Académique Documents
Professionnel Documents
Culture Documents
In the last 50 years, many bridges have been built as composite steel bridge decks, the influence of varying the sized steel
structures with decks of reinforced concrete that are supported by supporting beams has not been considered.9-11 Therefore,
longitudinal steel beams. The presence of the longitudinal steel the effect of varying the size of steel supporting beams on the
beams and the unloaded area of concrete slab cause the loaded loading capacity of the slab was one of the primary areas of
deck slabs to be restrained against lateral expansion. As a result, a
focus in this research. Using an improved QUB model,7 this
compressive membrane thrust is developed. In experimental tests,
the authors built a series of one-third scale steel-concrete paper details a practical design approach to account for the
composite bridge models with several varying structural parameters, enhancing effects of arching action.
including concrete compressive strength, reinforcement percentage,
and the size of steel supporting beams. After comparing the results EXPERIMENTAL PROGRAM
of different models, the influence of these structural parameters on Test models
the amount of compressive membrane action in the deck slab was eval- The one-third scaled models were designed to represent
uated. Furthermore, the improvement of an existing theoretical model
the lowest level of external restraint in the real bridge
provided accurate predictions for the loading-carrying capacities.
(external bay) and to verify the theory under realistic loading
conditions. Another aim was to provide information on the
Keywords: beam; bridge; compressive strength; membrane.
most effective restraint system, thereby enabling the most
efficient enhancement in the load-carrying capacity.
INTRODUCTION
Based on an elastic finite element analysis result, an
Bridge deck slabs in typical beam-and-slab-type bridges
overall length of 2 m (6.7 ft) was chosen for the test models,
have inherent strength due to in-plane forces set up as a result
and the cross-sectional dimensions were typical of a
of the restraint provided by the slab panel boundary
composite steel-concrete bridge deck slab at one-third scale
conditions, including beams, diaphragms, and slab
(Fig. 1). As shown in Table 1, the variables were concrete
continuity. This is known as compressive membrane action
compressive strength, reinforcement percentage, and
(CMA) or arching action.
external restraint stiffness (by varying the minor axis I value—
Although the effect of arching action in concrete bridge Iyy value of edge beam). The typical model consisted of a
deck slabs has been recognized for some time, it is only one-way spanning concrete slab of 50 mm (1.96 in.) thickness
recently that there has been acceptance of a rational treatment of supported by two steel I-beams connected at the end by
arching action in concrete slabs. Some design and assessment channel section diaphragms. Shear studs were simulated as
codes now acknowledge the benefits of CMA. These include 25 x 25 x 50 mm (0.98 x 0.98 x 1.96 in.) steel equal angles
the Department of Regional Development (NI), Design with a spacing of 150 mm (6 in.), which provided the equivalent
Specification for Bridge Decks1; The Canadian Bridge shear area as the full- scale shear studs. The composite effect
Design Code 2 ; and the UK Highways Agency Standard between the steel beam and concrete bridge deck was
BD81/02.3 The latter came about as a direct result of achieved by the steel stud weld on the top flange of the steel
research at Queen’s University Belfast (QUB).4-7 beam. The steel beams were connected by an end diaphragm
Composite steel-concrete bridges are one of the most consisting of parallel flange channel sections of size 125 x
common types of bridge form. The presence of longitudinal 65 x 15 kg (4.92 x 2.55 x 33.1 lb) connected with four 6 mm
steel beams, together with shear stud connectors, provide (0.24 in.) diameter bolts (refer to Fig. 1).
restraint against expansion of the deck slab. As a result, The models were reinforced with bars at one-third scale. In all
compressive membrane forces are developed (refer to of the test models, the steel reinforcement was positioned at the
Fig. 1(b)), which caused an enhancement in loading capacities in middepth of the concrete slab. The 8 mm (0.32 in.) reinforcing
concrete girder-slab-type bridge decks.8 bar with a yield strength of 597 N/mm2 (86,588 psi) was used
In this paper, the effects of several structural variables on in the model with 0.5% reinforcement percentage and 10 mm
the load capacities of the deck slabs of the composite steel- (0.39 in.) reinforcing bar with a yield strength of 560 N/mm2
concrete bridges were studied by experimental tests. A (81,221 psi) in those with a 1.0% reinforcement percentage.
design method is presented for assessing the restraint stiffness that
To understand the effect on the external restraint stiffness
exists in this typical steel composite bridge deck and thereby by the steel supporting beams, two types of steel I-beam were
the strength enhancement due to CMA.
Fig. 1—Typical test model (NTS). (Note: 1 in. = 25.4 mm; 1 kip = 4.448 kN.)
EXPERIMENTAL RESULTS
Observed behavior
Representative crack patterns at failure load for the
different loading levels are presented in Fig. 3. In all the
models—due to the line load—a longitudinal crack occurred
under the load, resulting in an obvious one-way spanning
effect. An increase in restraint stiffness or concrete compressive
strength reduced the ductility of the slab. As a result, the
punching effects became more obvious, and the failure mode
was more localized, brittle, and sudden.
In Models M77SB05, M38BB05, and M69BB05, a brittle
behavior was observed with punching the primary failure
mode. Crushing of the concrete top surface was a significant
aspect of failure for all of the experimental models. The other
Fig. 3—Crack patterns in Models M36SB05, M77SB05, and models, however, did not exhibit brittle failure modes,
M69BB05. particularly Model M36SB05, which exhibited a more
ductile behavior compared to the other slabs, as shown by its
load-deflection responses (refer to Fig. 4).
Model deflections
From the load versus deflection response at the slab
midspan (T3) in Fig. 4, it can be seen that, for the slabs with
higher external restraint stiffness or concrete compressive
strength, a delay in the stiffness degradation was evident.
This characteristic was associated with the deformation
response of a laterally restrained slab.
Figure 5 shows the deflected shape at the midsection at the
same level of applied load (50 kN [11.2 kips]) in the transverse
direction for Models M36SB05 and M38BB05. The deflections
have been magnified by 10 times with respect to the cross-
sectional dimensions. The deflected shapes clearly show the
horizontal deflections in the supporting beams, which is the
result of membrane action. Furthermore, it can be seen that
Fig. 4—Load-deflection response in validated experimental the vertical deflection in the midspan was lower as the size
test. (Note: 1 in. = 25.4 mm; 1 kip = 4.448 kN.) of the steel supporting beams increased.
A uniformly distributed load applied to a beam produced where T is torque, q is uniform load, and h is height of the beam.
a variation in the value of deflection along the length of the The rotational angle in the unit can be given by
beam, but this can be equated to an average value of deflection by
considering the deflected shape. Integrating the expression
h
for the deflected shape and dividing by the base length gives q ⋅ dx ⋅ --- ⋅ dx
the average deflection, and by equating the average dθ = dT ⋅ dx- = -------------------------------
---------------- 2 (6)
deflection in lateral bending to the stiffness of a spring of an GJ GJ
equivalent area, the lateral restraint can be expressed by7
Therefore, the rotational angle in the edge beam can be
obtained from the integration of Eq. (6) between 0 and L (or
EI yb
K b = ζ ---------
- (2) over beff) with the boundary conditions of θ = 0, whereas x =
3
b eff 0 and x = L. Therefore,
12GJ
where A1 is the cross-sectional area of slab outside effective K tor = ---------------- (10)
2 3
width, and A2 is all the areas of diaphragms. S b eff
1- = -----
1- + -----
1- + ---------
1 - or K = -----------------------------------------------------------------------
1
----- (11)
K r K b K d K tor r
( 1 ⁄ K b ) + ( 1 ⁄ K d ) + ( 1 ⁄ K tor )
Effective width
The restraint model for the bridge deck was assumed to be
the predominately one-way spanning of the deck slab and the
assumption that the diaphragms and surrounding area of
slab, in combination with edge beams, were resisting arching
thrust in the direction parallel to the slab span direction,
which is different from the approach by Rankin for slab and
Column.17 This assumption was validated by the NLFEA
model proposed by Zheng18 (refer to Fig. 7). In the model
proposed by Taylor et al.,7 it was estimated that the influence
of the arching force was sufficiently low at a distance equal
to the effective span plus the depth of the slab (Le + h) from
the face of the support. The effective width of slab subjected
to arching forces can be given by
Table 4(b)—Comparison of results of theoretical models and experimental tests—Taylor et al.’s 23 tests
Beam
Model fcu , psi width, in. %As d, in. Pf1, kips Pv1, kips Pf2, kips Pv2, kips Pt , kips Pt /Pp1 Pt /Pp2
D1 15,983 5.9 0.53 1.3 43.1 48.2 42.6 51.4 41.6 0.97 0.98
D2 14,533 5.9 0.53 1.3 41.1 45.3 42.3 48.1 45.0 1.09 1.06
D5 13,663 5.9 0.53 1.0 35.8 36.9 34.6 39.2 33.7 0.94 0.97
D6 14,214 7.9 0.53 1.0 44.5 39.0 44.6 43.1 40.9 1.05 0.95
D7 14,547 3.9 0.53 1.0 24.5 33.6 26.9 36.4 30.3 1.24 1.13
Average 1.06 1.02
Standard deviation 0.12 0.08
Coefficient of variation 0.11 0.07
Note: Pf1, Pv1, and Pp1 are flexural punching capacity, shear punching capacity, and loading capacity predicted by original theoretical models, respectively; Pf2,Pv2,and Pp2 are flex-
ural punching capacity, shear punching capacity, and loading capacity predicted by proposed theoretical models with modified determination methods for effective width; 1 in. =
25.4 mm; 1 kip = 4.448 kN; 1 lbf2 = 47.88 N/m2; and 1 lb = 0.45359 kg2.
Table 5—Comparison of results of theoretical models with other theoretical models in authors’ tests
Model Ppf , kips Ppv , kips PCanada , kips Pkirk , kips Ppark , kips Pt , kips Pt /Pp Pt /PCanada Pt /Pkirk Pt /Ppark
M36SB05 13.1 19.1 11.2 22.2 28.2 13 1 1.16 0.59 0.46
M77SB05 17.3 30.2 16.9 38.2 34.7 17.5 1.02 1.04 0.46 0.51
M38BB05 16.5 20.7 11.2 23.3 33.2 17.8 1.08 1.58 0.76 0.53
M69BB05 22.7 30.6 15.7 35.7 40.3 22.3 0.98 1.41 0.62 0.55
M33SB10 15.1 18.9 13.5 21 35.8 14.4 0.95 1.07 0.69 0.4
M34BB10 19.9 20.4 13.5 21.6 50.8 21.4 1.07 1.58 0.99 0.42
Average 1.02 1.31 0.68 0.48
Standard deviation 0.05 0.25 0.18 0.06
Note: 1 kip = 4.448 kN.
0.43 0.25 f y ⎞ ⎛ M a + M b⎞ ⎛ f y ⎞
P pv = ---------- f cu × ( critical perimeter ) × d ( 100ρ e ) (18) ρ e = ( ρ a + ρ ) ⋅ ⎛ --------
- = --------------------- ⋅ --------- ρ (20)
rf ⎝ 320⎠ ⎝ M b ⎠ ⎝ 320⎠
The term rf is a reduction factor that accounts for a where ρe is the equivalent arching reinforcement percentage,
variation in the shape of the column. A value of 1.15 was ρ is the actual slab reinforcement percentage, and the critical
used for square columns compared to circular columns due perimeter is as follows: b0 = 2(cx + cy + 6d) for a rectangular
to the stress concentrations in the corners. This formula load; or b0 = 4(c + 3d) for a square load; or b0 = π(φ + 6d) for
quantifies the shear punching strength in terms of the a circular load.
“equivalent” area of reinforcement due to the combined According to the flexural and shear punching mode
effects of bending and arching. This quantified the arching predictions, the ultimate strengths of the bridge deck slabs
moment in terms of an equivalent bending resistance, that is can be obtained as
M
ρ e = ⎛ ------a-⎞ ⋅ ρ a (19) Pp = lowest of Ppf and Ppv (21)
⎝ M b⎠
Modification of original model
A correction for the variation in the yield strength was Compared to the predicted capacities from Taylor’s
necessary, however. Therefore, the total “equivalent” percentage approach,23 the proposed analysis of the authors, which
of reinforcement can be described by incorporates the torsional stiffness of supporting beams and
CONCLUSIONS
Based on the aforementioned analysis, the following
conclusions have been drawn:
1. Experimental results indicate that CMA are influenced by
concrete compressive strength and lateral restraint stiffness.
2. The ultimate strengths of the experimental models are
directly influenced by the concrete compressive strength and
lateral restraint stiffness. The reinforcement percentage has
a less significant effect on the capacity of the slab in the
experimental study.
Fig. 10—Correlation of predicted with test results. 3. Current design standards are highly conservative in
(Note: 1 in. = 25.4 mm.) predicting the strength of laterally restrained bridge deck slabs.
4. In the proposed lateral restraint stiffness model, the
an improved effective width prediction, showed a better influence of the torsional stiffness in the supporting beams is
collection with the results of experimental tests (refer to incorporated. It was found, however, that for typical dimension,
Table 4). It was found, however, that the proposed restraint the torsional stiffness was not significant in the influencing the
model, with the incorporation of the torsional stiffness of the overall external restraint stiffness.
edge supporting beams, did not improve the prediction of 5. From studying the results from NLFEA, a more
load capacities significantly. This is due to Ktor being much accurate determination of the effective width (arching zone)
higher in value than Kb. Therefore, Kr was more significantly was achieved. The use of an improved effective width
influenced by the value of Kb. The horizontal displacement provided improved ultimate strength predictions.
due to minor-axis bending of the supporting beams 6. The proposed method, developed from the predictions
(concrete/steel beams) was far larger than the rotational approach of Rankin17 and Taylor,23 combined with the
displacement due to the torsional stiffness of the beams. The restraint model and concept of an “equivalent” area of
modification in determining the effective width, however, arching reinforcement, gave an accurate prediction for
provided better prediction of the load capacities compared to the strength of a wide range of laterally restrained bridge
those using the previous estimation. deck slabs. The method provided consistent but slightly
conservative predictions.
Comparison of proposed method with other
methods in authors’ tests REFERENCES
A summary of five predicted failure loads is given in Table 5. 1. Department of Regional Development for Northern Ireland (formerly
These are based on Kirkpatrick et al.’s approach,6 Park’s Department of the Environment or DOE), “Design of M-Beam Bridge
Decks—Amendment No. 3 to Bridge Design Code,” N.I. Roads Service
approach,25 the prediction method provided by Canadian Headquarters, 1986, pp. 11.1-11.5.
researchers (Desai et al.),11 and the authors’ proposed 2. CAN/CSA-S6-00(R2005), “Canadian Highway Bridge Design Code,”
method. Because the approach from Kirkpatrick et al.6 and Canadian Standards Association, Mississauga, ON, Canada, 706 pp.
Park25 can just provide one punching failure mode, some 3. BD 81/02, “Use of Compressive Membrane Action in Bridge Decks,”
Design Manual for Roads and Bridges, V. 3, Section 4, Part 20, UK Highways
unsafe capacities were predicted using these two methods. Agency, Aug. 2002.
Desai et al.11 provided a rational model to predict the ultimate 4. Taylor, S. E.; Rankin, G. I. B.; and Cleland, D. J., “Arching Action in
loads of the bridge deck with consideration of compressive High Strength Concrete Slabs,” ICE Proceedings—Structures and Buildings,
membrane action and showed a good correlation at low levels of No. 146, Nov. 2001, pp. 353-362.
lateral restraint stiffness but were slightly conservative at the 5. Rankin, G. I. B., and Long, A. E., “Arching Action Strength Enhancement in
Laterally Restrained Slab Strips,” ICE Proceedings—Structures and Buildings,
higher levels of external restraint stiffness. This is due to an No. 122, Nov. 1997, pp. 461-467.
assumption of constant restraint stiffness from the supporting 6. Kirkpatrick, J.; Rankin, G. I. B.; and Long, A. E., “Strength of Evaluation of
beams. In the comparisons with other predictions, the proposed M-Beam Bridge Deck Slabs,” Structural Engineer, V. 62b, No. 3, Sept.
method provided more accurate, reliable, and safe predictions 1984, pp. 60-68.
with an average ratio of test-to-predicted failure load of 1.02. 7. Taylor, S. E.; Rankin, G. I. B.; and Cleland, D. J., “Guide to Compressive
Membrane Action in Bridge Deck Slabs,” Technical Paper 3, UK Concrete
Bridge Development Group/British Cement Association, June 2002.
Comparison with other test data 8. Khanna, O. S.; Mufit, A. A.; and Bakht, B., “Experimental Investigation of
Because the punching capacity is highly influenced by the Role of Reinforcement in Strength of Concrete Deck Slabs,” Canadian
Journal of Civil Engineering, V. 27, No. 3, 2000, pp. 475-480.
loaded area, to fully confirm the accuracy of the proposed 9. Barrington, deV., B.; Hewitt, B. E.; Casgoly, P.; and Holowka, M.,
model, a larger number of test results6,8,9,19,20,23with “An Investigation of the Ultimate Strength of Deck Slabs of Composite
different loaded areas, particularly composite bridge deck Steel/Concrete Bridges,” Transportation Research Record No. 664,
tests in both laboratory and field tests, were used for Transportation Research Board, 1978, pp. 162-170.
comparison. In total, 67 bridge deck slabs were analyzed 10. Mufti, A. A., and Newhook, J. P., “Punching Shear Strength of
Restrained Concrete Bridge Deck Slabs,” ACI Structural Journal, V. 95,
with the following results: the average value of Pt /Pp = 1.09; No. 4, July-Aug. 1998, pp. 375-381.
the sample standard deviation is 0.10; and the coefficient of 11. Desai, Y. M.; Mufti, A. A.; and Tadros, G., “Finite Element Analysis of
variation is 0.09. Steel-Free Decks,” User Manual for FEM Punch Version (2.0), July 2002, pp. 3-8.
The sample was considered to have satisfactorily covered a 12. BS EN10056, “Specification for Structural Steel Equal and Unequal
Angles: Tolerances on Shape and Dimensions,” 1993, pp. 20-56.
wide range of variables and the good correlation obtained has 13. BS 5400, “Steel, Concrete and Composite Bridges,” Parts 2 and 4,
validated the proposed method, as shown in Fig. 10. In the British Standards Institute, London, UK, 1978 and 1990, pp. 20-43.
correlation models, more than 70% of the specimens were 14. ACI Committee 318, “Building Code Requirement for Reinforced