Concrete is cement, water, sand and stone; its engineering is everything that follows from the fact that it gets strong slowly, shrinks as it dries, creeps under sustained load and cracks at a tensile stress around a tenth of its compressive strength.

Bridge concrete is specified by 28-day compressive strength: 28–35 MPa (4–5 ksi) for decks and substructure, 40–70 MPa (6–10 ksi) for prestressed girders, where high strength lets more strand be used. Modulus of elasticity scales roughly with the square root of strength and matters for deflection and prestress loss. Creep can double the long-term deflection of a girder and transfers stress from concrete to steel; shrinkage restrained by the girder cracks the deck. Both are predicted by the AASHTO 5.4.2.3 models and both depend on humidity, member size and age at loading.

Durability, not strength, is what ends a concrete bridge's life. Chloride from de-icing salt or the sea reaches the reinforcement, the steel corrodes, the concrete spalls. The defences are cover (50–75 mm on exposed faces), low permeability (low water–cement ratio, supplementary cementitious materials such as fly ash, slag and silica fume), and keeping water off the structure in the first place.

Reinforcing bars carry the tension concrete cannot. Standard bars are deformed for bond, Grade 420 MPa (60 ksi) yield in most codes, with Grade 550 (80 ksi) permitted in AASHTO for many applications. The design questions are how much steel, where, and how it is anchored: development length is the embedment needed for a bar to reach its yield strength through bond, and it scales with bar diameter, bar stress, concrete strength, cover and coating.

plain deformedepoxy-coatedstainlessGFRP

Corrosion protection has produced a family of alternatives: epoxy-coated bars (cheap, but the coating is damaged in handling and the debate about their long-term value continues), galvanised bars, stainless steel (expensive, used at the worst exposures such as deck edges and splash zones), and glass-fibre-reinforced polymer (GFRP) bars, which cannot corrode at all but are elastic to failure and need their own design rules.

Stainless steel bars come in austenitic grades (1.4301/304 and 1.4401–1.4436/316) and duplex grades (1.4362/2304 and 1.4462/2205), to BS 6744 or ASTM A955. They tolerate chloride contents many times higher than carbon steel before corroding, and inside concrete they can be tied directly to carbon-steel bars without galvanic trouble, because both remain passive. The price per tonne is several times that of carbon steel, so they are used where they earn it: the outer layer in splash zones, deck edges, joints and parapet anchorages. The pier at Progreso in Mexico, built in 1937–41 with stainless reinforcement, is still in service, while a neighbouring pier built later with carbon steel has deteriorated badly.

GFRP bars (ASTM D7957) are elastic to failure, with tensile strengths of roughly 600–1,200 MPa that fall as the diameter grows and a modulus of only 45–60 GPa, about a quarter of steel's. Because they do not yield, sections are designed to fail by concrete crushing rather than by bar rupture, with lower resistance factors (0.55–0.65 in ACI 440), and the low modulus makes crack width and deflection govern more often than strength. Sustained stress is limited to a fraction of the guaranteed strength, 20% under ACI 440.1R, to avoid creep rupture, and bends must be made in the factory because the bars cannot be bent on site. Bridge rules exist in the AASHTO LRFD Bridge Design Guide Specifications for GFRP-Reinforced Concrete, ACI 440.11 and CSA S6 Section 16; Canada has the longest record of GFRP-reinforced bridge decks.

Prestressing works because the steel is so strong that the losses do not use it up. Standard seven-wire strand is 12.7 or 15.2 mm in diameter (0.5 or 0.6 in), Grade 1860 MPa (270 ksi), low-relaxation, stressed to about 75% of its ultimate strength at jacking. Bars (Grade 1035) and wires are used in special cases.

Pretensioning stresses the strand against abutments in a casting bed before the concrete is placed, and is the factory method for precast girders. Post-tensioning threads tendons through ducts in hardened concrete, jacks them from the ends, and fills the ducts with grout; it is how boxes, segmental bridges and long girders get their prestress on site.

The prestress you jack is not the prestress you keep. Losses come from elastic shortening of the concrete as it is compressed, from creep and shrinkage over years, from relaxation of the strand, and, in post-tensioning, from friction along the duct and from anchorage set as the wedges seat. Together they take 15–25% of the jacking force. Grouting protects the strand from corrosion, and when it is done badly the consequences are hidden: the Ynys-y-Gwas bridge (Wales, 1985) collapsed with no warning because water had reached ungrouted tendons at the segment joints.

Bridge steel is specified for strength, toughness and weldability. In North America, ASTM A709 covers Grade 36, 50, 50W (weathering) and the high-performance HPS 50W and 70W (yield 345 and 485 MPa) whose toughness and weldability let designers use them in fracture-critical members. Europe uses S355 and S460 to EN 10025.

Weathering steel forms a tight rust patina that protects it, saving paint for the life of the bridge, though only where it can dry: not in salt spray, not under a leaking joint, not in a permanently damp tunnel. Toughness is specified by Charpy V-notch energy at a temperature zone, because steel that is tough at room temperature can be brittle at −30 °C. Fatigue is governed by the detail, not the steel grade: welded attachments, cover plates and stiffener ends are assigned categories A to E' with sharply different allowable stress ranges.

Connections are welded in the shop and bolted in the field with high-strength bolts (ASTM F3125 Grades A325 and A490) designed as slip-critical where movement would matter. Composite action with the concrete deck through shear studs is what makes modern steel girders efficient.

Steel–concrete composite construction is the most important "composite" in bridges: the deck takes compression, the steel takes tension, and studs make them one section. It halves the steel in a positive-moment region.

shear studsUHPC jointCFRP strip

Fibre-reinforced polymers arrived in the 1990s: GFRP decks and bars, carbon-fibre (CFRP) strips and wraps for strengthening, and CFRP stay cables (Stork Bridge, Winterthur, 1996) that weigh a fifth of steel and cannot corrode but are hard to anchor and to inspect.

Ultra-high-performance concrete (UHPC), with compressive strengths of 150–200 MPa and steel fibres that give it tensile ductility, is used for joints between precast deck panels, for thin shell girders (Sherbrooke, 1997, was the first), and for overlays. Its cost keeps it in small volumes at critical places.

Stainless steel reinforcement, stainless-clad strand, and duplex stainless for cable components are the durability upgrades for marine environments, including the Gulf, where the combination of heat, humidity and chloride is as severe as anywhere in the world.

Masonry bridges are the largest stock of very old structures still in service in Europe, and assessing them is its own discipline: the arch is checked by thrust-line or limit analysis rather than by stress, and fill, spandrel walls and ring separation matter more than stone strength. They fail by scour of the piers far more often than by overload.

masonry arch, thrust line dashedglue-laminated timber

Timber is back, as glulam and cross-laminated panels, for footbridges and low-volume road bridges where a light, quick, low-carbon structure is wanted. Its rules are about moisture: keep it dry, detail the ends, and expect the connections, not the timber, to govern.

High-performance concrete (HPC) is not a strength class but a way of specifying: concrete engineered for the property a member actually needs, which on a bridge is more often low permeability, low shrinkage or early strength than ultimate strength. The tools are a low water–binder ratio, about 0.30–0.38, made workable with high-range water reducers, and supplementary cementitious materials. Silica fume at 5–10% of the binder refines the pore structure; ground granulated blast-furnace slag at 25–70% or fly ash at 15–35% cuts permeability and heat of hydration and keeps cutting permeability as the concrete ages. A good HPC deck passes fewer than 1,000 coulombs in the rapid chloride permeability test (ASTM C1202, the "very low" band), where an ordinary concrete with a water–cement ratio of 0.45 typically falls in the 2,000–4,000 "moderate" band.

HPC has its own ways of going wrong. Silica-fume concrete bleeds almost no water, so its surface dries before it sets and plastic shrinkage cracks appear within hours unless the deck is fogged and wet-cured at once. Low water–binder mixes dry themselves out from within and show autogenous shrinkage that ordinary concrete does not, and very high-strength concrete is more brittle and more prone to explosive spalling in fire. The mix design is the easy part; curing decides whether an HPC deck is crack-free.

Lightweight concrete, made with expanded shale, clay or slate aggregate, weighs about 17–19 kN/m³ instead of 23–24 and is the high-performance option for long spans, where self-weight is most of the load. The cast-in-place segmental New Benicia–Martinez Bridge in California (2007), with main spans of about 200 m, is the best-known lightweight example.

Ultra-high-performance concrete (UHPC) goes much further. FHWA defines it as a cementitious composite with a compressive strength of at least 150 MPa and a sustained post-cracking tensile strength above 5 MPa. Typical mixes contain no coarse aggregate: cement, silica fume and fine quartz, a water–binder ratio near 0.2, and about 2% by volume of short steel fibres about 13 mm long and 0.2 mm in diameter. The fibres bridge cracks, so the material keeps carrying tension after first cracking, at roughly 6–10 MPa, through many fine cracks rather than one wide one. The dense matrix lowers chloride diffusion by one to two orders of magnitude compared with normal concrete, and heat-treated UHPC has very little creep or long-term shrinkage.

Its uses follow its price, which per cubic metre is many times that of ordinary concrete:

• Field-cast joints between prefabricated deck panels and girders, where UHPC's bond lets reinforcing bars lap over about 150 mm, a fraction of the length ordinary concrete needs. This is now the most common use in North America.
• Thin overlays of 25–50 mm on existing decks, pioneered in Switzerland, which add a waterproof layer and stiffness in one pour.
• Precast girders and shells: the Mars Hill Bridge in Iowa (2006), the first US highway bridge with UHPC girders, and the 120 m arch of the Seonyu footbridge in Seoul (2002).

Design rules have arrived only recently: France (NF P18-710, 2016) and Switzerland (SIA 2052) have national documents, and AASHTO guidance has grown out of FHWA research. Two cautions apply everywhere. Fibre orientation, set by how the material flows into the form, changes the tensile strength in a given direction, so placement must be planned like reinforcement. And the published properties assume proper curing, which in a hot climate means controlling temperature as well as moisture.

The covers and water–cement ratios in a code table are a prescriptive answer to a question that can be asked directly: given this exposure and this concrete, how long until the reinforcement starts to corrode? On major bridges with 100- or 120-year design lives it is now normal to answer it with a model rather than a table.

Tuutti's model (1982) splits the life of reinforced concrete in two. During initiation, chloride or carbonation advances from the surface while the steel stays passive and nothing is visible. During propagation, the steel corrodes, the rust expands, and the cover cracks and spalls. Propagation can be short in wet, chloride-laden concrete, so design usually takes the end of initiation, depassivation, as the limit state.

Chloride ingress is modelled as diffusion by Fick's second law. With a constant surface chloride C_s and apparent diffusion coefficient D, the chloride content at depth x after time t is C(x,t) = C_s[1 − erf(x / 2√(Dt))], and corrosion starts when C at the bar reaches the critical content C_crit. Take a splash-zone pier with C_s = 4.0% and C_crit = 0.4%, both by mass of binder. Setting C/C_s = 0.1 gives x / 2√(Dt) = 1.163, so the time to depassivation is t = x² / (5.41D). With D = 0.5 × 10⁻¹² m²/s, representative of a good slag or fly-ash concrete in the long term:

The lesson is in the exponent: time grows with the square of the cover and inversely with D. Adding 25 mm of cover does more than most coatings, and halving D with slag or fly ash doubles the life. Construction tolerance works the same way in reverse: a bar placed 15 mm too shallow in a 75 mm cover loses more than a third of its initiation time.

Real calculations treat the inputs as time-dependent and uncertain. D falls as the binder keeps hydrating, D(t) = D₀(t₀/t)^α, with an ageing exponent α of roughly 0.3 for plain Portland cement and 0.45–0.6 for slag and fly-ash blends, which is why those binders perform so well. Diffusion also speeds up with temperature, roughly doubling between 20 °C and 35 °C, which matters in the Gulf. The fib Model Code for Service Life Design (fib Bulletin 34, 2006, and ISO 16204) treats C_s, D, α, C_crit and the cover as random variables and accepts a probability of depassivation within the design life of about 10% (β ≈ 1.3), with D measured on the project mix by the rapid migration test NT Build 492. The replacement Tappan Zee crossing in New York is one of the bridges designed this way; Life-365 is a freely available North American alternative.

Carbonation, the slow neutralisation of the concrete's alkalinity by carbon dioxide from the air, advances as x = k√t and is fastest at moderate humidity, so on bridges it matters on sheltered surfaces such as box-girder interiors and soffits rather than on wetted faces. The other threats are chemical: alkali–silica reaction, prevented by aggregate testing, low-alkali binders and slag or fly ash; sulfate attack from groundwater and the sabkha soils of the Gulf; and delayed ettringite formation in concrete that reached more than about 70 °C during curing, a real risk in large pours and steam-cured precast in hot climates.

Every welded steel structure contains small flaws at the toes of its welds, and every passing truck opens and closes them a little. The stress range that grows them is set by the geometry of the detail far more than by the strength of the steel: a crack at a weld toe does not care whether the plate yields at 345 or 485 MPa. Fatigue design is therefore a matter of choosing details, and AASHTO (Article 6.6.1.2, Table 6.6.1.2.3-1) and Eurocode (EN 1993-1-9) both catalogue them by category. Eurocode names each category by its stress range at 2 million cycles, from 160 MPa down to 36 MPa; a welded transverse stiffener is about category 80.

The finite-life resistance is (ΔF)_n = (A/N)^1/3: halve the stress range and the life increases eightfold. Below the threshold, cracks do not grow at all under constant-amplitude loading. A C' stiffener weld under Fatigue I (section 4.10) must keep 1.75Δf ≤ 82.7 MPa, which limits the unfactored stress range from the fatigue truck to 47 MPa. On a lightly trafficked road with an ADTT_SL of 300, N = 365 × 75 × 300 = 8.2 million cycles, (A/N)^1/3 = 56 MPa, and Fatigue II allows 56 / 0.80 = 70 MPa. For this detail the two checks give the same answer at just under 1,000 trucks per lane per day.

The categories are a record of what has cracked. Cover plates welded to the flanges of rolled beams, a common way of adding capacity in the 1950s and 1960s, put a Category E or E' detail where the stress is highest; the cracks found at the cover-plate ends of the Yellow Mill Pond Bridge on I-95 in Connecticut in the 1970s made them the textbook example, and they have largely disappeared from new design. The most common cracking in US steel bridges, however, has been distortion-induced fatigue. Connection plates for cross-frames were once deliberately left unattached to the tension flange, and the short web gap between plate and flange was bent back and forth by differential deflection of adjacent girders. AASHTO 6.6.1.3.1 now requires transverse connection plates to be welded or bolted to both flanges.

Fracture is fatigue's last act, and it depends on toughness (section 5.4) and on constraint. In December 2000 two of the three girders of the Hoan Bridge in Milwaukee fractured with no fatigue crack growth to give warning: intersecting welds at a lateral-bracing gusset had created a pocket of triaxial constraint in which the steel could not yield. Current practice avoids such intersecting welds and leaves gaps between gussets and transverse plates. Good fatigue design reads as a list of habits: put details in low-stress regions, keep attachments off tension flanges, bolt rather than weld secondary members to primary ones, and grind or peen weld toes where a detail must remain in a high-stress zone.

Unprotected carbon steel in a coastal atmosphere can lose a tenth of a millimetre of thickness a year or more. Steel bridges are protected in one of two ways: by an alloy that protects itself, or by barrier coats and sacrificial metal that protect it.

Weathering steel (ASTM A709 Grade 50W, EN 10025-5 S355J2W) contains small additions of copper, chromium, nickel and phosphorus. In alternating wet and dry conditions its rust forms a dense, adherent layer that slows further corrosion to a few micrometres a year, so it needs no paint for the life of the bridge. The patina needs dry periods and little chloride to form; where the steel stays wet, or salt arrives as sea spray or de-icing runoff, the rust stays loose and flaky and the steel corrodes like ordinary steel. FHWA Technical Advisory T 5140.22 (1989) lists the situations to avoid: marine coastal sites, persistently wet or humid locations, low clearance over salted roads, and polluted industrial atmospheres. Detailing does the rest: eliminate deck joints or paint the steel near them, keep drainage away from the girders, add drip bars so rusty runoff does not stain the piers, and avoid pockets that trap water and debris. Many authorities also add a sacrificial thickness allowance to exposed plates.

Everywhere else, steel is coated, and the performance of any coating is decided first by surface preparation: abrasive blast cleaning to Sa 2½ (ISO 8501-1) or SSPC-SP 10 near-white metal, with a sharp surface profile for adhesion. The common bridge system is three coats applied in the shop:

• a zinc-rich primer, about 75 µm, whose zinc corrodes sacrificially to protect the steel at any scratch;
• an epoxy intermediate coat, 100–150 µm, the barrier;
• a polyurethane or polysiloxane finish, 50–75 µm, for colour and for resistance to ultraviolet light, which chalks and degrades epoxies.

Thermal-sprayed zinc or zinc–aluminium (metallizing, 200–300 µm, sealed) and hot-dip galvanizing (ISO 1461 or ASTM A123) are the sacrificial alternatives. Galvanizing is limited by the size of the bath, so it suits secondary members, railings and short girders. A duplex system, paint over galvanizing or metallizing, lasts longer than the sum of its parts because each layer protects the other.

ISO 12944 ties the system to the site. It classifies atmospheres from C1 to C5 and CX, defined in ISO 9223 by the first-year thickness loss of carbon steel: 25–50 µm in C3, 50–80 µm in C4, 80–200 µm in C5 and 200–700 µm in CX. It then rates coating systems for durability: low (up to 7 years to the first major maintenance), medium (7–15), high (15–25) and very high (more than 25). Coastal sites in hot, humid, salt-laden climates such as the Gulf generally fall in C5 or above, where only very-high-durability systems make economic sense, and the splash zone of a marine pier needs an immersion-grade system or encasement.

Painting is one of the largest recurring maintenance costs of a steel bridge and among the most disruptive, because repainting needs containment, access and traffic management, and many older bridges carry lead-based paint that must be removed under full containment. A life-cycle comparison should therefore price at least one full repaint within a 100-year life, or justify the weathering steel, metallizing or duplex system that avoids it.

Bearings and joints are where a bridge's materials work hardest: squeezed, sheared, slid and weathered with every passing truck, and expected to last decades. Chapter 9 describes the components; this section is about what they are made of.

Elastomers for bearings are natural rubber (polyisoprene) or polychloroprene (neoprene). Natural rubber is cheaper and stays flexible in the cold; polychloroprene resists ozone, oil and weathering better. Bridge compounds have a hardness of about 50–60 Shore A and a shear modulus G of about 0.6–1.2 MPa; EN 1337-3 takes 0.9 MPa as standard. Steel shims vulcanised between the rubber layers stop them bulging, so a laminated bearing is stiff in compression and soft in shear. Its compressive stiffness and capacity depend on the shape factor S, the loaded area of one layer divided by the area free to bulge, LW / [2h_ri(L + W)]. For a 400 × 500 mm bearing with 12 mm layers, S = 9.3, typical for a bridge bearing.

Rubber's enemies are cold, ozone and poor bonding. In prolonged cold, polychloroprene in particular can crystallise and stiffen several times over, so AASHTO M 251 grades compounds for low-temperature zones; in hot climates the concern is heat ageing and ozone cracking, controlled by oven-ageing and ozone tests. A common failure is loss of bond between rubber and shims, which is why production bearings are sampled and proof-tested in compression and shear before they leave the factory. In a pot bearing the elastomer is confined in a steel cylinder and behaves almost like a fluid; sealing rings keep it from squeezing out past the piston, and their wear usually ends the bearing's life.

PTFE (polytetrafluoroethylene) slides on a mating plate of austenitic stainless steel polished to a mirror finish. Its friction falls as contact pressure rises and increases with cold and wear; dimples pressed into the sheet hold silicone grease and keep friction low for decades. EN 1337-2 design values for lubricated, dimpled PTFE run from about 3% at high contact pressure to about 8% at low pressure, and unlubricated sheet is higher. Friction matters twice: as a force on the substructure, and as the force needed to make the bearing move at all. Wear depends on accumulated slide path, which on a long, flexible bridge can be dominated by small traffic-induced movements rather than the daily thermal cycle. Modified ultra-high-molecular-weight polyethylene (UHMWPE) sliding materials, with higher allowable pressure and much lower wear, are now common on long spans and railway bridges.

Expansion-joint seals are extruded EPDM or polychloroprene glands held in steel or aluminium extrusions anchored to the deck. The steel parts of modular joints take every wheel as a separate impact and have cracked by fatigue at their welded centre-beam connections; NCHRP Report 402 (1997) set the fatigue design basis now used. Asphaltic plug joints rely on a polymer-modified binder that softens with heat, so in hot climates they rut and flow and are best kept to small movements and moderate traffic. Whatever the material, specify it with its test standard (AASHTO M 251 for elastomeric bearings, the EN 1337 series for bearings and their sliding materials) and keep the certificates: these are the parts that will be replaced, and the replacement needs the original data.

A design assumes material properties; a site has to prove them. The client-side engineer's job here is less to run the tests than to make sure the right tests are run, at the right frequency, on samples that represent what went into the structure, and that the results are acted on.

Concrete. Fresh concrete is checked at delivery for workability (slump or flow), temperature, air content where specified, and unit weight; hot-weather specifications commonly cap the placing temperature at about 32 °C. Strength is tested on 150 × 300 mm or 100 × 200 mm cylinders (ASTM C39) or 150 mm cubes (EN 12390-3). Cubes read about 20–25% higher, which is why a Eurocode class is written as cylinder/cube strength, C40/50. Under ACI 318 a strength test is the average of two 150 mm or three 100 mm cylinders at 28 days, sampled at least once a day, once per 110 m³ and once per 460 m² of slab or wall. Concrete is accepted when every average of three consecutive tests reaches f'_c and no single test falls more than 3.5 MPa below it, or below 0.9f'_c when f'_c exceeds 35 MPa.

Producers must therefore aim above f'_c. With a standard deviation s of 4 MPa, ACI 318 requires a mean strength of the larger of f'_c + 1.34s and 0.90f'_c + 2.33s, which for f'_c = 40 MPa is 45.4 MPa. When results fall short, cores are drilled from the structure, three for each low test, and the concrete is accepted if their mean is at least 0.85f'_c and no core is below 0.75f'_c. Durability-critical work adds acceptance tests on the actual mix: chloride migration (NT Build 492) or rapid chloride permeability (ASTM C1202), water penetration and, common in Gulf specifications, initial surface absorption. In mass concrete, embedded thermocouples check the default limits of ACI 301: a peak temperature of 70 °C and a difference of 19 °C between core and surface.

Steel. Plates arrive with mill certificates (EN 10204 type 3.1 or equivalent) giving chemistry, tensile properties and Charpy energy, which are checked against the specification and traced to the pieces cut from each plate. Welding follows AASHTO/AWS D1.5, the Bridge Welding Code: qualified procedures and welders, visual inspection of every weld, and radiographic or ultrasonic testing of complete-penetration groove welds, all of those in tension and a sample of those in compression. High-strength bolts are verified before installation in a tension calibrator, a few from each lot tightened to confirm that the chosen method (turn-of-nut, direct tension indicators or twist-off bolts) actually reaches the required pretension. Coatings are checked for surface profile and dry film thickness, coat by coat.

Prestressing. Strand certificates (ASTM A416 or the national equivalent) give the breaking load, the elongation at failure (at least 3.5% under ASTM A416) and the measured modulus, which is used in the elongation calculation. At stressing, each tendon's measured elongation is compared with the calculated value, and a difference beyond the specification's tolerance, typically 5–7%, is investigated before the duct is grouted, because it means that friction, modulus or force is not what the design assumed. Grout is accepted on fluidity, bleed and strength under the PTI M55 specification, with full-scale mock-ups for long or vertical ducts.

Two habits make acceptance work. Agree the inspection and test plan, with its hold points, before work starts. And treat a non-conformance as information about the process, not only about the piece: a low cylinder, a short elongation or a thin coating is rarely an isolated event.