The workhorse. A slab bridge is a solid or voided concrete plate spanning between supports, economical to about 15 m solid and 25 m voided, with nothing to inspect but the underside. Beyond that the slab becomes a deck resting on girders, and the girder does the spanning.

slabT-beambulb-teeplate girdercomposite

• Reinforced concrete T-beams cast in place: 12–25 m, common in older highway stock.
• Precast prestressed concrete girders, AASHTO I-sections, bulb-tees, NU and NEXT beams: 20–50 m per span, occasionally 60 m with spliced segments. Factory-made, erected in a night, the dominant highway bridge in most countries.
• Steel plate girders, welded I-sections acting compositely with the deck: 30–120 m, longer when continuous and haunched. Light to erect, need painting or weathering steel.

Girders carry load by bending, so depth is the main lever: typical span-to-depth ratios run from about 15 for simple concrete spans to 30 or more for continuous steel. The site's girder depth tool gives the AASHTO minimums.

Close the section into a hollow box and two things happen: the top and bottom flanges are both wide, so the section is efficient in both positive and negative bending; and the closed cell is enormously stiff in torsion, which lets the bridge be curved in plan or carry eccentric load without twisting. That is why almost every long-span concrete girder bridge and every curved elevated highway is a box.

single-cell boxmulti-cell boxsteel tub

Cast-in-place prestressed concrete boxes built by balanced cantilever reach 200–300 m; the Shibanpo Bridge in Chongqing holds the girder record at 330 m by using a steel box for the middle of its main span. Steel boxes with orthotropic decks are lighter and are used both as girders and as the deck of cable-supported bridges. Single-cell, multi-cell, and trapezoidal "tub" boxes are chosen by width and by how the deck will be built.

The price of the box is complexity: internal diaphragms, thermal gradients through the section, distortion of the cross-section under torsion, and a hollow interior that must be ventilated and inspected.

A truss is a beam with the web replaced by triangles, so every member carries axial force only. Members can be small, connections are the work, and the structure is transparent to wind. The classic forms are named for their diagonals: Pratt (diagonals in tension, the steel favourite), Howe (diagonals in compression, the timber favourite), and Warren (alternating, no verticals).

PrattHoweWarren

Trusses are deck, through or half-through depending on where the traffic runs. They dominated railway bridges from 1850 to 1950 and reached 549 m in cantilever form at Quebec. Today they survive where stiffness matters, long railway spans, the stiffening girders of some suspension bridges, and the Ikitsuki (400 m) and similar continuous truss spans in Japan.

Their weakness is maintenance. Hundreds of connections gather water and dirt; eyebars and pins cannot be inspected without dismantling; and a two-truss bridge has no redundancy, which is why the Silver Bridge collapse of 1967 created the fracture-critical member inspection rules that Chapter 10 describes.

An arch carries load in compression and pushes outward at its springings. If the ground can take that thrust, rock in a gorge, the arch is the natural form; if it cannot, a tie between the springings takes the thrust and the arch becomes a self-contained beam, the tied arch or bowstring.

deck archthrough (tied) archhalf-through arch

• Deck arch: traffic above, spandrel columns down to the rib. Gorges and valleys.
• Through arch: traffic hangs below the rib on hangers, needed where clearance below is limited. Usually tied.
• Half-through: deck cuts through the rib mid-height.

Hinges control how the arch responds to settlement and temperature: three-hinged arches are statically determinate and forgiving; two-hinged are stiffer; fixed arches are stiffest and most sensitive. Materials went from stone to concrete (Krk, 390 m) and steel (New River Gorge, 518 m; Chaotianmen, 552 m) to concrete-filled steel tube, which gives China the current record at the Tian'e Longtan Bridge, 600 m (2024), ahead of Pingnan Third at 575 m. Arches are built by cantilevering the rib out on temporary stays, or by rotating half-arches into place; see the Methods animations.

Strictly, the two longest arches use different technologies. Pingnan Third is a concrete-filled steel tube (CFST) arch: the steel tubes are erected first by cable crane, then filled with pumped concrete, and the filled tube is the finished rib. Tian'e Longtan is a reinforced concrete arch built around a stiff skeleton of CFST, which first serves as the erection arch and is then encased in concrete to form the final box rib. Both methods avoid the enormous centring that a 600 m concrete arch would otherwise need. The idea of a steel skeleton that ends up inside a concrete arch goes back to Josef Melan in the 1890s; building skeletons, and whole arch ribs, from concrete-filled tubes at this scale is a Chinese development of the last thirty-five years.

Straight cables run from a tower directly to points along the deck, so the deck is a beam on many elastic supports, and it is also a compressed column because the horizontal component of every stay pushes it toward the tower. That combination, deck in compression, stays in tension, tower in compression, is very stiff for its weight and builds itself: each new deck segment is hung on its stay and used as the platform for the next.

Stays are arranged as a fan (all converging at the tower top), a harp (parallel), or the semi-fan most large bridges use. Design choices: one plane of stays (needs a torsionally stiff box deck) or two; concrete, steel or composite deck; tower shapes from a single mast to H, A, inverted-Y and diamond. Spans of 200–600 m are routine; the largest (Changtai, 1,208 m; Russky, 1,104 m; Sutong, 1,088 m) needed stays over 500 m long and decks that resist buckling under the accumulated compression. Stays are replaceable strand bundles in sheathing, and their fatigue and vibration behaviour (rain–wind vibration especially) is a specialist subject.

Drape a main cable over two towers and anchor it in the ground; hang the deck from it on vertical hangers. The main cable is a pure tension member, the towers are columns, and the deck is a stiffening girder whose only job is to spread concentrated loads and keep the shape stable. Nothing spans farther for less material, which is why every span over about 1,250 m is a suspension bridge.

anchoragemain cablehangersstiffening girdertower

The parts:

• Main cable, thousands of 5 mm galvanised wires, spun in place from anchorage to anchorage (aerial spinning) or lifted as prefabricated parallel-wire strands (PPWS), then compacted and wrapped.
• Anchorages, gravity blocks of concrete, or tunnels into rock, holding a pull measured in hundreds of meganewtons.
• Towers, steel or concrete, 200–330 m tall on the largest bridges.
• Stiffening girder, a truss (Akashi, Golden Gate) or a streamlined steel box (Severn onwards). The box is lighter and aerodynamically stable; the truss is stiffer for railways.

Where there is no room for anchorages the cable can be tied to the deck itself, a self-anchored suspension bridge (San Francisco–Oakland Bay east span, Yeongjong), at the cost of building the deck first. The span record is 2,023 m (1915 Çanakkale); designs beyond 3,000 m exist and are limited by cable weight and aerodynamics rather than by strength.

An extradosed bridge looks like a cable-stayed bridge with the towers cut short: stays leave a low pylon at a shallow angle and act less like supports than like external prestressing tendons draped outside the girder. The deck stays deep and stiff, the stays see small stress ranges and can be designed at higher allowable stress, and the whole thing is built as a balanced cantilever. The form fits spans of 100–275 m where a box girder would be too deep and a cable-stayed bridge too tall; Japan built the first (Odawara, 1994) and many since, and it has become common in Korea, China and India.

cable-stayed: tall tower, steep staysextradosed: short pylon, shallow stays, deep girder

Other hybrids exist because real sites are awkward: cable-stayed with a suspended centre (a "cable-stayed suspension" system used at the Third Bosphorus and in several Chinese Yangtze crossings), arches with stays, and cantilever trusses with cable assistance. Each mixes two load paths, and the designer's job is to make sure the stiffer one does not steal all the load from the other.

When ships need more clearance than a fixed bridge can give, part of the bridge moves. Three families cover almost every case: the bascule tilts one or two leaves upward about a horizontal axis, balanced by a counterweight (Tower Bridge); the swing bridge rotates about a central pier to open two channels; the vertical lift raises a whole span between two towers on counterweighted cables, the choice for long spans and railways.

basculeswingvertical lift

Movable bridges are machines as much as structures. The engineering questions are balance, machinery and control, fatigue from thousands of openings, the alignment of the span when it lands, and what happens when the power fails. They cost more to run than to build, and are usually replaced by high-level fixed bridges when traffic allows.

Type selection is where a bridge is really designed. The factors, roughly in the order a designer meets them:

• Span layout, set by the obstacle, the navigation channel, the flood, and where piers can stand. It decides the family: under 50 m girders, 50–250 m boxes and arches, 250–1,000 m cable-stayed, beyond that suspension.
• Ground, rock allows arches and cheap anchorages; soft soil pushes toward light superstructures and fewer, larger foundations.
• Construction access, can cranes reach? can barges? Is the crossing over live traffic, so it must be built from above?
• Environment, seismicity, wind, ship impact, salt, ice, temperature range.
• Life-cycle cost and maintenance, joints and bearings, paint, cable replacement.
• Appearance, which is not last; bridges are the most public structures a society builds.

Two tools on this site encode the first pass: the span-type recommender and the method selector. They will not choose a bridge for you, but they will show you which options are not silly.

Appearance deserves more than a line, because it is the factor engineers are least trained for. A few rules hold across cultures. Proportion: the deck should read as slender against its span and against the height of the piers; a deep girder on short piers looks heavy, the same girder on tall piers does not. Order: one family of pier shapes, one rhythm of spans, one parapet line along the whole crossing, with changes made for a structural reason and visibly so. Edge and shadow: a deck overhang of 1.5–2.5 m throws the girder into shadow and makes the bridge look thinner than it is, and a clean, continuous edge matters more than any decoration. Odd numbers of spans usually look better than even ones, because they avoid a pier in the middle of the view.

Fritz Leonhardt set out these principles at length in Brücken (1982; in English Bridges: Aesthetics and Design), and David Billington called the best of such work structural art, in which efficiency, economy and elegance are pursued together. Good appearance costs little when it is considered at type selection and a great deal when it is added afterwards. Section 3.15 works through a type selection from start to finish.

Take away the bearings, cast the deck into its supports, and the bridge becomes a rigid frame. The joint between deck and pier or wall now carries moment, so the deck behaves as if partly fixed at its ends and the supports share the bending. Frames come in three very different sizes.

• Portal frames and box culverts, 6–25 m: a deck slab cast monolithically with two walls that double as abutments, the standard underpass beneath a road or railway embankment. On poor ground the bottom is closed and the frame becomes a box.
• Inclined-leg and V-leg frames, 40–120 m: legs raked toward the span shorten the effective span of the deck and give a valley crossing the outline of an arch with the simplicity of a girder.
• Continuous rigid frames, 100–270 m: prestressed concrete boxes built by balanced cantilever on tall piers, with no bearings at all over the main piers. The form is very common in China; the auxiliary channel bridge of the Humen crossing (270 m, 1997) is the classic example.

The gain is easy to show. For a uniform load w on a span L, a simply supported deck has a midspan moment of wL²/8. With fully fixed ends the same load gives wL²/12 at the supports and wL²/24 at midspan. A real frame lies between the two: if the legs are as stiff as the deck (EI/h of the leg equal to EI/L of the deck, fixed bases, no sway), moment distribution gives an end moment of wL²/18. For a 20 m portal carrying 100 kN/m that is 2,220 kNm at each corner and 2,780 kNm at midspan, against 5,000 kNm for the same deck on bearings. The peak moment falls by 44 per cent and the deck can be thinner. The corner, however, becomes the most heavily stressed and congested point of the structure, and its detailing (Eurocode 2, Annex J) decides whether the frame performs as designed.

What the frame gains in bending it pays for in restraint. Temperature, shrinkage and creep all try to change the length of the deck, and without bearings that movement has to be absorbed by bending the piers or walls. A 300 m concrete frame cooling by 30 °C shortens by 90 mm, and each end pier must follow its share. The moment induced in a pier fixed top and bottom by a sway δ is 6EIδ/h², so it falls with the square of the pier height. That is why continuous rigid frames need tall piers, or twin thin-wall piers that are stiff enough for the cantilever stage but flexible along the bridge, and why a frame on short, stocky piers usually ends up with bearings after all.

Long continuous rigid frames have also taught a hard lesson: in many spans of 200 m and more, long-term midspan deflection and web cracking turned out far greater than predicted, because creep, shear lag and prestress losses had been underestimated. Present practice adds prestress margin and camber, and provides anchorages and deviators for spare external tendons that can be added if the span keeps sagging (section 6.4).

To nineteenth-century engineers a continuous girder was a problem: it is statically indeterminate, laborious to analyse by hand, and a settling pier changes all its moments. In 1866 the German engineer Heinrich Gerber patented an answer: put hinges into the continuous girder near the points where its moment would be zero anyway. The girder becomes a chain of anchor spans with cantilevers projecting from them and short suspended spans resting on the cantilever tips. The bending diagram stays close to that of the continuous beam, so most of the material saving is kept, but every force now follows from statics alone, and a settling support simply tilts the spans without stressing them. The first such girder crossed the Main at Hassfurt in 1867, and the system spread worldwide; in Japan and Korea it is still simply called the Gerber girder.

The cantilever principle reached its peak in steel trusses. The Forth Bridge in Scotland (1890, John Fowler and Benjamin Baker) has two main spans of 521 m, each made of two 207 m cantilevers carrying a 107 m suspended span. Baker explained it with a famous living model, in which the Japanese engineer Kaichi Watanabe sat on a plank between two men whose braced arms were the cantilevers. The Quebec Bridge (549 m) collapsed twice during construction, in 1907 when a compression chord buckled and in 1916 when the suspended span fell while being hoisted; it opened in 1917 and still holds the cantilever record.

Concrete followed. Many post-war balanced cantilever bridges were built with a hinge at midspan, which avoided the continuity tendons that were then difficult to design; the Bendorf Bridge over the Rhine (1964, 208 m) is one. The hinge proved to be the weak point. Creep makes the two cantilever tips droop for decades, the kink at the hinge grows, and the bearings and joints there wear out. The Koror-Babeldaob Bridge in Palau, a record 241 m span when it opened in 1977, had sagged by more than a metre at its hinge by the 1990s; it collapsed in 1996, three months after a repair that had made it continuous. Balanced cantilevers are now closed at midspan with a cast stitch and continuity tendons, and new concrete Gerber hinges are rare.

Two inheritances of the Gerber idea need special care in existing bridges. The first is the half-joint, or dapped end, where a suspended span sits on a notch in the cantilever: water leaking through the joint above runs straight onto the most highly stressed reinforcement, which cannot be inspected without lifting the span. The de la Concorde overpass in Laval, Quebec, collapsed at such a support in 2006, killing five people. The second is unseating in earthquakes: a suspended span on a short seat can be pulled off its support when the piers move apart, as happened to several spans at Kobe in 1995. AASHTO LRFD Article 4.7.4.4 therefore requires a minimum support length

N = (200 + 0.0017 L + 0.0067 H)(1 + 0.000125 S²) mm

where L is the length of deck to the adjacent joint (for a hinge within a span, the sum of the lengths on either side), H the average height of the adjacent piers, both in mm, and S the skew in degrees; N is then factored by 75 to 150 per cent according to the seismic zone (Table 4.7.4.4-1). For an in-span hinge with 60 m of deck on each side (L = 120,000 mm), piers 10 m tall and no skew, N = 200 + 204 + 67 = 471 mm, or about 710 mm at 150 per cent. Older hinges often have far less, which is why seat extensions and cable restrainers are among the most common seismic retrofits.

Expansion joints and bearings are the parts of a bridge that fail first. Joints leak, and the water that passes through them carries de-icing salt or sea spray onto the bearings, the pier caps and the girder ends below, exactly where inspection is hardest (Chapter 9). An integral bridge removes them: the deck is cast monolithically into the abutments, and the whole structure lengthens and shortens with temperature as one piece, pushing its abutments into the backfill and pulling them away again.

• Fully integral: the abutment is a stub wall on a single row of flexible piles, often steel H-piles oriented to bend about their weak axis, so the abutment can translate and rotate with the deck.
• Semi-integral: the deck and a deep end diaphragm move together on bearings set on a fixed abutment; the diaphragm pushes the backfill, but the foundation does not bend. Used on rock, behind tall abutments and on longer bridges.
• Jointless decks: over the intermediate piers of a chain of simply supported girders, a continuous deck or a thin link slab replaces the joints while the girders keep their own bearings.

The design question is movement. With a coefficient of thermal expansion for concrete of about 10 × 10⁻⁶ per °C, a 90 m integral bridge that warms and cools 25 °C either side of the temperature at which it was made continuous moves 10 × 10⁻⁶ × 25 × 45,000 mm ≈ 11 mm at each abutment, each way. Shrinkage and creep add a one-way shortening: for precast girders made integral a few months after casting, a remaining strain of the order of 200 × 10⁻⁶ adds another 9 mm. The abutment therefore swings between about +11 mm and −20 mm, and the piles and backfill must accept that every season for a century without damage. The uniform temperature component comes from AASHTO LRFD Article 3.12.2 or EN 1991-1-5; AASHTO Article 11.6.1.3 sets the general requirements for integral abutments.

The backfill is the subtle part. Each summer the abutment pushes into the soil; each winter it pulls away and loose granular fill slumps into the gap. Over the years this ratcheting densifies the fill, and the pressure on the wall climbs from the at-rest value toward passive. British practice (formerly BA 42/96, now PD 6694-1) gives a design coefficient K* that grows with the cyclic movement the wall imposes; other owners design directly for a passive envelope. The same movement opens a void under the approach slab, which is why integral bridges always have approach slabs, a sleeper or flexible joint at the far end of the slab, and free-draining, well-compacted backfill.

Many American agencies allow fully integral bridges of roughly 100–200 m, depending on material, with skews of up to 20–30°; the former UK durability standard BD 57 made integral construction the default for bridges up to 60 m long and 30° skew. The price of fewer joints is more design effort for soil-structure interaction, and more care with skew, curvature and the casting sequence of the end diaphragms. In hot, saline climates such as the Gulf the case is strong: a large temperature range is no obstacle for short bridges, and every joint removed is one less path for chloride.

Footbridges are where new structural systems are tried first: the loads are light, the spans moderate, and the public expects more than a plain girder. Section 13.27 covers the vibration problem in outline; this section describes the systems themselves.

The purest is the stress ribbon: a thin concrete deck, 0.25–0.4 m thick, cast on or around tendons that hang between two abutments like a cable. Deck and cable are one element, with no towers, girders or hangers; the cost is in the anchorages. Because people have to walk on it, a stress ribbon can sag only about L/50, and the horizontal pull H = wL²/(8f) of section 1.6 is correspondingly enormous.

Jiří Stráský built many of the best-known stress ribbons, first in the Czech Republic and then abroad, including the Sacramento River Trail Bridge at Redding, California (1990), a single span of about 127 m. Variants rest the ribbon on intermediate saddles or prop it from below with a shallow arch, which reduces both the sag and the anchorage force.

The other footbridge systems are the road-bridge families at smaller scale. Prefabricated steel trusses and girders of 15–60 m, lifted in overnight, are the economic default; arches and cable-stayed footbridges of 40–200 m are chosen for appearance as much as economy, the light loads allowing single masts, inclined arches and curved decks; suspension footbridges reach several hundred metres in mountain regions.

Whatever the system, a footbridge is designed for people, not trucks. The AASHTO LRFD Guide Specifications for the Design of Pedestrian Bridges use a crowd load of 4.3 kPa (90 psf); EN 1991-2 recommends q_fk = 2.0 + 120/(L + 30) kN/m², between 2.5 and 5.0. The governing check is usually vibration. The AASHTO guide asks for a vertical natural frequency above 3.0 Hz and a lateral frequency above 1.3 Hz, or else a check of the response; the European guides (Sétra, HiVoSS) instead define critical ranges of about 1.25–2.3 Hz vertically and 0.5–1.2 Hz laterally, comfort limits on acceleration, and a lateral lock-in threshold of about 0.1 m/s². Stress ribbons and slender cable-supported footbridges often fall inside those ranges and need either tuned mass dampers or a full dynamic analysis with crowd loading.

A railway bridge and a road bridge of the same span can look alike, but the road bridge is governed by strength under rare heavy trucks and the railway bridge by stiffness, fatigue and the behaviour of the track it carries.

Load. Load Model 71 of EN 1991-2 (Article 6.3.2) represents heavy rail traffic as four 250 kN axles at 1.6 m spacing with 80 kN/m on either side, unlimited in length; North America uses AREMA's Cooper E80 for the same purpose. On a 30 m simple span one track of LM71 gives a midspan moment of about 12,300 kNm, or 13,400 kNm with the dynamic factor Φ₂ = 1.44/(√L_Φ − 0.2) + 0.82 = 1.09 for well-maintained track and a determinant length L_Φ of 30 m. One lane of AASHTO HL-93 on the same span gives about 3,800 kNm, including the 33 per cent dynamic allowance on the truck. A single track therefore carries roughly three and a half times the load of a highway lane, before any national factor α for heavier freight, and sees it with every loaded train.

Fatigue. A highway girder sees its full design truck rarely; a railway girder sees a large fraction of its design load millions of times, so fatigue governs much of the steel in railway bridges and their details are kept simple and conservative.

Stiffness and the track. Rails are continuously welded and tied to the bridge through ballast or fastenings, so when the deck expands, bends or is braked against, the rails take part of the force. EN 1991-2 (6.5.4) limits the additional rail stress, for the standard case, to 72 MPa in compression and 92 MPa in tension, or the track may buckle or break. Longitudinal loads are large: braking 20 kN/m up to 6,000 kN per track, traction 33 kN/m up to 1,000 kN. Beyond an expansion length of about 60 m for steel decks and 90 m for concrete, rail expansion devices are normally needed. Deflection is tightly controlled: EN 1990 Annex A2 limits the vertical deflection under rail traffic to L/600 as a basic safety requirement and adds stricter limits for deck twist, end rotation and passenger comfort.

Speed. Above about 200 km/h, the regularly spaced axles of a train can drive a short span into resonance. EN 1991-2 then requires a dynamic analysis with real train types, and EN 1990 limits the peak deck acceleration to 3.5 m/s² for ballasted track and 5 m/s² for direct-fastened track, because above that the ballast loses its grip. That is why high-speed lines run on stiff, heavy concrete boxes, typically 24–40 m spans with span-to-depth ratios of about L/10–L/15, where a highway girder of the same span would be L/20 or more.

Construction depth. Track level is fixed by the line, and in towns there is little room between the rail and whatever lies below. Railway bridges therefore often place the structure beside the track instead of under it: half-through plate girders acting as U-frames, concrete trough decks and through trusses. A road can climb a metre in 25 m; a main-line railway needs 100 m or more to do the same.

Section 3.9 listed the factors that decide a bridge type. Here they are applied to a crossing typical of Gulf coastal developments, with numbers representative of such sites rather than of a specific project.

1. Span layout. The 100 m channel plus pier protection sets a main span of about 140 m. The other 960 m is approach viaduct over shallow water, the natural place for precast spans of 40–50 m (sections 3.1 and 3.2); the decision that matters is the main span.

2. Candidates. From the span ladder of Fig. 3.9: a haunched concrete box built by balanced cantilever, an extradosed bridge, a steel tied arch, and a cable-stayed bridge. The last fails at once on economy: at 140 m the towers, stays and anchorages cost more than the girder depth they save, and a cable-stayed main span begins to pay at about 200–250 m. Three options remain.

3. Profile. The deck must clear an envelope 100 m wide and 20 m high, so the depth that matters is the structure depth 20 m in from each pier, at the edge of the envelope.

At a 4 per cent grade, every metre of road level costs 25 m of approach at each end. The box needs about 55 m more approach on each side than the extradosed option, and about 100 m more than the arch: at 30 m width, roughly 3,300 m² and 5,900 m² of extra approach deck. That cost belongs to the main-span option, not to the approaches, and it often decides the comparison.

4. Construction. The box and the extradosed bridge are both built by balanced cantilever from the main piers: no falsework in the channel, and navigation kept open. The arch must be built on temporary supports in the channel or floated in whole by heavy-lift barge, which needs a fabrication yard, a weather window and a channel closure; on soft ground its tie helps, because the foundations then carry vertical load only.

5. Durability. The box is all concrete, with an interior to ventilate and inspect. The extradosed bridge adds stays, but at low stress ranges and replaceable strand by strand. The steel arch needs a marine coating renewed every 15–25 years, and its hangers are sensitive to fatigue and corrosion. Over a 100-year life in severe chloride exposure, the concrete options score better.

6. Appearance. A flat coastal setting with a new waterfront usually asks for a recognisable crossing. The box is plain and heavy at the piers; the arch is a landmark; the extradosed bridge is a middle course, its short pylons, about 12–18 m above the deck (L/8 to L/12), marking the channel without dominating the view.

Verdict. In most cost studies of this kind the box is the cheapest main span; whether it is the cheapest crossing depends on the approach length it adds. The extradosed option costs a little more for the main span, recovers part of that in shorter approaches and gives a visible gateway; for an owner who wants a landmark without the maintenance of steel, it is the likely choice. The tied arch is justified only where the image matters more than life-cycle cost. Change one input and the answer moves: a 250 m channel with 40 m clearance makes this a cable-stayed site, and a 60 m channel lets a deeper version of the approach girder cross it with no special span at all. Type selection compares whole crossings, not main spans.