design wind force coefficients for hyperbolic paraboloid free roofs

Upload: physicaldavid

Post on 03-Jun-2018

260 views

Category:

Documents


4 download

TRANSCRIPT

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    1/19

    Journal of Physical Science and Application 4 (1) (2014) 1-19

    Design Wind Force Coefficients for Hyperbolic

    Paraboloid Free Roofs

    Fumiyoshi Takeda1, Tatsuya Yoshino

    1and Yasushi Uematsu

    2

    1. Technical Research Center, R&D Division, Taiyo Kogyo Corporation 3-20, Syodai-Tajika, Hirakatashi, Osaka, Japan

    2. Department of Architecture and Building Science, Tohoku University, Sendai, Japan

    Received: August 11, 2013 / Accepted: September 04, 2013 / Published: January 15, 2014.

    Abstract:This study discusses the wind force coefficients used to design hyperbolic paraboloid free roofs, which are obtained from

    wind tunnel experiments, computational fluid dynamics, and structural analyses. Design wind force coefficients are proposed on the

    basis of the wind tunnel experiment results with rigid models. The proposed wind force coefficients are compared with the

    specifications of the Australia/New Zealand Standard from the viewpoint of load effect. In addition, the application of the proposed

    wind force coefficients to membrane structures is investigated. Moreover, the effect of the deformation of membrane structures on

    wind force is verified. As a result, it is clarified that the proposed wind force coefficients need to be improved when designing

    membrane roofs, and show future work.

    Key words:Wind force coefficients, hyperbolic paraboloid roof, membrane structure, wind tunnel experiment, computational fluid

    dynamics, structural analysis.

    Nomenclature

    CD, CL: Drag and lift coefficients, respectively

    CMx, CMy: Moment coefficients about the x and y axes,respectively

    CNW, CNL: Wind force coefficients on the windward andleeward halves of the roof, respectively

    CNW0, CNL0: Basic values of CNWand CNL, respectively

    CNW, CNL: Design wind force coefficients on the windward

    and leeward halves of the roof, respectivelyD: Drag

    F1, F2: Frame models 1 and 2 for structural analysis,respectively

    Gf: Gust effect factor

    H: Mean roof height

    Iu: Turbulence intensity

    L: Lift

    Lx: Longitudinal length scale of turbulence

    Mx,My: Aerodynamic moments about the x and y axes,

    respectivelyN: Axial force in a column

    N*: Non-dimensional axial force in a column(=N/(qHa

    2/4))

    Nmean: Mean value ofN*values

    Corresponding author: Fumiyoshi Takeda, researchengineer, research fields: building science, civil engineering.

    E-mail: [email protected].

    NW,NL: Normal wind force on the windward and leewardhalves, respectively

    N1,N2,N3: Concentrated loads on the leaf springs of the

    force balanceS: Projection area of the roof

    S1: Suspension model 1 for structural analysis

    UH: Mean wind speed at the mean roof heightH

    WD1: The wind direction range of = 0 45

    WD2: The wind direction range of = 90 45

    a: Horizontal (projection) width of the roof

    gf: Peak factor

    h: Difference in height of the roof

    qH: Reference velocity pressure at the mean roof

    height

    Greek Letters: Power law exponent for the mean velocity profile

    : Correction factor for considering oblique winds

    n:Maximum displacement at the n-thstep ofstructural analysis

    ij: Absolute value of the deference between themaximum displacements at the iandjsteps ofstructural analysis

    : Wind direction

    : Correction factor of wind force coefficients for

    membrane structures

    DAVID PUBLISHING

    D

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    2/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs2

    1. Introduction

    Hyperbolic paraboloid (HP) free roofs are widely

    used for structures providing shade and weather

    protection in public spaces such as parks, playgrounds,and shopping areas. Fig. 1 shows an example of such

    a roof. Membrane structures are often used in the

    design of these roofs [1], because they are generally

    very lightweight structures. However, they are also

    vulnerable to wind loading. In practice, such roofs

    often experience damage during windstorms.

    Therefore, their resistance to wind is one of the

    greatest concerns for structural engineers when

    designing these roofs. Moreover, the wind force

    coefficients are important parameters in the design.

    The Australia/New Zealand (AS/NZ) Standard [2]

    has specified the design wind force coefficients on

    HP-shaped free roofs. However, the range of roof

    shapes for which the wind force coefficients are

    specified is rather limited. Regarding the

    wind-induced response of HP-shaped free roofs, Pun

    and Letchford reported the analytical results of an

    HP-shaped tension membrane roof subjected to

    fluctuating wind loads [3]. However, to the best of ourknowledge few studies of wind loads on HP-shaped

    free roofs have been conducted.

    The purpose of this study is to propose the wind

    force coefficients on HP-shaped free roofs for the

    design of structural elements such as columns, posts,

    beams, cables, and membranes. The paper consists of

    six chapters. Following the Introduction, Chapter 2

    presents the roof shapes and definitions of wind force

    coefficients on the HP-shaped free roofs. Chapter 3

    explains the wind tunnel experiments using rigid

    models. The arrangement, procedure, and results of the

    wind tunnel experiments are presented in Section 3.1.

    On the basis of these results, we propose the design

    wind force coefficients in Section 3.2, assuming that

    the roof is rigid and supported by four corner columns

    (Fig. 2). The wind force coefficients are represented as

    equivalent static loads, in which the dynamic load

    effect is considered in the evaluation of the gust effect

    factor. In Section 3.3, a comparison is made between

    the proposed design wind force coefficients and

    specified values in the AS/NZ Standard. Furthermore,

    the application of the proposed design wind force

    coefficients to membrane structures is described in

    Section 3.4.

    In practical membrane structures, there are several

    roof supporting systems such as a post and guy cable

    system shown in Fig. 1. Moreover, because the

    membrane roof is not rigid but rather flexible, the roof

    deformation becomes larger than that of conventional

    roofs such as metal roofs. Therefore, when the

    proposed wind force coefficients are used for the

    design of membrane structures, the following questionsmay arise. (1) Can we apply them to membrane

    structures with other supporting systems? The load

    path may change depending on the roof supporting

    system. (2) Can we apply them to flexible roofs that

    Fig. 1 HP-shaped membrane free roofs.

    Fig. 2 HP-shaped roof supported by four corner columns.

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    3/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 3

    deform under wind loading? The roof deformation

    may change the wind forces significantly. (3) Can we

    apply them to suspension structures for which the

    boundary may also deform? Therefore, in Chapter 4,

    the above subjects are investigated by computational

    fluid dynamics (CFD) and structural analyses for three

    roof models with different structural systems.

    Finally, in Chapter 5, we discuss the wind force

    coefficients to be applied to membrane roofs on the

    basis of the results obtained in Chapter 4. Note that

    the present paper is an extended and revised version of

    our previous papers [4-6].

    2. Wind Force and Moment Coefficients on

    HP-shaped Free Roofs

    2.1 Roof Shape

    Three models (Models A-C) with different rise/span

    (or sag/span) ratios are investigated in the present

    study (Fig. 3a). The layout of the roof is a square of

    15 m 15 m, and the mean roof height (H) is 8 m

    (Fig. 3b).

    2.2 Definition of Wind Force Coefficients

    Fig. 4 shows the definition of the aerodynamic

    forces and moments acting on the roof, whereDandL

    represent drag and lift and Mx and My represent the

    moments about the x and yaxes, respectively. These

    values are normalized as follows:

    (1)

    (2)

    (3)

    (4)

    where qH represents the reference velocity pressure at

    the mean roof height H,h represents the difference in

    height of the roof (Fig. 3a),a represents the horizontal

    (projection) width of the roof, and Srepresents the wind

    force coefficients of the projection area of the roof. For

    (a)

    (b)

    Fig. 3 Roof shapes: (a) rise/span ratio (b) dimension of

    Model A

    Fig. 4 Definition ofD,L,MxandMy.

    7.5 m

    5.0 m

    2.5 m

    h= a/2

    h= a/3

    h= a/6

    a: horizontal (projection) width of the roof

    h: difference in height of the roof

    21.2 m

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    4/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs4

    simplicity, the design roofs are specified by the two

    uniformly distributed values (CNW and CNL) over the

    windward and leeward halves (Fig. 5), respectively,

    which are defined as follows:

    (5)

    / (6)

    where, NW and NL represent the normal wind forces

    (positive downward) on the windward and leeward

    halves, respectively. The coefficients CNWand CNLfor

    the wind directions = 0 and 90 can be provided by

    CLand CMyas follows [5].

    = 0:

    32 (7)

    32 (8)= 90: 32 (9) 32 (10)

    3. Design Wind Force Coefficients

    3. 1 Wind Tunnel Experiments

    3.1.1 Experimental Arrangement

    The experiments were conducted in a boundary layer

    wind tunnel with a working section 1.4 m wide, 1.0 m

    high, and 6.5 m long at the Department of Architecture

    and Building Science, Tohoku University, Japan. A

    turbulent boundary layer with a power law exponent

    of= 0.18 for the mean velocity profile was generated

    on the wind tunnel floor. The turbulence intensity Iu

    and longitudinal length scaleLxof the flow at a height

    ofz= 100 mm were 0.17 and 0.16 m, respectively. The

    wind tunnel model was made of nylon resin with ageometric scale of 1/100. The thickness of the nylon

    resin was 1 mm. Fig. 6 shows a model mounted on a

    Y-shaped force balance designed, built, and gauged for

    this experiment.

    The force balance was made of 1.2-mm-thick

    phosphor bronze to measure the liftL and aerodynamic

    momentsMx and My (Fig. 7). The aluminum column

    base was pin-jointed to the end of a leaf spring. The

    Fig. 5 Definition of CNLand CNW.

    Fig. 6 Model mounted on a force balance (Model A).

    Fig. 7 Y-shaped force balance.

    bending stress at the base of each leaf spring was

    measured by strain gauges, from which the

    concentrated load at the end of each arm was computed.

    The liftL and aerodynamic momentsMxandMyabout

    thex andy axes were computed from the concentrated

    loadsN1toN3as follows:

    (11) (12) (13)

    Low

    Low

    HighHighy

    CNL CNW

    CNW CNL NW

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    5/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 5

    The definitions ofx1,x2, andx3are shown in Fig. 7.

    Note that the computed Mx and My values include

    additional moments induced by the drag. The origin of

    thexandyaxes is not located at the center of the roof

    surface. The distance between the origin and roof

    center causes moments about the x and y axes.

    Therefore, we estimated the effects of additional

    moments on the axial forces induced in the corner

    columns supporting the roof. Our estimation indicates

    that the eccentricity of the origin overestimates the

    maximum load effect by up to approximately 11% for

    Model A, 8% for Model B and 3 % for Model C [4].

    3.1.2 Experimental Procedure

    The measurements were performed at a wind speedof UH6 m/s and at a mean roof height ofH = 80 mm.

    The design wind speed is assumed to be 31.5 m/s,

    which is a typical value of strong wind events;

    therefore, the velocity scale is approximately 1/5.25.

    The geometric scale of the models (1/100) and this

    velocity scale yield a time scale of approximately 1/19.

    The wind direction was changed from 0 to 90 with

    increments of 15. The outputs of the strain meters

    were sampled simultaneously at a rate of 200 Hz for a

    period of 32 s, which approximately corresponds to 10

    min in full scale. The measurements were repeated six

    times under the same condition. The statistics of

    aerodynamic coefficients were evaluated by applying

    the ensemble average to the results of six consequent

    runs.

    3.1.3 Experimental Results

    Fig. 8 shows the statistical values of the lift and

    moment coefficients as a function of for Model A

    (h/a = 1/2); the mean and the maximum and minimumpeak values are plotted in each figure. The lift

    coefficient is maximum (upward) when 0 and

    minimum (downward) when 90. This feature is

    related to increased wind velocity along the convex

    surface, i.e., the top surface for 0 and the bottom

    surface for 90. The magnitude of the negative

    peak value of CMxis maximum when 90, whereas

    that of CMyis maximum when 0. The values of CMx

    (a)

    (b)

    (c)

    Fig. 8 Statistics of lift and moment coefficients (Model A).

    (a) CL (b) CMxand (c) CMy.

    for 0 and those of CMyfor 90 are relatively

    small in magnitude. The variation in CMx with is

    opposite to that of CMy.

    3.1.4 Load Effects

    The axial forceN induced in each column (Fig. 9) is

    computed from the time histories of CL, CMx,and CMy,

    assuming that the roof is rigid and supported by four

    corner columns [4]. In the present study, we focus on

    -1.00-0.75-0.50-0.25

    0.000.250.500.751.00

    0 15 30 45 60 75 90

    CL

    (deg.)

    CLmean

    CLmax

    CLmin

    -0.4

    -0.3

    -0.2

    -0.1

    0

    0.1

    0.2

    0 15 30 45 60 75 90

    CMx

    (deg.)

    CMxmean CMxmax CMxmin

    -0.4

    -0.3

    -0.2

    -0.1

    0

    0.1

    0.2

    0 15 30 45 60 75 90

    CMy

    (deg.)

    CMymean CMymax CMyminCMxmaxCMxminCMxmax

    CLmean

    CLmaxCLmin

    CMxmaxCMxmin CMxmax

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    6/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs6

    the axial forces induced in the columns by the wind

    forces as the most important load effect for evaluating

    the design wind force coefficients.

    The maximum and minimum peak values of the

    non-dimensional axial force N*(= N/(qHa

    2/4)) among

    the four columns are plotted in Fig. 10. The absolute

    values of the maximum and minimum N* values

    generally decrease with h/a. The variation in maximum

    tension (positive N*) with wind direction is relatively

    small, whereas that in maximum compression

    (negative N*) is significant. The maximum

    compression for all wind directions is induced when

    90.

    3.1.5 Gust Effect FactorThe gust effect factor Gfis defined as the ratio of the

    maximum or minimum axial force to the mean value

    induced in the column. The maximum value is used

    when the mean value is positive, and the minimum

    value is used when the mean value is negative. The Gf

    values are computed to investigate the dynamic effect

    of wind turbulence on the column axial forces. Fig. 11

    shows the results for Gfplotted against the mean

    reduced axial forceN*

    mean. When the value of |N*mean| is

    small, Gf exhibits large value with a large scatter.

    However, as |N*

    mean| increases, the Gfvalues collapse

    into a narrow range around Gf= 2.0, which corresponds

    to a peak factor ofgf2.5, on the basis of quasi-steady

    assumption, i.e., Gf(1 + 2.5 0.17)2[7-9]. Agfvalue

    of approximately 2.5 is somewhat smaller than that for

    gable, troughed, and mono-sloped free roofs, which is

    approximately 3.0 [10]. This difference may be due to

    the effect of flow separation from the leading edges of

    the roof on the wind loads. The turbulence induced by

    the flow separation appears to be lower for HP roofs

    than that for the other roofs. In the structural analyses

    in Section 3.4, a Gfvalue of 2.0 is used for evaluating

    the design wind forces that provide equivalent static

    loads.

    3.2 Proposed Design Wind Force Coefficients

    The roof is divided into two areas, i.e. the windward

    and leeward halves, and the design wind force

    coefficients C*NW and C

    *NL for these halves are

    specified. The following procedure provides the design

    wind force coefficients, assuming that the roof is rigid

    and supported by four corner columns (Fig. 9).Step1: The basic values of wind force coefficients

    CNWand CNL, denoted as CNW0and CNL0, are determined

    from a combination of the lift coefficient (CL) and

    moment coefficient (CMxorCMy), which produces the

    maximum load effect when = 0 or 90 (Fig. 12).

    Fig. 9 Four corner columns supporting the roof.

    (a) (b) (c)

    Fig. 10 Non-dimensional axial forces. (a) Model A (h/a= 1/2); (b) Model B (h/a= 1/3) and (c) Model C (h/a= 1/6).

    -2.0

    -1.0

    0.0

    1.0

    0 30 60 90

    N*

    (deg.)

    Max. (Tension)Min. (Compression)

    -2.0

    -1.0

    0.0

    1.0

    0 30 60 90

    N*

    (deg.)

    Max. (Tension)Min. (Compression)

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 30 60 90

    N*

    (deg.)

    Max. (Tension)

    Min. (Compression)

    = 90

    = 0

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    7/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 7

    Fig. 11 Gust effect factor based on the load effect.

    Step2: Considering that the axial force induced in a

    column may become maximum for oblique winds, weintroduced a correction factor , which is defined as the

    ratio of the actual peak force for a wind direction range

    of = 0 45 (WD1) or = 90 45 (WD2) to that

    computed from the CNW0and CNL0values.

    Step3: The design wind force coefficients (C*NWand

    C*NL) are provided as follows:

    (14) (15)

    These coefficients provide equivalent static wind

    loads.

    Fig. 13 shows the correction factors () of Models

    A-C for the two load cases I and II, which induce

    maximum tension and compression in the columns,

    respectively, for the wind direction ranges WD1 and

    WD2. When the h/aratio is small, the value of for

    WD1 is relatively large. In addition to this case, the

    value of is approximately 1.0. Similar features are

    observed for gable, troughed, and mono-sloped

    roofs [10].

    Figs. 14a and 14b show a phase-plane representation

    of the CL-CMyrelation for Models B and C, respectively,

    (a) = 0 (b) = 90

    Fig. 12 Wind direction for basic values of C*NWand C

    *NL.

    Fig. 13 Correction factor .

    when = 0. The circles in the figure represent the

    maximum and minimum peak values of CL(CLmaxand

    CLmin) during a 10 min period in full scale. Theenvelope of the trajectory appears like an ellipse with

    an inclined axis, indicating a positive correlation

    between CL and CMy. For Model C, the CL and CMy

    values are correlated well with each other. In this case,

    the CMyvalue at the instant when CLmaxor CLminoccurs

    is nearly equal to the maximum or minimum value of

    CMy(CMymaxor CMymin). The maximum load effect may

    be given by the combination of the two peak values. On

    the other hand, for Model B, the correlation between CL

    and CMy is relatively low. The peak + peak

    combination of CL and CMy does not necessarily

    produce the maximum load effect. The maximum load

    effect may be given by a certain combination of CLand

    CMy. The envelope of the CL-CMytrajectory for Models

    A and B is approximated by a hexagon (hereafter

    Hexagon) shown in Fig. 14c. The critical condition

    producing the maximum load effect may be given by

    one of these six apexes. The CL-CMxrelation for = 90

    exhibits a similar feature. From the combination of thelift and moment coefficients CMxor CMyobtained above,

    the basic wind force coefficients CNW0 and CNL0 are

    computed by Eqs. (7) and (8) or (9) and (10) for the two

    wind directions =0 and 90, respectively. For each

    wind direction, two sets of CNW0 and CNL0 values are

    selected from the six sets corresponding to the apexes

    of the Hexagon to evaluate the design wind force

    coefficients, which induce maximum tension (Load

    0.0

    0.5

    1.0

    1.5

    2.0

    0 0.2 0.4 0.6Correctionfactor

    h/a

    Load case (W 1)

    Load Case (WD 1)

    Load Case (WD 2)

    Load Case (WD 2)

    = 0 = 90o

    Correctionfactor

    h/a

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    8/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs8

    (a)

    (b)

    (c)

    Fig. 14 Phaseplane representation of the CL-CMyrelation

    (= 0) (Hexagon): (a) model B (h/a = 1/3), (b) model C

    (h/a = 1/6) and (c) model of the envelope of CL-CMy

    trajectory.

    case ) and maximum compression (Load case II)in the

    columns.

    3.3 Comparison with the Specifications of the

    Australia/New Zealand Standard

    The plots in Figs. 15 and 16 are the estimated values

    of C*NWand C

    *NLfor WD1 and WD2, respectively. In

    addition, the specifications of the AS/NZ Standard

    (2011) are shown by the dashed lines. The Standard

    provides two values of wind force coefficients

    (expressed as positive and negative) for each of the

    windward and leeward halves; the h/a ratio is limited to

    the range 0.1-0.3. The proposed values of C*NW for

    Load cases and II are relatively close to the specified

    values of the AS/NZ Standard. On the other hand,

    regarding the leeward half, the proposed wind force

    coefficients for the two load cases are similar to each

    other and are nearly equal to one of the specified values

    of the AS/NZ Standard. These features are similar to

    (a) (b)Load case : Maximum tensionLoad case II: Maximum compression

    Fig. 15 Wind force coefficients C*NWand C

    *NL (WD1). (a)

    windward half and (b) leeward half.

    (a) Windward half (b) Leeward half

    Load case Maximum tension

    Load case IIMaximum compression

    Fig. 16 Wind force coefficientsC*NWand C

    *NL (WD2).

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    C*NW

    h/a

    Load case

    Load case

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    C*NL

    h/a

    Load case

    Load case

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    C*NW

    h/a

    Load case

    Load case

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    C*NL

    h/a

    Load case

    Load case

    CMymax

    CMymean

    CMymin

    CLmaxCLmeanCLmin

    C y

    1

    CL

    2

    3

    4 6

    5

    CL0.5

    C y

    0.00

    0.05

    0.10

    -0.5

    0.15

    CMy

    0.05

    0.00-0.5

    -0.20

    1.0 CL

    AS/NZ (positive)AS/NZ (positive)

    AS/NZ (negative)AS/NZ (negative)

    AS/NZ (positive)

    AS/NZ (negative)

    AS/NZ

    (negative)

    AS/NZ (positive)

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    9/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 9

    those observed by Uematsu et al. [10] for gable,

    troughed and mono-sloped roofs.

    The axial forces induced in the columns are

    computed using C*NWand C

    *NLand are compared with

    those predicted from the AS/NZ specifications. The

    results are shown in Fig. 17. As mentioned above, the

    AS/NZ Standard generally provides four combinations

    of the wind force coefficients on the windward and

    leeward halves. The maximum and minimum axial

    forces among the four values are shown in the figure.

    Note that these values are consistent with the present

    results for Load cases and II, respectively, despite the

    existence of a difference in the wind force coefficients,

    as shown in Figs. 15 and 16.

    3.4 Application of the Proposed Wind Force

    Coefficients to Membrane Structures

    3.4.1 Structural Analysis

    To investigate the application of the proposed design

    wind force coefficients to membrane structures,

    structural analyses were conducted using three

    analytical models for the wind directions = 0 and

    90. Structural analysis was performed for six pairs of

    wind force coefficients (Table 1), which correspond to

    the apexes of the Hexagon (Fig. 14c), including the two

    design wind force coefficients proposed above. Each

    structural model has the same rise/span ratio of h/a =

    1/2, which is the same as that for Model A.

    In practice, three structural systems are often used

    for membrane structures, i.e., frame, suspension, and

    air-supported types [1]. This analysis focuses on the

    frame and suspension types. Fig. 18 shows the

    analytical models. In Frame Model 1 (F1), the roofstructure is constructed of perimeter girders and

    binding beams, which divide the roof area into 12

    zones (Fig. 18a). The roof frame is covered with

    pre-stressed membrane. The pre-stress is 4 kN/m in

    both the warp and fill (weft) directions of the

    membrane. The warp direction is shown in Fig. 18b,

    which is the same for all models. The fill direction is

    the membrane is assumed as 12 N/m2. Frame Model 2

    (a) (b)

    Fig. 17 Reduced axial force: (a) WD1 and (b) WD2.

    Table 1 Wind force coefficients corresponding to the

    apexes of Hexagons for Model A.

    Apex= 0 (WD1) = 90 (WD2)

    CNW CNL CNW CNL

    1 -0.33 -0.36 -0.19 0.06

    2 -0.16 -0.19 0.01 0.25

    3 -0.20 -0.49 -0.45 0.32

    4 0.40 -0.34 -0.28 1.16

    5 0.18 -0.11 0.06 0.82

    6 0.20 -0.54 -0.59 0.85

    Note: Considering the correction factor due to oblique winds

    (F2) consists of perimeter girders (beams) and

    pre-stressed membranes (Fig. 18b). In these frame

    models, the connection of the beam elements is rigid

    and the roof girders are supported by four corner

    columns. On the other hand, Suspension Model (S1)

    consists of curved perimeter cables and a pre-stressed

    membrane; the roof is supported by the posts and guy

    cables at the four corners, as shown in Fig. 18c. The

    column bases of the F1 and F2 models are fixed,

    whereas the posts of the S1 model are pin-supported.

    Therefore, not only the axial forces but also the

    bending moments are induced in the columns of the F1and F2 models. On the other hand, in the S1 model, no

    bending moments are induced in the columns.

    The material of the columns, beams, posts and cables

    is steel, whereas, that of the membranes is

    polytetrafluoroethylene-coated (PTFE-coated) glass

    fiber plain-weave fabric. The projection area of the S1

    model is approximately 82% of that of the frame

    models because of the curved perimeters. The S1 model

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    N*/Gf

    h/a

    Load case

    Load case

    -1.0

    -0.5

    0.0

    0.5

    1.0

    0 0.2 0.4 0.6

    N*/Gf

    h/a

    Load case

    Load case

    AS/NZ

    (negative)

    AS/NZ (positive)AS/NZ (positive)

    AS/NZ (negative)

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    10/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs10

    (a)

    (b)

    (c)

    Fig. 18 Analytical models. (a) Frame model 1 (F1); (b)

    Frame model 2 (F2) and (c) Suspension model 1 (S1).

    perpendicular to the warp direction. The self-weight of

    is the most flexible among the three models, which

    causes largest deformation to the roof under wind

    loading. On the other hand, the F1 model is relatively

    rigid. The roof membrane slightly deforms in the

    downward direction because of the self-weight.

    Therefore, the initial shapes of these three roofs are

    slightly different from each other because of the

    difference in the supporting system.

    Structural analyses were conducted by a program

    named MAGESTIC, which is a software program

    developed by Taiyo Kogyo Corp., based on the finite

    element method using the NewtonRaphson method,

    in which geometrical nonlinearity is considered [11].

    The membrane is assumed to be orthotropic and elastic.

    Furthermore, it is assumed that the membrane can only

    carry tension; in other words, it does not resist

    compression and bending moments. As mentioned

    above, the design wind speed is 31.5 m/s, which

    provides a velocity pressure of 605 N/m2. Six pairs of

    the wind force coefficients, which are calculated fromthe CNW0and CNL0 values for each wind direction (WD1

    or WD2), are applied to the roofs to determine the

    critical conditions that provide the maximum and

    minimum load effects. The load effects are provided by

    the combination of CL and CMy (or CMx), which

    corresponds to the six apexes of the Hexagon. In the

    analysis, the stresses are calculated on the basis of the

    Building Standard Law of Japan [12] and Design

    Standard for Steel Structures published by the

    Architectural Institute of Japan [13-14].

    For the membranes and cables, the tensile stresses

    are calculated from the tensile forces. However, for the

    beams and columns, the extreme fiber stresses were

    calculated by combining the axial forces and bending

    moments. For the posts of the S1 model, the axial

    stresses were calculated from the axial forces. The

    allowable stresses and material constants (Tables 2-4)

    were also determined on the basis of the Law and

    Standards [13-15]. Moreover, the ratio of thecomputed stress to allowable stress was calculated,

    which is called the stress ratio in the present paper.

    Figs. 19 and 20 show the results of the structural

    analyses for the F1 and S1 models, respectively. In

    these figures, the maximum stress ratios for the

    members, i.e., the column, post, beam cable, and

    membrane (Mem), are shown for the six apexes of the

    Hexagon in Fig. 14c.

    Mem

    C1: P-406.4 6.4

    C2: P-406.4 6.4

    B1: P-558.8 6.4

    Mem: Membrane

    Warp direction Mem

    C1: P-558.8 6.4

    C2: P-406.4 6.4

    B1: P-318.5 6.9

    B2: P-216.3 5.8

    B3: P-165.2 3.7

    Mem: Membrane

    Mem

    C1: P-406.4 6.4

    C2: P-216.3 4.5

    Ca1: 30(7 19) StrandCa2: 42.5(1 61) Spiral

    Ca3: 14(1 19) Spiral

    Ca4: 18(1 19) Spiral

    Ca5: 45(1 91) Spiral

    Mem: Membrane

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    11/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 11

    Table 2 Beam and Post.

    Elastic modulus E= 2.05 108kN/m2

    Poissons ratio = 0.3

    Table 3 Cable.

    Elastic modulusStrand rope E= 1.37 108kN/m2

    Spiral rope E= 1.57 108kN/m2

    Table 4 Membrane (t: Thickness).

    Tensional stiffnessWarp:Ew t= 1285 kN/m

    Fill:Ef t= 861 kN/m

    Apparent Poissons ratio Warp: w= 0.85 Fill: f= 0. 57

    Shear stiffness G t= 57 kN/m

    Note: Measured by MSAJ/M-02 1995 and MSAJ/M-01 1993

    in Standards of the Membrane Structures Association of Japan.

    3.4.2 Application of the Proposed Design WindForce Coefficients

    In Section 3.2, we proposed the design wind force

    coefficients C*NW and C

    *NL, focusing on the column

    axial forces as the load effect, with an assumption that

    the roof is rigid and supported by four corner columns.

    These wind force coefficients were obtained from the

    combination of CLand CMxor CMy, which provides the

    maximum tension and compression in the columns.

    They correspond to two apexes of the Hexagon shown

    in Fig. 14(c). For example, in the case of h/a = 1/2

    (Model A), Apexes 3 and 4 provide the maximum load

    effects for WD1 (= 0 45), whereas Apexes 4 and

    6 provides the maximum load effects for WD2 (= 90

    45). However, in the case of membrane structures,

    the roof is rather flexible, and not rigid. Furthermore,

    the roof supporting system may be different from that

    assumed when discussing the design wind force

    coefficients in Section 3.2. Wind forces acting on the

    roof are first transferred to the peripheral members

    (beams or cables) via membrane tension, and

    thereafter, they are transferred to the columns or the

    post and guy cables (Fig. 18). Therefore, it is expectedthat the other load effects should be considered for

    such structures. This subject is discussed below on the

    basis of the structural analysis results shown in Figs.

    19 and 20.

    Apex 4 or 6 provides the maximum stress ratio for

    all stresses of WD2, which indicates that the proposed

    wind force coefficients are also applicable to the

    membrane roof structures. On the other hand, Apexes 3

    (a1) WD1 (a2) WD1 (b1) WD2 (b2) WD2

    Fig. 19 Stress ratio for F1.

    (a1) WD1 (a2) WD1 (b1) WD2 (b2) WD2

    Fig. 20 Stress ratio for S1.

    0.0

    0.2

    0.4

    0.6

    1 2 3 4 5 6

    Stressratio

    Apex

    Mem(Warp) Mem(Fill)C1 C2

    0.0

    0.2

    0.4

    0.6

    1 2 3 4 5 6

    Stressratio

    Apex

    B1 B2 B3

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    1 2 3 4 5 6

    Stressratio

    Apex

    Mem(Warp) Mem(Fill)C1 C2

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    1 2 3 4 5 6

    Stressratio

    Apex

    B1 B2 B3

    0.0

    0.2

    0.4

    0.6

    0.8

    1 2 3 4 5 6

    Stressratio

    Apex

    Mem(Warp) Mem(Fill)C1 C2

    0.0

    0.2

    0.4

    0.6

    0.8

    1 2 3 4 5 6

    Stressratio

    Apex

    Ca1 Ca2 Ca3Ca4 Ca5

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    1 2 3 4 5 6

    Stressratio

    Apex

    Mem(Warp) Mem(Fill)C1 C2

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    1 2 3 4 5 6

    Stressratio

    Apex

    Ca1 Ca2 Ca3Ca4 Ca5

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    12/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs12

    or 4 do not always provide the maximum stress ratio

    for WD1. For example, the maximum stress is given at

    Apex 6 for the bending stress of B3 (Fig. 18a).These

    results imply that the proposed design wind forcecoefficients may underestimate the responses when

    applied to membrane roof structures. Table 5

    summarizes the stress ratios obtained at Apexes 3, 4,

    and 6 for such stresses of the members that show

    maximum stress ratios at Apex 6. These members,

    which include all beams (B1, B2, and B3) in the F1 and

    F2 models and the peripheral cables (Ca1) in the S1

    model, are connected to the membranes, and therefore,

    the stresses involved in the members may maximize

    because of the bending moments induced by membrane

    tensions. In Table 5, the ratio of the stress ratio at Apex

    6 to the larger of those at Apexes 3 and 4 is also shown.

    The value of this ratio ranges from 1.0 to 1.1 for most

    load effects, except for the bending stress in B3 of

    Model F1, which shows a ratio of approximately 1.3.

    The reason for such a large value for B3 of Model F1

    may be due to the discontinuity in the distribution of

    the wind force coefficient at the location of the B3

    beam (center line of the roof) for WD1, i.e., thebending stress involved in this member may be affected

    by the difference in the membrane tensions from the

    windward and leeward halves. Such a situation was not

    considered in the above discussion of design wind

    force coefficients.

    The above-mentioned feature, where the maximum

    values of some load effects such as the bending stresses

    of the beams and the membrane tensions are provided

    at some apexes of the Hexagon that are different from

    those for the column axial forces, implies that not only

    the column axial forces but also the other load effectsshould be considered when discussing the design wind

    force coefficients for membrane roof structures. To

    improve the wind force coefficients, it is necessary to

    identify the most important load effect for such

    structures. This is the subject of our future study.

    4. Effect of Roof Deformation on the Wind

    Forces

    4.1 Analytical Method

    The present study proposed the design wind force

    coefficients in Section 3.2, assuming that the roof is

    rigid. In addition, we investigate the application of the

    wind force coefficients to the three membrane

    structures, i.e., the F1, F2, and S1 models in Section 3.4.

    Because the membrane roofs are flexible enough to

    deform under wind loading, roof deformation may

    affect the wind forces significantly. However, the

    question of whether the proposed design wind forcecoefficients can be applied to such membrane roofs

    when considering the deformations still remains

    unresolved. Therefore, we conducted CFD and

    structural analyses using the F1, F2 and S1 models to

    investigate the effect of roof deformation on the wind

    loads.

    First, this section shows the effectiveness of CFD

    Table 5 Stress ratios of Apexes 3, 4, and 6 (WD1).

    Member Stress ratio (Apex 3) Stress ratio (Apex 4) Stress ratio (Apex 6) Ratio (Apex 6 to 3 or 4)

    F1

    Mem (Warp) 0.34 0.25 0.34 1.02

    B2 0.29 0.50 0.53 1.06

    B3 0.44 0.42 0.58 1.32

    F2Mem (Warp) 0.35 0.25 0.35 1.01

    B1 0.42 0.46 0.47 1.03

    S1

    Mem (Warp) 0.30 0.26 0.31 1.03

    Ca1 0.63 0.63 0.63 1.01

    Ca2 0.66 0.64 0.68 1.06

    Ca4 0.32 0.30 0.32 1.03

    C1 0.60 0.60 0.64 1.02

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    13/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 13

    analysis by comparing the mean wind force

    coefficients obtained from the analysis with those

    obtained from the experiments for Models A-C. CFD

    analysis simulates the wind tunnel experiment.

    Thereafter, CFD and structural analyses were repeated

    to compare the load effects before and after roof

    deformation. In practice, the membrane roofs vibrate

    under dynamic wind loads. However, the present study

    does not consider the influences of roof vibration. Only

    the effect of deformation corresponding to the mean

    (time-averaged) wind loads is considered here.

    Fig. 21 shows the algorithm for estimating the effect

    of roof deformation on the mean wind loads. First,CFD analysis is performed on the initial roof shape

    (CFD-1), which is a three dimensional analysis with

    the Reynolds-averaged Navier-Stokes (RANS) model.

    Next, to obtain the roof deformation corresponding to

    the mean wind loads, we conducted a structural

    analysis (SA-1) using the wind force coefficients

    obtained from CFD-1. This structural analysis is

    similar to that described in Section 3.4.1. In the

    analyses, the gust effect factor was assumed to beGf=

    1.0 because the focus is on the mean wind loads. CFD

    analysis (CFD-2) was repeated on the deformed roof

    obtained from SA-1 to obtain the mean wind force

    coefficients. Subsequently, the structural analysis

    (SA-2) with the mean wind force coefficients

    obtained from CFD-2 was repeated. Thus, CFD and

    structural analyses were repeated until a convergence

    of the response obtained. The criterion for convergence

    is based on the variation in the deformed roof shape;

    i.e., the following condition is used for the criterion:

    1300 (16)

    where n and n-1 represent the maximum

    displacements at the nand n1 steps of the structural

    analyses, respectively.

    4.2 Effectiveness of CFD Analysis

    4.2.1 Outline of CFD Analysis

    In the present study, we calculate the mean wind

    force and moment coefficients CL, CMxand CMy[16].

    We used an open-source software program named

    OpenFOAM version 1.5 [17]. The computational

    domain is 1.0 m wide, 1.4 m high, and 3.0 m long, inwhich an HP-shaped roof with the same configuration

    as that used in the wind tunnel experiments was placed,

    as shown in Fig. 22a. Fig. 22b shows a numerical

    model of Model A. Fig. 23 shows the resolution of the

    model grid. The computation is based on the finite

    volume method, in which the Semi-Implicit Method for

    Pressure Linked Equations (SIMPLE) algorithm and

    the renormalization group (RNG) k-model are used.

    The boundary condition is summarized in Table 6 [18].

    The turbulence intensities Iu for the analysis were

    determined on the basis of the wind tunnel experiment

    (Fig. 24). The wind direction changed from 0 to 90 at

    increments of 15 in the same manner as in the wind

    tunnel experiment.

    Fig. 21 Algorithm for investigation on the effect of roof

    deformation due to the wind loads.

    END

    Convergence?

    Structural analysis

    (SA-n)

    CFD analysis(CFD-n)

    Structural analysis(SA-1)

    CFD analysis(CFD-1)

    START

    No

    Yes

    n = 2 or more

    Mean wind forcecoefficients

    Mean wind forcecoefficients

    Deformed roof

    Deformed roof

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    14/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs14

    (a)

    (b)

    Fig. 22 Numerical model (Model A). (a) Simulated wind

    tunnel and (b) HP-shaped model (= 0)

    (a)

    (b) (c)

    Fig. 23 Mesh resolution. (a) Side view of the simulated wind

    tunnel model (b) Enlarged side view (c) Enlarged front view.

    4.2.2 CFD Analysis Result

    Figs. 25 to 27 show the results on the mean CL, CMx,

    and CMyvalues for Models A, B, and C, respectively. In

    these figures, the experimental results were also plotted

    for comparison. In general, the CFD results agree well

    with the experimental results for CLand CMy. However,

    regarding the mean CMx values, the agreement is

    poorer, particularly for larger values, although the CFD

    Table 6 Boundary condition of simulated wind tunnel.

    Surface atXmin

    ( Inlet )

    Reference height:ZG= 0.6m

    Wind velocity at the reference height:UG= 8m/sPower law index: = 0.18

    Turbulence intensity: Experimental values(Fig. 24)

    Surface atXmax( Outlet )

    Surface pressure at outlet: 0 Pa

    Surface at Ymin,

    YmaxandZmaxFree-slip wall

    Surface atZmin No-slip wall

    HP surface No-slip wall

    Fig. 24 Turbulence intensity and non-dimensional wind

    velocity profile.

    analysis captures the general trend of the experimental

    results. Although the reason for this difference is not

    yet clear, there are two possible reasons. One is related

    to the experimental method. The force balance used for

    measuring the wind forces may affect the wind flow

    under the roof. Furthermore, the effect of the drag force,

    which is unavoidable in the measurements, may affect

    the results for the CMx values. The other reason is

    related to the numerical model used and other factors,

    such as the grid resolution around the model,

    turbulence model, and the boundary condition. Further

    investigations are necessary to improve the agreement

    between the CFD analysis and wind tunnel experiment

    results.

    Furthermore, the dynamic effect should be

    considered appropriately. Nevertheless, the results

    show that CFD analysis is useful for the present

    study.

    HeightZ(mm)

    Turbulence intensity

    Mean wind velocity

    (Experimental value)

    Mean wind velocity

    (Approx. expression)

    0 0.2 0.4 0.6 0.8 1 1.2

    0

    100

    200

    300

    400

    500

    600

    700

    IuUz/UG

    UG : Mean wind velocity at a reference heightofZG = 600 mm

    Iu, Uz/UG

    HP-shaped roof

    xy

    z

    xz

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    15/19

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    16/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs16

    Scale factor for displacement: 5 times

    (a)

    Scale factor for displacement: 5 times(b)

    Scale factor for displacement: 5 times

    (c)

    Fig. 28 Deformation of the roof along the center line from

    SA-2 for the wind direction = 0. (a) F1 model wind dir. =

    0; (b) F2 model wind dir. = 0 and (c) S1model wind dir.

    = 0.

    respectively. The mean CDvalues from CFD-1 are also

    plotted in the figures. The values for the S1 model are

    corrected for the difference in the roof area. The CFD-1

    results are consistent with the experimental results. Fig.

    31 shows the distributions of the mean wind force

    coefficients on the deformed roofs. Comparing these

    results with those in Fig. 29, we observe that the region

    of upward wind forces expands in the leeward direction

    for the F2 and S1 models. This indicates that the roof

    deformation in the upward direction causes an increase

    in the wind force on the roof.

    4.4 Effect of Roof Deformation on the Wind Force

    Coefficients

    Fig. 32 shows the comparison of the mean CD, CL,

    and CMyvalues between CFD-1 and CFD-2 for = 0.

    (a)

    (b)

    (c)

    Fig. 29 Wind force coefficients obtained from CFD-1 for

    the initial roof shape. (a) F1 model (Wind dir. = 0); (b) F2

    model (Wind dir. = 0) and (c) S1 model (Wind dir. = 0).

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    0.99

    0.67

    0.35

    0.00

    +0.29

    +0.61

    +0.93

    +1.25

    +1.57

    +1.89

    +2.21

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    0.980.57

    0.15

    +0.27

    +0.68

    +1.10

    +1.51

    +1.93

    +2.34

    +2.76

    +3.17

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    0.89

    0.62

    0.36

    0.09

    +0.18

    +0.45

    +0.71

    +0.98

    +1.25

    +1.51

    +1.78

    1/a1/60

    2/a1/53

    21/a1/429

    1/a1/76

    2/a1/67

    21/a1/600

    1/a1/143

    2/a1/135

    21/a1/2500

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    17/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 17

    (a)

    (b)

    Fig. 30 Comparison between the experiment and CFD-1

    results for the mean aerodynamic coefficients. (a) = 0 and

    (b) = 90.

    In Fig. 32, the CL and CMy values obtained from

    CFD-2 for the F2 and S1 models are approximately

    4%18% larger than those obtained from CFD-1.

    These features may have been caused by the expansion

    of the upward wind force region accompanied by a

    change in curvature of the membrane surface. A similar

    feature was observed for = 90. Fig. 33 shows the

    ratio of the maximum stress obtained from SA-2 to that

    obtained from SA-1 for = 0. The values obtained

    from SA-2 are generally larger than those from SA-1.

    The ratios for the F1 model are up to 1.08, whereas

    those for the F2 and S1 models are up to 1.13. The ratio

    for column C1 of the F2 model is the largest, whereas

    that for column C2 of the S1 model is the lowest. This

    feature appears to be related to the structural system;

    i.e., the F1 and F2 models have rigid joints and fixed

    column bases, whereas the S1 model has only pin joints

    and pinned supports. Therefore, the bending moments

    in the columns of the F1 and F2 models appear to affect

    the ratios. Regarding the ratios for Mem, the S1 model

    provides the largest value. This feature is probably due

    to the wind load on the leeward half of the S1 model

    (Fig. 31c), which is the largest among all models.

    Furthermore, this phenomenon may be related to the

    (a)

    (b)

    (c)

    Fig. 31 Wind force coefficients obtained from CFD-2 for

    the initial roof shapes. (a) F1 model (= 0); (b) F2 model (

    = 0) and (c) S1 model (= 0).

    -0.1

    0.0

    0.1

    0.2

    0.3

    0.4

    1 2 3

    C

    oefficents

    F1 F2 S1

    Exp.-CL

    EXP.-CMy

    CFD1-CD

    CFD1-CL

    CFD1-CMy

    -0.3

    -0.2

    -0.1

    0.0

    0.1

    0.2

    1 2 3

    Coefficents

    F1 F2 S1

    Exp.-CL

    EXP.-CMx

    CFD1-CD

    CFD1-CL

    CFD1-CMx

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    0.99

    0.67

    0.35

    0.00

    +0.29

    +0.61+0.93

    +1.25

    +1.57

    +1.89

    +2.21

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    0.98

    0.57

    0.15

    +0.27

    +0.68

    +1.10

    +1.51

    +1.93

    +2.34

    +2.76

    +3.17

    0.89

    0.62

    0.36

    0.09

    +0.18

    +0.45

    +0.71

    +0.98

    +1.25

    +1.51

    +1.78

    1

    2

    34

    5

    6

    7

    8

    9

    10

    EXP. - CL

    EXP. - CMy

    CFD1- CD

    CFD1- CL

    CFD1- CMy

    EXP. - CL

    EXP. - CMx

    CFD1- CD

    CFD1- CL

    CFD1- CMx

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    18/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs18

    Fig. 32 Comparison between CFD-1 and CDF-2 (= 0).

    Fig. 33 Ratio of the maximum stress obtained from SA-2 to

    that obtained from SA-1 (= 0).

    member stiffness and member arrangement. In addition,

    a similar feature was observed for = 90. From these

    results, a correction factor for the effect of roof

    deformation should be introduced in the wind force

    coefficients.

    5. Conclusions and Future Study

    The present study discussed the wind force

    coefficients for the design of HP free roofs. Three roofs

    with different rise/span (or sag/span) ratios were

    analyzed. First, the design wind force coefficients for

    structural members were proposed on the basis of the

    wind tunnel experiments with rigid models. Regarding

    the wind force coefficients on HP-shaped free roofs,

    the AS/NZ Standard provides the specifications.

    However, the range of roof shapes, for which the wind

    force coefficients are specified, is rather limited. The

    proposed design wind force coefficients were

    compared with those provided in the AS/NZ Standard.

    In addition, structural analyses were performed for

    three membrane free roofs with the same rise/span

    ratio and different supporting systems, i.e., two frame

    types and one suspension type. In structural analyses,

    the proposed design wind force coefficients were used

    to investigate their application to membrane structures.

    The results suggested that the proposed design wind

    force coefficients should be improved by appropriately

    considering the structural system and load path.

    The effect of roof deformation on the wind forces

    was evaluated by an iterative analysis between CFD

    and structural analyses because the membrane roof is

    not rigid but rather flexible. Before starting the

    iterative analysis, we showed the effectiveness of CFD

    analysis using the RANS model by comparing the

    mean wind force coefficients obtained from the

    experiment and that obtained from CFD analysis.

    Thereafter, the mean wind force coefficients for theinitial roof shape were computed by CFD analysis.

    Then, we performed a structural analysis using the

    mean wind force coefficients obtained from the CFD

    analysis to predict the deformation of the membrane

    roofs corresponding to mean wind loads. The CFD

    analysis on the deformed roofs was repeated, followed

    by the structural analysis using the computed mean

    wind force coefficients on the deformed roof. The load

    effects obtained from the second structural analyses

    were compared with those obtained from the first

    structural analysis. The results suggested that to

    improve the wind force coefficients, it is necessary to

    consider the roof deformation.

    Finally, we proposed two methods for improving

    the wind force coefficients. First, a correction factor

    for membrane structures was introduced into the wind

    force coefficients, similar to the factor introduced in

    Eqs. (14) and (15) when considering the effect of wind

    direction. The modified wind force coefficients are asfollows:

    (17)

    (18)

    where, is the correction factor for membrane

    structures.

    Second, the basic values of CNWand CNL(CNW0and

    CNL0) were themselves modified to consider the effect

    -0.1

    0.0

    0.1

    0.2

    0.3

    0.4

    1 2 3

    Meanw

    indforce

    coefficients

    F1 F2 S1

    CFD1-CD

    CFD1-CL

    CFD1-CMy

    CFD2-CD

    CFD2-CL

    CFD2-CMy

    1.0

    1.1

    1.2

    1 2 3

    Ratio(SA2/SA1)

    F1 F2 S1

    Mem(Warp)

    Mem(Fill)

    C1

    C2

    CFD1 - CDCFD1 - CL

    CFD1- CMyCFD2- CD

    CFD2- CLCFD2- CMy

    Mem (Warp)

    Mem (Fill)

    C1

    C2

  • 8/12/2019 Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs

    19/19

    Design Wind Force Coefficients for Hyperbolic Paraboloid Free Roofs 19

    of membrane structures. The modified coefficients are

    as follows:

    (19)

    (20)

    where, CNW0mand CNL0mrepresent the basic values of

    CNW and CNL and include the effect of membrane

    structures on the structural systems and roof

    deformation. In both cases, we need to consider the

    parameters of structural systems, roof stiffness, and

    roof shape, such as aspect ratio and deformation. This

    will be the subject of future study.

    References[1] K. Ishii, Membrane Structures in Japan, SPS Publishing

    Company, Tokyo, Japan, 1995, pp. 1-374.

    [2] Standards Australia Limited/Standards New Zealand,Structural design actions Part 2: Wind actions,

    Australia/New Zealand Standard, AS/NZ 1170.2 (2011).

    [3] P.K.F. Pun, C.W. Letchford, Analysis of a tensionmembrane hypar roof subjected to fluctuating wind loads,

    in: Third Asia-Pacific Symposium on Wind Engineering,

    Hong Kong, 1993.

    [4] Y. Uematsu, F. Arakatsu, S. Matsumoto, F. Takeda, Windforce coefficients for the design of a hyperbolic paraboloid

    free roof, in: Proceeding of Seventh Asia-Pacific

    Conference on Wind Engineering (APCWE-VII), Taipei,

    Taiwan, 2009, pp. 635-638.

    [5] F. Takeda, T. Yoshino, Y. Uematsu, Wind forcecoefficients for the design of a hyperbolic paraboloid free

    roof, in: Proceedings of the 13th International Conference

    on Wind Engineering, Amsterdam, The Netherlands,

    2011.

    [6] F. Takeda, T. Yoshino, Y. Uematsu, Discussion of designwind force coefficients for hyperbolic paraboloid free

    roofs, in: Seventh International Colloquium on Bluff Body

    Aerodynamics & Applications (BBAA7), Shanghai,

    China, 2012.

    [7] Y. Uematsu, E. Iizumi, T. Stathopoulos, Wind loads onfree-standing canopy roofs: Part 1. Peak wind force

    coefficients for the design of cladding, Journal of Wind

    Engineering 30 (105) (2005) 91-102.

    [8] Y. Uematsu, Y. Iizumi, T. Stathopoulos, Wind loads onfree-standing canopy roofs: Part 2. Wind force coefficients

    for the design of main force resisting systems, Journal of

    Wind Engineering 31 (107) (2006) 35-49.

    [9] Y. Uematsu, E. Iizumi, T. Stathopoulos, Wind loads onfree-standing canopy roofs: Part 3. Validity and

    application of the proposed wind force coefficients,

    Journal of Wind Engineering, 31 (109) (2006) 115-122.

    [10] Y. Uematsu, E. Iizumi, T. Stathopoulos, Wind forcecoefficients for designing free-standing canopy roofs,

    Journal of Wind Engineering and Industrial Aerodynamics

    95 (2007) 1486-1510.

    [11] K. Ishii, State-of-the-art-report on the stress deformationanalysis of membrane structures, Research Report on

    Membrane Structures 4 (1990) 69-105, The Membrane

    Structures Association of Japan, 1990. (in Japanese)

    [12] Ministry of Land, Infrastructure and Transport, PublicNotice No.666, Japan, 2002.

    [13] Architectural Institute of Japan, Design Standard for SteelStructures Based on Allowable Stress Concept, Japan,

    2005. (in Japanese)

    [14] Architectural Institute of Japan, Recommendations forDesign of Cable Structures, Japan, 1994. (in Japanese)

    [15] Membrane Structures Association of Japan, TestingMethod for In-Plane Shear Properties of Membrane

    Materials (MSAJ/M-01), Standards of the Membrane

    Structures Association of Japan, 1993.

    [16] F. Takeda, T. Yoshino, Y. Uematsu, Wind forcecoefficients for the design of a hyperbolic paraboloid free

    roof, in: Proceeding of the International Association for

    Shell and Spatial Structures (IASS) Symposium, Shanghai,

    China, 2010.

    [17] OpenFOAM, http://www.openfoam.com/.[18] Working group for CFD prediction of pedestrian wind

    environment around building, Architectural Institute of

    Japan, Guidebook for Practical Applications of CFD to

    Pedestrian Wind Environment around Buildings, 2007. (in

    Japanese)