Vibration-based method to detect prestress-loss

in beam-type bridges
G

Jeong-Tae Kim*a , Yeon-Sun Ryua , Chung-Bang Yunc

aDepartment of Ocean Engineering, Pukyong National Univ., Pusan 608-737, Korea
bDepartment of Civil Engineering, KAIST, Taejon 305-600, Korea

ABSTRACT

The loss of the prestress force is an uncertain parameter that may jeopardize the safety of PSC bridges. The prestress
force is used to control crack formation in concrete, to reduce deflections, and to add strength to prestressed members;
therefore, a substantial prestress-loss can lead to severe problems in the serviceability and safety of the PSC structures.
A vibration-based method to detect prestress-loss in beam-type PSC bridges by monitoring changes in a few natural
frequencies is presented. A SID (system identification) model is formulated to estimate changes in natural frequencies
of the PSC bridges under various prestress forces. Also, an inverse-solution algorithm is proposed to identify the
prestress-loss in the PSC bridges by using the changes in natural frequencies. The feasibility of the approach is
evaluated using PSC beams for which a few natural frequencies are available for a set of prestress-loss cases.

Keywords: Damage identification, structural safety, prestressed concrete bridge, prestress-loss, modal test, natural
frequency

1. INTRODUCTION

This paper deals with the general problem of utilizing dynamic modal properties of structures to nondestructively detect
damage in these structures. Structural damage may be defined as any deviation of a geometric or material property
which may cause undesired responses of the structure. A solution to this problem is important for at least two reasons.
Firstly, damage detection is the first step in the broader category of safety assessment. Secondly, a timely safety
assessment could produce desirable consequences such as saving of human lives, protection of property, increased
reliability and productivity, and reduction in maintenance costs. Therefore, an accurate and reliable safety assessment
capability should be ensured in timely manner to maintain the integrity of structural systems (Stubbs and Osegueda,
1990; Pandey et al., 1991; Rytter, 1993; Kim et al, 2003).
In recent years, interest in the safety of existing prestressed concrete (PSC) bridges has been increased. Prestressed
concrete is defined as concrete in which there have been introduced internal stresses of such magnitudes and
distribution that the stresses resulting from given external loading are counteracted to a desired degree. The prestress
force which is one of unknown parameters in the PSC bridges is introduced to control crack formation in concrete, to
reduce deflections, and to add strength to the prestressed members. Therefore, a substantial difference between the
desired and the in-service prestress forces can lead to severe and critical serviceability and safety problems (Saiidi, et al,
1996; Saiidi et al., 1998; Aalami, 2000; Miyamoto et al., 2000). In other words, a PSC girder is considered as
irreparable as it is seriously damaged on the condition of the prestressing strands (Civjan et al., 1995). It is known that
the loss of the prestress force in tendon occurs due to elastic shortening and bending of concrete, creep and shrinkage of
concrete, steel relaxation, anchorage take-up, and frictional loss between tendon and its surrounding materials. Also,
the loss of the prestress force unexpectedly occurs due to damage or severing of prestress strands. Therefore, it is very
important to estimate the prestress-loss by considering the fact that a prestressed concrete member should keep effective
force at each significant of loading, together with appropriate material properties for that time in the life history of the
structure.
Unless a PSC bridge is instrumented at the time of construction, the existing prestress force cannot be directly
monitored and other alternative methods should be sought. Based on previous research works, nondestructive
evaluation methods using vibration test data can be used to estimate the prestress loss in the PSC bridges on the basis of

Smart Structures and Materials 2003: Smart Systems and Nondestructive Evaluation 559
for Civil Infrastructures, Shih-Chi Liu, Editor, Proceedings of SPIE Vol. 5057 (2003)
© 2003 SPIE · 0277-786X/03/$15.00

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 the following consideration: (1) the loss of the prestress force in the structure is related to the change in structural
stiffness (Lin, 1963; Saiidi et al., 1994); (2) the loss of the prestress force changes vibration characteristics of the
structure (Saiidi et al., 1994; Miyamoto et al., 2000); and (3) the change in structural stiffness can be estimated by
monitoring changes in vibration characteristics of the structure (Cawley and Adams, 1979; Kim and Stubbs, 1995;
Stubbs and Kim, 1996; Kim and Stubbs, 2002; Kim et al., 2003). However, to date, no successful attempts have been
made to identify the relationship between the prestress-loss and the change in modal parameters (Saiidi et al., 1994;
Miyamoto et al., 2000). A need remains to develop a practical method that can identify the prestress-loss via
monitoring changes in natural frequencies. The vibration-based method will prompt the primary, online, alarm on the
loss of the prestress force before other NDT techniques are enforced to diagnosis the state of the structure, in detail.
A vibration-based method is developed to nondestructively detect the loss of prestress force in beam-type PSC
bridges using a few natural frequencies. Firstly, an analytical model is formulated to identify the relationship between
changes in natural frequencies and the loss of prestress force in a PSC beam. Secondly, an inverse-solution algorithm is
proposed to identify the prestress-loss in the PSC bridges by using the changes in natural frequencies. The feasibility of
the proposed method is evaluated using a PSC beam, tested by Saiidi et al. (1994), for which the lower two natural
frequencies are available for a set of prestress-loss cases.

2. DERIVATION OF PSC BEAM MODEL

A simply supported rectangular beam is selected to analyze the effect of the prestress force introduced by internal
tendons. First, an axial-force model that represents a beam subject to axial forces is examined. Figure 1 illustrates an
analytic model of a PSC beam for which the prestress force, N, is applied at its anchoring edges. In the model, the
Euler-Bernoulli beam under flexure motion is analyzed to include effect of compressive loads.

N x

L
y

(a) PSC Beam under Axial-Force

G

E c , I c , Ac , ρ c , L
N

(b) Euler-Bernoulli Beam Under Axial-Force

Fig. 1. Axial-Force Model of PSC Beam

Assuming the axial compression force is uniform along the length of the member and not varying with time, the
equation of the free transverse vibration becomes (Clough, 1982):
∂2  ∂2 y  ∂2 y ∂2 y (1)
 Ec I c + N + ρ A =0
∂x 2  ∂x 2  ∂x 2 ∂t 2
c c

where y is transverse displacement; Ec I c is flexure rigidity of concrete beam-section; and ρ c Ac is mass of concrete
beam per unit length. On applying the boundary conditions, Eq. (1) leads to the nth natural frequency as follows:
 nπ  E I  nπ 
4 2
N (2)
ω n2 =   c c −  
 L  ρ c Ac  L  ρ c Ac
Eq. (2) shows that an increase in the axial compression reduces the frequency and vise versa. However, this is
contradictory to the behavior of the PSC beam with straight concentric tendons since natural frequencies are reduced as

560 Proc. of SPIE Vol. 5057

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 a result of the loss of prestress force (Saiidi, 1984).
As an alternative model, a tension-strength model is examined for the PSC beam with straight concentric tendons.
Figure 2 schematizes the tension-strength model for which a tendon is initially stretched and anchored to introduce
prestressing effect. Suppose that the structure is in compression due to the prestress loads applied at the anchorage
edges. Then we can model that the structure is initially deformed in compression (e.g., up to the deformed span length
Lr ) and the inside tendon is in tension due to the constraint after elastic stretching for prestressing effect. The initial
deformation of the beam results in the reduction of span length, δL(= L − Lr ) , and the expansion in the cross-section by
Poisson effect. The governing differential equation for the beam is expressed by:
∂2  ∂2 y  ∂2 y (3)
E I + m =0
 
∂x 2 ∂x 2 ∂t 2
r r r
 
where E r I r is flexural rigidity of PSC beam section and m r is mass per unit length of PSC beam. The flexural rigidity
and the mass of the PSC beam can be computed using formula for composite beam.
Er I r = Ec I c + Es I s (4-a)
mr = ρ c Ac + ρ s As (4-b)
where Es I s is the equivalent flexural rigidity corresponding to the contribution of the tendon on the flexural resistance
and ρ s As is the mass of the tendon per unit length. The equivalent flexural rigidity, E s I s , is derived from analyzing
flexural vibration of a cable under uniform tension as shown in Fig. 3(a). The governing differential equation of the
cable under tension is given by (Clough, 1982):
∂2 y ∂2 y (5)
N +ρ A =0
∂x 2 ∂t 2
s s

G Cable under Residual Tension

x

Lr = L(1 − δL L)
y

(a) PSC Beam under Initial Deformation

G

r , I r , m r , Lr

(b) Beam of Equivalent Flexural Rigidity

Fig. 2. Tension-Strength Model of PSC Beam

Furthermore, Eq. (5) leads the nth natural frequency of the cable under tension as follows:
2
 nπ  N (6)
ω nc 2 =  

 Lr  ρ s As
On assuming that there exists a beam equivalent to the cable under tension with respect to modal properties, as shown in
Fig. 3(b), the nth natural frequency of the equivalent beam is derived as follows:
4
 nπ  Es I s (7)
ω c2
= 



ρ s As
n
 Lr 
By setting Eq. (6) equal to Eq. (7), we obtain the equivalent flexural rigidity as follows:

Proc. of SPIE Vol. 5057 561

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 2

Es I s = 
 Lr 
N
(8)

 nπ 
On substituting Eq. (8) into Eq. (4), the total flexural rigidity of the PSC beam section can be expressed as:
2
 Lr  (9)
Er I r = Ec I c +   N
 nπ 
where E r I r is assumed constant along the entire length of the beam. Applying Eq. (9) and appropriate boundary
conditions to Eq. (3) leads the nth natural frequency of the residual-tension model as follows:
4
 nπ 
2
1   L   (10)
ω n2 =  

Ec I c +  r  N
 Lr  mr   nπ  


where the deformed span length Lr can be computed by:

 δL   N  (11)
Lr = L 1 −  = L1 − 

 L   Ac E c 

ρ s , As , Lr , ω nc
N

(a) Cable under Tension Load

G

, ρ s , As , Lr , ωn
c
Es , I s

(b) Beam with Equivalent Flexural Rigidity

Fig. 3. Flexural Vibration of Cable under Tension

In case of Ac L2 << 1 , the axial compression effect is negligible if the prestress force N is less than the beam’s yielding
point. Note that the residual tension effect of the tendon is reflected on the natural frequency equation of the PSC beam
via quantifying the equivalent bending rigidity which is in terms of the beam span length and the axial force. Note also
that Eq. (10) can be directly used to compute frequency changes due to changes in prestress forces. As an inverse
solution of Eq. (10), the prestress force can be identified by:
2
 nπ 
2

(N )n = ω n2 mr  Lr 
 − E c I c  

(12)
 nπ   Lr 
where ( N )n is the identified prestress force by using nth natural frequency and structural properties. By assuming no
changes in beam’s geometry and material properties occur due to changes in the prestress force, the first variation of the
prestress force can be expressed by:
2

(δN )n = δω n2 m r  Lr 

(13)
 nπ 
where (δN )n is the change in the prestress force that can be identified by the nth mode and δω n2 is the change in ω n2 due
to the prestress-loss. By dividing Eq. (13) by Eq. (12), the relative change in the prestress force that can be identified by
using the nth mode is obtained as:

562 Proc. of SPIE Vol. 5057

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 2
 Lr 
δω n2 m r   (14)
 δN   nπ 
  = 2
 nπ 
2
 N n  Lr 
ω n2 m r   − E c I c  

 nπ   Lr 

Dividing Eq. (14) by mr and ( Lr nπ ) gives:

2

 δN  δω n2 (15)
  =
 N n ω − ϖ n2
2
n

where ϖ n2 is the nth natural frequency of the beam with zero prestress force and is given by:
4
 nπ  Ec I c (16)
ϖ n2 =  

 Lr  mr
From Eq. (15), the relative change in prestress force, δN N , can be estimated by measuring natural frequency changes,
δω n , and natural frequencies of the beam with zero prestress force, ϖ n . However, in existing real structures, ϖ n is not
available unless measured at as-built state; therefore, ϖ n should be estimated from system identification process.

3. VERIFICATION USING PSC BEAMS

3.1 Description of Test Structure

The test structure is schematized in Fig. 4. Saiidi et al. (1994) performed a laboratory experiment on the structure to
measure changes in dynamic modal properties by adjusting prestress forces applied to the test structure. The beam was
reinforced longitudinally and 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 diameter duct that remained ungrouted. The concrete was made with type II Portland
cement and had a maximum aggregate size of 12.7-mm. The 28-day compressive strength of concrete was 20.3 Mpa.
Dynamic modal tests were conducted with different axial forces. For the modal tests, the beam was instrumented with
seven equally spaced accelerometers that measured the vertical acceleration responses of the beam. The structure was
excited by vertical impact from a hammer. Each dynamic test was performed after the desired prestress force had been
applied and the cable had been anchored. The jack was disconnected from the beam to avoid the effect of the jack mass
on the beam’s dynamic responses. The prestress force was varied from zero to a relatively large level, i.e., the maximum
axial force was 131.3kN. Four sets of modal data were collected for each axial force level, two with impact applied at
midspan and the other two with impact at the quarter point. Using fast Fourier transformation technique (FFT), the first
two frequencies were obtained for each in-situ prestress-force case as listed in Table 1. The listed frequencies are the
average values of the results measured at all seven channels.

10 cm
G
G Prestress
Cable
#3

1.27 cmG 12.5 cm

#2 @ 30 cm
3.6 m

Beam Section Stressing

Load Cell
GG Jack
Fig. 4. Prestressed Concrete Beam (Saiidi et al., 1994)

Proc. of SPIE Vol. 5057 563

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 3.2 System Identification of PSC Beam
The previously formulated PSC beam models were evaluated using the PSC beam shown in Fig. 4. We identified
natural frequencies of the PSC beam, for each case listed in Table 1, by using two different models: Math model and FE
model. The Math model is equivalent to Eq. (10) from which frequencies are directly computed for known prestress
forces. The span length Lr is approximated using Eq. (11). The elastic modulus of concrete was estimated as
E c = 21.52Gpa by using the 28-day compressive strength f 28 = 20 .3Mpa , the linear mass density of concrete
ρ c = 2400kg / m 3 , and cross-sectional area of concrete Ac = 1.24 × 10 −2 m 2 . For the maximum force (i.e., N = 132.8kN ) in
Table 1, Lr is computed as 0.9995L, i.e., the compressive strain resulted due to the force is less than 5 × 10 −4 . Therefore,
we neglect the axial deformation effect provided that Lr ≈ L ( = 3.6 m) . Also, the second moment of area is calculated as
I c = 1.734 × 10 −5 m 4 . The mass per unit length are calculated using Eq. (4-b) as mr = 30.75kg / m , in which the cross-sectional
area and the linear mass density of the steel tendon are As = 1.27 × 10 −4 m 2 and ρ s = 7850kg / m 3 , respectively.
On implementing the above structural data into Eq. (10), we can rewrite a math model to predict natural frequencies
of the PSC beam as follows:
ω n2 = 5133.3 × n 4 + 0.0247n 2 N (17)
where n is mode number and N is the prestress force in Newton. Here all ten scenarios in Table 1 were investigated by
using Eq. (17). The predicted frequency results are summarized in Table 1.

Table 1. Measured Frequencies and Predicted Frequencies of PSC Beam

In-situ Experimental Predicted Frequency (Hz)
Prestress Frequency (Hz)
Case Math Model FE Model
Force (Saiidi et al., 1994)
(kN) Mode 1 Mode 2 Mode 1 Mode 2 Mode 1 Mode 2
1 0 11.41 43.99 11.409 45.635 11.196 44.047
2 15.71 12.09 44.11 11.832 46.063 11.689 44.547
3 27.05 13.47 44.89 12.128 46.371 12.032 44.886
4 36.49 12.89 44.69 12.370 46.626 12.311 45.161
5 57.25 13.63 45.62 12.885 47.179 12.902 45.744
6 81.81 14.49 45.57 13.468 47.828 13.568 46.401
7 91.26 14.72 46.32 13.686 48.073 13.816 46.645
8 121.46 14.72 45.86 14.360 48.854 14.578 47.398
9 130.91 14.97 46.10 14.565 49.096 14.809 47.622
10 132.80 15.07 45.87 14.610 49.145 14.855 47.667

The FE model of the PSC beam is schematized in Fig. 5. For analysis purpose, we divided the beam into 28,512 block
elements. The steel tendon is placed to center by 288 elements along the beam and each element size is 1cm x 1.25cm x
1.25cm. All others are concrete elements and each element size is 1.125cm x 1.125cm x 1.25cm. Material properties of
the FE model were assigned as follows: (1) for concrete elements, the elastic modulus E c = 21.52Gpa , Poisson’s ratio
υ = 0.18 , and the linear mass density ρ c = 2400kg / m 3 ; and (2) for steel tendon elements, ρ s = 7850kg / m 3 , υ = 0.3 , and
the elastic modulus E s = 0.3155 ⋅ N ⋅ n −2 ( kN / m 2 ). Among those properties, the elastic modulus E s is estimated by using
the equivalent flexural rigidity formula, Es can be estimated by:
L2 N N (18)
Es = = 0.3155 2
n π 2Is
2
n
where the second moment of area of the steel tendon element is given by I s = 4.16 × 10 −3 m 2 . Note that E s is different
for modes, and that means we have to adjust E s for each mode. For example, E s = 0.3155 ⋅ N ( kN / m 2 ) for the first
bending mode and E s = 0.0789 ⋅ N ( kN / m 2 ) for the second bending mode. The contribution of the tendon to the flexural
stiffness occur maximum in the first bending mode and decreases by geometric progression as mode number increases.
Modal parameters of the FE model were generated numerically using the commercial software ANSYS. Here all ten

564 Proc. of SPIE Vol. 5057

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 cases in Table 1 were investigated. The predicted frequencies of the first two bending modes are also summarized in
Table 1. Mode shapes of the first two bending modes are plotted in Fig. 6. The predicted frequencies using those two
models were compared with the measured frequencies from the experiment as plotted in Figs. 7(a) and 7(b). In case of
the Math model, prediction errors are between 0.1% and 3% for mode one and between 3% and 7% for mode two. In
case of the FE model, prediction errors are between 1% and 3% for mode one and between 0.1% and 4% for mode two.
In mode one, the predicted results show relatively good accuracy in both models. In mode two, the results of the FE
model shows relatively better accuracy than the Math model.

[email protected]=4.5cm 1cm [email protected]=4.5cm

[email protected]=5.625cm
12.5cm
G

1.25cm
Steel Tendon Concrete

[email protected]=5.625cm
[email protected]=3.6m 10cm
G
Fig. 5. FE Model of PSC Beam

ͣ͟ ͡

ΖΕ ͢͟ ͡
ΦΚΥΝ
Ρ
Ξ
Ͳ͑Ν ͟͡ ͡
ΒΕ
Π;͢͟͞ ͡
ͣ͟͞ ͡
͡ ͟͡ ͢ ͟͡ ͣ ͟͡ ͤ ͟͡ ͥ ͟͡ ͦ ͟͡ ͧ ͟͡ ͨ ͟͡ ͩ ͟͡ ͪ ͢

ͽΠΔΒΥΚΠΟ͙͑Ω͠ͽ͚

;ΠΕΖ͑͢ ;ΠΕΖ͑ͣ

Fig. 6. Mode Shapes of FE Model of PSC Beam

3.3 Inverse Solution: Prestress-Loss Detection

The prestress-loss in the test structure is identified from the inverse solution of the PSC beam models. The relative
change in prestress force with reference to the full prestress force is identified by measuring the relative change in the
nth natural frequency with reference to the frequency margin between the full prestress state and the zero prestress state.
In order to predict the prestress-losses for the cases in Table 1, Eq. (15) is rewritten as follows:
δN δω n2 ω nf2 − ω nd
2
(19)
= =
Nf ω nf2 − ϖ n2 ω nf2 − ϖ n2
where the subscripts f and d denote the full prestress state and the damaged prestress-loss state, respectively. In real
structures, natural frequencies can be measured at those two states.

However, we have to estimate natural frequencies of the zero prestress state, ϖ n , since it is not realistic to measure the
frequencies in existing in-service structures. In this study, ϖ n is modeled by the math model for N = 0 (i.e., 11.409 Hz
for mode 1 and 45.635 Hz for mode 2) and also by the FE model for N = 0 (i.e., 11.196 Hz for mode 1 and 44.047 Hz
for mode 2). Here all ten cases in Table 1 were examined to detect prestress-loss and the results are outlined in Table 2.

Proc. of SPIE Vol. 5057 565

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 The predicted prestress-losses using those two models were compared with the measured in-situ prestress-losses.
The predicted prestress-losses versus the inflicted prestress-losses were plotted in Figs. 8(a) and 8(b). The correlation
between those two sets is relatively high both in FE model and in Math model. It means that the prestress-loss in the
PSC beam can be detected via monitoring changes in natural frequencies of a few basic modes. In case of the Math
model, prediction errors are between 1% and 72% for mode one and very high for mode two. In case of the FE model,
prediction errors are between 5% and 75% for mode one and between 10% and 70% for mode. It is observed that the
accuracy of the predicted prestress-loss results depends on the accuracy of measured frequencies and the accuracy of the
baseline modeling of the zero force state. In mode one, both models shows almost identical results. In mode two, the FE
model shows relatively better accuracy than the Math model.

ͧ͢ ͦ͡
ͥͪ
ͦ͢
͚Ϋ ͚Ϋ ͥͩ
͹
͙͑ ͥ͢ ͹
ΪΔ ͙͑Ϊ ͥͨ
ΟΖ ΔΟ
Φ΢ͤ͢ ΖΦ ͥͧ
ΖΣ ΢
ΖΣ
ͷ ͷͥͦ
ͣ͢
ͥͥ
͢͢ ͥͤ
͡ ͣ͡ ͥ͡ ͧ͡ ͩ͡ ͢͡͡ ͣ͢͡ ͥ͢͡ ͡ ͣ͡ ͥ͡ ͧ͡ ͩ͡ ͢͡͡ ͣ͢͡ ͥ͢͡
΁ΣΖΤΥΣΖΤΤ͑ͷΠΣΔΖ͙͑ΜͿ͚ ΁ΣΖΤΥΣΖΤΤ͑ͷΠΣΔΖ͙͑ΜͿ͚


    ͶΩΡΖΣΚΞΖΟΥ ;ΒΥΙ͑;ΠΕΖΝ ͷͶ͑;ΠΕΖΝ

(a) Mode 1 (b) Mode 2

Fig. 7. Measured Frequency Versus Predicted Frequency

͢ ͢
ΤΤ ΤΤ ͩ͟͡
Πͽ ͩ͟͡ Πͽ
͞Τ ͞Τ
ΤΖ ΤΖ
ΥΣΤ ͧ͟͡ ΣΥ ͧ͟͡
ΖΣ ΤΖ
΁
͑Ε ͥ͟͡ Σ΁
ΖΥ ͑Ε ͥ͟͡
ΔΚ ΥΖΔ
ΕΖ ͣ͟͡ ΚΕ
Σ΁ ΖΣ ͣ͟͡
΁
͡
͡ ͣ͟͡ ͥ͟͡ ͧ͟͡ ͩ͟͡ ͢ ͡
;ΖΒΤΦΣΖΕ͑΁ΣΖΤΥΣΖΤΤ͞ͽΠΤΤ ͡ ͣ͟͡ ͥ͟͡ ͧ͟͡ ͩ͟͡ ͢
;ΖΒΤΦΣΖΕ͑΁ΣΖΤΥΣΖΤΤ͞ͽΠΤΤ

   
;ΒΥΙ͑;ΠΕΖΝ ͷͶ͑;ΠΕΖΝ

(a) Mode 1 (b) Mode 2

Fig. 8. Measured Prestress-Loss Versus Predicted Prestress-Loss

566 Proc. of SPIE Vol. 5057

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 Table 2. Prestress-Loss Prediction Results
Experiment Math Model FE Model
Case
N (kN) δN N (δN N )1 (δN N ) 2 (δN N )1 (δN N ) 2
1 0 1.0 0.999 - 0.953 -
2 15.71 0.882 0.835 - 0.795 0.994
3 27.05 0.796 0.471 - 0.449 0.559
4 36.49 0.725 0.629 - 0.599 0.671
5 57.25 0.569 0.426 - 0.406 0.144
6 81.81 0.384 0.177 - 0.169 0.172
7 91.26 0.313 0.108 -1.930 0.103 -0.261
8 121.46 0.085 0.108 0.043 0.103 0.006
9 130.91 0.014 0.031 -0.934 0.029 -0.133
10 132.80 0.0 0.0 0.0 0.0 0.0

4. CONCLUSION

A methodology to nondestructively detect prestress-loss in beam-type PSC bridges using a few natural frequencies was
presented. An analytical model to estimate natural frequencies of the PSC bridges under various prestress forces was
developed. Also, an inverse-solution algorithm to predicted prestress-loss was formulated. The feasibility and
practicality of the model was evaluated using PSC beams for which a few natural frequencies were measured for a set
of prestress-loss cases.
By applying the proposed approach to the test structure, it was observed that the predicted natural frequencies
showed relatively accurate compared to the measured frequencies at different prestressing stages. It was also observed
that both in FE model and Math model, the correlation between the inflicted prestress-losses and the predicted prestress-
losses was relatively high. The prestress-loss in the PSC beam could be detected via monitoring changes in natural
frequencies of a few basic modes. Research to improve the accuracy of prestress-loss prediction should be continued
along two lines of inquires: the uncertainty in measurement of modal data and the accuracy of baseline modeling of the
initial zero-force state of the structure.

ACKNOWLEDGMENT

This work was partially supported by Smart Infra-Structure Technology Center and KOSEF in the program year of 2002.

REFERENCES

Aalami, B.O. (2000). “Structural Modeling of Posttentioned Members”, Journal of Structural Engineering, ASCE, Vol.
126, No. 2, pp.157-162.
Cawley, P., and Adams, R. D. (1979). “The Location of Defects in Structures from Measurements of Natural
Frequencies”, J. Strain Analysis, Vol. 14, No. 2, pp. 49-57.
Clough, R.W., and Penzien, J. (1982), Dynamic of Structures, McGraw Hill, USA.
Civjan, S.A., Jirsa, J.O., Carrrasquillo, R.L., and Fowler, D.W. (1995). “Method to Evaluate Remaining Prestress in
Damaged Prestressed Bridge Girders”, Research Report 1370-2, Center for Transportation Research, University of
Texas at Austin.
Kim, J.T., and Stubbs, N. (1995). “Model-Uncertainty Impact and Damage-Detection Accuracy in Plate Girder”, Journal
of Structural Engineering, ASCE, Vol. 121, No., 10, pp. 1409-1417.
Kim, J.T. and Stubbs, N. (2002). “Improved Damage Identification Method Based on Modal Information”, Journal of
Sound and Vibration, Vol. 252, No. 2, pp. 223-238.

Proc. of SPIE Vol. 5057 567

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms

 Kim, J.T., Ryu, Y.S., Cho. H.M, and Stubbs, N. (2003). “Damage Identification in Beam-Type Structures: Frequency-
Based Method Vs Mode-Shape-Based Method”, Engineering Structures, Vol. 25, pp. 57-67
Lin, T.Y. (1963), Design of Prestressed Concrete Structures, John Wiley & Sons, USA.
Miyamoto, A., Tei, K., Nakamura, H., and Bull, J.W. (2000). “Behavior of Prestressed Beam Strengthed with External
Tendons”, Journal of Structural Engineering, ASCE, Vol. 126, No. 9, pp. 1033-1044.
Pandey, A. K., Biswas, M., and Samman, M. M. (1991) “Damage Detection from Changes in Curvature Mode Shapes”,
Journal of Sound and Vibration, Vol. 145, No. 2, pp. 321-332.
Rytter, A. (1993). “Vibration Based Inspection of Civil Engineering”, Ph.D. Dissertation, University of Aalborg,
Denmark.
Saiidi, M., Douglas, B., and Feng, S. (1994). “Prestress Force Effect on Vibration Frequency of Concrete Bridges”,
Journal of Structural Engineering, ASCE, Vol. 120, No. 7, pp. 2233-2241.
Saiidi, M., Shield, J., O’Connor, D., and Hutchens, E. (1996). “Variation of Prestress Force in a Prestressed Concrete
Bridge During the First 30 Months”, PCI Journal, Vol. 41, No. 5, pp. 66-72.
Saiidi, M., Hutchens, E., and D. Gardella. (1998). “Bridge Prestress Losses in Dry Climate”, Journal of Bridge
Engineering, ASCE, Vol. 3, No. 3, pp. 111-116.
Stubbs, N., and Osegueda, R. (1990). “Global Nondestructive Damage Evaluation in Solids”, Int J. Analytical and
Experimental Modal Analysis, Vol. 5, No. 2, pp. 67-79.
Stubbs N., and Kim, J.T. (1996), “Damage Localization in Structures without Baseline Modal Parameters”, AIAA
Journal, Vol. 34, No. 8, pp. 1649-1644.

*[email protected]; phone 82 51 620-6227; fax 82 51 628-8146; structure.pknu.ac.kr

568 Proc. of SPIE Vol. 5057

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/15/2012 Terms of Use: http://spiedl.org/terms