Professional Documents
Culture Documents
2012 - Residual Prestress Forces and Shear Capacity of Salvaged Prestressed Concrete Bridge Girders
2012 - Residual Prestress Forces and Shear Capacity of Salvaged Prestressed Concrete Bridge Girders
Abstract: Seven prestressed concrete bridge girders that had been in service for 42 years, and represented two span lengths and reinforce-
ment designs, were tested to determine their effective prestress force and ultimate shear capacity. A cracking moment test was used to
determine the effective prestress force in the girders. The measured effective prestress force was compared with calculated values according
to the AASHTO LRFD prestress loss equations to investigate their adequacy. The AASHTO refined method was shown to provide the most
accurate results to within 10% of the measured values. An ultimate shear test was also performed on two of the girders. An external load
was applied near the support and increased until the girder failed in shear. The various procedures in the AASHTO LRFD specifications
were compared with the measured results. The AASHTO simplified procedure predicted only 51% and 39% of the average measured shear
capacity for the short and long span girders, respectively. The strut-and-tie models were found to estimate the shear capacity more accurately.
The AASHTO refined method was shown to provide the most accurate results. DOI: 10.1061/(ASCE)BE.1943-5592.0000212. © 2012
American Society of Civil Engineers.
CE Database subject headings: Prestressed concrete; Finite element method; Experimentation; Concrete bridges; Girder bridges.
Author keywords: Prestress concrete; Shear capacity; Prestress losses; Finite-element modeling; Experimental testing.
Introduction steel in prestressed concrete beams affected the overall shear behav-
ior of girders. They found that as the area of the prestressing strand
The in-service structural properties of prestressed concrete girders increased, the shear capacity increased even when the total
are difficult to accurately determine, especially when the girders prestress force was held constant. By increasing the area of the
have been in the field for an extended period of time. While mild flexural steel reinforcement, the overall shear capacity also
there are several, two key factors that affect the performance of increased. Oh and Kim (2004) tested the shear capacities of large-
prestressed concrete girders are the following: (1) the effective pre- scale, prestressed concrete beams and concluded that beams with
stress force and (2) the shear capacity near the supports. Since these higher concrete strength exhibited more diagonal cracks with a
properties are very difficult to measure in situ, accurate analytical smaller crack width and that the principal directions decreased
methods are required. as the load increased. They also concluded that the concept of aver-
Current design methodologies provide some guidance on how to age strains and the changing of principal directions according to the
determine these two properties. According to the AASHTO LRFD applied load can be used for a more realistic shear analysis and that
specifications (2009), total prestress losses can be determined by the girders with higher-strength concrete showed an increase in the
using a time-dependent or lump-sum procedure. The AASHTO ultimate shear capacity. Kaufman and Ramirez (1988) tested six
LRFD specifications also provide two sectional procedures and AASHTO girders made with high-strength concrete to gain a better
one strut-and-tie approach for calculating the shear capacity. understanding of both the flexural and the shear behavior of pre-
Various researchers have studied both of these issues. Saqan and stressed concrete girders. They found that the higher-strength con-
Frosch (2009) performed a study to determine how the flexural crete increased the ultimate capacities for both shear and flexure.
1
MacGregor et al. (1965) tested 104 prestressed concrete beams to
Graduate Research Assistant, Dept. of Civil and Environmental Engi- determine their ultimate shear capacities and to understand their
neering, Utah State Univ., 4110 Old Main Hill, Logan, UT 84332-4110. overall shear behavior. Many different beam configurations were
2
Associate Professor, Dept. of Civil and Environmental Engineering,
Utah State Univ., 4110 Old Main Hill, Logan, UT 84332-4110 (corre-
tested to quantify important variables that affect the shear behavior
sponding author). E-mail: paul.barr@usu.edu of prestressed concrete beams and, on the basis of the findings,
3
Graduate Research Assistant, Dept. of Civil and Environmental Engi- proposed relationships that have been used over the years. Other
neering, Utah State Univ., 4110 Old Main Hill, Logan, UT 84332-4110. studies include Naito et al. (2010), Kordina et al. (1989), and
4
Associate Professor, Dept. of Civil and Environmental Engineering, Elzanaty et al. (1986).
Utah State Univ., 4110 Old Main Hill, Logan, UT 84332-4110. E-mail: The effective prestress force for a prestressed concrete girder is
marv.halling@usu.edu an important property required to determine the service behavior
5
Graduate Research Assistant, Dept. of Civil and Environmental Engi- and the ultimate shear and flexural capacity of a girder. Several
neering, Utah State Univ., 4110 Old Main Hill, Logan, UT 84332-4110.
researchers have performed studies to determine the effective pre-
Note. This manuscript was submitted on July 15, 2010; approved on
December 22, 2010; published online on December 27, 2010. Discussion
stress force in a girder. Barr et al. (2008) instrumented five pre-
period open until August 1, 2012; separate discussions must be submitted stressed, high-performance concrete bridge girders and recorded
for individual papers. This paper is part of the Journal of Bridge Engineer- the changes in prestress for 3 years from the time of casting. They
ing, Vol. 17, No. 2, March 1, 2012. ©ASCE, ISSN 1084-0702/2012/2- found that methods provided in previous editions of the AASHTO
302–309/$25.00. LRFD specifications for predicting prestress losses overpredicted
sured from the bottom of the beam (Fig. 2). The ultimate capacity determined, it was then used to calculate the effective prestress
of the strands was the same as Girders 1 through 6 [1,720 MPa force in the girder. At service, the three factors that contribute
(250 ksi)]. The compressive strength of this girder was specified to the stress in the bottom of a prestressed girder are as follows:
as 34.5 MPa (5,000 psi) but was experimentally determined to (1) self-weight of the girder, (2) prestress force in the girder,
be 64.1 MPa (9,300 psi). The shear-reinforcing steel also had and (3) externally applied load. The equation to calculate this stress
the same properties as Girders 1 through 6 with a yield stress of can be written as Eq. (1). Each of the girders had slightly different
230 MPa (33.4 ksi). For Girder 7, the shear reinforcement initiated cross-sectional properties owing to the amount of decking that re-
(Si ) at 15.2 cm (6.0 in.) from the center of the support and was then mained above the top flange of the girder. Therefore, detailed sec-
subsequently spaced (S) at 43.2 cm (17.0 in.) on center (S) through- tion properties were measured to accurately calculate the prestress
out the remaining length of the girder (Fig. 2). force for each girder
Pe Pe eyg M sw yg M max yc
σ¼ þ ð1Þ
Cracking Moment Test Ag Ig Ig Ic
Fig. 4. Typical shear test setup Fig. 5. Typical shear test failure
of one end of Girder 7. This shear span was slightly larger at vertical surface strain on the concrete was found to be 0.004,
131 cm (51.5 in.) because of the deck material. indicating that the stress in the reinforcing steel at failure was above
The externally applied load was measured in two ways. This the yield stress of the steel showing that yielding likely occurred.
redundancy was desirable since the ultimate applied load was
the critical value for this testing. The load was measured by a Calculation of Shear Capacity
strain-gauge-based load cell and with a pressure transducer in line
with the hydraulic ram. In addition to the load, deflections were For this research, three predictive methods were used to compare
recorded with an LVDT, and strain gauges attached to the concrete current code practices with the measured results. The first method-
surface of the girder measured changes in strain. The LVDT was ology was the simplified method provided in the AASHTO LRFD
positioned on the top of the girder adjacent to the load cell and ram. specifications (2009). This is the preferred method for most state
The hydraulic ram was used to gradually apply an increasing DOTs when designing or analyzing bridges. The second predictive
load through failure. The applied load was continuously monitored method was the nominal method in the AASHTO LRFD specifi-
and recorded during the test. Once the applied maximum load had cation (2009). The third method used was a strut-and-tie model
decreased by more than 20%, the test was terminated by completely (STM). This method is described in the appendix of the American
removing the load, after which the girder and data were examined. Concrete Institute (ACI) building code (2008). The AASHTO
Table 3 lists the recorded ultimate shear capacities from each test. LRFD specifications also allow for a STM to be used in the design
The ultimate shear capacity was obtained on the basis of the and analysis of prestressed concrete girders for shear capacity.
maximum load and the load configuration. The ultimate load In general, the shear capacity (V n ) of a member is determined
was defined as the maximum recorded load and was used to cal- as a combination of the shear contribution of individual parts. In a
culate the ultimate shear force. The average shear capacity was de- prestressed concrete girder there are three main contributors to the
termined as 727.7 kN (163.6 kip) for Girder 6 and 1,163.2 kN shear strength, namely, transverse steel (V s ), the vertical component
(261.5 kip) for Girder 7. As all testing parameters were constant, of the prestressing force (V p ), and the shear resistance of the con-
the difference in capacities is believed to be because of the smaller crete (V c ). The general governing equation is provided as Eq. (4)
stirrup spacing and larger concrete strength of Girder 7.
For Girder 6, the failure mechanism was flexural shear where Vn ¼ Vc þ Vs þ VP ð4Þ
the cracks first developed at the bottom of the girder at an angle The simplified and nominal methods in the AASHTO LRFD
of 90° from the longitudinal axis. After the cracks propagated specifications (2009) allow the shear capacity of prestressed con-
up through the bottom flange, the orientation changed as the shear crete girders to be calculated on the basis of a sectional analysis.
forces dominated the flexural effects, and the cracks propagated For the simplified method, the calculations for determining V c are
at an angle of approximately 42°. The primary shear crack was very similar to the ACI procedure. The value of V c is calculated in
accompanied by other shear cracks but was smaller in size. The two different ways, depending on the manner the shear cracks de-
accompanying cracks were generally parallel to the primary shear velop, namely, flexure-shear cracking or web-shear cracking. If
crack but with less dilation. The cracks then opened until there flexure-shear cracks control the design, the value V ci [Eq. (5)]
was not enough aggregate interlock or friction to hold the girder should be used as V c . However, if web-shear cracks control the
together, at which point the girder experienced a sudden failure. design, V cw is to be used [Eq. (6)]. V c is defined to be the lesser
Girder 7 failed in a similar manner but at a larger load. More of V ci and V cw
information regarding the cracking behavior of the girders can
be found in Osborn (2010). Fig. 5 shows a picture of a typical pffiffiffiffi VM pffiffiffiffi
V ci ¼ 0:02 f 0 cbv d v þ V d þ i cre ≥ 0:06 f 0 cbv d v ð5Þ
web-cracking failure observed for Girder 7. M max
The strain was measured at the concrete surface. The strain mea-
pffiffiffiffiffiffi
surements on the concrete surface were used to estimate the strains in V cw ¼ ð0:06 f 0 c þ 0:30f pc Þbv d v þ V p ð6Þ
the shear-reinforcing steel at failure. On the basis of the tension test
results, the yield stress of the vertical stirrups was 230 MPa (33.4 ksi). where V d = shear force at section attributable to unfactored
This yield stress occurred at a strain of 0.001, yet the maximum dead load, including both the dead load of the structural
at extreme fiber of section where tensile stress is caused by exter- The calculated values are also compared to the measured values and
nally applied loads; M dnc = total unfactored dead load moment act- shown as a percentage of the measured values.
ing on the monolithic or noncomposite section; Sc = section The average measured shear capacity was 727.7 kN (163.6 kip)
modulus for the extreme fiber of the composite section where ten- for Girder 6 and 1,163.2 kN (261.5 kip) for Girder 7. Using
sile stress is caused by externally applied loads; Snc = section the AASHTO simplified methodology, calculated shear values
modulus for the extreme fiber of the monolithic or noncomposite of 366.0 kN (82.3 kip) and 446.1 kN (100.3 kip) were obtained
section where tensile stress is caused by externally applied loads; for Girders 6 and 7, respectively. The AASHTO equations pre-
f pc = compressive stress in concrete (after allowance for all pre- dicted only 51% of the average measured shear capacity for Girder
stress losses) at centroid of cross-section-resisting externally ap- 6 and 39% for Girder 7. The AASHTO nominal method predicted a
plied loads or at junction of web and flange. calculated shear capacity of 274.4 kN (61.7 kip) and 323.3 kN
The AASHTO nominal method is on the basis of the modified (72.7 kip) for Girders 6 and 7, respectively. These values are
compression field theory. Once cracking has occurred, according to 38% and 28% of the average measured values for Girders 6 and 7,
this method, the web reinforcement resists the tensile stresses, and respectively. Calculated shear capacities from the STM were
the concrete struts carry the compressive stresses. The nominal 616.3 kN (138.6 kip) and 1,150.8 kN (258.7 kip) for Girders 6
capacity of the concrete (V c ) is calculated by using Eq. (7) and 7, respectively. The STM was accurate to within 85% of the
pffiffiffiffi average measured shear capacity for Girder 6 and 98% of the
V c ¼ 0:0316β f 0 cbv d v ð7Þ average measured shear capacity for Girder 7.
In all cases, the calculated values using the sectional analyses in
where β = factor indicating ability of diagonally cracked concrete the AASHTO specifications were very conservative. The AASHTO
to transmit tension and shear. procedures are both on the basis of bending theory, implying that
Both the AASHTO methods are on the basis of bending theory the bending stresses are distributed linearly through the depth of the
with the assumption that plane sections remain plane. The D-region beam, which is typically the case as long as the section being
is defined as any section of a beam within a distance d (depth of the considered is not near the support or an applied concentrated load.
beam) of a concentrated applied load or a support. For this research, The girders that were tested for this research had loads that were
the shear capacity was examined near the D-region where stresses applied near the D-region (a=d ¼ 1:5), where plane sections do not
are not typically distributed linearly through the cross section of the typically remain plane. The STMs developed for this research were
girder. able to much more accurately predict the ultimate shear capacity of
To investigate the predictive shear capacity of the prestressed the girders. This accuracy is attributed to the strut-and-tie method-
concrete beams used in this research, a simple STM was developed ology more effectively predicting the behavior of members that are
following the AASHTO and ACI recommendations. STMs are loaded within the D-region. Applying this methodology may
rarely used in the design of new girders, but this methodology remove restricted load capacities of bridges where shear is the
can prove very useful in design and in the analysis of prestressed controlling failure mechanism.
concrete girders. In addition to comparing the experimentally determined capac-
The STM is an idealized model of a girder’s load paths consist- ity with the code-predicted values, the experimental values were
ing of struts that are compression members made of concrete par- also compared with a detailed finite-element model by using
allel to the expected cracks, ties, or stirrups, which are tension ANSYS. The concrete girder, shear reinforcement, and prestressing
members made of steel analogous to the reinforcement, and nodes strands were modeled by using solid, shell, and link elements,
made of concrete, which are connecting members. respectively. The properties assigned to the concrete and shear
For this research a STM was developed by using two struts and reinforcement were on the basis of experimentally determined
one tie connected at three nodes. This configuration formed a sim- values. The magnitude of residual prestress force for the model
ple triangular truss, which was analyzed to obtain ultimate shear was on the basis of the results of the cracking tests (Table 2). Each
Bridge plan specifications 782.9 176.0 55.6 12.5 108% (248.3 kip). This represented an increase of 52%. Subsequently, the
Cracking test (average) 727.3 163.5 0.0 0.0 100% shear-steel spacing for the Girder 6 model was decreased from
58.4 cm (23.0 in.) to 43.2 cm (17.0 in.), and the ultimate shear
capacity was additionally increased to 1,174.4 kN (264.2 kip). This
Table 3. Measured and Estimated Shear Capacities for Girder 6 new capacity represented an increase of approximately 10% in
(48.9 MPa) and Girder 7 (64.1 MPa) comparison with the original capacity. These results show that
Girder Girder the difference in concrete compressive strength between the two
Percentage Percentage girders had a much larger influence on shear capacity in compari-
6A shear 6B shear
of of son with the differences in stirrup spacing.
Method kN kip measured kN kip measured
AASHTO 366 82.3 55% 366 82.3 47%
simplified Summary and Conclusion
AASHTO 274.4 61.7 41% 274.4 61.7 35%
nominal
Seven decommissioned bridge girders were tested to investigate
the effective prestress force and shear capacity of 42-year-old,
Strut-and-tie 616.3 138.6 92% 616.3 138.6 78%
prestressed concrete, bridge girders. This set of girders was com-
Measured value 668.3 150.3 786.7 176.9
posed of two span lengths and reinforcing designs. A cracking test
was performed on Girders 1 through 5 to determine the residual
finite-element model was loaded incrementally up through failure. prestressing force in the prestressing strands. The experimentally
The calculated deflections at each load step were compared with determined prestress losses were then compared with estimated val-
the measured values and are shown in Fig. 7. ues that were calculated on the basis of procedures recommended in
Fig. 7 shows a very close comparison between experimental and the AASHTO LRFD specifications. Ultimate shear tests were also
analytical values. In all cases, the load-deflection plots showed performed on Girders 6 and 7 to quantify the shear capacity of the
an initially stiff region that was reduced once cracking occurred. girders with the applied loading near the D-region (a=d ¼ 1:5). The
The shear capacity from the analytical models was selected as measured shear capacities were compared with estimated values
the largest applied load and is shown on the figure as the last plotted that were calculated by using sectional procedures recommended
value. For both girders, the maximum difference in ultimate capac- in the AASHTO LRFD specifications. In addition, the measured
ity between experimental and analytical was less than 5%. shear capacities were compared with estimated values by using
The experimental results showed a large difference in shear strut-and-tie methodologies. The following conclusions were ob-
capacities between Girder 6 [727.7 kN (163.6 kip)] and Girder tained on the basis of the research results: