weighting of relevant parameters and even the inclusion or not of
one or more parameters. One example is the Danish Design Reference Year (D-DRY) [26],
which includes specific parameters, such as 5-minute values for direct normal radiation, or
forecast information to be used for the simulation of energy management systems. To create
a D-DRY the data set of basis months are ranked according to the distance (measured in
standard deviations) of each variable per month from the value of the long-term mean.
The Chartered Institution of Building Services Engineers (CIBSE) in association with the
UK Met Office produces CIBSE TRYs for the United Kingdom [27]. In the case of the UK
1 The information of various weather files is regarding to their corresponding updated formats. This
is the case of TMY, updated to TMY2 and TMY3; WYEC, updated to WYEC2; and IWEC, updated
to IWEC2.
