Conference PaperPDF Available

Predicting space usage by multi-objective assessment of outdoor thermal comfort around a university campus

Authors:

Abstract and Figures

With the impending issues regarding global warming, urban design is considered a key driver to improve the microclimate in cities. For public spaces, studies suggest that outdoor thermal comfort may be seen as a proxy for space usage, and in turn, its attractiveness to people. Although the topic has gained interest in recent years, the discussion so far has focused on computing the metrics rather than deriving interventions from them. Here, we use the tool Eddy3D to model and analyze the outdoor thermal comfort of a designated area around a university campus. Further, we demonstrate how to estimate space usage from those results. Finally, we conduct a spatial sensitivity analysis of the underlying results as a step towards decision aiding. Our work demonstrates how decision-makers may derive areas where interventions will likely have the largest impact on outdoor thermal comfort performance.
Content may be subject to copyright.
Predicting space usage by multi-objective assessment of
outdoor thermal comfort around a university campus
Patrick Kastner1, Timur Dogan1
1Environmental Systems Lab, Cornell University, Ithaca, NY, USA, pk373@cornell.edu
ABSTRACT
With the impending issues regarding global warming, urban
design is considered a key driver to improve the microclimate in
cities. For public spaces, studies suggest that outdoor thermal
comfort may be seen as a proxy for space usage, and in turn,
its attractiveness to people. Although the topic has gained
interest in recent years, the discussion so far has focused on
computing the metrics rather than deriving interventions from
them. Here, we use the tool Eddy3D to model and analyze
the outdoor thermal comfort of a designated area around a
university campus. Further, we demonstrate how to estimate
space usage from those results. Finally, we conduct a spatial
sensitivity analysis of the underlying results as a step towards
decision aiding. Our work demonstrates how decision-makers
may derive areas where interventions will likely have the largest
impact on outdoor thermal comfort performance.
Author Keywords
Outdoor Thermal Comfort, Sensitivity Analysis, CFD, MRT,
Space Usage
1 INTRODUCTION
Climate change scenarios suggest disconcerting outcomes for
the future of cities, especially in urbanized and dense locations.
As a result of changing urban microclimates, there is not only
increased competition over the wind, sun, and daylight [6], but
also a growing risk for detrimental impacts on aspects of the
city life; public health, commercial activities, active mobility,
and space usage among them [17]. However, the fact that
urban microclimates are shaped by the way cities are built also
opens new opportunities. For these reasons, the prediction of
outdoor thermal comfort has become an increasingly relevant
area of research.
In the early 1980s, Whyte [19] devoted a significant amount
of time to studying how people use public spaces in US
cities, and why some plazas show different usage patterns than
others. From video recordings, they counted the number of
dwellers which revealed that access to sunlight can be a major
factor for plaza use. For NYC, they found that there is a strong
correlation between space usage and exposure to sunlight while
in other months the microclimate was less relevant. Direct
exposure to sunlight is preferred through May during which
SimAUD 2020 May 25-27 Vienna, Austria
©2020 Society for Modeling & Simulation International (SCS)
time the ambient temperature has yet reached comfortable
levels. By contrast, in months from June through August,
people show little preference between sun and shade and other
social parameters and amenities were determining factors [18].
This study frames simulated-based microclimate assessment
as a multiple-criteria evaluation problem. It aims to assess the
microclimate conditions on a university campus from which
outdoor space use, a proxy for the attractiveness of outdoor
spaces, is derived. This is motivated by the fact that the
majority of the student population is on campus during spring
and fall when outdoor thermal comfort is highly dependent
on sun and wind exposure. The framework of this analysis is
justified with a two-at-a-time and a global sensitivity analysis.
The spatial sensitivity adds a layer to the results that allow
decision-makers to deduce where geometric interventions will
likely have the largest impact on the performance.
2 METHODOLOGY
This study utilizes a simulation tool called Eddy3D to simu-
late the annual outdoor thermal comfort around an existing
building arrangement on the Cornell University campus. The
weather data for this study were retrieved from the EnergyPlus
repository by the Department of Energy
1
and classifies as
warm-summer humid continental climate (Dfb) according to
the Köppen classification.
2.1 Case study: Engineering Quad at Cornell University
The CAD model used is comprised of Cornell’s Engineering
Quad in Ithaca, upstate NY, see Figure 1. As the elevation
differences between the buildings are insignificant, they have
been neglected in the CAD model.
2.2 Universal Thermal Climate Index (UTCI)
Like other outdoor comfort metrics, the UTCI was developed
as an equivalent temperature measure. It is based on a multi-
node thermo-physical model that was coupled with an adaptive
clothing model [8]. Thus, for any combination of air temper-
ature, wind, radiation, and humidity, the UTCI is defined as
the air temperature of a particular reference condition which
would cause the same thermal sensation as predicted by the
model [5]. We chose the UTCI as a metric, as several studies
have shown the UTCI to be a suitable metric that is sensitive
and accurate for cold temperatures while also achieving good
1Syracuse-Hancock.Intl.AP.725190_TMY3
Figure 1. CAD model of the Cornell University engineering quad.
agreement with human responses in tropical climates [14].
Mathematically, the UTCI is computed with a polynomial
approximation of four input variables:
𝑈𝑇𝐶𝐼 =𝑇𝑎𝑚𝑏 +𝑓(𝑇𝑎𝑚𝑏, 𝑇𝑀 𝑅𝑇 , 𝑈𝑊 𝑖𝑛 𝑑 , 𝑝𝑣𝑎 𝑝 𝑜𝑢𝑟 )(1)
where
𝑇𝑎𝑚𝑏
is the ambient temperature,
𝑇𝑀 𝑅𝑇
is the mean
radiant temperature,
𝑈𝑊 𝑖𝑛𝑑
is the wind velocity, and
𝑝𝑣 𝑎 𝑝𝑜𝑢 𝑟
refers to the vapor pressure, which is a function of the dry-
bulb temperature. The resulting UTCI temperatures may be
classified and reported as thermal stress categories ranging
from “extreme cold stress” to “extreme heat stress”. Ideally,
one would strive for maximizing the annual urban climate for
the “No thermal stress” condition, which falls in the range of
9C and 26C.
Empirical data that has been collected by Reinhart et al. [15],
confirmed studies undertaken by White. Over one year, the
number of people with WiFi devices in a public courtyard on
a university campus has been recorded. It has been shown
that people take longer lunch breaks when in “no thermal
stress” conditions. This suggests that people prefer being in
outdoor spaces if the microclimatic conditions are within the
“no thermal stress” category. Their findings are summarized in
Figure 2.
Figure 2
. Collocation of empirical outdoor thermal comfort and space usage
correlations collected by Reinhart et al.
2.3 Sensitivity analysis
To identify the dominant input parameters of the UTCI metric
in this particular temperate climate, and to deduce actionable
suggestions on how the microclimate could be improved,
sensitivity analyses were carried out in three stages. First, a
two-at-a-time sensitivity analysis was carried out by varying
the ambient temperature with the remaining three parameters,
see Figure 3. For this, the hidden variables were fixed to 20
C
for the dry-bulb temperature, 1
𝑚/𝑠
for the wind velocity and
50 % for the relative humidity. The code was adapted from
[13]. Further, we ran a global, variance-based Sobol sensitivity
analysis. It is a method that decomposes the variance in
the UTCI into fractions which can be attributed to the four
input variables. Generally, it is used to derive the rank-order
importance of input variables in multivariate analyses. Beyond
that, it is also helpful in not only gaining an understanding of
how the variables affect the solution in sets of two but allow
us to quantify the total effect when varying all parameters
simultaneously (which cannot be plotted in two dimensions).
The Saltelli implementation in SALib [9] was used to evaluate
100’000 samples of the UTCI problem space with the upper
and lower bounds from Table 1. The upper and lower bounds
were derived from the Syracuse weather data, except for the
MRT, which has been assumed to be within a
±
5
𝐶
offset
from the dry-bulb temperature.
Table 1
. Variables and bounds for the global sensitivity analysis from the
Syracuse weather data.
Variable Unit Min Max
AmbTemp [𝐶]-19.4 37.8
MRT [𝐶]-24.4 42.8
Wind [𝑚/𝑠]0 17
RelHum [%]0 100
The aforementioned methods provide a good starting point to
understand the dominance of the input variables in a theoretical
framework. However, as cities shape their microclimates, they
also shape the upper and lower bounds of the UTCI input
variables in time and space. In this regard, the wind velocity
and the MRT can be especially spatially-inhomogeneous due
to wind sheltering effects or extreme surface temperatures.
Therefore, we conducted a spatially-resolved global sensitivity
analysis, for which the annual upper and lower bounds for
each probing point serve as an input. For this, we probed the
wind velocity and the MRT from the outdoor thermal comfort
simulation results and made use of the weather data for ambient
temperature and relative humidity.
2.4 Annual outdoor thermal comfort simulation frame-
work
The simulation framework consists of two simulation en-
gines, namely: OpenFOAM and Radiance. Both engines
are centered around a toolkit called Eddy3D
2
that is imple-
mented in Rhinoceros and Grasshopper which handles pre-,
post-processing and the data handling between the simulation
engines. Specifically, Eddy3D creates the simulation domain,
specifies the boundary conditions, and takes care of processing
the weather data based on the building geometry in Rhinoceros.
Similarly, it uses Rhinoceros’ meshing capabilities to export
2www.eddy3d.com
building meshes for both OpenFOAM and Radiance. Further,
it uses Radiance to calculate irradiation and view factors for
each sensor point.
Wind velocity
Due to the computationally-expensive nature of Computa-
tional Fluid Dynamics (CFD) simulations, it is infeasible to
run a single analysis for an entire year. Considering this,
we made use of the wind reduction factors method which
has been implemented in Eddy3D. The tool utilizes Open-
FOAM’s
𝑏𝑙 𝑜𝑐 𝑘 𝑀𝑒 𝑠ℎ
utility for the background mesh and
𝑠𝑛𝑎 𝑝 𝑝 𝑦 𝐻𝑒 𝑥𝑀 𝑒 𝑠ℎ
to subsequently snap the background mesh
to the building geometry. For the background mesh, we used a
cylindrical simulation domain approach which allows reusing
the same computational mesh for every wind direction, thus
reducing the computation time and storage space [11]. Within
the cylindrical mesh, we further refined the mesh within a
refinement box that surrounds the buildings of interest. The
simulation domain was set up according to best practices
while taking into account all relevant surrounding buildings
which resulted in 6
.
6
·
10
6
cells for the global mesh. This
methodology makes use of a set of CFD simulations from
several wind directions. For this study, we used 8 RANS
simulations (0
°
, 45
°
, 90
°
, 135
°
, 180
°
, 225
°
, 270
°
, 315
°
) in a
45
°
interval. Depending on the direction, we mapped the inlets
on the one-half circle of the simulation domain and the outlet
on the opposite side. We used an incompressible, isothermal,
steady-state solver from OpenFOAM in combination with a
𝑘𝜔𝑆𝑆𝑇
turbulence model. The half-circular domain inlet
was set to an atmospheric boundary layer (ABL) profile for
U, k, and 𝜔, and a roughness length z0=1that corresponds
to a suburban environment. At the outlet of the computa-
tional domain, constant pressure is assumed, while the other
variables are imposed to be zero-gradient. The ground and
the building geometry used the same boundary conditions, a
no-slip condition for velocity, a zero-gradient condition for the
pressure and wall functions for
𝑈
,
𝑘
, and
𝜔
. Going forward,
the 8 RANS simulations served as a nearest neighbor lookup
table of wind velocities in concert with the annual weather
data. For each probing point, we probed the simulated velocity
from the 8 CFD simulations. This multidimensional array is
used to calculate the dimensionless wind velocity for every
probing point by dividing the simulated velocity magnitude by
the scaled-down inlet velocity with the logarithmic wind power
profile. This yields a spatial wind reduction matrix with infor-
mation for every probing point for each of the 8 wind directions.
From here, we converted the spatial matrix into a temporal
matrix. For every hour of the year and its corresponding wind
direction, we looked up the nearest neighbor CFD simulation
and multiplied the velocities from the spatial velocity matrix
with the wind velocity from the weather data that has been
scaled down to the probing height. This operation yields a
temporal velocity matrix with wind reduction data from which
the wind velocities for the UTCI calculation are retrieved [10].
For cases where the wind velocity was outside the bounds of
the UTCI calculation
(
0
.
5
𝑚/𝑠<
applicable range
<
17
𝑚/𝑠)
,
we replaced the values with lower and upper bounds and lifted
the velocity at a height of 10 𝑚as advised in [4].
Mean radiant temperature
The mean radiant temperature (MRT) calculated in this study
consists of three components: the sky temperature
𝑇𝑠𝑘 𝑦
, the
solar gain (
Δ𝑀 𝑅𝑇𝑑𝑠
) of being exposed to direct sun, and the
ground and building surface temperature 𝑇𝑔𝑏.
To estimate the ratio between
𝑇𝑠𝑘 𝑦
and
𝑇𝑔𝑏
, we used Radiance
to run a view factor analysis for every probing point. We
calculated
𝑇𝑠𝑘 𝑦
through the sky emissivity and the horizontal
infrared radiation intensity [7]. We approximated
𝑇𝑔𝑏
with
the ambient temperature as it has been shown that typical
differences are less than
±
5
𝐾
[12]. The solar gain to the
human body is calculated using the Effective Radiant Field
(ERF) [1] which we adapted for an outdoor setting and from
which we derived (
Δ𝑀 𝑅𝑇𝑑𝑠
). For the irradiance that is used as
an input for the ERF, we implemented a Radiance-based Two-
Phase (DDS) method. The DDS approach was chosen because
it provides a better spatial resolution of the direct component.
The Two-Phase DDS method is a daylight-coefficient-based
simulation with an all-weather dynamic sky model (Perez Sky
model). Instead of approximating the position and shape of
the sun with few sky patches, we used 577 sun patches for
the direct and diffuse simulation and 2305 direct sun patches.
The illuminance is a linear combination of: (1) an annual
daylight coefficient simulation, (2) annual direct-only daylight
coefficients, and (3) an annual sun-coefficients simulation [3]:
𝐸=𝐶𝑑𝑐 ·𝑆𝐶𝑑 𝑐𝑑 ·𝑆𝑑+𝐶𝑠𝑢 𝑛 ·𝑆𝑠𝑢 𝑛 (2)
where
𝐶𝑑𝑐
and
𝑆
denote the daylight coefficient matrix and the
sky vector,
𝐶𝑑𝑐 𝑑
and
𝑆𝑑
denote the direct-sky coefficient and
the direct sky matrix, and
𝐶𝑠𝑢𝑛
and
𝑆𝑠𝑢𝑛
denote the direct-sun
coefficient and the sun matrix, respectively. [16]. Finally,
we computed the MRT for every hour
and every probe
𝑖
according to Equation (3).
𝑀 𝑅𝑇ℎ,𝑖 =𝑇𝑠𝑘 𝑦 , ℎ ·𝑓𝑠𝑘 𝑦 ,𝑖 +𝑇𝑔𝑏 ,ℎ ·𝑓𝑠 𝑘 𝑦 ,𝑖 +Δ𝑀 𝑅𝑇𝑑𝑠 ,ℎ (3)
2.5 Space usage
We used the data collected by Reinhart et al. [15] to estimate
space usage around the Cornell University Engineering Quad
to provide a meaningful metric for campus decision-makers.
For this, we calculated the number of people for each hour and
probing point as a function of the UTCI. The space usage is
calculated for the three seasons spring, summer, and fall for
occupancy periods between 10 am - 7 pm. All values were
then normalized between 0 - 1 as the empirical data cannot
be directly applied to a different university campus without
further investigation.
3 RESULTS
Both non-spatial sensitivity analyses justify the assumptions
that have been made when setting up the simulation framework.
Two-at-a-time sensitivity analysis
In Figure 3, each input variable to the UTCI metric shows
a non-linear behavior when varying it with the ambient tem-
perature, however, their amplitude and parameter range differ
(a) (b) (c)
Figure 3. Two-at-a-time sensitivity analysis of the UTCI input variables.
significantly. When varying the MRT with the ambient tem-
perature in Figure 3 (a), high and low MRTs create significant
discomfort through temperature asymmetry. For the wind
velocity in Figure 3 (b), we see highly non-linear behavior for
both low ambient temperatures and high wind velocities. This
effect is commonly known as the wind chill due to increased
forced convection. When varying the ambient temperature
with relative humidity, non-linear behavior is evident only for
high levels of both relative humidity and ambient temperature.
Global sensitivity analysis
Figure 4 quantifies the first-, second-, and total-order sensitivity
indices illustrated as black dots, gray lines, and black circles,
respectively. The confidence intervals for each index and
variable were found to be below 1
·
10
2
. We confirm that the
dominant variables contributing to the UTCI in a climate in
upstate NY were ambient temperature, MRT, and wind velocity.
Those three variables show total-order sensitivity indices of
83 %, 5 %, 16 %, respectively. The least dominant factor
in this climate is the relative humidity which only accounts
for
1 %
of the variance. This suggests that the humidity may
be neglected for the Dfb climate zone. Note that the total
sensitivity indices sum to a value greater than one, indicating
that the input variables are to some extent correlated. The
second-order sensitivity indices range from
1 - 5 % with
wind and ambient temperature showing the largest interaction
effects. This confirms the findings from the two-at-a-time
sensitivity analysis, see Figure 3 (b).
Annual outdoor thermal comfort
All results in this section have been computed and probed at a
height of 2 m above ground. Figure 5 (a) shows the averaged
annual wind velocity magnitudes. We see that points that
are sheltered from the wind experience a lower pedestrian
comfort rating and vice versa. It is also evident that venturi
effects exist in two areas, channeling the wind from north to
south. Figure 5 (b) shows the averaged annual mean radiant
temperature. The average MRT ranges from
4.5-6.5C
where
higher temperatures are seen closed to building geometries and
vice versa. Figure 5 (c) shows the averaged annual UTCI. While
points close to building geometries generally see favorable
conditions in terms of wind velocities and MRTs, it is evident
that the venturi effect sports are rendered less comfortable.
Overall, the seating arrangements on the Engineering Quad
are within areas that show a rather high annual averaged UTCI.
l
< 1 % 1 %
78 % 83 %< 1 %5 %
Figure 4
. Global sensitivity indices where S1 is the first-order, S2 is the
second-order, and ST is the total-order effect for the UTCI estimated with the
Sobol method.
Space usage
In Figure 5 (d) we show the predicted space usage around
the campus on June 30, 16:00 h, a hot and calm day from
the weather file for which the average UTCI was estimated to
33
C. We focus on hourly data in this part of the results as
an annual average would not allow to highlight the temporal
variance and its effect on space usage. Areas with exposure
to direct sun receive a penalty in terms of space usage due to
wind not being present. By contrast, shaded areas experience
increased numbers of space usage as those are the only ones
with a comfortable UTCI. Here, the seating arrangements on
the Engineering Quad suffer from direct exposure to sunlight.
Given the already uncomfortable conditions (UTCI = 33
C),
the seating arrangements and other areas exposed to direct
sunlight suffer a penalty.
Spatially-resolved global sensitivity analysis
Figure 6 (a) and Figure 6 (b) show the spatially-resolved total
sensitivity indices for the wind velocity and the MRT. The
upper bound for ST for the wind velocity is close to 40 %,
(a) (b)
(c) (d)
Figure 5
. (a) Annual wind velocity magnitudes; (b) average annual mean radiant temperature; (c) average annual UTCI; (d) estimated space usage on June 30,
16:00 h, a hot and calm day from the weather file. The dashed markers indicate actual outdoor seating arrangements on campus.
(a) (b)
Figure 6. Spatially-resolved total sensitivity index for the (a) the wind velocity and (b) the mean radiant temperature.
whereas the upper bound for the MRT is around 10 %. For
example, the wind velocity around two north-south corridors
in Figure 6 (a), explains a large part (
40 %) of the variance
in the UTCI output considering an entire year.
4 DISCUSSION
We used weather data from Syracuse, NY as boundary con-
ditions for this study as there is limited data available for the
site in Ithaca, NY. While ambient temperature and irradiation
of both sites are comparable, it is conceivable that Ithaca, NY
experiences deviating wind velocity distributions due to its
overall differences in elevation and its location at the edge of a
valley. Further, using empirical data from another university
campus in Boston to estimate space usage has limitations.
Despite the overall climate in Boston being similar, it lies in
a different sub-climate zone with hotter summers than those
in upstate NY. Although Reinhardt et al. [15] suggest their re-
sults hold for climate comparable to Boston’s, the relationship
between UTCI and space usage in upstate NY be likely differ-
ent. Nonetheless, in this study, we argue for the importance
of decision aiding based on outdoor thermal comfort results
rather than their validation.
We have shown that local, global, and spatially-global sensitiv-
ity analyses can help understand the dominant input variables
of a complex metric such as the UTCI. The results of the global
sensitivity analysis in Figure 4 reveal the strong second-order
interaction between the ambient temperature and the wind
velocity. That is to say that, without plotting the relation-
ship between both variables, we know that a large part of the
variance in the UTCI can be attributed to varying both the
ambient temperature and the wind velocity simultaneously. It
is worth noting that the global analysis itself does not reveal the
inverse nature of this relationship (an increase in wind velocity
and a decrease in dry-bulb temperature yields a non-linear
decrease in UTCI). As sensitivity analyses help assess arbitrary
functions (such as the UTCI), generalizing this insight might
be helpful for practitioners in the design process. In that
regard, we suggest to use a global sensitivity analysis as a first
step when analyzing a (potentially unknown) metric, and to
derive a factor prioritization from it. Identifying second-order
effects (gray lines in Figure 4) will help to determine which
"slices" of the solution space to plot in two dimensions to
better understand the relationship of the variables.
The 16 % total sensitivity of the wind velocity warrants a
discussion of the limitation of the annual wind velocity method,
see Figure 5 (a). First, the number of wind directions simulated
determines the accuracy of the result, although, the increase
shows diminishing returns. Although the authors are aware that
a higher number of wind directions will increase the overall
accuracy of the annual wind velocity magnitudes, we refrained
from simulating additional wind direction in the interest of
time. Further, for each of the 8 simulations, a constant inlet
wind velocity of
5𝑚/𝑠
at
10 m
was assumed, regardless of the
distribution of wind directions and their corresponding hourly
wind velocities. This was justified by Becker et al. [2] that
have shown that the reattachment length behind a cube does
not change significantly for Reynolds numbers greater than
1
·105
which applies for this study. In general, however, the
wind velocities between the buildings on campus are likely
overestimated as not all surrounding buildings on campus have
been included in the CFD simulation.
The range of the averaged annual MRT in Figure 5 (b) seems
reasonable as the average dry-bulb temperature is 10
C. A
closer inspection of the annual MRT time series revealed
that the shaded MRT is generally very close to the dry-bulb
temperature, whereas the nighttime MRT lies
10
C below.
This confirms the findings by [12]. The current simulation
framework, however, does not take into account radiative heat
exchange and therefore does not take into account urban heat
islands that are likely to occur during summer. Although
Kessling et al. [12] argue that the difference between surface
and dry-bulb temperatures usually below
±
5
𝐾
even in very
hot climates, it is not clear if those findings are generalizable
for arbitrary building densities, other than Riyadh. Here, more
research is required to establish a more holistic and robust
workflow.
It is worth noting that the ranges for the spatially-resolved
sensitivity analysis in Figure 6 (a) and Figure 6 (b) differ from
those in Figure 4. That is to say that locally, the bounds for each
point differ significantly from the assumptions made in Table 1.
Moreover, the areas with high total-order sensitivity indices in
Figure 6 (a) and Figure 6 (b) confirm what was expected by
the average annual UTCI distribution. The areas colored in
red may be interpreted as the areas for which the respective
variable (wind velocity / MRT) is controlling most of the
variance in the UTCI output. For example, in areas where
the wind velocity is high on average (north-south corridors),
the wind velocity is relatively dominant. Besides, Figure 6
(b) suggests that wind shelters might be beneficial to improve
the annual outdoor comfort for the seating arrangements to
the west of the building as the MRT variance in the output is
relatively low. Insights from such visualizations will prove
to be helpful in other studies when decision-makers decide
about possible interventions to promote good outdoor thermal
comfort.
Future studies should combine spatial and temporal sensitivity
studies. A preliminary sensitivity analysis of the UTCI for
different climate zones has shown that the complexity of
outdoor comfort metrics typically causes a subset of parameters
to be inactive at any particular time; and that the inactive input
variables differ from climate zone to climate zone. This
sparsity of activation may lead to needless computational
complexity and inappropriate assumptions of parameters that
are inactive in the first place. Combining spatial and temporal
sensitivity analyses (ideally by season) might present a valuable
opportunity to overcome the complexity of simulation engines.
In effect, this would be achieved by restricting the variable input
space to only those parameters which are actively contributing
to the output at a specific time and location for the particular
climate zone.
5 CONCLUSIONS
Exploring parameter activation at the spatial and temporal
scales is important not only for diagnostic analyses of biome-
teorological indices such as the UTCI but also to provide
actionable insights when deciding about potential interven-
tions. This study represents a novel step in this direction by
visualizing the spatial sensitivity of the UTCI and exempli-
fies how to derive the outdoor space usage as a proxy for
the attractiveness of outdoor spaces. We conclude that the
spatial variability of any outdoor comfort metric can easily
be visualized which provides valuable information about its
behavior. As the availability of computing power continues to
increase, we anticipate the community to look beyond results
only; instead, we anticipate using them to derive actionable
insights which to date remained largely unexplored.
REFERENCES
1. Arens, E., Hoyt, T., Zhou, X., Huang, L., Zhang, H., and
Schiavon, S. Modeling the comfort effects of short-wave
solar radiation indoors. Building and Environment 88
(2015), 3–9.
2. Becker, S., Lienhart, H., and Durst, F. Flow around
three-dimensional obstacles in boundary layers. Journal
of Wind Engineering and Industrial Aerodynamics 90,
4-5 (2002), 265–279.
3. Bourgeois, D., Reinhart, C. F., and Ward, G. Standard
daylight coefficient model for dynamic daylighting
simulations. Building Research & Information 36, 1
(2008), 68–82.
4. Bröde, P., Fiala, D., Błażejczyk, K., Holmér, I.,
Jendritzky, G., Kampmann, B., Tinz, B., and Havenith, G.
Deriving the operational procedure for the universal
thermal climate index (utci). International journal of
biometeorology 56, 3 (2012), 481–494.
5. Broede, P., Blazejczyk, K., Fiala, D., Havenith, G.,
Holmer, I., Jendritzky, G., Kuklane, K., and Kampmann,
B. The universal thermal climate index utci compared to
ergonomics standards for assessing the thermal
environment. Industrial health 51, 1 (2013), 16–24.
6. Bui, Q., and White, J. Mapping the shadows of new york
city: Every building every block. New York Times (2016).
7. EnergyPlus Documentation. Engineering Reference -
EnergyPlus 8.9. The Reference to EnergyPlus Calculation
(2019).
8. Fiala, D., Havenith, G., Bröde, P., Kampmann, B., and
Jendritzky, G. Utci-fiala multi-node model of human heat
transfer and temperature regulation. International journal
of biometeorology 56, 3 (2012), 429–441.
9. Herman, J., and Usher, W. Salib: an open-source python
library for sensitivity analysis. Journal of Open Source
Software 2, 9 (2017), 97.
10. Kastner, P., and Dogan, T. Towards high-resolution
annual outdoor thermal comfort mapping in urban design.
In Proceedings of Building Simulation 2019: 16th
Conference of IBPSA (2019).
11. Kastner, P., and Dogan, T. A cylindrical meshing
methodology for annual urban computational fluid
dynamics simulations. Journal of Building Performance
Simulation 13, 1 (2020), 59–68.
12. KESSLING, W., ENGELHARDT, M., and
KIEHLMANN, D. The human bio-meteorological chart.
In PLEA (2013).
13. Kongsgaard, C. Utci parameter sensitivity, 2012.
14. Provençal, S., Bergeron, O., Leduc, R., and Barrette, N.
Thermal comfort in quebec city, canada: sensitivity
analysis of the utci and other popular thermal comfort
indices in a mid-latitude continental city. International
journal of biometeorology 60, 4 (2016), 591–603.
15. Reinhart, C. F., Dhariwal, J., and Gero, K.
Biometeorological indices explain outside dwelling
patterns based on wi-fi data in support of sustainable
urban planning. Building and Environment 126 (2017),
422–430.
16. Subramaniam, S. Daylighting simulations with radiance
using matrix-based methods. Lawrence Berke-ley
National Laboratory (2017).
17. Tumini, I., and Rubio-Bellido, C. Measuring climate
change impact on urban microclimate: A case study of
concepción. Procedia engineering 161 (2016),
2290–2296.
18. Whyte, W. H. City: Rediscovering the center. University
of Pennsylvania Press, 2012.
19.
Whyte, W. H., et al. The social life of small urban spaces.
Conservation Foundation Washington, DC, 1980.
... Moreover, simulations considering the effect of windthermal coupling on the velocity distribution and flow field have been conducted (Hamdan and de Oliveira, 2019;Utkucu and Sözer, 2020a;Loh and Bhiwapurkar, 2021). Several studies have coupled airflow, solar radiation, and thermal simulations to perform thermal comfort assessments (Johansson and Yahia, 2020;Sun and Rao, 2020;Lin et al., 2021) and energy consumption calculations Xu et al., 2019;Graham et al., 2020;Kastner and Dogan, 2020b;Natanian et al., 2020;Ibrahim et al., 2021b). However, pollutant dispersion simulations using CFD plug-ins during the design stages have not been widely studied (Ma et al., 2018;Jia et al., 2021). ...
... Statistically, Butterfly and Autodesk CFD are popular plug-ins typically applied in urban block studies. Moreover, the performance simulation tool Ladybug is often coupled with CFD plug-ins in urban block studies to perform solar radiation access calculations and daylight analysis, which supplements the outdoor thermal comfort assessments of urban blocks (Ma et al., 2018;Kastner and Dogan, 2020b). ...
... The turbulence characteristics in the vertical direction and the concentration distribution law of pollutants through the simulation results obtained by Autodesk CFD were analyzed (Fig. 5(b and c)), supporting the authors in identifying the best cluster type that is conducive to PM 2.5 dispersion from the inside to the outside (Fig. 5(d)). Furthermore, previous studies have also analyzed total building energy consumption inside the street based on design platforms (Sharmin and Steemers, 2019;Kastner and Dogan, 2020b ...
Article
Full-text available
The transformation of urban and building design into green development is conducive to alleviating resource and environmental problems. Building design largely determines pollutant emissions and energy consumption throughout the building life cycle. Full consideration of the impact of urban geometries on the microclimate will help construct livable and healthy cities. Computational fluid dynamics (CFD) simulations significantly improve the efficiency of assessing the microclimate and the performance of design schemes. The integration of CFD into design platforms by plug-ins marks a landmark development for the interaction of computer-aided design (CAD) and CFD, allowing architects to perform CFD simulations in their familiar design environments. This review provides a systematic overview of the classification and comprehensive comparison of CFD plug-ins in Autodesk Revit, Rhinoceros/Grasshopper, and SketchUp. The applications of CFD plug-ins in urban and building design are reviewed according to three types: single-objective, multi-objective, and coupling simulations. Two primary roles of CFD plug-ins integrated into the design process, including providing various micro-scale numerical simulations and optimizing the original design via feedback results, are analyzed. The issues of mesh generation, boundary conditions, turbulence models, and simulation accuracy during CFD plug-in applications are discussed. Finally, the limitations and future possibilities of CFD plug-ins are proposed.
... The use of the wind factors is presentedin the next section. Even though a limited number of wind directions were used, they were more than the minimum of eight required to represent all the possible wind directions with an adequate level of accuracy to obtain reliable wind patterns for outdoor thermal comfort analysis[65]. ...
... A dimensionless matrix (wind factors) was obtained, dividing the wind velocities simulated for all the 16 directions and each grid sensor used, by the fixed wind velocity used for CFD simulations. Finally, the hourly measured wind directions were associated with the closest among the 16 directions used for simulations and the hourly measured wind speeds from the weather file were multiplied by the previously obtained wind factors[65]. ...
Article
Full-text available
Considering climate change, controlling outdoor microclimates is an increasingly pressing concern. Microclimates have a significant effect on both outdoor and indoor comfort, and on the energy efficiency of buildings. This concern is particularly important as current climate conditions reveal that warmer summers are threatening the comfort of pedestrians and causing overheating in office environments, which is consequently increasing cooling energy consumption. A further concern is that this trend now extends to Nordic latitudes. Existing literature demonstrates how a local microclimate depends on many factors such as urban density, shape and orientation of buildings, the types of materials present, the number of green areas and anthropogenic activities. However, there is little research focusing on how reciprocal distances among tall buildings, and their relative position, affect outdoor and indoor comfort, and the associated energy consumption of buildings. This paper presents a unique and comprehensive insight into the interconnected nature of indoor and outdoor comfort via coupled simulations. It presents a study of clusters of tall commercial buildings located in the Nordic climate of Tallinn (Estonia) with different microclimates, and shows that the differences are due to variable shadowing and reflections and different wind patterns. The results, which focus on summer conditions, show that small variations of cluster layout strongly affect the local indoor and outdoor comfort, thus highlighting the need to conduct both studies simultaneously in research aiming to increase pedestrian and indoor comfort and resource efficiency.
... Nasrollahi et al. [20] investigated the role of urban geometry on outdoor perception in the hot climate of Ahvaz, Iran. A few studies have addressed this aspect in cold climates [21]. Therefore, there is a lack of research considering the impact of urban typologies on the outdoor environment in cold climatic contexts that can be served as representative models of such approaches. ...
Conference Paper
In the context of climate change, the architectural sector slowly turns its interest toward holistically assessing effects of design on the neighborhood scale. In addition to the importance of residential energy performance, improving well-being in outdoor spaces should be considered in the design stage that can potentially decrease the cost and environmental impacts of buildings. Consequently, this work implements energy and comfort simulation tools to search for the best performing results of the existing condition of the built environment in Syracuse, New York, United States. This work explores the design features and characteristics of interconnected aspects of Low-rise, Mid-rise, and High-rise urban typologies, including energy performance and outdoor thermal comfort. Mid-rise multifamily buildings have shown the best performing results. The findings of this study contribute to identifying better design solutions that further will guarantee the predefined requirements of more energy efficient and responsive buildings in this climatic context.
Article
The architectural community needs holistic, evidence-based planning tools to promote urban resilience in the face of global warming. To ensure maximum impact, simulation-driven microclimate analysis methods must be integrated early in the design process. With Eddy3D, we present a toolkit to simulate outdoor thermal comfort (OTC) metrics with a decoupled approach. We motivate the decoupled systems framework with meteorological measurements and local and global sensitivity analyses of three different climates. For a real-world case study on a university campus, we present results for both wind velocity and mean radiant temperature simulations. Finally, we discuss the advantages and disadvantages of a decoupled simulation approach considering design aiding and the architectural. Our findings support reduced simulation time and flexibility, with the caveat of reduced accuracy due to neglecting forced convection, albeit this being less relevant in the early stages of design. The framework presented in this manuscript has been implemented and released as Eddy3D, a plugin for Rhino & Grasshopper.
Article
In the physics-based simulation of urban geometries, the outdoor environment was usually simulated separately from buildings – until recently, when the holistic assessment of the urban environment began to attract more attention. Although analyzing design alternatives with multiple objectives is still a challenge, computational tools enable generating thousands of scenarios to rapidly assess performance corresponding to a specific goal. In this study, we developed a multi-phase optimization framework for conceptual urban design. We tested this framework for urban typologies in Syracuse. The energy performance of each alternative was compared with a baseline. The alternatives that generate wasteful energy performance were filtered out first, then remaining scenarios that performed better than the baseline were analyzed using outdoor thermal comfort autonomy (OTCA). Mid-rise multifamily buildings showed the best performance (55.8% energy improvement compared to the baseline). Although hot week outdoor comfort satisfaction among selected mid-rise typologies was high (92.9–98.5%), the satisfaction in cold week was very low (between 8.4–11.6%) among them. This framework contributes to identifying an acceptable range of design solutions by broadening the perspective of the field toward using a more customized optimization framework in early design that will further guarantee the requirements of energy efficient and sustainable cities.
Conference Paper
Full-text available
Global warming and increasingly dense cities lead to poor outdoor thermal comfort that may not only be detrimental to our health and well-being but also decreases social and commercial activities. Although workflows for the analysis of thermal comfort exist, they have yet transitioned into the quotidian architectural design process. Our work-flow allows for annual outdoor comfort analyses that are seamlessly integrated into a commonly-used CAD environment. We simulated the annual outdoor thermal comfort on a university campus and discuss which simplifications seem appropriate by means of preliminary on-site measurements. The results exemplify the possibility to conduct such analyses within reasonable time and accuracy if some simplifications to the UTCI estimation are acceptable.
Technical Report
Full-text available
A tutorial for performing annual matrix-based simulations with the Radiance rendering system. The following simulations are covered: Daylight Coefficients, The Three-Phase Method, The Five-Phase Method, The Four-Phase Method with non-coplanar modeling and The Six-Phase Method. The exercise files for the examples detailed in the tutorial can be downloaded from: https://www.radiance-online.org/learning/tutorials/radTutorialFiles-master.zip
Article
Full-text available
SALib contains Python implementations of commonly used global sensitivity analysis methods, including Sobol (Sobol’ 2001, Andrea Saltelli (2002), Andrea Saltelli et al. (2010)), Morris (Morris 1991, Campolongo, Cariboni, and Saltelli (2007)), FAST (Cukier et al. 1973, A. Saltelli, Tarantola, and Chan (1999)), Delta Moment-Independent Measure (E. Borgonovo 2007, Plischke, Borgonovo, and Smith (2013)) Derivative-based Global Sensitivity Measure (DGSM) (Sobol’ and Kucherenko 2009) , and Fractional Factorial Sensitivity Analysis (Andrea Saltelli et al. 2008) methods. SALib is useful in simulation, optimisation and systems modelling to calculate the influence of model inputs or exogenous factors on outputs of interest. SALib exposes a range of global sensitivity analysis techniques to the scientist, researcher and modeller, making it very easy to easily implement the range of techniques into typical modelling workflows. The library facilitates the generation of samples associated with a model’s inputs, and then provides functions to analyse the outputs from a model and visualise those results.
Article
Full-text available
Cities, due to the concentration of population, economic activities and built infrastructures, are high-risk and potential damage areas in global warming scenarios. Specific climate conditions, which occurs in urban areas, are expected to change with almost certainly disadvantageous effects on quality of life at outdoor spaces. This paper presents an evaluation of climate change impact on cities, by simulating the thermal performance of outdoor spaces, using different climate scenarios. The Plaza de Armas of Concepción (Chile) has been used as case study, and climate conditions at 2020, 2050 and 2080 have been produced for the acknowledged A2’medium-high’ emissions Greenhouse Gases GHG scenario, according to Intergovernmental Panel on Climate Change IPCC. To simulate the urban microclimate, the ENVI-met v4 software, a three-dimensional numerical model based primarily on fluid dynamic and thermodynamic models has been used to estimate interaction between surfaces, plant and atmosphere at the micro-scale. The comparison between different scenarios and discussion about results are presented as well. The results and discussion are a first step to advance on the knowledge of climate change impact for proposing adaptation strategies in the urban context.
Article
Full-text available
The growing need for valid assessment procedures of the outdoor thermal environment in the fields of public weather services, public health systems, urban planning, tourism & recreation and climate impact research raised the idea to develop the Universal Thermal Climate Index UTCI based on the most recent scientific progress both in thermo-physiology and in heat exchange theory. Following extensive validation of accessible models of human thermoregulation, the advanced multi-node 'Fiala' model was selected to form the basis of UTCI. This model was coupled with an adaptive clothing model which considers clothing habits by the general urban population and behavioral changes in clothing insulation related to actual environmental temperature. UTCI was developed conceptually as an equivalent temperature. Thus, for any combination of air temperature, wind, radiation, and humidity, UTCI is defined as the air temperature in the reference condition which would elicit the same dynamic response of the physiological model. This review analyses the sensitivity of UTCI to humidity and radiation in the heat and to wind in the cold and compares the results with observational studies and internationally standardized assessment procedures. The capabilities, restrictions and potential future extensions of UTCI are discussed.
Article
For urban CFD simulations, it is considered a best practice to use a box-shaped simulation domain. Box-shaped domains, however, show drawbacks for airflow from several wind directions as remeshing and additional preprocessing steps become necessary. We introduce a routine to create a cylindrical mesh that expedites the simulation of arbitrary wind directions using OpenFOAM. Results computed with the cylindrical domain are validated against wind tunnel data. We report that the cylindrical method yields comparable results in terms of accuracy and convergence behaviour. Further, run time comparisons in a real-world scenario are conducted to discuss its advantages and limitations. Based on the findings, we recommend using the cylindrical approach if at least eight wind directions are analyzed for which we report 18% run time savings. The cylindrical domain along with automated best practice boundary conditions has been implemented in Eddy3D – a plugin for Rhinoceros.
Article
This study presents a novel approach to validate the capability of biometeorological indices to predict the likelihood of urban dwellers to be outside during midday. Over a period of ten months three Wi-Fi scanners were used in a public courtyard in Cambridge, MA, to record outside dwelling patterns. Based on encrypted MacIDs courtyard attendees could be divided into 16,000 regulars and 676,000 visitors. Universal Thermal Climate Index (UTCI) predictions based on a combination of measured microclimatic conditions and mean radiant temperature simulations using ENVI-met were shown to strongly correlate with the number of regulars present during lunchtime with coefficients of determination (R2) of 92% during spring and 70% during summer/fall, respectively. Lunchtime attendance peaked for UTCI values in the thermal comfort and moderate heat stress ranges. In parallel, the probability for regulars to have lunch outside more than doubled during those UTCI conditions and the median lunchbreak length increased from 8 min to 12 min. These findings suggest that UTCI can be used as a reliable environmental performance metric to support the design and preservation of comfortable outdoor spaces. The reported use of public Wi-Fi data can help city governments to better understand – and potentially improve – the use of outdoor spaces while maintaining the privacy of their constituents.
Article
Named by Newsweek magazine to its list of "Fifty Books for Our Time." For sixteen years William Whyte walked the streets of New York and other major cities. With a group of young observers, camera and notebook in hand, he conducted pioneering studies of street life, pedestrian behavior, and city dynamics. City: Rediscovering the Center is the result of that research, a humane, often amusing view of what is staggeringly obvious about the urban environment but seemingly invisible to those responsible for planning it. Whyte uses time-lapse photography to chart the anatomy of metropolitan congestion. Why is traffic so badly distributed on city streets? Why do New Yorkers walk so fast-and jaywalk so incorrigibly? Why aren't there more collisions on the busiest walkways? Why do people who stop to talk gravitate to the center of the pedestrian traffic stream? Why do places designed primarily for security actually worsen it? Why are public restrooms disappearing? "The city is full of vexations," Whyte avers: "Steps too steep; doors too tough to open; ledges you cannot sit on. . . . It is difficult to design an urban space so maladroitly that people will not use it, but there are many such spaces." Yet Whyte finds encouragement in the widespread rediscovery of the city center. The future is not in the suburbs, he believes, but in that center. Like a Greek agora, the city must reassert its most ancient function as a place where people come together face-to-face.
Article
The newly developed Universal Thermal Climate Index (UTCI), along with the physiological equivalent temperature (PET), the humidex (HX) and the wind chill index (WC), was calculated in Quebec City, Canada, a city with a strong seasonal climatic variability, over a 1-year period. The objective of this study is twofold: evaluate the operational benefits of implementing the UTCI for a climate monitoring program of public comfort and health awareness as opposed to relying on traditional and simple indices, and determine whether thermal comfort monitoring specific to dense urban neighborhoods is necessary to adequately fulfill the goals of the program. In order to do so, an analysis is performed to evaluate each of these indices' sensitivity to the meteorological variables that regulate them in different environments. Overall, the UTCI was found to be slightly more sensitive to mean radiant temperature, moderately more sensitive to humidity and much more sensitive to wind speed than the PET. This dynamic changed slightly depending on the environment and the season. In hot weather, the PET was found to be more sensitive to mean radiant temperature and therefore reached high values that could potentially be hazardous more frequently than the UTCI and the HX. In turn, the UTCI's stronger sensitivity to wind speed makes it a superior index to identify potentially hazardous weather in winter compared to the PET and the WC. Adopting the UTCI broadly would be an improvement over the traditionally popular HX and WC indices. The urban environment produced favorable conditions to sustain heat stress conditions, where the indices reached high values more frequently there than in suburban locations, which advocates for weather monitoring specific to denser urban areas.