Hertz contact stress calculator
When a ball, roller or wheel presses on another curved or flat surface, the contact shrinks to a tiny circle, strip or ellipse and the pressure there runs into thousands of MPa. Enter the radii, materials and load to get the contact size, the peak and mean pressure, how far the bodies close up, the stresses below the surface and the load at which the material first yields. Then check your finite element result against it. Free, no sign-up, runs in your browser.
Every result re-derived by an independent solver (surface-displacement check, Boussinesq and Flamant integration) · Updated 21 Nov 2026
Radius 0 means flat. A concave surface (a ball in a seat, a ball in a bearing groove) takes a negative radius. For the general case, x and y are the two principal directions shared by both bodies: a wheel has its rolling radius along x and a flat tread (0) along y; a rail is straight along x (0) and has its crown radius along y. Frictionless, linear elastic, smooth surfaces, contact small next to the radii.
What is Hertz contact stress?
Two curved bodies pressed together touch first at a point or along a line. Under load they flatten into a small patch: a circle for two spheres, a strip for two parallel cylinders, an ellipse for anything else, such as a wheel on a rail. Heinrich Hertz worked out the size of that patch and the pressure on it in 1882. The pressure is an ellipsoid: zero at the edge and largest at the centre, p0.
Because the patch is tiny, p0 is huge. A 25 mm steel ball resting on a steel plate under 1 kN, about the weight of a 100 kg person, makes a contact circle 0.88 mm across with a peak pressure of 2,464 MPa. The patch grows as the load rises, which is why contact stress is not proportional to load.
Hertz assumes the materials stay linear elastic, the surfaces are smooth and frictionless, and the patch is small compared with the radii and the bodies. Every result below follows from two numbers that combine the bodies:
A flat surface has 1/R = 0. A concave one, like a bearing race groove or a ball seat, has a negative radius, which makes R larger, so the contact is more conforming and the pressure is lower.
Hertz contact formulas for a sphere, a cylinder and an ellipse
Sphere on sphere or on a flat (circular contact)
δ is the approach: how far two distant points in the bodies move towards each other. The force grows with δ1.5. That is the Hertz contact law used by multibody codes (MotionView, Adams) for impacts. Its slope, the contact stiffness, is 2E*a.
Two parallel cylinders, or a cylinder on a flat (strip)
b is the half-width of the strip and P the load per unit length. A line contact has no finite Hertz approach: how far the bodies close up depends on their size and supports, not only on the contact. So this calculator gives δ for circles and ellipses only.
General bodies (elliptical contact)
K and E are the complete elliptic integrals of the first and second kind. The calculator solves the first equation for the eccentricity e exactly and does not use the curve-fit approximations found in handbooks. The long semi-axis a lies in the direction of the smaller relative curvature. For a wheel on a rail that is the rolling direction.
Why contact failure starts below the surface
On the axis under the centre of a circular contact, all three stresses are compressive and nearly equal at the surface, so the shear there is small. Lower down, σz fades more slowly than σr, and the difference between them, the shear, peaks inside the material:
That buried maximum is why rolling bearings, gear teeth and rails fail by rolling contact fatigue: a crack starts a fraction of a millimetre down, grows parallel to the surface and finally lifts out a pit or a spall. The one place a sphere contact pulls rather than pushes is the edge of the circle, where the radial stress is tensile, (1 − 2ν)p0/3 = 329 MPa for the 1 kN ball. In glass or ceramics that edge stress starts the ring cracks you see under a hard indenter.
The load at first yield
Set the peak von Mises equal to the yield strength Y. For a sphere (ν = 0.3):
The cube is the surprise. For the 25 mm ball on a plate of Ansys Structural Steel (Y = 250 MPa), first yield comes at 4.4 N, with a contact radius of 0.072 mm and an approach of 0.42 µm. Real bearings survive because bearing steel is hardened to several times that yield strength, and because a contact loaded somewhat beyond first yield can still settle back to elastic behaviour after the first few cycles (shakedown).
Worked example: a 25 mm steel ball on a steel plate under 1 kN
Both bodies are steel, E = 200 GPa and ν = 0.3; R1 = 12.5 mm and the plate is flat, so R = 12.5 mm.
Max shear 764 MPa and peak von Mises 1,528 MPa, 0.212 mm below the surface. The contact stiffness at this load is 2E*a = 96,752 N/mm. Press the first preset to see it in the calculator.
The "Sphere pressing on a plate" challenge: what Hertz says about a 4 mm push
Skill-Lync students have posted 154 reports for this Ansys Workbench challenge. A sphere is driven 4 mm into a plate. The plate is Ansys "Structural Steel NL", bilinear plastic (the reports quote a 250 MPa yield and a 1,450 MPa tangent modulus), and in most reports the sphere is plain linear-elastic Structural Steel. 129 reports use the nonlinear plate material and 99 set a friction coefficient of 0.74. None of them mentions Hertz. Three things a Hertz hand calculation tells you before you mesh:
- The elastic stage ends almost immediately. For a 25 mm ball on that plate, yield begins at an approach of 0.42 µm. The 4 mm push is about 9,634 times that, so nearly the whole run is plasticity and Hertz only covers its first instant. Enter your own sphere radius in the calculator, with the load set to "Approach δ", to get your number.
- A peak of 3,000 MPa is not a stress the steel can carry. 75 reports type the stresses they found. 63 of them quote a peak above 2,000 MPa; the median is 3,318 MPa. That is 13 times the 250 MPa yield and 7 times the 460 MPa ultimate strength of the material they assigned. A linear-elastic sphere never yields, so it reports whatever stress the strain implies. A bilinear plate with a gentle tangent slope keeps adding stress far past the strains where real steel would neck or tear. Report the plastic strain and the size of the dent, and treat the stress peak as a property of the material model, not of the steel.
- Friction does not matter in the elastic stage. With the same steel on both sides, the two surfaces would slide sideways by exactly the same amount under a normal load, so friction has nothing to resist. 0.74 and frictionless give the same Hertz contact. Friction starts to matter once the plate yields and the material flows outwards past the sphere.
Wheel on rail: the "Rail-wheel and track" challenge and the 35 mm mesh
Take a 915 mm wheel (the new diameter of Indian Railways' ICF and LHB coach wheels) with a flat tread on a 60E1 (UIC 60) rail, whose head has a 300 mm crown radius, both steel. In the calculator the wheel has R1x = 457.5 mm and R1y = 0 (flat), and the rail has R2x = 0 (straight) and R2y = 300 mm. At 100 kN, a little under the 112 kN static wheel load of a 22.9 t axle (the axle-load limit on most Indian Railways freight routes), the contact is an ellipse 14.5 × 11.0 mm with p0 = 1,200 MPa. At 500 kN it grows to 24.8 × 18.7 mm and p0 = 2,052 MPa, which is 1.71 times as much for five times the load, the cube root of 5.
The challenge asks for exactly those two loads. 39 reports are published, and 31 of them mesh the contact faces with 35 mm elements. One element is wider than the whole contact patch at either load, so the model cannot represent the pressure peak. In the 15 reports that type both results, the peak equivalent stress goes from a median of 146 MPa to 551 MPa, between 3.49 and 4.36 times as much (median 3.88) for five times the load. The contact pressure itself grows only 1.71 times. A peak that rises almost in step with the load comes from bending and bearing stresses away from the contact. It is not the contact stress, which on this geometry is close to 1,200 MPa at 100 kN.
To capture it, refine the rail head and wheel tread to well under a millimetre over a region a few times the patch size, with a frictional or frictionless contact and the load stepped up. Then compare the peak contact pressure with the Hertz value. With Structural Steel the rail would start to yield at 3.8 kN, so a 100 kN result above 250 MPa in a linear model means the material has already yielded. Real rail and wheel steels are much stronger, and their running surfaces work-harden in service.
What this calculator does not cover
- Plasticity beyond first yield, wear and residual stress: use an elastic-plastic FE model (it is what the sphere challenge is about).
- Sliding or rolling with traction. Surface shear moves the peak towards the surface and raises it; for a sliding line contact with a friction coefficient above about 0.3, yield starts at the surface.
- Conforming contacts (a pin in a hole with a small clearance, a ball deep in a close groove) where the patch is not small next to the radii, and edge loading of short rollers. Real rollers are crowned for that reason.
- Rough surfaces, coatings and layers, adhesion (JKR) at very small scales, and impact dynamics beyond the quasi-static Hertz law.
- Principal planes of the two bodies at an angle to each other: here they are aligned (or the bodies are crossed at 90°, which is the same thing).
Skill-Lync's CAE courses go from the theory behind contact, meshing and nonlinear materials to running and checking real contact and crash models in Ansys, HyperMesh and LS-DYNA, the work CAE teams hire for.
Frequently asked questions
What is Hertz contact stress?
It is the pressure and stress where two curved elastic bodies touch, worked out by Heinrich Hertz in 1882. The contact spreads into a small circle, strip or ellipse, the pressure peaks at the centre at 1.5 times the mean (circle) or 4/π times the mean (strip), and the most stressed point lies just below the surface, not on it.
How do you calculate the contact stress between a sphere and a flat plate?
Combine the elastic constants into E* and the radii into R (a flat has 1/R = 0). The contact radius is a = (3FR/4E*)^(1/3) and the peak pressure p0 = 3F/(2πa²). A 25 mm steel ball pressed on a steel plate with 1 kN gives a = 0.44 mm and p0 = 2,464 MPa.
Where does contact failure start?
Below the surface. Under a sphere the largest shear stress is 0.31 p0 at a depth of 0.48 times the contact radius, and von Mises peaks at 0.62 p0 at the same depth (for ν = 0.3). Under a cylinder it is 0.30 p0 at 0.79 times the half-width. That is why rolling contact fatigue starts as subsurface cracks and ends as pits or spalls.
What load makes a Hertz contact yield?
Yield starts below the surface when p0 reaches Y/0.62 = 1.61Y for a sphere, or about 1.79Y for a cylinder (von Mises, ν = 0.3). Because p0 grows only as the cube root of the load, the load at first yield goes as Y³: a 25 mm ball on a 250 MPa steel plate yields at just 4.4 N.
Does contact stress increase in proportion to the load?
No. For a sphere or an elliptical contact the patch grows with the load, so the peak pressure rises with the cube root of the load: five times the load gives 1.71 times the pressure. For a cylinder it rises with the square root. A finite element result that scales one to one with load has not resolved the contact.
What element size do I need for a contact analysis?
Small next to the contact patch, which is usually a few millimetres or less. Work out the Hertz half-width first and put several elements across it. The 100 kN wheel-on-rail contact here is 14.5 by 11.0 mm, so a 35 mm element is bigger than the whole patch and cannot show the pressure peak.
Does friction change the Hertz contact stress?
Not for a normal load on two bodies of the same material: their surfaces would slide sideways by the same amount, so friction has nothing to resist and the frictionless Hertz answer stands. With dissimilar materials friction changes the pressure only slightly. Sliding or rolling with traction is different: the surface shear moves the peak stress towards the surface.
Related on Skill-Lync
- A student report on the sphere pressing challenge
- A student report on the rail-wheel and track challenge
- Gear design calculator (Hertz contact at the pitch point of spur and helical gears)
- Mesh quality calculator
- Stiffness calculator
- Consistent units calculator for FE models
- All free engineering tools
Sources
- H. Hertz, "Über die Berührung fester elastischer Körper", Journal für die reine und angewandte Mathematik 92 (1882), pp. 156-171.
- K. L. Johnson, Contact Mechanics, Cambridge University Press (1985): chapters 3 and 4 (Hertz theory for circular, line and elliptical contact, the stresses on the axis), chapter 5 (normal contact with friction: no slip when the two bodies have the same elastic constants) and chapter 6 (onset of yield: p0 = 1.60Y under a sphere, von Mises, ν = 0.3).
- I. Green, "Poisson ratio effects and critical values in spherical and cylindrical Hertzian contacts", International Journal of Applied Mechanics and Engineering 10(3), 2005, pp. 451-462: first-yield p0/Y = 1.61 at 0.48a (sphere) and 1.79 at 0.70b (cylinder) for ν = 0.3, which our calculator reproduces.
- R. G. Budynas and J. K. Nisbett, Shigley's Mechanical Engineering Design, section 3-19, Contact Stresses (spherical and cylindrical contact, the stresses along the axis and the stress-depth charts; max shear 0.30pmax at z = 0.786b).
- S. P. Timoshenko and J. N. Goodier, Theory of Elasticity, 3rd ed. (Boussinesq's point load on a half-space and Flamant's line load): the solutions our verifier integrates over the contact pressure.
- Rail and wheel data: Techniques de l'Ingénieur M3070 "Railway rails" (60E1 head: 300 mm crown radius, then 80 mm and 13 mm); RDSO specification for LHB brake discs (wheel diameter 915 mm new); Ansys Workbench help (Structural Steel: E = 200 GPa, ν = 0.3; yield 250 MPa, ultimate 460 MPa).
- Skill-Lync student project reports for the "Sphere pressing on a plate" and "Rail-wheel and track" challenges (154 and 39 reports read in full on 21 Nov 2026; counts above).