-~' '';;

,
', ' :' ' '

!'r. ,
,, , ,

'C:~

PRACTICAL FORMULAS FOR ESTIMATION OF CABLE TENSION BY VmRATION METROD By Hiroshi Zui; Tohru Shinke,z and Yoshio Namita3
ABSTRACT: The vibration metbod is usualIy used for field measurement of cable tension during tbe construction of cable system' bridges such as an arch bridge stiffened witb inclined cables or a cable-stayed bridge. Practical formuIs for tbe vibration metbod are proposed herein taking theeffects of flexural rigidity and sag of a cable into account. The formulas are based on the approximate solutions of high accuracy for tbe equation of inclined cable witb flexural rigidity. Cable tensions are easily estimated by tbese formulas using measured natural frequencies of l~-order modes. The practical formulas presented herein are applicable to various cables, regardless of lengtb ind tension as far as tbe vibration of first- or second-order mode is measurable. As to a very long cable tbat cannot be easily excited artificially, a formula is presented by using natural frequencies of high-order modes obtained from statinary microvibrations. The accuracy is confirmed through comparison of tbe values obtained by practical formulas witb measured values and calculated values by the finite element method.

,!,

j
'

d
'.

J
,g

;1

i
j :

1 f

i
:

INTRODUCTION

The cable forces must be adjusted during construction of cable system bridges such as the Nielsen bridge (an arch bridge stiffened with inclined cables) or a cable-stayed bridge so that cable forces and bridge geometry may be optimized. Thereforc, accurate measurement of cable forces has practical ! importance and a simple, speeciy,and rcliablc metbod of meaj surement is needed by the field engineers. The vibration 1 method by which cable forces ~estimated from measured ; natural frequencies is often used for tIle measurement of cable
,

: forces due to its simplic:ity andspeediness. N81ural frequencies

,

of a cable are inftuenced not only by cable force but also by flexural rigidity, sag-spari ratio, and inclination of tIle cable, and tIlese effects shouldbe taken into account for tIle esti~', BA$IC GOVERNING EQUATIONS ANO SOWTIONS mation of cable forces. Practical formulas for this purpose were proposed by tIle writers' in Shin1ceet al. (1980). Cable forces can be easily estimated from tIlese formulas using mea- ' Baslc G()vemlng Equatlons sured,natural frequencies of cables, and, tIlus, tIleljeformulas Fig. 1 shows an inclined catile model and its coordinates. are widely used in Japan. These formulas, however, havc a As shown in, tIle figure, tIle left support O is. the origin of certain limit of application and do not yield good results when coordinates and tIle direction OP is ta1C:en tIle x-coordinate. as tIle cable is not slender or not sufficientlytensioned.When we The direction nomalto OP is tIle y-coordinate of which tIle introduce a dimensionless parameter ~ = VTIEI'I, whereT, descending direction is taken as positive. Inaddition, 9 and 1 El, and 1 are cable force, flexural rigidity of cable,and span are tIle,angle 'Of cable inclination and cable lengtll, respeclengtll, respectively, tIle applicable range o tIle formulas is tively. specified as' 3 s Earid 10 s ~ for tIle first and second modes The following assumptions are made in the analysis: of vibration, respectively. Actually, Corexample, in case of the Nielsen bridge, tIle formulas become inapplicable for certain cables. Therefore, more reliable formulas, which can be apT plied to wide ranges of parameter ~, are needed from tIle practical viewpoint. In this paper, we pre:;ent new formulas exclusively of use for small values of ~ and reformulated tIle practical formulas to cope with tIle aforementioned problems. The new formulas presented herein are sufficiently accurate and applicableto any region of ~ so far as first- or second-order mode vibrations can be detected or artificially caused. In tIle case of a very
' ' ,

slender cable such as tIlat used at tIle uppermost pan of a cable-stayed bridge, it is not easy to excite tIle cable artificially in firsF or second-order mode vibration. In such case, stationarymici'ovibratiris are USed."n which high-frequency i modes are usually dominant. Shimada (l994)proposed an estimating metllod using high natural freque1cymodes by tIle exact soluticin o cable vibration considcring tbe effects of flexural rigidity, sag, and inclination of cable. However, in tIlis metllod, iti,s necessary to solve a nonlinear equation and use of a computer is unavoidable. In tIle present paper, a simpler formula is plcsented for tbe region ,ol.ZOO Ein high natural :s frequency uiodes,considering that ~becomes fairlylarge in case o a slendercable.

I Prof.. Cepe. Civ. Engrg.; Setsunan Univ., Ikeda-nakamachi, Neyagawa. Osaka, 572. Japan. 2Prof., Dcpt. Civ. Engrg.. Kobe Tech. Coll., Gakuen-higashimachi. Nishi-ku. Kobe. 651-21. Japan., . 2Prof.. Dept. Civ. Engrg., Setsunan Univ., Ikeda-nakamachi, Neyagawa. Osalca, 572, Japan. Note. Auoc:iate Editor: James M. Nau. Discussion open until November 1, 1996. To exteDd the closing date one monm.a Wrltten request must be ftled witb tbe ASeE Manager of Joumals. The manuscript for this paper was submitted for review and possible publicatioD on June S, 1995. This paper is part of the JDIU1Jl / Structurai E"peCI'I', Vol. 122, D No. 6. June. 1996. CASCE, ISSN 0733-944S1961OOO6-0651-06561$4.00 + $;50 per pap. Paper No. 10893.
' ,

() -sIlo .lo/2

T
,Y FlG. 1. Inclln8d9.~I.
JOURNAL OF STRUCTUAAL , .

800 ha 'F..ur..
I JUNE 1996 /651

ENGINEERING

>',

I.The sag-to-span ratio 8 ==sIlo is sufficient1ysmall (8 silo 1).. . 2. The cable vib~atesonly within.the xy-plane and its motion in the.x:-direction.isnegligibly small (v y). 3. The geometric shape of the cable is e'tpressed by a second-order parabola.

"

al

= lv'v'C. + 'Y.- ,2

(11) (12)

~l.. lv'V',. + 'Y.+ ,2

ApproxlmateFormulas

The equation of motion inthe y-direction becomes (Shimada . 1994)

El a.v(x, t).- T a2v(x, t) - h a2y + ~ a2v(x, t) = O ax. ax2 (t) ax2 g af (1)

Since (10) is a tran~ndental equation, cable forces cannot beobtained direct1yfrem measured frequencies. When the dynamic characteristic of a cable is similar to that of a string, namely, the nondimensionalparameter E.. V'TIEI'1 is large, we introduce a nondimensional parameter TI"in the frm
TI. == flf~

by vibration; w weight of. cable per unit length; and g = gravitational acceleration. In (1), v y is assumed frOIDassumption 2. The geometric shape of cable is expressed by a parabolic formula from assumption 3 4d
Y =f where d x(1

where El =. flexural rigidity of cable; v(x, t) ==deftection in the y-direction due to vibration; T cable force in the direction OP as shown in Fig. 1; h(t) ==derivative cable force caused

(13)
(14)

f~ = !!.. /Ti 21"V-;;

where f = <a>I2'11'; f~ =. theoretical values of the nth order and natural frequency of a string (Humar, 1990, pp. 689-690). By substituting these formulas into (11). and (12), al and ~l are trans. formed. into the forms

x)

(2)

= the cable sag defined in the x-y coordinates (Fig. 1). Substituting (2) into (1)
El a.v(x, t) -T a2v(x,t) + ~ a2v(x,t)

aJ=J.
~z
.

h(t) -(3) ax. ax1 g af. 12 When cable force T is small and sag becomes relatively Jarge, the effect of derivative cable force due to vibration h(t) cannot be n~glected forthe first-order mode. However, the effect of h(t) is negligibly small for second- or higher-order modes even when cable force. T is smalL The effect of h(t) is, therefore, . ignored for simplicity. Then, (3) becomes El a.v(x, t) ax.

=~

~ ~l + (~y -) J. ~ ~1+ ('"''"y + 1

=O

(15)

(16)

and by substituting these equations into (10), the following nondimensional expression is obtained:

2n'll'TJ,,(1 cos al cosh ~l) + ~ sin al sinh ~l -

(17)

- T a2v(x, t) + ~ a2v(x, t) ==O ax2 g at2

(4)
.

Though (17) is also a transcendental equation, the solution 11. for a givenvalueo Ecan be obtained by an iterative method . such as the Newton-RaphsonmethOd(Humar, 1990, pp. 342(18)-(21). Within the limit of the fint- and second-order mOde,approximate solutions of (17) can be expressed in the form

344).The initialvaluesofiterative calculationsare givenby

The accuracy and applicability of (4) is discussed later. Eq. (4) coincides with the equation of motion of a beam with axial tension T (Humar, 1990, pp. 653-656). By means of variable separation of v(x, t) = v*(x)q(t) Eq. (4) can be transformed into the form d.v*(x)
cJx.

(5) W
g

111

=f=""2.2;

(17:S~)

(18)

EI--T--w2-=0
.

d2v*(x)
cJx2

(6)

1J1

=1.075VI

d:~t) + (a)2q(t)V*(x)O =

(7)

+ e~8r;(6:S ~ :S 17) ~ Tl2 =~ - 2.2; (6O:s~)

(17:S ~ :S 60)

(19)

(20)

with introduction of parameter (a)2.Eq. (7) is an equation of free vibration of a single degree of freedom, and (a)indicates the circular frequency ofthe system. Thc: general solution of (6) is
v*(x)

112 0.985 ~ -~3.1;

(21)

=Al

sinh(~)

+ A2 cosh(h)

+ A3 sin(ax)

+ A. cos(ax)

(8) where

with

By assuming that the cable is clamped at both ends, the following equation of free vibration is obtained:
.2(al)(~I)[1

e = TI2El; and 'Y.

a2

=(,' + -y.il2 ==

,2; ~2= (t. + -y.)If1. t2 +

ww2IgE/.

(9a,b)

- cos(al)cosh(~l)] + [(~1)2 - (ali]sin(al)sinh(~l) =O

where (18) and (19) are the approximate formulas for the firslorder mOdeand (20) and (21) are fer the second-order modc. These approximateformulas are obtained in the manner mentioned in the following. 1. In case oflarge E,the solutionof (17), Tlnis nearly equal to land can be expressed as 1 +4. By substituting 1 -+:A into (17) and by using the relationsof 2n'l'l'TlnlE 1 and sl~h ~l COI: ~l 1, (18) and (20) are obtained. Eq. (21) IS cosh formed after adjusting the coefficient so as to minimizc tbt difference betwccn theoretical and approXimatesolutions. 2. In case of relatively small E, Tlnbecomes similar to tbt value of an axially tensioned beam. We get (19) by using tbt axially tensioned beam, and also by adjusting the coeffiCICII in the aforementioned way. Fig. 2 shows the comparisonbe~

~ o! ';' ~'

(10)

equivalentstaticbendingrigidityEl' = E/(1 + E 2) for.th~ 2/4'11'

where
6521 JOURNAL OF STRUCTURAL ENGINEERING/ JUNE 1996

1 '" 1.1 11/1 1.8

1.5 1A

Exact Approx.

1.3
1.2
1.1

::: ~.'m:'
1.4 1.2 1
......................

~o' FlG. 2.

10'

1rJ

10'

,1;

o 2 4

10 12 14 16 18

Exactand Approxlmate Solutlons of 11~ versus

;
FlG. 3. Exact and ApprQxlmate Solutlons of cp~ versus E

tween the exact and approximate solutions of (17). The approximate solutions coincide with the exact ones wi.thinthe When Ebecomes small, the solution of (17), 1'I~,increases very rapidly, and it becomesdifficult to obtain exact solutions. Thus, these equations ate not suitable for the region of small
similar to that of a beam and \Ve introduce another nondimen'f~=
. .

j error of 0.4%.
I

1.

l sional parameter:
i
!

'

values of E.When E is Small,the characteristics of a cable are

.

order and the second-order mode, reipectively. Eqs. (27) and (28) are obtained by using the equivalent static bending rigidity for the axially tensi.onedbeam and by adjusting the coefficient so as to malee minimum the difference between theoretical and aproxmate solutions. Fig. 3 shows the comparison between the exact and approximate solutions of (26). The approximate solutions' agree fairly well with the exact solutions within the error of 0.4% for the region O :S E:S 8 for the mstorder mode and O :S E:S 18 for the second-order mode.

tI/:

(22)
."

where t:

= the theoretical value of the nth order natural fre.

quency of a beam clamped at both ends and is given as follows

(Humar, 1990, pp. 672-674):

t: = 21flz -y-;a~ fEii

where
al

(23)

When a cable is very long, it is not easy to. excite ii artificially in the mst- or second-order mode and stationary microvibrations are necessarily used. High-frequency. modes are usualIy dominant in the stationary microvibrations and approximate solutions for high-ordet modes are practically needed. In such a case, Etakes a large value and it is regarded as 200 :S E.ConsequentIy,the.same approximate solution as (20) can be used with sufficient accuracy, that is ~ = E _E2.2; (2oo:s E)
(29)

I When the cable

By substituting these equations into (11) and (12),al and~l are transformed into

=4.730; az =7.853 forceapproaches zero (E =O),'f~ becomes 1.

E

Approxlmate Formulas for Cables wlth Large Sag When a cable has a relatively large sag, it is necessary to examine the accuracy and applicability of approximate solutions especia11yfor the first-order mode. Irvine and Caughey (1974) showed theoretical natural frequencies for an inclined cable with relatively large sag considering derivative cable forces due to vibrations as follows (flexural rigidity of cable is not taken into account):

al

= V2

~1 + e;~'f~Y-l

(24)

E ~l = V2

~1 + e;'~'f~Y + 1

(25)

Substitution of these equations into (10) yields the following nondimensional equation of free vibration:
2a~'f~(1

t: =:; -va;
where
n1f (n

(30)
tricmode (31)

- cos alcosh

~l) + EZsin al sinh ~l

=O

(26) >..=

The solution 'f~ for a given value of Ecan be obtained by an iteration such as the Newton-Raphson method. Approximate solutions of (26) can be obtained as follows:
'ft

A-tan>" ,:i { solution of f

=1,2, ...)

~orunsymme
.

= fo wl

for symmetric mode (32)

= ~1 +
.

1~;

0:s E :S 8

(27)

= 128E;A8J

COS'!!

'P2=

~ 1 + :;;

O:s E:S 18

(28)
.'

where (27) and (28) are approximate solutions for the mst..

These equations show that extension. and contraction' of a cable do not occur andthat orily cable sliape'changes, and the solutions coincide with those of a striog foe unsymmetric modes. On the other hand, the effect of extension and contraction of
JOURNAL OF STRUCTURAL ENGINEERING 1JUNE 1996/653

TABLE . Parametrlc Calculatlon8 of Natural Frequenclea 1.

I (m) (1) 10.0 10.0 10.0 10.0 10.0 10.0 20.0 30.0 8 (2) fY' fY' fY' 3fY' 45. 6fY' rf fY' EA (kN) (3) 1.58 x 1.568 x 3.136 x 1.568 x 1.568 x 1.568 x 1.568 x 1.568 x w (kNIm) (4) 0;0784 0.157 0.157 0.157 0.157 0.157. 0.157 0.157 El (kN mI) (5) 7.223 11.47 9.096 6.557 5.734 4.548 42.50 125.2

10' 10'. 10' 10' 10' 10' lOS 10'

T (kN) (6) 28.89 45.85 36.39 26.25 22.93 18.19 42.48 60.09

8 (7) 0.0034 0.0043 0;0027 0,0043 0.0060 0.0107 0.0043 0.0049

"'11

(8) 1.252 1.252 1.252 1.255 1.253 1.248 1.252 1.252

.'It (9) 1.164 1.164 1.115 1.165 1.165 1.164 1.164 1.164

cp (10) 1.113 1.113 1.113 1.116 1.114 1.110 1.113 1.113

cp; (11) 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

Note: E.. 20.0; r' .. 1.00;r .. U 75.

cp~ 3.0 FEM ~

o 5.0 .. 10.0 e 20.0 t> 50.0

2.0

. 100.0

FlRsT SYMO'TRIC 000'

..

(~) -taD}~)r2 =
(W~i)

1.0

i O

F1RST UNSYMMETRIC MODE 1.0

. 2.0

3.0

4.0
.

r = { J 128EA63V1lc:os6 B . 0.31~ 0.5 } 0.31~+ 0.5

FIO. 4. Relatlonbetween((J:and r a cable appears' significantly, and derivative cable forces due to vibrations cannot be neglected for symmetric modes. The natural frequency o symmetric mode is determined by nondimensional parameter ro,which isthe function of length, sag, weight, extensional rigidity, and inclination angle of cable. To check these characteristics for a cable with flexural rigidity, parametric calculations were carried out by the finite element method taking into account the effect of flexural rigidity of cable, and results are shown in Table 1 (Shinke et al. 1980).
In Table 1, f' =

AIso, as te the solutions of symmetric mode, 111 1910 and the region of applicability can be decided by parameter f' an4 t.
Let us, then, introduce a new parameter

r including

f' and

=f' 0.3l~ + 0.5

0.3l~ - 0.5

(33)

Transforming Irvine's equation by using (33)

-vr;; and ~ are

constant (f' = 1.0;

~=

20.0), and length, sag-to-span ratio, and inclination of angle of cable are varied. In this table, as mentioned before, 11.is the ratio of cable natural frequencies to those of string. The new parameter 19: expresses the ratio of cable natural frequencies to.those of beam with axial tension. Tab1e 1 shows the following characteristics of the solution: 1. The values of 11.and 19: are kept almost constant, independent of the values of length, sag-to-span ratio, and inclination angle of a cable so far as f' and ~ are constant. 2. For unsymmetric mode, all of 19: become 1, which means the effects of sag-to-span ratio and inclination angle of cable are negligible for the unsymmetric mode.
The approximate solutions of the unsymmetric mode, 112and 192,therefore, are applicable for cables with any sag and inclination..
.."'" I 11"'1I1I'I d' n:: <::Tl'lllrTl 'I'Idl ::""(~I"'::CI:.I'''Ir:>, 11'Me:: .""'"

(w;r)- bn (w;r) r' - { -

wl

0.31~+ 0.5

'

l28EA83 cos'e' 0.31E - 0.5 } (34)

Fig. 4 shows the relation between 19 and

expresses (34) and coincides well with calculated values for various values of~. This means that relationship between cable force and natural frequency can be well explained by (34) for
various values of

r.

A solid line

" ~ ~

~ and r.

..

However, (34) is nonlinear and, moreover, two values of cable force give ttle same first-order (first symmetric) nanual frequency [as shown later in Fig. 6 (f < 3)], and the value of cable force is very sensitive 10the change of first-order nanual frequency in the region where the effects of cable sag and derivative cable force are large. This means that the slighl measurement error of natural frequency causes a large errorof Gableforce. On the other hand, even in this region, the effeclS of cable sag and derivative cable force are negligibly smallfO; second-order mode (first unsymmetric mode), and the valueo

.~'r
J(-

cable force is not so sensitive to the change of second-order 1. natural frequency (r < 3 in Fig. 6). "lberefore, for the region

ii,,; "..,' ...:: /,

lo;
,

j where the eff~~ of c.ablesag are large, namely, the values of

are small, lt lS desirable to use second-order mode for the sneasurement of cable forees. Practically, it is necessary to set In this paper, the applicable regon fOl the first-order sag and inclination are negligbly small even for tint-

1e results based on these practical formulas coincide with exact solutions o the theoretical formula o cable vibration with the accuracy of 0.4%. and cable forces can be calculated directly &om measured natural frequencies. "lbough it is possible, in principie, to derive practical formulas fOl high-order modes in the regon of

a limit of applicable regon of the formula fOl the first-order

'snode.

~ ::S200, Erarely

; cable
.~

snode is set as r ;;::3, because. in this region. the effects.of

order mode and it is easicr to excite a cable in tint-oIder mode than in second-order mode.
PRACTICAL FORMULAS OF CABLE FORCE First, parameters
,

==

becomes less than 200 when a cable is very long and hardly excited artificially in low-order modes. Also, it becomes Complicated to maleepractical formulas of high-order modes in the regon of E ::S200, because it is inevitable to subdivide the regionof applicationof formulas depending on the values o Eand n (number ofvibration modes). As far as 200 :S E,(41) can be applied for any values of ~
'

~ and r are calculated &om cable prop-

and n.

erties and design conditions. When r is greater than 3. vibra-

; tions of first-order mode are used based on the aforementioned : reasons. When is smaller than 3, vibrations of second-order

V.riflcatlon of Practlcal Formulas Let us vefy the accuracy of practical formulas by comparing the values obtained by practical formulas with experimental values and calculated values by' the finite element methodshown in Shinke et al. (1980)and Shimada (1994). Table 2 shows cable properties used in.verification. Experiments in Shinke et al. (1980) were carried out in the following way. Cable s~imens were placed horizontally and tensioned by a hydraulic'jack, where cable tensions were measured by a load cell. Specimens were excited artificially in the first or second mode of vibration, and natural frequencieswere measured,by accelerometers. Fig. S sOOwsthe relation between cable force T and natural frequency f in the case of short sufficiently tensioned cable (small E). In these cases, characteristics of cables are similar to those of beams. A solid line shows the values of practical formulas and the eireles show experimental values. "lbelength of cables are 3.40 m, 7.15 m, and 9.95 m. "lbe values of practical formulasagree very well with experimental values. Fig. 6 shows the relation between cable force T and natural frequencyf in the case of cables of medium length (medium E).All the cables are 31.5'm long. Numerals shown inside Fig.
TABLE 2- Cable Propertle. U88d In VerltlC8tJon I w El (m) (kNlm) (kN m2) (1) (2) (3) 0.118 23.5 7.15,9.95 0.144 34.5 3.40,31.5 301.9 1.028 380.3

mode 8!e used. 'H a cable is vcry long like,in a case of upper
side cables of a long-span cable-stayed bridge, it becomes difficult to excite it artificially in low-order modes. :aven if such a cable can be excited in low-order modes, it talces much time tO measure natural frequencies due to their long periods of vibration. Therefc;re, it is desirable to use high-frequency
'

modes of stationary microvibratiODS. By considering theseconditions. the following practical formulas for the estimation of cable forces are derived from (18)-(21) and (27)-(29): 1. In the case of using the natural frequency of first-order mode (cable with sufliciently small sag 3 :sT)
T

(fal)2 [1 -2.20

..:.

0.550 ()

2];

(17:S

E>

(35)

=~w (fa1)2 [0.865 -

11.6(fYJ;

(6 $ E $ 17)
(0$ E $ 6)

(36) (37)

=4; (fal)20.828 10.5 2]; [ (f)

'

he : 2. In the case of using the natural frequency of seeooo-order E. mode (cable with relatively large sag r :S 3) 2 1E w ,C i T = (Jil)21_"": 4.40,.- 1.10 - ; (60 $ E> (3_8)

3)'
,

.12

(f)]
2, 2

'

=~ (IzI)21.03- 6.33 - 1.58 ] j [ X ~)

T

(17

E :S
17)

60) (39)

800

r
4) Re :
j

=~ (Izli [0.882- 85.0~) 2]; (O$ ~$

4w

(40)

T
]N

700 600 500 400 300 200 100

3. In the case of using the natural frequencies of high-order : modes (very long cable 2 :S n)

'or .Ie I
Ot

where f., Iz, and f,. are nieasured natural frequencies of firsti order. second-order, ,and nth-order modes 'respectively; C defined in x-y coordinates (=sIlo);and 8 = inclination angle of cable (see Fif' ,1). (37) is obtained in the following way. From (27), ~ 42(cp 1) and by substituting E2 = TIEl

T = g (/.1)2 l' - 2.20

;r

T ); .
'

nC

(200

E)

(41)

ol V(Elg)/(wJ4); r=
'81
al al Id ht al

(0.3IE- 0.5)];EA =extensional rigidity;8

V(wl)/(128EA83 cosJ8)[(0.3IE 0.5)/ +

=sag-to-span

ratio

= ~ -

T =42~1
)

(192

1)

(42)

o
,O
quency (Small
,lnllRNAI

I 1- Measured
10 20 . 30 l. Hz
'

'Substituting (22) and (23) nto (42), we obtain (37). We get other equations inthe same manner.
"

FIG. 5. Relatlon betw8en Cable Force T and Nllturat Fr.

ti
n: C:::TRllrTIIRAI r::"I~I",r::r::P''''~

I "I"'C

100<: I ;c:c:

350

/eN 300 250 200 150 100 50 O O

FIG. 6. Relatlon betwaen quency f(Medlum~)

good agreement with calculatedvalues for any order modes. NatUral frequencies of low-:order mode are small and it takes much time 10 measure natural ftequencies of low-order mode. Also, slight errors of measurement affect the accuracy of cable forcessignificantly in low-order modes. Hence, it is practically desirable to use high-order modes for the purpose of decreasing. the measurement time and increasing the accuracy of estimated cable forces.

CONCLUSIONS

.1 t

I
2 3

3 l.Sm

MeasuredI 4
Hz Natural Fre-

f.

Cable Force T .nd

12000

kN10000
8000 6000 4000 2000 O O 0.5 1 1.5 2 In
Natural Frequency

Practical formulas 10 estimate cable forces by the vibration method are proposed. Since these formulas are in algebraic form, cable forces are calculated directly ftom measured natural ftequencies. The formulas are applicable for any cable independent of the length and the intemal force of the cable as lar as the vibration of first- or second-order mode is measurable. As to a very long cable that cannot be easily excite<! artificially, a formula is presented by using natural ftequencies of high-order modes obtained ftom stationary microvibrations. The accuracy of the formulas are confirmed by the comparlson of the values of practical formulas with experimental values for short- and middle-Iength cables and calculated values by finite element method for very long cables. Formulas proposed herein can be usefully and conveniently applied to the adjustment work of cable forces during the erection of cable system bridges. APPENDIX l.
.

REFERENCES

It

301.9m

. FEMI
2.5 Hz

Humar, J. L. (1990). Dynamics 01 structures. Prentice-Hall, Inc., Englewood Cliffs, N.J. Irvine, H. M., and Caughey, T. K. (1974). "The linear tbeory of free vibration of a suspended cable." Proc., Royal Soc., London, England, Series A, Vol. 341. Shimada, T. (1994). "Estimating metbod of cable tension from natural frequency of high mode," Proc., JSCE, 50111-29, 163-171 (in Japanese). Shinke, T., Hironaka, K., Zui, H., and Nishimura, H. (1980). "Practical formulas for estimation of cable tension by vibration metbod." Proc., JSCE, 294, 25-'-34 (in Japanese).

APPENDIX

11. NOTATION

FIG. 7. Relatlon between Long Incllned Cable Force T and f (Large C)

The following symbols are used in this paper:

is 6 are numbers of vibration order. In the region where greater than 3, the values of practical formulas agree well with experimental ones of first-order mode. In the region where is less than 3, the effects of sag and derivative forees due to vibration become larger for first-order mode, and the differences between the values of practical formulas and experimental values become significantly large. On the other hand, the values of practical formulas agree well with experimental values of second-order mode in the whole region. Finally, Fig. 7 shows the relation between cable force T and natural frequency f in the case of long cables (large ~). The cable is 301.9 m long and has an angle of inclination of 23. In this case, natural ftequencies of high-order modes are used on the condition of using stationary microvibrations. Numerals shown inside Fig. 7 are numbers of vibration order. Circles show calculated values by the finite element method. In Fig. 7, the verification of practical formulas is made only for calculated values, since measured values are not easily available for very long cables. The values of practical formulas are in

e
EA

El
f

= V(Elg)/(wl'); = extensional rigidity of cable; = flexuralrigidity of cable;

= theoretical natural frequency
of cable; value of nth-order natural frequency of a beam clamped at both ends; theoretical values of nth-order natural frequency of a string; gravitational acceleration;

f: = theoretical
f: =
g

J. = measured natural frequencies of cable in nth-order mode; 1

= span length; = cable force; v(x, t) = deftection in y-di.rection due te vibration; w = weight of cable per unit length; 8 = sag-to-span ratio; ~ = dimensionless parameter =VT/El'l; 'T}. = nondimensional parametef = flf:; e = angle of cable inclination; and <P. = nondimensional parameter =fiJ:.
1
T

h(t)

= derivativecable force caused by vibration;

656/ JOURNAL OF STRUCTURAL

ENGINEERING

/ JUNF. 1996