Elsevier

Engineering Structures

Volume 136, 1 April 2017, Pages 133-148
Engineering Structures

Enhancing wind performance of tall buildings using corner aerodynamic optimization

https://doi.org/10.1016/j.engstruct.2017.01.019Get rights and content

Highlights

  • Automated optimization procedure for reducing wind load on tall buildings is developed.

  • Turbulent atmospheric boundary flow and wind directionality are considered.

  • Two application examples for minimizing drag and lift, respectively, are presented.

  • Neural networks trained with 3D LES of 2D flows are used to evaluate the objective functions.

  • Optimal shapes are verified by comparing wind-induced responses using 3D LES of 3D flow.

Abstract

Wind-induced loads and motions of tall buildings usually govern the design of the lateral load resisting systems. The outer shape of the building is one of the many parameters that affect these design wind loads and responses. This study presents building corner aerodynamic optimization procedure (AOP) to reduce the wind load, by coupling an optimization algorithm, large eddy simulation (LES) and an artificial neural network (ANN) based surrogate model. As an illustration, corner mitigation that has limited effect on both the structural and the architectural design is presented. Two aerodynamic optimization examples focusing on drag and lift minimization that consider wind directionality and turbulence are presented. For example, reductions in the order of more than 30% both in along- and across-wind responses are obtained through a two-surface chamfering that was constrained to 20% of the building width.

Introduction

New generations of tall buildings are becoming increasingly taller, flexible and slender primarily driven by novel developments in design methods and new construction materials and techniques. This in turn makes tall buildings more sensitive to lateral loads such as wind. In addition, there is a need to lower the building weight in order to decrease the gravity loads to control the inertial forces developed by earthquake. This further contributes to an increase in the wind-induced forces and motions. As a result, wind-induced loads and motions typically govern the design of the lateral load resisting systems in tall buildings. The outer shape of the building is one of the main parameters that affect these loads and responses. The dependence of the wind load on the building shape makes the generalizations of wind load for tall buildings almost impossible, because every complex shape and surroundings produce a unique set of design wind loads. On the other hand, this dependency on the shape provides a unique opportunity to reduce the wind load through outer shape modifications either globally or locally. In that context, global modification involves major changes on the form of the building, which has a considerable effect on the overall architectural and structural design. This includes large openings, tapering, twisting, set-backing, etc. The architects can implement global modifications at the early conceptual design of the building if the modifications fit with the major functionalities of the building. On the other hand, local modifications result in minor changes on the building shape that have limited effects on the structural and architectural designs. Thus, the architects can introduce the local mitigations at a later stage of the conceptual design. One such local mitigation is corner modification; whish is the focus of the present study.

The outer shape of tall buildings is typically aerodynamically “bluff” and characterized with sharp corners. Wind loads for tall buildings with various shapes have been widely investigated in many numerical and experimental wind engineering studies, few examples include Vickery [1], Lee [2], Okajima [3], Igarashi [4], Nakamura and Ohya [5], and Merrick and Bitsuamlak [6]. Many researchers have reported that careful modification of the shape of the corners can provide better aerodynamic performance [7], [8], [9], [10]. Fig. 1 summarizes the widely used corner modifications in literature. Boundary Layer Wind Tunnel (BLWT) based studies [11], [12], [13], [10] reported chamfered, recessed and rounded corners to be effective in reducing the along- and across-wind forces. Kwok and Bailey [14] reported that finned corners increase the along-wind and decrease the across-wind responses, while slotted corners reduce responses in both directions. Tamura and Miyagi [9] reported that 2D flow BLWT tests were sufficient to indicate the aerodynamic improvements by corner modifications similar to ABL flow tests. Table 1 summarizes the scope and main findings of previous experimental and computational studies focusing on aerodynamic modifications of tall building corners.

As summarized in Table 1, BLWT has been widely used for studying building aerodynamic mitigations. This approach is reliable but only useful to compare limited number of feasible building shapes in addition to being costly for repetitive investigation. A wide portion of the search space remains unexplored as the search space is only limited to the tested options [19]. On the other hand, integrating CFD with an optimization algorithm can be more useful to explore wider geometric alternatives to find near optimal shapes. This is inspiring an increased number of researchers to work on building aerodynamic optimization applications. For example, Kareem et al. [20], [21], [22] introduced an approach for tall building corner optimization to reduce drag and lift by adopting low-dimensional CFD models. This approach is useful to overcome the computational cost associated with the iterative procedure required for optimization. Bernardini et al. [19] investigated the efficiency of utilizing Kriging model as a surrogate model for the objective function evaluation. The utilization of a surrogate model reduced the computational time. In these studies, Unsteady Reynolds-Averaged Navier–Stokes (URANS) equations were used. Although these studies developed a very promising and useful approach for building aerodynamic optimizations, some limitations are observed. For example, (i) wind directionality effect is not considered, (ii) low-order CFD models are used to evaluate shape alternatives, although wind performance assessment usually requires the use of high accuracy CFD- or BLWT-based evaluations. Using these novel approaches, it is possible to infer the relative performance of the various geometric alternatives (i.e. comparing alternatives) adopting the reduced order 2D simulations. A similar conclusion was also reported by Tamura and Miyagi [9]. However, adopting a simplified low order simulation can significantly reduce the analysis accuracy that may affect the conclusions observed under such simplified scenarios. Particularly when simulating the turbulent atmospheric boundary layer (ABL) flow and its interaction with a tall building. In the author’s opinion, the CFD simulations used to assess wind loads on buildings shall be commensurate with the complexity encountered in urban flows. These complex interactions can be realistically captured through LES as reported by Nozawa and Tamura [23], Dagnew and Bitsuamlak [24], [25], Aboshosha et al. [26] and Elshaer et al. [27]. It is to be noted that the accuracy of LES depends on the proper selection of the inflow boundary conditions and the adopted grid resolution. Thus, the consistent discrete random flow generator (CDRFG) technique developed by the authors is utilized to validate the wind responses for the best performing shapes. This technique was previously adopted to study a low-rise building [28], a standalone tall building [29] and a surrounded tall building in a city citer [27].

Building on these interesting benchmark aerodynamic optimization studies and targeting to address their shortcomings, the current study presents a new Aerodynamic Optimization Procedure (AOP) that uses LES and accounts for the wind directionality effects. In this procedure, 3D LES models of a 2D flow are utilized to generate the seed aerodynamics database used to train surrogate models. The wind responses of the selected shapes are further verified through accurate 3D LES simulation of an ABL flow (i.e. 3D turbulent flow) interacting with the study building.

The paper is organized in four sections. Section 1 (this section) presents an introduction and literature review on building aerodynamic mitigations. In Section 2, a description for the developed AOP is provided. Section 3 presents two optimization application examples focusing on minimizing drag and lift, respectively. Section 4 presents results and discussions of the optimization examples, and verification for the near optimal solutions using ABL flow based wind response.

Section snippets

Aerodynamic Optimization Procedure (AOP)

The AOP can be adopted for examining various types of mitigations, including corner rounding, chamfering, slotting, building twisting, tapering, etc. It is to be noted that the building shape usually bounded by other architectural and structural design considerations in addition to improving the aerodynamic performance. Thus, the proper selection of the design variables and their upper and lower bounds (constraints) will ensure that the optimal shape will satisfy other architectural and

Aerodynamic optimization application examples

The efficacy of the proposed aerodynamic optimization procedure is examined through two examples. Example 1 aims at finding a cross-section that minimizes the drag forces, while Example 2 aims at finding a cross-section that minimizes the across-wind vibration (or load). Thus, the objective functions are set to be the mean drag coefficient (CD) and the standard deviation of the lift coefficient (CL) in Examples 1 and 2, respectively. For each combination of design variables (candidate), the

Optimization results and discussions

The optimization procedure is conducted for the two optimization examples until the optimal solutions are obtained after 40 generations. Fig. 14 shows the fitness curves for the optimization examples where the objective function value of the best fitness candidate in each generation is plotted versus the generation number. This figure illustrates the improvement of the aerodynamic properties (objective functions) over optimization generations. For optimization Example 1, the mean drag

Conclusions

The current study introduces a robust aerodynamic optimization procedure that combines Genetic Algorithm, Large Eddy Simulation and Artificial Neural Network models. During the optimization procedure, ANN model is used to evaluate the objective function once trained with the aerodynamic data generated through 3D LES analyses of a 2D flow. Two optimization examples are presented to demonstrate the proposed optimization procedure aiming at reducing the drag and lift forces, respectively. A final

Acknowledgment

The authors would like to acknowledge the financial support from the National Research Council of Canada (NSERC), Ontario Center of Excellence, Canada Research Chair (for the second author) and Queen Elizabeth II Scholarship (to the first author). The authors are grateful for access to SHARCNET (a high performance computing facility) and support received from their excellent technical support team.

References (54)

  • K. Nozawa et al.

    Large eddy simulation of the flow around a low-rise building immersed in a rough-wall turbulent boundary layer

    J Wind Eng Ind Aerodyn

    (2002)
  • H. Aboshosha et al.

    LES of wind in the built-environment: inflow and roughness induced by fractal surfaces

    Sustain Cities Soc

    (2015)
  • A. Elshaer et al.

    LES evaluation of wind-induced responses for an isolated and a surrounded tall building

    Eng Struct

    (2016)
  • H. Aboshosha et al.

    Consistent inflow turbulence generator for LES evaluation of wind-induced responses for tall buildings

    J Wind Eng Ind Aerodyn

    (2015)
  • S. Murakami

    Overview of turbulence models applied in CWE-1997

    J Wind Eng Ind Aerodyn

    (1998)
  • Y. Tominaga et al.

    AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings

    J Wind Eng Ind Aerodyn

    (2008)
  • X. Chen et al.

    Evaluation of peak resultant response for wind-excited tall buildings

    Eng Struct

    (2009)
  • W. Cui et al.

    Simulation and analysis of intervention costs due to wind-induced damage on tall buildings

    Eng Struct

    (2015)
  • B.J. Vickery

    Fluctuating lift and drag on a long cylinder of square cross-section in a smooth and in a turbulent stream

    J Fluid Mech

    (1966)
  • B.E. Lee

    The effect of turbulence on the surface pressure field of a square prism

    J Fluid Mech

    (1975)
  • A. Okajima

    Strouhal numbers of rectangular cylinders

    J Fluid Mech

    (1982)
  • T. Igarashi

    Characteristics of the flow around a square prism

    Bull JSME

    (1984)
  • Y. Nakamura et al.

    The effects of turbulence on the mean flow past two-dimensional rectangular cylinders

    J Fluid Mech

    (1984)
  • R. Merrick et al.

    Shape effects on the wind-induced response of high-rise buildings

    J Wind Eng

    (2009)
  • A. Kareem et al.

    Mitigation of motions of tall buildings with specific examples of recent applications

    Wind Struct

    (1999)
  • K.C.S. Kwok et al.

    Aerodynamic devices for tall building and structures

    J Eng Mech ASCE

    (1987)
  • A. Elshaer et al.

    Wind load reductions due to building corner modifications

  • Cited by (138)

    View all citing articles on Scopus
    View full text