Kubilay 2017

UCLIM-00299; No of Pages 21
Urban Climate xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Urban Climate
journal homepage: http://www.elsevier.com/locate/uclim
Coupling of physical phenomena in urban microclimate: A

model integrating air flow, wind-driven rain, radiation and
transport in building materials
A. Kubilay a,b,⁎, D. Derome b, J. Carmeliet a,b
a
Chair of Building Physics, Swiss Federal Institute of Technology ETHZ, Zurich, Switzerland
b
Laboratory for Multiscale Studies in Building Physics, Swiss Federal Laboratories for Materials Science and Technology (Empa), Dübendorf,
Switzerland
a r t i c l e i n f o a b s t r a c t
Article history: Building materials play an important role in the absorption, transport
Received 17 October 2016 and storage of heat and moisture in the built environment. A fully-inte-
Received in revised form 12 April 2017 grated urban microclimate model is proposed, which solves for wind
Accepted 28 April 2017
flow and for the transport of heat and moisture in the air and building
Available online xxxx
materials. The model includes long-wave and short-wave radiative ex-
change between surfaces and the distribution of wind-driven rain inten-
Keywords: sity. Transport in air and building materials are coupled in such a way
Urban microclimate
that the steady Reynolds-averaged Navier-Stokes (RANS) is solved iter-
Computational fluid dynamics (CFD)
atively with the unsteady heat and moisture transfer in building mate-
Porous media
Heat and moisture transport rials. The proposed approach provides the information required for
Radiation analyzing different contributions of convective cooling, sensible heat
Wind-driven rain transfer due to rain, evaporation, in addition to the thermal storage
throughout the day. This approach is demonstrated with a case study in-
vestigating the impact of rain deposition on an isolated three-dimen-
sional street canyon lined with porous building materials. The study
shows different rate of evaporation and duration of evaporative cooling
for a change in wind speed during a rain event. The distribution of wind-
driven rain particularly influences the spatial and temporal distribution
of surface and air temperatures. A significant influence of neighboring
surfaces is found on the surface temperatures.
© 2017 Elsevier B.V. All rights reserved.
⁎ Corresponding author at: Laboratory for Multiscale Studies in Building Physics, Swiss Federal Laboratories for Materials Science and
Technology (Empa), Überlandstrasse 129, CH-8600 Dübendorf, Switzerland.
E-mail address: aytac.kubilay@empa.ch (A. Kubilay).
http://dx.doi.org/10.1016/j.uclim.2017.04.012
2212-0955 © 2017 Elsevier B.V. All rights reserved.
Please cite this article as: Kubilay, A., et al., Coupling of physical phenomena in urban microclimate: A
model integrating air fl..., Urban Climate (2017), http://dx.doi.org/10.1016/j.uclim.2017.04.012
2 A. Kubilay et al. / Urban Climate xxx (2017) xxx–xxx
1. Introduction
Urban heat island (UHI), which describes the increase in temperature of the urbanized areas compared to
the rural one, is observed to intensify due to on-going urbanization and climate change. UHI has adverse ef-
fects on thermal comfort and health and leads to increases in energy use for space cooling and in greenhouse
gas emissions (Moonen et al., 2012). The main causes of UHI can be summarized as: 1) decreased long-wave
radiation loss to the sky due to increase in blockage, 2) increased sensible heat storage due to use of materials
with high thermal admittance, e.g. concrete, asphalt, brick, 3) increased absorption of short-wave radiation
due to increased reflections and use of low-albedo materials, 4) decreased evapotranspiration due to the re-
duced presence of soil, water bodies and vegetation, 5) increased anthropogenic heat production due to build-
ings and traffic, 6) decreased convective heat transport due to reduction in wind speed and 7) increased
absorption of long-wave radiation due to air pollution, as proposed by Oke (1982).
A considerable amount of research is currently performed with the aim to decrease the causes that lead to
UHI and to assess various mitigation measures. Several recent studies include extensive literature overviews
on microclimate (Moonen et al., 2012; Toparlar et al., 2015) and on UHI (Arnfield, 2003; Mirzaei and
Haghighat, 2010; Mirzaei, 2015) which mainly categorize the investigation methods in two general groups:
observational and numerical methods. Observational methods such as field measurements and thermal re-
mote sensing provide significant information on actual conditions in urban areas. Their main drawbacks are
that they are time-consuming and document only the meteorological conditions taking place during the mea-
surements. While it is technically possible to model different environmental conditions in wind-tunnel mea-
surements, such investigations can be very demanding to carry out. Nevertheless, observational methods are
widely used to validate and provide boundary conditions for numerical models. More frequently, UHI studies
resort to numerical methods. The first type of numerical models consists of urban canopy models (UCM),
which solve energy balance equations for a control volume representing typical urban geometries, e.g. two ad-
jacent buildings. UCMs do not model local urban airflow, which is necessary for an accurate prediction of sen-
sible and latent heat fluxes. Instead, such models mostly relate the convective transfer coefficients to a
reference wind speed based on empirical correlations (Masson, 2000; Kusaka et al., 2001; de Morais et al.,
2016) or drag coefficients (Martilli et al., 2002; Lemonsu et al., 2012). Furthermore, spatial variation of param-
eters is usually simplified by assuming uniform values over large surfaces, disregarding local effects that are
significant at street-canyon scale. An example of local effects is the local temperature difference between
air and building surfaces, which is responsible for buoyancy to occur and may thus significantly affect the
flow regime (Allegrini et al., 2013) as well as the local air temperature. Air temperature is also influenced
by a complex set of parameters such as prevailing wind speed and direction, building and urban environment
geometry, building materials and orientation with respect to the sun. To take into account such local effects,
the second group of numerical models is based on computational fluid dynamics (CFD), which can resolve
buildings and provide detailed information on the local spatial distribution of temperature and wind speed.
CFD methods are used commonly in various applications related to urban physics, such as wind comfort, ther-
mal comfort, pollutant dispersion, wind-driven rain (WDR) and ventilation (Blocken, 2014). However, efforts
in including all physics are yet to come. Current efforts are mentioned next.
The advantages of CFD are its ability to provide accurate spatial and temporal information and the given
possibility to perform parametric studies based on different scenarios. On the other hand, CFD simulations
are complex and demanding to perform, which often leads to considerable simplifications in the physics in-
volved in microclimate studies. For example, several numerical studies, which focus on the effect of various
urban geometries on the microclimate, use fixed surface temperatures or fixed surface heat fluxes as bound-
ary conditions (Kim and Baik, 2001; Xie et al., 2006). The spatial gradients of these values on the building sur-
faces, which exist even in the simplest cases, are not considered, nor are their variation during the day and the
influence of the thermal storage. However, taking into account heat transport in building materials when
modeling urban environments yields the heat stored during day-time due to solar heat flux and the heat re-
leased during night-time. These phenomena are the main causes of the UHI. Coupling heat transport in the air
with the heat transport in building materials, compared to the cases where a fixed temperature or a fixed heat
flux is imposed at building surfaces, gives more realistic boundary conditions for wind flow where buoyancy is
significant. Recently, several studies have coupled CFD simulations with heat conduction within the building
facades and the ground, using simplified models such as one-directional heat transport (Tominaga et al.,
2015; Toparlar et al., 2015) or surface energy-balance models (Li et al., 2005; Kaoru et al., 2011; Ma et al.,
A. Kubilay et al. / Urban Climate xxx (2017) xxx–xxx 3
2012; Qu et al., 2012). Also CFD simulations were coupled with building energy simulation (BES) models to
assess the relation between the microclimate and the energy demand of buildings (Bouyer et al., 2011;
Allegrini et al., 2012a). More recently, Allegrini et al. (2015) evaluated the impact of buildings on the local cli-
mate, based on the surface temperatures calculated with a BES model. These studies focused mainly on ther-
mal transport in air, except for Tominaga et al. (2015), and either neglected latent heat flux altogether or used
a simplified uniform latent heat flux at surfaces.
Moisture redistribution within the built environment is a further complicating factor in urban microcli-
mate, as moisture is the agent of evaporative cooling, while affecting the thermal capacity and conductivity
of building materials. Therefore, urban microclimate studies have to take into account wetting and drying
as well as heating and cooling of urban building materials, such as concrete, asphalt, clay brick, which play
an important role in the absorption, transport and storage of heat and moisture. Modeling moisture transport
in building materials allows to take into account the cooling provided by evaporation. Saneinejad et al. (2014)
presented a two-dimensional coupling of CFD simulations with a building-envelope heat-air-moisture
(BE-HAM) model to evaluate evaporative cooling and the resulting thermal comfort, where the building
facades are considered initially at moisture saturation due to rain. Such consideration of moisture is
particularly important because rain is the main moisture source in an urban environment. On the other
hand, uniform wetting is not realistic as WDR intensity undergoes a spatial distribution even in the
simplest cases (Blocken and Carmeliet, 2002; Kubilay et al., 2014), and is influenced by the geometry
of the building or building neighborhood, environment geometry, wind profile, speed and direction
and rainfall intensity (Blocken and Carmeliet, 2004). Modeling the actual surface wetting due to WDR
and the resulting moisture absorption will result in different wetting and drying conditions for building
materials lining the microclimate.
In the view of the current lack of holistic microclimate modeling approach, the present study proposes a
fully-integrated microclimate model that takes into account the interactions between various physical phe-
nomena including wind flow, solar radiation, buoyancy, rain, air temperature and relative humidity. This
way, detailed three-dimensional redistribution of heat and moisture is modeled within the urban areas
(Fig. 1). The model solves for the wind flow, the temperature and relative humidity in the air, and the temper-
ature and moisture content in building materials. The distribution of WDR intensity is calculated using an
Eulerian multiphase model (Kubilay et al., 2014, 2015b, 2017). Direct and diffuse solar radiation as well as
thermal radiation with diffuse reflections are taken into account. The heat and mass transport in air and build-
ing materials are coupled in such a way that the steady Reynolds-averaged Navier-Stokes (RANS) is solved it-
eratively with the unsteady heat and moisture transfer in building materials for each timestep on a typical
day. This model yields local flow patterns, air conditions and the spatial distribution of heat and moisture in
urban materials with higher resolution in comparison to more simplified UCM and CFD models. The model
allows for the detailed spatial analysis of the cooling effect on the surfaces and in the air, in addition to the dif-
ferent contributions to cooling, such as convective cooling, sensible heat transfer due to rain and evaporation.
Detailed evaluation of the local effects of short-term phenomena, e.g. heat waves, can be performed at
Fig. 1. Schematic of the main physics implemented in the coupled urban microclimate model. Air domain, modeling wind flow and wind-
driven rain (WDR), exchanges information with the HAM (heat and moisture transport in porous media) model, which iterates with the
radiation model.
neighborhood scale. In the present paper, a case study is performed for an isolated street canyon using the
proposed modeling approach. The daily variations of surface temperatures in a street canyon, in dry and in
rain-wetted states, are compared for two different reference wind speeds.
The remainder of the paper is organized as follows: Section 2 describes the numerical models for the dif-
ferent physics involved in the study. The governing equations and the numerical settings are described in de-
tail. Furthermore, the coupling algorithm is explained. Section 3 describes the details of the case study and the
material properties. In Section 4, the results of the study are presented. Finally, Section 5 provides a conclusion
and perspectives.
2. Description of the numerical model and the coupling strategy
2.1. Numerical model for wind flow
Steady RANS with the standard k-ε model (Launder and Sharma, 1974) is chosen for wind-flow simula-
tions. In general, RANS k-ε models have drawbacks related to the overprediction of the size of the wake and
the location of reattachment due to the underestimation of turbulence kinetic energy in the wake of the build-
ing (Murakami, 1993). On the other hand, more complicated turbulence models such as large eddy simulation
(LES) result in a higher computational cost, especially considering the three-dimensional coupling and daily
simulations intended in the present modeling approach. In addition to RANS k-ε equations, governing equa-
tions for heat (as an active scalar) and moisture (as a passive scalar) transport are solved in the air domain.
The governing equations in the air domain use a compressible formulation of the Navier-Stokes equations.
Thus, the buoyancy is taken into account by directly calculating the density variations based on temperature
and pressure, instead of using the Boussinesq approximation, where the density is considered constant except
for the buoyancy term.
2.2. Numerical model for wind-driven rain (WDR)
2.2.1. Governing equations for WDR

WDR intensity is calculated with an Eulerian multiphase (EM) model (Huang and Li, 2010; Kubilay et al.,
2013). In the EM model, the rain phase is regarded as a continuum, as is the wind phase. Each raindrop-size
class is treated as a different phase, as raindrops of similar size will interact with the wind-flow field in a sim-
ilar way. Rain phase calculations are one-way coupled with the air phase, given that the effect of raindrops on
the wind flow is ignored. This is a valid assumption as the volumetric ratio of rain in air is below 1 × 10−3 for
rainfall intensities up to 20 mm/h and below 1 × 10−2 for even the most severe cases according to de Wolf
(2001). For each rain phase, the following continuity and momentum equations are solved (Kubilay et al.,
2013):
∂α d ∂α d ud; j
þ ¼0 ð1Þ
∂t ∂x j
3μ a C d ReR
0 0
∂α d ud;i ∂α d ud;i ud; j ∂α d u d;i u d; j
þ þ ¼ α d gi þ α d ui ud;i ð2Þ
∂t ∂x j ∂x j ρw d 2 4
where αd is the phase fraction of rain phase d, which represents a specific class of raindrop size, ud,j denotes
the velocity component of rain phase d, ui the velocity component of wind in direction i (where i = 1, 2, 3),
ρw the density of the raindrops, μa the dynamic air viscosity, gi the gravitational acceleration, Cd the drag co-
efficient. ReR denotes the relative Reynolds number calculated using the relative velocity between the air and
rain phases. The terms on the right-hand side represent the gravity and the drag forces. Drag coefficients are
obtained based on the measurements of terminal velocities of water droplets free falling in stagnant air by
Gunn and Kinzer (1949). The third term on the left-hand side in Eq. (2) corresponds to the turbulent mass
flux, which is the transport of mass due to turbulent motions in the flow. Turbulent dispersion of droplets
is due to these turbulent motions. By defining a response coefficient, Ct, velocity fluctuations in rain phases
can be related to the velocity fluctuations in the wind:
0 0
2 u d;i u d; j 1
Ct ¼ ¼ ð3Þ
u iu
0 0 tp
j 1þ
t fl
where tp denotes the particle relaxation time and tfl denotes the Lagrangian fluid time scale. As the particle
relaxation time gets smaller compared to the Lagrangian fluid time scale, e.g. for smaller raindrops, Ct ap-
proaches the value of 1. In such cases, raindrops are more influenced by the fluctuations in wind velocity.
For a more detailed description of the turbulent dispersion modeling of raindrops, the reader is referred to
Kubilay et al. (2015a).
The EM model requires less computational effort compared to Lagrangian particle tracking models which
can require very high number of raindrops to achieve accurate results (Blocken and Carmeliet, 2007) especial-
ly in the case of geometries involving multiple buildings. Furthermore, the EM model allows the estimation of
the WDR intensity and the raindrop impact parameters, such as impact speed and impact angle, on all surfaces
even in complex geometries. The EM model is validated against field measurements on various geometries
during rain events with different characteristics (Kubilay et al., 2014, 2015b). The WDR solver is implemented
into OpenFOAM (2015), which is an open-source CFD package, and can be accessed at the following website:
http://www.carmeliet.arch.ethz.ch/ResearchDatabase/WindDrivenRainFoam
2.2.2. Numerical settings for WDR

The EM model requires the definition of the values of rain-phase fraction, αd, and the rain-phase velocity,
ud,j, at the inlet and top boundaries for each rain phase d. The phase fraction can be defined as:
Rh f h ðRh ; dÞ
αd ¼ ð4Þ
V t ðdÞ
where Vt(d) represents the terminal velocity of a raindrop with diameter d, Rh the horizontal rain intensity
through the horizontal plane, fh(Rh,d) the raindrop-size distribution through the horizontal plane (Blocken
and Carmeliet, 2004). In the present study, the raindrop-size distribution by Best (1950) is considered,
which uses an empirical formula based on a wide bibliographical survey and on measurements for a large
number of rain events at various locations. The terminal velocities are based on the experimental study of
Gunn and Kinzer (1949).
The vertical component of rain-phase velocity is set equal to the terminal velocity for that phase at the
boundaries. The horizontal component of rain-phase velocity is set such that there is a local force equilibrium
between rain and wind phases so that the droplets neither accelerate nor decelerate in an unobstructed com-
putational domain. The boundary conditions for the rain phases at the building walls, on the ground and at the
outlet are set in such way that the normal gradient of the volumetric ratio, ∂αd/∂n, equals zero when the nor-
mal wind velocity vector is pointing out of the domain, and the values of the volumetric ratio, αd, are equal to
zero when the normal wind velocity vector is pointing into the domain. With these boundary conditions at
walls, interactions between the raindrops and the walls, e.g. splashing, bouncing, are not modeled.
2.3. Numerical model for coupled heat and moisture transport in building materials
Absorption, transport and storage of heat and moisture are simulated within the building materials using
coupled heat and moisture transport equations for porous media. A porous material consists of a solid mate-
rial matrix and a pore space (porosity), which can be filled with gas (water vapor or dry air) or liquid. In the
present study, the continuum modeling approach (Whitaker, 1977) is applied, where the different phases are
not distinguished separately at a certain point in the material but, instead, the macroscopic behavior of the
porous material is modeled.
The governing equations for moisture transport within the porous domains are given in Eq. (5a), where gl
and gv denote liquid and vapor moisture transfer and are defined in Eqs. (5b) and (5c), respectively.
∂w ∂pc
¼ −∇ðg l þ g v Þ ð5aÞ
∂pc ∂t
g l ¼ −K l ∇pc ð5bÞ
δv pv δ p
gv ¼ − ∇pc − v v2 ðρl Lv −pc Þ∇T ð5cÞ
ρl RT ρl RT
In Eqs. (5a)–(5c), w denotes moisture content, pc capillary pressure, T absolute temperature, Kl the liquid
permeability, δv the vapor permeability, pv the vapor pressure, ρl the liquid density, R the gas constant and Lv
the heat of vaporization. The derivative ∂w/∂pc represents the moisture capacity, i.e. the derivative of the
moisture retention curve which gives the moisture content w as a function of the capillary pressure pc.
Note that Kl and δv are dependent on moisture content and are obtained by using experimental techniques
(Carmeliet et al., 2004).
The governing equations for heat transport within the porous domains are given in Eq. (6a), where qc and
qa denote conductive and advective heat transfer and are defined in Eqs. (6b) and (6c), respectively.

∂T ∂w ∂pc
ðc0 ρ0 þ cl wÞ þ cl T ¼ −∇ðqc þ qa Þ ð6aÞ
∂t ∂pc ∂t
qc ¼ −λ∇T ð6bÞ
qa ¼ ðcl T Þg l þ ðcv T þ Lv Þg v ð6cÞ
In Eqs. (6a)–(6c), c0 denotes the specific heat of the dry material, cv the specific heat of water vapor, cl the
specific heat of liquid water, ρ0 density of the dry material, and λ the thermal conductivity. The advective heat
transfer, qa, represents advective heat flow due to vapor and liquid flow including latent heat transport. A de-
tailed description of the derivation of the governing equations and the underlying assumptions can be found
in Janssen et al. (2007) and Defraeye (2011).
The heat and moisture fluxes calculated in the air domain are set as boundary conditions for the exterior
surfaces of the building materials. The moisture exchange at the exterior surfaces of building materials, gext,
comprises the convective vapor exchange, gconv, and the rain flux, Rwdr, as shown in Eq. (7). The heat ex-
change, qext, comprises the convective heat transfer, qconv, the radiative heat transfer, S, the sensible heat
transfer due to rain, Rs, and the latent and sensible heat transfer due to vapor exchange, qls, as shown in
Eq. (8), where Twb denotes the wet-bulb temperature and Ts the surface temperature.
g ext ¼ g conv þ Rwdr ð7Þ
qext ¼ qconv þ S þ Rs þ qls ; Rs ¼ cl T wb Rwdr ; qls ¼ ðcv T s þ Lv Þg conv ð8Þ
The WDR flux, Rwdr, is calculated with the WDR model in Section 2.2. The calculation of radiative heat flux,
S, is given in Section 2.4. Note that the convective heat and mass fluxes, qconv and gconv, are the values obtained
in the CFD calculation. The maximum moisture content in the building materials is the capillary saturated
moisture content. In the present model, film forming and runoff on the exterior surfaces are not modeled,
which is a valid simplification as long as the exterior surfaces do not reach the capillary saturated moisture
content, which is the case here. Further extension of the model is required to account for these phenomena.
The transport model described in Eqs. (5a)–(5c) and (6a)–(6c) is implemented into OpenFOAM. The model
is validated with existing results of moisture uptake experiments on porous stones (Derluyn et al., 2013)
and verified by comparing with HAMSTAD (Heat Air and Moisture STAndards Development) benchmark
cases (Hagentoft et al., 2004) which are specifically developed for moisture transport in building materials.
2.4. Numerical model for radiative heat transfer
Short-wave (solar) and long-wave (thermal) radiation heat transfer are modeled based on a radiosity ap-
proach. The model considers the radiation exchange between the domain surfaces and with the sky. Air is con-
sidered as a non-participating medium, i.e. absorption, scattering and emission of radiation by air are
neglected. All surfaces are assumed to be opaque to both long-wave and short-wave radiation, i.e. windows
on building facades are not considered and, therefore, transmissivity is zero. The model further assumes
that surfaces are gray and diffuse, i.e. emissivity and absorptivity are equal and independent of wavelength
and direction. The following equations are solved to calculate the net radiative heat flux at boundaries:
N
L 4 L
qout;k ¼ εk σT k þ ρk ∑ F kj qout; j ð9Þ
j¼1
N
S S
qout;k ¼ −ð1−ak ÞI sol þ ak ∑ F kj qout; j ð10Þ
j¼1
where qLout,k denotes the energy leaving the surface k due to long-wave radiation and qSout,k due to short-
wave radiation, εk the emissivity of surface k, σ the Stephan-Boltzmann constant, Tk the surface tem-
perature, ρk the reflectivity of surface k, ak the albedo, i.e. short-wave reflectivity, of surface k. Fkj de-
notes the view factor between surfaces k and j, i.e. the fraction of energy leaving surface j that is
incident on surface k. View factors between surfaces are calculated once at the beginning of the simu-
lation using OpenFOAM based on an algebraic formula. Isol denotes the total solar radiation incoming to
surface k, which is composed of direct, Idir, and diffuse, Idiff, components:
B
Isol ¼ Idir þ Idiff ; Idir ¼ I DN cosθ; IDN ¼ Ae sinβ ; Idiff ¼ CIDN F ss ð11Þ
where IDN denotes the direct normal radiation, θ the incidence angle between the incoming solar rays and the
receiving surface normal, β the solar altitude and Fss the angle factor between surface and sky. A, B and
C are constants depending on the time of the year (ASHRAE, 1985). The direct component of incoming
solar radiation, Idir, is calculated with ray tracing and takes into account the variation of shading and
incidence angle over the day.
The sky temperature, Tsky, is calculated based on the cloud cover and the ambient temperature, Ta, as fol-
lows (Cole, 1976):
εc ¼ ð1−0:84cÞð0:527 þ 0:161 exp½8:45ð1−273=T a ÞÞ þ 0:84c ð12aÞ

4 −6 6 4
T sky ¼ 9:365574 10 ð1−cÞT a þ T a cεc ð12bÞ
where c denotes the cloudiness and εc the cloud emissivity.
The system of linear equations in Eqs. (9) and (10) allows for an infinite number of reflections between
surfaces. In Eqs. (9) and (10), the net radiative heat flux on surface k is calculated by the net emitted radiation
from surface k and the total reflected radiation that comes from all other surfaces j. Reflections due to both
long-wave and short-wave radiation are assumed to be diffuse.
2.5. Solution algorithm
OpenFOAM 2.4 is used as the CFD solver with three main add-ons that are implemented by the authors:
the WDR solver, the solar radiation model and the governing equations for the coupled heat and moisture
transport in porous materials. The coupling between the air and porous domains is performed by sequentially
solving the steady governing equations in the air and the unsteady governing equations in the porous do-
mains for each exchange timestep. The exchange timestep is defined as the timestep where the information
is exchanged at the air/surface interfaces (Saneinejad et al., 2014). This approach is valid due to the fact that
the time scale of transport in building materials is larger than the time scale of transport in air. During the
transient simulation within the solid porous materials, the solution of air domain remains constant. Therefore,
the exchange timestep is chosen to be 10 min in this study, as an hourly exchange time step is too long.
For each exchange timestep, first, the steady air flow is solved. The pressure-velocity coupling for the
wind-flow solution is done with the Semi-Implicit Method for Pressure Linked Equations (SIMPLE) algorithm.
Second order discretization schemes are used for both the convection terms and the viscous terms of the
governing equations. The simulations in the air domain are terminated when all the scaled residuals reach
a value which is below the following values: 10−5 for all velocity components, turbulence terms and continu-
ity, 10−4 for heat and moisture. Afterwards, the calculated heat and moisture fluxes at the street-canyon
boundaries in the air domain are applied as boundary conditions in the coupled porous domains representing
the building materials (step 1 in Fig. 1). Transient heat and mass transport in porous domains are simulated
using adaptive timesteps (Janssen et al., 2007). For this, an internal iteration is performed, during which the
thermal radiative heat fluxes are updated until temperature and moisture content values converge (step 2 in
Fig. 1). Finally, the new values for temperature and moisture at the boundaries of the porous domain are used
to solve the steady air flow for the next exchange timestep (step 3 in Fig. 1). The calculation time for a 24 h
period took between 40 and 50 h in a parallel run using 64 processors. Approximately 90% of the total calcu-
lation time was spent for the CFD calculations in the air domain.
3. Description of the case study
The present study compares the microclimatic conditions within an isolated street canyon under dry and
wet conditions and different wind speeds to demonstrate the capability and the strength of the proposed
coupled urban microclimate model. Specifically, the impact of rain on microclimate is investigated by compar-
ing the resulting thermal conditions due to rain to those of a dry day. The governing equations for wind, heat,
moisture and WDR are solved in the air domain. The transport of heat and moisture in the building materials
are solved in separate computational domains.
3.1. Computational domain and grid for air
The isolated street canyon is composed of two buildings with flat facades and horizontal roofs. Both build-
ings have the dimensions of height × length × width of 10 × 10 × 50 m3. The computational domain for air
and the orientation of the street canyon with respect to the sun are shown in Fig. 2(a). For simplicity, wind
speed and wind direction are assumed constant during the day. Wind direction is from due west, so perpen-
dicular to the street canyon. The distances of buildings to the air-domain boundaries satisfy the guidelines
stated in Tominaga et al. (2008) and Franke et al. (2011). After sunrise, the solar rays first hit the leeward
wall of the street canyon, while the street-canyon ground and the windward wall are shaded. Around the
solar noon, the ground is receiving solar rays with an angle close to the normal. Later, the solar rays hit the
windward wall of the street canyon.
The computational grid for the air domain is generated following a grid-sensitivity study. Five structural
grids with hexahedral cells are created by refining the grid further by a constant ratio at each step. Constant
temperature boundary conditions are applied on the street-canyon surfaces for the grid-sensitivity study. The
lateral and vertical profiles of magnitude of wind speed are compared at the center of the street canyon. The
grid consisting of 1,178,080 cells is selected as further refining resulted in a difference of only 3.1% along the
horizontal line and 1.3% along the vertical line on average. The first cell height on the walls is 10 cm.
The inlet profiles of mean wind speed U, turbulence kinetic energy k and turbulence dissipation rate ε are
shown in Eq. (13) following the definitions by Richards and Hoxey (1993), where U(y) denotes the horizontal
wind speed at height y, u*ABL the atmospheric-boundary-layer friction velocity, y0 the aerodynamic roughness
length, κ the von Karman constant and Cμ (=0.09) is a model constant. The log-law expression for the inlet
mean wind speed profile is characterized by the normalized expression u*ABL/U = 0.072 and y0 is chosen
0.03 m, representing a levelled country with low vegetation, e.g. grass, and isolated obstacles (Wieringa,
Fig. 2. a) Computational domain and orientation of the street canyon with respect to wind direction and the sun. b) Computational grid for
air domain showing building surfaces and part of the ground (1,178,080 cells).
1992). The same value of u*ABL is used for the profiles of turbulence kinetic energy k(y) and turbulence dissi-
pation rate ε(y).

uABL y þ y0 uABL 2 uABL 3
U ðyÞ ¼ ln ; kðyÞ ¼ q ffiffiffiffiffiffi ; εðyÞ ¼ ð13Þ
κ y0 C κ ðy þ y0 Þ
μ
For wall treatment, standard wall functions by Launder and Sharma (1974) are used. For the ground sur-
face, sand-grain roughness modification (Cebeci and Bradshaw, 1977) is considered. The equivalent sand-
grain roughness height, ks, is taken to be 0.03 m and the roughness constant, Cs, is set as 9.793. Building sur-
faces are assumed smooth. Note that the standard wall functions have limitations in terms of the accuracy of
predicted heat flux and several studies suggested improvements by modifying the turbulent Prandtl number,
Prt (Defraeye et al., 2011; Allegrini et al., 2012b). For the top boundary, constant values are set for U, k and ε by
using the values from the inlet profiles at the same height as suggested by Blocken et al. (2007) in order to
limit the horizontal inhomogeneity. A constant static gauge pressure of 0 Pa is used at the outlet boundary.
Symmetry conditions are applied on both sides of the domain.
3.2. Computational domain and grid for building materials
Heat and moisture transport is considered within the materials lining the street canyon, i.e. the leeward
facade, the street and the windward facade. As the main focus of the present study is the microclimate within
the street canyon, the surfaces outside the street canyon are assumed to be adiabatic and impermeable. The
leeward and windward facades of the street canyon are finished with a 0.09 m layer of clay brick masonry
as shown in Fig. 3(a). The interior surfaces are modeled with a thermal resistance of 2.5 m2 K/W. An indoor
temperature of 20 °C is chosen. The interior surfaces of the facades are assumed to be impermeable to mois-
ture. The outer surfaces of the facades have an emissivity of 0.9 and an albedo of 0.4. The outer layer of the
street-canyon ground is assumed to consist of 0.10 m clay brick, which is chosen to emphasize moisture ab-
sorption in comparison to asphalt or concrete. Beneath the brick layer, soil with a depth of 1.90 m is located.
Soil is chosen to consist of a balanced proportion of sand, silt and clay. Soil temperature at the bottom is set to
10 °C. The bottom surface of the soil layer is assumed to be impermeable to moisture. The outer surface of the
ground has an emissivity of 0.9 and an albedo of 0.2, representing a dark colored material. The lateral bound-
aries of the porous material layers are assumed to be adiabatic and impermeable.
The cross-sections of the computational grids for the street-canyon porous domains in the direction be-
tween exterior and interior surfaces are given in Fig. 3(b). The computational grids for windward and leeward
walls have cell refinement towards the interior and exterior boundaries. They consist of 62,500 cells in total
Fig. 3. a) Configuration of the 3D models for the street-canyon facades and the ground (not to scale). b) Cross-section of the computational
grids for the street-canyon facades (62,500 cells) and ground (125,000 cells).
Table 1
Thermal properties of the dry materials used in porous media.
Material Density [kg/m3] Specific heat [J/kg K] Thermal conductivity [W/m K] Emissivity [−] Albedo [−]
Brick (facades) 1600 1000 0.682 0.9 0.4

Brick (ground) 2280 800 0.682 0.9 0.2
Soil 1150 650 1.500 – –
(20 in x-direction, 25 in y-direction and 125 in z-direction). The computational cells are uniform in y and z
directions. The computational grid for the ground domain has cell grading towards the exterior surface and
at the interface between different materials. It consists of 125,000 cells in total (25 in x-direction, 40 in y-di-
rection and 125 in z-direction). The computational cells are uniform in x and z directions. Thermal properties
for the dry materials are given in Table 1 (Hagentoft et al., 2004) and the moisture properties such as moisture
retention and the liquid permeability are found in Fig. 4.
3.3. Atmospheric conditions
For the first case, the domain is subjected to meteorological conditions on a dry day, thus without rain, at
typical early summer, with moderate ambient temperatures. The meteorological data are based on a typical
meteorological year (Meteonorm, 2000) and total solar radiation intensity for clear sky (ASHRAE, 1985) for
21st of June in the city of Zurich, Switzerland. Typical meteorological year represents a collection of standard-
ized data obtained from long-term mean values. The simulations for the dry case have been run initially for
several days using the meteorological data for 21st of June as boundary conditions until the conditions
reach a daily thermal cycle that is independent from the initial values. Two approach wind speeds are consid-
ered: Uref = 2 and 5 m/s at the building height. Fig. 5 shows the variations of ambient temperature, relative
humidity and the solar radiation. The solar noon occurs at around 13:28, considering daylight saving time.
The maximum ambient temperature is around 19 °C, while the minimum is around 11 °C. The relative humid-
ity varies between 62% and 86%.
For the wet case, a single rain event is considered. The rain event starts at sunset and continues uniformly
until sunrise for a duration of 10 h (between 19:30 and 05:30) as indicated in Fig. 5(a). The horizontal rainfall
intensity, i.e. the rainfall intensity through a horizontal plane, is assumed to be constant at 1 mm/h. The dis-
tribution of WDR intensity on surfaces, i.e. building facades and ground, is calculated using the Eulerian mul-
tiphase model discussed in Section 2.2. All the other conditions are the same as in the dry case. The resulting
WDR intensity, or the liquid flux reaching each surface within the street canyon, is applied as a boundary con-
dition described by Eq. (7). For a total rainfall amount of 10 mm, none of the surfaces reach the capillary sat-
urated moisture content. The initial values for the simulation of the rain event are the state at 19:30 calculated
for the dry day. Afterwards, the simulations for the rainy case are run until the end of the following day. The
overview of all the simulations is given in Table 2.
Fig. 4. a) Wetting moisture retention and b) liquid permeability for clay brick and soil.
Fig. 5. The daily variation of a) total solar radiation intensity and b) ambient temperature and relative humidity.
4. Results
4.1. Analysis of climatic loading
Initially, the spatial variation of thermal conditions is discussed for the street-canyon surfaces. Fig. 6 shows
the distributions of net short-wave radiation and surface temperatures during the dry and wet cases on the
surfaces of the street canyon at different times of the day with respect to the solar noon (SN). The windward
and leeward facades and the street surface are shown using a perspective view from above the street canyon.
The distribution of surface temperature depends mainly on the short-wave radiation heat transfer when there
is sunlight on the surface. The remaining factors are the heat exchange due to convection, reflections within
the street canyon and the heat stored in the building materials. Fig. 6 shows temperature patterns with strong
gradients along the height and width in addition to the strong effects of shadowing. The highest surface tem-
peratures are mainly observed towards the center of the facades. The lower surface temperature at the upper
and side edges of the buildings are due to the larger convective heat transfer and larger thermal radiation
losses, inducing more cooling at those parts. Fig. 6 also shows a clear distinction between the surface temper-
atures for cases with and without rain. The influence of rain on the surface temperatures can be observed, es-
pecially on the street surface, even long time after the rain event stops as shown in Fig. 6(e–f).
Large differences are observed in the distribution of WDR intensity at different reference wind speeds due
to the interaction of raindrops with wind-flow patterns. The distribution of WDR intensity on the surfaces of
Table 2
Overview of the simulations in the case study.
Wind speed Rain event Rain duration Simulation duration
5 m/s No – 0 h–48 h
5 m/s Yes 19.5 h–29.5 h 19.5 h–48 h
2 m/s No – 0 h–48 h
2 m/s Yes 19.5 h–29.5 h 19.5 h–48 h
Fig. 6. The distribution of net short-wave radiation and temperature on the street-canyon surfaces for the dry and wet cases with a per-
spective view from top at time instants with respect to the solar noon (SN): a) t = SN − 12 h, b) SN − 6 h, c) SN, d) SN + 2 h, e) SN + 4 h
and f) SN + 6 h for Uref = 5 m/s on June 21 (α: solar azimuth, β: solar altitude).
the street canyon is shown in Fig. 7 for a rainfall intensity of 1 mm/h. The street surface is wetted the highest,
whereas the leeward wall is sheltered. As wind speed increases, wetting increases on the windward facade
and decreases on the street surface. For Uref = 2 m/s, surface-averaged WDR intensity is about 0.92 mm/h
on the street, 0.06 mm/h on the windward facade and 0 mm/h on the leeward facade. For Uref = 5 m/s,
0.81 mm/h on the street, 0.16 mm/h on the windward facade and 0 mm/h on the leeward facade. Because
of the higher wetting of the street surface, the decrease in street temperature lasts longer compared to the
other surfaces as seen in Fig. 6. The gradients in the spatial distribution of WDR intensity is a result of the in-
ertia of different droplet sizes. Larger droplets have higher inertia and they are less influenced by the local
wind-flow patterns within the street canyon. As a result, the WDR intensity on the street surface has larger
values near the windward facade. This is because of the fact that only smaller droplets can reach the portion
of the street surface near the leeward facade, which is sheltered for larger droplets. There is also a large gra-
dient of WDR intensity on the windward facade, where the upper and side edges are exposed to higher WDR
amounts, especially for Uref = 5 m/s. This is a result of larger wind speed around the edges of the building.
The daily variation in surface temperatures has an influence on the local wind-flow patterns within the
street canyon. The variation of the wind-flow field is shown in Fig. 8 in the vertical centerplane and the
Fig. 7. a) The distribution of WDR intensity on the street-canyon surfaces (windward facade, street and leeward facade) with a perspec-
tive view from top for a) Uref = 2 m/s and b) 5 m/s.
resulting air temperature difference with the ambient air is compared for the dry and wet cases for Uref = 5 m/
s. The upward air motion due to buoyancy outwards the street canyon is clearly visible. At t = 05:00, there is a
small vortex next to the windward facade at the bottom, which is also visible in the temperature distribution
for the dry case in Fig. 8(b). At t = 11:00, air temperature distribution indicates an increase in surface temper-
atures at the ground near the leeward facade, which changes the wind-flow field by pushing the vortex fur-
ther towards the windward facade. This is due to the street surface being partially in the shade. As the surface
temperatures of the windward facade increase in the late afternoon (t = 17:00), buoyancy effects get stron-
ger and the vortex completely disappears. For the dry case, the highest temperature difference with the am-
bient is observed at t = 17:00 by about 5 °C near the leeward facade. For the wet case, the air temperatures are
lower by up to 2–3 °C compared to the dry case. The influence of rain on air temperature is clearly visible even
at t = 11:00, 5.5 h after the rain event stops.
4.2. Analysis of heat fluxes
Here, the heat fluxes at the street-canyon surfaces due to their interactions with the outside environment
are presented. Fig. 9 shows the variation of the radiative, convective and latent heat fluxes on different street-
canyon surfaces before, during and after the rain event for Uref = 5 m/s. Positive values denote heat gains
Fig. 8. a) Wind-flow field and temperature distribution for the b) dry and c) wet cases in the vertical centerplane for a reference wind
velocity Uref = 5 m/s at building height.
Fig. 9. Variation of the radiative, convective and latent heat fluxes on a) leeward facade, b) street surface and c) windward facade before,
during and after the rain event for Uref = 5 m/s.
towards the urban surfaces. Peaks in heat fluxes throughout the day occur in an order that follows the move-
ment of the sun around the street canyon, i.e. first peak occurs at the leeward facade, followed by peaks at the
street surface and then at the windward facade. Note that there is a delay of 1–2 h between the peak positions
of solar radiation and the ones of thermal radiation and convective heat flux as a result of the dynamics of heat
storage. At all surfaces, the main heat gain is due to the solar radiation, while the remaining heat fluxes act as
heat-removal mechanisms. Before the rain event, the removal of heat provided by convection is found to be
mostly larger than the amount due to thermal radiation. At the beginning of the rain event, the cooling due
to latent heat flux increases sharply at the street and the windward facade, which slowly decreases as the
rain event goes on. This is due to a decrease in drying rate caused by the accompanying increase in local rel-
ative humidity in the air and also due to the overall decrease in surface temperature during the rain event. The
main effect of rain is observed due the latent heat flux at the street surface, which has its peak after the rain
event stops (Fig. 9(b)). This is due to the fact that the solar radiation enhances evaporation and, as a result,
latent heat flux gets as large as the half of solar radiative flux. The decrease in surface temperature leads to
a decrease in thermal radiation and convective heat flux compared to the conditions before the rain event.
The variations of heat fluxes for Uref = 2 m/s show similar behavior. In general, the decrease in wind speed
Fig. 10. Variation of the subsurface heat flux at the street-canyon surfaces before, during and after the rain event for Uref = 5 m/s.
leads to a decrease in convective heat flux (by up to 32%) at the street canyon surfaces and as a consequence to
larger thermal radiation (by up to 22%). For each surface in Fig. 9, during most of the day-time, the solar radi-
ation input is larger than the sum of the heat removal due the remaining mechanisms. On the other hand, dur-
ing the night-time, there is a net negative heat flux which indicates cooling.
This net heat flux at the surface can also be confirmed by the subsurface heat flux, i.e. the total heat flux at
the exterior boundaries of porous domains, which governs the heat storage and release in the street-canyon.
Fig. 10 shows the variation of the subsurface heat flux. Negative values in Fig. 10 denote heat storage to the
building materials. Note that the subsurface heat flux changes direction for all surfaces during the day-
night cycle. Specifically, heat is stored in the street canyon starting at about t = 06:00 until afternoon, after
which heat is removed by air or lost by long-wave radiation. Heat storage to street decreases significantly
after the rain event due to evaporative cooling. The variation shows similar behavior for Uref = 2 m/s,
which results in up to 15% more heat storage to building materials.
4.3. Analysis of temperature and moisture content
This subsection focuses on the changes at specific positions on the street-canyon surfaces. Fig. 11 presents
the daily variation of exterior surface temperatures at the center and at the sides of the street-canyon surfaces
for the dry case for Uref = 2 m/s. Note that only the first 24-hour period is shown, as the following 24-hour
period is identical for the dry case. At the center of the surfaces in Fig. 11(b), the first peak occurs for the lee-
ward wall as the temperature increases to 39 °C. Later around the solar noon, the highest temperature occurs
on the street among all the other surfaces (69 °C) due to the lower albedo value of the street surface. Similar
observations are made for Uref = 5 m/s but peak temperatures are lower by about 5–15 °C. Finally, during late
in the afternoon, the peak temperature on the windward facade (53 °C) occurs. The peak temperature ob-
served at the center of the leeward facade is about 13 °C lower than the one on the windward facade. A sec-
ondary peak is visible on the leeward facade, and also to a lesser extent on the street surface. Such secondary
peaks are due to thermal radiation exchange and reflections between the street canyon surfaces. At the south
side of the street canyon in Fig. 11(a), the surface temperatures are in general lower than the ones at the cen-
ter of the facades. The peak temperatures are considerably lower especially for the street surface, windward
facade and, to a lower extent, leeward facade. The lower surface temperatures at the side are due to larger
thermal radiation exchange with the environment and due to the cooler air coming through the sides of
the street canyon. Note also that the peaks are not as sharp as at the center of the surfaces and they cover a
longer time span. This is due to the fact that there is less shadowing from solar radiation at the southern
side compared to the center of the street canyon. At the north side of the street canyon in Fig. 11(c), the tem-
perature range is similar to the one at the south side although the shapes of the curves are different. This dif-
ference is due to fact that solar rays come from northeast-east after sunrise and from northwest-west before
sunset.
Surface temperature and moisture content can vary significantly as a result of the rain event. The varia-
tions of exterior surface temperature and moisture content before, during and after the rain event at the
Fig. 11. Exterior surface temperatures for the dry case at the center of each street-canyon surface a) at the south side, b) at the center and
c) at the north side for Uref = 2 m/s.
center of different street-canyon surfaces are shown in Fig. 12 for Uref = 5 m/s. For the street surface in Fig.
12(b), as the rain event starts, the moisture content increases rapidly. This is accompanied by the decrease
in the surface temperature due to evaporative cooling and sensible heat transfer due to rain. The resulting
temperature decrease on the street surface is 4–5 °C during the rain event as shown in Fig. 12(b). Once the
rain event stops, the surface moisture content starts to decrease. As solar radiation increases during the
day, the rate of evaporation increases, resulting in a decrease of 25 °C in surface temperature. After about
t = 38 h, street surface is shaded. The small variations in surface moisture content after t = 38 h are due to
the differences between the rate of evaporation and the rate of moisture transport inside brick. Fig. 12(c)
shows the same analysis for the center of the windward facade. The increase in surface moisture content is
smaller, which leads to about 3 °C lower surface temperatures during the rain event. After the rain event
ends, moisture evaporates completely within a few hours due to much lower surface wetting of the windward
facade compared to the street surface. Nevertheless, the surface temperature remains at about 2 °C lower
compared to the conditions before the rain event due to radiation exchange with the cooler street surface.
A similar effect is observed on the leeward surface as shown in Fig. 12(a). Even though the leeward facade
is almost not wet at all, there is a difference of up to 2 °C before and after the rain event.
The vertical variations of the surface temperature and moisture content on the windward facade are pre-
sented in Fig. 13, considering the large vertical gradient in WDR intensity as shown in Fig. 7(b). The top part of
Fig. 12. Variation of surface temperature and moisture content at the centers of the a) leeward facade, b) street surface and c) windward
facade during and after the rain event (solid lines) in comparison with the dry day (dashed lines) for Uref = 5 m/s.
the facade is exposed to a higher amount of WDR and, as expected, the increase in surface moisture content is
considerably higher during the rain event. This leads to a decrease of 6 °C in maximum surface temperature
compared to the conditions before the rain event. On the contrary, the bottom of the windward facade is ex-
posed to a much smaller amount of WDR. Nevertheless, the decrease in surface temperature compared to the
dry case reaches up to 13 °C. This is due to the lower air temperature near the street surface, which is cooled
down by evaporative cooling, combined with radiative heat exchange with the cooler street surface. In com-
parison, the smallest influence of WDR on the windward facade in terms of surface temperature is found at
the center of the facade, which is about 2 °C compared to the conditions before the rain event.
4.4. Impact of wind speed on evaporative cooling by rain
Fig. 14 evaluates the impact of wind speed on the evaporative cooling by showing the differences in sur-
face temperature and moisture content following the rain event. For the street surface, the cooling effect peaks
in the afternoon (t = 38 h), where the decrease in surface-averaged temperature reaches 20.5 °C for Uref =
Fig. 13. Variation of a) surface temperature and b) moisture content at different heights on the windward facade during and after the rain
event (solid lines) in comparison with the dry day (dashed lines) for Uref = 5 m/s.
Fig. 14. Difference of surface-averaged temperature, T, and moisture content, w, between rainy and dry cases on a) the street surface and
b) the windward facade for Uref = 2 m/s (solid lines) and Uref = 5 m/s (dashed lines).
2 m/s and 17.2 °C for Uref = 5 m/s. Note that the higher wind speed leads to more rain deposition on the wind-
ward facade and less on the street compared to lower wind speed. The higher cooling effect for Uref = 2 m/s is
caused by the higher wetting amount on the street surface, which leads to about 3 °C lower surface temper-
ature than for Uref = 5 m/s. The cooling effect decreases towards the end of the 48 h period as the surface
moisture content and, as a consequence, the latent heat flux decrease. At the end of the 48 h period, the re-
maining moisture in the street surface is about 3 times higher for Uref = 2 m/s than it is for Uref = 5 m/s.
For the windward facade, the temperature decrease due to rain is similar for both wind speeds, although
the moisture content is higher for Uref = 5 m/s. This may be due to several factors. First, the total wetting of
the windward facade is clearly lower than the street surface, which limits the cooling due to the latent heat
flux. Second, most of the wetting on the windward facade occurs at the top and side edges of the facade.
These are also the regions with higher convective transfer coefficient. Therefore, the drying also occurs faster.
The decrease in surface temperature accompanies the decrease in air temperature. The decrease in the air
temperature due to rain is compared in Fig. 15 for the two reference wind speeds. The air temperature is av-
eraged within the street canyon at the pedestrian height (1.75 m). For the dry case, the air temperature shows
less variation for Uref = 2 m/s with lower maximum and higher minimum values. The maximum air temper-
ature difference with the ambient is 1.9–2.0 °C for dry case while it is 1.4–1.5 °C after the rain event. Overall,
the decrease in air temperature due to the 10 h-long rain event is observed for the whole day long after the
rain stops for both wind speeds. In fact, the decrease in air temperature due to rain is larger after the rain
event compared to during the rain event. On the other hand, the influence of the wind speed on the decrease
in air temperature due to rain is found to be limited. Both curves shift downward by a similar amount due the
rain event.
5. Summary and conclusions
In this study, a fully-integrated 3D numerical urban microclimate model is developed and presented. For
this, wind flow in the air is solved together with heat and moisture transport both in the air and in the porous
building materials, in addition to the solar and thermal radiation with exchanges. Steady RANS simulations for
air flow are coupled with unsteady heat and moisture transfer in building materials for each exchange
timestep. Spatial distribution of surface wetting due to WDR is calculated using an Eulerian multiphase
model. The influence of environmental loading (wind, rain, solar radiation) on the surface and air tempera-
tures in an isolated street canyon is investigated in detail for two different reference wind speeds in order
to demonstrate the model. The impact of a single rain event on microclimate is presented by comparing the
temporal and spatial variation of thermal conditions within the street canyon with a day without rain. The
model gives detailed information on the thermal storage within urban materials and allows to compare
heat-removal mechanisms at different times of a day. The results provided by the model are valuable when
comparing different strategies that aim to improve urban microclimate. This modeling approach proposes a
Fig. 15. Difference of average street-canyon air temperature at pedestrian height with the ambient temperature during and after the rain
event (solid lines) and dry (dashed lines) cases.
frontward avenue for a better understanding of evaporative cooling in urban context due to its physics-based
coupling of the fate of WDR-moisture with wind flow and buoyancy.
In the case study, high spatial gradients in surface temperatures are observed during the day time for the
dry case as a result of solar radiation and shadowing. For both night and day times, the highest surface tem-
peratures tend to be located at the center of the street-canyon surfaces. Heat is stored in the street canyon be-
tween sunrise and afternoon, after which the stored heat is released in the environment. At lower wind speed,
lower convective heat flux, higher thermal radiation and larger heat storage are observed.
Also in the case study, surface temperatures are found to markedly decrease when a rain event occurs, by
up to 25 °C, depending on the WDR deposition patterns which depend on wind velocity. The decrease in sur-
face temperatures following a short and low-intensity rain event with a total rainfall amount of 10 mm is
found to be the highest on the street surface and for the lower wind speed. Furthermore, a significant influ-
ence of neighboring surfaces on the surface temperature within the street canyon is observed due to thermal
radiation exchange. The decrease in surface temperatures due to evaporative cooling results in lower air tem-
perature in the street canyon and the effect of evaporative cooling is visible long after the rain event stops. Dif-
ferent rates of evaporation and different durations of evaporative cooling are found by varying wind speed.
The primary aim of the case study was to demonstrate the extent of the capabilities of the coupled model.
The present study neglects daily variations in wind speed and wind direction, while considering single type of
building material. This study also does not take into account the droplet physics after impact such as spreading
or splashing, nor the film forming or runoff that may occur when the surface of the porous materials is capil-
lary saturated. The present case study focuses on a situation where all the rain water impinging on a surface is
absorbed. On-going work is considering to take into account the different aspects of environmental loading,
e.g. spreading and splashing at droplet impact to better determine the surface rain water locally available, sur-
face film occurrences such as rain puddles, which would further delay the drying process, and film runoff on
building facade, which in effect redistributes the water over the surface. Furthermore, influence of vegetation,
e.g. trees, green roofs, is not included in the present study. Further studies will implement more accurate
wind-flow conditions based on meteorological data, take into account different orientations with respect to
the sun and rain, couple with comfort models and study the influences of vegetation. Also, the effects of dif-
ferent building materials and different surface albedos on the thermal comfort throughout the day need to be
investigated. One aspect where the model is expected to bring significant insights is in the understanding of
the separate, but often counteracting, influences of materials, geometry, vegetation, etc. on the urban micro-
climate and thermal comfort.
References
Allegrini, J., Dorer, V., Carmeliet, J., 2012a. Analysis of convective heat transfer at building façades in street canyons and its influence on
the predictions of space cooling demand in buildings. J. Wind Eng. Ind. Aerodyn. 104–106, 464–473.
Allegrini, J., Dorer, V., Defraeye, T., Carmeliet, J., 2012b. An adaptive temperature wall function for mixed convective flows at exterior sur-
faces of buildings in street canyons. Build. Environ. 49, 55–66.
Allegrini, J., Dorer, V., Carmeliet, J., 2013. Wind tunnel measurements of buoyant flows in street canyons. Build. Environ. 59, 315–326.
Allegrini, J., Dorer, V., Carmeliet, J., 2015. Coupled CFD, radiation and building energy model for studying heat fluxes in an urban environ-
ment with generic building configurations. Sustain. Cities Soc. 19, 385–394.
Arnfield, A.J., 2003. Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat
island. Int. J. Climatol. 23 (1), 1–26.
ASHRAE, 1985. ASHRAE Handbook—Fundamentals: American Society of Heating. Refrigerating and Air-Conditioning Engineers Inc.
Best, A.C., 1950. The size distribution of raindrops. Q. J. R. Meteorol. Soc. 76 (327), 16–36.
Blocken, B., 2014. 50 years of computational wind engineering: past, present and future. J. Wind Eng. Ind. Aerodyn. 129, 69–102.
Blocken, B., Carmeliet, J., 2002. Spatial and temporal distribution of driving rain on a low-rise building. Wind Struct. 5 (5), 441–462.
Blocken, B., Carmeliet, J., 2004. A review of wind-driven rain research in building science. J. Wind Eng. Ind. Aerodyn. 92 (13), 1079–1130.
Blocken, B., Carmeliet, J., 2007. Validation of CFD simulations of'wind-driven rain on a low-rise building facade. Build. Environ. 42 (7),
2530–2548.
Blocken, B., Stathopoulos, T., Carmeliet, J., 2007. CFD simulation of the atmospheric boundary layer: wall function problems. Atmos. En-
viron. 41 (2), 238–252.
Bouyer, J., Inard, C., Musy, M., 2011. Microclimatic coupling as a solution to improve building energy simulation in an urban context.
Energ. Buildings 43 (7), 1549–1559.
Carmeliet, J., Hens, H., Roels, S., Adan, O., Brocken, H., Cerny, R., Pavlik, Z., Hall, C., Kumaran, K., Pel, L., 2004. Determination of the liquid
water diffusivity from transient moisture transfer experiments. J. Therm. Envel. Build. Sci. 27 (4), 277–305.
Cebeci, T., Bradshaw, P., 1977. Momentum Transfer in Boundary Layers. Hemisphere Publishing Corporation, New York.
Cole, R.J., 1976. The longwave radiative environment around buildings. Build. Environ. 11 (1), 3–13.
Defraeye, T.W.J., 2011. Convective heat and mass transfer at exterior building surfaces. Ph.D. Thesis. Katholieke Universiteit Leuven, Leu-
ven, Belgium.
Defraeye, T., Blocken, B., Carmeliet, J., 2011. An adjusted temperature wall function for turbulent forced convective heat transfer for bluff
bodies in the atmospheric boundary layer. Build. Environ. 46 (11), 2130–2141.
Derluyn, H., Griffa, M., Mannes, D., Jerjen, I., Dewanckele, J., Vontobel, P., Sheppard, A., et al., 2013. Characterizing saline uptake and salt
distributions in porous limestone with neutron radiography and X-ray micro-tomography. J. Build. Phys. 36 (4), 353–374.
Franke, J., Hellsten, A., Schlunzen, K.H., Carissimo, B., 2011. The COST 732 best practice guideline for CFD simulation of flows in the urban
environment: a summary. Int. J. Environ. Pollut. 44 (1–4), 419–427.
Gunn, R., Kinzer, G.D., 1949. The terminal velocity of fall for water droplets in stagnant air. J. Meteorol. 6 (4), 243–248.
Hagentoft, C.-E., Kalagasidis, A.S., Adl-Zarrabi, B., Roels, S., Carmeliet, J., Hens, H., Grunewald, J., et al., 2004. Assessment method of numer-
ical prediction models for combined heat, air and moisture transfer in building components: benchmarks for one-dimensional cases.
J. Therm. Envel. Build. Sci. 27 (4), 327–352.
Huang, S.H., Li, Q.S., 2010. Numerical simulations of wind-driven rain on building envelopes based on Eulerian multiphase model. J. Wind
Eng. Ind. Aerodyn. 98 (12), 843–857.
Janssen, H., Blocken, B., Carmeliet, J., 2007. Conservative modelling of the moisture and heat transfer in building components under at-
mospheric excitation. Int. J. Heat Mass Transf. 50 (5–6), 1128–1140.
Kaoru, I., Akira, K., Akikazu, K., 2011. The 24-h unsteady analysis of air flow and temperature in a real city by high-speed radiation cal-
culation method. Build. Environ. 46 (8), 1632–1638.
Kim, J.-J., Baik, J.-J., 2001. Urban street-canyon flows with bottom heating. Atmos. Environ. 35 (20), 3395–3404.
Kubilay, A., Derome, D., Blocken, B., Carmeliet, J., 2013. CFD simulation and validation of wind-driven rain on a building facade with an
Eulerian multiphase model. Build. Environ. 61, 69–81.
Kubilay, A., Derome, D., Blocken, B., Carmeliet, J., 2014. Numerical simulations of wind-driven rain on an array of low-rise cubic buildings
and validation by field measurements. Build. Environ. 81 (0), 283–295.
Kubilay, A., Derome, D., Blocken, B., Carmeliet, J., 2015a. Numerical modeling of turbulent dispersion for wind-driven rain on building
facades. Environ. Fluid Mech. 15 (1), 109–133.
Kubilay, A., Derome, D., Blocken, B., Carmeliet, J., 2015b. Wind-driven rain on two parallel wide buildings: field measurements and CFD
simulations. J. Wind Eng. Ind. Aerodyn. 146, 11–28.
Kubilay, A., Carmeliet, J., Derome, D., 2017. Computational fluid dynamics simulations of wind-driven rain on a mid-rise residential build-
ing with various types of facade details. J. Build. Perform. Simul. 10 (2), 125–143.
Kusaka, H., Kondo, H., Kikegawa, Y., Kimura, F., 2001. A simple single-layer urban canopy model for atmospheric models: comparison
with multi-layer and slab models. Bound.-Layer Meteorol. 101, 329–358.
Launder, B.E., Sharma, B.I., 1974. Application of the energy-dissipation model of turbulence to the calculation of flow near a spinning disc.
Lett. Heat Mass Transf. 1 (2), 131–138.
Lemonsu, A., Masson, V., Shashua-Bar, L., Erell, E., Pearlmutter, D., 2012. Inclusion of vegetation in the Town Energy Balance model for
modelling urban green areas. Geosci. Model Dev. 5 (6), 1377–1393.
Li, X., Yu, Z., Zhao, B., Li, Y., 2005. Numerical analysis of outdoor thermal environment around buildings. Build. Environ. 40 (6), 853–866.
Ma, J., Li, X., Zhu, Y., 2012. A simplified method to predict the outdoor thermal environment in residential district. Build. Simul. 5 (2),
157–167.
Martilli, A., Clappier, A., Rotach, M.W., 2002. An urban surface exchange parameterisation for mesoscale models. Bound.-Layer Meteorol.
104 (2), 261–304.
Masson, V., 2000. A physically-based scheme for the urban energy budget in atmospheric models. Bound.-Layer Meteorol. 94 (3),
357–397.
Meteonorm, 2000. Global Meteorological Database for Solar Energy and Applied Climatology. Meteotest, Bern, Switzerland.
Mirzaei, P.A., 2015. Recent challenges in modeling of urban heat island. Sustain. Cities Soc. 19, 200–206.
Mirzaei, P.A., Haghighat, F., 2010. Approaches to study Urban Heat Island—abilities and limitations. Build. Environ. 45 (10), 2192–2201.
Moonen, P., Defraeye, T.W.J., Dorer, V., Blocken, B., Carmeliet, J., 2012. Urban physics: effect of the microclimate on comfort, health and
energy demand. Front. Archit. Res. 1 (3), 197–228.
de Morais, M.V.B., de Freitas, E.D., Guerrero, V.V.U., Martins, L.D., 2016. A modeling analysis of urban canopy parameterization
representing the vegetation effects in the megacity of Sao Paulo. Urban Clim. 17, 102–115.
Murakami, S., 1993. Comparison of various turbulence models applied to a bluff-body. J. Wind Eng. Ind. Aerodyn. 46-7, 21–36.
Oke, T.R., 1982. The energetic basis of the urban heat island. Q. J. R. Meteorol. Soc. 108, 1–24.
OpenFOAM, 2015. OpenCFD Ltd. OpenFOAM v2.4.0 User Guide. Retrieved from http://www.openfoam.com.
Qu, Y., Milliez, M., Musson-Genon, L., Carissimo, B., 2012. Numerical study of the thermal effects of buildings on low-speed airflow taking
into account 3D atmospheric radiation in urban canopy. J. Wind Eng. Ind. Aerodyn. 104–106, 474–483.
Richards, P.J., Hoxey, R.P., 1993. Appropriate boundary conditions for computational wind engineering models using the k-ϵ turbulence
model. J. Wind Eng. Ind. Aerodyn. 46–47 (0), 145–153.
Saneinejad, S., Moonen, P., Carmeliet, J., 2014. Coupled CFD, radiation and porous media model for evaluating the micro-climate in an
urban environment. J. Wind Eng. Ind. Aerodyn. 128, 1–11.
Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M., Shirasawa, T., 2008. AIJ guidelines for practical applications of
CFD to pedestrian wind environment around buildings. J. Wind Eng. Ind. Aerodyn. 96 (10−11), 1749–1761.
Tominaga, Y., Sato, Y., Sadohara, S., 2015. CFD simulations of the effect of evaporative cooling from water bodies in a micro-scale urban
environment: validation and application studies. Sustain. Cities Soc. 19, 259–270.
Toparlar, Y., Blocken, B., Vos, P., van Heijst, G.J.F., Janssen, W.D., van Hooff, T., Montazeri, H., Timmermans, H.J.P., 2015. CFD simulation and
validation of urban microclimate: a case study for Bergpolder Zuid, Rotterdam. Build. Environ. 83, 79–90.
Whitaker, S., 1977. Simultaneous heat, mass, and momentum transfer in porous media: a theory of drying. In: Hartnett James, P., Irvine
Thomas, F. (Eds.), Advances in Heat Transfer. Elsevier, pp. 119–203.
Wieringa, J., 1992. Updating the Davenport roughness classification. J. Wind Eng. Ind. Aerodyn. 41 (1–3), 357–368.
de Wolf, D.A., 2001. On the Laws-Parsons distribution of raindrop sizes. Radio Sci. 36 (4), 639–642.
Xie, X., Liu, C.-H., Leung, D.Y.C., Leung, M.K.H., 2006. Characteristics of air exchange in a street canyon with ground heating. Atmos. En-
viron. 40 (33), 6396–6409.

Kubilay 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kubilay 2017

Uploaded by

Copyright:

Available Formats

UCLIM-00299; No of Pages 21

Urban Climate xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

journal homepage: http://www.elsevier.com/locate/uclim

Coupling of physical phenomena in urban microclimate: A

2. Description of the numerical model and the coupling strategy

2.1. Numerical model for wind ﬂow

2.2. Numerical model for wind-driven rain (WDR)

2.2.1. Governing equations for WDR

can be related to the velocity ﬂuctuations in the wind:

2.2.2. Numerical settings for WDR

qa ¼ ðcl T Þg l þ ðcv T þ Lv Þg v ð6cÞ

g ext ¼ g conv þ Rwdr ð7Þ

qext ¼ qconv þ S þ Rs þ qls ; Rs ¼ cl T wb Rwdr ; qls ¼ ðcv T s þ Lv Þg conv ð8Þ

2.4. Numerical model for radiative heat transfer

εc ¼ ð1−0:84cÞð0:527 þ 0:161 exp½8:45ð1−273=T a ÞÞ þ 0:84c ð12aÞ

2.5. Solution algorithm

3. Description of the case study

3.1. Computational domain and grid for air

3.2. Computational domain and grid for building materials

Brick (facades) 1600 1000 0.682 0.9 0.4

3.3. Atmospheric conditions

4.1. Analysis of climatic loading

Wind speed Rain event Rain duration Simulation duration

4.2. Analysis of heat ﬂuxes

4.3. Analysis of temperature and moisture content

4.4. Impact of wind speed on evaporative cooling by rain

5. Summary and conclusions

You might also like