Computer fluid dynamics application for establish the wind loading on the surfaces of tall buildings

Document Sample
Computer fluid dynamics application for establish the wind loading on the surfaces of tall buildings Powered By Docstoc
					                                              Recent Researches in Automatic Control

 Computer fluid dynamics application for establish the wind loading on the
                       surfaces of tall buildings
       Civil Engineering Faculty,(b) Civil Engineering Faculty, Institute of Structural Engineering, Structural Mechanics
                              University Transilvania of Brasov, (b) Technical University of Košice
                                  Str. Turnului Nr. 5. Brasov. Romania, (b) KSM UIS SvF TU v Kosiciach,, ,

Abstract: - The scope of this paper consists in the Computer Fluid Dynamics apply to establish the wind effects on
the surfaces of the tall buildings systems. The tall buildings are in present more and more frequently in the modern
cities. The topology of the buildings and the wind directions are more and more important in the case of new tall
buildings design. The new materials used in structural engineering with their high mechanical properties, the
construction machinery the technologies advances and the economical scope are the principal reasons in the design
and execution of taller and more slender structures. In practice the quasi-static computing method used for current
structure is exceed by inherit structural flexibility. The wind forces systems that load the buildings surfaces are in
practices extremely difficult and expensive. Even the use of wind tunnel is extremely difficult in principal by the
problems in the dumping values for the aero-elastic model assuming and Reynolds number scaling. The butt of this
article consist in development of a CFD method very useful as a complimentary tool for easy establish the wind
tunnel parameters work conditions. In the present paper the CFD modeling is based on Volume Finite Method (VFM)
with Total Validation Diminishing Method (TVD) for modeling the flow around complexes buildings surfaces and
establishes the pressure variation around them. The new CFD VFM algorithm is developed for minimal errors and
based on Riemann and Godunov solvers. The results of this method are compared with the quasi-static models and
with the most used numerical method.

Key words: Volume finite method, CFD, surface pressure, TVD method, tall buildings, steel structure.

       1. Introduction
The wind loading is considered by Devenport(1998)                    The Eurocode and most used standards are
in three categories:                                                 highlighted by Allsop (2009) as the most flexible
a) Extraneously-induced loading based on naturally                   and inclusive code for normal buildings. The quasi-
turbulent oncoming wind. The weak of upstream                        static methods offered by these codes are only
obstructions enhance this categories buffeting.                      applicable for buildings with structural properties
b) Unstable flow phenomenon such as separations,                     such that they are not susceptible to dynamic
reattachments and vortex shedding generate a                         excitation (Metha, 1998). Thus, the tall buildings,
secondary type of forces.                                            those with high slenderness ratios and/or
c) The movement-induced excitation of the body                       asymmetric planes, exceed limitations and are
generate by the deflection of the structure create                   advised to be tested in the wind tunnel.
fluid flow too. This phenomenon with a strong                        This pattern induces inherent structural flexibility
unsteady states character gives the complexity of the                and heightens concerns regarding the aero-elastic
fluid flow around the flexible tall structures. The                  fluid-structure interaction between the wind and the
modern design of flexible tall structures must                       tall building. The codes of practice have been
request to earth quakes events and wind loads, cases                 formulated with a view to providing an acceptable
that represent a state of the art of the civil                       balance between the overly complex reality and
engineering.                                                         oversimplified approach.

           ISBN: 978-1-61804-004-6                             433
                                        Recent Researches in Automatic Control

Scaled-model wind tunnel testing is an established              literature. It proceeds to highlight the user-defined
tool among industry design practices. Boundary                  criteria that must also be satisfied. Finally, the scope
layer wind tunnels are capable of quantifying time-             for future research on simplifying CFD analyses for
dependent surface pressures, including the complex              tall buildings is discussed, with a view to producing
types of loadings (torsion and acrosswind). The                 a more efficient and practical solution.
model can be used to determine the best orientation             The modern simulation is based on turbulence
of the proposed building, the case of Burj Dubai                models, based on different models of turbulence
analysis (Irwin and Baker, 2006). The simulation,               used for solving the conservation of mass,
nevertheless, it has its own limitations that include           momentum and energy equations. Sun et al., 2009
the difficulty to maintain proportionality between              and Castro, 2003 use different CFD turbulence
the scaled turbulence characteristics and the scaled            model for analyze the wind loads on the tall
building model, especially if the topography is                 buildings. The selection of the turbulence model is
significant (Taranath 1998). Furthermore, it is                 made based on accuracy, computational cost,
important to ensure Reynolds number effects on the              accessibility and available time for simulations. The
pressures are kept to a minimum. It is noted by Sun             most complete form of CFD is the Direct Numerical
et al. (2009) that a computational approach has the             Simulation (DNS) method that uses the direct
capability of being more flexible than traditional              solution of Navier-Stokes equations for each control
wind tunnel experiments. For example, a fully                   volume. The disadvantage of this method consists in
coupled solution between computational fluid                    the mesh size dimension conditions. The cell mesh
dynamics (CFD) and finite element modelling                     must be smallest that the vortex eddy within the
(FEM) can be developed to model the fluid-structure             flow for capturing the turbulent effects. So, the cost
interaction (FSI). A wind tunnel test relies on the             of DNS become extremely high and Knapp (2007)
simplified assumption that the scaled aeroelastic               conclude that the DNS method should be limited by
model can satisfactorily replicate the dynamic                  a small scale simulation and low Reynolds numbers.
properties of the full-scale design. It neglects the            The other methods are implemented in ANSYS Inc,
influence of higher modes. The application of CFD               (2005) and is based on Reynolds-averaged Navier-
for practical wind engineering problems has                     Stokes (RANS) and Large Eddy Simulation (LES)
received a lot of research attention over the last              and are the two most used method for wind load
three decades and has made major progress due to                simulations. The RANS methods are based on the
the advancement of computer technology. Thus far,               two popular most used models, κ−ε and κ−ω, where
the leading applications for the built environment              κ represent the kinetic energy and ε the turbulent
have concentrated on mean wind speeds for areas                 dissipation rate or ω the specific dissipation rate.
including: natural ventilation; pollution dispersion;           The RANS method is based on additional empirical
and human comfort at street and balcony level                   equations for establish the turbulent viscosity and
(Stathopoulos, 1997). It has proven very difficult for          are relatively simple to use and robust and can
CFD to acceptably model the complex flow                        describe the full spectra of turbulence scale. Other
interference phenomena induced from buildings.                  methods are based on transient solutions based on
Typical features of this unsteady flow regime                   spatial filtering approach adapted with subgrid-scale
include turbulent length scales and separation                  for smallest cell dimension eddy. The mesh in that
regions larger than the body size of the structure.             method must be smallest that in the RANS method
This is the reason less work has been performed on              but the filtering approach can give more information
predicting time-dependent surface pressures on                  about turbulence areas of the model. In the last
these man-made bluff bodies. CFD has not                        years, was developed a new method based on the
developed enough to suggest it could replace wind               RANS and LES models, named Detached Eddy
tunnel testing in this respect. It does, however, offer         Simulation (DES), method where the simplest
encouraging potential to act as a complimentary                 RANS algorithms are used for majority flow
tool. In this paper, the various turbulence models              domains simulate and LES is used only in the area
will be discussed with respect to their ability to              of separated flow. The use of DES method in the
predict surface pressures and resulting wind loads              wind load estimation of tall structures consists in the
for a tall building. This includes a detailed review of
previous validation studies performed within the

       ISBN: 978-1-61804-004-6                            434
                                                    Recent Researches in Automatic Control

high turbulent area developed around the buildings                         and in the specificity of wind flow.

   2. Numerical analysis

                                                                                           (   )
In this paper we establish the pressure distribution                                      r
on the façades of the tall buildings using the TVD                                 + ∇ ρ ⋅V = 0                                     (1)
algorithms for gas dynamics based on Euler PDE
                                                                                     (     ) (             )
                                                                                      r        r r               r
system of equation for a situation of wind loading                                 ρ ⋅V + ∇ ρ ⋅V × V = −∇p + ρ ⋅ g
more closely of the reality. The domain of                                      ∂t                                                  (2)
computing is established in the figure 1 and the
                                                                                ∂                                r
wind input speed diagram in the figure 2. The                                      ( ρ ⋅ E ) + ∇ ( ρ ⋅ E + p ) ⋅ V = ∇ ( λ ⋅ ∇T )   (3)
system of partial differential equations is give on the                         ∂t
forms of mass conservation (1), momentum
conservation (2) and energy conservation (3):

                                                                             Index       Value Index      Value      Index Value
                                                                              Ox          [m]   Oy         [m]        Oz    [m]
                                                                               L           96    Y          96         Z     96
                                                                               L1           6   Y/2         48        H1     21
                                                                               L2          21   Y1          36        H2     69
                                                                               L3          46   Y2          33        H3     12
                                                                               L4           3   Y3          24
                                                                               L5          36   Y4          24
                                                                               L6          18

                            a)                                                                             b)
   Figure 1. Geometrical domain dimensions.

   The system of equations is write as:                                       The total energy E represent the sum of internal
                                                                              and kinetic energy and is expression is:
          q = f ( q ) + g ( q ) + h( q ) (4)
   Where:                                                                            E=
                                                                                      γ −1 2
                                                                                           + ρ u 2 + v 2 + w 2 (6)     )
            ρ                 ρu 
            ρu                ρu 2 + p                                    And the equation of state for a g-law polytropic
                                                                          gas considered in the present work has the shape
   q =  ρv ; f ( q ) =  ρuv ;
                                         
            ρ w               ρuw                                                γ=
                             u (E + p )
                                                                                     cv
                ρv                      ρw                             Where cp and cv are the specific heat at constant
                ρuv                     ρuw                            pressure, respectively constant volume.
                                                 
   g ( q ) =  ρv 2 + p ; h( q ) =  ρvw 
                                        2        
                ρvw                     ρw + p 
               v(E + p )
                                        w(E + p )
                                                   

   2.1.           Boundary and Initial Conditions and input particularities

       ISBN: 978-1-61804-004-6                                       435
                                     Recent Researches in Automatic Control

The Boundary conditions used in the solving                 India, in agreement with the design norms for wind
problems consist in free output from the planes             load. The speed distribution on Ox is considered
P2,P3,P4 and P5, as in figure 2a and in input from          linear on all the inlet area for verification the
plane P1. The inlet gas in the computation domain           turbulence that can appear in the fluid trap between
is made from plane P1 with a speed distribution as          the two buildings with different height. The inputs
in figure 2b. The speed distribution is made in             from plane P1 have a pulse shape with equal time
accord with the maximum speed at 60 year                    between the pulse duration and pause between two
measurement on wind in the most affected area on            pulses.

                           a)                                                    b)
  Figure 2. The plane notation of the domain analyzed, a) and the inlet speed distribution on Ox direction.

  3. Results
The analyze give the maps of gas speeds, u,v,w              the pressures on the façade and the other surfaces
energy and pressure inside the computing domain,            of the tall buildings.

                                    Figure 3. The streams line of the gas flow and the pressure in the
                                    façade section at time 0.915 s of simulation.

        a) t=0.152484 s           b) t=0.299869 s                c) t=0.446137 s           d) t=0.592342 s

     ISBN: 978-1-61804-004-6                          436
                                      Recent Researches in Automatic Control

      e) t=0.731194 s              f) t=0.871237 s            g) t=1.534835 s              h) t=5.39784 s
 Figure 4. Gas speed U [m/s] in the xOz symmetry plane of the domain of analyze for different time moments.

      a) t=0.299869 s              b) t=0.446137 s                  c) t=0.592342 s          d) t=0.731194 s

      e) t=0.871237 s              f) t=0.985921 s                  g) t=1.107586 s          h) t=1.245531 s

      i) t=1.390198 s             j) t=1.534835 s                k) t=1.679116 s               l) t=5.39784 s
     Figure 5. The map of pressure on the facade of the tall building for different moment of the aplication.

    a) t=0.871237 s             b) t=1.390198 s                c) t=2.114628 s              d) t=5.39784 s
Figure 6. The pressure map on the back surface of tall building for diverse moments of the application.

   ISBN: 978-1-61804-004-6                             437
                                          Recent Researches in Automatic Control

          a) t=0.871237 s            b) t=1.390198 s                c) t=2.114628 s             d) t=5.39784 s
      Figure 7. The pressure map on the right surface of the tall building for diverse moments of the application

        a) t=0.871237 s           b) t=1.390198 s               c) t=2.114628 s           d) t=5.39784 s
     Figure 8. The pressure maps on the left surface of the tall building for diverse moments of application.

   4. Conclusion
 The gas dynamics modelling based on Euler PDE                   wind loads and the gas is practical incompressible.
 system of equation can solve the problems of wind               The turbulence area of flow, that in the civil
 loads on tall structures without using the Navier-              engineering have a huge area of the domain (60-
 Stokes PDE system of equation with diverse                      80%) in the cases of wind loads on tall buildings
 turbulence flow models for accurate flow                        can be simulate using the Euler system of
 dynamics.                                                       equations and the complicated turbulences models
 A combination between the two modelles can be                   can be avoided.
 used because the gas speeds are low in the case of

   5. References

1.Allsop, A. BS EN 1991-1-4 Tall Buildings. In: ICE (Institution of Civil Engineers), New Eurocode on Wind
Loading. (London, UK, 11 May 2009).
2. ANSYS Inc. FLUENT 12.0 Theory Guide. (2009)
3. Building Research Establishment. Wind around tall buildings, BRE Digest 390, (BRE: Watford, UK, 1994).
4. British Standards Institution. BS 6399-2:1997 Loadings for buildings – Part 2: Code of practice for wind loads.
(BSI: London, UK, 1997).
5. British Standards Institution. BS EN 1991-1-4:2005 Eurocode 1: Actions on structures – Part 1-4:General actions
– Wind actions. (BSI: London, UK, 2005).
6. Castro, I.P. CFD for External Aerodynamics in the Built Environment. The QNET-CFD Network Newsletter. Vol.
2: No. 2 – July 2003, pp 4-7 (2003).
7. Davenport, A.G. What makes a structure wind sensitive? in Proc. The Jubileum Conference on Wind Effects on
Buildings and Structures. 8. Porto Alegre, Brazil 25-29 May 1998, eds. A. A. Balkema, (Rotterdam, Netherlands,
1998) pp 1-14.
9. Irwin, P.A. and Baker, W.R. The Burj Dubai Tower Wind Engineering. Structure Magazine, June 2006 pp 28-31
10. Knapp, G.A. Improved methods for structural wind engineering. Ph. D. University of Nottingham(2007).
11. Nozu, T., Tamura, T., Okuda, Y., and Sanada, S. LES of the flow and building wall pressures in the center of
Tokyo. Journal of Wind Engineering and Industrial Aerodynamics, Elsevier, 96 pp 1762-1773 (2008).

       ISBN: 978-1-61804-004-6                             438