Negative Moment Capacity in early Reinforced Concrete T-Beam

Bridges

Geoff Taplin and Rladh Al-Mahaidi, Monash University

Geoff Boldly, Armando Giufre & David Payne, VicRoads Design

SYNOPSIS

The behaviour of a continuous reinforced concrete T-beam bridge was investigated to assess
the manner in which bending moments were redistributed from the support to midspan at the
ultimate limit state. The results of a full scale and a half scale beam test are presented.

1 INTRODUCTION

An extensive experimental and theoretical investigation is being undertaken by VicRoads and

Monash University to assess the ultimate strength of reinforced concrete T-beam bridges.
The early focus of this work was on the shear strength, as this was often deficient when
assessed using current code provisions. As a result of the work that has been carried out to
date, recommendations have been developed and implemented for the application of advanced
methods for shear strength assessment of T-beam bridges. Further work on shear strength
assessment is outlined in a companion paper. The application of advanced methods for shear
strength assessment has resulted in a number of bridges being re-classified as having
satisfactory shear strength. As a consequence, inadequate flexural capacity has now become
the control on the rated strength of these bridges.

Frequently the early T-beam bridges have adequate flexural capacity in positive moment
regions, but insufficient and poorly detailed negative moment reinforcing. The rated strength
of the bridges can be improved if significant redistributions of moment from the negative to
the positive moment regions are allowed.

This paper reports on the negative moment redistribution that was observed in the full scale
testing of the Baranduda Bridge, and in the testing of two half-scale T-beams in the Civil
Engineering Laboratory at Monash University.

2 NEGATIVE BENDING BEHAVIOUR OF THE BARANDUDA BRIDGE

The Baranduda Bridge on the Kiewa Valley Highway is a reinforced concrete T-beam bridge
constructed around 1916. It has three spans of approximately 7.7 metres, with piers, beams
and abutments cast monolithically. In 1997 VicRoads tested the bridge for the purpose of
assessing the shear strength. The bridge testing and shear strength results were presented at
the 1997 Austroads Conference (1,2) and at the 2000 Transportation Research Board (USA)
Bridge Engineering Conference (3). A general view of the bridge during testing is shown in
Figure I.

275
 Figure 1: Baranduda bridge during testing

Although the bridge was principally tested to establish its shear strength, valuable information
was also obtained on the flexural behaviour of the bridge. In all, eleven separate load cases
were applied to the Baranduda Bridge. The first ten cases involved loads in an end span. In
all of these cases, rotations of the abutment and of the ends of the beams were measured.
These measurements indicated that the beam was behaving as simply supported at the
abutment, even though the joint was monolithic. This was explained when the joint was
broken out after testing and it was discovered that, for this particular bridge, there was no
continuity between the abutment reinforcing and the beams (Figure 2). Therefore, no
negative moments were developed at the ends of the beams.

Figure 2: Detail showing discontinuity of reinforcing at the abutment

276
As well as measuring rotations at the abutment, rotations were measured at the pier to beam
joint in the loaded end span. These measurements indicated that this joint was only effective
in transferring approximately 50% of the beam rotation into the pier. Figure 3 plots the
rotations for the case of midspan loading of an end span beam, in the linear elastic range. It
can reasonably be assumed that, although the bridge was constructed monolithically with the
abutment and piers, the design intent was for a simply supported continuous bridge.

200 _
180 _
160
140
120 _
100
0 80
60 beam rotation
40
pier rotation
20
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
rotation (radians)

Figure 3: Rotations of the pier and the beam in a nominally monolithic joint

In the final of the eleven load cases each of the four T-beams was loaded at 1.81 metres into
span 2 (ie at quarter span). The beams were loaded uniformly, so as to simulate a line load
across the deck. The purpose of this loading was predominantly to establish the shear
strength of the T-beams, as well as to generate significant negative bending at the pier. Some
of the top reinforcing bars over the pier were strain gauged for this loading case.

Before presenting the values for the measured strains, it is necessary to understand the manner
in which the top reinforcing was detailed. The top bars were formed by bending up the
midspan flexural bars (as was common practice in the early development of reinforced
concrete). The points at which the bars were bent up were relatively close to the support,
much nearer in fact than the theoretical point of contraflexure for an elastic bending moment
diagram. The arrangement is drawn in Figure 4, and Figure 5 shows a plan view of the bars
after testing, with the concrete broken away. Note that bars were not continuous from one
span into the next, rather the bars from each span terminated at the pier, with an overlap
length approximately equal to the pier width. While the bending down of the top bars before
the point of contraflexure is common in early T-beam bridges, it is not considered that
terminating the bars at the pier is a common detail. All bars were plain bars.

As a consequence of the fact that the bars bend down before the point of contraflexure, the
full elastic negative moment cannot develop. Once the cracking moment of the concrete at
the bend down point is exceeded in negative bending, a 'hinge' effectively develops at that
location, and the beam acts as a simply supported beam between the points at which the top
bars bend down. This behaviour was predicted by non linear finite element analysis of the
beam. In the finite element model the actual arrangement of the bars was modelled discretely,

277
 with all of the bent-up bars being accurately represented. It was assumed in the modelling
that there was no slip between the bars and the concrete. Based upon these assumptions the
maximum strain predicted in the top bars was approximately 400 microstrain. The measured
strains, plotted in Figure 6, show that values as high as 900 microstrain were recorded in three
of the instrumented bars, while the fourth bar recorded much lower strain. This may be partly
due to the locally reduced cross section where the concrete was broken away for fixing of the
strain gauges, or it may be partly due to slip in the reinforcing bar.

To better understand the behaviour of these beams in negative bending, two half scale two-
span beams were built and tested in the laboratory.

2 x 28 mm dia. 660mm 140mm

1 x 28 mm dia.

Figure 4: Arrangement of the flexural reinforcing bars at the pier

Figure 5: Photograph of the flexural reinforcing bars at the pier

278
 800
700 .0 -
Att
.
load on the beam (kN)
600 A'
500
400
. . gauge 1
300 A gauge 2
200 _m__ gauge 3
100 . A . - gauge 4

0
0 200 400 600 800 1000
microstrain

Figure 6: Measured strains in top bars over the pier Baranduda Bridge

3 EXPERIMENTAL INVESTIGATION OF THE NEGATIVE BENDING

BEHAVIOUR

Following the results of the Baranduda test, it was decided to test two half scale beams under
controlled conditions in the laboratory, in order to monitor the redistribution of negative
moment that occurred as a consequence of the bars being bent down closer to the support than
the theoretical point of contraflexure for an elastic bending moment diagram. The reinforcing
details were faithfully replicated from the Baranduda Bridge. Figure 7 shows the two beams
before casting, with the bent-up bars clearly visible.

Figure 7: Half scale two-span beams prior to pouring showing the top bars

279
 In order to investigate negative moment behaviour, it was beneficial to have the load applied
on one span only, and see how effective moment transfer was into the adjacent span.
Therefore, different loading arrangements were used for the two different beams, as illustrated
in Figures 8 and 9.

1715 1535 1535 1715

load load

3000 3000

dimensions in mm

Figure 8: Test set up for symmetric loading

1715 1535 1535 1715

.4 /4 41

load reaction

A
3000 3000

dimensions in mm

Figure 9: Test set up for one span loaded

The mechanism for moment redistribution will be explained by referring to the results from
the symmetrically loaded beam test. The elastic bending moment diagram for central point
loads is simply obtained as in Figure 10.

P (kN) 0.558P (kNm) p "

0.477P (kNm) 0.477P (kNm)

><
3m 3m

Figure 10: Elastic bending moment diagram for symmetric loading

280
As described previously, the top bars in the beam bend down close to the support, so that at a
distance of 0.5 metres from the support (in the half scale model), there is minimal effective
negative reinforcement. Therefore, once the concrete has cracked, the beams behave as if
pinned at this position (for negative bending). The effect that this has on the bending moment
diagram can be modelled by analysing the structure with pins inserted at 0.5 metres from the
support (Figure 11). The negative moment redistributes to the midspan region.

P (kN) P (kN)
0.293P (kNm)

0.607P (kNm) 0.607P (kNm)

3m 3m
Figure 11: Elastic bending moment diagram - with pins at 0.5 m from the support

From the measured strains in the flexural reinforcing it is possible to calculate the bending
moment, assuming that the cross section is either cracked or uncracked. This was done for
both the span cross section (ie at the load - Figure 12) and the support cross section (Figure
13), using realistic stress-strain curves for the steel (as measured), and the concrete.

Referring to Figure 12, the span bending moments (ie the bending moments under the load)
were calculated from the measured strains assuming that the cross section was cracked
throughout the entire load range. This is a good approximation, as cracking at midspan
occurred once the load reached 10 kN. Referring to Figure 13, the support cross section
remained uncracked until the load reached 62 kN, and so the support moments were
calculated assuming that the cross section was uncracked up to 62 kN, and cracked for all
loads in excess of 62 kN.

The validity of this interpretation of the measured strains can be checked by summing the
support and span moments that were calculated from the strains. The load was applied at
1.465 metres into the 3 metre span, therefore for equilibrium,

\ 1.465 P x 1.465 x 1.535

(span moment) + (sup port moment) x = 0.75P eqn. I
3 3

This was plotted in Figure 14, where the span moment was taken from Figure 12, and the
support moment was taken from Figure 13. The results show that equilibrium was maintained
throughout the loading, as the moments redistributed from support to midspan, and the
moments calculated from the measured strains are accurate.

The behaviour of the beam will now be explained with reference to Figures 10 to 13.

281
 based on measured
90 strain (assumes
cracked section)
5 80
based on elastic
70 analysis - no internal
pins
v_.: 60
E based on elastic
50 _ analysis - with
internal pins
40
RE;
9
4 30
...g 20 .
0.
10
0
0 20 40 60 80 100 120 140
load per span, P (kN)

Figure 12: Load versus span bending moment

90 based on measured
strain (assumes
3 80 section uncracked
then cracked)
*V 70 .based on elastic
analysis - no internal
0 60 pins
5
to 50
based on elastic
"""
.0
.- 40 analysis - with
internal pins
t 30
0
" 20
10
0
0 20 40 60 80 100 120 140
load per span, P (kN)

Figure 13: Load versus support bending moment

1. Initially the beam was uncracked.

2. At a load of 10 kN the beam cracked at midspan. From 10 kN to 60 kN the beam behaved
as predicted by linear elastic continuous beam analysis the span moments calculated
from the measured strains agree well with the linear elastic analysis results, while over the
supports the moments calculated from the measured strains are following the linear elastic
analysis results, although there is an offset due to low strain readings at low loads.
3. At 60 kN the beam cracked in the negative moment region, at the location where the top

282
 bars turned down in span 1. Once cracking occurred at this section, the beam was
effectively pinned here for negative bending. This lead to a redistribution of moment
from the support to midspan, so (as can be observed in the moments calculated from the
measured strains) the span moment increased, and started to exceed the value based upon
a linear elastic analysis. The reduction in negative moment was dramatic, the moments
calculated from the measured strains being less, in fact, than those determined from a
linear elastic analysis model of the structure with internal pins at 0.5 metres from the
support. Cracking continued as the load increased to 77 kN, and the span bending
moment continued to diverge from the linear elastic analysis value.
4. At 77 kN the beam cracked in the negative moment region at the location where the top
bars turned down in span 2. Now the formation of internal pins was complete. The span
moments calculated from the measured strains agree well with those determined from a
linear elastic analysis model of the structure with internal pins at 0.5 metres from the
support. The support moments are less than those predicted by this model, but the support
moments calculated by the model are particularly sensitive to the location of the internal
pins, and it may be that the assumption of 0.5 metres from the support for the location of
the pins is not the best representation of the physical location in the beam.

90 _
80
Z
70
60
5 50 sum of moments based
on measured strains as
40 per eqn. I
t 30
20 . simply supported
moment (0.75P kNm)
10
E 0
0 20 40 60 80 100 120 140
load per span (kN)

Figure 14: Equilibrium check on moments calculated from measured strains

It is considered that this explanation of the beam's behaviour accounts well for the observed
redistribution of moment from support to midspan. It is very important to note that the
mechanism for redistribution is solely related to the detailing of the reinforcing. It is not
attributable to yield in the reinforcing, which is the conventional mechanism for redistribution
in reinforced concrete. The bars over the supports did not yield. At failure (which was at
midspan) the maximum strain in any of the bars over the supports was 900 microstrain. Yield
occurred at 1600 microstrain. For this mechanism of redistribution the ductility of the steel is
not a consideration. Crack widths over the support are the limiting factor. A large proportion
of early T-beam bridges have detailing similar to the Baranduda Bridge (and replicated in the
half scale tests), in that the reinforcing over the support turns down before the point of
contraflexure in the elastic bending moment diagram.

283
 The redistribution is summarised in Table 1.

Table 1: Comparison of measured moments and moments from elastic theory

peak moment calculated peak moment based upon

from elastic theory measured strains (kNm)
(no internal pins) (kNm)
span moment 60 (at P = 125 kN) 79 (at P = 125 kN)
support moment 70 (at P = 125 kN) 24 (at P = 62 kN)

4 CONCLUSIONS

The redistribution of moment from support to midspan was determined by the reinforcing
detailing that was used, and did not require yielding of the reinforcing over the support.
Because the redistribution was caused by the detailing of the negative reinforcement and not
by the yielding of the bars, the ductility of the reinforcing steel, although excellent, was
irrelevant. The amount of redistribution that could occur was limited by the positive moment
capacity, and by the acceptability or otherwise of cracking at the support, but not by the
ductility of the negative moment region. The detailing in the Baranduda Bridge, and
replicated in the half scale beam tests described here, whereby the reinforcing over the
support turns down before the point of contraflexure in the elastic bending moment diagram,
is found in a large number of the early T-beam bridges.

REFERENCES

1.TAPLIN, G. & AL-MAHAIDI, R. "Theoretical analysis of the reinforced concrete T-beam

bridge at Baranduda" Austroads Bridge Conference, Sydney, December 1997
2. GIUFRE, A. "Ultimate load testing of the Baranduda Bridge" Austroads Bridge
Conference, Sydney, December 1997
3. AL-MAHAIDI, R., TAPLIN, G. & GIUFRE, A. "Load distribution and shear strength
evaluation of old concrete T-beam bridges" National Research Council Transportation
Research Board Fifth International Bridge Engineering Conference, Tampa Florida, April
2000

284