Computational assessment of wind flow field on and around atypical buildings

. This article provides an overview of pressure coefficients ( Cp ) on atypical tall buildings with the application of CFD. Various modifications in architectural shapes on tall buildings eventually lead to a reduction in the wind load on building surfaces. The surface pressure on conventional (Square and rectangular) buildings is relatively different in comparison to other tall buildings. This study is to evaluate the surface pressure coefficient over rectangular, taper and setback buildings. The computational results show that the taper building has 7% Cp rise at ground level ( y/H= 0.225) in the windward face, and 34% Cp fall at the middle level ( y/H= 0.475) in the side face when compared with the rectangular building. Whereas for the setback building, Cp at ground level near setback ( y/H= 0.225) has reduced to about 25% and about 6% at the middle level ( y/H= 0.475) in windward than that in the rectangle building. Also, the side faces of the setback showed a 15% drop in Cp than other buildings. In leeward face, Cp is reduced to 56% near setback at the top of the building ( y/H= 0.725). This valuation of the Cp on these buildings shows that the effect of setbacks on building reduces the pressure variation on all faces and the downstream wake vortices.


Introduction
In developed countries, there is an extensive emergence of tall structures. They are highly sensitive to the action of wind force which is a demanding task for the structural engineers and architects in connection with habitability, serviceability, and survivability. And the assurance of essential structural performance also becomes difficult to vindicate. The influence of strong wind excitation intensity on tall buildings is effectively felt on adjacent buildings and wind environments outdoor. It inevitably affects wind acceleration at the ground level, wind noise, vegetation and the comfort of the pedestrian. Wind tunnel testing and field investigations are the two notable methods used to explore and study the wind flows. Both the assessments are costly and time-consuming.
An alternative as well as a redundant tool to study and investigate wind loads on buildings in recent times, is CFD (Computational Fluid Dynamics). CFD can be referred to as a virtual wind tunnel in wind engineering which is capable of generating environmental flows to predict the flow physics. The CFD procedures in wind engineering studies has recognized a plethora of importance from several international engineering communities. It has motivated the establishment of CFD practice guidelines by Franke et al.
[1&2], Tominaga, Y [3] and Blocken et al. [4]. 3D RANS model approach was used by various researchers in the past to study wind flow around buildings (Jian Hang et al. [5], W.D.Janssen [6], Jianlin Liu and Jianlei Niu [7], Blocken et al. [8], K.T. Tse et al. [9]). H. Tanaka et al. [10] presented the flow features like pressure coefficients, overturning moment coefficients and PSD (Power Spectral Density) on and around aerodynamically modified tall buildings using numerical and experimental results. From the experimental analysis of suburban and open terrain wind flow, Yong Chul Kim & Jun Kanda [11] concluded that the setbacked models are more practical to reduce the fluctuating lift force than the tapered or square models. Later, time-domain and frequency-domain analysis were performed by Yong Chul Kim [12] on a square, tapered and setback models with the side ratio of 1:1. A lot of unconventional building studies are being carried out. Yong Chul Kim et al. [13] examined the wind-induced coupled motion on plan varying tall buildings vs square buildings and found that the along-wind and torsional accelerations are smaller in the first. Using DDES (Delayed Detached Eddy Simulation) and IDDES (Improved Delayed Detached Eddy Simulation) turbulence models, Rajasekarababu et al. [14] explored the velocity profiles, upstream stagnation, and surface pressure distribution around a setback building. They concluded that DDES under predicted the downstream recirculation compared with IDDES turbulence model. Concerning recent literature studies, complex time-varying 3D wind flow field on and around the buildings have been estimated accurately with the support of IDDES turbulence model in CFD techniques.
The following section generally outlines several computational parameters that may affect both the accuracy and efficiency of computational simulation, such as the computational domain and its setting parameters. Meanwhile, pressure coefficients assessed in four different heights (y/H=0.225, 0.475, 0.725 and 0.975 respectively) are shown in fig1. ANSYS Fluent 18 is used to perform the CFD simulation, and IDDES turbulence model is used to assure the validity and reliability of this computational assessment. Section 3 comprehensively compared wind pressure coefficients on and around tall buildings throughout their perimeter. Overall, the outcome of this assessment is to enlighten the engineers and architects with a elementary understanding of tall buildings. It helps to study indoor and outdoor natural ventilation environments.

Model Description and Computational Parameter settings
This work considered a setback building of 210 m full scale height with side ratio 1:1.5 and roof-floor to base -floor area ratio 1:6.25 with a roof and three setbacks. The geometric scale chosen in the wind tunnel experiment is 1/300 for the open terrain environmental wind flow. The longer building face is 0. 015m, and the shorter face is 0.01m (D). H is the total height of the model, which is taken as 7D (H is same for all buildings); base floor area is 0.15m x 0. 1m and subsequent setbacks in higher floors are d' =D/1.25, d" =D/1.66 and d'''=D/2.5. For the rectangular building, roof dimension is same as that of base floor dimension and for taper building roof dimension, d=D/2.5 and is as shown in Fig.1(a) (For all building the base floor dimensions are the same (0.15m x 0. 1m)). The grid distribution was done in ANSYS ICEM CFD for the 3 buildings is shown in Fig.1 (b). A grid independence test was done, which includes a comparison of results from grids of various factors like, first wall distance, stretching ratio and density. The refined grid for rect, taper and setback buildings is 8.9, 8.2, 8.7 million cells respectively. Further, the solver settings are performed on ANSYS FLUENT 18.0 with SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) pressure-velocity coupling algorithm. With the combination of Shear Stress Transport, 3-D unsteady RANS equation is solved in IDDES following Rajasekarababu et al., [15]. For spatial discretization in the pressure-based solver, Least Square Cell-Based method was used as gradient, and the second-order scheme was chosen for pressure. The second-order upwind scheme is used for turbulent kinetic energy and turbulent dissipation rate terms. Bounded Central Differencing for momentum and transient formulation second-order implicit was used in the numerical methods. The convergence is assumed to be achieved when all the scaled residuals are levelled off and when the XYZ momentum, K, ω and continuity are set to at least 10 -5 . Whereas, the time step size is (Δt=Typical cell size / Characteristic flow mean velocity) Δt= 0.00014. The assumed residuals of the computational simulation were achieved and monitored over a considerable period.

The Pressure coefficient (Cp) on the tall building:
The below-given equation used to calculate the pressure coefficient (Cp).
Where p denotes is the extracted pressure from the required height. p0 is the static pressure at the reference height, ρ indicates the density of air (1.225 Kg/m 3 ), U 2 H is the mean wind velocity at the building's reference height. The pressure coefficients are measured below 10% of each setback height (see d, d', d" in Fig.1). The pressure coefficient extracted at three different heights on the building are compared and are shown in Fig.3. The assessed pressure coefficient on taper and setback building are compared with rectangular (rect) building. In the windward face, taper building experienced high pressure compared with rect building and increased 3% at y/H=0.225. In contrast, setback Cp fell 21% at the same height. In the windward face, at y/H=0.475, both tapered and rect building showed the same positive pressure whereas at y/H = 0.725 taper building showed a 7% increase concerning rect building. Similarly, setback windward face at the corresponding heights mentioned before falls by 4% and 17% respectively. The Cp for windward face is positive, and the windward edges experienced negative Cp due to wind shear. The side face of the building experienced fluctuating pressure due to strong shear wind force on its leading edges. The pressure in the right side face reduced by 11%, at the same time, there was a sudden pressure fall of 32% in the leeward side when compared with rect building at y/H=0.225 for both taper and setback. The Cp in the right side face of taper and the setback building were seen reduced by 5% and 12% respectively at y/H=0.475. At the same instance, there is a dramatic reduction of 51% in the leeward face for setback and just 16% for taper. Left side face of the taper and setback building was observed to be reduced by 12% and 32% respectively. The wind profile exponentially showed changing along with height; the increase of wind velocity influences the wind flow around the building. The pressure coefficients on the edges of the rect building at y/H=0.725 observed a considerable drop, as shown in Fig.3. (c). Shows that a huge suction created on the side faces. Similarly, a large fluctuation was noticed in the leeward face, which indicates the vortex formation. Furthermore, at y/H=0.725 on the right side face of the taper and setback building, Cp decreased by 70% and 90% respectively. Moreover, there is also a 54% and 82% reduction of Cp on the left side faces of the taper and setback buildings respectively. At y/H=0.975, taper and rect buildings evidenced vortex shedding and wake formation in downstream; as a result, the pressure drops on the leeward face as shown in Fig.3. The pressure distribution on each side of the building shown in Fig.4. For rect building Cp= -2.2 is the maximum negative pressure on the top face (roof), which depicted in Fig.4 (a), the strong upwind shear layer creates the cavity zone on the roof of the rect building which leads to recirculation. In Fig.4 (b) and (c) taper and setback roof has maximum negative Cp of -1.2 and -0.8 respectively, this shows modifications along with the height of the building, and there is also moderations in the wind loads on roof, thereby decreasing the recirculation on roofs. The vortex formations on buildings are visualized in CFD to compare the vortex structures behind them; a Q-criterion used to identify the wake vortices using Eq. (2).

Q r r s s 
Isosurface Q-criterion with non-dimensional velocity contour is used to obtain the preferred large eddy flow structure (excluding small-scale vortices of no interest). The instantaneous vortex structures identified behind the building (with Q=750) by IDDES turbulence model using fine mesh are shown in Fig.4. This indicates the capability of the simulation to solve turbulence by providing shielding function for the SSTbased IDDES model in order to achieve highly chaotic large and small scale vortices. The IDDES captures the vastly separated flow of 3D stream wise and transverse vertical structures. This flow structure reveals the streams behind the unconventional building physics, especially in the sides of the separation and wake regions. The visualization of vortices indicates the intensity of the turbulence downstream for the buildings. The upstream horseshoe (HS) vortex in the windward face of the rect building has been compared with the other two buildings. At the same time, the trailing vortex structure in rect building has an un-symmetrical wake region which shows the presence of a wake oscillation with lower frequency. This phenomenon opposes the higher frequency vortex shedding with higher energy turbulence which is shown in Fig.5 (a).
In taper building, the area of the wake region reduces along with the height (like a wake funnel as in Fig.5  (b)). The trailing wake regions in the setback building differs from the other two buildings as the setbacks on building break the vortices into small parts which eventually slows down the downstream driving parts. Moreover, it reduces the vortex stretching in downstream, sees Fig.5 (c). Simultaneously, the permanently attached upstream edge vortices in windward face and roof of the setbacks are relatively smaller compared to the rest of the buildings.

Comparison of Turbulence Intensity around the buildings:
(a) (b) (c)

Fig. 6 Turbulence intensity around buildings (a) Rectangle, (b) Taper & (c) Setback.
Turbulence Intensity (TI) contour for three buildings is presented in Fig.6. The TI changes along with the height in the wake zone. Regarding the intensity of the turbulence on rect building, the maximum intensity is located in the recirculation region due to viscous shear separated from the leading edge. In taper, the roof leading edge shear stress influences intensely in downstream wake zone ( Fig.6 (b)) and the TI core extended up to 5D. The presence of set-backs design on building breaks the core intensity of the turbulence substantially and dissipates into small eddies in downstream. (Fig.6 (c)). Where Cd and Cl are the drag and lift force coefficients, FD is the streamwise wind force, FL is the transverse wind force, and U 2 H is the reference velocity at the measured height and D is the projected width parallel to the wind direction. Fig.7. Shows the comparison of aerodynamic force coefficients of the buildings.

Fig. 7 Aerodynamic coefficient of atypical buildings (a) Cd, (b) Cl
The aerodynamic coefficients are measured at 8 different heights and all the aerodynamic coeffcient curves varied with height. Especially in setback building Cd values reduced near the setback due to flow break down. In rect building, the value of Cd is high nearly 2.3 at ground level, and it reduces along with height as seen in Fig.7. Furthermore, taper building Cd value is high at 8 th level (y/H=0.975), which is quite close to 0.9H measured by Kim et al., [16] in their wind tunnel testing. The main cause of these drag variation along the height is due to variation in wind pressure distribution on the leeward face as a result of the characteristics of geometric shapes of the building. The advantage of height modification in tall buildings is that, as a downdraft is suppressed on account of increasing the width with low wind speed (ground level wind) and an up-flow is accelerated on account of increasing width with high wind speed, the height variation of leeward distribution is large, which also influence on the drag force coefficients which is shown in Fig.7 (a). It is also noted that the lift coefficient (across wind force) values also show that setback and taper building is effective in across winds compare to rect building.

Conclusion
The purpose of this study is to assess and compare the wind flow features around conventional (rectangle) tall building over un-conventional (taper and setback) buildings using IDDES turbulence model. The substantial recommendations of this study derive the following conclusions.
 At y/H=0.475, presence of setback on the building along with height shows 51% Cp reduction and just 16% for taper in comparison with rect building for the leeward face. This is because the setbacks slow down the downstream.  Dimensional changes along the height of the building showed a reduction of 70 and 90 percentage in Cp on the right side face of the taper and setback buildings at y/H=0.725. Moreover, there is a 54% and 82% reduction in Cp on the left side faces of both the buildings respectively.  Instantaneous vortex structure indicates the formation of hairpin vortices with high energy at downstream due to the driving and lagging vortices. Due to largely expanded wakes, wake regions in the taper building was observed falling along with the height.  The features of trailing vortices of setback building differ from other buildings due to downstream vortex breaks down, stretching of vortices reduced by the setbacks and the intensity of the turbulence also reduces along with the height.  The reduction of drag coefficient is mainly initiated from the leeward pressure distribution.
Overall, from this assessment, setbacks on building reduce the wind load and ensure comfort compared with conventional tall buildings. Also, IDDES turbulence model can enumerate detailed complex flow features with limited grid spacing.