Redevelopment of Prestressing Force

in Severed Prestressed Strands

Jarret Kasan, S.M.ASCE1; and Kent A. Harries2
Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The flexural capacity rating of prestressed girders having severed or corroded strands often progresses by simply neglecting the
contribution of the affected strands from the girder section and assessing the capacity. While this is an adequate approach at the affected
section, it is conservative elsewhere along the span, assuming that fully bonded strands are used. Once it reenters sound concrete, the severed
or corroded strand continues to be bonded to the concrete; thus stress transfer between the concrete and strand is possible and the strand
prestress force may be “redeveloped” (in the sense of transfer length) by bond transfer at a distance from the damage location. This paper
reports an experimental study that clearly shows that prestress force is redeveloped at a distance from the location at which a strand is severed.
Furthermore, American Association of State Highway and Transportation Officials (AASHTO) and American Concrete Institute (ACI) code-
prescribed transfer-length calculations appear to remain valid for this redevelopment behavior. DOI: 10.1061/(ASCE)BE.1943-5592
.0000177. © 2011 American Society of Civil Engineers.
CE Database subject headings: Prestressed concrete; Corrosion; Girders; Bonding.
Author keywords: Prestressed concrete; Corrosion; Development length; Transfer length.

Introduction following testing to failure (Harries 2009). In this case, the lowest
layer of strand in the box, having a clear cover of only 32 mm
The deteriorating condition of the nation’s bridge infrastructure (1.25 in.) and partially exposed because of spalling likely resulting
cannot be overstated. Prestressed concrete (PC) bridges, however, from vehicle impacts, exhibits marked corrosion. The second layer
generally contradict this trend. These often newer structures have of steel, 38 mm (1.5 in.) above the lower layer remains bright.
thus far demonstrated exceptional durability and represent only a Figs. 1(c) and 1(d) show representative damage to prestressed
small fraction of deficient bridges in the nation (FHWA 2007). box girders (Harries 2009). The damage shown in Fig. 1(c) was
Nonetheless, PC bridges are susceptible to potentially catastrophic almost certainly initiated by a recent vehicle impact and does
damage, the extent of which is often difficult to assess until it has not exhibit significant corrosion yet. Left unpatched, the damage
progressed to the point of collapse (Harries 2009). Deterioration of in Fig. 1(c) is likely to progress to that shown in Fig. 1(a).
PC bridges usually results from corrosion of the strand in a region Fig. 1(d) shows relatively typical soffit corrosion. In this case,
where the concrete has been damaged, often by vehicle impact the initial cause is less clear, although the corrosion is most cer-
(Harries et al. 2009). Because of its refined chemistry and typically tainly accelerated by salt spray from the road below. In the structure
small diameter, prestressing steel is particularly susceptible to cor- shown, the vertical clearance to the interstate carriageway below
rosion (ACI 222 2001). Nonetheless, provided that sound concrete was only 4.42 m (14.5 ft) making both vehicle impact and salt spray
ensuring resistance to chloride attack is present, the passivating problematic (Harries 2009).
layer on the prestressing strand is maintained and corrosion is Damage such as that seen in Fig. 1 clearly affects the capacity of
mitigated. Examples of this are shown in Fig. 1, which shows the individual girder and, therefore, the bridge. Furthermore, when
damage is caused by vehicle impact, it will often be located in the
the observed damage from the collapsed Lake View Drive Bridge
critical middle third of the flexural span. When rating a girder
reported by Harries (2009). In Fig. 1(a), a 9.5 mm (3=8 in:)-
exhibiting such damage, typically a sections approach will be used
diameter strand is entirely corroded at the location of a vehicle
and the contribution of the severed and corroded strands will be
impact; the steel crumbled to the touch. A short distance away,
neglected. While this is an adequate approach at the affected sec-
where the strand reenters sound concrete, the corrosion is only sur- tion, it is conservative elsewhere along the span, assuming that fully
face corrosion. Finally at the location at which the strand was again bonded strands are used. The severed or corroded strand, once it
encased in sound concrete, only bright steel is found. A similar reenters sound concrete, continues to be bonded to the concrete;
situation is seen in Fig. 1(b), showing the soffit of a box girder thus stress transfer between the concrete and strand is possible.
1
If this is the case, the strand may be redeveloped (in the sense
Doctoral candidate, Dept. of Civil and Environmental Engineering, of development length) by bond transfer at a distance from the dam-
Univ. of Pittsburgh, Pittsburgh, PA 15261.
2 age location. By the same argument, the prestress force in a bonded
Associate Professor, Dept. of Civil and Environmental Engineering,
Univ. of Pittsburgh, Pittsburgh, PA 15621 (corresponding author). E-mail:
strand is not lost at a distance from the damage, since it too may be
[email protected] redeveloped.
Note. This manuscript was submitted on February 23, 2010; approved
on August 31, 2010; published online on September 6, 2010. Discussion Development and Transfer Lengths
period open until October 1, 2011; separate discussions must be submitted
for individual papers. This paper is part of the Journal of Bridge Engineer- Conceptually, this redevelopment is no different from the original
ing, Vol. 16, No. 3, May 1, 2011. ©ASCE, ISSN 1084-0702/2011/3-431– stress transfer from the tensioned strand to the concrete section
437/$25.00. at the ends of the member at the time of prestress force release,

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011 / 431

J. Bridge Eng., 2011, 16(3): 431-437

Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. In situ corrosion of prestressing strand

only in this case, the free surface of the section is not at the end For typical cases, f pe is on the order of 1,240 MPa (180 ksi), thus
of the member but at the edge of the damaged region. Thus, the both AASHTO and ACI requirements give the same transfer
concepts of transfer length and development length should remain length. The development length, also measured from the point
valid (ACI Committee 2008; AASHTO 2007). at which bond commences, is given by both AASHTO and ACI as
In a prestressed concrete element, the stress in the prestressing
steel may be assumed to vary linearly from zero at the point where λd ¼ κðf ps
0:66f pe Þd b ðksi and in:Þ
bonding commences to the effective stress after losses, f pe , at the ð2Þ
λd ¼ 0:145κðf ps
0:66f pe Þd b ðMPa and mmÞ
end of the transfer length, ℓtr . In a member loaded to its capacity,
strand stress may be assumed to increase linearly from the end of
the transfer length to the development length, ℓd , reaching the stress where κ ¼ 1:0, except for members deeper than 610 mm (24 in.)
at nominal resistance, f ps , at the development length. This idealized designed based on AASHTO, where κ ¼ 1:6. Both transfer and
relationship is shown in Fig. 2. The transfer length for a bonded development length recommendations were developed in the
seven-wire strand is given as 60 strand diameters (60d b ) by 1950s and 1960s and adopted by ACI and AASHTO in 1963
AASHTO (2007) and as shown in Eq. (1) by ACI (2008). and 1973, respectively (Tabatabai and Dickson 1993). The κ factor
was introduced by AASHTO in 1988 to address perceived worst-
case characteristics of older strand material (AASHTO 2007). All
ℓtr ¼ 0:33f pe d b ðksi and in:Þ previous calculations assume a conventional minimum concrete
ℓtr ¼ 0:05f pe d b ðMPa and mmÞ ð1Þ cover over the strand; for flexural members, this is typically
38 mm (1.5 in.) (ACI Committee 2008; AASHTO 2007).

Severed Strands and Redevelopment

Permitting strand redevelopment allows that at a distance greater

than ℓd from the observed strand damage, girder capacity is re-
stored since (a) the strand is still present in sound concrete and thus
is reinforcing the section, and (b) the prestress force attributed to
the damaged strand is still present. This, however, is not typically
taken into account in rating damaged prestressed girders, where the
common practice is to assume that once a strand is severed, it is lost
over the entire length of the girder. By accounting for redevelop-
ment and considering appropriate sections along a damaged girder,
it is possible to significantly increase a girder’s rating over that
obtained for the damaged section.
Fig. 3 illustrates this conceptually for a simply supported PC
girder. In Fig. 3, an undamaged girder has a moment capacity
of M. The lines in Fig. 3 represent damage reducing the sectional
capacity from 95 to 75% of the undamaged capacity. The actual
Fig. 2. Idealized transfer and development lengths
girder capacity (based on a uniformly distributed load) is given

432 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011

J. Bridge Eng., 2011, 16(3): 431-437

Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Girder capacity accounting for location of damage

based on the location of this sectional damage over the half-span of (f pu ¼ 263 ksi), having a standard deviation of 27.6 MPa
the girder. For instance, severe sectional damage may reduce capac- (4 ksi), confirming the grade and quality of the almost 50-year
ity to 0.75 M. If this damage is located at midspan (x ¼ 0:5), the old strand. A stress-strain curve obtained from one of the center
girder flexural capacity is 0.75 M. If, however, the same damage is (straight) wires tested is presented in Fig. 5.
located at x ¼ 0:35, the overall girder load carrying capacity is Initial prestress in all strands is reported to be f pi ¼ 0:7 f pu ¼
0.82 M, and if x < 0:25, the girder flexural capacity is unaffected. 175 ksi ¼ 1;207 MPa (Spancrete 1960). Calculations of estimated
Thus, by considering the various undamaged sections away from prestress loss based on the method reported by PCI (1999) were
the damage location, the load-carrying capacity of the girder may 0:13f pu ¼ 33:7 ksi ¼ 232 MPa, resulting in an effective prestress
be improved over the case of considering only the critical section. force of f pe ¼ 0:57 f pu ¼ 141:3 ksi ¼ 974 MPa. The force in a sin-
To permit this approach, the damaged strand must be redeveloped. gle corner strand on the test girder was experimentally established
to be 0:44f pu ¼ 110:5 ksi ¼ 762 MPa by Naito et al. (2006) In the
present study, the force in an interior bottom layer strand (H)
Experimental Program was similarly found to be 0:47f pu ¼ 118:3 ksi ¼ 816 MPa
(see Table 1).
If the beneficial effects of accounting for the redeveloped strand are The experimentally obtained prestress force values necessarily
to be used in load rating or assessing the need for repair or replace- include both the effects of prestress (f pe ) and the effects of applied
ment, the effectiveness of the redevelopment, indeed its existence, load. The latter can be assessed based on a sections analysis and
must be established. The following experimental program was is subtracted from the experimentally obtained force in the
carried out to assess strand redevelopment behavior. strand to assess the effective prestress. In this study, the only load
carried by the girder is its own self-weight, determined to be
Test Girder 1;355 kg=m(909 plf). The girder is supported as shown in
Tests were conducted on a girder recovered from the decommis- Fig. 4(a), and the location at which strand force is determined is
sioned Lake View Drive bridge (Harries 2006). Originally built located only 4.1 m (13.6 ft) from the nearest support. Based on
in 1960, the test girder was an exterior girder of an adjacent these conditions, the stress in a single lower-layer strand that is
box girder bridge having no composite topping. The girder is a attributed to applied load is only 3.4 MPa (0.5 ksi).
1070 mm (42 in.) deep by 1220 mm (48 in.) wide hollow PC The relatively low apparent values of effective prestress force
box 27.8 m (91.2 ft) long (Spancrete 1960). Primary reinforcement were confirmed in two other girders from the Lake View Drive
consisted of 60 9.5 mm 3=8 in: diameter stress-relieved 1720 MPa bridge that had effective prestress forces of 696 and 662 MPa
(250 ksi) seven-wire strands placed in five layers. A section of the (101 and 96 ksi) (Naito et al. 2006). Additional confirmation of
girder is shown in Fig. 4(b). The asphalt topping and barrier wall these values was made by direct measurements of girder camber,
had been removed from the girder when it was decommissioned. which were all smaller than the long-term calculations (PCI 1999)
The original bridge had a 39° skew. At the time of the experimental and fabricator drawings (Spancrete 1960) indicated (Harries 2006).
program described, the girder was supported on four timber cribs, No reason is offered for these low values, although we noted that
as shown in Fig. 4(a). The test region itself was in excellent shape, both strand and concrete materials were in generally excellent
there was little evidence of deterioration or damage of any kind. shape (apart from isolated damage). Therefore, the authors suggest
Subsequent removal of the concrete cover for instrumentation that the initial prestress, f pi , may have been lower than intended.
and testing (described in the following) revealed only bright strands
Test Procedure and Protocol
in sound concrete with no evidence of corrosion.
Although the concrete strength of the test girder was not estab- The basic premise of the test procedure was to intentionally sever a
lished in this study, the specified concrete compressive strength for prestressing strand while measuring the strain in the same strand at
all girders of the Lake View Drive bridge was 40.7 MPa (5,900 psi). some distance from the cut. The drop in strain at a distance from the
Cores taken from two other girders (Harries 2006) revealed very cut is an indication of the transfer of prestress between this location
consistent strengths of 50.3 MPa (7,300 psi). A single core and that of the cut, at which the stress is now zero. In this manner,
extracted from the test girder in a previous study yielded a com- the transfer length may be experimentally determined.
pressive strength of 58.2 MPa (8,440 psi) (Naito et al. 2006). Thus To access the strands for instrumentation, a shallow notch was
the concrete is considered to be of very good quality. Two samples cut transversely across the girder soffit, removing the concrete
of prestressing strand were removed from the test girder and the cover and exposing the surface of the lower layer of prestressing
individual wires were tested. The average ultimate strength of strand [Fig. 4(e)]. Considerable care was taken to avoid damage to
both center (straight) and outer (helical) wires was 1,813 MPa the strands and to maintain as much concrete around the strand as

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011 / 433

J. Bridge Eng., 2011, 16(3): 431-437

Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Test and girder details

possible. Very small electrical resistance strain gauges (micromea- at a shorter distance ℓcut (cut No. 5, for instance). Strands were cut
surements type EA-06-062AQ-350/P) having a grid width of using an abrasive cut-off wheel on an 8,000 rpm grinder. Each cut
1.57 mm (0.062 in.) and grid length of 2.90 mm (0.114 in. were was accomplished in about 15 s. Strain readings of all gauges were
applied to a single wire of the partially exposed strand. To facilitate taken before and after each cut. Approximately 10 min elapsed
accurate alignment, the strain gauge matrix was trimmed to be just between subsequent cuts. No data is reported for gauge G, since
greater than the grid width. The gauges were applied within the this gauge was damaged while making cut No. 1.
notch as close as possible to the edge from which the cut distance
was to be measured [distance x shown in Fig. 4(d) and given in
Table 1]. Experimental Results
Eight strands were instrumented; these were labeled A–H as
indicated in Fig. 4. Individual strands were cut at a distance ℓcut All results are presented in terms of stress change. Because a strand
from the edge of notch. The order in which strands were cut is in tension when it is cut, the result of the cut is a negative stress
and the respective cut distances, ℓcut , are given in Table 1. In some change. All stress-change data is calculated from strain-gauge
cases, where ℓcut was sufficiently long to have little observed effect data assuming 200 GPa (E ¼ 29;000 ksi). All normalized stress
on the strain at the gauge location, the strand was cut a second time values reported have been normalized based on 1,720 MPa

434 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011

J. Bridge Eng., 2011, 16(3): 431-437

 (f pu ¼ 250 ksi). Fig. 6 shows a stress history of all gauges reported
in this study. In each case, the significant stress drop represents that
associated with cutting the strand noted. The stress change attrib-
utable to cutting each strand is also given in Table 1.
The stress in strands not cut at a given time increases marginally
(increased tension) since the act of cutting a strand eliminates a
portion of the section reinforcing, redistributing stress to the re-
maining strands. For instance, based on the results of strand cut
No. 1, the stress in the strands attributable to prestress and girder
dead load was 816 MPa (118.3 ksi). Strand cut Nos. 11 and 12, also
both in the notch, yielded larger apparent stresses since these were
made to a section having six and seven fewer effective strands, re-
Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

spectively. Indeed, the total stress drop of strand E during the

course of testing was 988 MPa (143.3 ksi), which accounts for
the increase in stress associated with cutting the adjacent strands.
Strands that lost only some of their prestress following cutting
show some further reduction in stress with subsequent readings.
Fig. 5. Stress-strain curve obtained for strand from test girder This is attributed to the creep of the transfer forces (as described
in the next section) rather than any effect from subsequent
Table 1. Strand Cutting Order, Location, and Resulting Stress Change
strand cuts.

Stress changea Equilibration of Transfer Forces

at strain gauge owing
Cut no. Strand x mm (in.) ℓcut mm (in.) to cut ksi (MPa) Where partial stress drop was observed, it was not instantaneous.
Upon cutting the strand at some distance ℓcut , a drop in stress at the
1 H 17.8 (0.7) 0 (cut in slot)
118:3 (
816)
strain gauge occurs. With time, the stress continues to fall as the
2 F 15.2 (0.6) 59:5db ¼ 566 (22.3) none
transfer of remaining prestress forces equilibrate. This behavior is
3 E 12.7 (0.5) 52:0db ¼ 495 (19.5)
0:3 (
2:1) believed to result from the progressive redistribution of bond stress
4 D 15.2 (0.6) 34:7db ¼ 330 (13.0)
0:3 (
2:1) along the strand in the vicinity of the cut. Fig. 7 shows the contin-
5 Fb 15.2 (0.6) 16:3d b ¼ 155 (6.1)
27:0 (
186) ued stress drop in Strand A following the initial drop associated
6 C 15.2 (0.6) 13:3d b ¼ 127 (5.0)
73:1 (
504) with cutting this strand [714 MPa (103.6 ksi)]. It takes about
7 B 22.9 (0.9) 11:5d b ¼ 109 (4.3)
112:4 (
775) 10 min for the stress in the strand to reequilibrate at a level almost
8 A 20.3 (0.8) 4:3d b ¼ 41 (1.6)
130:6 (
900) 186 MPa (27 ksi) below the prestress remaining immediately
9 Eb 12.7 (0.5) 26:9db ¼ 257 (10.1)
25:6 (
177) following the cut; the final stress drop because of cutting Strand A
10 Db 15.2 (0.6) 20:8d b ¼ 198 (7.8)
4:5 (
31) (cut No. 8) was 900 MPa (130.6 ksi), as indicated in Table 1.
11 Db 15.2 (0.6) 0 (cut in slot)
125:0 (
862)
Apparent Transfer Length
12 Eb 12.7 (0.5) 0 (cut in slot)
117:4 (
809)
a
Stress calculated from strain gauge data assuming E ¼ 29;000 ksi Fig. 8 shows the stress change resulting from increasing cut distan-
(200 GPa). ces. There was no observed stress drop associated with any cuts
b
Strand recut after initial cut had little effect. made beyond ℓcut ¼ 35d b ¼ 13:1 in: ¼ 333 mm. The results

25 0.1

0 0
gauge F gauge D
stress change (normalized to fpu)

-25 -0.1
stress change (ksi)

-50 -0.2
gauge C
gauge E
-75 -0.3

gauge H gauge B
-100 -0.4

-125 -0.5
gauge A

-150 -0.6
initial stress, fpi =175ksi = 0.70fpu
-175 -0.7
0 2 4 6 8 10 12 14
cut number

Fig. 6. Stress change-cut number history for all gauges

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011 / 435

J. Bridge Eng., 2011, 16(3): 431-437

Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Stress change-time history for strand A following cut No. 8

Fig. 8. Stress drop versus cut distance

appear to validate the ACI transfer length equation [Eq. (1)], which capacity, this would be expected to affect serviceability of these
is equal to 39d b in this case. girders. Beyond destructive tests similar to those conducted here,
The effective prestress in the strands found in this study is there is no practical method for assessing prestressing force in situ.
lower than would otherwise be predicted: 816 MPa (118 ksi) versus Finally, the writers acknowledge that this paper presents exper-
a predicted value of 974 MPa (141 ksi) calculated previously. Addi- imental data from a single decommissioned girder. The work has
tionally, the cuts in this study were discrete and did not disturb the demonstrated that the redevelopment of severed strands that should
surrounding concrete to a great degree. Considering the less occur, does, in fact, occur in a manner that is essentially intuitive.
discrete nature of real strand damage, one might expect a longer Nonetheless, opportunities to verify these results are relatively rare
apparent transfer length in situ. Thus, the authors contend that and should be capitalized on when they become available. In par-
the use of the AASHTO-prescribed value of ℓtr ¼ 60d b is appro- ticular, the effect of reduced concrete cover on prestress force trans-
priate to conservatively establish the transfer length required to fer and strand development requires investigation. Harries (2006)
reestablish prestress force at a distance beyond the extent of the noted significant variation in concrete cover in the Lake View Drive
damaged concrete. girders, sometimes as small as 14.5 mm (0.57 in.). While both the
In this study, it has been shown that the transfer-length calcu- concrete and strand seemed sound in this case, the effect on effec-
lation used for design remains valid for the redevelopment of ef- tive prestress force is unknown.
fective prestress in a severed strand. The development length of the
severed strand [Eq. (1)] has not been validated, but the validation of
the transfer length suggests no changes in strand bond behavior, Acknowledgments
and therefore no effect on development length is anticipated.
The writers gratefully acknowledge the Pennsylvania Department
of Transportation for allowing the writers continued access to the
Discussion and Conclusions Lake View Drive girders.

The findings of this study demonstrate that severing prestressing

strand is a local effect. That is, the effects of the lost strand affect References
the section at the damage location. To either side of the damaged
concrete region, the effective prestress in the strand is redeveloped American Association of State Highway and Transportation Officials
(AASHTO). (2007). LRFD bridge design specifications, 4th Ed. and
over the transfer length, ℓtr . This approach implies the need to con-
Interims, Washington, DC.
sider not only the critical section of a girder but also all sections American Concrete Institute (ACI) Committee 222. (2001). “Corrosion of
along its length when rating the girder or designing repair measures prestressing steels.” ACI 222.2R-01, ACI, Farmington Hills, MI, 43.
for a damaged girder (Harries et al. 2009). American Concrete Institute (ACI) Committee 318. (2008). “Building code
This particular study also identified lower than expected values requirements for structural concrete and commentary.” ACI 318-08,
of prestress force in the sound strands. While not affecting ultimate ACI, Farmington Hills, MI, 43.

436 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011

J. Bridge Eng., 2011, 16(3): 431-437

 Federal Highway Administration (FHWA). (2007). “National bridge inven- “Forensic evaluation of prestressed box beams from the Lake
tory (NBI).” 〈www.fhwa.dot.gov/bridge/nbi.htm〉 (Feb. 12, 2010). View Drive over I-70 bridge.” ATLSS Rep. No. 06-13, Lehigh Univ.,
Harries, K. A. (2006). “Full-scale testing program on de-commissioned Bethlehem, PA, 62.
girders from the Lake View Drive bridge.” Rep. FHWA-PA-2006- Precast/Prestressed Concrete Institute (PCI). (1999). PCI design handbook,
008-EMG001, Pennsylvania Department of Transportation, Harrisburg, 5th Ed, PCI, Chicago.
PA, 158. Spancrete. (1960). “Washington County L.R. 798-1 bridge at STA. 1205 þ
Harries, K. A. (2009). “Structural testing of prestressed concrete girders 50:00 Drawings (3 sheets).” June 28, 1960. Approved by PADoH
from the Lake View Drive bridge.” J. Bridge Eng., 14(2), 78–92. August 10, 1960.
Harries, K. A., Kasan, J., and Aktas, C. (2009). “Repair methods for Tabatabai, H., and Dickson, T. J. (1993). “The history of the prestressing
prestressed girder bridges.” Rep. FHWA-PA-2009-008-PIT 006,
strand development length equation.” PCI J., 38(6), Nov.–Dec. 1993,
Pennsylvania Department of Transportation, Harrisburg, PA, 169.
64–75.
Naito, C., Sause, R., Hodgson, I., Pessiki, S., and Desai, C. (2006).
Downloaded from ascelibrary.org by College of Engineering Trivandrum on 12/29/15. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF BRIDGE ENGINEERING © ASCE / MAY/JUNE 2011 / 437

J. Bridge Eng., 2011, 16(3): 431-437