Introduction to Bridge Engineering

Every bridge you will design, inspect, or rate breaks down into the same small set of load-carrying elements. Learning their names, their purpose, and the practical construction issues that go with each is the vocabulary you will use for every conversation with a contractor, DOT reviewer, or client for the rest of your career. This section walks the anatomy top-down — from the wearing surface a car actually drives on, through the girders and bearings, down into the pier, and finally into the foundation.

1. Superstructure vs. substructure — the load path

The single most important idea in bridge engineering is the load path: every wheel load, every pound of self-weight, and every gust of wind must find a continuous path from where it is applied down to competent soil or rock. The bridge is divided into two families of parts that pass this load like a bucket brigade:

• Superstructure — everything above the bearings that carries the roadway: wearing surface, deck slab, primary members (girders / stringers), and secondary members (diaphragms, cross-frames, cover plates, stiffeners). It gathers vehicular loads and delivers them to a discrete set of bearing points.
• Substructure — everything from the bearings down: pedestals, cap beams, piers, abutments, and foundations (spread footings, drilled shafts, or driven piles). It gathers bearing reactions and spreads them into the ground under an allowable pressure.

2. The wearing surface — the sacrificial layer

The wearing surface is a sacrificial layer poured or laid on top of the structural deck. It protects the deck from tire abrasion, chloride ingress (from de-icing salt and coastal spray), UV, and the hydraulic action of tires driving through standing water. Two families are used on US bridges today:

• Bituminous — 50–75 mm of hot-mix asphalt over a waterproofing membrane bonded to the deck. Cheap, fast to place, easy to mill and replace, but the membrane failure rate governs long-term deck durability.
• Latex-modified concrete (LMC) — a 30–50 mm high-density concrete overlay bonded monolithically to the structural deck. More expensive, higher fatigue resistance, typical service life of 25+ years on interstate bridges.

Because the wearing surface is renewable — most agencies replace it two or three times over a 75-year bridge life — AASHTO classifies it as a separate load DW with the higher load factor γ_DW = 1.50 in Strength I, versus γ_DC = 1.25 for the permanent structural dead load. That difference exists because overlays often end up 30–50% thicker than the original design as pavement is added over the years.

3. The deck slab — the structural driving surface

Below the sacrificial wearing surface sits the structural deck slab itself — typically a 200–250 mm reinforced concrete slab cast on stay-in-place metal forms spanning between the girders. Its job is threefold: (i) carry the wheel load between girders in transverse bending, (ii) distribute concentrated tire loads to several girders through membrane action, and (iii) act compositely with the girders as the top flange of the overall girder section, adding a huge amount of moment capacity to the composite girder.

Reinforcement is placed in two mats: a top mat resists negative bending over the girders (and takes the salt-water attack, so it is almost always epoxy-coated — the green bars in the photo), and a bottom mat resists positive bending in mid-panel. Typical spacings are 150–200 mm each way. The deck slab is the shortest-lived structural element on the bridge — national inventories show a mean deck life of 35–45 years before deep rehabilitation.

4. Primary members — girders and stringers

Primary members are the load-carrying beams that span between piers or abutments. They collect the deck reactions and deliver them to the bearings. In US practice you will see three families:

• Prestressed concrete I-girders (AASHTO Types II–VI, Bulb-T, Florida FIB) — precast in a yard, trucked to site, and set with a crane. The workhorse for 20–60 m simple spans. Fast erection, low maintenance, but limited to straight or gently curved alignment.
• Steel plate girders and rolled beams — welded I-sections tailored to the moment envelope. Handle curved alignment, heavy skew, and long spans (30–130 m) but need painting and periodic fatigue inspection.
• Concrete or steel box girders — closed-cell sections with far higher torsional stiffness. Used on curved elevated freeways and for segmental construction from 40 m up to 250+ m.

On multi-girder bridges the girders are laid out at a spacing of 2.4–3.7 m so that the deck slab design becomes a repetitive one-way transverse problem. The word stringer is used for the same idea in truss and floor-beam systems: a stringer is a longitudinal beam supported by transverse floor-beams, which in turn frame into the truss chord.

5. Secondary members — diaphragms, cross-frames, stiffeners

Secondary members do not carry the primary bending moment, but the bridge cannot function without them. Their job is to tie the girders together into a system:

• Diaphragms — solid transverse beams (concrete diaphragms in prestressed girder bridges, or bolted channel diaphragms in shallow steel girders) placed at supports and at mid-span. They distribute wheel loads laterally to adjacent girders, share torsional reactions, and stabilize the girders against lateral-torsional buckling during erection and service.
• Cross-frames — X- or K-shaped bracing of steel angles bolted between the webs of steel plate girders (visible in the photo). Function is the same as a diaphragm but they are lighter and easier to fabricate for deep girders. On curved girder bridges, cross-frames are primary load-carrying members because they resist the twist of the curve.
• Cover plates — extra plates welded to the flanges of a rolled steel beam in the high-moment region only. They add capacity where the moment demands it without paying for a heavier section end-to-end. Historic detail; new designs prefer welded plate girders because cover-plate weld terminations are notorious fatigue points.
• Stiffeners — vertical or longitudinal plates welded to the web of a steel plate girder. Transverse (vertical) stiffeners prevent web shear buckling, especially near bearings and at cross-frame locations. Longitudinal stiffeners are used on deep girders to push up the web-bend-buckling limit.

6. Substructure — piers, bents, and abutments

The choice of pier type is driven by three practical constraints: (i) the height of clearance that must be maintained beneath the deck, (ii) the depth and velocity of water at the site (which drives scour design — see Ch. 14), and (iii) the horizontal loads (wind, ship impact, seismic) the pier must resist. The four types you will see 95% of the time are:

Abutments serve two jobs at once: they support the end reactions of the superstructure, and they act as retaining walls holding back the approach embankment. Three configurations dominate:

Between the girder and the top of the pier or abutment sits a pedestal — a plain-concrete block, usually 100–200 mm high, that raises the bearing off the concrete surface so ponding water cannot sit against it. Above the pedestal, a bearing(elastomeric pad, pot, or rocker/pin) accommodates the rotations and translations produced by live load, temperature, creep, and shrinkage. Bearings are the single most inspected element on the bridge — they wear out, and their replacement drives the deck-joint replacement schedule.

7. Deck joints — where movement is absorbed

Concrete and steel both change length with temperature (approximately α = 11 × 10⁻⁶ /°C for steel and 10 × 10⁻⁶ /°C for concrete). Over a 50 m simple span through a 60 °C temperature range the deck grows and shrinks about ±16 mm. Add creep and shrinkage of the concrete and you routinely need to accommodate ±25 mm of longitudinal movement at each end of the span. The deck joint is the detail that absorbs it.

Selection is a trade-off between movement capacity, water-tightness, ride quality, and life-cycle cost. Every joint you specify becomes a maintenance item — the table below summarises the honest disadvantages of each family so you can pick the one whose failure mode you can live with.

Where possible, modern practice eliminates the joint entirely by making the deck continuous over the piers (link slab, jointless bridge, or integral abutment). Every joint you can remove is one less maintenance-nightmare item and one less path for chloride-laden water onto the bearings.

8. Span-length terminology

• Span > 6 m ⇒ classified as a bridge (AASHTO / FHWA definition)
• Span < 6 m ⇒ classified as a culvert
• Short span: 6 – 30 m  ·  Medium span: 30 – 100 m  ·  Long span: > 100 m

9. The family of structural systems

Bridges are classified four independent ways: by traffic (highway, pedestrian, rail, transit, utility), by the position of traffic (deck, through, half-through), by material (concrete, steel, timber, FRP), and — most importantly for design — by structural system. Each system has a practical span range within which it is the economic choice.

10. How the engineer picks a system

The system-selection matrix is a balance of ten factors, roughly in the order they usually govern:

1. Span length — determines which families are even feasible
2. Total bridge length and pier count
3. Girder spacing — fewer deep girders vs many shallow girders
4. Locally available materials and shipping cost
5. Site conditions — foundations, vertical clearance, alignment curvature, shipping channel, urban footprint
6. Speed and constructability — precast vs cast-in-place, gantry / launched deck options
7. Contractor equipment and workforce experience
8. Aesthetics and community context
9. Access for future inspection and maintenance
10. Initial + life-cycle cost — steel typically needs more paint / repair in coastal regions; concrete typically the least

Site geometry can flip the answer in a heartbeat. A curved alignment eliminates precast I-girders entirely — you either accept the segmentation cost of precast segmental, or you cast in place. A required shipping channel of 200 m forces a single span across it, at which point cable-stayed often beats girder construction on total cost even though the unit cost per m² of deck is higher. In tight urban corridors, precast segmental pieces trucked in at night beat cast-in-place falsework schemes that block traffic for months. The Millau Viaduct (shown above) is a landmark example — the deck was launched incrementally from one abutment across temporary piers because the 270 m valley depth made conventional falsework impossible.

The deck is the structural element that receives the wheel loads directly and transfers them to the supporting girders. Its material choice drives self-weight, durability, construction speed, ride quality, and future maintenance cost. In North American practice three materials dominate: timber, concrete, and steel. Each carries a different history, mechanical behaviour, and failure mode.

Timber is the oldest deck material and remains competitive on low-volume rural routes and short spans (typically ≤ 20 ft plank, ≤ 40 ft glulam). Two families are used today:

• Sawn-lumber plank decks — nailed or bolted transverse planks over longitudinal stringers. Simple, field-repairable, but rides rough and channels water into the substructure.
• Glued-laminated (glulam) panel decks — factory-laminated Douglas fir or Southern pine panels post-tensioned transversely with high-strength bars. Behaves as an orthotropic plate; ride quality approaches concrete.

Design is governed by AASHTO LRFD Section 8 and the NDS Supplement. Live-load distribution uses the equivalent-strip approach with an impact-reduction factor(IM is taken as zero for wood — timber's damping absorbs the dynamic component, per §3.6.2.3).

Concrete is the dominant U.S. deck material — roughly 85 % of the National Bridge Inventory (FHWA, 2023). Four sub-types deserve mention:

• Cast-in-place (CIP) reinforced concrete — typical thickness 8–9 in, f′_c = 4 ksi, εᴮ epoxy-coated or stainless bars. Designed by AASHTO's equivalent-strip method (§4.6.2.1) or the empirical design method that exploits internal arching action (§9.7.2).
• Precast concrete deck panels — 3–4 in-thick stay-in-place panels topped with a CIP composite pour; used for accelerated construction.
• Full-depth precast panels — 8-in panels post-tensioned longitudinally; connected to girders through pockets grouted with UHPC.
• Prestressed concrete decks — pretensioned or post-tensioned; common in segmental box girders and long-span cable-stayed bridges.

Steel decks are chosen where self-weight is critical: long-span, movable, and rehabilitation projects that must not add dead load to existing girders. The three common forms are:

• Orthotropic-plate deck — a stiffened steel plate that serves simultaneously as the top flange of the girders, the deck, and (with wearing surface) the riding surface. Weighs ≈ 25 % of an equivalent RC deck. Used on Golden Gate re-deck, Verrazzano, San Mateo-Hayward.
• Open steel grid — welded rectangular bars forming an open grating; wheel loads bear directly on the steel. Very light (≈ 20 psf) but noisy and skid-prone in rain.
• Filled/partially-filled grid — the open grid is filled with concrete; combines light weight with a paved riding surface.

FRP composites — glass or carbon fibres in a polymer matrix — offer a corrosion-free alternative in aggressive environments (marine, chemical, high-salt). Two families:

• FRP deck panels — pultruded or vacuum-infused sandwich panels 4–8 in deep; γ ≈ 12–20 pcf, i.e., one-fifth of concrete. AASHTO published the LRFD Guide Specifications for FRP Vehicular Bridges (1st ed., 2018).
• FRP reinforcing bars (GFRP, CFRP) — replace steel rebar in concrete decks where chlorides are unavoidable (e.g., northern deck overlays). Covered by AASHTO's LRFD Bridge Design Guide Specifications for GFRP-Reinforced Concrete.

Segmental construction builds long-span concrete box girders from short precast or cast-in-place segments joined by longitudinal post-tensioning. The two erection variants are balanced-cantilever (segments hung symmetrically from a pier) and span-by-span (segments assembled on an underslung truss). Landmark examples: Sunshine Skyway, Seven Mile Bridge, and the U.S. 331 Choctawhatchee Bay Bridge.

• Eliminates falsework over navigable water, deep valleys, or live traffic.
• Repeatable factory-cast segments give tight quality control.
• Post-tensioning ducts must be perfectly grouted or corrosion of tendons — see the FDOT Ringling Bridge and the Mid-Bay Bridge inspection findings — becomes a life-cycle liability.

Accelerated Bridge Construction is FHWA's umbrella program for methods that reduce on-site closure from months to hours. Core techniques:

• Prefabricated Bridge Elements and Systems (PBES) — full-depth precast deck panels, precast pier caps, precast abutment stems, and even full superstructure modules.
• SPMT moves — a temporary bridge is built off-line on multi-axle transporters, then rolled into place during a weekend closure. Utah DOT's I-15 Sam White bridge (2011) set the North-American benchmark: a 354-ft, 2,400-ton superstructure moved 700 ft in one night.
• Lateral slide-in — the new bridge is built parallel to the existing one, then slid transversely onto the piers.
• Geosynthetic-Reinforced-Soil Integrated Bridge System (GRS-IBS) — abutments of alternating compacted fill and geotextile; suited to short-span local bridges.

UHPC (f′_c > 21 ksi, tensile strength > 1 ksi, fibre-reinforced) allows a 6-in lap splice to develop a #6 bar — one-third of the length required in conventional concrete. That single property has made full-depth precast panelconstruction economically viable and is now the FHWA-recommended detail for ABC deck-to-girder and panel-to-panel connections.

• 3-D-printed concrete abutments and formwork — first live-load demonstration bridges in the Netherlands (Gemert, 2017) and Iowa DOT (2022).
• Shape-memory-alloy (SMA) reinforcement — self-centring seismic columns being piloted by Caltrans.
• Structural health monitoring (SHM) — fibre-optic strain sensors embedded in girders (e.g., the New I-35W in Minneapolis) feed a digital-twin model that updates load rating in real time.
• Bio-based and low-carbon binders — calcined-clay/limestone (LC³) and geopolymer concretes are moving from research into pilot deck pours.

A completed bridge and a bridge under construction are two different structures. In the finished state the load path is continuous, every member is braced by its neighbours, and the design engineer has spent months proving under AASHTO LRFD that every limit state is satisfied. During construction that same bridge passes through dozens of transient states — partial cantilevers, unshored girders, unstressed tendons, half-cast decks — and the loading is dominated by construction equipment (crawler cranes, launching gantries, form travellers) that the designer never saw. Statistically this is when most bridges fail. This section is not a list of collapses; it is an introduction to construction sequencing, temporary stability, erection engineering, and the professional responsibility that binds them.

1. Two different structures

The first thing a young engineer must internalise is that the finished-bridge model studied in class is only the final frame of a long movie. Every earlier frame is a different structural system, often less redundant and with a different governing limit state.

2. Why bridges fail during construction

Post-collapse investigations by the NTSB, FHWA, and international counterparts converge on seven recurring causes. Any one of them can bring the structure down; most real collapses are the intersection of two or three.

3. The construction load path

The clearest way to see the risk is to draw the load path twice — once for the crane lifting a girder into place, and once for the same girder in service. The finished path is short and redundant; the construction path is long, passes through temporarybearings, and often depends on a single crane mat or false-bent that was designed by the contractor, not the engineer of record.

Notice the construction path has two extra links (temporary bearing and false-bent) and passes through a member the engineer of record may never have designed. Every extra link is a place the chain can break.

4. Major case studies

Case study

FIU–Sweetwater Pedestrian Bridge

Miami, FL, USA · 2018 · 6 killed, 10 injured

A 174 ft pre-cast concrete truss span built alongside SW 8th Street using Accelerated Bridge Construction (ABC) and moved into position with self-propelled modular transporters. Cracks up to 40 mm wide were photographed at the north diagonal-to-deck node two days before the collapse. Post-tensioning of the cracked diagonal was under way with the roadway open to traffic when the node failed and the span dropped onto vehicles below.

Timeline

1. 10 Mar 2018 — Span moved into place using SPMTs; overnight closure of SW 8th Street.

2. 13 Mar — Cracks reported at north diagonal; design team judges structure stable, does not close road.

3. 15 Mar 09:00 — Peer-review call — 'not a safety concern.'

4. 15 Mar 13:47 — Re-tensioning of diagonal #11 in progress; node fails; span collapses in <2 s.

Engineering lessons

• ABC does not eliminate the need for redundancy — the truss was non-redundant at the failed node.
• Cracks of that magnitude in a post-tensioned concrete member are a stop-work condition, full stop.
• Peer review must have authority to close the road. In this case it did not.
• Ethical obligation of the licensed EOR to disclose risk overrides schedule and cost pressure.

Case study

Quebec Bridge (first & second collapses)

Québec, Canada · 1907 & 1916 · 75 (1907) + 13 (1916)

The world's longest cantilever span at the time (549 m). During erection the consulting engineer Theodore Cooper approved an increase in span length after fabrication had started, but the dead-load calculations were not revised. As the south cantilever grew, chord compression members began to buckle visibly. Cooper's telegram ordering work to stop arrived after the collapse. In 1916, during re-erection with a new design, the 5,000-tonne central span fell into the river while being hoisted because of a casting failure in a temporary lifting shoe.

Engineering lessons

• Dead load must be recalculated for every geometry change — never scaled by intuition.
• Compression-member buckling is the classic failure mode of long cantilevers under construction.
• The engineer of record must have physical presence and unambiguous stop-work authority.
• This disaster gave rise to Canada's 'Ritual of the Calling of an Engineer' (Iron Ring) in 1922 — a lasting professional obligation.

Case study

Koror–Babeldaob (K–B) Bridge

Palau · Built 1977 · retrofit 1996 · collapsed 1996 · 2 killed, 4 injured

At the time of construction the longest prestressed-concrete box-girder span in the world (241 m). Long-term creep and shrinkage caused the midspan sag to reach 1.6 m — far more than the design predicted. A 1996 retrofit added external post-tensioning cables and cast additional concrete at midspan; three months later the bridge collapsed. Investigations point to loss of prestress from creep interacting with the sequence in which new tendons were stressed against a deformed structure.

Engineering lessons

• Long-term creep of prestressed concrete is time-dependent and non-linear; single-value estimates are inadequate.
• Retrofit stressing sequence changes the load path — must be re-analysed on the current (deformed) geometry.
• Field measurement of camber and prestress loss should be part of any long-span PC bridge's inspection régime.

Case study

I-35W Mississippi River Bridge

Minneapolis, MN, USA · Built 1967 · collapsed 2007 · 13 killed, 145 injured

Although a service failure, the collapse is a construction-loading case study. The NTSB found that under-designed 12 mm gusset plates at node U10 had been operating near capacity for 40 years. On the day of failure, MnDOT was resurfacing the deck: 287 tonnes of construction aggregate, mixer trucks, and a paving machine were concentrated over the critical node, and the added dead load from prior overlays had never been recalculated against the original gussets.

Engineering lessons

• Every resurfacing / rehab must recalculate DL against the original design — γ_DW = 1.50 is not a decorative number.
• Non-redundant steel truss bridges (fracture-critical members) demand FCM inspection procedures — see Ch. 19.
• Construction staging on an in-service bridge is a design task, not a contractor detail.
• The collapse accelerated the FHWA gusset-plate re-evaluation programme across the U.S. inventory.

Case study

Hintze Ribeiro Bridge

Entre-os-Rios, Portugal · Built 1887 · collapsed 2001 · 59 killed

A 19th-century wrought-iron truss on stone piers over the Douro River. Decades of upstream sand mining had lowered the riverbed by 6 – 8 m, exposing and undermining the shallow spread footing of Pier 4. On a rainy March night the pier tilted, one span dropped, and a passenger bus and cars fell into the river. Not a construction failure per se, but a case study in how maintenance, inspection, and adjacent construction (dredging) interact with an old structure.

Engineering lessons

• Scour is the #1 cause of bridge collapse in the U.S. and worldwide — see Ch. 14.
• Underwater inspection of pier footings is mandatory for scour-critical bridges (NBIS Item 113).
• Adjacent human activity — sand mining, dredging, channel realignment — is a construction issue for existing bridges.
• Historic bridges need scour and inspection régimes matched to their actual foundation type, not the modern one.

5. Failure mechanisms — a working vocabulary

Every case study above is an instance of one or two of the mechanisms below. Learn them as a vocabulary; when you inspect a bridge in construction, you are looking for these specific initiators.