Creation and application of future typical weather files in the evaluation of indoor overheating in free-floating buildings

https://doi.org/10.1016/j.buildenv.2022.109059Get rights and content

Highlights

  • A practical workflow to construct yearly and typical weather files from EURO-CORDEX is presented.

  • Contrary to assumptions, medium future typical years are not as intense as 2003 heatwave year in Nantes.

  • Buildings located in sparsely built locations that are less affected by UHI are also at risk of overheating.

Abstract

Expected Global warming and heatwaves coupled with the urban heat island effect (UHI) can overheat indoor environments of free-floating buildings in temperate climate regions. Overheating assessment requires practitioners to use appropriate climate data and suitable measurement indices. The aim of this article is first, to propose a practical approach to generate yearly and typical ready-to-use future typical weather datasets (FTWY) using high-resolution Regional Climate Model (RCM) data from Coordinated Regional Climate Downscaling Experiment (CORDEX), and second, investigate the potential of FTWYs in the assessment of indoor overheating, considering UHI effect.

To achieve these objectives, three dynamically downscaled (DDS) FTWYs generated from RCMs (IPSL-SMHI, CNRM-ALADIN, MPI-REMO) were compared with one statistically downscaled (ESD) FTWY from Meteonorm, and observed heatwave weather data of 2003. Comparative analysis was performed in two stages: comparison of monthly statistical distribution of climate variables, and analysis of heatwave presence. Urban weather generator (UWG) was used to project UHI effect on two weather files for two buildings, and three overheating measurement indices were used to exploit results. Comparative analysis of weather files show that temperature in a FTWY in the medium future (2040–2070) is likely not as intense as the heatwave of 2003 for Nantes. Results also confirm that it is better to use two weather files, and at least two overheating indices to obtain reliable outputs. This study also revealed that indoor overheating is not limited to densely built areas where impact of UHI is highest; buildings located in sparsely built neighbourhoods are also at risk.

Introduction

2011–2020 was the warmest decade on record and the global average temperature in 2020 was approximately 14.9 °C, which is 1.2 °C higher than the pre-industrial (1850–1900) level [1]. The projections of IPCC in their latest reports warns that even with the best estimates, regardless of what emission scenario is considered, the warming will reach 1.5 °C before 2040 [2]. National Climate Assessment report states that the number and strength of heatwaves, heavy downpours, and major hurricanes have increased over the last decades and will continue to do so in the future. This increase will disrupt and damage critical infrastructures and vitality of communities, putting vulnerable population disproportionately at risk of climate-related adverse consequences [3].

Europe in particular is more likely to be affected by heatwaves and cold snaps compared to other extreme weather events such as hurricanes that form in tropical and subtropical latitudes.

An example is the exceptional heatwave in the summer of 2003 that resulted in at least 30,000 excess deaths in Europe, of which nearly 15,000 were in France, between August 1 and 20, 2003 [4].

Heatwaves simulated using EURO-CORDEX regional multi-model demonstrate that under future climate conditions, the frequency, duration and intensity of heatwaves will increase across France and other parts of Europe. Heatwave events could occur during a larger span of summertime and the 2003 event would be a typical event by the end of the century [5].

Influences of changing frequencies and intensities of extreme weather events are amplified in the urban areas by a distinct urban microclimate feature known as urban heat island (UHI) effect. This effect is characterized by higher temperatures within the build-up urban area as compared to rural surroundings [6]. UHI is the result of many factors that modify the climatic exchanges in the city: shape and density of urban fabric, thermo-physical specifications of artificial urban surfaces and heat generated from anthropogenic human activities. These factors change the urban energy balance by entrapping solar radiation, changing humidity and intensity of air circulation, increasing thermal storage capacity, and decreasing the latent heat transfer due to the reduced presence of vegetation and water bodies. As a result, urban air and surface temperatures cool down slowly in the evening, maintaining a hot environment for buildings and people. UHI has a significant impact on heat stress, thermal comfort and energy demand of the buildings and people in urban areas. Therefore, it is important to consider it in building design and simulations [7].

This study uses Urban Weather Generator (UWG) method to project the influence of UHI effect on typical future weather data and observed weather data. UWG is a methodology and software tool that estimates hourly urban canopy air temperature and humidity ratio based on urban morphological parameters and urban land use [8]. It can be used alone or in conjunction with other existing programs to account for the impact of UHIs.

UWG model contains four interacting components: Rural Station Model (RSM) which estimates sensible heat fluxes; Vertical Diffusion Model (VDM) that calculates vertical air temperature profiles at a rural weather station; Urban Boundary Layer (UBL) that accounts for vertical histograms of the air temperature above the urban coverage; and Urban Canopy and Building Energy Model (UC-BEM) that allows taking into consideration temperature and humidity ratio of the air in the urban canyon [8,9]. UWG has previously been validated in several studies for Basel, Singapore, Toulouse, Rome, Barcelona, and Abu Dhabi [10].

Comprehensive time-activity studies in Europe and US have shown that people on average spend 16 h/day indoors. This number increases to approximately 20 h/day for those above 64 years old [11], asserting the importance of indoor air quality and indoor thermal conditions.

Projected variations in extreme weather events and global temperature increase will further add pressure on buildings, making them uncomfortable or even potentially dangerous to occupants’ wellbeing [[12], [13], [14]]. Heatwaves in particular can cause severe overheating in buildings that are not equipped to cope with it. It could lead to several problems ranging from thermal discomfort and productivity reduction to illnesses and even death of occupants [13].

Furthermore, recent buildings have been designed to answer to past climate conditions and often with typical meteorological year (TMY) weather files that are obtained by means and thus do not include heatwaves [15] or heat island effect. Moreover, past construction regulations and most of the current ones in western and northern Europe are not focused on summer conditions and as a result, even recent buildings can already be highly uncomfortable during the heatwaves [16]; R [17,18].

The magnitude of occupant vulnerability inside the building due to overheating depends on several parameters such as duration and intensity of exposure to heat, as well as, on personal adaptation capacity of the occupant. Installation of cooling systems in the already energy-intensive building sector could mitigate associated risks. However, the resulting increase in energy demand would affect global climate change. Moreover, if installed in every household, these systems would dramatically increase the electricity demand for cooling at peak time and at the same time discharge hot air that will further intensify urban warming. Another factor that affects occupants’ vulnerability to future climate conditions and heatwaves is energy precariousness [19]. This is especially relevant for naturally ventilated buildings that have traditionally not relied on energy to keep occupants safe from overheating during summer.

For practitioners in the thermal evaluation of buildings, this means that typical weather files created from historical records collected from rural weather stations may no longer be suitable to assess buildings’ resilience in the context of future climate and heatwaves.

Indoor overheating, similar to thermal comfort, is a dynamic phenomenon that varies both spatially and temporally. Researchers over the years have proposed numerous methods to describe indoor overheating, but still, there is no consensus on how to evaluate it through simulation or measurement [16]. Therefore, there is a need to specify what we mean by overheating or heat stress in buildings. A literature review on indoor overheating indices by Epstein et al. collected and analysed more than 40 different heat stress indices [20]. Authors argue that too much emphasis has been placed on the academic accuracy of many of these indices at the expense of practicality. They recommend using simple and easy to use indices. Their literature review covered indices that were in use or proposed until 2005. Since then indoor overheating and comfort measurement indices have evolved and a number of new indices have been introduced. The most important change in that front is probably the adoption of adaptive comfort indices by ANSI/ASHRAE Standard 55 and ISSO 74 (Dutch Guidelines) in 2004 and 2005. European standard adopted it in 2007 in EN 15251-1. Adaptive indices were slightly modified and updated in ANSI/ASHRAE Standard 55 in 2017, and EN 16978-1 was introduced in 2019 to replace EN 15251. A review of indices by Ref. [21] proposes to classify homogeneous indices into four families: (1) percentage indices that demonstrate comfort as a percentage [of time] inside a range such as PMV and CIBSE guidelines; (2) cumulative indices such as degree-hour criterion in EN 15251, and exceedance metrics illustrated by Ref. [22]; (3) risk indices such as Nicol et al.‘s overheating risk [23] that suggests thermal discomfort is related to the difference between operative temperature and EN adaptive thresholds, and Robinson and Hadi's overheating risk index [24] which is based on the analogy that (the storage) of human tolerance to overheating stimuli may be equivalent to storage of charge in an electrical capacitor, and proposes a simple mathematical model to predict overheating risk given a set of measured environmental conditions; (4) averaging indices such as the average predicted percentage of dissatisfied (PPD), and the difference between peak temperature and annual average. A more recent review on time-integrated overheating evaluation methods for temperate climate regions by Rahif, Amaripadath, and Attia [25] state that most standards recommend using adaptive comfort methods in the assessment of indoor thermal conditions for free-floating buildings, and static approaches for air-conditioned buildings. Their review analyzes 11 international standards, 5 national building codes, and a number of scientific literature that present overheating indices for temperate climate regions. The large number of indices in the literature and national standards indicate that researchers are trying to quantify the relation between human body and climatic stress in a single formula or by a single index. It is, however, obvious that there is a complex relation between the two and using a single index could mask or even exaggerate the indoor thermal conditions. Using multiple indices could probably better explain this relationship because some indices give complementary results and highlight certain aspect of occupants' sensation that are ignored or given less attention by other indices. In this study, we used three easy to use indices to describe what constitutes indoor over-temperature in a naturally ventilated building.

Three methods of thermal simulations are commonly used nowadays for building performance evaluations: static, semi-dynamic, and dynamic. Of the three, the dynamic method is considered more appropriate in building thermal assessment. This method requires at least hourly weather data that contains temperature, humidity, radiation, wind, atmospheric pressure, etc. Depending on the objective of dynamic BPS, two distinctive weather data types are used: synthetized and observed data. The latter is often used in the performance monitoring phase and is collected from weather stations or by in-situ measurements. The former is more frequently used in the design phase and is synthetically generated from climate normals. WMO defines climate normals as a period that covers at least 30 years of data. For evaluation of buildings’ climate resilience, future weather data are required. These data are based on future emission scenarios and projections produced using climate models. Emission scenarios are used as input for General Circulations Models also referred to as Global Climate Models (GCMs). GCMs cover the entire surface of the globe and their spatial resolution is coarse, typically between 150 and 600 km [26]. Application of given GCMs for building thermal evaluation requires downscaling to a finer spatial and temporal resolution to consider regional and local scale estimates of climate variability and change. As can be seen in Fig. 1, there are two main approaches to downscale GCMs: Dynamical downscaling (DDS) and empirical-statistical downscaling (ESD).

DDS and ESD stand on two distinctive philosophies: DDS relies on climate data that are based on our knowledge of physical processes (solving equations for humidity, temperature, local wind, etc.) and ESD makes use of information obtained from the statistical analysis (e.g. regression relationships) of previously observed climate data.

Erlandsen et al. and Moazami et al. [27,28] in their studies have also discussed a hybrid approach in which the results of dynamically downscaled GCMs, also referred to as Regional climate model (RCM), stored at a coarse resolution undergo further downscaling using statistical approach.

Several studies have been conducted on the relative difference of DDS and ESD on climate impact assessments. P.Tootkabani et al. [26] in their paper on comparative analysis of different future weather data for building energy simulations, compared weather files from three tools that are based on ESD (WeatherShift, Meteonorm, and CCWorldWeatherGen) with one DDS future typical meteorological year. Their results show that all ESD weather files have relatively similar operation in predicting thermal comfort and energy consumption in buildings in comparison to DDS weather files. Their paper also states that the ESD method, regardless of how it is used can provide sufficient information to perform comparative analysis on long-term variations in energy consumption of buildings, but existing inconsistency within the method can lead to significant prediction error. Under such conditions, they found the DDS method more reliable when the objective of the study is to investigate and communicate the resilience of buildings to future climate conditions. Ramon et al. [29] in their paper state that ESD method is more suited to investigate average energy performance in future climate realization but less suited to assess extreme conditions. DDS, on the other hand, can be used for both average and extreme assessment purposes. Moazami et al. [28] in their study on the impact of future weather data types on building energy performance concluded that weather files generated using DDS that take into account both typical and extreme climatic conditions are most reliable to evaluate energy robustness in the context of future climate uncertainties.

In the present study, availability of open-source RCMs (dynamically downscaled GCM) data from EURO-CORDEX presented an opportunity to systematically compare three future DDS climate models and one future ESD model, assuming the high-emission scenario [representative concentration pathway (RCP) 8.5], with observed weather data of 2003. The latter was accessed from MeteoFrance archives and transformed into EnergyPlus (.epw) file format that can be used in BPS.

The aim of this study is, first, to provide insights for practitioners in BPS on how to generate future and present ready-to-use weather files using open-source RCMs, and second, through a comparative analysis with heatwave weather data of 2003 show their potential in indoor overheating assessment of buildings’ considering urban heat island effect. The method described here does not illustrate in detail the uncertainties associated with emission scenarios, climate projections, climate models, and bias adjustment of climate models as they were addressed previously by Refs. [[30], [31], [32]].

The method/workflow to generate typical weather files is of particular interest for practitioners in building simulations that use current or future weather data in thermal performance evaluation of buildings. Implementation of the approach we used to compare and evaluate indoor overheating of two free-floating buildings with two weather files and three indices contribute to the existing body of knowledge in the assessment and study of climate-change-proof buildings.

The next section of this document, materials and methods, is structured as follows: first sub-section describes the workflow to extract yearly weather data for any location; second sub-section demonstrates the methodology to assemble typical weather file from downloaded yearly data; third sub-section presents the method used in comparative analysis of weather files; fourth sub-section depicts the application of weather files on two buildings case study. The final section summarizes and discusses the results, limitations and prospects.

Section snippets

Extracting yearly weather data

Coordinated Regional climate Downscaling Experiment CORDEX (www.cordex.org) is an international coordinated effort supported by World Climate Research Programme's Working Group on Regional Climate. As a part of CORDEX, EURO-CORDEX is today the main reference framework for regional downscaling research of climate data. The main goals of this experiment are: (1) to evaluate and improve different RCMs, (2) to better understand regional and local climate phenomena through downscaling, (3) to

Comparative analysis of weather files

The objective of this sub-section is to analyse the differences in various climate models at monthly scale with different climate variables. Monthly statistical distribution of dry bulb temperature, global horizontal irradiance, and relative humidity of three DDS models, one ESD (meteonorm 2050), and 2003 observed weather data are shown in Fig. 11.

Observed weather of 2003 was selected for comparison because it was the severest heatwave recorded in France up until June 28, 2019, when a

Limitations and prospects

In this study, the raw data of all necessary climate variables used to construct typical weather data for the future are not bias-adjusted, therefore only a comparative assessment can be done with them. In addition, the thresholds used in heatwave detection, are fixed absolute thresholds for maximum and minimum daily temperatures and are based on the worst historical heatwave event. In future studies, for France or any other locations around the world, other methods of heatwaves measurement

Conclusions

In the first part of this paper, a methodology/workflow was described to generate yearly and typical ready-to-use weather files for BPS using EURO-CORDEX database of dynamically downscaled climate data of the past, present, and future.

We generated three FTWYs using this methodology, then compared them with future weather file of Meteonorm 2050, and measured weather data of 2003. Comparative analysis of future weather files showed a difference not only between the weather files that were

CRediT authorship contribution statement

Obaidullah Yaqubi: Writing – original draft, Validation, Methodology, Investigation, Formal analysis. Auline Rodler: Writing – review & editing, Conceptualization. Sihem Guernouti: Writing – review & editing, Conceptualization. Marjorie Musy: Writing – review & editing, Conceptualization.

Declaration of competing interest

The authors declare that they have no known conflicting financial interests or personal relationships that could have influenced the work reported in this paper.

References (50)

  • Amin Moazami et al.

    Impacts of future weather data typology on building energy performance - investigating long-term patterns of climate change and extreme weather conditions

    Appl. Energy

    (2019)
  • Mirata Hosseini et al.

    Generating future weather files under climate change scenarios to support building energy simulation - a machine learning approach

    Energy Build.

    (2021)
  • Thomas K. Thiis

    Monitoring and simulation of diurnal surface conditions of a wooden façade

    Procedia Environ. Sci.

    (2017)
  • Mayank Bhatnagar et al.

    Determining base temperature for heating and cooling degree-days for India

    J. Build. Eng.

    (2018)
  • Kathryn J. McCartney et al.

    Developing an adaptive control algorithm for Europe

    Energy Build.

    (2002)
  • Sanjay Kumar

    Comparative study of thermal comfort and adaptive actions for modern and traditional multi-storey naturally ventilated hostel buildings during monsoon season in India

    J. Build. Eng.

    (2019)
  • Dann Mitchell et al.

    The day the 2003 European heatwave record was broken

    Lancet Planet. Health

    (2019)
  • Daniel Fosas

    Mitigation versus adaptation: does insulating dwellings increase overheating risk?

    Build. Environ.

    (2018)
  • Wmo

    State of the Global Climate 2020

    (2020)
    (2021)
  • Ipcc
  • Sarah Doherty

    National Climate Assessment Report

    (2018)
  • Vérène Wagner

    Evolution of heatwaves and associated mortality in France, 2004-2014/Évolution des vagues de chaleur et de la mortalité associée en France, 2004-2014

    (2018)
  • Gabriele Manoli

    Magnitude of urban heat islands largely explained by climate and population

    Nature

    (2019)
  • Bruno Bueno et al.

    The urban weather generator

    J. Build. Perform. Simulat.

    (2013)
  • Lindita Bande

    Validation of UWG and ENVI-met models in an Abu Dhabi district, based on site measurements

    Sustainability

    (2019)
  • Cited by (5)

    View full text