Вы находитесь на странице: 1из 8

1

Fluid Structure Interaction (FSI)


on a Tapered Tall Building
Andrea Sanzone
School of Mechanical and Industrial Engineering
Concordia University
Montréal, Québec
Email: andrea.sanzone@polymtl.ca
an sanzo@encs.concordia.ca

Abstract—A simple 3D finite element analysis is con- buildings. Anyway up to now most of the investigations
ducted in order to simulate the fluid structure interaction have been conducted only thorough experimental wind
on a tapered tall building. The analysis is made using tunnel testing and not through numerical simulations.
COMSOL Multyphisics . R In this simplified fluid structure
interaction it is assumed that the deflection of the solid does In this paper, to investigate the behaviour of a tall ta-
not significantly change the flow. The simulation uses the pered building, a simplify analysis using a finite element
Incompressible Navier-Stokes equation for the fluid and the model is used. This preliminary investigation is focused
linear elastic structural mechanics equations for the building. on the equilibrium solution and on the effects that some
fundamental parameters (i.e. velocity, density, viscosity)
or simply a change in the geometry could have on it.
November, 2010
I. Introduction and literature review

N OWDAYS the tall buildings with efficient struc-


tural system are made of high-strength materials
in order to reduce weight. This make them more slen-
II. Analysis
The flow over the building creates forces on its surface
due to pressure and shear stress. These forces load the
der and flexible, but unfortunately due to their lower structure and deflect it. The deformation depends on
damping values, they are more exposed to wind-induced the forces exercised by the fluid and thus depends on
excitations. These excitations could cause discomfort to the flow distribution. The flow field in turn depends
the building occupant and reduce the structural safety on the deformation of the structure. The problem is
of the building. said to be two-way coupled. In the real physics the
As mentioned in [1] a lot of solutions were tested in vibratory response of the building when the current
order to reduce such excitations like softened corners, velocity increased to the point that vortices were forming
varying cross section shape, spoilers, porosity or open- and shedding on alternating sides of the building could
ings, setbacks and of course tapering. be very destructive for the structure.
Tapering consists in changing the cross section with The civil engineering structure are typically bluff bodies
height. Cooper et al. [2], showed that the taper ef- immersed in turbulent flows; the aeroelastic phenomena
fect increased the vortex-shedding frequency over the related to this type of structures are usually divided in
height of the building and the local Strouhal number 4 categories:
increased with height and did not remain constant. • Static aeroelasticity:
Kim and You [3] showed that the tapering was more
− torisional divergenge.
efficient in reducing across-wind responses then along-
wind responses . Moreover they discovered that the • Dynamic aeroelasticity:
effect of the tapering is more effective in a suburban − vortex shedding and lock-in;
environment than in an urban environment. Kim, You − galloping;
and Ko [4] also showed with an aereoelastic model that − flutter.
with an adequate structural damping is possible to use In order to estimate the effects of the interaction between
the taper ratio in order to spreads the vortex-shedding aerodynamic forces and structural motions it is first fun-
over a broad range of frequencies so as to suppress the damental to solve the Navier-Stokes equations for tur-
vortex-induced excitations that facilitate the reduction of bulent flow with time-dependent boundary conditions
across-wind responses, but if the dumping is too small dependent on the solution itself. Due to the complexity
the tapering could have an adverse effect causing an of the problem and given the limitations of analytical and
increase of across-wind responses. These and others ar- numerical procedures the aeroelastic characterization of
ticles confirm the importance of the study of the tapered civil engineering structures relies largely on laboratory
2

testing and empirical modeling 1 . B. Numerical modelling


In this preliminary study we consider only the equilib-
rium solution for a 3D structure and we consider the two As in Kim and You’s experiment [4], we consider
physics linked in only way: the deformation depends a building having a square base which side is b =
on the fluid drag force and thus the flow distribution, 80m and height h = 320m. The side at the top is a
but the flow does not depend on the deformation. This = 60m, thus the taper ratio is a/b = 0.75. As showed
is done in order to simplify the problem, reducing the in the figure 1 the buildings is located in the ’test
problem size and the run time, otherwise a much denser chamber’ that is much bigger than the model itself:
mesh and an additional mesh in the flow subdomain that the dimensions are 520mx520mx1040m, this is done
moves with the bending of the structure tube should be in order to clearly see the flow pattern because the
added. There improvements could be object of a further subsonic fluid starts feel the presence of the building
analysis. before running into it and the fluid after the structure is
obviously deeply affected by the presence of an obstacle.
III. Modelling
A. Physical modelling
Hereinafter are reported some of the equations and
relations governing the fluid flow and the structure
considered as linear and elastic.

1) Fluid flow: The behaviour of a fluid flow is fully


described by the Navier-Stokes equations. For this
simple problem the fluid flow (newtonian) is described
by a less complicated version of the Navier-Stokes
equations, valid only for incompressible fluids (Mach
number < 0.3) [6]. The first equation is the so called
continuity equation and represents the conservation of
the mass through the control volume, while the second
is a vectorial equation which represent the conservation
of momentum in all the three dimensions (assuming Figure 1. Tapered model
constant viscosity):

∇ · u = 0, (1) Since the various fluid dynamics phenomena are re-


∂u lated to the Reynold number and thus to the velocity,
ρ + ρ(u · ∇)u = −∇ p + µ∇2 u + F, (2) it is extremely important characterized as well as one
∂t
can the wind velocity profile. In this study we consider
where:
the fluid going along the x direction (the velocity has
• ρ is the density [kg/m3 ]; only the U component), so the first face perpendicular
• u is the velocity [m/s]; to the x axis is the inlet (u = −U0 n). Here the velocity is
• p is the pressure [ Pa]; considered to have only the x component and it follow
• µ is the viscosity [ Pas] the wind profile power law [8]:
• F is the external force vector [ N ];
 z α
2) Elastic structure: considering a linear elastic struc- U (z) = U0 (zr ) , (5)
zr
ture the strain tensor written in terms of the displace-
ment gradient is [7]:
where:
1
e = (∇u + ∇uT ). (3)
2
The stress-strain constitutive relation for linear materials • U is the wind velocity along x direction;
commonly known as Hooke’s law is the following:
• z is the height;
σ = D : e, (4)
• zr is the reference height;
where σ is the Cauchy stress tensor and D is the fourth-
order stiffness tensor. • α is an empirically derived coefficient (= 0.3 for
1A good reference that gives more detailed explanations of these urban area with 15% of buildings higher than 15m).
phenomena and how to model some aeroelastic problems is [5]
3

such as in [4], that is ρs = 192 kg/m3 2 . The study


of the equilibrium solution for the tapered building
showed in figure 1 is done using a ’coarser’ mesh
with 248 elements 1594 degrees of freedom. As already
mentioned this is done in order reduce the time of
the simulation for this preliminary simulation. The
system is solved with the Stabilized Bi-Conjugate
Gradient Method [9] with Geometric Multigrid Method
as preconditioner with a relative tolerance of 1e − 5.
The Reynolds number (Re = ρuD/µ) for this simulation
is equal to Re = 1.2e8. It is calculated taking u = 25.1m/s
(the mean of the inlet wind profile) and D = 70m (the
mean of a side). The maximum cell Reynolds number is
7.59e8 .

Figure 2. Wind profile

On the outlet instead we impose a pressure p0 =


0Pa as for incompressible Navier-Stokes equation the
pressure arises as a Lagrange multiplier that serves to
satisfy the incompressibility constraint coming from the IV. Results and comparison
continuity equation. The side an the top of the chamber
are considered as open boundaries and so a no stress
condition is set (∂u/∂n = 0). The floor and the sides
of the building are considered as wall and so no-slip
condition is imposed (u = 0); for viscous fluids this
A. Tapered building
condition states that at a solid boundary, the fluid will
have zero velocity relative to the boundary.

For the wind profile we have chosen a U0 (zr ) equal


to 5 m/s and a zre f equal to 10 m.
As one can see in the first image the velocity is zero in
the boundary layer (blue color) showing that it respects
the no-slip condition and increase moving away from
the wall (red color). The maximum velocity registered
is 36.8m/s. It is important to notice how the fluid feels
the presence of the structure before run into it. This
happens because the fluid is subsonic and thus it starts
gradually to reduce its velocity due to the presence of
the building. The influence of the body in the fluid flow
is underlined by the deviation of the streamlines. After
the structure the fluid can not follow the contour of
the bluff body any more and a wake is generated with
Figure 3. Boundary conditions great increase in the vorticity as it is displayed in the
sixth figure.
The density of the air is approximated as a The total deflection of the structure as expected is
constant (ρ a = 1.225 kg/m3 ) and so the viscosity almost nil (4.39e − 5m), this is because we considered
(µ = 1.78 ∗ 10−5 Pas). These assumption are taken in the building made entirely of a glass silica.
order to simplify the problem because in the atmosphere
these quantities vary with the altitude, and because of
the considerable height of a tall building, for a more
accurate study this should be taken into account. This
will be investigated later on.
2 It is obviously not true that a building could be made entirely of
Concerning the structure, the properties of silica glass
this material. We use this solution in order to make up to the lack of
are assumed such as E = 73.1e9Pa (Young’s modulus), information. If one wants to determine the E,ν and ρ of a skyscraper,
ν = 0.17 (Poisson’s ratio), but assuming the density a detailed analysis should be needed.
4

Figure 4. Tapered building: velocity, streamline, deformation Figure 7. Tapered building: Cell Reynolds number

Direction Max force load [N/m2]


|x| 160
|y| 162
|z| 144

B. Parametric study: velocity


In this section we have study how a change in velocity
could affect the equilibrium solution. We have chosen
three different U0 (zr ) to be used in the power law in
order to determinate the wind profile; the values chosen
are respectively 5, 7.5 and 10 m/s. The figure below show
the three different wind profile. As one can see even if
the change in the velocity at the reference height is not
so big, at high altitudes there is a remarkable difference.

Figure 5. Tapered building: top view

Figure 8. Wind profile parametric study

In order to better understand the differences in the


velocity, the figure represents the three different velocity
Figure 6. Tapered building: vorticity fields at z = 260m.
5

Figure 11. Maximum total displacement (parametric)


Figure 9. Velocity field (parametric)

As could be noticed in figure 10, the change in the


velocity affects considerably the Reynolds number; the
greater is the value of the velocity, the greater will be
the Reynolds number. Here in the following figure are
reported the values of the cell Reynolds number at z =
260m.

Figure 12. Total displacement (parametric)

For what concern the vorticity, even in this case we


observe, as expected, an increment.

Figure 10. Cell Reynolds number (parametric)

In figure 11 the maximum total displacement values


are plotted: obviously the faster is the fluid the greater
will be the displacement; we can notice that the in-
crement in the total displacement follow a quasi-linear
curve. The other image shows the distribution of the total
displacement for the face perpendicular to the flow. On
the left side of the figure we have the value of the top
and on the right side the value of the bottom. Even if the
values are very small there is a big difference between
the first case and the last one. Figure 13. Vorticity (parametric)
6

C. Effect of geometry: No tapered building

In this study we want to investigate how the geometry


affects the equilibrium solution. We thus consider all the
variables exactly alike the case of the tapered building
but we use a structure with no taper as show in the figure
below.

Figure 15. No Tapered building: velocity, streamline, deformation

Figure 14. No Tapered model

Figure 16. No Tapered building: top view

The considerations about the flow pattern in the case


of a no tapered building are similar as the previous
ones and as showed in the images the velocity field
and the maximum velocity detected is almost the
same. However in this case the geometry of the building
generate a wake bigger than before; this is quite relevant
because the bigger is the wake, the greater will be the
drag and the more the structure will interact with the
turbulent structure. In this static analysis there is no
presence of the time dependent vortex interaction, but it
is almost sure that a bigger wake will result in a bigger
interaction with the edifice.
It can be also seen that the distribution of the whirling
structures is quite different with respect of the previous
analysis; as one can see in this case it seems that
behind the body the vorticity is more developed than
the previous case as it confirms the circular structure.
It seems that with this type of geometry an intenser
vorticity is generated.
For what concern the total maximum force loads in this
case are higher, however the displacement is still close
to zero.
Figure 17. No Tapered building: vorticity
7

Figure 18. No Tapered building: cell Reynolds number Figure 19. Temperature (ISA)

Direction Max force load [N/m2]


|x| 185
|y| 182
|z| 184

D. Parametric study: density and viscosity

In this section we improve our model introducing the


variation of density and viscosity in accordance with the
ISA model of the atmosphere [10]. In this model for an
altitude less than 11000 m, we have:

Figure 20. Density (ISA)


T = T0 + λz, (6)

where z is the altitude, T0 = 288.15K and λ is a constant


equal to −0.0065K/m.
For the density we have:

ρ = ρ0 (1 + λz/T0 )(−1− g0 /( Rλ)) , (7)

where ρ0 = 1.225kg/m3 , g0 = 9.80655m/s2 , R =


287.05m2 /(Ks2 ). For the viscosity we utilize the Suther-
land’s law which is based on kinetic theory of ideal gases
and an idealized intermolecular-force potential:

 T 1.5 ( T + S)
0
µ = µ0 , (8)
T0 ( T + S)

where µ0 = 1.78e − 5Pas and S = 110K. Figure 21. Viscosity (ISA)


8

It can be noticed that there are no big variation in this


quantities.
In our parametric analysis we considered 3 altitudes (0m
, 160m, 320m) at which calculate the density and the
viscosity and than we use them for the three different
simulations. A more accurate model should be made
considering the variation of density and viscosity at
every altitude step in accordance with the ISA equations.
The results are here reported:

Figure 24. Cell Reynolds number

V. Conclusion
This simple static 3D analysis shows that a tapered
building experience less forces than a non tapered
one. What is more interesting is that the for a non
tapered building the wake is bigger and the vorticity
distribution is different.
The parametric studies showed us the importance to
use a wind velocity model and to characterize it in the
best way in order to reproduce correctly the physics of
the problem: as we have seen in the velocity parametric
study there a slight difference in the U0 (zre f ), will easily
Figure 22. Maximum total displacement lead to different results.
Another observation is that the variation of density and
viscosity in accordance to the ISA model are not so
indispensable, as they are so low, and they could be
quietly neglect.
A further analysis could start from this model and
As one can see there is a very low linear variation in use a denser and moving mesh in order to simulate
the maximum total displacement: the lower the viscosity a time dependent analysis capable to characterize the
and the density, the lower will be the displacement. The fluid structure interaction focusing on the interaction
other parameters observed do not vary so much between between the vortexes generated and the building.
the three cases, but if we compare them to the first
study (tapered building with no ISA atmosphere), we
can observe slight variation for example in the maximum References
velocity and the cell Reynolds number. [1] Irwin P. A., 2009. Wind engineering challenges of the new gen-
eration of super-tall buildings, Journal of Wind Engineering and
Industrial Aerodynamics 97, 328-334.
[2] Cooper K.R., Nakayama M., Sasaki Y., Fediw A.A., Resende-Ide
S., Zan S.J., 1997, Unsteady Aerodynamic force measurement on
a super-tall building with tapered cross section. Journal of Wind
Engineering and Industrial Aerodynamics 72, 199-212.
[3] Kim Y.M., You K.P., 2002. Dynamic responses of a tapered tall
building to wind loads. Journal of Wind Engineering and Industrial
Aerodynamics 90, 1771-1782.
[4] Kim Y.M., You K.P., Ko N.H., 2008. Across-wind responses of an
aeroelastic tapered tall building. Journal of Wind Engineering and
Industrial Aerodynamics 96, 1307-1319.
[5] Dowell E. H., 2004. A Modern Course in Aeroelasticity (4th edi-
tion).Kluwer Academic Publishers.
[6] Panton R.L., 1996. Incompressible Flow, second edition. John Wiley
and Sons Inc..
[7] Reddy, J. N., 2008. An introduction to continuum mechanics : with
applications. Cambridge University Press.
[8] Beychok, Milton R., 2005. Fundamentals Of Stack Gas Dispersion
(4th Edition ed.). Author-published.
[9] Saad, Y., 2003. Iterative Methods for Sparse Linear Systems (2nd
ed.). SIAM.
[10] U.S. Standard Atmosphere, 1976, U.S. Government Printing Of-
Figure 23. Velocity fice, Washington, D.C.

Вам также может понравиться