EUROSTEEL 2011. August 31 - September 2, 2011. Budapest, Hungary
COMPOSITE TRUSS BRIDGES: NEW TRENDS, DESIGN AND RESEARCH a
António Reis , José J. Oliveira Pedro a
a
GRID Consulting Engineers and Tech. Univ. of Lisbon – Dept. of Civil E ngineering, Portugal
INTRODUCTION Composite truss bridges are one of the most efficient and aesthetically attractive design solutions in bridge engineering [1]. Structural steel and concrete materials are combined in the most efficient way to highlight the art of structural engineering in bridge design. Composite action in trusses may be explored in a number of different ways, producing a variety of design solutions. The most adopted and efficient solution consists in a under slung truss, where the deck slab acts in composite action with the compression chord. The concept only holds at positive bending moment sections; at internal supports of continuous girder bridges, the deck slab is under tension and no composite action can be considered at ULS. However, double composite action in composite truss bridges may also be adopted at negative bending moment regions. A review on new trends was presented by Reis [2], [ 2], including double composite action in continuous trusses with variable depth as adopted for the Nantenbach bridge [3], triangular tubular deck bridges as adopted in Switzerland [4], with very large overhangs up to 6.0 m, and the Italian hybrid truss with full web systems [5]. Composite box girders may be adopted for wide decks, up to 30 m or even more, using struts to support the overhangs. These struts may be arranged in a longitudinal truss, which participates in the overall resistance of the superstructure [2]. A different concept is proposed in [2] for wide curved decks, by adopting two main vertical Warren trusses and two inclined trusses to support the overhangs. The double composite action solution in truss bridges yields in the limit a box girder with a concrete lower flange, as adopted by the authors for the cable-stayed bridge in Coimbra [6]. Truss decks with a reinforced concrete slab at the bottom chord level are adopted in rail bridges mainly to overcome strict vertical clearance requirements over highways. For road-rail bridge decks, a classical option is a double deck made of two Warren trusses in composite action with the deck slabs. This was the solution adopted for the Oresund Bridge between Denmark and Sweden and it is proposed as Base Case Design for the new crossing of the Tagus River in Lisbon [7]. Truss decks with triangular cross-sections may represent the most efficient composite decks, at least for road and pedestrian bridges where fatigue issues are not so critical. Some issues may be raised about the redistribution bending moment capacities of these trusses from the supports to the span sections. The torsion behaviour of the triangular trusses, and deformability under eccentric traffic loading, are also issues deserving a detailing analysis [2].
1
DESIGN OPTIONS AND NEW TRENDS
Trusses in bridges have been adopted since the first generations of metal bridges. However, in the last two decades some attempts have been made to implement different concepts, mainly related to - composite action with the concrete deck slab; - double composite action with a lower concrete flange at the internal supports; - composite trusses with triangular cross sections reducing the lower chords to a single tube; - hybrid full web trusses (HFWT), exploring the combination of a full plate girder with a triangular truss; - main and secondary trusses, with different arrangements at the cross-section for wide decks. Double composite action in truss bridges has been already adopted for long-span variable depth girders [3]. Composite trusses with triangular cross sections as in Fig. 1. to be discussed in more detail in the next section, have been one of the most interesting developments in the last two decades. The overall shear deformations in composite truss bridge decks at the internal support sections induce relevant secondary bending moments in the chords and diagonal members of Warren trusses.
The Italian concept of a hybrid truss 4.0 4.0 with full web system ( Fig. 2) explores 4.0 the combination of a full plate girder at the cross section axis, reducing shear deformation effects, with a triangular [m] truss to achieve enough torsion resistance under eccentric loadings. The concept Fig. 1. Lully viaduct, Switzerland. Half deck cross-section. had been adopted in road viaducts with Design by Dauner Ingénieurs conseils [4] spans up to 40 m, achieving very low 12.6 m 2 1.6 0.5 5.25 5.25 steelwork quantities, below 100 kg/m . Main and secondary trusses may be combined for wide decks namely for curved bridges where an equivalent box section is needed [2]. For a curved [m] 3.75 3.75 roadway viaduct, with a continuous Fig. 2. HFWT deck system, Italy. Design by REDESCO [5] deck, 20 m width and 60 m long typical spans, the authors have developed a 20 m composite truss deck ( Fig. 3) with transverse prestressing. The main and secondary trusses, supporting the overhangs, are all Warren type trusses. 9.0 [m] Between the lower chords, a horizontal tubular bracing is introduced, forming Fig. 3. Study of a composite truss for a curved roadway an equivalent box type behaviour, under viaduct, Portugal. Cross-section and model. Design by GRID torsion. The chords and diagonals are made of rectangular tubular welded sections, while the lateral trusses are made of circular hollow sections. Avoiding intermediate internal diaphragms for construction simplicity and aesthetics, one of the main issues was the differential displacement between main trusses under eccentric loading. The secondary bending moments developed at the transverse crossbeams, connecting the upper and lower chords, were investigated by a FEM and some results were presented in [2]. The attained steel 2 S355 J2 amount was only 128 kg/m . 12 m
5 2 . 0
0 4 6 3 . . 0 0
9 . 2
2 . 2
5 . 3
2
COMPOSITE 3D TRIANGULAR TRUSS DECKS: DESIGN AND RESEARCH
Composite trusses with triangular cross sections were proposed by Jean Muller and were adopted by the first time, to the authors knowledge, in Roize bridge [8]. The concept was spread through Europe, namely in Switzerland (Lully viaduct – Fig. 1) and in Germany (Korntal-Münchingen bridge and St Kilian viaduct) [9.10]. The nodes, in these tubular structures, may be made by shop welding between chords and diagonals or may be cast steel prefabricated nodes as shown in (Fig. 4). Nodes are designed for ultimate and fatigue resistance under static and dynamic loading. Member chords and diagonal are mainly subjected to axial forces, but secondary moments needed to be considered due to the rigidity of the welded connections, namely for fatigue design. Detailed FEM for the stress analysis under fatigue traffic loading models may be developed. Some issues may be raised with respect to these triangular section tubular composite truss bridges: 1) the shear failure mode of the lower chord near internal supports where large equivalent shear forces in the girder are taken as axial forces in the diagonals ( Fig. 5), 2) the redistribution of negative bending moment at support sections at ULS, and 3) the torsion stiffness capacity of the deck under eccentric traffic loading. Research on the behaviour of composite bridge trusses with triangular cross sections, has been developed at the Tech. Univ. of Lisbon under the supervision of the first author [12, 13, 14]. Due to paper length limits, only the first issue is discussed. Two tests were carried out for bridge models ( Fig. 6 ) at about 1/5 scale, under two point loads to simulate the evolution of bending moments between the support section S1 and the mid span section S2. Loads P1 and P2 were increased independently independently up to the failure, accordingly to the following sequence:
first, loads P1 and P2 were equally increased up to the plastic load capacity of section S1 and, then, load P1 was kept constant and load P2 was increased up to the failure load of mid span section S2. In the first test, the ultimate load capacity at S2 was not reached due to the small amount of reinforcement at section S1. The theoretical load displacement curves ( Fig. 7 ) were evaluated by a nonlinear analysis [12, 13]. The theoretical ultimate load in st the 1 test was about 16% higher than the experimental one. The collapse was initiated by shear failure mode ( Fig. 5) of the bottom chord at support section and then by buckling of the diagonals at this section. The amount of reinforcement was 3 times increased for the 2nd test in order to improve ductility of section S1. The steel diagonals diameter was also increased, and the theoretical ultimate load at S2 could be reached. The theoretical values were P2u= 341 kN for P1u= 206 kN; the experimental load values obtained were P2u= 345 kN for P1u= 210 kN. The theoretical and experimental displacements at the mid span section were 77 mm and 72 mm, respectively. The ultimate load and displacements were the ones predicted by a non-linear elasto-plastic analysis, but as in the first test, the failure mode induces the shear failure mode of the compressed chord at the support section. As a conclusion, a total redistribution of bending moments can be achieved, what may be considered for practical design at ULS, provided the amount of steel reinforcement at the support sections is sufficient for the redistribution, and shear failure mode at the bottom chord near the intermediate supports of continuous girders is properly prevented. Shear failure load of the bottom chord may be increased by either raise its thickness or reduce the gap between diagonals at the nodes.
/ h1 Model H h1 / t t [mm] B/b1/b2 [mm]
Di / ti i=1.2.3 [mm]
Fig. 4. Welded nodes and cast steel nodes for tubular triangular truss bridges [9,11]
Fig. 5. Node shear failure: Test at Tech. Univ. Lisbon [12]
Reinforcement [mm]
Steel/Concrete
Test 1
570/409/70
1100/486/307
114.3/5; 76.1/5; 32/3.5
(φ5+φ5)//100 f sy sy=500MPa f y=320MPa; f c=48.1MPa
Test 2
650/485/70
1300/500/400
114.3/5; 76.1/5; 38/5.6
(φ10+ φ8)//75 f sy y=293MPa; f c=42.2MPa sy=500MPa f
Fig. 6. Test bridge models at the Tech. Univ. of Lisbon [12,13] 280
400
240
Predicted ultimate load
300
200 ]
N k
250
[ 160
d a o 120 L
200
Experimental
Experimental Numerical
150
Numerical
80
100 Load Test 1
40 0
350
Load Test 2
50 0
0
10
20
30
40 50 60 70 Displacement [mm]
80
90
100
0
10
20
30 40 50 Displacement [mm]
60
Fig. 7. Load P2 –displacement chart at mid-span section of Fig. 6 test bridge models [12,13]
70
80
3
SEMI-THROUGH COMPOSITE TRUSSES FOR RAILWAY BRIDGES
A semi-through composite truss of Fig. 8, with a reinforced concrete deck slab at the bottom chord level, is less efficient than an under slung truss by two reasons: r easons: the slab for simply supported spans, or at span sections of continuous decks, is under tension and the cross-girders induce transverse bending of the diagonals of the truss. However, semi-through trusses are currently adopted in rail bridges, since vertical clearances are quite often a main constraint for truss decks. Low slenderness’s of 10 to 12 are currently required, not only to reduce the steel amount in chord sections, but also to satisfy deformability and vibration requirements at SLS in rail bridges. For High Speed Railways (HSR) the limits for maximum vertical accelerations at the deck are only 2 3.5 m/s to guarantee ballast stability under traffic. On the other hand, for a very good or good 2 2 comfort level, the maximum allowable vertical acceleration are respectively 1.0 m/s or 1.3 m/s . Deck vertical deflections up to L /2000 and L /2500 for spans between 40 and and 60 meters under LM71 2 traffic load model, is an indirect way to achieve the limit of 1.0 m/s for train speeds of 350 km/h. For HSR semi through-truss decks, SLS requirements are likely to be the design control criteria. The main issue is how to define the maximum vertical acceleration criterion in the deck. Due to the deformability of the slab, local vertical accelerations resulting from the i nteraction of the global and local vibration modes of the superstructure, tend to be the governing design criterion. The transversal deformation of the slab may be reduced by introducing steel stringers connected to the cross-girders along the alignments of the tracks. The l ocal vertical accelerations are then reduced, in particular in skew decks due to bearing effects at the support sections. In any case, the design 2 criterion for maximum accelerations of 3.5 m/s may be too conservative if interpreted as maximum peak acceleration at any point of the deck. Since it is a ballast stability problem, it appears more logical to adopt a design criterion based on an average acceleration after a few cycles or on an average acceleration at the deck cross-section. Some proposals to control maximum accelerations in 2 the track were made in [15]. A criterion based on a maximum acceleration of 3.5m/s reached after a sequence of 10 cycles, looks more realistic. A recent semi-through composite truss railway viaduct designed for HSR adopted a slenderness, main-span/distance between chord axes, of 13.5 ( Fig. 8). The top and lower chords are welded rectangular tubular sections 800 x400 mm and 800 x600 mm respectively (plate thicknesses between 16 and 60 mm). Welded I sections were used in the diagonals (plate thicknesses between 16 and 80 mm). Steel S355 N or S355 NL was used depending on the thicknesses. The reinforced concrete slab, 0.35 m thick, is supported by 4 stringers and cross girders at every 4.5 m. A dynamic analysis was performed for the High Speed Loads Models (HSLM) specified in the Eurocodes. The maximum peak deck accelerations were usually obtained for 430 km/h. The study 2 has shown a maximum acceleration at the bottom chord alignments of approximately 3.2 m/s . 2 However, the peak acceleration tends to increase to almost 5 m/s due to transverse cross-section deformations. However, a criterion as previously referred to – maximum peak accelerations measured in a sequence of 10 cycles, show an acceptable behaviour. 153,00m 49,50
54,00
49,50
5
) s 4 / m ( n 2 o i t a r e 0 l e ‐ c c a ‐‐ 2 l a ‐ c i t r ‐4 e ‐ V ‐
2
2
+3,5 m/s
4
3
2
) 2 1 s m ( n o i 0 t a r e l e c c A 1
‐
4,0
2
3
‐3,5 m/s2
4
12,95m
5
0
0
1.0 1
2,0 2
3,0 3
4,0 4
Time (s)
5,0 5
Fig. 8. A semi-through composite truss railway bridge. Design by GRID – General layout and peak vertical acceleration vs. time for HSL Models at 430 km/h
4
COMPOSITE TRUSSES FOR CABLE-STAYED RAILWAY BRIDGES
One of the main issues for railway cable-stayed bridges is deformability and vibration under traffic loading. In Table 1 the main characteristics of railway composite cable-stayed bridges are presented. At this stage, only a few bridges for HSR have been built. The Oresund Bridge for two traffic rails and the recently built Chinese Tianxingzhou Bridge with 4 traffic rails are perhaps the most well known projects in this field. Both bridges were built for combined rail and road traffic and double decks were adopted. The Chinese bridge has 3 planes of stays and 3 trusses. With its 504 m main span, it is presently the world record for railway cable-stayed bridges. rd A recent Base Case Design, the so-called 3 Crossing of the Tagus River in Lisbon [7] ( Fig. 9), may be considered one of the most demanding infrastructure projects in this area. The total crossing includes a bridge of about 7.3 km with three requirements for navigation channels – the one near Lisbon requiring a main span of 540 m. A cable-stayed bridge for this main navigation channel is needed, as shown in Fig. 10. A variety of design options have been studied for the river crossing [7]. The retained option was a double deck bridge, with 6 lanes for highway traffic at the upper deck and 4 tracks at the lower deck – 2 for conventional rail traffic and 2 for HSR ( Fig. 9). A ballasted track was the preferred option of the Owner. What makes the Fig. 9. The cable-stayed bridge designed for the structure unique compared to other similar projects 3rd Tagus crossing in Lisbon. Design by GRID [5], is the need to accommodate 4 tracks at the lower deck. The project is presently suspended by financial reasons. The deck of the cable-stayed bridge is a steel-concrete composite Warren truss, with cross girders at a distance of 15 m, and constant height between chord axes of 9.6 m (span/depth ratio of 12.5). Chords and diagonals are made of welded tubular sections in steel S460 M/ML. Transverse bending moments induced by the cross-girders and secondary bending moments in the main chords are 30.2
6 . 9
6 . 1 1
6.9
21.0
Table 1. Rail-road cable-stayed bridges built, under construction or design Name Country
End of construction Main span Deck type Deck Deck width / depth Rail use type Total length suspension type cross-section / Slenderness
Zárate-Brazo Largo I e II Argentina Iwakurojima/Hitsuishijima
Japan Kap Shui Mun Hong-Kong, China Øresund Sweden / Denmark Wuhu China Orinoco River II Venezuela Tianxingzhou China Orinoco River III Venezuela rd 3 Tagus River Crossing Portugal Fehmarn Belt Germany / Denmark
1978 1 Light train 1988 2 HS Railway 1996 2 Light train 2000 2 HS Railway 2000 2 Conventional 2006 1 Conventional 2009 HSR+Conv. Under constr. 1 Conventional Under design 2 HSR + 2 Conv. Under design 2 HS Railway
330 m 550 m 420 m 790 m 430 m 820 m 490 m 1092 m 312 m 672 m 300 m 1200 m 504 m 1092 m 360 m 600 m 540 m 1140 m 724 m 2414 m
One level 2 lateral steel box-girders Lateral with an ortotropic slab Double deck 2 lateral Warren Lateral steel trusses Double deck Composite box girder with Lateral 2 interior Viereendel trusses Double deck 2 lateral Warren Lateral composite trusses Double deck 2 lateral Warren Lateral composite trusses One level Composite box-girder Lateral with external struts Double deck 2 lateral + 1 central 3 planes steel trusses Double deck 2 lateral Warren Lateral composite trusses Double deck 2 lateral Warren Lateral composite trusses Double deck 2 lateral Warren Lateral composite trusses
22.60m 2.6m / 127 29.1m 13.9m / 30 32.5m 7.70m / 56 23.5m 10.2m / 46 21.94m 13.5m / 23 24.70m 5.75m / 52 30.0 m 15m / 34 19.2 m 12m / 30 30.2 m 11.6m / 47 28.7 m 12.9m / 56
Fig. 10. Span layout for the cable-stayed bridge of the 3 rd Tagus river crossing in Lisbon. Design by GRID
particularly relevant near the support sections due to global shear deformations of the trusses. Fatigue resistance mainly controls truss node zones. The stringers of the upper and lower decks participate in the overall bending resistance of the cross-section. The stays have a number of strands varying between 47 and 109 strands of 15 mm each. In what concerns the deformability of the bridge deck, the maximum static deflection under rail design traffic loading is about 1.0 m. The vertical vibration frequency of the deck is 0.31 Hz. A dynamic analysis was made under the HSLM at 120 km/h (reduced speed at the proximity of 2 2 Lisbon) and the maximum vertical acceleration was only 1 m/s , well below the 3.5 m/s limit. The aerodynamic of the deck was tested in a wind tunnel with a sectional model at scale 1/70. The results show a good aerodynamic stability for wind speeds up to 292 km/h, which was defined as the minimum required critical wind speed for aerodynamic instability. The drag aerodynamic coefficient, referred to the depth of the deck, is 0.86 < C D < 1.14 for of wind angles between –5º and +5º. The lift coefficient, referred to the width of the deck, is around –0.10 for the wind flow at 0º. 5
CONCLUSIONS AND ACKNOWLEDGMENT
The aesthetic and structural advantages of composite truss bridges were highlight. Composite triangular trusses were discussed; experimental and numerical results were presented. Some main issues for semi-through composite trusses for HSR bridges were discussed based on a recent design rd case. A double deck composite cable-stayed bridge for the 3 Tagus River crossing in Lisbon was presented. Special thank are due to L. Salvador who developed the dynamic analysis referred in 3. So to former Msc students J. Braz and O. Videira by the experimental studies reported in section 2. REFERENCES
[1] Reis, AJ “Bridge decks: composite systems for improved aesthetics and environmental impact”, Proc.3rd Int. Meeting on Composite Bridges, pp. 645-59. Madrid 2001. [2] Reis, AJ “Steel concrete composite bridges: options and design issues”, 7 th ECCS Int. Conf. on Steel Bridges, pp. I3-28. Guimarães 2008. [3] Saul, R “Bridges with double composite action”, SEI Vol.6 Nº1, pp. 32-36(5). February 1996. [4] Dauner, G., et al. “The Lully Viaduct, A Composite Bridge with Steel Tube Truss”, Journal of Constructional Steel Research, Vol. 46. Nº1, pp. 67-68(2), April 1998 [5] Giulianni, M “Hybrid truss and full web systems”, Proc.3rd Int.Meet. on Composite Bridges, p.79-94. Madrid 2001. [6] Reis, AJ; Pedro, JO “The Europe bridge in Portugal: concept and structural design”, Journal of Constructional Steel Research Vol.º60, Issuesº3-5, pp. 363-72, 2004. [7] Reis, AJ “Two large bridge projects in Portugal in environmentally constrained and urbanised areas”, Proceeds. th 34 IABSE Symposium, Venice 2010. [8] Montens, S, et al. “Le pont expérimental sur la Roize” Bulletin Ponts Métalliques”, nº 15, pp. 69-96. 1992. [9] Hirt, M “Tubular trusses for steel-concrete composite bridges”, Int. Conf. on Fatigue and Fracture in the Infrastructure, EPFL, 2006. [10] Hanswille, G; Sedlachek, G Steel and composite bridges in Germany: State of the Art , Institute for Steel and Composite Structures, University of Wuppertal. Germany 2007. [11] Shumacher, A “Fatigue behaviour of welded circular hollow section joints in bridges”, PhD Thesis, EPFL. 2003. [12] Braz, J “Composite truss bridge decks”, Msc. Thesis, IST - TULisbon, 2009. [13] Videira, O “Composite truss bridge decks”, Msc. Thesis IST - TULisbon, 2009. [14] Almeida, A “Torsion in three-dimensional composite truss bridge decks”, Msc. Thesis, IST - TULisbon, 2010. [15] Zacher, M; Baeßler, M Dynamic behaviour of ballast on railway bridges - Chap. 7 of Dynamics of High-Speed Railway Bridges, Ed. Delgado, R. et al. Taylor & Francis Group, pp. 99-112. London 2009.
