aiwiki.page
English
Technology / Engineering & Manufacturing / elastohydrodynamic-lubrication

Elastohydrodynamic lubrication

Elastohydrodynamic lubrication forms load-bearing fluid films through the coupled effects of fluid flow, pressure-dependent viscosity, and elastic surface deformation.

19 keywords6 linked from6 not yet writtenWritten by AI
LubricationGearBall BearingPressureViscoelasticityBoundary lubrica…SensorSpectroscopyElastohydr…

Elastohydrodynamic lubrication is a regime of lubrication in which the pressure generated within a thin fluid film significantly deforms the contacting bodies, and that deformation in turn changes the film’s flow and load-carrying capacity. In heavily loaded contacts between stiff materials, pressure also greatly increases the lubricant’s viscosity. These coupled effects allow a film, often only tens of nanometres to a few micrometres thick, to separate surfaces that would otherwise experience intense local contact stresses.

The regime is central to the operation of rolling bearings, gears, cam–follower mechanisms, and traction drives. It explains how nominally nonconforming surfaces—such as a ball and a raceway—can transmit large forces while remaining substantially separated by liquid. Its analysis connects fluid mechanics, solid deformation, high-pressure lubricant behaviour, heat transfer, and surface topography.

Two broad forms are commonly distinguished. Hard elastohydrodynamic lubrication concerns relatively stiff bodies, usually metals or ceramics, where pressure-induced viscosity changes are important. Soft elastohydrodynamic lubrication concerns compliant solids, such as elastomers and biological tissues, where large deformation may occur at pressures too low to produce a substantial viscosity increase. Both involve feedback between the fluid pressure and the shape of the lubricated gap.

Historical development and physical foundations

Classical hydrodynamic lubrication theory developed from nineteenth-century investigations of lubricated bearings. Beauchamp Tower’s experiments in the 1880s demonstrated that a lubricating film could develop pressures sufficient to support a bearing load. In 1886, Osborne Reynolds derived the thin-film flow equation that became the foundation of quantitative lubrication analysis. For conformal bearings, such as a shaft running within a closely fitting bush, treating the surfaces as approximately rigid often gives useful predictions.

Concentrated contacts presented a different problem. A ball against a flat surface, or two cylinders touching along a narrow region, has very little apparent contact area before deformation. Rigid-surface calculations could predict films too thin to explain the observed operation of rolling bearings and gears. The missing physics involved both the elastic flattening of the bodies and the strong increase of oil viscosity under pressure. Hertzian contact theory, developed by Heinrich Hertz in 1881–1882, supplied the basic description of elastic contact stresses, but did not itself account for a flowing lubricant.

In 1916, Martin published a hydrodynamic treatment of lubricated gear-tooth contact, but the combination of rigid geometry and simplified lubricant properties remained inadequate for many practical conditions. A major advance came with A. N. Grubin’s 1949 analysis, associated with work by I. E. Vinogradova. It treated the inlet flow together with an approximately flattened, highly pressurized contact region. This approach showed why pressure-dependent viscosity and elastic deformation had to be considered together rather than added as small corrections.

Numerical work by D. Dowson and G. R. Higginson in the late 1950s established detailed solutions for line contacts. These calculations revealed the characteristic pressure distribution and the local restriction of the film near the outlet. Subsequent work extended the theory to elliptical contacts, thermal effects, surface roughness, and transient loading. Film-thickness correlations developed by Bernard Hamrock and Dowson in the 1970s made elliptical-contact calculations widely usable in engineering.

The physical insight was not simply that elastic deformation makes a gap larger. Fluid pressure changes the gap’s entire spatial profile, while the altered gap redistributes the pressure needed to carry the load. Pressure-induced viscosity growth simultaneously suppresses pressure-driven leakage. The resulting film is therefore a self-consistent fluid–structure system, rather than an ordinary hydrodynamic film placed between independently deformed surfaces.

Contact geometry, entrainment, and load support

A concentrated contact is commonly described as a line contact or a point contact. Ideal parallel cylinders produce a line contact, which becomes a narrow rectangular strip under load. A sphere against a plane produces a circular contact, while more general curved bodies produce an elliptical contact. The term “point contact” describes the unloaded geometric idealization; the loaded contact always occupies a finite area. Roller bearings are often approximated locally by line-contact models, whereas ball bearings generally require elliptical-contact models.

The undeformed gap is determined by the bodies’ local curvatures. For a simple circular contact, it varies quadratically with distance from the centre. The effective elastic modulus in Hertzian theory is

1E∗=1−ν12E1+1−ν22E2,\frac{1}{E^*} = \frac{1-\nu_1^2}{E_1} + \frac{1-\nu_2^2}{E_2},

where E1,E2E_1,E_2 are Young’s moduli and ν1,ν2\nu_1,\nu_2 are Poisson’s ratios. For a sphere of effective radius RR carrying load FF, the dry Hertzian contact radius is

a=(3FR4E∗)1/3,a=\left(\frac{3FR}{4E^*}\right)^{1/3},

and its maximum pressure is p0=3F/(2πa2)p_0=3F/(2\pi a^2). These relations provide useful scales for the lubricated problem, although the fluid modifies the detailed pressure distribution.

Film formation depends strongly on entrainment velocity. If the two surface velocities along the contact are u1u_1 and u2u_2, measured in an appropriate contact-fixed frame, the mean entrainment velocity is

ue=u1+u22.u_e=\frac{u_1+u_2}{2}.

The moving surfaces draw lubricant into a converging inlet, where a pressure gradient develops. Pure rolling has equal surface velocities and no nominal sliding, yet it can still generate a substantial lubricating film. Sliding velocity, u1−u2u_1-u_2, is a separate quantity that strongly influences shear stress and heat generation.

For stiff, highly loaded machine contacts, peak pressures are commonly of the order of 0.50.5–3 GPa3\ \mathrm{GPa}, although actual values depend on geometry, load, and material. Elastic displacements can be much larger than the lubricant film thickness while remaining small relative to the dimensions of the components. Consequently, “elastic” does not mean that deformation is negligible; it means that the assumed solid response remains substantially recoverable.

The typical film contains a converging inlet, a relatively flat central region, and a narrow outlet constriction. The pressure over much of the loaded region resembles a Hertzian distribution, but a localized pressure spike can appear near the outlet. In elliptical contacts, transverse flow also produces thin side-lobe regions. The smallest film thickness may therefore lie away from the centreline and need not coincide with the point of maximum pressure.

Coupled equations and numerical solution

The principal flow model is the Reynolds equation, obtained from the equations of viscous fluid motion by exploiting the small ratio of film thickness to contact length. Under an isothermal, Newtonian, no-slip approximation, with surface motion directed along xx, a compressible form is

∂∂x(ρh312η∂p∂x)+∂∂y(ρh312η∂p∂y)=∂(ρhue)∂x+∂(ρh)∂t.\frac{\partial}{\partial x} \left( \frac{\rho h^3}{12\eta}\frac{\partial p}{\partial x} \right) + \frac{\partial}{\partial y} \left( \frac{\rho h^3}{12\eta}\frac{\partial p}{\partial y} \right) = \frac{\partial(\rho h u_e)}{\partial x} + \frac{\partial(\rho h)}{\partial t}.

Here pp is pressure, hh is film thickness, ρ\rho is density, and η\eta is dynamic viscosity. The left side represents pressure-driven flow, while the right side accounts for entrainment and time-dependent storage. More general formulations include motion in both in-plane directions and variations of viscosity across the film.

The gap is not prescribed independently. It is calculated from a relation of the form

h(x,y,t)=h0(t)+g(x,y)+δ[p](x,y,t),h(x,y,t)=h_0(t)+g(x,y)+\delta[p](x,y,t),

where gg is the undeformed geometric gap and δ[p]\delta[p] is the elastic displacement generated by the distributed pressure. For elastic half-spaces, displacement is obtained through an integral whose kernel depends on the distance between the pressure source and the displacement point. Finite layers, coated surfaces, and complex component shapes require different elastic models.

A further constraint enforces load balance:

∬Ap(x,y) dA=F\iint_A p(x,y)\,\mathrm{d}A=F

for a fully fluid-supported contact. In mixed lubrication, the total load includes both fluid pressure and solid-contact forces. The unknown offset h0h_0 is adjusted until the required load is supported. Pressure-dependent density and viscosity close the isothermal system; thermal calculations additionally require energy equations and heat transfer into the solids.

Boundary conditions are essential. The inlet may be prescribed as flooded, with lubricant available sufficiently far upstream, or its supply may be limited. At the outlet, the liquid can cavitate rather than sustain the large tensile pressures that an unconstrained calculation might predict. Cavitation models based on mass conservation track liquid content as well as pressure, improving treatment of film rupture and subsequent reformation.

The resulting equations are strongly nonlinear: a small film-thickness change greatly alters hydraulic conductance because of the h3h^3 term, and pressure can change viscosity by orders of magnitude. Numerical methods include finite differences, finite elements, Newton-type iteration, multigrid acceleration, and fast evaluation of elastic deformation integrals. Reliable solutions require grid refinement and conservation checks, particularly near the outlet constriction, sharp surface features, or rapidly moving disturbances. A converged numerical residual alone does not establish that the chosen rheology, inlet condition, or elastic approximation represents the physical contact.

Film-thickness prediction and engineering parameters

Engineering calculations often use dimensionless film-thickness correlations instead of solving the full coupled problem. These correlations summarize numerical solutions for particular contact geometries and assumptions. Their main inputs are entrainment speed, inlet viscosity, pressure–viscosity response, elastic modulus, effective radius, load, and contact ellipticity. Their convenience does not remove their dependence on the conditions under which they were fitted.

For an elliptical contact, a widely used Hamrock–Dowson expression for minimum film thickness is

hmin⁡Rx=3.63 U0.68G0.49W−0.073(1−exp⁡[−0.68k]),\frac{h_{\min}}{R_x} = 3.63\,U^{0.68}G^{0.49}W^{-0.073} \left(1-\exp[-0.68k]\right),

with

U=η0ueE′Rx,G=αE′,W=FE′Rx2.U=\frac{\eta_0u_e}{E'R_x}, \qquad G=\alpha E', \qquad W=\frac{F}{E'R_x^2}.

In this convention, E′=2E∗E'=2E^*, RxR_x is the effective radius in the entrainment direction, η0\eta_0 is inlet viscosity at the reference pressure, and α\alpha is a pressure–viscosity parameter. The ellipticity parameter kk describes the contact’s transverse-to-longitudinal extent within the correlation’s geometric convention. Mixing modulus or ellipticity definitions from different sources can produce substantial errors.

The exponents express important physical tendencies. Increasing entrainment speed or inlet viscosity substantially increases film thickness. Load has a much weaker direct exponent because increasing load also enlarges the elastically flattened contact. This does not imply that load is unimportant: it strongly affects pressure, subsurface stress, fatigue, heat generation, and the validity of the underlying assumptions. The formula applies to smooth, fully flooded, steady, isothermal conditions within its intended range, not universally to every lubricated concentrated contact.

Central film thickness and minimum film thickness serve different purposes. Central thickness characterizes much of the flattened region and is often convenient for optical measurement. Minimum thickness is more directly relevant to the risk of local surface interaction. The ratio between them is not fixed; it varies with operating conditions and contact shape. Substituting one for the other can therefore distort an assessment of surface separation.

A common screening measure is the film parameter

λ=hmin⁡Rq12+Rq22,\lambda= \frac{h_{\min}} {\sqrt{R_{q1}^2+R_{q2}^2}},

where Rq1R_{q1} and Rq2R_{q2} are the root-mean-square roughnesses of the surfaces. Values above roughly 3 are often associated with full-film separation, values around 1–3 with mixed lubrication, and lower values with substantial asperity interaction. These are approximate engineering categories, not universal transition boundaries. Roughness wavelength, directionality, deformation, running-in, and surface chemistry all affect whether a given nominal λ\lambda corresponds to actual separation.

Lubricant rheology, traction, and thermal effects

The defining material property of many hard EHL contacts is the lubricant’s piezoviscous response: viscosity increases with pressure. A simple representation is the Barus relation,

η(p)=η0exp⁡(αp).\eta(p)=\eta_0\exp(\alpha p).

It is useful for explaining the mechanism and constructing approximate models, but extrapolating a single constant α\alpha across the full pressure range can be inaccurate. Real lubricants have pressure–viscosity behaviour that depends on temperature, molecular composition, and the pressure interval being considered. More elaborate empirical relations and measured high-pressure data are used when detailed prediction is required.

The inlet and the central contact play different roles. Inlet flow largely determines how much lubricant enters and thus strongly influences film thickness. The high-pressure region determines much of the shear resistance and transmitted tangential force. Consequently, two oils with similar ambient viscosities can produce different film thicknesses and traction characteristics. Ambient viscosity alone is not a complete descriptor of EHL performance.

At high pressure and shear rate, many lubricants do not behave as Newtonian liquids. Shear thinning reduces effective viscosity as deformation rate increases; some models also represent stress saturation through a pressure-dependent limiting shear stress. Viscoelastic effects become relevant when molecular relaxation times are comparable to the time spent passing through the contact. These responses should not be treated as interchangeable: a shear-thinning law, a viscoelastic model, and a limiting-stress model represent different aspects of material behaviour.

For predominantly sliding shear, a useful scale is

γ˙∼∣u1−u2∣h.\dot{\gamma}\sim \frac{|u_1-u_2|}{h}.

A sliding speed of 1 m s−11\ \mathrm{m\,s^{-1}} across a 0.1 μm0.1\ \mu\mathrm{m} film corresponds to a nominal shear rate of about 107 s−110^7\ \mathrm{s^{-1}}. Actual shear may vary through the film because viscosity and temperature are not uniform. Integrating the local shear stress over the contact gives the tangential force; its ratio to normal load is commonly called the traction coefficient in EHL studies.

Viscous dissipation generates heat, while compression, conduction, and convection also enter the thermal balance. Thermal EHL therefore couples the flow and deformation equations to lubricant and solid energy equations. Inlet heating can reduce the viscosity available for film formation, and heating within the contact can alter traction. The two solids need not receive equal fractions of the generated heat, especially when their speeds or thermal properties differ.

Traction curves commonly plot tangential force against slide-to-roll ratio,

SRR=u1−u2ue,\mathrm{SRR}=\frac{u_1-u_2}{u_e},

with sign and percentage conventions stated explicitly. Their initial slope and subsequent nonlinear behaviour reflect lubricant rheology and heating together. Extracting a material law from a measured traction curve therefore requires a model of the contact rather than a direct conversion of force into a single viscosity.

Rough surfaces, starvation, and transient operation

Classical smooth-surface EHL assumes continuous separation and an ample lubricant supply. Real surfaces contain asperities, machining marks, dents, and waviness. When these features are comparable in height to the film thickness, they perturb both pressure and deformation. Micro-elastohydrodynamic lubrication describes these local interactions. An asperity entering the contact can generate a highly concentrated pressure disturbance, elastically flatten, and modify lubricant flow around itself.

In mixed lubrication, fluid pressure and solid asperity contacts carry the load together. A model must then distinguish the fluid-filled gap from actual solid contact and account for their mechanical interaction. Boundary lubrication concerns the protection and friction provided by molecularly thin surface layers when bulk-fluid separation is insufficient. The regimes can coexist within one nominal contact, especially during running-in or under nonuniform lubrication.

Surface roughness cannot always be reduced to a single amplitude. Features aligned with the direction of motion influence side leakage and local pressure differently from transverse features. Wavelength matters because deformation and flow respond differently to fine asperities and broad waviness. Running-in can remove or plastically deform high points, but the evolving surface may also acquire dents or fatigue damage. A low average roughness therefore does not by itself guarantee a favourable local gap.

Starvation occurs when the available lubricant cannot sustain the fully flooded inlet condition. The inlet meniscus moves towards the contact, and less fluid is entrained than a fully flooded calculation assumes. In grease-lubricated bearings, lubricant availability depends on base-oil release, displacement of grease, replenishment between rolling elements, and migration along the raceway. Total lubricant quantity and local inlet supply are distinct variables; a component can contain substantial lubricant while an individual contact remains starved.

Time dependence introduces squeeze-film action through the storage term in the Reynolds equation. Approaching surfaces expel fluid and generate additional pressure, whereas separating surfaces can draw fluid into the gap or cause cavitation if replenishment is inadequate. During rapid speed or load changes, the film retains a history of earlier conditions rather than instantaneously adopting a steady-state thickness.

These effects are important in oscillating bearings, cam mechanisms, and contacts passing over dents or textured surfaces. At motion reversal, entrainment may momentarily vanish, but residual and squeeze films can still contribute to separation. Conversely, a steady-state calculation based on the average speed can miss the most severe part of a cycle. Detailed transient models follow the moving pressure field, surface deformation, liquid supply, and any asperity contacts through time.

Experimental observation and model validation

Optical interferometry is a principal method for measuring EHL film thickness. A common apparatus rolls a steel ball against a transparent glass or sapphire disc. A partially reflective coating at the contact surface allows reflected light from different interfaces to interfere. The resulting optical signal depends on optical path length, enabling the spatial film profile to be reconstructed. Colour interferometry shows the broad film shape, while spectrometric and spacer-layer techniques extend measurement to thinner films.

Converting optical data into physical thickness requires knowledge of refractive index, coating properties, and the optical arrangement. The lubricant’s refractive index changes under pressure, and temperature can also affect the conversion. Carefully calibrated methods can resolve nanometre-scale thickness changes, but the reported result remains dependent on these corrections. The transparent counterface also has elastic and thermal properties different from those of many engineering steels.

Electrical methods provide complementary information. Contact resistance can detect intermittent conductive paths associated with loss of separation, whereas capacitance can be related to film geometry and dielectric properties. Neither signal is a simple universal measure of minimum thickness. Conductive additives, insulating surface films, contamination, and the distribution of the gap can influence the reading. Electrical measurements are particularly valuable when combined with optical observations or controlled surface roughness.

Twin-disc and ball-on-disc machines measure traction under specified load, entrainment speed, sliding, and temperature. Small pressure transducers and thin-film sensors can investigate local conditions, although their dimensions and mechanical properties may disturb the pressure field they are intended to measure. Spectroscopic methods can provide information about temperature, molecular orientation, or chemical changes, depending on the lubricant and optical access.

Validation is strongest when a model reproduces several independent observables. Agreement with central film thickness alone does not establish that the calculated minimum thickness, pressure spike, or traction is correct. For example, an isothermal calculation may match film thickness under mild sliding while failing to predict the traction curve at larger sliding speeds. Measurements of inlet supply and surface topography help distinguish rheological errors from incorrect boundary conditions.

Uncertainty arises from load and speed calibration, temperature control, surface geometry, viscosity measurements, optical conversion, and data processing. Because film thickness is highly sensitive to inlet viscosity, even a modest temperature difference between the nominal reservoir condition and the actual inlet can matter. Experimental reports therefore need sufficiently detailed operating conditions to support meaningful comparisons between laboratories and numerical models.

Machine applications, friction, and surface damage

Rolling bearings are a central application because their internal contacts combine high local pressure with repeated rolling and some sliding. EHL determines separation between rolling elements and raceways, but bearing operation also depends on cage motion, lubricant distribution, internal clearance, and component temperature. In angular-contact bearings, spin and differential sliding make the shear field more complex than that of a purely rolling ball on a flat surface.

Gear teeth experience continuously changing curvature, load, and surface velocity through the meshing cycle. In an ideal involute gear pair, the sliding direction reverses at the pitch point, while entrainment generally remains nonzero. Local EHL calculations are therefore combined with gear geometry and load sharing. Tooth modification, misalignment, dynamic loading, and roughness alter the conditions at which the film must support the load.

Cam–follower contacts add strongly varying speed and force, sometimes with entrainment reversal or very low speed near heavily loaded portions of the cycle. In engines, thermal conditions and lubricant composition further complicate the film. EHL calculations help distinguish a failure caused primarily by insufficient fluid supply from one associated with excessive temperature, surface geometry, or local load concentration.

Traction drives exploit rather than merely tolerate lubricant shear. In these devices, pressurized lubricant transmits tangential force between rolling elements, allowing torque transfer without gear teeth. Their useful traction depends on high-pressure rheology, while efficiency and temperature depend on the accompanying dissipation. The same lubricant characteristic that supports torque transmission may be undesirable in a bearing or gear application where lower friction is the objective.

EHL is closely related to rolling-contact fatigue, wear, micropitting, and scuffing, but it does not by itself determine component life. Even with full separation, repeated subsurface stresses can initiate fatigue around material imperfections. With inadequate separation, asperity stresses and near-surface damage become more important. Micropitting consists of small fatigue-related surface pits and is influenced by roughness, sliding, local stresses, and lubricant chemistry. Scuffing involves severe adhesive damage under adverse thermal and lubrication conditions.

Reducing lubricant viscosity can lower churning and some viscous losses, but it also reduces film thickness. This trade-off is relevant to electric-vehicle transmissions and high-speed industrial machinery, where efficiency targets coexist with demanding contact conditions. System-level analysis also considers pumping losses, heat rejection, lubricant ageing, and material compatibility. A film-thickness calculation is consequently one component of a broader mechanical and thermal assessment, not a stand-alone predictor of durability.

Soft contacts, biological systems, and limits of the continuum model

Soft EHL occurs when compliant surfaces deform substantially under hydrodynamic pressure even though the lubricant’s viscosity changes little with pressure. Elastomers, hydrogels, and biological tissues can have moduli many orders of magnitude below those of bearing steels. Water-based liquids can therefore generate meaningful fluid–structure coupling at relatively modest pressures. The governing mechanism remains coupled flow and deformation, but the balance of effects differs from conventional hard EHL.

Layer thickness and support conditions become especially important. A thin compliant coating bonded to a rigid substrate does not deform like an elastic half-space. Its response depends on the ratio of layer thickness to the pressure footprint, its compressibility, and the constraint imposed by the substrate. In nearly incompressible layers, lateral restraint can greatly increase apparent stiffness. Time-dependent polymer behaviour adds relaxation and energy loss, making the deformation dependent on speed and loading history.

A characteristic soft-lubrication effect is the generation of lift through deformation-induced asymmetry. A rigid object moving beside a compliant wall can create a pressure field that deforms the wall, changing the gap so that positive and negative pressure contributions no longer cancel as they would in a corresponding symmetric rigid configuration. The resulting force depends on compliance, speed, geometry, and cavitation conditions. Such effects are relevant to particle motion near soft boundaries and to some microfluidic systems.

Biological joints require a broader framework than ordinary elastic lubrication. Articular cartilage is a porous solid containing interstitial fluid, and its load support depends on fluid pressurization within the tissue as well as flow in the interfacial gap. Poroelasticity describes the coupled deformation and fluid transport of such materials. Surface-bound molecules, synovial-fluid constituents, and boundary layers also contribute to low friction. Joint lubrication is therefore not adequately represented by treating cartilage as a smooth rubber bearing lubricated by a single Newtonian liquid.

At very small separations, the continuum assumptions used in the Reynolds equation become less secure. Molecular layering, wall slip, adsorption, and surface forces can alter flow and normal stress. A lubricant confined to a few molecular diameters may not possess the same effective properties as the bulk liquid. The relevant crossover depends on molecular size, surface chemistry, pressure, and confinement; there is no single thickness at which every continuum model fails.

Research at these limits combines continuum EHL, molecular simulation, surface-force measurements, and chemically resolved experiments. Other active developments include deformable textured surfaces, multiphase lubrication, non-Newtonian aqueous fluids, and multiscale descriptions linking individual asperities to component-scale contacts. Each requires identifying which parts of the classical description—thin-film flow, elastic response, bulk rheology, or boundary conditions—remain appropriate at the scale being studied.