You are on page 1of 10

Journal of Wind Engineering and Industrial Aerodynamics, 36 (1990) 539-548 539

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

A Generalized Definition of Gust Factor

Giovanni Solari*

ABSTRACT

This paper offers a critical survey of the procedures at


present most used for calculating the gust factors of alongwlnd
velocity, local pressure, equivalent pressure, alongwind displace-
ment and equivalent static pressure. Although the physical pheno-
menon and the theoretical background are obviously common to all
these subjects, the study reveals substantial inhomogeneities and
noticeable incongruities of treatment, especially in the field of
boundary situations.
In the light of these considerations the Author has recently
developed a research.program aimed at formulating a unitary calcu-
lation technique leading to a generalized definition of the gust
factor. The basic hypotheses and the preliminary results of this
study are reported herein, illustrating the advantages of this
method and the aspects which still require further improvements.

INTRODUCTION

In the field of wind engineering, "gust factor" is a term


usually adopted to define the ratio between the expected maximum
and the mean value of an effect associated with the gust buffet-
ing. More precisely it is called valocity gust factor, local
pressure gust factor, equivalent pressure gust factor, displace-
ment gust factor and equivalent static pressure gust factor,
according to whether the effect taken into consideration is a wind
speed, a local pressure, an equivalent pressure, a structural
displacement or an equivalent static pressure.
Although the physical phenomenon and the theoretical back-
ground are obviously common to all these subjects, actually, dif-
ferent procedures have been developed in the technical literature
with reference to each of them. This fact gives rise to substan-
tial inhomogeneities of treatment and noticeable incongruities
between results, especially when these procedures are applied to
boundary situations. This is reflected in the standards sector
always precluding the formulation of clear and theoretically
consistent structural and methodological classification criteria.
Starting from these considerations this paper proposes a
revised and unitary definition of the gust factor. The adoption of
the equivalent wind spectrum technique (Solari,1988) makes it pos-

* Associate Professor, Istituto di Scienza delle Costruzioni,


University of Genova, Via Montallegro i, 16145 Genova, Italy.

0167-6105/90/$03.50 © 1990--Elsevier Science Publishers B.V.


540

sible to express this quantity in closed form. In the light of


this treatment the classical formulae of the velocity, local
pressure, equivalent pressure, displacement and equivalent static
pressure gust factors can be regarded and derived as simple
particular cases of this generalized solution.
The formulae proposed herein represent the first prelimi-
nary results of a research program at present still in progress
(Solari,1989a,1989b).

GUST BUFFETING A N D A L O N G W I N D RESPONSE

The structure of the wind is usually represented by assi-


gning the vectoral temporal law of the speed. In reality, in most
problems related to gust buffeting and alongwind response, the
lateral and vertical components of turbulence play a secondary
role. In this situation, the instantaneous wind velocity can be
treated as a scalar quantity and expressed through the formula:

V(M;t) = V(z) + v(S;t) (i)

V(z) = the mean wind speed at height z of point M (of coordinate


x, y, z); v(M;t) = the longitudinal component of turbulence
(directed towards y), random function of space (M) and time (t).
This latter term can be treated as a stochastic stationary
Gaussian process and characterized, in the domain of frequencies
n, by its cross-power spectral density function:

Sv(M,M' ;n) = %/Sv(z;n) Sv(z';n) Coh(M,M';n) (2)

S (z;n) = the power spectrum of the alongwind turbulence; Coh


(~,M';n) the coherence function.
Consider now a fixed bluff body (for instance a cladding
element or a rigid construction) immersed in a wind velocity
field. The instantaneous local pressure developed by the flow over
the body can be expressed as:

e(M;t) = P(M) + p(M;t) (3)

P(M), p(M,t) = respectively, the mean static pressure and the zero
mean fluctuating pressure acting over the point M of the bluff su[
face A. Using the strip and quasi static theory (Davenport,1961):
i
P(M) = ~ ~ Cp(M) V2(z) (4)

p(M;t) = p C (M) V(z) v(M;t) (5)


P
p = the air density; C_(M) = the mean pressure coefficient in M.
It is particularly meaningful to notice that, based upon Eqs.4 and
5, local pressure, as well as wind velocity, can be considered as
a stochastic stationary Gaussian process.
In the light of this treatment, the equivalent pressure
Q(M;t) is, by definition, that ~ressure having a spatial distribu-
tion directly proportional to P(M), and producing, over the bluff
body, an instantaneous resultant force equal to the one due to the
541

actual dynamic wind buffeting. It is simple to verify that Q(M;t)


is the stochastic stationary Gaussian process:

Q(M;t) = Q(M) + q(M;t) (6)


Q(M) = P(M) (7)
q(M;t) = P(M) ~(t) (8)
a(t) = the ratio between the integrals of p(M;t) and P(M) over A.
Consider finally a flexible structure subjected to the pres-
sure field defined by Eqs.3,4 and 5, and free to vibrate in the
alongwind direction. The structural behaviour can be characterized
by the temporal law of the instantaneous displacement:

Y(M;t) = Y(M) + y(M;t) (9)

Y(M), y(M;t) = the mean static displacement and the zero mean flu£
tuating displacement due to the mean static pressure (Eq.4) and to
the zero mean fluctuating pressure (Eq.5), respectively. It is re-
levant to observe that introducing the hypothesis that structural
behaviour is linear, displacement Y(M;t) can be treated, as well
as wind velocity V(M;t), local pressure P(M;t) and equivalent
pressure Q(M;t), as a stochastic stationary Gaussian process.
In the field of structural engineering applications the
analysis of the dynamic alongwind response is often further
simplified by using the modal analysis and neglecting the
contribution of second and higher vibration modes. In this way:

y(M;t) = Y(M) ~(t) (i0)

On the basis of this definition it is extremely convenient to


introduce the following concept of equivalent static pressure
W(M;t). By definition, W(M;t) is that ~ressure having a spatial
distribution directly proportional to P(M), and producing, when
statically applied over the .bluff structure, an instantaneous
displacement Y(M;t) equal to the one due to the actual dynamic
wind action. The equivalent static pressure is therefore the
stochastic stationary Gaussian process:

W(M;t) = W(M) + w(M;t) (Ii)

W(M) = P(M) (12)

w(M;t) = P(M) B(t) (13)

Eqs.ll,12 and 13 generalize Eqs.6,7 and 8.

MAXIMUM VALUES AND GUST FACTORS

Consider the stochastic stationary Gaussian process:

E(M;t) = E(M) + e(M;t) (14)


542

E(M;t) = a generic quantity related to the gust buffeting; E(M) =


the mean value of E(M;t); e(M;t) = the zero mean fluctuating part
of E(M;t); Se.(M;n) = the power spectrum of e(M;t); S~(M;n) = the
power spectrum e(M;t).
Adopting the method proposed by Davenport (1964), the
expected maximum, Emax(M), of E(M;t) is given by the formulae:

(M) = E(M) + ge(M) o (M) (15)


max e
ge(M) = "%/i in[T We(M)] + 0.5772 (16)
%/2 in[T 9 (M)]
e
1 o~(M)
We(M) - 2~ o ~ (17)
e

o~(M) = f=o Se(M;n) an (18)

o~(S) = f: S~(M;n) an (19)

g (M), v (M), o (M) = respectively, the peak factor, the expected


e e
frequency and ~he standard deviation of e(M;t); o.(M) = the
• • . e .
standard deviatlon of e(M;t); T = the duratlon of gust buffetlng.
In the context of this formulation the following quantity:

(M)
G (M) - max (20)
e fi(M)
is referred to as the gust factor of E(M;t).
Noticeable particular expressions of G (M) are obtained by
making E = V,P,Q,Y,W; these quantities are c~lled, respectively,
velocity gust factor ( G ) , local pressure gust factor ( G ) , equi-
valent pressure gust ~actor ( G ) , displacement gust factor ( G )
q
and equivalent static pressure gust factor ( G ) . They origina[e
• . W . .
from a common physlcal phenomenon, are derlved within the same
mathematical treatment, are therefore strictly correlated.
Adopting the basic hypotheses previously introduced:

Gw(M) = Gy(M) = G(M) (21)

G (M) = lim G(M) (22)


q
n->~
O

G (M) = lim G (M) (23)


P S+0 q
1 + G (M)
Gv(M) = 2p (24)

n = the fundamental frequency of the structure; S = the largest


O
side of the structure. G , G , G = quantities independent of M.
Eqs.21 to 24 show ~hat Y a c~rrect and general expression of G
(Eq.21) contains all the elements to obtain, as simple particular
cases, correct and general expressions of Gq, Gp and G v.
543

CURRENT CALCULATION PROCEDURES

A critical survey of the methods proposed in the technical


literature reveals that previously stated general concepts are
quite illusory. The current calculation criteria have in fact
been developed with the aim of estimating the gust factor of one,
or at the most two, among the stochastic processes V, P, Q, Y, W.
The application of these procedures in the limit conditions
stumnarized by Eqs.21 to 24 usually involve mathematically and
physically unacceptable solutions•
A relevant part of these incongruities originates in the
choice of the mathematical law representing the power spectrum of
the alongwind turbulence. All spectral equations at present used
have in fact a common limit. They correctly reproduce the physical
phenomenon of the energy cascade (Tennekes & Lumley,1972) in the
lower frequency range, in which S (z;n) = n o , as well as in the
inertial subrange, where S (z;n) V n- 513 . On the other hand, they
maintain this latter spectral form unchanged a~so in the high
frequency range, in which, in reality, S (z;n) = n , k ~ 7.
• V .
This circumstance has mznor importance In the analysis of the
physical phenomena directly related to S (z;n); in these cases the
use of traditional spectral equations i~ the high frequency range
gives rise to negligible errors on the safety side. This same
reasoning fails when generalized towards the physical phenomena
associated with the spectral content of ~(M;t); in these cases the
application of traditional spectral equations beyond the inertial
subrange leads to infinite spectral values, S.(z;n) = (2~n) 2
V
S (z;n), for n tending towards the infinite. This incongruity is
V
more or less implicitly present in all the analyses concerning the
maximum values and therefore the gust factors•
This problem initially arises when calculating the velocity
gust factor G (M). In this case the direct use of current spectral
equations calls for an infinite value of c.(M) (Eq.19), of v (M)
(Eq.17), of g (M) (Eq.16) and therefore ~f G (M) (Eq.20). ~his
• .V . . V
clrcumstance is usually avolded by replaclng the gust factor
G (M), as defined by Eq.20, by the quantity:
V

Vmax(M; ~ )
G (M;~) - (25)
v ~(M)

(M;~) = the expected maximum of the V(M;t) averaged over the


_max
ulme inteval T<< T. This method eliminates the incongruity related
to the spectral tail by establishing a meaningful analogy with the
measured peak values (T = the finite response time of the record-
ing equipment). Noticeable expressions of G (M;T) are reported by
Durst (1960), Deacon (1965), Brook & Spiilane (1970), Wieringa
(1973), Greenway (1980), Wood (1983).
Analogous problems arise when calculating the equivalent pre~
sure factor, G (M), in the limit situation in which the largest
side of the stru@ture, S, tends towards zero (Eq.23). Also in this
case the direct use of current spectral formulae involves infinite
values of o~(M) = o~(M) (Eq.19), of Vq(M) = ~p(M) (Eq.17), of gq
544

(M) = g_(M) (Eq.16), and therefore of G (M) = G (M) (Eq.20). This


incongruity is often avoided by applyingqtwo different methods.
The first method, frequently used by standards, attributes a
conventional value to the peak factor g (M). For example ISO
(1986) suggests g = 3; Eurocode'No. 9 (1988)qassigns g = 3.5.
The second m~thod expresses Qmax(M) as: q

Qmax(M) = ~1 p Cp(M) Q~ax(M;~ ) (26)

= an averaging time of the peak wind speed selected according to


the size of the surface of the body considered. BSI (1972) recom-
mends ~ = 3 s for structural elements such as cladding, T = 5 s
for structures with the largest vertical and horizontal dimension
less than 50 m, • = 15 s for structures with the largest dimension
greater than 50 m. Greenway (1979) suggested T = 4.5S/V.
In the light of the Eqs.22 and 23 the same incongruities as
above arise when calculating G (M) in the limit case in which n
tends to the infinite and S tends towards zero. They are usuall~
avoided by adopting three different classes of procedures.
The first method consists of assigning a conventional value
to the peak factor g (M). Vellozzi & Cohen (1968) suggested gy =
3; Vickery (1970) recommended g. =3.5.
The second class of procedures comprises the methods which
can be applied, as a rule, only to flexible structures. Davenport
(1967) neglected the quasi-static part of both o2(M) and o~(M),
assigning ~ (M) = n . Simiu (1976,1980) perfected this formula
O . 2
neglecting only the quasl-static part of o.(M); in this case:

Uy(M) = n o (27)

~ , N = quantities proportional to the quasi-static part and to


th~ resonant part of c2(M), respectively. Also Solari (1982,1983)
adopted Eq.27 giving Y~ and ~ i n closed form. It is relevant to
notice that Davenport ! s, 0 Simiu ! s and Solari I s methods fail in the
range of rigid structures, independently of the structural size.
As a third alternative procedure, ESDU (1987,1988) recently
proposed the assignation of two distinct peak factors (g, = 3.5;
g given by Eq. 16 using v = n ) to the quasi-static par£ and to
O
t~e resonant part of fluctuating displacement.
The above considerations highlight the conceptual and appli-
cative limits of the methos proposed in the technical literature.
Each of them furnishes reliable results limitedly to one specific
problem, often not clearly delimited. On the other hand they lead
to mathematically and physically unacceptable solutions when
extrapolated to boundary problems, or, even worse, outside their
own applicability range. This situation, however criticizable from
a general point of view, involves particularly unfavourable
consequencies in the standards sector, precluding the formulation
of quantitative and theoretically consistent structural and
methodological classification criteria. Furthermore, it frequently
happens that the application of two different methods to the same
frontier problem leads to quite contradictory results.
545

PROPOSED METHOD

In the light of the problems discussed in the preceding


paragraph, the Author has recently carried out a research program
aimed at formulating a unitary calculation procedure (Solari,
1989a,1989b). The method is based upon the same hypotheses
previously formulated, adopts the equivalent wind spectrum
technique (Solari,1988), generalizes the gust factor definition
given by Eq.25 limitedly to V, to the quantities P, Q, Y, W:

Emax(M; ~)
Ge(M;~) = (E = V,P,Q,Y,W) (28)
~(M)
(M;~) = the expected maximum of E(M;t), averaged over ~.
max From a conceptual and formal point of view, the use of the
proposed procedure is equivalent to keeping the classical gust
factor definition unaltered (Eqs.15 to 20), adopting a fictitious
wind turbulence characterized by an identically unitary coherence
function, Coh(M,M';n) ~ i, and by a revised power spectrum
invariable in the space and given by the formula:
nC B nC D nCzH sin2(~n~ )
Svr(n;~)=Sv(h;n) 8{.4 x }[l+y ~{ Y }-X] g{.4 } (29)
V(h) V(h) V(h) (~n~) =
8{n} = ! _ I (i - e -2n) (30)
2n=
S = the classical power spectrum of turbulence; C , C , C = re-
v x z
spectively, the lateral, longitudinal and vertical expo~entlal de-
cay coefficients; y = 2C CI/(C +CI) 2 ; C , CI = the absolute values
I . w 1 w .I
of the mean pressure coe~fzczents on the wxndward and leeward side
of the construction, respectively; B, D, H = respectively, the
width, the depth and the height of the structural surface exposed
to wind; h = the reference height of the structure (Solari,1988).
The above problem has been solved in closed form with
reference to the following spectral formula (Solari,1989a):
n Sv(z;n) 6.868 n Tv(Z)
- (31)
a2(z) [i + 10.302 n T (z)] 51s
v v
o2(z) = u~ 8(z), u, = the shear velocity; Tv(Z) = Lv(Z)/V(z) , L =
t~e integral length scale of turbulence, v
The final result of the analysis is represented by the
following expression of the gust factor (Solari,1989b):
a (h)
G : Gy(M;T) : 1 + 2 g v--l----%/~ + ~ (32)
o
V(h)
g = %/2 in(~T) + 0.5772 (33)
%/2 in(vT)

: ~w ~¥~
1 i/~z + (2~no)2~
(3~)
546

~n Svr(no; T)
_ o (35)
4~ o2(h )
V

~{0.98 ~ + 0.027 72
= ) S < 0.82 ~" (36)
O
I + 0.29 ~~
2o = ~f{0.6 B + 0.02 ~2 + 0.6 H + 0.02 H 2} ~ > 0.82 ~ (37)
1 + 0.35 B i + 0.035 H
-@1 = [ _ ~ y ] 2 0.03 S <_- 3 ~ (38)

~1 = [~-~cv] 2 0.14 S > 3 ~ (39)

0.321 C B 0.321 C H ~
~_ x , ~_ z , ~ = B + H + 4%B~i
= (~-~-~;
Tv T v (h) V(h) Tv(h) V(h) 6 (40)

Eq.32 completely satisfies Eqs. 22,23 and 24. From this point
of view it contains all the informations needed to obtain, as simp
le limit cases, reliable analytical expressions of G , G , and G .
It is relevant to notice that in the field of f~exi~le stru v-
tures ~ < < ( 2 ~ n ) z ~ and Eq.34 coincides with Eq.27. In this case
making ~ = 0 an~ T (h) = 0.321hB~S(h)/V(h) (Solari, 1989a), Eq.32
offers analogous reVults to those given by Simiu (1980) and Solari
(1982,1983). In the opposite limit case of very rigid construc-
tions, ~ m 0 and Eq.34 becomes:

v = ~l ~ o (41)

In this situation G tends towards G (Eq.22) and Eq.32 leads to


results quite similar to those given ~y Greenway (1979). Finally,
when B and H tend towards zero, Eq.41 becomes:
0.173
- (42)
~0.7 T °'3(h)
V

and G tends towards G (Eq.23). In this case Eqs.24 and 32 offer


solutions almost coincident with those obtained by Wood (1983).
Eqs.32 to 40 represent preliminary results of a still out-
standing research program. Although they already furnish reliable
approximations on the safety side, the Author maintains that
further improvements could be introduced in the final reports
(Solari, 1989a, 1989b).

CONCLUSIONS

The calculation method set out in this paper gives a general,


homogeneous and unitary picture of the problem of the gust buffet-
ing and of the alongwind response. Its formal simplicity makes it
especially adapted to fast engineering evaluations and standards
applications. It also gives the possibility of formulating, in its
context, theoretically consistent criteria of structural and
methodological classification. However, a correct use of this
547

method is often conditioned by a suitable choice of the temporal


parameter ~. This choice is obvious when the method is used to
compare the results of experimental measuring and theoretical
evaluations, while, instead, it involves attention-worthy
implications (Solari,1989b) when used in the forecast sphere. The
Author maintains that a correct assigning of parameter • cannot
avoid the development of a deeper analysis of the frequency
content of the spectral tail of the atmospheric turbulence.

REFERENCES

Brook,R.R.; and Spillane,K.T., "On the Variation of Maximum Wind


Gusts with Height," Journal of Applied Meteorology, Vol.9,
pp.72-78, Feb., 1970.

BSI, 1972: "Code of Basic Data for the Design of Buildings",


Chapter V: "Loading," Part 2: "Wind Loads," British Standards
Institution, London, UK.

Davenport,A.G., "The Application of Statistical Concepts to the


Wind Loading of Structures," Proceedings of the Institution
of Civil Engineers, Vol.19, pp.449-472, Aug., 1961.

Davenport,A.G., "Note on the Distribution of the Largest Value of


a Random Function with Application to Gust Loading," Procee-
dings of the Institution of Civil Engineers, Voi.24, pp.187-
196, Jun., 1964.

Davenport,A.G., "Gust Loading Factor," Journal of the Structural


Division, ASCE, Voi.93, No.3, pp.ll-34, Mar., 1967.

Deacon,E.L., "Wind Gust Speed: Averaging Time Relationship,"


Australian Meteorological Magazine, Vol.51, pp.ll-14, 1965.

Durst,C.D., "Wind Speeds over Short Periods of Time,"


Meteorological Magazine, Voi.89, pp.181-186, 1960.

ESDU, 1987: "Calculation Methods for Along-Wind Loading. Part 2:


Response of Line-like Structures to Atmospheric Turbulence,"
ESDU 87035, Engineering Sciences Data Unit, London, UK.

ESDU, 1988: "Calculation Methods for Along-Wind Loading. Part 3:


Response of Buildings and Plate-llke Structures to
Atmospheric Turbulence," ESDU 88019, Engineering Sciences
Data Unit, London, UK.

Eurocode No.9, 1988: "Actions on Structures," Part 8: "Wind


Loads," Final Draft for Static Action, Konig und Heunisch,
Beratende Ingenieure, Frankfurt; Lehrstuhl fur Stahlbau der
RWTH Aachen, West Germany.

Greenway,M.E., "An analytical Approach to Wind Velocity Gust


Factors," Journal of Industrial Aerod~rnamics, Vol. S, pp.61-
91, 1979.
548

Greenway,M.E., "The Effects of Finite Observation Time and Finite


Averaging Time on the R.M.S. and Extreme Wind Velocity,"
Journal of Wind Engineering and Industrial Aerodynamics,
Vol.6, pp.175-179, 1980.

ISO, 1986: "Wind Loads on Structures," Working Draft for DP 4354,


International Organization for Standardization, Berlin,
Germany.

Simiu,E., "Equivalent Static Wind Loads for Tall Building Design,"


Journal of the Structural Division, ASCE, Vol.102, No.4,
pp.719-737, Apr., 1976.

Simiu,E., "Revised Procedure for Estimating Along-Wind Response,"


Journal of the Structural Division, ASCE, Voi.106, No.i,
pp.l-10, Jan., 1980.

Solari,G., "Alongwind Response Estimation: Closed Form Solution,"


Journal of the Structural Division, ASCE, Voi.108, No.i,
pp.225-244, Jan., 1982.

Solari,G., "Analytical Estimation of the Alongwind Response of


Structures," Journal of Wind Engineering and Industrial
Aerodynamics, Vol.14, pp.467-477, 1983.

Solari,G., "Equivalent Wind Spectrum Technique: Theory and


Applications," Journal of Structural Engineering, ASCE,
Vol.l14, No.6, pp.1303-1323, Jun., 1988.

Solari,G., "Wind Buffeting and Gust Factors. Part i: Peak Velocity


and Equivalent Pressure," 1989a, in preparation.

Solari,G., "Wind Buffeting and Gust Factors. Part 2: Dynamic


Alongwind Response and Equivalent Static Pressure," 1989b, in
preparation.

Tennekes,H.; and Lumley,J.L., "A First Course in Turbulence," MIT


Press, Cambridge, MA, 1972.

Vellozzi,J.; and Cohen,E., "Gust Response Factors," Journal of the


Structural Division, ASCE, Voi.97, No.6, pp.1295-1313, 1968.

Vickery,B.J., "On the Reliability of Gust Loading Factors,"


Proceedings of the Technical Meeting Concerning Wind Loads on
Buildings and Structures, Building Sciences Series 30, Natio-
nal Bureau of Standards, Washington, DC, pp.93-I04, 1970.

Wieringa,J., "Gust Factors over Open Water and Built-up Country,"


Boundary-Layer Meteorology, Vol.3, pp.424-441, 1973.

Wood,C.J., "A Simplified Calculation Method for Gust Factors,"


Journal of Wind Engineering and Industrial Aerodynamics,
Vol.12, pp.385-387, 1983.

You might also like