Torsion Bar

A spring that is just a straight rod

Most springs announce themselves visually. A coil is an obvious bundle of helical wire; a leaf spring is a stack of curved steel straps. A torsion bar hides in plain sight: it is a plain straight rod, often indistinguishable from a length of round bar stock, and yet it is a complete suspension spring. The trick is the direction of the load. You do not push on a torsion bar; you twist it.

Clamp one end of the bar rigidly to the chassis — usually through a hex socket or a fine spline so the bar cannot slip — and attach a lever arm to the other end. The lever is the suspension control arm itself, or a separate crank pinned to it. When the wheel rises over a bump, the control arm swings, the lever rotates the free end of the bar, and the bar winds up along its length. The elastic material resists that twist with a restoring torque, exactly the way a stretched coil resists extension, and that restoring torque is the spring force that holds the car up. Release the load and the bar unwinds, returning the stored energy.

Crucially, all of the energy is stored as shear strain in the metal, not as bending. Every fiber along the length of the bar is sheared a little, and the sum of those tiny shear deformations over a long bar adds up to a usefully large twist at the lever. That is why length matters so much: a longer bar gives more material to share the twist, so it deflects more for the same torque and behaves as a softer spring.

The governing equations

The behavior of a round torsion bar reduces to two clean formulas. The first gives the angular deflection (in radians) under an applied torque T:

θ = T · L / (G · J)

where L is the working length of the bar, G is the shear modulus of the material (about 79 GPa for spring steel), and J is the polar moment of inertia of the cross-section. For a solid round bar of diameter d:

J = π · d⁴ / 32

Torsional spring rate:   k = T / θ = G · J / L  =  G · π · d⁴ / (32 · L)

The second formula gives the stress. Shear stress in the cross-section is zero on the axis and grows linearly outward, reaching its maximum at the surface:

τ(r) = T · r / J          (linear from 0 at center to τ_max at surface)

τ_max = T · (d/2) / J  =  16 · T / (π · d³)

Two consequences fall straight out of these. First, the spring rate scales with the fourth power of diameter (k ∝ d⁴) but only inversely with length (k ∝ 1/L), so diameter is by far the most sensitive knob. Doubling diameter makes the bar sixteen times stiffer; doubling length only halves the rate. Second, the surface does all the work — the core carries almost no shear stress, which is why a solid torsion bar is partly dead weight and why hollow bars are attractive when mass matters.

Worked example — sizing a road-car front bar

Suppose we want a torsional rate of about 760 Nm/rad at the bar, in spring steel with G = 79 GPa, and we have 1.0 m of length available along the frame rail. Solve k = G·π·d⁴ / (32·L) for d:

k = 760 Nm/rad,  G = 79e9 Pa,  L = 1.0 m

d⁴ = 32 · k · L / (G · π)
   = 32 · 760 · 1.0 / (79e9 · π)
   = 24320 / (2.482e11)
   = 9.80e-8 m⁴
d  = (9.80e-8)^0.25 ≈ 0.0177 m ≈ 17.7 mm   →  round up to 25 mm? check stress

Take d = 25 mm,  L = 1.0 m:
  J = π · (0.025)⁴ / 32 = 3.83e-8 m⁴
  k = 79e9 · 3.83e-8 / 1.0 ≈ 3026 Nm/rad   (too stiff at 25 mm/1 m)

To hit 760 Nm/rad at d = 25 mm, lengthen the bar:
  L = G · J / k = 79e9 · 3.83e-8 / 760 ≈ 3.98 m   (too long to package)

Practical compromise — d = 20 mm, L = 1.1 m:
  J = π · (0.020)⁴ / 32 = 1.571e-8 m⁴
  k = 79e9 · 1.571e-8 / 1.1 ≈ 1128 Nm/rad

Check peak stress at a working torque of, say, T = 450 Nm:
  τ_max = 16 · 450 / (π · 0.020³) = 7200 / (2.513e-5) ≈ 286 MPa  ✓ (well under ~600 MPa allowable)

The example shows the real design tension: the d⁴ term means a fat short bar is brutally stiff, so achieving a soft road rate usually pushes the designer toward a smaller diameter and a longer bar than first intuition suggests. Stress then sets the floor on diameter — you cannot keep shrinking d, because τ_max climbs with 1/d³ and will eventually exceed the fatigue allowance of the steel. The bar that ships is the smallest diameter (lightest, softest) that still keeps surface stress under the heat-treated allowable across the full bump-to-droop travel.

Materials, heat treatment and the surface

Because the surface carries the highest shear stress and is where every fatigue crack starts, torsion-bar metallurgy is obsessively focused on the skin. The workhorse alloys are silicon-manganese and chromium-vanadium spring steels — SAE 9260, 51CrV4 (also called 6150), and similar grades — hardened and tempered to give a high yield strength with retained toughness. A typical heat-treated spring steel reaches an ultimate tensile strength around 1,400 MPa, and the bar is run at a working surface shear stress of roughly 400 to 700 MPa.

• Shot peening. Blasting the surface with steel shot plastically dents the skin and leaves a compressive residual stress layer perhaps 0.2 mm deep. Because fatigue cracks open under tension, this compressive pre-load must be overcome before any service stress can crack the surface, which can more than double fatigue life.
• Presetting (scragging). The bar is deliberately twisted past its elastic limit in the operating direction during manufacture. When released, the outer fibers that yielded are left with a favorable residual shear stress that opposes the service load, so the bar can run at higher working stress without fatigue. This is why a torsion bar is directional — it is marked L or R and must be installed twisting the correct way.
• Surface finish. Grinding marks, decarburized skin from heat treatment, and corrosion pits are all crack initiators at the high-stress surface. Production bars are ground, sometimes polished, and protected against corrosion; a rusty pit on the working surface is a classic field-failure origin.

Solid vs hollow — chasing the dead core

Since shear stress falls to zero at the center, the inner material of a solid bar barely contributes to either stiffness or strength while still carrying its full mass. A hollow torsion bar removes that lazy core. For a tube with outer diameter d_o and inner diameter d_i, J = π·(d_o⁴ − d_i⁴) / 32, so a tube with a bore at 60 percent of the outer diameter loses only about 13 percent of its J (and stiffness) while shedding around 36 percent of its weight. The penalty is cost and the surface-stress sensitivity of the inner bore, which is harder to peen and inspect. Hollow bars therefore appear where mass is precious — motorsport, aerospace control linkages, and the lightest performance anti-roll bars — while mass-market suspension and tank torsion bars stay solid for cost and robustness.

How a torsion bar compares to other springs

A coil spring is, kinematically, a torsion bar wound into a helix: the wire of a coil is loaded almost entirely in torsion as the coil compresses. So the comparison below is really about packaging and tuning, not about a different physics.

Property	Torsion bar	Coil spring	Leaf spring
Energy storage mode	Pure shear (torsion)	Pure shear (torsion in the wire)	Bending
Packaging	Long thin rod along a frame rail	Compact vertical cylinder	Long flat stack under an axle
Energy per unit mass	High	High (≈ same as torsion bar)	Low — bending wastes the neutral axis
Ride-height adjustment	Trivial — rotate the anchor splines	Requires a different spring or perch	Requires re-arching or shims
Rate character	Linear (progressive is hard)	Easily progressive (variable pitch)	Mildly progressive (leaves stack up)
Failure mode	Fatigue crack at surface → corner drops	Fatigue crack → corner drops	Cracked leaf, often partial sag
Typical use	Tanks, classic VW/Porsche, anti-roll bars	Most modern passenger cars	Trucks, trailers, solid rear axles

Where torsion bars actually show up

• Classic passenger cars. The VW Beetle and Type 2 bus, the early Porsche 911 and 912, the Renault 4, and the Citroen 2CV (which ran a clever horizontal bar interconnecting front and rear) all used torsion-bar suspension. Chrysler's "Torsion-Aire" front suspension put longitudinal bars on a generation of Dodge and Plymouth cars.
• Trucks and SUVs. Many body-on-frame trucks (Ford Ranger, GM and Dodge pickups, the Hilux) used front torsion bars for decades precisely because ride height is adjustable with a bolt — invaluable for load leveling and trim.
• Tracked military vehicles. Transverse torsion bars running across the hull floor are nearly universal: the WWII Panther and Tiger II pioneered them, and the T-72, Leopard 2, and M1 Abrams all suspend their road wheels this way. Laying the spring flat across the floor keeps the hull low and leaves room above for the turret basket.
• Anti-roll (sway) bars. Almost every car on the road carries a torsion bar you rarely think about. A transverse bar links the left and right suspension; when both wheels rise together it merely rotates, but when one wheel rises and the other does not, the bar twists and resists body roll. Same equation, different duty cycle.
• Mechanisms and instruments. Torsional springs scale down to clock balance-wheel hairsprings, torsion pendulums (the 400-day "anniversary" clock), and the suspension fibers of sensitive torsion-balance instruments — the same family Cavendish used to weigh the Earth.

Failure modes and trade-offs

• Surface fatigue cracking. The dominant failure. The bar sees fully reversed or pulsating shear at the surface millions of times; a crack nucleates at a surface defect, corrosion pit, or decarburized layer and propagates around the bar until it snaps. Shot peening and presetting exist specifically to push this out past the design life.
• Total loss when it breaks. Unlike a multi-leaf spring that can limp on a cracked leaf, a snapped torsion bar drops that corner of the vehicle completely. There is no graceful degradation, which is why inspection and corrosion protection are taken seriously.
• Mass. For the same energy stored, a torsion bar is generally heavier than an equivalent coil, mostly because of that under-stressed central core. Hollow bars claw some of that back at a cost penalty.
• Linear rate. A plain bar gives a constant rate. Achieving a progressive (rising-rate) curve needs tricks — tapered bars, multiple bars engaging in stages, or rising-rate lever geometry — whereas a coil gets it almost for free with variable pitch.
• Anchor and lever loads. All of the torque has to react somewhere. The fixed anchor and the lever splines see high local stress and must not slip or fret; a slipped spline quietly changes the ride height and corrupts the geometry.
• Set (creep). Run at too high a stress for too long and the bar takes a permanent twist — "set" — sagging the corner. Conservative working stress and presetting guard against it.

Common pitfalls when designing with torsion bars

• Underestimating the d⁴ sensitivity. A few hundredths of a millimetre of grinding error on diameter shifts the spring rate by several percent. Production bars are ground to tight tolerance, tested, and matched in left/right sets.
• Installing a directional bar the wrong way. Preset bars are handed (L/R). Fit one backwards and the favorable residual stress now opposes the wrong direction, slashing fatigue life.
• Forgetting the lever ratio. The bar sees torque, the wheel sees force. Wheel rate depends on the square of the control-arm lever length, so a geometry change quietly rescales the effective spring rate even with the same bar.
• Ignoring stress at the spline roots. The transition from the round working section into the splined or hex anchor is a stress concentration. Generous fillet radii and rolled or peened transitions are essential.
• Neglecting corrosion protection. A torsion bar lives under the car in the spray zone. An unprotected surface pits, and a pit on the high-stress skin is a fatigue crack waiting to start.