Hardness and Fatigue: How Metal Hardness Governs Fatigue Strength and Damage Tolerance

Hardness is cheap and fast to measure — but how far can it really predict fatigue behavior?

Introduction

Hardness is one of the cheapest and fastest mechanical properties to measure — a Rockwell or Brinell indenter, a few seconds, and a single number. Because hardness testing is so accessible, engineers have long used it as a proxy for other properties, most notably tensile strength and fatigue strength. This correlation is genuinely useful, but it is also easy to over-apply. This post looks at why hardness correlates with fatigue strength, where that correlation breaks down, and how hardness affects fatigue damage accumulation and crack growth — not just the number of cycles to crack initiation.

Why Hardness Correlates with Fatigue Strength

The Physical Link

Hardness is fundamentally a measure of a material’s resistance to localized plastic deformation. The same microstructural features that resist indentation — fine grain size, dislocation density, precipitate hardening, martensitic transformation — also resist the cyclic plastic deformation that drives fatigue crack initiation. This shared physical origin is why hardness tracks reasonably well with both ultimate tensile strength (UTS) and fatigue strength.

The Classical Approximation

For many wrought steels, the empirical relationship between hardness and ultimate tensile strength is approximately linear:

\[ \text{UTS (MPa)} \approx 3.45 \times HB \]

where \(HB\) is the Brinell hardness number. From there, the fatigue limit (endurance limit) of steels is often estimated as a fraction of UTS:

\[ S_e \approx 0.5 \times \text{UTS} \quad (\text{for UTS} \lesssim 1400\ \text{MPa}) \]

Combining these gives a commonly used rule of thumb:

\[ S_e\ (\text{MPa}) \approx 1.6\ \text{to}\ 1.75 \times HB \]

This is a useful first-pass estimate when full S/N data are not available — but it is an approximation with real limitations, discussed below.

Where the Correlation Holds Well

Through-hardened, homogeneous steels — the hardness measurement reflects the bulk microstructure that also controls fatigue crack initiation.
Moderate hardness ranges (roughly 200–400 HB) — where the material remains reasonably ductile and fatigue is initiation-dominated.
Smooth, polished specimens — free of stress concentrations, where the fatigue limit is genuinely controlled by microstructural resistance to slip band formation.

Where the Correlation Breaks Down

High Hardness and Notch Sensitivity

As hardness increases beyond roughly 400–450 HB, the linear \(S_e\)-versus-hardness relationship becomes unreliable and often overestimates fatigue strength. High-hardness, high-strength steels become increasingly notch sensitive — a small surface defect, inclusion, or machining mark that would be tolerated in a softer, more ductile material becomes a potent stress concentrator and crack initiation site in a hard one. The fatigue notch sensitivity factor \(q\) approaches 1.0 (full notch effect) as hardness and strength increase, whereas softer, more ductile steels have \(q\) values well below 1.0 because local plasticity blunts the stress concentration.

Inclusions and Subsurface Initiation

In very high strength, high hardness steels (typically above 50 HRC), fatigue cracks frequently initiate not at the surface but at subsurface non-metallic inclusions — oxides, sulfides, or carbides. This shifts the governing fatigue mechanism away from surface slip-band nucleation (which hardness correlates with) toward inclusion size and distribution statistics (which hardness does not directly capture). This is the basis of the Murakami \(\sqrt{\text{area}}\) model, which predicts fatigue limit from inclusion size and hardness jointly rather than hardness alone:

\[ \sigma_w = \frac{C \cdot (HV + 120)}{(\sqrt{\text{area}})^{1/6}} \]

where \(\sigma_w\) is the fatigue limit, \(HV\) is Vickers hardness, \(\sqrt{\text{area}}\) characterizes the projected inclusion size, and \(C\) depends on inclusion location (surface, subsurface, or interior) and loading type.

Surface Treatments Decouple Hardness from Bulk Behavior

Case hardening (carburizing, nitriding, induction hardening) creates a hardness gradient rather than a uniform value. The surface hardness controls resistance to crack initiation, but the case depth and the residual compressive stress profile control how far a crack can grow before it reaches the softer, tougher core. A single bulk hardness number cannot describe this. Two parts with identical surface hardness but different case depths can have substantially different fatigue lives.

Overaging and Microstructural Instability

In some precipitation-hardened aluminum and nickel alloys, hardness and fatigue strength can diverge if the microstructure is metastable. Cyclic loading at elevated temperature can cause precipitate coarsening (overaging) during service, softening the material locally even though the as-manufactured hardness was high. The initial hardness measurement does not capture this time-dependent degradation.

Hardness and Crack Growth — Not Just Initiation

Most hardness–fatigue correlations focus on the fatigue limit, which is dominated by crack initiation resistance. But hardness also affects the crack propagation phase, and often in the opposite direction.

The Strength–Toughness Trade-off

Increasing hardness generally increases yield and tensile strength, which raises resistance to crack initiation. But it typically decreases fracture toughness \((K_{IC})\) and can increase the Paris Law crack growth rate constant \(C\) for a given stress intensity range \(\Delta K\):

\[ \frac{da}{dN} = C \, (\Delta K)^m \]

This means:

Harder materials often initiate cracks later (higher fatigue limit)
Harder materials often propagate cracks faster once initiated, and fail at a smaller critical crack length (lower fracture toughness)

The practical consequence is that a harder material can have a longer crack initiation life but a shorter, more abrupt crack propagation life — meaning less warning between detectable crack size and final fracture. This is a critical consideration in damage-tolerant design, where inspection intervals depend on the time available between a detectable crack and the critical crack length:

\[ a_c = \frac{1}{\pi} \left( \frac{K_{IC}}{Y \sigma} \right)^2 \]

where \(a_c\) is the critical crack length, \(Y\) is a geometry factor, and \(\sigma\) is the applied stress.

Residual Stress Interactions

Many hardening processes (shot peening, case hardening, cold rolling) introduce beneficial compressive residual stresses at the surface, in addition to raising local hardness. These residual stresses suppress crack initiation and slow early crack growth independent of the hardness increase itself. This is why shot-peened parts often show much greater fatigue improvement than a hardness-based \(S_e\) estimate alone would predict — the benefit is a combination of hardness increase and residual stress, and separating the two contributions requires residual stress measurement (X-ray diffraction), not hardness testing alone.

Practical Guidance for Fatigue Analysis

Use hardness-based S_e estimates only as a screening tool: Reasonable for preliminary sizing or comparing candidate materials in the same class, but should not replace S/N test data for fatigue-critical design.
Apply notch sensitivity corrections for high-hardness materials: As hardness rises above roughly 400 HB, apply the fatigue notch sensitivity factor \(q\) explicitly rather than assuming a smooth-specimen fatigue limit applies at stress concentrations.
Check for subsurface initiation risk in high-strength steels: For hardness above roughly 50 HRC, consider inclusion-based fatigue limit models (e.g., Murakami) in addition to or instead of bulk hardness correlations, particularly for components with high cleanliness requirements (bearings, gears, fasteners).
Characterize case-hardened parts by profile, not a single number: Case depth, core hardness, and the residual stress profile all matter — a surface hardness spec alone is insufficient for fatigue life prediction in carburized or nitrided components.
Remember the toughness trade-off in damage-tolerant design: A harder material with a higher fatigue limit may have less residual life once a crack is detected. Inspection intervals should account for the higher Paris Law crack growth rate and lower fracture toughness typical of hardened states.
Separate hardness effects from residual stress effects: When evaluating shot peening, cold working, or case hardening, do not attribute the full fatigue benefit to hardness alone — residual compressive stress is often the larger contributor and behaves differently under elevated temperature or overload relaxation.

Summary

Hardness is a convenient and physically meaningful proxy for fatigue strength because both properties share a common root in a material’s resistance to localized plastic deformation. The classical \(S_e \approx 1.6\text{–}1.75 \times HB\) relationship is a useful first estimate for through-hardened, moderate-strength steels.

But the correlation weakens or breaks down at high hardness levels, where notch sensitivity rises, subsurface inclusion-driven initiation dominates, and — critically — increasing hardness often comes at the cost of reduced fracture toughness and faster crack propagation. Fatigue-critical design should treat hardness as one input among several: notch sensitivity, inclusion cleanliness, residual stress state, and fracture toughness all need to be considered alongside the hardness number to get a complete picture of fatigue performance.