APPLICATION OF COMPUTATIONAL FLUID DYNAMICS (CFD) IN WIND ANALYSIS OF TALL BUILDINGS Damith Mohotti, Priyan Mendis, Tuan Ngo Department of Infrastructures Engineering, The University of Melbourne, Victoria, Australia Abstract Computational Fluid Dynamics (CFD) has wide range of applications in modern engineering community. Use of CFD in wind engineering has significantly increased over the last few decades. However, very limited work has been done on the application of CFD in simulating the wind behaviour around tall structures. With the development of high speed supercomputers, the possibility of performing such analysis within reasonable time period has become a reality. This paper presents an outline of such study performed using an advanced finite element (FE) code. An isolated rectangular building model has been considered as the based model in the analysis. A constant velocity profile has been used with k-ε turbulent model. An atmospheric boundary layer based wind profile has been adopted using a user defined function in the simulations. The effect of neighbouring buildings onto the nearby tall building has also been discussed. Wind pressure development, velocity profile in the close vicinity of the building was studied and presented. Keywords: Computational Fluid Dynamics, wind loading, tall buildings. 1.0 Introduction There is a high demand for tall buildings in the central business districts (CBD s) of major cities all over the world. With the recent trend of building structures over 500 m with different complex geometries, it has created a great challenge for practicing engineers to design such flexible structures to withstand wind and other similar lateral loadings. As the geometry of a typical floor depends on many different constraints such as land area and optimum viewing angle for occupants, most of these structures have considerably high aspect ratios. The major concerns in designing of tall buildings are not limited to the resistance of the structural system to the lateral loads, but also the comfort of the building s occupants in relation to the wind-induced motion of the buildings. The current practice wind codes and standards are limited to the prediction of wind induced responses of tall buildings with square or rectangular cross sections and aspect ratios of not more than 6. Australian standard AS/NZS-1170.2 (Standards Australia, 2011) can be considered as one of the frequently used standards for such analysis all over the world. However, the standard clearly emphasises that the values given in the code are only applicable for structures below 200 m height. In addition, for free standing tall buildings the first-mode fundamental frequency shall be larger than 0.2 Hz. Therefore, tall buildings outside these limits, the AS-1170.2 has recommended to use wind tunnel tests as a supplementary technique in order to estimate the wind loads. However, performing wind tunnel tests require considerable resources and it is a time consuming and expensive effort. Therefore, it is worthwhile to look into alternative solutions to replace such experimental procedures. With the advancement of computer capabilities, it is now possible to simulate considerable complex numerical simulations within a feasible time period. CFD (Computational Fluid Dynamics) is getting popular among wind engineering research community due to its capability of modeling wind flow in different domains. Computational fluid dynamics, usually abbreviated as CFD, is a branch of fluid mechanics that uses numerical methods and algorithms to solve and analyze problems that involve fluid flows. Computers are used to perform the calculations required to simulate the interaction of liquids and gases with surfaces defined by boundary conditions. With high-speed supercomputers, better solutions can be achieved within a reasonable time period. Ongoing research yields software
that improves the accuracy and speed of complex simulation scenarios such as transonic or turbulent flows. CFD has been used effectively in modeling aerodynamics effect on automotives. Therefore, it has shown a considerable accuracy in simulating atmospheric boundary layer effect. This highlights the possibility of using a similar approach to simulate the wind behaviour around the buildings (Blocken et al., 2012; Hang et al., 2012). The fundamental basis of almost all CFD problems are the Navier Stokes equations, which define any single-phase (gas or liquid, but not both) fluid flow. Unlike flow around streamed line object, analysis of flow around sharp edged bluff-body involves many difficulties as pointed out by previous researchers. Accurate representation of flow separation, turbulent formation are vital for achieving better representation of actual scenario of virtual wind models. In this study advanced Finite element code, ANSYS FLUENT 14.5 (ANSYS Inc, 2012) has been used as the solver. Six different turbulence models are available in FLUENT which are incorporated with equations to solve the transported variable, turbulent viscosity, turbulent production and turbulent destruction terms. LES (large Eddy simulation) and K- ε models are the most commonly used turbulent models to represent the wind flow around building domains. There are considerable advantages and disadvantages of these models in terms of using in wind simulations. This paper only presents the results obtained using K- ε model even though the LES model (Zhang, 1994; Franke et al., 2004; Blocken et al., 2007; Blocken et al., 2007; Parente et al., 2011; Lou et al., 2012) studies are continuing in the current research study. This paper presents the preliminary work of an ongoing research project to simulate the wind pressure acting on tall buildings, in three main sections. Section 1 gives a brief introduction on CFD applications in wind engineering. Section 2 of this paper presents the details of the numerical model used in this study with a brief introduction to numerical simulation process. Section 3 of this paper discusses some of the findings of this study. A general summary and future plan of the current study are given under the conclusions. 2.0 Numerical Modeling A full scale geometrical model has been used in this study. The building has a rectangular prismatic shape with dimensions 100 m (x) by 150 m (y) by 600 m (z) height representing a true scale building in an open terrain. The flow is described in a Cartesian coordinate system (x, y, z), in which the Z-axis is aligned with the stream flow direction, the X-axis is in the perpendicular direction and the Y-axis is in the vertical direction. The computational domain dimensions, boundary conditions and the wind tunnel configurations are given in Figure 1. A half of the model has been used in the analysis to save computation cost using the symmetricity of the building and the fluid domain. As shown in Figure 1 the inlet and outlet boundaries have been extended to 8 times and 25 times the width of the building in order to obtain the undisturbed flow near the fluid domain boundaries. Top wall Inlet Principle building Outlet Bottom wall (a)
25 B 8B (b) Figure 1 Geometry of the numerical model H 2H (c) B D An engineering wind model for Australia has been developed in Melbourne from the Deaves and Harris model (D&H model,1978) (Mendis et al., 2007). This model has been developed based on full scale data and on the classic logarithmic law from which a mean velocity profile in strong winds applicable in non-cyclonic regions is derived, as given in Eq.(1). V z u 0.4 [log e ( z ) + 5.75 ( z ) 1.88 ( z 2 ) 1.33 ( z 3 ) + 0.25 ( z 4 ) z 0 z g z g z g z g V z u 0.4 log e ( z z 0 ) ] (1) (2) The numerical values are based on a mean gradient wind speed of 50 m/s. For values of z<30.0 m, the z/z g values become insignificant and the Eq.(1) can be simplified to Eq.(2). Where, V z is the design hourly mean wind speed at height z, u is the friction velocity as described in Mendis et al.(2007). This wind profile has been integrated as a user defined function (UDF) in addition to a constant velocity profile in simulation of incoming wind. Turbulent intensity of 1% and turbulent intensity ratio of 10% were adopted at the inlet while turbulent intensity was increased to 5% at the outlet. Figure 2 Wind velocity profile used in this study, Mendis et al. (2007) Table 1 Configurations used in the study Case Configuration Velocity profile Case 1 Isolated Continuous flow Case 2 Isolated Use D&H model Case 3 Full height adj. Bldg upwind of PB (principal Use D&H model building) Case 4 half height adj. Bldg upwind of PB Use D&H model Case 5 Full height adj. Bldg downwind of PB Use D&H model Case 6 half height adj. Bldg downwind of PB Use D&H model
In the present study, six different configurations under two different flow fields have been investigated. Table 1 summarises all the configurations considered for the numerical simulation in this study. Case I simulates the wind effects on an isolated building with a constant velocity profile. Case 2 uses same building model with the D &H velocity profile described above. Case 3 and 5 investigate the flow pattern and the pressure distribution on the two adjacent buildings of same heights both in upwind and downwind directions. The two remaining configurations given in table 1 investigate the effect of wind flow due to the presence of a building with half in height to the principal building considered. Although not proven in a systematic study, it is believed that, in general, a half height building could provide maximum interference effects. 3.0 Results and Discussion Flow behaviour around the building under a constant velocity profile is shown in Figure 3. The preliminary simulations clearly show the vortex formation in the wake of the bluff body. Even with the initial coarse mesh, the simulations were able to capture flow separations and vortex formation quiet well (Figure 3(b)). Even though the incident velocity is around 40 m/s, the sharp edges near the building corners have caused an increment in the velocity of the stream lines and jumped approximately closer to 50 m/s in the vicinity of the front edges. The above results agree with the results published by Baetke et al. (1990). Inlet 3D flow domain Velocity Stream lines Building model Outlet (a) Z= 150 m (b) (c) Figure 3 Velocity stream lines of the flow (Case 1) (a) 3D flow domain (b) At z=150 m (c) At the symmetrical plane The pressure coefficients of the building surface along the symmetrical plane of the building were obtained and presented as shown in Figure 4. A constant velocity profile was used in case 1 of the analysis. Therefore the pressure induced on the front surface (A- B) does not show considerable variation as expected in a real building. However, this simulation was used to explain the difference in
Cp (presure coefficients) pressure on the building with the application of atmospheric boundary layer wind profile. According to the results obtained from the analysis shows that the building front face has a maximum pressure coefficient of 1.26 while downwind face has a pressure coefficient of -0.95. These results are in agreement with the experimental results published by Dagnew et al. (2009). 1.5 1 B C 0.5 0-0.5-1 A D -1.5 A B C Position along the symmetrical plane-case 1 Figure 4 Pressure coefficients along the symmetrical plane (A-B) Obtaining the correct pressure development on the building surface is very important in the designing of tall structures to predict the behaviour of the structures correctly. This will help the designers to accurately predict the acceleration of the building which is one of the key elements in the designing of such structures. The user defined velocity (UDF) profile was used in the analysis (case 2-6) to represent the atmospheric boundary layer wind profile. The velocity vectors formation in the symmetrical plane of the domain is presented in Figure 5(a). Figure 5 (b) further elaborates the velocity contours developed in two directional symmetrical planes. D Velocity builds up according to D&H model Symmetrical plane (a) (b)
Figure 5 Velocity distribution incorporating the D&H model(a) velocity vector formation (b) velocity contours (Case 2) Figure 6 Velocity stream lines at two different levels of the buildings (a) z=150 m (b) z=480m (Case 2) Figure 6 shows the wake region formation at two different elevations of the buildings. As the velocity of the incoming wind is different at those two planes, the development of flows separation and the outgoing velocity are considerably different. In addition, the results obtained from Case 2 analysis show a considerable difference with the results obtained from the uniform wind velocity. The pressure distributions obtained from the simulations are presented in Figure 7 and 8. The horseshoe vortex shape contours generation on the front wall agree with the results presented by Dagnew et al.(2009) and Huang et al.(2012). The distribution of pressure iso-surfaces shows a considerable disagreement with the code adopted procedure for the pressure distribution of tall buildings. The pressure values increase with the increasing height. However, results show that by utilising the distribution given in CFD analysis, the design values can be optimised by considering high and low pressure zones in the buildings. Figure 7 Pressure distribution along the symmetric plane (Case 2) (a) (b) (c)
Figure 8 Pressure induced on the building surface (case 1) (a) 3D pressure contour map (b) windward face (b) leeward face The wind flow pattern of two close buildings with same size was studies in Cases 3 and 5. Shielding effect to a particular building from another can be positively used to reduce the vulnerability to wind induced loadings. The model was able to capture the low pressure zone generated due to the shielding effect from the neighbouring building and the turbulence created in between the buildings. This turbulent wind zone can affect the comfort of the pedestrians walking in the close vicinity. Therefore it is important to have some sort of vegetation or landscape those areas to improve the comfort of the pedestrians. The results for Case 5 given in Table 1 are presented in Figure 9. Inlet 3D flow domain (a) Principal building Wake region Low pressure zone (b) (c) Figure 9 (a) 3D representation of stream line distribution of two adjacent buildings with same size (b) at z=150 m (c) along the symmetrical plane (case 3) Figure 10 Influence of adjacent building to the pressure distribution (case 3) As shown in Figure 9, the presence of nearby building has a considerable influence on the aerodynamic response of the neighbouring buildings. Wind flow around different buildings located in
different geographical topologies therefore should be investigated on case by case basis. One of the other main configurations considered in this paper was the assessment of the influence of the adjacent buildings with half of the height to the principal building. In Case 5, the adjacent building was placed in windward (upwind) direction (As shown in Figure 11-12). In case 6, the smaller building was located in the down wind direction of the principle building. In both scenarios, the heights of the buildings were same. As reported by Dagnew et al (2009), a reduction of 125-150% in mean pressure has been observed in the principle building when it is shielded by the shorter building, when compared with the isolated building model. This is due to the complete sheltering effect coming from the adjacent building into the principle building model. However in the case of locating the adjacent building in the leeward direction, there is no considerable reduction in pressure observed in the front wall but shows a considerable reduction in the back wall. This low pressure region can cause considerable suction on the building which increases the deflection and acceleration of the principal building. And also these low pressure regions can cause discomfort to the pedestrians in the vicinity. Figure 11 Velocity vectors development around the buildings (case 4) Inlet Principal building (a) Wake region (b) (c) Figure 12 Velocity stream line distribution along a plane at z=175 m from the ground surface
A considerable velocity reduction was observed near the taller building. This causes less pressure on the front surfaces of the principal building. However a similar pressure distribution contour pattern has been observed in two buildings. When compared with the isolated building model, the centre of the pressure contours has moved in the vertical direction towards the top of the building. Therefore lesser loads can be expected on the building due to the wind pressure. There is a possibility of using these simple phenomena in future tall building design to protect against severe wind effects. Figure 13 Pressure development on the surface of the buildings in the windward face and leeward face Figure 14 Pressure development in the vicinity of the building along the symmetrical plane 4.0 Concluding Remarks A preliminary analyse of CFD for wind analysis of tall buildings is performed and presented. Several geometrical configurations were analysed in order to understand the effect of wind on isolated buildings and others with some surrounding buildings. The results show that CFD has a great potential to be used in wind engineering for tall structures. In order to achieve accurate atmospheric boundary layer wind profiles, a group of roughness elements were created on the bottom surface. This shows much improvement in the implementation of boundary layer wind profile in wind analysis. This study is in progress at the University of Melbourne, to investigate the possibility of incorporating the FSI (fluid-structure interaction) computational method into wind simulation of tall buildings.
References ANSYS Inc (2012). "ANSYS ", Release 14.5, Century Dynamics Inc, CA,USA. Baetke, F., H. Werner and H. Wengle (1990). "Numerical-Simulation of Turbulent-Flow over Surface-Mounted Obstacles with Sharp Edges and Corners." Journal of Wind Engineering and Industrial Aerodynamics, 35(1-3): 129-147. Blocken, B., J. Carmeliet and T. Stathopoulos (2007). "CFD evaluation of wind speed conditions in passages between parallel buildings effect of wall-function roughness modifications for the atmospheric boundary layer flow." Journal of Wind Engineering & Industrial Aerodynamics, 95(9-11): 941-962. Blocken, B., W. D. Janssen and T. van Hooff (2012). "CFD simulation for pedestrian wind comfort and wind safety in urban areas: General decision framework and case study for the Eindhoven University campus." Environmental Modelling and Software, 30: 15-34. Blocken, B., T. Stathopoulos and J. Carmeliet (2007). "CFD simulation of the atmospheric boundary layer: wall function problems." Atmospheric Environment, 41(2): 238-252. Dagnew, A. K., G. T. Bitsuamalk and R. Merrick [2009]. Computational Evaluation of Wind Presure On tall Buildings. 11th Americas Conference on Wind Engineering. San Juan, Puerto Rico. Franke, J., C. Hirsch, A. G. Jensen, H. W. Krüs, M. Schatzmann, P. S. Westbury, S. D. Miles, J. A. Wisse and N. G. Wright (2004). Recommendations on the use of CFD in wind engineering. Proceedings of the International Conference on Urban Wind Engineering and Building Aerodynamics, von Karman Institute, Sint-Genesius-Rode, Belgium,. Hang, J., Y. Li, R. Buccolieri, M. Sandberg and S. Di Sabatino (2012). "On the contribution of mean flow and turbulence to city breathability: The case of long streets with tall buildings." Science of The Total Environment, 416(0): 362-373. Lou, W., M. Huang, M. Zhang and N. Lin (2012). "Experimental and zonal modeling for wind pressures on double-skin facades of a tall building." Energy and Buildings, 54(0): 179-191. Mendis, P., T. Ngo, N. Haritos and A. Hira (2007). "Wind Loading on Tall Buildings." Electronic Journal of Structural Engineering: 41-54. Parente, A., C. Gorlé, J. van Beeck and C. Benocci (2011). "Improved k ε model and wall function formulation for the RANS simulation of ABL flows." Journal of Wind Engineering & Industrial Aerodynamics, 99(4): 267-278. Standards Australia (2011). AS/NZS-1170.2. Structural Design Action-Part 2: Wind Actions. Sydney. Zhang, C. X. (1994). "Numerical predictions of turbulent recirculating flows with a k-e model." Journal of Wind Engineering and Industrial Aerodynamics, 51(1994): 177-201.