How Bridges Carry Load

July 2, 2026 14 min read Engineering the Everyday

structural-engineeringcivil-engineeringbridgesmechanicsmaterials-science

Contents

A bridge has exactly one job — move load from a point in space where there’s nothing underneath it to a point where there is solid ground — and every bridge design in existence is a different strategy for making sure no single structural member ever has to fight its own material properties to do that job. Steel and steel cable are superb in tension and mediocre in compression when unsupported (they buckle). Concrete and masonry are excellent in compression and nearly useless in tension (they crack). The entire discipline of bridge typology — why some spans are a flat beam on piers, some are a soaring arch, some hang from cables off towers — comes down to how cleverly each design routes force through its members so that tension goes to something good at tension and compression goes to something good at compression. Get that routing wrong, or let it degrade unnoticed, and you get catastrophic failure — which is why the history of bridge engineering is written as much in its failures as its successes.

The Two Forces That Explain Everything

Any structural member under load experiences some combination of two basic states. Compression squeezes a member — pushes its two ends toward each other — and pure compression is what plain concrete, stone, and masonry handle extremely well, because unreinforced brittle materials fail by cracking apart under tension, not by being squeezed. Tension stretches a member — pulls its two ends apart — and this is where steel dominates, because a steel cable or rod can be pulled to enormous force with minimal elongation before it yields, while masonry under the same pull simply splits.

   COMPRESSION                    TENSION
   (squeezing)                   (stretching)

   ═══►    ◄═══                ◄═══    ═══►
   [========]                  [========]
   good material: concrete,    good material: steel,
   stone, masonry              cable, steel rod

   fails by: crushing/         fails by: yielding/
   buckling                    snapping

A third condition, bending (or flexure), is really just tension and compression happening simultaneously within the same member — one side of a bent beam stretches while the opposite side compresses, with a neutral axis in between experiencing neither. This is precisely why a plain concrete beam is a bad structural idea on its own: the bottom face of a beam under downward load is in tension, exactly the condition unreinforced concrete is worst at, which is the direct throughline to why steel rebar gets embedded specifically in the tension zones of concrete members — a mechanism covered in detail in Concrete and Cement: The Chemistry of Civilization. Bridge design, at its core, is the art of choosing a structural shape that routes bending loads into pure tension in some members and pure compression in others, rather than leaving any one member to suffer both simultaneously.

Beam Bridges: The Simplest Answer, and Its Limit

A beam bridge is the most primitive form — a rigid horizontal span resting on supports at each end (or at intermediate piers), carrying load purely through its own bending stiffness. Under a load in the middle of the span, the top of the beam goes into compression and the bottom goes into tension, exactly the bending condition described above, with the piers or abutments underneath absorbing the resulting vertical compression straight down into the ground.

The limitation is structural, not aesthetic: bending stress in a beam increases with the square of the span length for a given load, so a beam’s practical span is capped by how much material you’re willing to add to keep the beam stiff enough not to sag or crack. Beyond a few hundred feet, a solid beam becomes impractically heavy, which is why long beam-type spans are almost always built as a truss (a beam with material hollowed out into a triangulated lattice, discussed below) or handed off to an arch or cable-based system instead.

Arch Bridges: Turning Bending Into Pure Compression

An arch bridge’s entire trick is geometric: instead of resisting bending directly, it routes the load down through a curved compression path so that, ideally, every point along the arch experiences pure compression and nothing else. Load applied anywhere on the deck travels down through the arch ring toward the keystone at the crown, then outward and down along the curve to the abutments at each end, which must resist both the vertical load and a significant outward horizontal thrust the curved shape generates.

This is why arch bridges are the oldest long-span technology in the historical record — Roman engineers built durable masonry arches specifically because masonry is excellent in compression and the arch shape is the one geometry that can carry a distributed load using compression alone, without needing any tensile reinforcement at all. The trade-off is that an arch demands genuinely solid abutments capable of resisting that horizontal thrust; an arch built on soft or unstable ground can fail not because the arch ring itself crushed, but because the foundations spread apart under the outward push and let the whole curve flatten and collapse.

Suspension Bridges: An Arch Turned Inside Out

A suspension bridge can be understood, almost literally, as an arch flipped upside down: instead of a compression curve pushing outward into abutments, a suspension bridge hangs the deck from a tension curve — the main cable — that pulls inward and downward on two towers. Vertical hanger cables drop from the main cable to the deck, transferring the deck’s load up into the main cable, which carries it in pure tension along its curved length to the towers. The towers then carry that force straight down to the ground in compression, and the ends of the main cable are locked into massive anchorages that resist the substantial inward pull the cable’s tension exerts.

The mathematics of that main cable shape is a genuinely elegant piece of physics: a cable hanging under only its own weight, uniformly distributed along its own length, settles into a catenary curve; but a suspension bridge’s main cable is carrying a deck load that’s uniformly distributed across the horizontal span rather than along the cable’s own length, and under that different loading condition the cable instead settles into a parabola. In practice, for bridges where the cable’s sag is a reasonably large fraction of the span, the two curves are close enough that engineers treat the main cable as parabolic for design purposes — a genuine mathematical simplification, not just a convenient approximation for hand sketches.

Because the entire load-bearing element is high-strength steel cable working purely in tension, suspension bridges achieve by far the longest clear spans of any bridge type — the Akashi Kaikyō Bridge in Japan held the world’s longest suspension main span for over two decades at 1,991 meters (roughly 6,532 feet) before being surpassed by Turkey’s 2022 Çanakkale Bridge — because tension-loaded steel cable can be made enormously strong relative to its weight in a way that no compression member of comparable span could match without becoming impossibly massive.

Cable-Stayed Bridges: Tension Without the Catenary

A cable-stayed bridge looks superficially similar to a suspension bridge — towers, cables, a suspended deck — but the load path is structurally distinct. Instead of hanging the deck from one continuous main cable via vertical hangers, cable-stayed bridges run individual diagonal cables directly from the towers to specific points along the deck, each cable carrying its own local section of deck load in essentially straight-line tension back to the tower. The towers then carry the combined cable forces down to the foundation in compression, exactly as in a suspension bridge — but because the diagonal cables also pull the deck inward toward the towers, cable-stayed decks additionally end up under horizontal compression along their own length, a load path suspension bridge decks don’t experience in the same way.

The practical consequence is a real trade-off between the two tension-cable bridge types: cable-stayed bridges need no massive cable anchorages (each cable’s tension is resolved locally at the tower and the deck, rather than requiring a giant ground anchorage to react against), which makes them cheaper and faster to build at moderate spans, but the direct diagonal cable geometry doesn’t scale to the extreme clear spans a continuous suspension main cable can achieve, which is why the very longest bridge spans in the world remain suspension designs rather than cable-stayed ones.

Bridge type	Primary load path	Key member in tension	Key member in compression	Practical span range
Beam	Bending within the beam	Bottom of beam	Top of beam, piers	Short (tens to ~250 ft without trussing)
Truss	Bending resolved into axial member forces	Bottom chord, select diagonals	Top chord, select diagonals/verticals	Short–medium (up to ~1,000+ ft)
Arch	Curved compression path	None (ideally pure compression)	Entire arch ring, abutments	Medium–long (up to ~1,800 ft)
Suspension	Cable tension to towers/anchorages	Main cable, hangers	Towers	Longest (world record ~6,500+ ft)
Cable-stayed	Direct diagonal cable tension to towers	Stay cables	Towers, deck (horizontal compression)	Medium–long (up to ~3,500 ft)

Truss Bridges: Turning a Beam Into a Triangle Machine

A truss bridge takes the beam concept and hollows it out into a triangulated framework of individual straight members, each one designed to carry force purely axially — in pure tension or pure compression along its own length — rather than in bending. Triangles are the geometric key: a triangle is the only polygon that can’t change shape without one of its sides changing length, which means a triangulated framework is inherently rigid using far less material than a solid beam of equivalent stiffness.

Under a load causing the overall truss to bend (as any beam-like span does), the top chord running along the truss’s upper edge goes into compression, the bottom chord goes into tension, and the diagonal and vertical web members in between take on tension or compression depending on their specific orientation and position relative to the load — some diagonals working in tension, others in compression, by design. Because each member carries a simple, calculable axial force rather than complex bending stress, trusses can span considerably farther than a solid beam of the same total weight, which made them the dominant technology for railway and highway bridges throughout the 19th and much of the 20th century, particularly before high-strength cable and modern arch/suspension engineering matured.

       top chord (compression) ═══════════════
              ╱╲    ╱╲    ╱╲    ╱╲    ╱╲
             ╱  ╲  ╱  ╲  ╱  ╲  ╱  ╲  ╱  ╲
            ╱    ╲╱    ╲╱    ╲╱    ╲╱    ╲
       bottom chord (tension) ═══════════════
            ▲                              ▲
         support                        support

   diagonals alternate tension/compression
   depending on load direction and position

How Engineers Reason About Failure

Every bridge is designed with a safety factor — the ratio between a member’s actual load capacity and the maximum load it’s ever expected to see in service — precisely because loads, material properties, and construction quality all carry real-world uncertainty. But safety factors only work if every load path in the design was correctly identified and correctly sized in the first place, and the two most instructive bridge failures in modern engineering history are lessons in exactly how that assumption breaks down.

Tacoma Narrows (1940) is the failure most people know, and it’s also the one most commonly misexplained: it’s frequently taught as simple mechanical resonance, but the actual mechanism was aeroelastic flutter — a self-reinforcing feedback loop between wind and the bridge deck’s own motion, distinct from resonance driven by a periodic external force at a structure’s natural frequency. The bridge’s deck was a solid, shallow plate girder, chosen for a slender, economical appearance, with a depth-to-span ratio far below prevailing engineering norms of the era. That solid girder shape behaved aerodynamically like an airfoil rather than letting wind pass through it the way an open truss deck does, and under sustained wind it built up a self-amplifying twisting oscillation that grew from an initial 1.5-foot vertical wave motion into a catastrophic 28-foot torsional twist within hours, tearing the deck apart. The direct engineering legacy is concrete and current: wind-tunnel testing of proposed suspension bridge deck sections is now standard practice, and modern suspension bridge decks are either aerodynamically shaped or built with open-truss stiffening specifically to prevent the same flutter feedback loop from ever developing.

I-35W Minneapolis (2007) is a different category of failure entirely — not aerodynamics, but a straightforward, decades-latent design error. The National Transportation Safety Board determined the probable cause was a set of undersized gusset plates — the flat steel connector plates joining truss members together — at a specific set of nodes on the bridge’s deck truss, plates that were roughly half the thickness the design loads actually required due to a calculation error made during the bridge’s original design in the 1960s. That error sat undetected for over three decades, through routine inspections that — as the NTSB specifically noted — gave inadequate scrutiny to gusset plates and excluded them from load-rating analysis entirely, treating them as incidental hardware rather than primary structural members. The plates finally failed under the combined weight of accumulated deck resurfacing from prior renovations and concentrated construction equipment and material staged on the bridge on the day of the collapse — extra load the original 1960s design margin was never sized to absorb on top of an already-undersized connection. The direct engineering legacy: gusset plates are now explicitly required to be included in bridge load-rating calculations and inspection protocols nationwide, closing exactly the blind spot that let a 1960s calculation error go undetected for over forty years.

Failure	Year	Root mechanism	Direct engineering legacy
Tacoma Narrows	1940	Aeroelastic flutter from a solid, under-stiffened deck girder	Mandatory wind-tunnel testing for suspension bridge deck sections
I-35W Minneapolis	2007	Undersized gusset plates from a 1960s design calculation error, compounded by added deck weight and concentrated construction load	Gusset plates now mandatory in load-rating and inspection protocols

Honest Trade-offs

Longer spans are never free. Every increase in achievable span length — beam to truss to arch to cable-stayed to suspension — comes with a corresponding increase in construction complexity, foundation demands (arch abutments, suspension anchorages), and long-term maintenance burden (cable corrosion protection, deck aerodynamics), not just a bigger version of the same structure.
Redundancy is a design choice, not a given. The I-35W truss was a “fracture-critical” design, meaning the failure of certain single members (like the undersized gusset plates) could cascade into total collapse with no alternate load path — a design category modern bridge engineering treats far more cautiously specifically because of that collapse.
Aerodynamics matters more as spans get longer and decks get slenderer. The same slenderness that makes a modern cable-supported bridge efficient and visually striking is exactly what makes flutter and wind-induced oscillation a live design concern, not a solved historical footnote.
Inspection scope is only as good as its assumptions. The I-35W collapse wasn’t purely a design failure — it was also an inspection-and-maintenance-process failure, because the process itself didn’t consider gusset plates worth scrutinizing, a reminder that a structurally sound design on paper can still fail if the ongoing inspection regime doesn’t cover every load-bearing element the original calculations relied on.

Verdict

Every bridge type is a different geometric strategy for the same underlying problem: get load from the deck to the ground while asking each material to do only what it’s good at — tension for steel and cable, compression for concrete, stone, and steel columns. Beams bend and accept some of both; trusses triangulate that bending into clean axial tension and compression in separate members; arches route load into pure compression through curvature; suspension and cable-stayed designs invert that idea entirely, routing load into pure cable tension and letting towers carry the resulting compression to the ground. The failures that shaped modern bridge codes — Tacoma Narrows and I-35W — weren’t failures of these basic principles; they were failures to correctly account for a force (aerodynamic flutter) or a connection (undersized gusset plates) within an otherwise sound structural concept, which is exactly why modern bridge engineering treats wind-tunnel testing and full-structure load rating as non-negotiable rather than optional refinements.