Plate Girder Bridge

What a plate girder bridge actually is

A plate girder is an I-beam too deep to roll in a steel mill, so it is welded together from three flat plates: two horizontal flanges and one vertical web. Lay one or two such girders side by side, drop a concrete deck slab across them, and you have a plate girder bridge — the most common medium-span configuration on the world's highway network. Drive a freeway anywhere in the United States, Europe, Japan, or Australia, and you cross plate-girder overpasses every few kilometres without giving them a second thought, which is exactly the point: they are cheap, fast to erect, and span the awkward 30 to 100 m range that rolled sections cannot reach and trusses or cable-stayed bridges would over-build.

The Pratt and Warren trusses that once dominated this range are now rare in new construction. Welded plate girders, made possible by automated submerged-arc welding in the 1950s and validated for fatigue by half a century of service, have replaced them. A modern interstate overpass typically uses two to four parallel girders, each 1.5 to 3 m deep, fabricated in 25 to 40 m segments, trucked to site, and field-spliced with high-strength bolts before the deck is cast.

Why an I, not a rectangle

Bending stress in a beam varies linearly with distance from the neutral axis, so the extreme fibres at the top and bottom feel the highest stress and the centreline feels nothing. Section modulus S = I/c quantifies how much bending moment a section can carry per unit of material, where I is the moment of inertia and c is the distance to the extreme fibre. Place steel far from the neutral axis and I grows as the square of the depth; the section modulus rises with it. A 50 mm × 1500 mm flange placed 750 mm above the centreline contributes Ad² = 75000 × 750² ≈ 4.2 × 10¹⁰ mm⁴ to the moment of inertia by itself — more than a solid rectangle of the same total weight could manage.

The I-section is the mathematical winner: thick flanges far apart, with just enough web to keep them at the right distance and to carry shear. The web typically holds 10 to 20% of the cross-sectional area and contributes 10 to 20% of the bending capacity; the flanges carry the rest. That is also why a deep girder is far more efficient than a shallow one — depth costs only web steel, but pays off as depth-squared in bending capacity.

Forces in the cross-section

At mid-span of a simply supported plate girder under uniform load, bending moment is maximum and shear is zero. At the supports, the reverse: maximum shear, zero moment. Each fibre of the cross-section feels both stresses, but the magnitudes vary as you walk along the bridge.

Component	Role	Stress carried	Typical thickness
Top flange	Compression block in positive moment	Up to fy (350 to 480 MPa)	25 to 75 mm
Bottom flange	Tension block	Up to fy	25 to 75 mm
Web	Shear transfer between flanges	τ up to 0.6 fy	10 to 20 mm
Transverse stiffener	Prevent web shear buckling	Stabilising — no service stress	10 to 16 mm
Bearing stiffener	Take concentrated reactions over supports	Compression to support reaction	16 to 30 mm
Shear stud (on top flange)	Lock deck slab to flange for composite action	Horizontal shear in stud cluster	19 to 22 mm Ø

The two flanges work like a couple: the top is shoved together along the bridge's length while the bottom is stretched. The web has to ferry vertical shear between them, but does not need to be thick — most of the time it is only just thick enough to avoid buckling, with transverse stiffeners spaced every 1.5 to 3 m to break the web into shorter, more stable panels.

Going composite — the concrete deck joins the team

A bare steel girder bends about its own neutral axis. Now imagine a concrete deck slab 200 to 250 mm thick sitting on top. If the slab is merely resting on the flange — friction only — it slides relative to the flange as the girder bends, and contributes nothing. But weld 22 mm stud connectors to the top flange at 150 to 300 mm spacing, embed them in the slab, and the slab and girder become a single composite section. Strain compatibility forces the slab to compress with the top flange; the neutral axis shifts upward into the slab; bending capacity nearly doubles.

Why does composite help so much? The slab's effective width — typically 8 to 12 times the slab thickness on each side of the girder — is added to the compression zone with its concrete strength (f'_c ≈ 28 to 40 MPa) acting like a wide, thick top flange. The neutral axis moves up, the bottom flange now sits at a larger distance c from the neutral axis, and section modulus on the tension side jumps. The result: a 1.5 m deep composite girder reaches the bending capacity of a 2.5 m deep bare steel girder, saving steel weight and reducing girder depth, which means less material in the abutments and approaches.

Worked example: a 40 m span composite plate girder

Two-girder composite bridge: 40 m simply supported span, 9 m deck width, two girders at 5 m spacing, 200 mm composite deck slab. Dead load (slab + barriers + future overlay) ≈ 12 kN/m². Live load (HL-93 lane + truck) governs at design.

Tributary load per girder:
  w_DL = 12 kN/m² × 4.5 m = 54 kN/m
  w_LL = 9.3 kN/m (lane) + 145 kN truck wheel × distribution

Design moment at midspan (factored):
  M_u  = (1.25 × 54 + 1.75 × ~26) × 40² / 8
       ≈ (67.5 + 45.5) × 200
       ≈ 22,600 kN·m

Provide composite section: flange 600 × 50, web 1500 × 14, flange 800 × 60
Composite plastic moment capacity (with 9 m × 200 mm slab, f'_c = 35 MPa):
  M_p ≈ 26,000 kN·m  →  M_u / M_p ≈ 0.87  ✓

Live-load deflection check:
  Δ_LL ≤ L / 800 = 40,000 / 800 = 50 mm
  Computed Δ_LL ≈ 32 mm  ✓

Camber for dead load:
  Δ_DL ≈ 65 mm   →   girder fabricated 65 mm upward at mid-span

The composite cross-section is asymmetric — wider bottom flange than top — because the concrete slab supplies most of the compression capacity, so only a modest top flange is needed for stability and bearing on the slab. The bottom flange does the heavy tension lifting. Camber is built into the girder so it lands flat once the deck and overlays have settled in.

Erection — the moment everything is most dangerous

A plate girder is most vulnerable not in service, but during the few hours between sitting it on its bearings and casting the deck. With no slab and no lateral bracing yet in place, the unbraced top flange can buckle sideways under its own weight, twisting the girder out of plane — lateral-torsional buckling. Temporary cross-frames between girders, ground bracing to falsework, or cable stays are all standard remedies. The girders are usually delivered to site in 25 to 40 m segments matching truck and rail clearances, lifted by crane, set on the bearings, and field-bolted with high-strength A325 or A490 bolts at the splice plates.

Once all girders are set and bridged with cross-frames, formwork hangs below the top flanges and the deck slab is poured. The slab cures, the shear studs lock in, and the composite section is born. Until that moment, the bare steel must carry its own self-weight, the wet concrete, the formwork, and any construction live load — a load case that often controls the design of the top flange and the lateral bracing.

Variants and detail choices

Number of girders. Two-girder bridges are economical for narrow decks (under 12 m) and use deeper girders. Multi-girder layouts (four or more) spread the load to shallower, lighter girders and tolerate damage to a single girder without collapse — important for redundancy in highway service.

Skew and curvature. Plate girders accept moderate skew (the bridge crossing the road below at an angle other than 90°) without redesign, but heavily skewed or horizontally curved bridges introduce torsion the open I-section is poor at carrying. Curved plate girder bridges need diaphragms between girders to redistribute torsion, and very tight curves switch to box girders for their closed-section torsional stiffness.

Continuous spans. Two-span and three-span continuous plate girders are common over rivers and rail crossings: a continuous girder is more efficient than two separate simple spans, but the negative moment over interior supports puts the top flange in tension and the bottom flange in compression — which means the deck slab no longer helps as a composite flange there, and the bottom flange must be sized for compression with the web stiffened against buckling. Many designs use heavier bottom flanges and additional longitudinal web stiffeners in the negative-moment regions.

Hybrid girders. The web is loaded only in shear and contributes little to bending. So use a lower-strength steel (350 MPa) for the web and a high-strength steel (480 to 690 MPa) for the flanges, where it pays off. Hybrid plate girders save 10 to 20% of total steel weight on long spans.

Weathering steel. Many plate-girder bridges now use weathering steel (ASTM A588 or Corten), which forms a tight oxide skin that protects against further corrosion — eliminating painting in most exposures. Drainage detailing is critical to avoid persistent wetting at deck joints and under bearings, which can still rust through.

Common failure modes

• Lateral-torsional buckling during erection. Unbraced top flange bows sideways and twists. Prevention: temporary cross-frames, ground bracing, or sequential erection so each girder is braced to the next before lifting the load off the crane.
• Web shear buckling. Slender webs under high shear at supports buckle in a diagonal tension-field pattern. Transverse stiffeners and properly sized end panels are the standard control.
• Fatigue cracking at weld details. Stiffener terminations, cope cuts, splice plates, and stud welds are all classified by AASHTO into fatigue categories. Category E details (poor fatigue performance) are now mostly designed out; Category B and C details (continuous welds, smooth transitions) dominate modern fabrication.
• Deck-joint leakage. Water and de-icing salt running through expansion joints onto bearings, girder ends, and pier caps causes the corrosion that drives most plate-girder bridge replacements. Joint-less integral abutments are now preferred where span length permits.
• Fracture of a tension flange. A single critical flange fracture (e.g. due to undetected fatigue crack or weld defect) can cause partial or full collapse in a non-redundant two-girder bridge. Modern designs favour redundancy (3+ girders) or use fracture-critical steel grades and supplemental inspection.
• Bearing failure. Worn elastomeric or roller bearings lock up and transmit unintended forces into the substructure. Periodic bearing inspection and replacement is part of standard maintenance.

Real-world plate girder bridges

• Sydney Harbour Bridge approach spans (Australia, 1932). Steel plate-girder approach viaducts feed the main arch — among the earliest large-scale plate-girder applications still in service.
• I-35W St Anthony Falls Bridge (USA, 2008). The replacement for the 2007 collapsed truss is a concrete box girder bridge, but the surrounding interstate uses dozens of composite plate-girder overpasses.
• Akashi Kaikyo approach viaducts (Japan). The world's longest suspension bridge's approach structure uses continuous steel plate girders up to 100 m span.
• UK Highways Agency M25 / M1 / M6 overpasses. Tens of thousands of standard composite plate-girder bridges form the structural backbone of the British motorway network.
• Train viaducts on Japan's Shinkansen high-speed lines. Continuous welded plate-girder spans dominate flat-terrain segments where bridge cost rather than span length is the driver.

Common pitfalls

• Underestimating erection loads. The bare-steel construction stage with wet concrete on the slab is often the controlling load case for the top flange and bracing. Designing only for service load can lead to costly retrofit during construction.
• Skipping the negative-moment composite check. Over interior supports of continuous spans, the deck is in tension and provides no composite help — the bottom flange does it alone, and longitudinal slab reinforcement carries the tension contribution.
• Ignoring stud shear strength under fatigue. Headed shear studs have a finite fatigue life, especially under variable-amplitude traffic. Modern AASHTO requires fatigue checks of stud welds, not just static.
• Using full passive resistance at bearings. Bearing stiffeners must transfer the full reaction into the web without the web crippling locally. Designs that skip the bearing stiffener calculation invariably crush the web under heavy reactions.
• Forgetting deck-replacement load cases. Highway decks need replacement every 30 to 50 years. The plate girder must be able to carry the bridge during slab demolition and recasting, often with traffic running on the other half. Stage-construction checks are mandatory.