Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

PROCEEDING

BOOK

“IArcSAS”
1st INTERNATIONAL
ARCHITECTURAL SCIENCES
AND APPLICATION
SYMPOSIUM

27-29 OCTOBER, 2021


ISPARTA / TURKEY

EDITOR
PROF. DR. ATİLA GÜL
PROF. DR. ÖNER DEMİREL
ASSOC. PROF. DR. SEYİTHAN SEYDOŞOĞLU
“IArcSAS”
1st International Architectural Sciences
and Application Symposium

October 27-29, 2021


Isparta, Turkey

CONFERENCE PROCEEDINGS BOOK

EDITOR
Prof. Dr. Atila GUL
Prof. Dr. Oner DEMIREL
Assoc. Prof. Dr. Seyithan SEYDOSOGLU

All rights of this book belongs to ISPEC Publishing House.


Without permission can’t be duplicate or copied.
Authors of chapters are responsible both
ethically and juridically.

ISSUED: 30/11/2021
ISBN: 978-625-7464-51-2
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

THE ANALYSIS OF THE WIND DESIGN’S PERFORMANCE


IN THE CHURCH OF SAINT FRANCIS OF ASSISI BY OSCAR NIEMEYER

Guilherme Silva TEIXEIRA,


Federal University of Mato Grosso, Institute of Exact and Earth Sciences, Brazil
ORCID No: 0000-0001-8225-4204

Marco Donisete de CAMPOS.


Federal University of Mato Grosso, Institute of Exact and Earth Sciences, Brazil
ORCID No: 0000-0003-4365-0129

ABSTRACT
Often, in architectural creation, the design approach neglects the action of the wind in the structure as a
result of the norms and codes that only address traditional and generic geometries. Thus, computational
studies are viable alternatives to analyze unique geometries, combining architectural innovations with
the study of dynamic actions due to wind. Here, was considered the wind action in an innovative project
constituted of parabolic and circumferential generatrices: the Church of Saint Francis of Assisi.
Designed by Brazilian architect Oscar Niemeyer in Belo Horizonte, Brazil, two paraboloid vaults and
three circular arches of reinforced concrete composed its structure. This work generated great
international recognition for the architect after 1943, as the design of the roofs did not require walls. For
geometry modeling, were adopted the Autodesk AutoCAD software and the fluid domain for low-rise
buildings. The simulations took place with the CFX solver of the Ansys Workbench software, and the
RNG K-Epsilon turbulence model was employed. The input velocity was estimated using the power-law
approximation and, in all simulations, tetrahedral meshes. To validate the methodology were considered
four hyperbolic paraboloid roof models. The pressure coefficients were analyzed and, for the flow
visualization, highlighting the detachment points and the recirculation zones.

Keywords: Wind loads, Parabolic roof, Ansys, Pressure coefficients, Oscar Niemeyer.

1054
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

1. INTRODUCTION

The Paraboloid Roofs


Hyperbolic parabolic shell, widely applied in architecture, especially in large-span buildings whose roof
and wall are of one whole object, and the axis of the arch is a parabola. And, this kind of building made
of particular materials like reinforced concrete will be very stable and durable (Shen et al., 2020). Many
architects have made use of large-span constructions as, for example, in the restaurant of the "City of
Arts and Sciences" in Valencia, Spain, and the "Los Manantiales Restaurant" in Xochimilco, Mexico,
designed by Felix Candela. Also, Coetzee Steyn in the "Bosjes Chapel" in South Africa. In Brazil, the
most representative example of this is the Saint Francis of Assisi Church by Oscar Niemeyer.

The Church of Saint Francis of Assisi


Brazilian architect Oscar Niemeyer (1907-2012) is one of the architectural masters of the twentieth
century, and a pioneer of Modernism admired for his highly inventive, organic forms. Oscar Niemeyer's
preoccupation with the spatial integration with the surroundings led him to create buildings cut through
transversally by transparency in his "classical" architecture (1940's to 1960s). The Pampulha project in
Belo Horizonte (1940) is a turning point both in (Brazilian and world) modern architecture and in his
work ("my architecture begins at Pampulha"). The curving lakeside defines the location of the buildings:
they are on virtual peninsulas advancing on the lake surface, thus thus maximizing the surrounding
landscape views. The main façade of the Saint Francis of Assisi Church, the focus of this study, is
entirely transparent (partly offset by brises soleil) and faces straight onto the lake, the waterscape being
one with the interior of the nave. The abutment by the main façade is higher, dropping down towards
the high altar. The dome is interrupted in the nave, and the high altar stands under another slightly
higher dome out of synch with the former at that point, allowing light to radiate on the high altar. The
transparency of the principal west façade and the greater height of the nave at the entrance to the church
give nature (the view of the lake) precedence over transcendence (the lower space of the high altar and
the Saint Francis panel behind it) (Holanda, 2009).

Numerical determination of wind loads for buildings with paraboloid roofs


In the literature, few academic articles approached the hyperbolic paraboloid shells simulation, in
particular, the wind effects. The first phase of the investigation reported in Teixeira and Campos (2021)
was focused in numerical tests performed in Ansys Workbench software on in-scale models, to analyse
the wind load on buildings describing the external pressure coefficients for different angles (0°, 45°, 90º,
180º, and 270°) of wind incidence on significant regions of the paraboloid roof. In this work, the pressure
coefficient with numerical tests on different geometries of hyperbolic paraboloid roofs were investigated
considering four significant wind angles of attack (i.e., 0º, 135º, 225º, and 315º) as well architetural
elements influence (the bell tower and marquee) in the fluid flow allow the study.

2. METHODOLOGY
In this work, were done meshes and post-processing with the Ansys Workbench software, and the
simulations took place with the CFX solver. For geometry modeling, was used Autodesk AutoCAD
software with dimensions proposed by Macedo (2008), as shown in Fig. 1a. According to Franke et al.
(2007) was used a control volume for low-rise buildings (i.e., H~B~L), whose adopted dimensions were:
length of 5H+L+15H, the width of 5H+B+5H and height of H+5H, dependent on the building height of
(H = 9.16 m), of the length L of the building in the flow direction and width B (Fig 1c). For local refining,
a body of influence was used (Fig 1b). According to Such (2018), was considered L for the length of the
downstream region and half of this length upstream. Also, twice this dimension for the width and the
height.

1055
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(a)

(b) (c)

Figure 1. (a) Dimensions of the Church of Saint Francis of Assisi, (b) geometry and body of influence, and
(c) geometry and control volume.

The wind velocity at different heights was estimated using the power-law approximation given by:
𝛼
𝑈𝑧 𝑍
=( )
𝑈𝑟𝑒𝑓 𝑍𝑟𝑒𝑓
being Uz is the wind speed (in meters per second) at height Z (in meters), and Uref is the known wind
speed at a reference height Zref, here adopted as 10 m. The exponent α is an empirically derived
coefficient that varies dependent on the terrain roughness and the time interval. An unstructured mesh
with curvature and proximity capture, composed of tetrahedrons, was applied (Figure 2a-b), which the
first level defines the dimensions of the elements in the fluid domain. The second level describes a local
refinement (with the body of influence), and the third, the refinement of the geometry faces that intersect
the flow.

(a) (b)
Figure 2. (a) Top view (model 5, Application 2) and (b) Elevation of longitudinal section (model 5,

1056
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

Application 2)

Discretization is used to solve governing equations of problems involving Ccomputational Fluid


Dynamics (CFD) to transform complex equations into algebraic equations. Faced with this, Franke et
al. (2004) recommend high-order schemes for discretizing advective terms. In this work, the advection
scheme and turbulence numeric was high resolution. The convergence criterion established was the
residual RMS equal to 10-4, which, although considered low-converged, can be sufficient for many
engineering applications (ANSYS, Inc, 2009) with 400 and 500 for the minimum and the maximum
number of iterations, respectively. The Double Precision scheme (16 digits of numerical precision) was
adopted to improve convergence, as the simulations involve a notably free surface (ANSYS, Inc, 2009).
Table 1 shows the rest boundary conditions adopted.

Table 1. Boundary conditions and non-dimensional parameters.

Condition Parameters
Method of mesh Tetrahedron
Capture curvature and proximity On
Reference pressure 101325 [Pa]
Air temperature 25º [C]
Turbulence intensity Medium (5%)
Flow regime Subsonic
Inlet U/Uref = (Z/Zref)α
Relative pressure of outlet 0 [Pa]
Wall Rough wall
Model wall roughness Smooth wall
Roughness 0.0025 [m] (Application 1)
0.01 [m] (Applications 2 and 3)
Zref 0.1 [m] (Application 1)
10 [m] (Applications 2 and 3)
Uref 16.7 [m/s] (Application 1)
30 [m/s] (Applications 2 and 3)
Advection scheme High resolution
Turbulence numeric High resolution

In this work, were adopted the following convergence criteria. The first concerns the residual RMS (also
used as stopping criterion) of energy, mass, momentum, and additional turbulence equations due to the
adopted RNG K-Epsilon model (Fig. 3a-b). The second was mass conservation monitoring, given by the
IMBALANCE monitor (mass difference in inlet and outlet), which showed consistent values (<1%),
according to ANSYS (2009) (Fig 3c).

1057
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(a) (b)

(c)
Figure 3. (a) Residual mass and momentum RMS monitor (model 2, Application 1), (b) Residual turbulence RMS monitor
(model 2, Application 1) and (c) IMBALANCE monitor (model 2, Application 1)

For comparison of applications, the local pressure coefficients were calculated, defined by Cpe=Δp/q
where Cpe is the external pressure coefficient; Δp is the difference in external pressure coefficient, and
q is the dynamic pressure. The analysis of pressure coefficients defines the third criterion for judging
convergence: physically coherent results.High variations in Cpe values can indicate convergence
problems, as there is a coherent range for this parameter. For positive values, which indicate
overpressures, is expected a maximum of Cpe=1.0 (disregarding errors associated with CFD), while for
negative values, which indicate suction, in defined regions of geometry, the magnitude can be from 6 to
8 times the pressure obstruction (Manfrim, 2006).

3. NUMERICAL APPLICATIONS

Application 1: Validation Methodology


For the validation methodology, hyperbolic paraboloid roof and square footprint presented by Rizzo and
Sepe (2015) were adopted, according to Fig. 4(a) and Table 2. Here, were accounted three situations:
the first with the wind falling at 0º, orthogonally the face whose elevation presents a parabola of positive
concavity, the second with the wind falling obliquely at 45º, reaching the corner of the model, and the

1058
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

third with the wind falling at 90º, orthogonally the face whose elevation presents a parabola of negative
concavity (Fig. 4b).

(a) (b)
Figure 4. (a) Geometrical parameters, and (b) different angles of incidence.

Table 2. Dimensions of geometric configurations.

Geometric H h f1 f2 L
configuration [m] [m] [m] [m] [m]
1 0.2133 0.1333 0.0267 0.0533 0.8000
2 0.2666 0.1333 0.4440 0.0890 0.8000

For the choice of mesh, analyzed aspects such as the element quality, the skewness, and the orthogonal
quality (Santana et al., 2020). The element quality is the metric that accounts for a relationship between
element area and border length (recommended values close to 1). The skewness indicates how close to
the ideal geometry (in this case, tetrahedron) the mesh cells or faces are (recommended values between
0 and 0.5). And, finally, the orthogonal quality metrics the element's orthogonality (recommended values
close to 1). Table 3 shows the models adopted with mesh dimensions and with their respective quality
metrics. The wind profile was defined using the power-law approximation for Zref = 0.1 m, α = 0.233,
and a wind speed of 16.7 m/s. As for the terrain, the roughness has adopted the value of 0.0025 m.

Table 3. Results of models 1, 2, 3, and 4, with their respective quality metrics.

Model 1 2 3 4
Geometric configuration 1 2 2 2
Wind direction 90º 0º 45º 90º
Element size in the fluid domain (m) 0.08 0.1 0.1 0.1
Element size in the body of influence (m) 0.04 0.05 0.05 0.05
Element size on geometry faces (m) 0.005 0.01 0.01 0.02
Nodes 402417 177443 242653 126447
Elements 2207163 984627 1363700 712401
Element quality (average) 0.8426 0.84623 0.84851 0.85038
Skewness (average) 0.22013 0.21448 0.21095 0.20805
Orthogonal quality (average) 0.77867 0.78435 0.78792 0.79083

Figure 5 shows the top view of the Cpe contours for all models, in which the color hue represents the
pressures acting on the surfaces corresponding to the ranges of the pressure coefficients: warm colors

1059
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

represent overpressures, while cold colors represent suctions. All models showed agreement in the
distribution of isobaric lines. In model 1 (Fig. Qa-b), the windward roof region presented a suction zone
with Cpe ranging from -1.1 to -0.7 (approximately), in agreement with Rizzo and Sepe (2015), which,
for the same region, range from -1.2 to -0.6. Also, for the middle of the coverage, the -0.2 magnitude
contour coincided. In model 2 (Fig. Qc-d), the suction peak, Cpemax= -0.9, in the central region of the
roof, coinciding with Rizzo and Sepe (2015). In both cases, the smallest suction region occurred in a
small windward (Cpe = -0.1). The suction zone in model 3 (Fig. Qe-f) occurred near the midpoint of the
left edge (top view), with a suction peak, near the contour, of -2.3, which is greater than the value -1.82
presented by Rizzo and Sepe (2015), the maximum difference among the cases analyzed in this
application. Finally, model 4 (Fig. Qg-h) presented an absolute difference in the coefficients of
approximately 0.3 (with a range of -0.06 to -1.18 (Rizzo and Sepe, 2015), and 0.3 to-1.3 (present work)),
the second maximum difference in the validation. Thus, the validation methodology showed good
agreement for most samples and showed a maximum variation of magnitude 0.5 in absolute values for
the external pressure coefficients (model 3).

(a) Model 1 (Rizzo and Sepe, 2015) (b) Model 1 (Present work)

(c) Model 2 (Rizzo and Sepe, 2015) (d) Model 2 (Present work)

1060
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(e) Model 3 (Rizzo and Sepe, 2015) (f) Model 3 (Present work)

(g) Model 4 (Rizzo and Sepe, 2015) (h) Model 4 (Present work)

Figure 5.Cpe contours for (a),(b) model 1, (c),(d) model 2, (e),(f) model 3, and (g),(h) model 4.

Application 2: Different wind incidence angles (0º, 135º, 225º, and 315º)
In this application, we studied the different angles of wind incidence on significant regions of the
paraboloid roof, namely: 0º, 135º, 225º, and 315º (Fig. 6). The wind speed of 30 m/s, according to Vallis
(2019), was adopted. In model 5, the wind reaches orthogonally on the main face of the main dome (A
in Fig. 1a), made with glass and brise soleil, where the fluid intercepts the marquee (F in Fig. 1a) before
reaching the edification. Model 6 analyzed the distribution of Cpe with the wind blowing diagonally on
the sideward arch (I) and secondary dome (C and B in Fig. 1a). Model 7, similarly, analyzes the effects
due to the action of the wind that diagonally hits the sideward arches (II) (D in Fig. 1a) and secondary
dome. Finally, model 8, whose wind hits the bell tower (E in Fig. 1a) and marquee before reaching the
building. The meshes were composed of 4.0 m tetrahedrons in the fluid domain, 2.0 m in the body of
influence, and 0.5 m in the building faces. Table 4 presents the mesh results for these models with their
respective quality metrics.

1061
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

Figure 6. Different angles of incidence of the wind in the Church of Saint Francis of Assisi.

Table 4. Results of models 5, 6, 7, and 8, with their respective quality metrics.

Model Directio Nodes Element Element Skewness Orthogon Reynolds Number


n s quality (average) al quality
(average) (average)
5 0º 118465 666741 0.82183 0.24691 0.75202 2.1491E+08
6 135º 140317 793734 0.82035 0.24854 0.75039 2.3631E+08
7 225º 144692 819670 0.81924 0.24989 0.74905 2.3957E+08
8 315º 138645 783688 0.82025 0.24876 0.75018 2.4006E+08
For this application, according to Blessmann (1995), α=0.25 was adopted considering an average 10 m
elevation of the top of the obstacles surrounding the building. For the terrain, defined as the rough wall,
with a roughness of 0.01 m was adopted. The roughness adopted represents the most critical situation
among the different materials that make up the terrain, such as concrete, asphalt, and grass. Figure 7
shows the wind power-law profile.

Figure 7. Wind profile using the power-law approximation (α=0.25)

Figure 8 presents the top view for the Cpe contours. Flows in which the wind partially falls (models 6,
7, and 8) on the parabolic cover (main dome shell, A in Fig. 1a) presented suction peaks in the concrete
shell structure region of inflection, the behavior observed in Teixeira and Campos (2021) for similar

1062
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

situations. These models show two distinct zones: windward, mainly under overpressure, and the other
to leeward under suction, with a characteristic related to the position of occurrence of the detachment
point. With a uniform distribution of coefficients of range from 0.00 to -0.66 across the entire coverage,
Model 5 was no great values for peaks. The most critical situation for suction happens in the main dome
shell (A in Fig 1a) of model 7 with Cpemin= -1.77, while the maximum overpressure of the cover is
between models 6 and 8 with equivalent Cpe values at the CFD level ( 0.41 and 0.39). Model 5
represents, simultaneously, the least critical situation for suction and overpressure, as the coverage
presented coefficients with negative values and magnitude lower than the other cases.
Figure 9 shows the Cpe contours for each model in the direction of their respective flows. In model 5,
the only one in which the fluid focuses orthogonally to a face orthogonal to the ground, presented an
increase in the values of the coefficients from the lowest level towards the highest point of the building,
with Cpemax = 0.69 on the face of the main dome (A in Fig. 1a). One of the causes of this effect is the
distribution of the wind speed profile modeled by the power law, which intensifies with increasing
height. On the other hand, models 6 and 7 that received the wind diagonally in the secondary dome-
sideward arches (I-II) set (B, C, and D in Fig. 1a) and model 8, whose wind obliquely hits the main
dome (A in Fig. 1a) presented an opposite behavior to that previously observed: Cpe's of greater
magnitude decrease towards the maximum height, where the suction peak occurred. This ambiguity
between the two situations is due to the fact that the orthogonality of the main face of the main dome
(model 5) causes considerable obstruction in fluid movement, unlike the slanted covers in models 6, 7,
and 8, which allow the fluid drain parallel to (or nearly so) the concrete surface.

(a) Model 5 (b) Model 6

1063
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(c) Model 7 (d) Model 8


Figure 8. Top view for the Cpe contour for (a) model 5, (b) model 6, (c) model 7, and (d) model 8.

(a) Model 5 (b) Model 6

(c) Model 7 (d) Model 8


Figure 9. External pressure coefficient isobaric lines for (a) model 5, (b) model 6, (c) model 7, and (d) model 8.

Figure 10 shows the streamlines that cross the building longitudinally. Similar to the distribution of
pressure coefficients, warm colors represent higher velocities, and cold colors represent lower velocities.
The fluid reaches the face of the main dome (A in Fig. 1a), and at about 60%~70% of the building height
is deflected to the soil when it meets the marquee (F in Fig. 1a), forming a small zone of windward
recirculation. In the opposite direction, the remaining accelerated fluid at the maximum height (H=9.16
m) occurred the detachment. Finally, one can notice the formation of a large recirculation zone in the

1064
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

paraboloid roof. This fact is in agreement with the negative coefficients presented in the previous
discussion.

Figure 10. Streamlines in the longitudinal plane.

In Figure 11, we have the velocity contour parallel to the ground that crosses the building longitudinally
at half height (XY-plane) and, also, the two contours orthogonal to the ground (ZY-plane) that intersect
the main and secondary domes (A and B in fig. 1a) on model 6. Between the two large zones of low
velocity downstream, we can notice a portion of the fluid undisturbed, which maintains the speed of 25
m/s. The highlights in Fig. 11 show that this behavior is related to the wind lifting point, which occurs
near the main dome maximum point. However, the fluid hits the secondary dome at a speed of 25 m/s
and accelerates at the maximum point of the parabola (reaching approximately 34 m/s) and does not
take off. Downstream reduces its speed, returning to that of approach.

Figure 11. Velocity contour (Model 6).

Model 7 presented an analogous behavior, despite not geometric symmetry between the two cases. The
fluid incident on the secondary dome (B in Fig 1a) accelerated and flowed parallel to the concrete
surface, returning to the approach velocity. On the other hand, the fluid incident on the main dome (A

1065
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

in Fig. 1a) presented a clear detachment point, due to its sharp deceleration (from 34 m/s for zero
velocity) (Fig. 12).

Figure 12. Velocity contour (Model 7).

Application 3: architetural elements influence (the bell tower and marquee)


Solid borders around buildings have a considerable influence on the flow velocity fields (Cóstula, 2006).
Architectural and constructive elements, such as sun visor devices and projection in the structure, are
relevant factors in the thermal comfort study and natural ventilation in buildings and their influence in
this type of analysis. However, they can be neglected during the pressure coefficients analysis, going
through simplifications. This application analyzed the bell tower and marquee influence (E and F in Fig.
1a) present in the church architecture. The bell tower has the shape of an inverted pyramid trunk and
contributes to supporting the marquee, which has a slight inclination. The elements are the only ones
that have straight lines in their design, creating contrast between the circular and parabolic curves of the
domes and sideward arches (Fig. 13). The 315º direction was chosen (Fig. 6) because, in this case, the
fluid intersects two architectural elements before the building. These elements were disregarded in the
simulation, and the results were compared with Model 8. The meshes were composed of 4.0 m
tetrahedrons in the fluid domain, 2.0 m in the body of influence, and 0.5 m in the building faces. For
this application, according to Blessmann (1995), α=0.25 (Fig. 7) was also adopted and the terrain was
defined as a rough wall with a roughness of 0.01 m. Table 5 presents the mesh results for these models
with their respective quality metrics.

1066
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(a) (Souza, 2012) (b) (Souza, 2012)


Figure 13. The architetural elements:the bell tower and the marquee

Table 5. Results of model 9 with their respective quality metrics.

Model Direction Nodes Elements Element Skewness Orthogonal Reynolds


Quality (average) Quality Number
(average) (average)

9 315º 115680 656071 0.81684 0.25258 0.74638 2.5513E+08

Model 9 presented a distribution of pressure coefficients in the coverage similar to model 8, being, for
the case of the complete model, the range from -0.03 to -1.68, equivalent to -0.07 to -1.77 (Fig. 14). The
distinction of the main dome's overpressure and suction regions (A in Fig. 1a) was conserved, and the
suction peak happens in the parabola inflection region for both cases.
According to Figure 15, on the main face of the main dome (A in Fig 1a), the overpressure peak occupies
equivalent portions for models 8 and 9, with values of Cpe=0.64 and Cpe=0.56, respectively.

1067
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(a) (b)
Figure 14. Top view for the Cpe contour for (a) model 8, and (b) model 9.

(a) (b)
Figure 15. External pressure coefficient isobaric lines for (a) model 8, and (b) model 9

The analysis of the interferences of the marquee and bell tower alone (E and F in Fig. 1a) in the fluid
flow allow the study of the slight difference in the results between these two simulations. Figure T
presents the velocity contours parallel to the terrain at different levels (2m, 3.5m, and 5m) intersecting
the architectural elements (model 8). On the first level (Fig. 16a), where the contour intersects the bell
tower below the marquee, a large part of the fluid, including that near the bell tower, was at low speeds
(approximately 14 m/s). This behavior is expected, given that for a height of 2m, the power-law will
return low values for the velocity component. However, a small amount of fluid accelerates past the
tower bell, reaching 25 m/s. On the other hand, downstream of the architectural element, the color scale
indicated values close to zero. The disturbed flow quickly returns to 14 m/s, returning to the initial
configuration, before reaching the main dome shell (A in Fig. 1a). This behavior repeated for the 3.5 m
and 5 m levels (Fig. 16c), with their respective speed variations. Thus, these architectural elements
caused a small local change in the flow. Their positions and geometric configurations allowed the fluid
to regain its initial characteristics before generating effects on the parabolic coverage.

1068
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

(a) (b)

(c)
Figure 16. Ground-parallel velocity contours for model 8, assuming a height of (a) 2 m, (b) 3.5 m, and (c) 5 m.

5. CONCLUSION

This paper used computational studies in the wind action analyses in an innovative project constituted
of parabolic and circumferential generatrices: the Church of Saint Francis of Assisi. Designed by
Brazilian architect Oscar Niemeyer in Belo Horizonte, Brazil, with two paraboloid vaults and three
reinforced-concrete circular arches composed its structure. The numerical tests performed with Ansys
Workbench software in the first phase of the investigation was focused on in-scale models to analyze
the wind load on buildings for different angles of wind incidence (0°, 45°, 90º, 180º, and 270°) on
significant regions of the paraboloid roof (Teixeira and Campos, 2021). In this work, was investigated
the pressure coefficient with numerical tests on different geometries of hyperbolic paraboloid roofs
considering four significant wind angles of attack (i.e., 0º, 135º, 225º, and 315º) as well architectural
elements influence (the bell tower and marquee) in the fluid flow.
For validation methodology, four models were considered involving hyperbolic paraboloid roof with
the square plan, according to Rizzo and Sepe (2015). The comparison of the distribution of isobaric lines
showed good agreement. The maximum difference, in absolute value, was 0.5 for suction in a 45º model.
However, the model still represents the most critical situation among all the cases analyzed by the

1069
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

authors and, for a significant number of contours, the values coincided. Flows in which the wind is
partially incident on the parabolic cover presented suction peaks in the shell concrete structure region of
inflection. The most critical situation for suction occurred in the shell of the main dome of the 225º
model, while the 135º and 315º models presented behavior considerably similar for the highest
overpressures. For the 0º direction occurred the displacement point in the section closest to the flow and
without high detachable peaks. The marquee and bell tower influence was application analyzed. The
results showed that the architectural components caused a low local disturbance in the flow, not enough
to generate high effects on the structure. The maximum difference in Cpe between the models (with or
not a marquee and a bell tower) was 0.1.
These results will represent a helpful guide to designing roofs with paraboloid shapes.

1070
“IArcSAS” 1st INTERNATIONAL
ARCHITECTURAL
SCIENCES and APPLICATION SYMPOSIUM
27-29 October 2021
ISPARTA, TURKEY

REFERENCES

ANSYS, Inc., (2009). ANSYS CFX-Solver Modeling Guide. Canonsburg.

Blessmann, J. (1995). O vento na Engenheria Estrutural. Ed. Universidade/UFRGS (in Portuguese).

Cóstola, D., (2006). Ventilação por Ação do Vento no Edifício: Procedimentos para Quantificação. (Doctoral dissertation,
São Paulo University, São Paulo, São Paulo, Brazil). Retrieved from
https://www.teses.usp.br/teses/disponiveis/16/16132/tde-26102010-144530/publico/dissertacao_final_low.pdf (in
Portuguese).

Franke, J., Hellsten, A., Schlünzen, H., & Carissimo, B. (2007). Best practice guide for the CFD simulation of flows in the
urban environment, COST Action 732: Quality assurance and improvement of microscale meteorological models.
COST Office.

Franke, J., Hirsch, C., Jensen, A. G., Krüs, H. W., Schatzmann, M., Westbury, P. S., Miles, S. D., Wisse, J.A., & Wright,
N.G., (2004). Recommendations on the use of CFD in predicting pedestrian wind environment , COST Action C14:
Impact of Wind and Storms on City Life and Built Environment. Hamburg: COST Office.

Holanda, F. (2009). Of glass and concrete internal versus external space relations in Oscar Niemeyer's architecture. In D.
Koch, L. Marcus & Jesper Steen(Eds.), Proceedings of the 7th International Space Syntax Symposium (Ref 043).

Macedo, D. M. (2008). Da matéria à invenção: As obras de Oscar Niemeyer em Minas Gerais, 1938-1955. Câmara dos
Deputados (in Portuguese).

Manfrim, S. T., (2006). Estudo Numérico para a determinação das pressões à ação do vento em edifícios industriais.
(Master's thesis, Faculdade de Engenharia - Câmpus de Ilha Solteira, Ilha Solteira, São Paulo, Brazil). Retrieved
from https://repositorio.unesp.br/bitstream/handle/11449/91483/manfrim_st_me_ilha.pdf (in Portuguese).

Rizzo, F., & Sepe, V. (2015). Static loads to simulate dynamic effects of wind on hyperbolic paraboloid roofs with square
plan. Journal of Wind Engineering and Industrial Aerodynamics, 137, 46–57.
https://doi.org/10.1016/j.jweia.2014.11.012

Santana, H. S., Silva, A. G. P., Lopes, M. G. M., Rodrigues, A. C., Taranto, O. P., & Silva Jr., J. L., (2020). Computational
methodology for the development of microdevices and microreactors with ANSYS CFX. MethodsX, 7, 100765.
https://doi.org/10.1016/j.mex.2019.12.006

Shen, X., Yang, Q., Li, L., Gao, Z., Wang, T. (2020), Numerical approximation of the dynamic Koiter’s model for the
hyperbolic parabolic shell. Applied Numerical Mathematics, 150, 194-205.
https://doi.org/10.1016/j.apnum.2019.10.003

Souza, M. H. (2012). Clássicos da Arquitetura: Igreja da Pampulha/Oscar Niemeyer. Retrieved from:


https://www.archdaily.com.br/br/01-83469/classicos-da-arquitetura-igreja-da-pampulha-slash-oscar-niemeyer

Such, M. R. (2018). Análise aerodinâmica de um veículo de eficiência energética. (Undergraduate thesis, Federal
University of Santa Catarina, Joinvile, Brazil). Retrieved from
https://repositorio.ufsc.br/handle/123456789/188252 (in Portuguese).

Teixeira, G. S., & Campos, M. D. (2021). Determination of wind loads for buildings with paraboloid roofs. In M. J. Iqbal
(Ed.), Proceedings of the 11th International Conference on Engineering & Natural Sciences (pp. 444-459).

Vallis, M. T. (2019). Brazilian Extreme Wind Climate. (Doctoral dissertation, Federal University of Rio Grande do Sul,
Porto Alegre, Brazil). Retrieved from https://www.lume.ufrgs.br/bitstream/handle/10183/198303/001099204.pdf

1071

You might also like