Sie sind auf Seite 1von 12

Shear and Torsion - A Look at Super T's

Joe Wyche, Associate, Bruechle Gilchrist and Evans Ply Ltd

SYNOPSIS

There are a number of issues arising out of the design of Super T's relating to shear and
torsion. Firstly, the forces generated are sufficiently large to require torsion design, but it is
argued that the Austroads Bridge Code (1) may require excessive reinforcement, or
longitudinal capacity, even if "torsionless" analysis is used. A less conservative alternative is
suggested. It is then suggested that the Code is unconservative in its treatment of compressive
combined shear and torsion capacity, and an alternative approach is offered. Finally end
anchorage of longitudinal forces is discussed, and some deficiencies in the Code provisions
highlighted.

1 INTRODUCTION

1.1 General

In the last six to eight years Super T's have captured a huge share of the market in bridges.
They are an Australian development, and their success is a tribute to the skill of the precasting
industry, because the thin webs and the cluster of prestressing strands make it quite a
challenge to produce the consistent high quality product required. For bridge designers there
are challenges as well. The Super T's have all but superseded multi-beam precast I-girder
bridges, and have captured much of the voided slab market. However they are not quite the
same as an I-girder and certainly not a slab. In many senses they are like the traditional large
box girder bridges, but their behavior falls somewhere between these and the precast I-girder.
They are something new — a multi-box girder bridge. They are extremely efficient for flexure,
but this paper will argue that current design Codes, including Austroads (1), need some
interpretation and maybe extrapolation to design them for shear and torsion. The paper
discusses what Austroads (1) "allows", but it is also a matter of some debate as to whether the
relevant Articles [Articles or Sections refer to Austroads (1)] were actually intended to take
the design paths down which Super T's tend to lead us. The paper is built around a typical
design case for a 34 m span Super T bridge with six girders, and end diaphragms, carrying
T/L 44 trucks and HLP320.

1.2 The Need to Design for Torsion

Figure 1 shows how torsion and bending in the Super T's interact in a real design case.
Matters would be greatly simplified if the analyses produced design torsion values, T*, low
enough to be ignored. There are several conditions which allow this in Article 5.8.3.4 (a), and
the only really relevant one is if

T* < 0.25 (13 Tuc, where Tuc is the "torsion cracking strength".

199
AS J REDUCES
DEFLECTION
INCREASES,
TORSION REDUCES
AND MOMENT
INCREASES

118001.2,
1M 1180010IM

34 M 34M 34M
CASE1 J=100% CASE2 J=0% CASE3 J=20%

MOMENT ENVELOPE

250101M
150KNM

\ 0 KNM
34M I34M MM

CASE1 J=100% CASE2 J=0% CASE3 J=20%

IDEALISED TORSION ENVELOPE

1400 101 1450104 I 1400104

34M 34M MM

CASE1 J=100% CASE2 J=0% CASE3 J=20%

IDEALISED SHEAR ENVELOPE

Figure 1: Interaction of Moment, Torsion and Shear

200
For 1.5 m deep Super T's, Tuc is typically 550 to 700 kNm, which means any torsions
exceeding about 100 kNm from a full elastic analysis have to be designed for. As implied in
Figure 1, this is usually the case for Super T's.

It is tempting to simplify things by analysing the structure as a grillage of torsionless beams ie


make the J value of the Super T's zero. As Figure 1 shows, this will result in increased
longitudinal bending moments, and the analysis then becomes very simple and similar to that
for precast I-girders. Austroads (1) allows such an analysis (Article 5.8.3.2), but then requires
a very substantial amount of "minimum torsion reinforcement" (Articles 5.8.3.7 and 5.8.3.6).
The only other option allowed by the Code is to analyse for torsion and design for it, using the
rules in Section 5.8.3. Either way the box shape of the Super T's demands that torsion be
considered, and the thin walls and longitudinal torsion forces also need special consideration,
especially near the ends of the girders.

2 THE AUSTROADS CODE RULES FOR ANALYSIS

2.1 Design Values for Torsion

What torsion values need we design for? This obviously depends on what comes out of the
analysis, so the question then devolves to what we may assume in the analysis. The results of
three different approaches are given in Figure 1, for a recent case designed by the author.

2.2 Torsion Compatibility Analysis

As previously mentioned, Austroads (1) Article 5.8.3.2 allows "torsionless" analysis (or more
correctly a "Torsion Compatibility" analysis), which the Austroads Commentary (1) states is
based on Collins and Mitchell (2). In fact Collins and Mitchell (2) in their paper refer to
spandrel beams as an example, and one wonders what they might think of the idea of
redistributing torsion to moment across a whole bridge superstructure, where the elements
doing the redistributing are also the flexural elements receiving the load. Nonetheless
Austroads certainly "allows" a Torsion Compatibility analysis, simply based on zero torsional
stiffness, which produces an analysis very similar to precast I-girders. Collins and Mitchell's
(2) Torsion Compatibility Analysis is different from Austroads (1) because it redistributes
only some of the torsion to bending, so it is not a completely "torsionless" approach. What
they allow is a redistribution of the maximum torsion (near the end of the Super T) down to
2/3 of the concrete cracking torsion strength, Tuc, and then associated lower torsions (near the
middle of the Super T) to be designed for at their calculated value. All these values are then
factored up by 1.3. Either way the "minimum torsion reinforcement requirement" is Tuc
(Article 5.8.3.7), and both methods assume that the difference between 0.25 1 Tuc and Tuc
is not significant to the design. This is not true for Super T's.

2.3 Full Torsion Analysis

To do a full torsion analysis the most obvious approach is the traditional one of an elastic
distribution. This assumes the full bending stiffness, I, and torsional stiffness, J, of the Super
T's, and also of the transverse slabs joining them. However, it is now suggested that there is
another approach which falls between the full elastic analysis and the Torsion Compatibility
approach described above. Austroads (1) Article 5.7.2.5 allows "any reasonable assumption"
of relative stiffness in an analysis - so what might be considered "reasonable" for torsional

201
stiffness? Clearly the implication of Article 5.8.3.2 allowing "torsionless" analysis is that
torsional cracking can produce a massive reduction in the torsional stiffness without a
corresponding reduction in flexural stiffness. The point is mirrored in Article 5.7.2.5 (d)
which warns that the reverse is not likely to be true ie if for some reason a part of a structure
uses a reduced stiffness in the analysis, other sections may not in reality be capable of
attracting resultant increased torsions produced by the analysis. So is there any basis, if it is
helpful to do so, to reduce the torsional stiffness, J, of the Super T's, without correspondingly
reducing their bending stiffness, I? Figure 1 includes as its Case 3 the results of analysis in
which J has been reduced to 20% of its original value, while the flexural stiffnesses of the
beams have been left unaltered.

3 MINIMUM TORSION REINFORCEMENT

3.1 Calculated Torsion Values for Super T's

As implied by the results in Figure 1, 1500 deep Super T's are generally unable to meet the
conditions of Article 5.8.3.4 and therefore torsional reinforcement must be provided.
However, it is likely that the calculated values of torsion will be substantially less than the
"torsion cracking strength", Tuc, of the Super T's. The T* values in the example are up to
about 250 kNm with Tuc varying from 560 kN m at the end of the beam up to 655 kNm from
4.5 m into the beam after no strands are debonded. In other cases T* values up to about 400
kNm have been calculated. The torsion varies with

span length
width of bridge (and beam spacing)
degree of skew
vehicle types eg T/L 44, T/L 54, HLP320 or HLP400, and now S/M 1600.

The only certain way to determine a reasonable level of design torsion is to individually
analyse the structure, but it seems likely that only "minimum torsion" will ever be required.

3.2 Elimination of Vuc

Once any torsion reinforcement is required the interaction of torsional strength governed by
steel, Tus, and shear strength governed by steel, Vus of Article 5.8.3.4 (b) must be satisfied
for the critical shear face where torsional and vertical shear are additive. One immediate
implication is that even if T* is zero from a "torsionless" analysis, no part of the "concrete
contribution" to shear strength, Vuc may be counted. This is presumably because a combined
shear and torsion failure may not allow it to be mobilised.

3.3 Longitudinal Force and "Minimum Torsion Reinforcement"

The extra longitudinal torsion force required for minimum torsion reinforcement (Articles
5.8.3.7 and 5.8.3.6) is substantial. For example, for the ligatures as designed for the bridge in
Figure 1, the extra torsion longitudinal forces are as shown in Table I.

202
kN at end kN at centre
Case from Figure 1 (8 = 45°) (0 = 30°) Comments

Casc I, Full J value 987 kN 44 kN includes Collins & Mitchell 1.3 factor
Case 2, Zero J value 593 kN 1198 kN compatible with minimum = Tuc
Case 3, 20% J 592 kN 55 kN includes Collins & Mitchell 1.3 factor

Table I: Longitudinal Forces for Various Compatibility Analyses

Of course Austroads actually requires that all cases be provided with minimum reinforcement
as in Case 2. The choice of values for 0 , the angle of the shear compression field will be
discussed later in the paper.

3.4 Why is "minimum torsional reinforcement needed"?

With Super T's the "torsional cracking strength" Tuc will theoretically never be reached, but
Austroads provides a factor of safety, reflecting that cracking in some form is almost
inevitable at the ultimate limit state. The conservatism of the T* < 0.25 (1) Tuc limit chosen
reflects both this uncertainty, and the uncertainty of the tension strength of concrete. It is
analogous to Article 5.8.1.4, which ensures that where the cracking moment exceeds the
design load, the beam is not left prone to a steel tensile stress failure if the cracking moment is
somehow exceeded.

However it seems to the author that Article 5.8.3.7 requiring minimum torsional
reinforcement is unnecessarily conservative in the case of these Super T's. This is probably
also more generally true for cases where torsion is between the lower limit allowed without
torsion steel and the torsional cracking strength, Tuc.

4 SUGGESTED ALTERNATIVE MINIMUM TORSIONAL REQUIREMENT

4.1 Alternative to Tuc

If a reasonable analysis has been carried out which defines a statically admissable set of
moments, torsions and shears and the beam is designed for the forces calculated, the beam
should not fail if it does crack in torsion. To provide an extra level of assurance, the design
torsion T*, could be increased by a factor of 1.3. Note that Collins and Mitchell (2) proposed
a factor of 1.3 for torsion steel calculated in their procedure, but mainly to control
serviceability limit state cracking in a Torsion Compatibility analysis, used to reduce torsions
which were above the cracking limit anyway.

4.2 Effect on Super T Design

Table I shows the longitudinal forces calculated in this way for Figure 1, Cases 1 and 3,
which compare with the Article 5.8.3.7 forces in Case 2. For Super T's the implication of
Table I at the centre of the beam is obvious, because the 1198 kN force required for Case 2

203
uses up 5 prestress strands out of the limited "budget" available (42, maybe up to 50 for
"special" beams). This must be added to the extra strands required because the moment has
been increased from 10900 to 11800 kNm when the torsion constant, J has been made zero
(see Figure 1). The magnitude of the Case 2 values can be reduced by increasing the truss
angle, and hence the number and size of ligatures, but a large increase in reinforcement is
required to have a significant impact.

At the end of the beam, the analysis with the full J gives the largest force, and, as will be
discussed later in the paper, this force is difficult to anchor. On the other hand this analysis
gives the lowest maximum bending moment, which could be helpful with a limited budget of
strands. However the increased torsions along much of the length of the beam cause larger
longitudinal forces which tend to negate this advantage. It has been suggested earlier in the
paper that torsional stiffness, J can reasonably be reduced relative to bending stiffness, I,
which will give larger bending moments but smaller torsions. This produces much the same
overall design as the full J, but has a much smaller force to anchor at the end of the beam.

5 IMPLICATIONS OF THE STRUT AND TIE SHEAR/TORSION MODEL ON


SUPER T'S

5.1 Background

This paper will not attempt to deal with the theory on which the provisions for shear and
torsion in Austroads (1) are based. However it must be borne in mind that it is a strut and tie
model. Although Collins and Mitchell (2) is referred to in the Commentary, Austroads
appears to be more closely based on the European CEB/FIP (3), which is also referred to in
the Commentary. For source papers on that document, refer to Thurlimann (4 & 5).

5.2 Concrete Compressive Strength

In general the truss formulae in Austroads (1) for Vus, Tus, and longitudinal torsion force are
as in the references cited. The longitudinal shear force is correctly treated by the "D shift"
rule (Article 5.8.1.8 (b)). The exception is the limiting compressive strength values,

Vu.max = 0.2 f c by do + Pv, (where Pv = 0 in Super T's) (Article 5.8.2.6), and

Tu.max = 0.2 f c Jt, which becomes Tu.max = 0.4 f c Am bw for the Super T (Article
5.8.3.3).

These are the same as the CEB/FIP (3) values when the truss angle 0 is 45°, except that the
0.2 in the Vu.max formula is 0.3, and the 0.4 in the Tu.max formula is 0.5. Austroads (1) is
therefore more conservative. Austroads (1) then has the formulation which drives the choice
of 0 towards an upper limit of 45° from a lower limit of 30° by the proximity of V* to
Vu.max (Article 5.8.2.10), and T* to Tu.max (Article 5.8.3.5 (b)). This becomes very
difficult to use where combined shear and torsion is involved (which almost always occurs
with any torsion). Article 5.8.3.3 has the compression failure interaction check, but the
Austroads (1) definition of Vu.max and Tu.max really imply that this valid only when the
truss angle is 45°, and other cases should also be checked. Austroads (1) is also confusing
because it identifies "0 v" for shear, and "0 t" for torsion, although this angle must be the
same on a combined shear and torsion face. Figure 2 illustrates this. These problems are

204
TOPTRUSS CHORD ISRF OR TOP PS
FLUSBENDING COMPRESSION

TRUSS
ANGLE

-31 lE END ANCHORAGE


VERTICAL TIES, SPAC ING
LENGTH
CLOSES WITH LARGER
LOWER TRUSS CHORD SHEAR AND TORSION,AND
STEEPER TRUSSANGLE
COMPRISES SOME OF
BONDED PS STRANDS

Figure 2: Strut and Tie Model on Web Elevation

easily resolved by using the CEB/FIP (3) truss formulae (retaining the extra Austroads
conservatism), and stating that any value of O in the range 30° to 60° may be chosen such that

T*/(4:1) Tu.max) + V*/( (1) Vu.max) <= 1 (Article 5.8.3.3), where

Vu.max = 0.2 f c bv do sin 2

Tu.max = 0.4 f c Am bw sin 2 O.

Using this approach the author has found that 100 mm thick Super T webs give interaction
values up to about 0.8 i.e. they are reasonably highly stressed, but work. It is a simple check
to make, and should be done for each new design.

5.3 Required End of Beam Anchorage of Torsion and Shear Reinforcement

The "D shift" rule for longitudinal shear force is supplemented by the detailing requirement of
Article 5.8.1.8 (c) (i), which requires that a tensile force equal to 1.5 V* be anchored beyond
the "face of the support", where V* is shear at a distance, d from that face. No direct
guidance is given for anchorage of the torsion reinforcement at the ends of the beams.

205
5.4 End Shear

For the 1.5 V* rule there is no requirement for a strength reduction factor, 421), and the author
interprets the requirement as a 45° shear truss with zero concrete contribution (ie Vuc = 0) and
a factor of safety of 1.5, which inverts to an equivalent cl) = 0.67. This then conservatively
ties all the end shear in the lower truss chord, and the 45° reflects the steepening truss angle as
the compression fan forms toward the support, as shown in Figure 2. Because Article 5.8.1.8
assumes the beam is a free end, no longitudinal shear tensile force can be carried in the top
chord. However Super T's in internal spans are different. This because they have a link slab
which can probably carry half this force through to the next span, which may be a bonus if
end anchorage is a problem. Currently, however the author recommends following Austroads
as directly as possible, anchoring all the force in the lower chord, as must occur at the ends of
the bridge.

5.5 End Torsion

There seems no choice but to add the longitudinal torsion force calculated in Article 5.8.3.6 to
the 1.5 V* (or maybe half this for internal span ends), and anchor all this force. The author's
understanding of the free end situation is that the top chord can still carry longitudinal tensile
forces because they are internally balanced within the torsion truss force system, so that
Article 5.8.3.6, which produces either the lower or upper chord force gives the correct value
to be added to the end longitudinal shear force.

A possible way of reducing this total longitudinal force is suggested in Collins and Mitchell
(2), who point out that the torsion shear is in the opposite direction on the opposite face, and
therefore is reduced by the vertical shear, which in turn, they suggest will reduce the total
longitudinal force. Therefore they propose getting the square root of the sum of the squares of
the longitudinal tensile shear and torsion forces, instead of simply adding them. This could
also be used along the length of the beam to increase the "budget" of prestress strands
slightly, although it would need a direct calculation of the longitudinal shear force instead of
using the "D shift" rule. However both Austroads (1) and CEB/FIP (3) require direct
addition of the longitudinal torsion force.

5.6 "Face of the Support" ???

Presumably the reasoning behind anchoring beyond the "face of the support" is the
assumption that over the support compression will have fanned to a vertical stress field (see
Figure 2) which will have already formed at the face of the support, meaning no horizontal
force components will exist beyond that point. This may be reasonably true if the support is
say a 700 mm wide steel top plate on a pot bearing, especially if there is a cluster of post-
tensioning anchors and cogged reinforcing bars on the end of the bridge. Contrast this with a
Super T which sits on a 250 mm wide elastomeric bearing, and relies solely on the partial
length development of those prestress strands which have been left unbonded. Reference to
Collins and Mitchell (2) who describe ideal and less than ideal end "compression fans" and
anchorage leads to the conclusion that Super T's should be treated cautiously, and anchorage
should be developed beyond the centre of the support rather than the face. This is similar to
the strut and tie provisions of Section 12.

It might be argued that at least the torsion part of the longitudinal force need be anchored only
beyond where it has transferred into the end diaphragm. This is a complicated area where the

206
beam is changing abruptly from hollow to solid, and the top slab is debonded for the link on
the internal spans. It is difficult to say precisely how the forces change, and choosing the
centre of the bearing for the torsion node is reasonably certain to be adequate. The problem is
that there is such a short anchorage distance available that even 100 to 150 mm becomes very
important.

Figure 3: Development of Anchorage in Prestress Strand

6 DEVELOPMENT IN PRETENSIONED TENDONS

This is treated in Article 5.13.3.2. Figure 3 shows the force per 15.2 mm strand developed
with length of bonding, allowing for the time/stress changes described in that Article. These
development characteristics look simplistic to the author, especially as Super T's rely on the
partial development of a short segment for such fundamentally important anchorage. For the
most optimistic Case 3 in Figure 1 and Table I, a total force of

1.5 V* (=1.5 x 1400 kN) + 692 kN = 2792 kN.

With 22 unbonded strands to anchor this force, the distance from the end of the beam where
the strands are cropped to the centre of the support is 478 mm. This distance is designed to
the absolute minimum using and even going beyond present Austroads (1992) rules.
However there may be some relief from the following

• there is a case for halving the 1.5 V* in internal spans.

207
These calculations have been performed on idealised envelopes of moment shear
and torsion. Careful consideration of associated actions from particular load cases
may result in less conservative results.
there is a case for using the square root of the sum of the squares of the forces
instead of the simple addition, but this has not been universally accepted, and
would require major changes to the present Austroads (1) approach, which need
thorough justification.
the bond characteristics of the pretensioning strands may actually be improved by
the presence of the vertical compression stress field. On the other hand the
Austroads (1) approach may be simplistic and even unconservative for partial
development lengths.

7 CONCLUSION

From anecdotal evidence the author believes that many designers simply ignore torsion when
designing Super T's. Some use J = 0 values but there may even be others who use the full J
value and still ignore torsion. Both would violate a number of Austroads (1) rules. However
there are a number of aspects of Austroads (1) and many other Codes which cause problems
designing Super T's, which are a unique Australian innovation. A summary of these points
which have been presented in the paper will be given below.

The final conclusion to be drawn is that either Austroads or a leading Road Authority should
provide research funds and brief a team, including a quality engineering academic institution,
to do the research on which to base revised Austroads (1) rules for shear and torsion. Apart
from the benefits to Australia of doing some leading edge research, the sheer volume of the
beams which have been produced and which will undoubtedly be produced in the future
strongly justifies the expenditure.

8 SUMMARY

Super T's must be designed for torsion.


Austroads (1) "torsionless" analysis (making J = 0) still requires "minimum
torsion reinforcement" and associated longitudinal torsion tensile force capacity.
The level of torsion can only be determined by analyzing the individual structure
and load cases.
Only "minimum torsion" levels will be reached, but they still require significant
reinforcement according to Austroads (1).
The critical design case is where the web vertical shear is enhanced by the torsion
shear.
Whenever torsion is present as well as shear, the shear "concrete contribution"
Vuc must be taken as zero - even for "torsionless" analysis, where J = 0.
It may be reasonable (and useful) to analyse for a reduced value of J relative to I
of the Super T's.
Minimum torsion reinforcement is necessary when there are small but significant
torsions less than the "torsional cracking strength", Tuc, to provide a torsion
"safety net" at the strength limit state.
Austroads (1) minimum torsion reinforcement provisions significantly reduce the
moment capacity of Super T's and/or demand a large increase in reinforcement.

208
• The author considers that the Austroads (1) provisions for minimum torsion
reinforcement are conservative and suggests an alternative level of designed
torsion capacity.
• It is considered that there are deficiencies in the Austroads (1) rules for web
compressive failure, especially for the interaction of shear and torsion. It is
suggested that the formulae for web vertical shear crushing and web torsion shear
crushing include the truss strut angle, to correct and clarify this process.
• There is a case for halving end horizontal anchorage for shear for internal span
supports, but Austroads (1) does not permit this as yet.
• Austroads (1) makes no direct provision for end anchorage of torsion
reinforcement.
• The longitudinal torsion forces must be added to the horizontal shear end
anchorage force and the total anchored.
• There may be a case for using less than the total of the two forces if a suitably
complex consideration of the geometry and force interactions in the end zone is
given.
• For Super T's the total horizontal end force should be anchored beyond the centre
of the elastomeric bearing, and not beyond the "face of the support", as Austroads
(1) permits for shear. This is the simple safe approach, and a more complex
assessment would be needed to justify the "face of support".
• End anchorage is provided solely by a partial development of prestress strands.
The Austroads (1) model for this development looks simplistic to use such a
limited length for this important anchorage role.
• It is likely that development lengths of up to 450 mm will be required from the
ends of the beam to the centre of the bearing.
• There is an urgent need for research, including the involvement of a quality
engineering academic institution, to develop new rules for Austroads (1).

9 SYMBOLS

Am torsion cell area (Article 5.8.3.3)


by web shear thickness (Article 5.8.2.6)
bw torsion wall thickness (Article 5.8.3.3)
d effective depth of beam (Article 5.1.6.3)
do depth from compressive face to steel tensile outer layer
fc concrete characteristic compressive strength (Article 5.6.1.1 (a))
I bending stiffness of a Super T
J torsional stiffness of a Super T
Jt torsional modulus (Article 5.8.3.3)
Pv vertical component of inclined prestress cable force
T* design value of torsion
Tuc torsion cracking strength (Article 5.8.3.5 (a))
Tu.max torsion web crushing strength (Article 5.8.3.3)
Tus torsion strength from steel (Article 5.8.3.5 (b))
Vuc concrete contribution to shear strength (Article 5.8.2.7)
Vu.max shear web crushing strength (Article 5.8.2.6)
Vus shear strength from steel (Article 5.8.2.10)
(1) Strength Reduction Factor (Article 5.2.3)
O theta, the compressive strut angle (Articles 5.8.2.10, 5.8.3.5)

209
10 REFERENCES

1 AUSTROADS. Austroads Bridge Design Code. Austroads, Sydney, 1992.

2 COLLINS M.P., and MITCHELL D. "Shear and Torsion Design of Prestressed and
Non-Prestressed Concrete Beams". Prestressed Concrete Institute Journal, PCI, Vol
25, No 5, Sept/Oct, 1980, pp 32-100.

3 CEB/FIP. "Model Code for Concrete Structures". CEB/FIP International


Recommendations, Third Edition, Comite Euro-International du Beton, 1978, 348 pp.

4 THURLIMANN B. "Shear Strength of Reinforced and Prestressed Concrete Beams".


CEB Bulletin d'Information No 126, Comite Euro-International Du Beton, June, 1978,
pp 16-38.

5 THURLIMANN B. "Torsional Strength of Reinforced and Prestressed Concrete


Beams". CEB Bulletin d'Information No 126, Comite Euro-International Du Beton,
June, 1978 pp 39-65.

210

Das könnte Ihnen auch gefallen