PRESTRESS FORCE EFFECT ON VIBRATION

FREQUENCY OF C O N C R E T E BRIDGES
By M. Saiidi, l Member, ASCE, B. Douglas, 2 Member, ASCE, and S. Feng3

INTRODUCTION
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

Nondestructive testing methods to determine the condition of existing

bridges are used in assessing the need for preventive maintenance, repair,
or replacement of the structure. Some of these methods rely on vibration
characteristics of the structure to identify its in-situ condition. The measured
dynamic data have been used to determine the stiffness of foundation and
abutments as well as superstructure elements (Douglas and Reid 1982; Pin-
jarkar 1988). One of the unknown parameters in existing prestressed con-
crete (PC) bridges is the prestress force. A substantial difference between
the design and the actual prestress force can lead to serviceability and safety
problems. Unless the bridge is instrumented at the time of construction,
the existing prestress force cannot be directly estimated. In two bridges for
which the existence of an adequate prestress force was in doubt, the vibration
characteristics were used to determine if the bridges were cracked (Buckle
1990). No attempt was made to quantify the effect of prestress force on
stiffness.
Theoretically, the presence of an axial force in a homogeneous beam
alters its frequencies of vibration because of the "compression softening"
effect (Tse 1978). For a simply supported prismatic beam, the solution of
the dynamic equilibrium equation after the application of boundary con-
ditions leads to the following equation for the natural frequency of vibration:

2 _N + __ (1)
(On = -- /72 it/

where n= mode number; L = span length; N = axial compressive force

(positive); m = beam mass per unit length; E = modulus of elasticity; and
I = moment of inertia for the beam section.
Eq. (1) shows that an increase in the axial compression reduces the fre-
quencyand vice versa. This equation can be changed into a nondimension-
alized form
1
Z = 1 - ~-sX (2)

where Z = Y/Yo, with Y = [~o~,/(EI/mL4)] and Yo = (nlr)4; and X = [N/

(~r2eI/L2)l .
Parameter Z is an index that shows the sensitivity of the square of fie-
lprof, and Chron., Civ. Engrg. Dept., Univ. of Nevada, Reno, NV 89557.
ZProf. of Civ. Engrg., Univ. of Nevada, Reno, NV.
3Visiting Scholar, Civ. Engrg. Dept., Univ. of Nevada, Reno, NV.
Note. Discussion open until December 1, 1994. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this technical note was submitted for review and possible publication
on August 20, 1992. This technical note is part of the Journal of Structural Engi-
neering, Vol. 120, No. 7, July, 1994. 9 ISSN 0733-9445/94/0007-2233/$2.00
+ $.25 per page. Technical note No. 4625.

2233

J. Struct. Eng., 1994, 120(7): 2233-2241

 quency to the changes in an axial-load index. Parameter X is the ratio of
the axial load to the buckling load for the beam. Eq. (2) shows that the
sensitivity of the frequency to changes in axial loads decreases as the mode
number increases. Therefore, only the frequencies for the lower modes can
potentially be used to detect changes in the axiaMoad effect on stiffness.
The primary objective of the present study was to attempt to quantify
the effect of the existing prestress force on stiffness, based on the measured
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

vibration frequency of prestressed concrete bridges.

EXPERIMENTAL STUDIES
Both field and laboratory tests were conducted. The tests in the laboratory
were necessitated because the measured field data showed a trend opposite
to that predicted by theory.

Field Testing
The study was initially conducted on an actual posttensioned concrete
bridge (called the Golden Valley Bridge). This bridge was the subject of
another study with the aim of measuring prestress losses directly on the
tendons. Because the tendons were instrumented from the time of stressing,
a direct estimate of the magnitude of the actual prestress force in the bridge
was available. It was, therefore, possible to correlate the vibration fre-
quencies to the existing prestress force. The Golden Valley Bridge is a 47.2-
m- (155-ft-) long, simply supported, posttensioned concrete box girder struc-
ture. The superstructure is 13.7-m (45-ft) wide and has a skew angle of 30 ~
The cross section of the bridge is shown in Fig. 1. The tendons consist of
484 13-mm- (0.5-in.-) diameter seven-wire low-relaxation strands. Two strands
in each of girders G1 and G2 were instrumented by electrical strain gages
(Saiidi and Hutchens 1992). The measured 28-day concrete compressive
strength was 42.1 MPa (5,300 psi).
The dynamic data were collected on days 105, 202, and 455 after the
completion of stressing. Only the vertical modes of vibration were of in-
terest. The first two sets of data were collected when the bridge was closed
to traffic, whereas the third set was obtained with the bridge open to normal
traffic. Eight channels of vertical accelerometers with a low-pass filter setting
of 50 Hz were used in the tests. To excite the bridge in the first two tests,
wooden blocks were placed on the deck and a relatively heavy truck was
driven over the block to apply an impact load. The impact location was at
approximately the quarter point of the bridge along the longitudinal axis.
The accelerometers were placed near the edge of the bridge deck during

- 45'-0"
~3'-9"

I
l
7'-0"

-2'-0"
I
,--3 sp, ~ 7 ' - 6 "
FIG. 1. Cross Section of Golden Valley Bridge (1 ft = 0.305 m; 1 in. = 25.4 mm)

2234
J. Struct. Eng., 1994, 120(7): 2233-2241
 the first two tests. The bridge had been opened to traffic by the time of the
third test, for which the accelerometers were placed in one row inside one
of the exterior cells over half of the length of the superstructure. To obtain
amplitudes that were comparable to those in tests 1 and 2, the data were
collected when relatively large trucks were traveling on the bridge at normal
speed [say 80 kn/h (50 mi/hr) or higher]. More detailed information about
the instrumentation of the bridge is provided in Saiidi et al. (1992).
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

Using the fast Fourier transformation (FFT) method (Bracewell 1978),

the frequencies for the first several natural modes of vibration were extracted
for the bridge. A sample FFT result for a record measured near midspan
in test 3 is shown in Fig. 2. Because the bridge had a 30 ~ skew, the torsional
mode of vibration was excited considerably (frequency of 3.58 Hz). The
fundamental bending-mode frequency is indicated by the peak at approxi-
mately 2 Hz. The second bending-mode frequency is represented by the
third peak. These modes were identified based on the mode shapes. Fre-
quencies for modes beyond the second bending mode could not be deter-
mined well and were not of interest.
The tendon forces on the days of tests 1 through 3 are shown in Table
1. The measured first and second bending-mode frequencies are also shown.
Note that the average first-mode frequencies reduced as the prestress force
decreased. No consistent trend in the second-mode frequency can be ob-
served. The reduction in the first-mode frequency as a result of the loss of
prestress force was contrary to the prediction of (1). Because of this dis-
crepancy, a specimen was built and tested in the laboratory.

~_ .3579E i RMPLITUDE
9~ .160 .1885E 0 CH I [R1 ]
XFORM:I024 PTS.
HgNNING WINDON
t .1983E t loA~r

.7?4rE t ./
_ .080+ ~, .4422E -t/ .1424E 2
.o4o // o .2194E 2
\ ] ~ _ _ . . _ ~ 2 7 5 3 E -1
'0 I0.'0 20/0 30.'0 40.:0
Frequency, Hz.
FIG. 2. Sample Frequency Analysis Result

TABLE 1. Frequencies for Golden Valley Bridge

Measured
Measured second Calculated
first mode mode first mode Ratio of
Day of Prestress force frequency frequency frequency column 5
test [kips (kN)] (Hz) (Hz) (Hz) to column 3
(1) (2) (3) (4) (5) (6)
105 12,026 2.028 7.590 2.155 1.063
(53,492)
202 11,685 2.024 7.655 2.149 1.062
(51,975)
455 11,115 2,011 7.063 2.139 1.064
(49,440)

2235

J. Struct. Eng., 1994, 120(7): 2233-2241

 Laboratory Testing
A sketch of the specimen tested in the lab is shown in Fig. 3. The beam
was reinforced longitudinally and in transverse direction with Grade 60 bars.
The stirrups were used to facilitate the positioning of the top bars. A Grade
250 seven-wire straight concentric strand was used as the prestressing steel.
The strand was placed in a 25-mm- (1-in.-) diameter duct that remained
ungrouted. The concrete was made with type II portland cement and had
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

a maximum aggregates size of 12.7 mm (0.5 in.). The 28-day compressive

strength of concrete was 20.3 MPa (2,950 psi). The specimen developed a
small crack at midspan under its own weight during handling:
Both free-vibration and static tests were conducted with different axial
forces. For the free-vibration tests, the specimen was instrumented with
seven equally spaced accelerometers that measured the vertical acceleration
response of the beam. In the static tests, a mechanical gage was mounted
at the center of the beam to measure vertical deflections. The specimen was
set into free vibration by vertical impact from a hammer. Each dynamic test
was carried out after the desired prestress force had been applied and the
cable had been anchored. The jack was disconnected from the beam in
order to avoid the influence of the jack weight on the response of the beam.
The static load was applied by slowly placing metal disks of known weights
on the beam. The prestress force was varied from zero to a relatively large
level. The maximum axial force was 131.3 kN (29,510 lb), which correspotids
to a compressive stress ratio of N / ( f ' c A ) = 0.5, where N, f ' , and A are the
axial force, the concrete compressive strength, and the gross area of the
beam section, respectively. The estimated axial buckling load for the beam
was 174.8 kN (39,300 lb), which was based on the gross section moment of
inertia. The maximum applied load was an upper bound of the expected
prestress force. The relatively large maximum prestress force allowed study
of the beam for a wide range of axial loads.
Four sets of free-vibration data were collected for each axial force, two
with impact applied at midspan and the other two with impact at the quarter
point. The fast Fourier transformation technique (FFT) was used for the
analysis of the frequencies and modal amplitudes. Fig. 4 shows a typical
result of the FFT analysis. A review of mode shapes showed that the two
peaks in the figure represent the frequencies of the first and second bending
modes.

14in- I

0.5 in. Strand in.

/}2 @ 12 i n . J i l l
Beam Section Prestress
f Cable

1B ft.

oad Cell Stressing--/

Jack
FiG. 3. Details of Prestressed Concrete Beam (1 ft = 0.305 m; 1 in. = 25.4 ram)

2236

J. Struct. Eng., 1994, 120(7): 2233-2241

 ' ~~ .1507E 2 flHPLITUDE
9 300-- .3461E 0 CH 7 [BCH43
XFOBH:I024 PTS.
HRNNING HINOOW
200-
"d
"~ . tO0- 94502E 2
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

_JL. ,.L -'

20.'0 40,'0 SO?O 8070
Frequency, Hz
FIG. 4. Typical FFT Result for Beam with Impact at Quarter Point

o 400 -- P=28,4 (I.los)

v~

~ ""~e~-- P=55.5 ( t b s )

200 ' 0

! I !
*p-,~

~ 0 I0 s 30 40
Axial Force, kips
FIG. 5. Variation of Center-Span Deflection with Prestress Force (1 in. = 25.4
mm; 1 kip = 4,448 kN)

The static vertical displacement of the beam for different prestress levels
was measured to determine how the effective stiffness of the beam is affected
by different axial forces. The beam was loaded by a nearly concentrated
load at midspan. Two relatively small load values were used in the tests:
126 N (28.4 lb) and 247 N (55.5 lb). The variation of displacement as a
function of the axial load is displayed in Fig. 5, which shows that the beam
stiffened as the axial load increased. The data indicate a slight degree of
nonlinearity in the response, which is expected of concrete elements even
at small loads.
Using FFT, the first two frequencies were obtained for the beam (Table
2). The frequencies listed in the table are the average values of the results
measured at all seven channels of accelerometers. The data in the table
clearly indicate that the first natural frequency increased as the axial force
was raised. The second-mode frequencies also generally increased, but they
were not as sensitive. The somewhat erratic trend in the second-mode fre-
quency for higher axial loads occurs because this mode is insensitive to the
closing of the crack at midspan. The center of the beam forms a stationary
node when the beam vibrates in its second mode only. Unlike the funda-
mental frequencies, the mode shapes were not affected significantly by the
prestress force. There was only a slight increase in the bulging of the fun-
damental mode shape at higher loads.
2237

J. Struct. Eng., 1994, 120(7): 2233-2241

 TABLE 2. Measured Frequencies and Effective Rigidity of Beam
Axial force L .f2 (E1)e • 10 -8
(Ib) (Hz) (Hz) (Ib-sq in.)
(1) (2) (3) (4)
0 11.41 43.99 1.021
6,010 13.47 44.89 1.549
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

12,720 14.15 45.71 1.837

18,180 14.49 45.57 2.029
26,990 14.72 45.86 2.266
29,090 14.97 46.10 2.369
29,510 15.07 45.87 2.401
29,200 14.78 45.86 2.327
20,280 14.72 46.20 2.215
19,230 14.95 46.32 2.160
16,396 14.72 46.05 2.044
12,723 13.63 45.42 1.724
8,108 12.89 44.69 1.473
3,492 12.09 44.11 1.220
Note: 1 lb = 4.448 N; 1 in. = 25.4 mm.

To attempt (1) for the beam specimen, it may seem reasonable to use
the effective moment of inertia of the beam, which is specified in ACI 318-
89. Because the maximum dead-load moment in the beam is less than the
cracking moment (by approximately 33%, based on the measured 28-day
concrete strength), the gross moment of inertia controls. Using a unit weight
of 23.6 kN/m 3 (150 pcf) and the measured concrete strength, (1) leads to a
reduction of nearly 40% in the first-mode frequency as the axial load changes
from zero to 131.3 kN (29,510 lb). In contrast, the measured data showed
a 32% increase in the first mode frequency (Table 2).

EFFECTIVE RIGIDITY
The conflict between the measured and calculated results is attributed to
the fact that the element is a concrete specimen with initial microcracks that
are primarily due to shrinkage. The axial force enhances the stiffness by
closing these cracks, The theory on which (1) is based does not account for
this effect. To reproduce the experimental results, one simple approach is
to develop a relationship for the rigidity E1 of prestressed concrete members
that is a function of the prestress force. The results of the FFT analysis were
used to determine the effective rigidity of the beam that would result in the
same frequency as the measured data when used in (1). This was done by
substituting the measured fundamental frequencies and axial loads for the
beam specimen in (1) and calculating El. The results are shown in Fig. 6.
A linear regression analysis of the data was conducted to develop an
empirical equation for the effective El. The data for prestress force of 18
kN (4 kips) or less were excluded beCause they were affected by the midspan
crack that is not normally expected in PC members and because the average
prestress force for these points was small [less than 1.4 MPa (200 psi)]. In
place of these data points, an artificial point was added at (EI)e = (EI)g
for zero prestress force. The least-square best fit line was forced to pass
through this point. The result was a line with a slope of 1.77. This value
was rounded down to the nearest 0.05. The resulting equation is

2238

J. Struct. Eng., 1994, 120(7): 2233-2241

 a5

o , /
20 O ~J
/

15
d \
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

(EI) e= (l § .75-~c) EIg

10
~J
5
9 Disregarded Points
9 Added Point
0 5 10 15 20 25 30
AXIAL FORCE, kips
FIG. 6. Variation of Rigidity in Terms of Prestress Force (1 in. = 25.4 mm; 1 kip
= 4,448 kN)

(EI)e = (1 + 1.75~) Elg (3)

This equation accounts for the gain in stiffness as a result of an increase

in the axial force. The effective E1 from this equation needs to be used in
place of E1 in (1).
To examine the effect of using the proposed expression for (EI)e on the
calculated frequencies of the Golden Valley Bridge, the fundamental fre-
quencies of the bridge were found. The effective prestress forces used in
the calculations were based on the measured strains on the tendons plus
the relaxation loss. A separate account of relaxation losses was necessary
because these losses are not associated with strain changes, and hence they
cannot be obtained from the tendon-strain data. The relaxation losses for
the days of testing were estimated using a time-step prestress-loss calculation
method that is described in detail in Saiidi and Hutchens (1992). On day
455, the relaxation loss was less than 14 MPa (2 ksi). The modulus of
elasticity for concrete was found using the relationship for normal-weight
concrete in ACI 318-89 and the measured 28-day concrete compressive
strength. Using the axial forces shown in Table 1 and the proposed effective
EI, (1) leads to the frequencies listed in column 5 in Table 1. The calculated
results are within 6% of the measured data and, as prestress forces are
reduced, the calculated frequencies decrease at about the same rate as the
measured values.

SUMMARY AND CONCLUSIONS

The present study is an attempt to determine if vibration frequencies of

prestressed concrete members can be used to establish prestress losses. Both
field and laboratory tests were carried out. A posttensioned simply sup-
ported, multicell box girder concrete bridge, the Golden Valley Bridge, and
a posttensioned beam were used. Because the actual prestresses in the bridge
and the beam were known, it was possible to determine if the changes in
2239

J. Struct. Eng., 1994, 120(7): 2233-2241

 the measured frequencies would correlate with the prestress forces. Both
elements were primarily tested under free-vibration conditions.
The theoretical prediction for homogenous members was that, as the
prestress force decreases, the frequency increases, because a reduction in
the axial compressive load should stiffen the element. The measured fre-
quencies for both the bridge and the beam showed an opposite trend. As
the prestress force decreased, apparently more microcracks opened, and
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

the members softened. Based on the measured data for the laboratory
specimen, an empirical equation was developed that accounts for the effect
of axial force on the rigidity of the element. This expression was used for
the Golden Valley Bridge and led to a reasonable estimate of the funda-
mental frequency at different prestress forces.
The proposed expression for the effective rigidity of prestressed concrete
members as a function of the axial force can be used to estimate the fre-
quency of vibration for the members. The data for the Golden Valley Bridge
showed that, for practical ranges of prestress force, the change in natural
frequencies of prestressed concrete members is very small even for the
fundamental mode of vibration. Because small changes in structural param-
eters, such as degree of fixity at the supports, mass, and section properties,
can affect the frequencies and introduce variations in the results that are
on the order of the measured changes in frequency, it does not appear
feasible to use vibration data to determine the actual prestress losses. As a
relative measure, however, the dynamic signature of the bridge may be
obtained after major prestress losses have occurred (say, after three years)
and periodic monitoring of the dynamic characteristics can be used to de-
termine if there has been a substantial loss of prestress. Such monitoring,
in combination with visual inspection of the bridge, may be useful in as-
sessing the existing condition of the bridge.

ACKNOWLEDGMENTS
The study presented in this report was funded by the National Science
Foundation (NSF) Grant No. 8900266. The support of Jack Scalzi, who was
the program manager for the grant, is gratefully acknowledged. The co-
operation of the staff of the Bridge Division at the Nevada Department of
Transportation for making the bridge available for the study is much ap-
preciated. Many thanks are due E. Hwang, E. Maragakis, and A. Ebra-
himpour for their assistance. Thanks are also due civil engineering students,
S. Abdel-Ghaffar, Y. Jiang, R. Pollard, D. Schwartz, D. Straw, and S.
Vrontinos, who participated in the tests.

APPENDIX. REFERENCES
Bracewell, R. N. (1978). The Fourier transform and its applications. McGraw-Hill
Book Co., New York, N.Y.
Buckle, I. (1990). "Evaluation of two concrete highway bridges using a modal testing
method." Proc., 4th Annu. Bridge Mgmt. Systems Workshop, Transportation Re-
search Board, Washington, D.C.
"Building code requirement for reinforced concrete." (1989). AC1318-89, American
Concrete Institute (ACI), Detroit, Mich.
Douglas, B. M., and Reid, W. H. (1982). "Dynamic tests and system identification
of bridges." J. Struct. Div., ASCE, 108(10), 2295-2312.
Pinjarkar, S. G. (1988). "An overview of current worldwide practices for non-de-
structive load testing for bridge rating and evaluation." Proc., Fifth Annu. Int.
Bridge Conf., Pittsburgh, Pa.
2240

J. Struct. Eng., 1994, 120(7): 2233-2241

 Saiidi, M., and Hutchens, E. (1992). "A study of prestress changes in a post-tensioned
bridge during the first 30 months." Rep. No. CCEER-92-3, Civ. Engrg. Dept.,
Univ. of Nevada, Reno, Nev.
Saiidi, M., Douglas, B., Feng, S., Hwang, E., and Maragakis, E. (1992). "Effects
of axial force on frequency of prestressed concrete bridges." Rep. No. CCEER-
92-4, Civ. Engrg. Dept., Univ. of Nevada, Reno, Nev.
Tse, F. S., Morse, I. E., and Hinkle, R. T. (1978). Mechanical vibrations, theory
and applications. Allyn and Bacon, Inc., Boston, Mass.
Downloaded from ascelibrary.org by Tongji University on 12/02/21. Copyright ASCE. For personal use only; all rights reserved.

2241

J. Struct. Eng., 1994, 120(7): 2233-2241