Wind engineering in the integrated design of princess Elisabeth ...

Wind engineering in the integrated design of princess Elisabeth Antarctic base
Javier Sanz Rodrigo *, Jeroen van Beeck, Jean-Marie Buchlin
von Karman Institute for Fluid Dynamics (VKI), Chaussée de Waterloo 72, B-1640 Rhode-St-Genèse, Belgium
article info
Article history:
Received 15 October 2011
Received in revised form
19 December 2011
Accepted 29 December 2011
Keywords:
Antarctica
Integrated design
Wind engineering
Snowdrift
Wind loading
Sand erosion
1. Introduction
abstract
The concern about Climate Change has significantly increased
the interest of scientific research in Antarctica. Coinciding with the
International Polar Year (IPY) 2007e2008, three European research
bases were designed and have recently been deployed in Antarctica:
Halley VI (United Kingdom), Newmayer III (Germany) and
Princess Elisabeth (Belgium).
The new stations put in evidence the primary role that the
environmental conditions play on their design, especially when
higher levels of sustainability and energy efficiency are pursued, as
it is the case for the Princess Elisabeth “Zero Emission” research
station.
The integrated design process, implemented in the case of the
Princess Elisabeth base, aims at making best use of the ambient
conditions in the design of energy efficient buildings with the least
impact on the environment throughout their lifetime. The use of
renewable energies in Antarctica as primary energy source
provides more autonomy, minimizing the fuel consumption with
the corresponding savings in logistics, CO2 emissions and risks of oil
spill contamination.
* Corresponding author. National Renewable Energy Centre of Spain (CENER),
C/Ciudad de la Innovación 7, 31621 Sarriguren, Spain. Tel.: þ34 948 25 28 00;
fax: þ34948270774.
E-mail address: jsrodrigo@cener.com (J. Sanz Rodrigo).
0360-1323/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2011.12.023
Building and Environment 52 (2012) 1e18
Contents lists available at SciVerse ScienceDirect
Building and Environment
journal homepage: www.elsevier.com/locate/buildenv
The Belgian Antarctic Base Princess Elisabeth is based on an elevated building on top of sloping terrain
and connected to an under-snow garage. The integrated design of the base was supported by wind
engineering testing that looked into building aerodynamics (pressure taps) and snowdrift management.
Wind tunnel modeling using sand erosion technique allowed efficient evaluation of the snow erosion
and deposition around different building-block shapes during the conceptual design phase. Parametric
testing shows that the positioning of the main building on the ridge has a significant impact on wind
loading and snow erosion and deposition. Important reductions in wind loading and snow deposition
can be obtained by elevating the building and reducing the windward cantilever. The positioning of the
garage roof can further decrease the wind loading by acting as a diffuser in the back of the building. This
study shows that, not only for safety and cost reduction but also for the integration of renewable
energies, important benefits in the design of a building can be achieved if wind engineering is considered
since the conceptual phase of the integrated building design process.
Ó 2012 Elsevier Ltd. All rights reserved.
The extreme weather conditions make Antarctic construction
the forefront of Bioclimatic Architecture, with katabatic winds up to
70 m s 1 that induce important structural loading; snow drifting
that can produce annual build-ups as high as 1.5 m, making
accessibility difficult and increasing maintenance works; temperatures
as low as 60 C that induce important heat losses; and very
low humidity that increases the risk of fire. Furthermore, the
Antarctic environment constitutes an excellent test bench for the
demonstration of renewable energies and energy efficiency
technology.
All in all, it is evident that building design in Antarctica
requires careful consideration of the environment in order to find
safe and cost-effective solutions with the least impact on the
environment, a key aspect of the Antarctic-Environmental
Protocol (1991).
1.1. Philosophy of a zero emission station
The Princess Elisabeth Antarctic research station is situated
approximately 1 km North of Utsteinen Nunatak, on a small and
relatively flat granite ridge (71 57 0 S23 20 0 E, 1390 m a.s.l.), 173 km
inland from the former Roi Baudouin base and 55 km from former
Japanese Asuka station. The new station occupies the empty space,
in the 20e30 East sector, left by the closing of Asuka station in
1992. The nearest permanent stations are Syowa (Japan), 684 km to
the west, and the Novolazarevskaya (Russia), 431 km to the east.
The nearest coast is some 190 km north.

2
The selected site is very convenient as it provides stable ground
for anchoring. The station design makes best use of terrain conditions
for the integration of the building following a hybrid design.
The main building, above ground-level and anchored onto snowfree
rock area is connected, with a weather protected bridge, to
an adjacent garage/storage building, constructed under the snow
surface (Fig. 1). The summer station is designed for optimal use by
12 people with a surface area (living, technical, research, storage) of
800 m 2 . An extension, based on heated shelters, make it possible to
accommodate another 8 to 18 people.
The system design of the station is based on sustainable technology
and high-energy efficiency, with full-year monitoring and
remote sensing capability. The station aims at being zeroemissions,
making use of renewable energy as the primary
energy source and integrating passive building design in
a comprehensive energy management regime, thereby minimizing
the use of fossil fuels. The power budget of the station is composed
of 48% of wind power from nine wind turbines, 20% of solar
photovoltaic from 380 m 2 of solar panels and 12% solar thermal
with 22 m 2 of solar panels.
The internal layout of the main building is designed with
concentric layers around a central technical core, which holds the
control systems, the water treatment unit and the batteries for
energy storage. Around the technical core, the kitchen and laundry
rooms and the sleeping and living rooms are distributed. A
substantial contribution to the zero-emissions target is met by
having very good insulation, with a stainless steel outer skin and 7
insulation layers in the walls and triple glazed windows. Passive
heating is also an important energy saver because it recycles the
heat produced inside the building.
The interested reader should refer to the Comprehensive Environmental
Evaluation CEE report [1] for a broader description of the
scope of the base and its design particularities. A dedicated website
(http://www.antarcticstation.org) is also available for the follow up
of the station activities.
1.2. The integrated building design process
The Princess Elisabeth base was designed by the International
Polar Foundation (IPF). An integrated building design approach was
followed whereby multiple design disciplines were assembled
from the beginning of the project to obtain a highly synergic design
that allows optimizing the performance and efficiency of the
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
building. This kind of approach is the state of the art in building
design and it is being adopted in singular buildings or building
complexes where high levels of energy efficiency and sustainability
are pursued.
A key to successful integrated building design is the early
participation of experts from different specialties: civil engineering,
architecture, interior and landscape design, energy and
wind engineering, etc. The early collaboration among them
allows finding opportunities at conceptual level that can produce
a very significant impact in the final performance and cost of the
design.
In the Antarctic integrated building design process, wind engineering
(wind and snowdrift assessment) and energy efficiency
lead decisions about the optimal positioning, orientation and shape
of the building. The integration of renewable energies and operational
aspects like the positioning of entrances, emergency exits or
snow collection facilities (for water consumption) are also determined
after careful assessment of the environmental conditions of
the building envelop.
The aerodynamic design of Princess Elisabeth station had three
phases: the conceptual design phase, the building envelop design
and the optimization phase. In the conceptual phase, the designers
had to decide about the optimum building typology by trading off
basic design parameters on a number of building-block concepts.
Once the basic typology was selected, the building envelope was
shaped and positioned on the ridge considering both internal
constraints (internal layout and system integration) and external
environmental aspects. From the wind engineering point of view,
snowdrift control and wind loading had to be tested in order to
assess the aerodynamic performance of different building prototypes
and ridge integration alternatives. Finally, an optimization
phase looked at more detailed elements of the building like, for
instance, localized forces on the corners of the building or the
integration of the under-snow garage.
1.3. Wind engineering strategy
Fig. 1. Sketch of the building and its integration in the ridge.
Wind engineering constitutes a crucial aspect in the design of
modern Antarctic bases [3]. Not only for safety and cost reduction
but also for the integration of renewable energies, important
benefits in the design of a building can be achieved if wind engineering
is part of the integrated building design process since the
very beginning.

All modern buildings have undergone aerodynamic studies at
different stages of their design. Delpech et al. [4] simulated the
snowdrift around Concordia station using real snow in a climatic
wind tunnel. Waechter and Williams [5] used water flume and CFD
modeling of snowdrifts to support the design of a new building for
the Amundsen Scott base at the South Pole. Beyers and Harms [6]
made field tests of snow accumulation in the vicinity of SANAE IV
station using a reduced scale model of the building. Leitl et al. [7]
performed wind tunnel simulations of Neumayer III station on
snowdrift and wind loading. The design of the new Halley VI station
has also been supported with CFD simulations (not published).
The wind engineering aspects of Princess Elisabeth station were
tested at the von Karman Institute L1-B wind tunnel with support
from numerical CFD models at different stages of the design. This
paper will only deal with the experimental part. Information about
the numerical aspects can be found in Ref. [2].
Snowdrift modeling in wind tunnel is a difficult task due to the
inherent impossibility to maintain similarity of all the driving
forces. It is also very costly and time consuming as it takes several
hours to develop significant build-ups. As a result, it was not
feasible to use this technique in the very demanding conceptual
design phase, where many building configurations had to be tested
in a very limited amount of time. Instead, the sand erosion technique
proved to be a cost-effective solution to evaluate the wind
conditions at ground-level and, at the same time, identify snow
accumulation and erosion regions. The application of this technique
to snowdrift assessment is published on this paper for the first time.
The sand erosion technique allowed the evaluation of six
different building-block concepts with several ridge integration
strategies. Prismatic blocks were used in order to offer the best
perspective for a systematic comparison of the different design
concepts. Two block concepts offered similar aerodynamic performance:
a square-based one-storey building or a rectangular-based
two-storey building aligned with the wind direction.
After a site visit during Belare-2005 expedition, IPF selected the
location on the ridge offering the best conditions for anchoring and
accessibility. It also turned out that a one-storey building would be
preferable for a better compatibility with the internal layout of the
building, which would be based on a concentric architecture
around a technical core reserved for the system installations. The
square-based model had better compatibility with this concentric
concept than the rectangular-based concept. Hence, it constituted
the reference for the building envelop design phase in which the
square building would be shaped and positioned to obtain better
aerodynamic performance.
The reference building-block was instrumented with pressure
taps to measure wind loading at different building-ridge integrations.
A parametric study showed that the height of the building
legs and the positioning across the ridge were sensitive parameters
for both aerodynamics and snowdrift control.
While elevating the main building with legs solves the problem
of snow accumulation, it might generate problems of snow erosion
around the under-snow garage. Hence, dune visualization tests
were performed to check the impact of the snow erosion generated
by the building on the integrity of garage.
Several evolutions of the envelop design were tested until the
adoption of a final one, whose aerodynamic performance was again
tested with pressure taps to assess the final loads and to optimize
the final positioning on the ridge. During this phase, wind tunnel
testing also revealed some aerodynamic aspects of the garage roof,
which could be used as a diffuser to modify the wind loading on the
building.
These wind tunnel testing phases will be summarized in this
paper. The test case shows how wind engineering can aid the
integrated building design process and the benefits of considering
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 3
these aspects from an early stage. The originality of the test case can
be focused on the study of the influence of the flow under the
building on the aerodynamic aspects of the building. Elevated
buildings on pillars are the typical design solution in snow regions
to cope with snowdrifts but they are typically placed in flat terrain.
Having sloping terrain under the building makes the building
aerodynamics less predictable. It is shown in the paper that by
carefully positioning the building on top of the ridge important
reductions in wind loading and snowdrift can be obtained.
2. Experimental setup
2.1. Wind tunnel site and ABL model
A model of the southern part of the ridge topography at a scale
1:100 was built and installed at the 2.8 m diameter turning table of
the VKI L1-B atmospheric boundary layer wind tunnel (Fig. 2, left).
The wind tunnel is of the close-circuit type, equipped with two
contrarotating fans of 580 kW that forces wind speeds up to
60 m s 1 . The rectangular wind engineering test section is 2 m high,
3 m wide and 20 m long with a roughed floor to allow the growth of
a turbulent boundary layer similar to the lower part of the atmospheric
boundary layer (ABL).
The site topography is made of 3 mm thick wooden layers
smoothed up with high definition plaster within the steps. Of
course, the model represents both the ridge and the snow surface
as it was found in Antarctica. Once the building is situated on top of
the ridge, the snow surface behind it will change under the action
of the erosion and deposition generated by the building.
Since no Coriolis forces and thermal stratification can be
reproduced, the test section is suitable for the characterization of
the surface boundary layer in neutral conditions. This is suitable for
the purposes of this investigation since we are interested in
snowdrift and wind loading at high wind speeds in the first 20 m
above the ground. At low wind speeds the Antarctic boundary layer
is characterized by stable stratification due to a quasi-permanent
temperature inversion due to radiative cooling.
The incoming surface boundary layer is modeled in the wind
tunnel using a 20 m rough floor that reproduces the logarithmic
wind profile according to Monin Obukhov theory:
U ¼ u * z
ln (1)
k z0 where u * ¼ ffiffiffiffiffiffiffi p
s=r is the friction velocity, s is the shear stress, r is
the air density, k ¼ 0.4 is the von Karman constant, and z0 is the
roughness length. Beyers and Harms [6] also used this logarithmic
profile to fit measurements in the vicinity of SANAE IV station. The
effective roughness height above snow covered terrain is proportional
to the saltation layer thickness, which varies with the square
of the friction velocity.
u
z0;eff ¼ C1 2 *
2g
where g ¼ 9.8 m s 2 is the acceleration of gravity and the constant
C1 depends on the snow properties. Beyers and Harms [3] present
values from different authors ranging from 0.035 to 0.12. The
effective roughness is low for wet snow and high for fresh snow
cover after snowfall. At SANAE IV they found a value of 0.0035 over
wet snow for friction velocities above 0.4 m s 1 . The effective
roughness height was between 2.8$10 5 m and 1.8$10 4 m for
friction velocities ranging from 0.4 to 1 m s 1 respectively. For dry
uncompacted snow the saltation layer and roughness height will be
higher. Considering the upper limit of C1 for high effective
(2)

4
roughness conditions, the values of z0 can range between 10 4 m
and 6$10 3 m for the same range of friction velocities. Therefore,
the variability of the effective roughness length can be quite large,
although always very low, as it depends on the snow properties and
the friction velocity.
The sensitivity associated to the ABL is assessed by testing two
ABL classes: Class I or ‘smooth ABL’ (z0 ¼ 2$10 4 m, 7% turbulence
intensity at building height, in full-scale units), close to the expected
site conditions, and Class IV or ‘rough ABL’ (z0 ¼ 0.4 m, 20%
turbulence intensity at building height, in full-scale units), typical
of the urban environment. The class I boundary layer is obtained by
removing the roughness elements of the class IV boundary layer.
The urban boundary layer is of course an unrealistic extreme case
for Antarctica but it is interesting to check the sensitivity of the
building aerodynamics to the incoming wind profile since the
incoming ABL is not known a priori. Nevertheless, it is expected
that snow erosion and deposition will be dominated by the
disturbance generated by the building rather than by the incoming
flow conditions.
2.2. Wind climate from onsite AWS data
An automatic weather station (AWS) was placed at the southernmost
edge of the ridge since December 2004. During the winter
2005, a station malfunction resulted in loss of measurements
between the 2nd of July and the 14th of August. Nevertheless, the
annual data availability is sufficient to assess the local wind climate
throughout the year 2005, with an annual 4-m mean velocity of
5.9 m s 1 and a prevailing wind direction sector from E to SSE. The
most energetic wind direction is E with 90% of the energy
content (Fig. 3). A CFD simulation of the mean flow from this wind
direction results in the 4-m high speed-up contour map of Fig. 2
(right) with respect to the AWS mean velocity. It is observed that
the building area in top of the ridge has 20% higher wind speeds
than the ones observed at the AWS position.
The estimated mean velocity is rather low due to the presence of
nearby Utsteinen and the Sor Rondane Mountains to the S-SE, that
shelter from the intense katabatic winds of the region. Wind speed
measured at Asuka Station between 1986 and 1991, 60 km northeast,
were twice as high due to a more exposed site to the katabatic
winds.
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
Fig. 2. Topographic model of the ridge at a scale 1/100 in the L1-B wind tunnel (left) and Eastern speed-up contours at 4 m with respect to the AWS velocity (right).
Even though the mean wind speed is not high, the site is characterized
by the presence of intense wind storms, especially during
the winter season. This is revealed by an annual velocity distribution
with a long tail characterized by a Weibull shape factor of 1.35.
2.3. Snowdrift assessment: sand erosion tests
The presence of strong winds in Antarctica is responsible for the
transport of large quantities of snow in the form of snowdrifts,
producing a variety of operational problems around buildings and
structures.
It is normal practice in Polar Regions to use elevated buildings as
they can passively remove the snow through the wind flow passing
under the building. Almost all modern stations use this basic
principle in their design. Some of them like Halley V, where the
snow accumulation rate is 1.5 m per year, even include jackable legs
to adapt to the progressively higher snow level.
The size of the building generated snowdrifts will depend,
among other factors, on the height of the pillars and the orientation
of the building against the wind. General guidelines about snowdrifts
around prismatic buildings on flat terrain can be found in
Refs. [8,9]. Elevated buildings present smaller snowdrifts by
increasing the height or the length of the building in the direction
of the wind. On the contrary, increasing the width perpendicular to
the wind direction increases the size of the snowdrifts. Kwok et al.
[10] studied the optimum spacing between groups of buildings in
order to minimize snowdrift. Kwok and Smedley [11] studied the
effect of the corner geometry and the wind incidence on the
snowdrift size. A very significant reduction in snowdrift is obtained
by chamfering or rounding the corners with respect to a sharpedged
building model. Hence, the chamfer should be as large as
the internal layout allows. Irrespectively of the corner geometry
they also recommended to align the building with the main wind
direction and use longer pillar heights to reduce the volume of the
snowdrift. Chamfering the corners is also a well known technique
to decrease wind loading on the building facades.
In the case of Princess Elisabeth base, the presence of the ridge
sticking out of the snow surface constitutes a particular situation in
which the wind speed-up generated between the building and the
ridge can be used to enhance the snow removal efficiency of the
building, allowing for the use of shorter pillars. The same principle is

used in the new AmundseneScott base in the South Pole, where the
bottom edge of the building facing the wind is shaped like an airfoil
to accelerate the flow and increase the erosive action on the snow.
The snow drifting process is described in detail in Ref. [12]. The
requirements for wind tunnel modeling of snowdrifts around
obstacles can be found from different authors [9,13e15]. Geometric
similitude of snowdrifts is fulfilled if the model and prototype wind
fields are similar because the snowdrift development is driven by
the wind shear stress. Hence, standard wind engineering requirements
can be adopted to obtain flow similarity [16]. Apart from
scaling the topography and building geometry and the ABL profile,
In the case of bluff body aerodynamics it is standard practice to
assume flow similarity when a fully turbulent regime is attained,
which is achieved with Re > 10 4 .
Testing the snowdrift development rate requires length experiments
at low wind speeds in order to avoid too much Froude
distortion [9]. Since the interest of the design process was rather on
the comparison of different building typologies, in the conceptual
and envelope design phases, than on the prediction of the amount
of snow build up, partial snowdrift modeling was adopted. The
alternative model is based on the sand erosion or scour technique,
which is used in wind engineering studies related to the assessment
of the wind comfort around buildings [17e19], i.e. the
prediction of the occurrence of high level winds at ground (i.e.
pedestrian) level in the urban environment.
The test procedure is straightforward. A thin layer of sand
(w3 mm) is spread all over the floor of the test section. A uniform
freestream velocity is set in the wind tunnel for a certain time
(1 min) such that a quasi-steady erosion contour is obtained. This
contour indicates lines where the friction velocity reaches the
threshold for saltation. By increasing the freestream velocity in
progressive steps (from 5 to 13 m s 1 in steps of 0.5 m s 1 ),
a family of contours is obtained, each one associated to a particular
value of the freestream velocity and the same value of the local
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 5
Fig. 3. AWS wind speed and direction distributions (JJA corresponds to summer season and DJF to winter season).
(threshold) friction velocity. A picture is taken from the top at the
end of each time step and a contour detection algorithm extracts
the contours that are gridded to obtain a contour map of erosion
wind speeds.
Even though some advances have been made in the quantification
of the sand erosion tests as a measurement of the shear
stress [20,21], the technique is still mostly used as a visualization
tool in order to spot areas with high or low skin friction. Livesey
et al. [18] compared the wind speeds inferred from the sand erosion
technique with those measured with a hot-film anemometer. He
found that the sand erosion measurement, rather than the mean
speed, was related to the gust speed, defined as the mean plus one
standard deviation. Nevertheless, the wind speeds inferred had
considerably large variability. As a result, it was concluded that the
technique was suitable to identify areas of high relative winds and,
as such it shall be used in comparative studies. This capability will
be used in the present context to infer the action of the building on
the snow surface around it by comparison with the situation
without the building and, as a result, distinguish areas of snow
deposition and erosion. Therefore, the sand erosion technique is
performed in two steps: first without the building (reference case)
and then with the building model at the position and orientation of
interest.
Being the erosion contours related to a gust factor [18] the
contours never reach a steady state shape because they are all the
time eroded under the action of turbulence gusts. Nevertheless, in
the first minute after a change in the freestream velocity, the largest
changes are produced as a result of the new mean wind speed [21].
Therefore, the contours recorded during the test are assumed to be
contributed mostly by the mean wind speed.
Once the two erosion speed contour maps are obtained, the
ratio between both will reflect an amplification factor, Ae,
a measure of the action of the building on the reference or background
flow filed:

6
Ae ¼ U 0
U b
where U0 is the erosion wind speed from the building-free case and
Ub is the erosion wind speed from the building case. That is, Ae > 1
implies a relative increase of the wind shear and Ae < 1 a decrease
of the wind shear. Then, in areas where Ae is equal or close to 1, the
building does not change the background surface wind flow.
The sand, as the snow, is eroded when the shear stress of the
wind acting on the ground exceeds a certain threshold. This
condition is satisfied, in the case of the sand employed in this study,
for a threshold friction velocity of 0.23 m s 1 . The contours obtained
from the sand erosion patterns delimit the positions at which the
friction velocity is near the threshold value, separating the erosion
(friction velocity higher than the threshold) from the deposition
areas (friction velocities bellow the threshold) for a given velocity.
According to Anno [14] “the geometrical similitude of a border line
between the eroded area and the deposited area constitute the
most important similitude in the modeling of a snowdrift since the
snowdrift would be formed as a combination of erosion and
deposition”. Hence, provided that the flow similitude parameters
are satisfied, the extension of the snow deposition areas will be
delimited with the contour at which the local friction velocity
equals the threshold friction velocity. The scaling of the snowdrift
areas is directly related to the ratio of friction velocity to threshold
friction velocity (u*/u*t, ratio between the wind shear stress and the
particle cohesive forces). Hence, the scale factor between the field
velocity and the wind tunnel velocity is directly the ratio of
threshold friction velocities:
u * ¼ u *; WT
u *t
u *t;WT
; U ¼ U WT
u *t
u *t;WT
where the subscript WT denotes wind tunnel parameters.
The threshold friction velocity of the snow in the filed is difficult
to predict and ranges between 0.1 m s 1 for dry uncompacted snow
and 0.4 m s 1 for wind hardened snow [22]. Li and Pomeroy [23]
found threshold 10 m wind speeds ranging from 7 to 14 m s 1 for
wet snow and from 4 to 11 m s 1 for dry snow using meteorological
data from 16 stations in western Canada. Beyers and Harms [6]
found a threshold 10 m wind speed of 8 m s 1 , equivalent to
a threshold friction velocity of 0.28 m s 1 , from profile measurements
at SANAE IV base (Droning Maud Land, Antarctica). Mann
et al. [24] analyzed particle counter profiles of snow at Halley (Coats
Land, Antarctica) and found threshold friction velocities in the
range 0.22e0.36 m s 1 , depending on the time since the last
blowing snow deposition. Hence it seems that an average value of
0.3 m s 1 can be assumed for the threshold friction velocity of the
snow in Antarctica. Since it is 30% higher than that of the sand
particles used in the sand erosion tests, it means that the field
velocities will be scaled by a factor 1.3 of the wind tunnel velocities
according to Eq. (5).
Of course, the sand erosion technique is a surface visualization
and does not provide information about the development rate of
the height of erosion or deposition. Therefore, the comparisons
between the different building typologies will be based on the size
of the areas with Ae < 1, as an indicator of the efficiency of the
building to accumulate snow. Considering a typical threshold wind
velocity for snowdrift of 5 m s 1 at 4 m in the AWS position, 91% of
the snowdrift will come from the sector E-ESE (63%) and SE-SSE
(28%) sectors. Two prevailing snowdrift directions where selected
for testing during the design phase: 101 and 145 . The first one is
the most energetic and will bring the largest contribution to
snowdrifts as it is noticed from the snow patterns behind the ridge.
Fig. 4 shows some examples of photos obtained from erosion
tests. The wind speed is progressively increased from the top
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
(4)
(5)
(6 m s 1 ) to the bottom (12 m s 1 ). The left column presents the
building-free test which is used as a reference to compute the
amplification factor. The middle and right columns show two
erosion tests for a rectangular building, situated on top of the ridge
with two orientations: parallel and perpendicular to the incoming
wind direction (145 ). The building in both cases is elevated from
the ground leaving a clearance of one (full-scale) meter between
the building and the top of the ridge.
The areas out of the influence of the building present similar
erosion patterns to the building-free case. The small differences are
due to the inherent lack of repeatability of the manual sand
spreading process. Nevertheless, the reproducibility of the erosion
tests is quite good, with uncertainties on the erosion velocities
bellow 0.5 m s 1 .
At low erosion velocities, the first erosion patterns appear near
the corners and bellow the building, where the turbulence and
wind speed-up are the highest. At high velocities, only the sand in
the best sheltered areas remains. These areas will have good wind
comfort but will also build up snow. As expected, when the building
is oriented perpendicular to the prevailing wind direction, the
snow accumulation areas are increased both behind and in from of
the building. The accumulation in the back is not so problematic if
there is some space left between the building and the snowdrift,
where the main entrance would be placed. The snowdrift accumulated
in the front is more risky as it will progressively block the
flow underneath the building creating more build up in the back.
2.4. Garage integration: dune formation tests
The surface erosion tests are very helpful in the analysis of the
wind conditions at surface level but they do not offer much information
about the vertical development of erosion or deposition. As
the elevated building concept seems to have the snow accumulation
problem under control, it was important to assess the effect
that the building aerodynamics would have on the snow erosion
around the under-snow garage, situated just behind the main
building. Aerodynamically speaking the garage roof serves as
a platform that avoids the development of too large snow erosion in
the vicinity of the main building, providing a more stable ground to
give access to the building from the West.
To study the garage integration, an 80 cm wide portion of the
wooden ridge model was removed from the lee side at the location
of the building. A vertical cut of 8 cm, following the rock profile as it
was found on the site, gives room for the accommodation of
a45 11 5cm 3 garage. The height between the garage roof and
the ridge top was kept at 1 m in full scale. The empty space left was
filled up with sand plus an additional centimeter to cover the
garage entirely. This situation constitutes the setup for the dune
formation visualizations. These tests are done, similar to the surface
erosion tests, at progressively increasing speeds but now the time
for each velocity step is extended to 5 min to account for the longer
development of the dune patterns. The edges of the cut have been
placed sufficiently far away from the action of the building, making
use of the results from the sand erosion tests. This will ensure that
the dunes are only driven by the action of the wind and not by the
presence of extra vorticity from the edges of the cut. Interestingly,
at the end of the experiments the erosion patterns show a smooth
transition from the snow bed to the wooden profiles at the edges.
Since these profiles are obtained in real scale by erosion and
deposition generated by the ridge alone, this can be considered as
a good indication of the similarity of the snow erosion patterns in
the wind tunnel. In fact, by measuring the snow surface topography
before and after the construction of the station, Pattyn et al. [25]
confirmed that, apart from the immediate vicinity, the building
has a limited effect on the patterns of snowdrift behind the ridge.

The sand bed is very erodible, compared with the wind hardened
snow and ice bed, so the results cannot be directly related to
the field conditions. As with the erosion tests, the dune formation
test can be used to qualitatively compare the relative difference
between different building configurations.
2.5. Wind loading assessment: pressure taps tests
To study the building-ridge integration from the wind loading
point of view, a model of the reference and final building models
was manufactured with a distribution of pressure taps to map the
pressure acting on the building skin (Fig. 5).
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 7
Fig. 4. Example of erosion tests: Building-free test as reference (left column), rectangular building oriented along the incoming wind direction (middle column) and perpendicular
to the incoming wind (right column). The wind direction (145 ) is indicated with an arrow.
The number of pressure taps was limited by the size of the
evacuation bridge situated in the back of the building, through
which all the 1 mm diameter tubes had to pass in order to reach
three scanning valves situated under the wind tunnel. The low
frequency response of the tubing only allowed the measurement of
mean pressures for the characterization the mean wind loading.
Uematsu and Isyumov [26] state that, when the characteristic
dimension of the building is small compared with the turbulence
integral scale of the incoming winds, the maximum load effects can
be evaluated using a quasi-steady approach. This is the standard
practice of many codes, like the Eurocode [27] or the Japanese AIJ
[32] recommendations, which deduce the design load from the
mean pressure field. It is also convenient when the turbulence field

8
is unknown, as in the present case. The quasi-steady method
assumes that the pressure fluctuations on the building are fully
correlated with the velocity fluctuations, i.e. the maximum pressure
will appear at the maximum velocity peak. Then, the design
pressure is calculated by multiplying the mean pressure coefficient
with the peak dynamic pressure calculated with the extreme
(design) wind velocity. This approach fails when the building
generated turbulence, at separation regions, contributes significantly
to the pressure fluctuations. This building generated turbulence
has smaller scales that can be important for small elements
on the skin of the building. As a result, the quasi-steady approach
can be safely used to calculate design loads for the main structure
or for large elements of the building but not for localized small
elements.
Tieleman et al. [33] conclude, in their review of the wind tunnel
requirements for wind loading assessment of low-rise buildings,
that the incoming wind profile does not influence the mean pressure
coefficient. It is more important to match the incoming horizontal
turbulence intensities if one is interested in reproducing well
the fluctuations of the pressure coefficients. The independence of
the mean pressure coefficient with the incoming turbulence was
also demonstrated in Ref. [28]. Hence, the uncertainty associated
with the incoming boundary layer conditions might not be that
important for the modeling of the mean pressure field. Nevertheless
two limiting classes of neutral ABL were tested in order to
verify this hypothesis.
The mean pressure coefficient is defined according to the
incoming dynamic pressure at freestream level:
Cp 10 ¼
Dp
1
2 rU2 AWS
where Dp is the measured mean pressure relative to a common
undisturbed reference outside the wind tunnel, r is the air density
and UAWS is the reference wind speed velocity measured at the
AWS position. Notice that the AWS is located at a lower elevation
than the building of around 10 m. The subscript 10 is used in
accordance with the Eurocode [27] to stand for global pressure
coefficient, to be used for structural design, or local pressure
coefficient for surfaces greater or equal to 10 m 2 . The AWS position
is used as a reference for the wind speed because it can be related
later with the field measurements to obtain the design dynamic
pressure:
1
Dp50 ¼ Cp10 2 rU2 AWS;50
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
Fig. 5. Reference (left) and final (right) building models instrumented with 133 and 159 pressure taps respectively.
(6)
(7)
where UAWS,50 is the design velocity estimated using extreme value
analysis at the AWS position. The mean yearly air density measured
at the site is 1.13 kg m 3 .
The overall forces acting on the building are obtained by integration
of the mean pressure coefficients on the building facades.
The force coefficient along i axis, CFi, isdefined as follows:
Fi CFi ¼
1
2 rU2 AWSA (8)
i
where Ai is the frontal area perpendicular to the incoming wind
direction along the i axis.
3. Conceptual design
The conceptual design of the base started from the selection of
a number of block geometries, with the least detail level, offering
the best perspective for a systematic comparison of the design
prototypes. Six different block concepts were selected for testing
during the conceptual design phase, all of them with the same
living area. The differences between them were on the shape of
the base (square or rectangular), the number of storeys (one or
two) and whether the building was integrated on the ridge or
elevated. Several wind incidences and ridge positions were
tested.
3.1. Selection of reference building-block concept
Fig. 6 shows three models, made of transparent plexiglas to
look through, that provided similar performance from the
snowdrift point of view (two building orientations are tested).
Two other building concepts integrated on the ridge presented
snow accumulation areas next to the building, indicating problems
of accessibility [3]. It is clearly noticed the impact of the
building orientation on the winds at surface level. When the
rectangular two-storey building is placed perpendicular to the
incoming wind it presents a much larger obstruction to the flow
and generates higher speed-ups (erosion areas) and also larger
areas for snow accumulation in the from and in the back of the
building. When the frontal area is minimized by aligning the
building with the wind direction, the snow deposition area is very
much reduced, as indicated by Kim et al. [8] in flat terrain
conditions.
The square-based buildings present similar snow accumulation
areas. The orientation at 45 seems to present fewer problems for
snow build-up in front of the building.

The rectangular model presents the smallest snowdrift area of
the three models, when it is aligned with the incoming wind. In
the other hand, it is also more sensitive to the wind direction
variability. Therefore it was concluded from these tests that the
three concepts had similar performance with respect to snow.
Hence, the selection of the building typology to follow the design
process could be left to other design criteria. Because of an easier
construction and a better compatibility with the internal layout
the elevated square-based and one-storey building model was
selected.
A 20 20 5 m 3 square-based model, selected in the
conceptual phase, was modified with 2 m depth chamfered at the
four vertical edges and 1 m depth chamfer at the top and bottom
edges, to offer much better performance in reducing snowdrifts, as
it was observed in Ref. [11]. This model shall constitute the reference
for further optimizations of the design.
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 9
Fig. 6. Amplification factor contours for models oriented parallel (left) and at an incidence angle (right) to the incoming wind direction from SE þ 10 .
3.2. Building-ridge integration
After Belare-2005 expedition to Antarctica, IPF selected the area
with the best terrain conditions for the anchoring of the building.
This area is situated some 40 m north of the previously selected
area for the erosion tests presented in Fig. 6. The positioning of the
building on the ridge is constrained by the anchoring conditions on
the rock, which are more difficult at the lee side of the ridge due to
an almost vertical slope. Therefore, the building can be better
positioned with the back side on top of the ridge and some cantilever
upstream.
Fig. 7 presents a sensitivity analysis of the amplification factor
versus the integration of the reference building, considering the
height of the clearance and the position across the ridge. In the topleft
figure, the building is at a reference position X0 and is elevated
1 m from the top of the ridge. When the building is lifted to 2 m at

10
Fig. 7. Amplification factor contours for the reference building model. Sensitivity to elevation height and across-ridge positioning.
the same position (top-right figure), the deposition area is reduced.
In the contrary, if the building is moved backwards by just 2.5 m
(bottom-left figure), the snow deposition area is significantly
enhanced. This situation is partly alleviated if the elevation is
increased to 2 m (bottom-right figure).
The positioning of the building across the ridge appears to be an
important aspect for controlling snow accumulation and erosion. A
building with larger windward cantilever offers more speed-up
underneath producing more aggressive erosion behind the
building. When the building is shifted backwards, the flow finds
more resistance to go under it and deviates more to the sides
producing less erosion behind the building. In both situations
elevating the building improves the erosion. It will be explained
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
later how the positioning of the building also affects the wind
loading as both aspects are controlled by the drag of the buildingridge
configuration.
The building at 45 incidence has a “delta wing” configuration
generating more vorticity and hence more erosion than the
building aligned with the incoming wind direction [3]. This was
also concluded by Beyers [31] from numerical CFD simulations of
the snowdrifts around SANAE IV station.
The wind loading on the building will be also influenced by its
positioning on the ridge. To study the building-ridge integration
from the wind loading point of view, a model of the reference
building model was manufactured with 133 pressure taps to map
the pressure acting on the building facades.
Fig. 8. Mean pressure coefficient at reference position, with 1 m clearance and wind incidences 0 (left) and 45 (right). The wind direction is 101 parallel to the X axis.

Fig. 8 shows the mean pressure coefficient distribution
throughout the building facades, obtained by linear interpolation
in the top and bottom faces and by nearest neighbor interpolation
in the rest, for two wind incidences: 0 and 45 . The positioning of
the taps is also indicated. The typical pressure build-up in the
stagnation area of the frontal face is depicted as well as the high
suction in the separation region at the leading edge of the top
surface. The latter is much higher for the building oriented at 45
wind incidence due to the formation of “delta wing” vortices,
similar to the situation of ground-based buildings with flat roofs
[28].
The pressure map at the top surface does not vary much with
the positioning of the building. The high suction generated at the
leading edge is increased at higher building elevations. This is due
to the higher surface velocities that create larger suction, as it is the
case for ground-based buildings [29]. A slight decrease in the top
surface suction is observed when the building is positioned backwards,
probably due to lower local velocities. Large suction areas
occur very close to the roof edges in the upwind corners and is
typically alleviated using parapets [30]. Unfortunately this remedy
will also create snow accumulation on the roof and thus it is not
a good solution in this case. Due to the sloping terrain, the wind
approaches the building with a small up-flow angle. Hence, a small
negative inclination of the roof would decrease the relative up-flow
angle, reducing the flow separation area and decreasing the overall
lift on the rooftop.
The most interesting results happen in the bottom surface,
where the flow interacts with the building and the ridge. Here, it
can be noticed a positive pressure at the leading edge due to the upflow
generated by the sloping terrain. Then, the pressure turns to
negative, due to the speed-up generated by a convergent clearance,
as the flow approaches the ridge top. At the ridge top the speed-up
and the suction reaches a maximum value which is then decreased
by a divergent clearance at the rear of the building.
Fig. 9 shows the effect of the clearance height on the pressure
distribution on the bottom surface. The left column shows the
sensitivity of the pressure distribution to the elevation height,
when this one is set to 0, 1 and 2 m. The right column shows the
sensitivity to the across-ridge positioning of the building, when this
one is placed 3 m backwards, at the reference position (X0) and 3 m
forward.
When the building sits in top of the ridge (Fig. 9, top-left) a high
blockage is created on the flow under the building, which generates
high pressure build-up and a net positive lift force on the building.
The lift force is two times higher than the one obtained with 1 m
clearance. Using a clearance of 2 m (Fig. 9, bottom-left) instead of
1 m further decreases the lift force. Even though the lift force is
counteracted with the self weight of the building, a positive aerodynamic
lift will lead to vibrations on the structure, so it should be
avoided. Owing to the ridge slope, the elevated building concept is
not only a good solution to decrease snow deposition but also
a more convenient configuration from the aerodynamic point of
view.
The sensitivity of the positioning of the building across the ridge
is also quite remarkable. The highest suction on the bottom face is
always generated at the ridge top, where the flow accelerates due to
the contraction of the clearance. The larger the area situated at the
ridge top the higher the suction on the bottom surface and,
therefore, the lower the overall lift force on the building.
4. Building envelope design
Once the reference concept has been developed, the building
envelope evolves with the integration of the interior layout and the
technical systems. At this stage of the process, the internal layout
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 11
becomes the main design driver of both the internal and external
geometry of the base.
The aim for a modular base, which would allow easier
manufacturing and transportation, lead to an octagonal building.
From the aerodynamic point of view, this geometry could be
considered as an intermediate case between the square building at
0 and 45 wind incidence. The integration of solar panels also
required an inclination of the side walls of 30 for a more efficient
energy capture. The accommodation of some units of the system in
the center of the building required some extra space, which was
found by adding a second storey that sticks out of the roof. This
add-on would also be used to give access to the roof and to
accommodate some more solar panels.
The new building envelope was tested using sand erosion tests.
Fig. 10 shows the amplification factor erosion maps obtained for the
octagonal building with the same configurations of Fig. 7 for the
reference building. At first glance, it is clear that the new geometry
is much more effective at removing snow. Indeed, the building in its
back position and with an elevation of 2 m removes almost entirely
all the deposition area situated behind the building, leaving only
the two deposition tails delimiting the building wake.
The erosion strength is better visualized with dune tests (Fig.11).
The visualizations reveal the formation of two dunes delimited at
both sides by erosion streams originated by the interaction of the
vortex shedding from the edges of the building with the wake of the
building. The two most external erosion streams are generated by
the corner vortices generated at the front of the building. The third
erosion stream, in between the two dunes, is attributed to the vortex
shedding from the trailing edge of the building, where the flow
passing underneath meets the low pressure in the wake. These two
vortex streams are very intense at the exit of the building producing
very intense wind shear that removes the sand on top of the garage
roof very early in the erosion tests. The intensity of these vortices
depends again on the speed-up generated under the building and
hence on the building-ridge positioning.
The dune formation tests are consistent with the surface erosion
test: the erosion strength is more important when the building is
elevated or is moved forward. Even with the very erodible sand bed
it seems that the garage integrity is preserved because the erosion
does not expose the garage too much in the reference building. In
the other hand, the octagonal building presents much deeper
snowdrifts. In effect, by reducing the surface in the back of the
octagonal building the suction created at the ridge top is acting on
a smaller area and the overall lift force is increased. This V shape at
the exit of the clearance also enhances the vorticity which is the
responsible for the strong erosion streams just behind the building.
To cope with potential erosion problems and lift forces, the back
of the building was modified back to its original extension, still
leaving the inclined walls for the solar panels. Dune formation tests
on this new building envelope are shown in Fig. 11 (right). It is
remarkable the important effect of the geometry at the back of the
building on the erosion patterns. The larger area on top of the ridge
attenuates the trailing edge vorticity decreasing the erosion very
significantly.
The new building geometry was considered definitive and no
mayor changes were done on it. As with the reference building,
wind loading tests were also conducted to verify the sensitivity of
the final geometry on the building positioning.
5. Design optimization and final assessment
A model of the final building geometry, obtained at the end of
the envelope design phase, was instrumented with 159 pressure
taps in order to assess the final wind loading specifications (Fig. 5).
This time, the positioning of the pressure taps was determined from

12
Fig. 9. Mean pressure coefficient in the bottom face of the reference building. Sensitivity to elevation height and across-ridge positioning. The axes grid size is 10 m. Wind direction
is 101 aligned with the building, from right to left.
CFD simulations of the pressure distribution on the skin of the
building. By using CFD, the uncertainty on the global forces due to
tap resolution was reduced from 30%, in the reference building
when CFD simulations were not available, to 3% in the final building
model. The positioning of the taps and the associated uncertainty
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
was estimated by comparing the integrated CFD forces obtained
from all the grid cells of the skin and based only on the cells of the
tap positions. The garage was also included in the final wind
loading tests. The height between the garage rooftop and the ridge
was another parameter to be tested.

J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 13
Fig. 10. Amplification factor contours for octagonal building with inclined walls. Sensitivity to elevation height and across-ridge positioning.
Fig. 11. Dune formation tests for the reference building (left), octagonal building (center) and final building (right) at X ¼ X0 2.5 m and 2 m pillars height. The wind direction is
101 , parallel to the building.

14
Fig. 12 shows the pressure field on the final building geometry,
this time using linear interpolation in all faces because of the better
repartition of the pressure taps. The final geometry offers a significant
reduction of drag force due to a more aerodynamic shape in
the front.
Again, building-ridge integration was studied for the final
geometry. Fig. 13 shows the effect of the elevation of the building on
drag and lift coefficients for different across-ridge positions in the
reference building and the final building, which is always situated
in the backwards position (X0e3 m).
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
Fig. 12. Mean pressure coefficient on final building model from pressure taps.
The sensitivity to the incoming ABL is also studied, with
measurements using the smooth (snow type) and rough (urban
type) test section. The effect on the lift and drag forces are indeed
not significant in agreement with similar studies [28,33]. However,
the final geometry offers 40% reduction on drag with respect to the
reference building, due to the aerodynamic frontal shape. The drag
decreases a further 20e30% as the building moves backwards due
to the effect of a smaller building-ridge wake.
The high positive lift force is also significantly decreased when
the reference building is shifted backwards due to the larger
Fig. 13. Lift (left) and drag (right) coefficients for the reference and final building models. Effect of clearance height, across-ridge positioning and incoming ABL.

suction on the bottom surface. For the same reason, when the
clearance height is increased, the lift force decreases.
Therefore, the elevated building is not only a good solution to
cope with snow accumulation but also a more convenient configuration,
from the structural point of view, in this particular situation
in which the building is installed in sloping terrain. Above 2 m
the decrease in lift force is not very significant, so it seems to be
a good choice for the pillars height, also providing comfortable
accessibility under the building. The final building geometry offers
higher lift coefficient due to a lower suction under the building.
Besides, the inclined faces on the sides, most of them under suction,
also increase the lift.
Having the back of the building partly on top of the garage roof
makes the positioning of the latter another design parameter to
take into account. Fig. 14 shows the influence of the step height
between the garage roof and the ridge top on the aerodynamic
coefficients.
An optimum is noticed around 1 m, with 60% lower lift and 20%
lower drag than the position with the garage roof leveled at the
ridge top. With a fixed step of 1 m, tilting the garage roof down by
a small angle of 3.5 results in a reduction of 20% in lift and drag. In
fact, the garage is acting as a diffuser, enhancing the speed-up
under the building. At 9.5 garage tilt the aerodynamic forces are
not reduced any further. This angle is close to the mean slope
observed in the snow surface at the lee side of the ridge, which was
the surface geometry present during the testing of the reference
building as the garage was not included.
Regarding wind direction variability (Fig. 15) it is remarkable
how low dependency is of the lift coefficient with wind direction in
the final building geometry. Surprisingly, the drag coefficient is the
lowest, not at 101 when the building has a parallel orientation to
the wind, but at 135 .At101 the ridge is perpendicular to the
incoming wind and the frontal area offered by the building-ridge
ensemble is the largest, producing a larger wake and therefore
higher drag. In the contrary, when the building is aligned with the
wind direction the highest speed-up is generated under the
building and the lowest lift is produced. The presence of the tower
increases the drag in the SE-S sector.
6. Evaluation of the design in the field
Unfortunately there are no field measurements that could be
used for validation of the wind tunnel design. Nevertheless it is
worth looking at some photos that were taken behind the Princess
Elisabeth station during the first years of operation of the station
that allow a visualization of the final capabilities of the building to
cope with snowdrifts.
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 15
Fig. 14. Influence of garage roof positioning. Final building geometry with 2 m clearance height at 101 wind direction.
Fig. 16 shows the situation of the snowdrifts behind the station
at the arrival of the BELARE 2008e09 season in November 2008.
The station was built in the previous season and was left inhabited
during the winter. Therefore, the situation shown in the photo
reflects the state of the snow surface after 7 months of untouched
snowdrift development. The situation is fairly good with a clean
garage surface and no apparent snowdrift build-up behind the
building. As expected, the erosion patterns produced by the
building are not as severe as showed by the volumetric sand
erosion tests. Overall, the snowdrift aspects seem under control.
Aiming for more energy independence, during the BELARE
2008e09 season the station was equipped with an array of solar
panels that covered the entire surface of the garage roof just
behind the building. This had important consequences in the
building aerodynamics as it is clearly noticed in the photos of
Fig. 17, taken one year later, at the beginning of the BELARE
2009e10.
The solar panels increase the overall drag of the building and
reduce the wind speed-up underneath. This reduces the efficiency
of the building aerodynamics to passively remove snowdrift. Also
snow accumulation dunes develop and bury some of the solar
panels increasing the maintenance works. Interestingly, the
asymmetric development of the two snowdrift dunes also
appeared in the wind tunnel tests, where larger accumulations
were observed in the southern side of the building (Fig. 11).
The snowdrifts might be different from year to year
depending on the interannual variability of the surface wind
Fig. 15. Lift and drag coefficients dependency on wind direction for the reference and
final building models.

16
Fig. 16. Situation of snowdrifts behind the princess Elisabeth station at the beginning of BELARE 2008e09 season. Courtesy of René Robert, Ó International Polar Foundation,
November 2008.
speed. Since AWS measurements are not available during the
operational phase of the station, ERA-Interim reanalysis data
has been used to compute the mean 10 m wind speed in the E-
SSE snowdrift sector over the wintering unmanned period
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
AprileSeptember as a proxy to snowdrift transport. The snow
transport can be estimated based on the empirical
function found during the STABLE2 experiment in Halley (Antarctica)
[34].
Fig. 17. Situation of snowdrifts behind the princess Elisabeth station at the beginning of BELARE 2009e10 season. Courtesy of René Robert, Ó International Polar Foundation,
Novemeber 2009.

Fig. 18. Mean velocity and snow transport anomalies with respect to the long term
period 1989e2010 considering the wintering unmanned season AprileSeptember and
the prevailing wind direction E-SSE.
logðQÞ ¼0:632U 3:69 (9)
where Q is the snow mass flux in g m 1 s 1 and U is the mean
velocity. The interannual variability of the mean velocity and snow
transport is provided in Fig. 18 in terms of anomalies, i.e. the ratio of
the seasonal mean to the long term seasonal mean 1989e2010.
It is observed that the snow transport was 14% lower in 2009
than in 2008 which means that the generation of snow build-ups
during the 2009 winter season can be attributed to the presence of
solar panels rather than to changing climatological conditions.
Snowdrift build-ups where also found at the beginning of
BELARE 2010e11 and 2011e12 seasons. Regardless of the large
interannual variability between 2009 and 2010 (58% increase) the
size of the snowdrifts was quite similar which means that the
cumulated volume reaches a quasi-steady state within the winter
season. Nevertheless, the main building always remains snow-free
ensuring accessibility. The maintenance works for removing the
snow around the solar panels and garage entrance take a few days
at the beginning of each season as reported in the station’s website.
7. Conclusions
The integrated design of an Antarctic building requires careful
consideration of the environmental conditions in which it will be
immersed. The conceptual design requires a rapid prototyping
process where the sand erosion technique proves useful to assess
snowdrift and wind conditions around different building concepts.
Dune formation tests help visualizing the erosion strength of the
building wake around the under-snow garage. Sand erosion and
wind loading tests showed the high sensitivity of the aerodynamics
of the building with the positioning on the ridge, all controlled by
the behavior of the flow under the building.
The envelope and optimization design phases were also supported
with wind tunnel testing and CFD modeling aiding the
decision making process until the assessment of the final design.
The aerodynamic shape of the final building results in 40% drag
reduction. The overall forces can be further lowered by using the
garage roof as a diffuser. Increasing the trailing edge area by
introducing a step between the ridge top and the garage rood
results in 60% lower lift and 20% lower drag. A small inclination of
the roof by 3.5 reduced further lift and drag by 20%.
After the first years of operation of the Princess Elisabeth base, it
can be concluded that the station can effectively cope with snowdrift,
producing no significant impact in the snow patterns behind
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18 17
the ridge as predicted by the wind tunnel sand erosion tests and
confirmed with measurements of the snow surface topography
before and after the construction of the building [25].
This study constitutes a good example of the advantages of
including wind engineering since the beginning of the integrated
building design process, giving the opportunity to consider the
most effective use of the environment while achieving important
structural and maintenance savings by carefully selecting the
optimum positioning and shape of the building.
References
[1] Belgian Science Policy and International Polar Foundation, 2007, Construction
and operation of the New Belgian Research Station, Dronning Maud Land,
Antarctica. Final Comprehensive Environmental Evaluation (CEE), May 2007.
[2] Gorlé C, Sanz Rodrigo J, van Beeck J. Design of Belgian polar base: CFD analysis
and validation, ICWE-12 proceedings, Cairns (Australia); July 2007.
[3] Sanz Rodrigo J. On Antarctic wind engineering, PhD thesis, Université Libre de
Bruxelles, Belgium; 2011.
[4] Delpech P, Palier P, Gandemer J. Snowdrifting simulation around Antarctic
buildings. J Wind Eng Ind Aerodyn 1998;74e76:567e76.
[5] Waechter BF, Williams CJ. Snowdrift design guidance for the New South Pole
Station. Proceedings of the 10th international conference on cold regions
engineering. New Hampshire (USA); August 1999.
[6] Beyers JHM, Harms TM. Outdoors modelling of snowdrift at SANAE IV
research station, Antarctica. J Wind Eng Ind Aerodyn 2003;91:551e69.
[7] Leitl B, Schatzmann M, Baur T, Koening-Langlo G. Physical modeling of snow
drift and wind pressure distribution at the proposed German Antarctic Station
Neumayer III. In: Proceedings of OMAE2006, Hamburg (Germany); June 2006.
[8] Kim DH, Kwok KCS, Smedley DJ, Rhode HF. Modelling of snowdrift around
prismatic buildings for Antarctic environment. Int J Offshore Polar Eng. ISOPE
1992;2(1):73e9.
[9] Kwok KCS, Kim DH, Smedley DJ, Rohde HF. Snowdrift around buildings for
Antarctic environment. J Wind Eng Ind Aerodyn 1992a;41e44:2797e808.
[10] Kwok KCS, Smedley DJ, Kim DH. Snowdrift Around Groups of Antarctic
Buildings, 11th Australasian fluid mechanics conference, University of
Tasmania, Hobart, Australia; 1992.
[11] Kwok KCS, Smedley DJ. Snowdrift around Antarctic buildings e effects of
corner geometry and wind incidence. Int J Offshore Polar Eng ISOPE 1993;13:
61e5.
[12] Kind RJ. Snow drifting. In: Gray DM, Male DH, editors. Handbook of snow,
principles, processes, management and use. Toronto: Pergamon Press; 1981.
[13] Kind RJ. A critical examination of the requirements for model simulation of
wind-induced erosion/deposition phenomena such as snowdrifting. Atmos
Environ 1976;10:219e27.
[14] Anno Y. Requirements for modeling of snow drift. Cold Regions Sci Technol
1984;8:241e52.
[15] Haehnel, Lever JH. Scaling snowdrift development rate, In: 52th Eastern snow
conference, Toronto, Ontario, Canada; 1995.
[16] Simiu E, Scanlan RH. Wind effects on structures: an introduction to wind
engineering. New York: John Wiley and Sons; 1978. p. 468.
[17] Beranek WJ, van Koten H. Visual techniques for the determination of wind
environment. J Wind Eng Ind Aerodyn 1979;4:295e306.
[18] Livesey F, Inculet D, Isyumov N, Daveport AG. A scour technique for the
evaluation of pedestrian winds. J Wind Eng Ind Aerodyn 1990;36:779e89.
[19] Dezsö G. On assessment of wind confort with sand erosion. Ph.D. thesis,
University of Eindhoven, The Netherlands; 2006.
[20] Viegas DX, Janerio Borges AR. An erosion technique for the measurement of
the shear stress field on a flat plate. J Phys E Sci Instrum 1986;19:625e30.
[21] van Beeck J, Dezs}o G, Planquart Ph. Microclimate assessment by sand erosion
and Irwin probes for atmospheric boundary layer wind tunnels. In: Conf. Proc.
on physical modelling of flow and dispersion phenomena, PHYSMOD2009,
24e26 Aug. 2009, von Karman Institute for Fluid Dynamics, Sint-Genesius-
Rode, Belgium, pp. I.3.1eI.3.8.
[22] Jaedicke C. Drifting snow and snow accumulation in complex arctic terrain.
Ph.D. thesis, Geographical Institute, University of Bergen; 2001.
[23] Li L, Pomeroy JW. Estimates of threshold wind speed for snow transport using
meteorological data. J Appl Meteorol 1997;36:205e13.
[24] Mann GW, Anderson PS, Mobbs SD. Profile measurements of blowing snow at
Halley, Antarctica. J Geophys Res 2000;105:24491e508.
[25] Pattyn F, Matsuoka K, Berte J. Glacio-meteorological conditions in the vicinity of
the Belgian princess Elisabeth station, Antarctica. Antarctic Sci 2010;22:79e85.
[26] Uematsu Y, Isyumov N. Wind pressures acting on low-rise buildings. J Wind
Eng Ind Aerodyn 1999;82:1e25.
[27] EN 1991 Eurocode 1 actions on structures, Part 1e4: General actions e wind
actions, 2008, European Standard.
[28] Bienkiewicz B, Sun Y. Local wind loading on the roof of a low-rise building.
J Wind Eng Ind Aerodyn 1992;45:11e24.
[29] Gerhardt HJ, Kramer C. Effect of building geometry on roof windloading.
J Wind Eng Ind Aerodyn 1992;41e44:1765e73.
[30] Kind RJ. Worst suction near edges of flat rooftops with parapets. J Wind Eng
Ind Aerodyn 1988;31:251e64.

18
[31] Beyers JHM. Numerical modelling of the snow flow characteristics
surrounding SANAE IV Research Station, Antarctica. Phd Thesis, South Africa;
2004.
[32] Tamura Y, Ohkuma T, Okada H, Kanda J. Wind loading standards and design
criteria in Japan. J Wind Eng Ind Aerodyn 1999;83:555e66.
J. Sanz Rodrigo et al. / Building and Environment 52 (2012) 1e18
[33] Tieleman HW, Hajj MR, Reinhold TA. Wind tunnel simulation requirements to
assess wind loads on low-rise buildings. J Wind Eng Ind Aerodyn 1998;74e76:
675e85.
[34] Dover AJ. Numerical modeling of blowing snow. Ph.D. thesis, Dept. of Applied
Maths, University of Leeds, Leeds, UK; 1993.