Induction-Hardened Raceways in Slewing Ring Bearings

2. The slewing ring as a structural bearing

2.1 The combined load system

An ordinary rolling bearing is sized against a single dominant load direction and a revolution count. A slewing ring is sized against a load combination. On an excavator upper structure the bearing carries the weight of the house, boom, arm and load (F_a), the digging and slew reaction forces (F_r), and above all the overturning moment M produced by the offset of those forces from the slew axis. M is normally the governing term: a modest machine can develop an overturning moment whose equivalent rolling-element load exceeds anything the direct axial load produces.

Because three quantities act at once, capacity is not published as a single number. Manufacturers publish static load curves: a boundary in the F_a–M plane, plotted for a stated F_r and a stated static safety factor, inside which the combination must lie. The underlying limit is the one expressed generically in ISO 76¹ — a permanent deformation at the most heavily loaded contact of about 1/10 000 of the rolling element diameter, corresponding to a maximum contact pressure of roughly 4 200 MPa for ball/steel contact and about 4 000 MPa for roller contact. A second, separate curve is usually given for the bolts, and a third for the gear teeth, because the three limits are independent and any of them can govern.

It is worth being explicit about a gap: there is no single, broadly adopted international standard that rates slewing rings end to end in the way ISO 281² and ISO 76¹ rate conventional bearings. Rating practice is a blend of ISO 76¹/ISO 281² principles, manufacturer-specific static curves derived from their own raceway geometry and hardening specification, and application-sector structural codes. Load curves from two suppliers are therefore not directly comparable unless the static safety factor, the assumed mounting stiffness and the hardening specification behind them are all known.

2.2 Raceway arrangements

Single-row four-point contact ball. The commonest arrangement. Both rings carry a gothic-arch (ogival) groove, so each ball can contact at four points, typically at a nominal contact angle α of 45°. Under any given load direction only one diagonal pair of contacts is active, so the same row reacts axial load in both directions plus moment plus radial load. Groove conformity (groove radius divided by ball diameter) is typically in the 0.52–0.53 region; tighter conformity lowers contact pressure but raises spin friction. The known penalty of four-point geometry is kinematic: the contact geometry cannot roll purely at all contact points, so spin sliding and its associated heat and wear are inherent, not a defect.

Multi-row ball. Two rows of balls at opposing contact angles, usually across three rings, frequently with the upper row of larger diameter because it takes the predominant downward axial-plus-moment load. This separates load directions, raises capacity for a given envelope and increases tilting stiffness.

Crossed cylindrical roller. Rollers alternate at 90° to one another on a common raceway. Line contact gives markedly higher static capacity and tilting stiffness than a ball row of comparable size, and rolling is essentially pure. The trade is sensitivity: line contact concentrates stress at the roller ends unless the rollers are logarithmically or otherwise crowned, and it is far less tolerant of raceway waviness, ring distortion and out-of-flat mounting than a ball row, which can accommodate small geometric errors by redistributing among many point contacts.

Three-row roller. Separate axial rows for up-load and down-load plus a dedicated radial row, decoupling all three load directions. Used at the top of the size range on large cranes and mining machines.

2.3 Contact mechanics and why depth matters

For a Hertzian point contact the maximum shear stress does not occur at the surface. The orthogonal shear stress peaks at roughly 0.5 b below the surface at about 0.25 p₀, and the von Mises equivalent stress peaks at a similar depth at roughly 0.6 p₀. Here b is the semi-minor axis of the contact ellipse, its half-width in the rolling direction, and it is b rather than a that sets the depth: in a groove of 0.52–0.53 conformity the semi-major axis a runs across the track and is six to eight times longer. For a large, heavily loaded slewing ring ball, b is commonly of the order of 1.5–3 mm, placing the peak subsurface stress roughly 0.7–1.5 mm below the raceway. A hardened case must therefore do two things: resist surface-initiated damage, and — more importantly — extend well beyond the depth of peak shear so that the case/core transition, where hardness falls steeply, is not itself loaded near its yield strength. If it is, the failure mode is case crushing: subsurface plastic collapse at the transition, seen at the raceway as local subsidence followed by large-flake spalling. This is a different mode from classic subsurface-initiated rolling contact fatigue, and ISO 15243³ is the appropriate vocabulary for distinguishing it and the other modes on a failed raceway.

3. Raceway induction hardening

3.1 Why induction rather than carburizing

Carburizing is metallurgically the stronger process: carbon is diffused into a low-carbon alloy steel at 900–950 °C, producing a case of roughly 0.7–0.9 % C at 58–62 HRC over a tough core typically in the 30–40 HRC band, with high compressive residual stress. It is the default for high-duty gearing, and the ISO 6336-5¹³ material tables reflect that superiority in the allowable stresses granted to carburized gearing. Carburizing is also used for some rolling-bearing rings, although most conventional bearing rings are through-hardened.

It is nonetheless the wrong process for a 1–3 m diameter ring, for reasons that are practical rather than metallurgical. The whole part must sit in a furnace for many hours and then be quenched in bulk, which for a large thin-section annulus produces ovality and axial coning that cannot be economically ground out; the entire ring including bolt-hole regions is hardened, complicating machining; and the furnace, fixture and cycle cost scale badly with diameter.

Induction hardening inverts these constraints. The ring is made from a medium-carbon alloy steel that already contains the carbon needed — 42CrMo4 / AISI 4140-class material to ISO 683-2⁹ and SAE J404¹⁰ is the industry workhorse — supplied in the quenched-and-tempered condition at roughly 250–320 HB. Only the raceway track is austenitized, by an inductor that scans around the ring while the ring rotates, followed immediately by a spray quench. Heat input is localized and brief, so bulk distortion is small; the unhardened body remains tough and machinable; and the process scales to any diameter simply by running the scan for longer. The compressive residual stress produced in the hardened layer, arising from the volume expansion of the martensite transformation against a constraining cold substrate, is a genuine structural benefit and is one reason the raceway tolerates the tensile hoop stresses imposed by bolting and by ring bending.

The compromises are real and should be stated. Attainable hardness is limited by the base carbon content: about 0.42 % C gives roughly 57–62 HRC at best, against 58–62 HRC from a properly carburized 0.8 % C case with a considerably more favorable retained austenite and carbide structure. The core beneath the case is the quenched-and-tempered base steel, generally softer and less tough than a carburized core, which changes the case/core stress balance. And the process creates a soft zone (Section 4) that carburizing does not.

3.2 Process physics

Induced current concentrates within the electromagnetic reference depth δ = 503·√(ρ/μ_rf) meters, with resistivity ρ in Ω·m and f in Hz. Steel above the Curie temperature is paramagnetic, so μ_r → 1 and ρ ≈ 1.2 × 10⁻⁶ Ω·m; at 10 kHz this gives δ ≈ 5.5 mm, at 3 kHz roughly 10 mm and at 25 kHz roughly 3.5 mm. Frequency is therefore the primary lever on hardening depth, with power density and scan speed setting how much of the heated depth reaches austenitizing temperature before the quench arrives. Two consequences follow. First, depth cannot be increased without limit at a fixed frequency: pushing more energy in mostly overheats the surface. Second, the transition from hardened case to unaffected core is gradual, not the step change a schematic implies.

3.3 Specifying and verifying depth and hardness

Surface hardening depth (SHD) is defined against a limiting hardness that is conventionally 80 % of the minimum specified surface hardness — for a 58 HRC minimum (≈ 653 HV) that is a limit of roughly 520 HV. ISO 18203⁴ sets out the measurement of surface-hardened layer thickness; SAE J423⁵ covers case depth measurement methods generally. The hardness traverse itself is made by microindentation to ASTM E384⁷ or ISO 6507-1⁸ on a sectioned and polished sample; bulk surface hardness is verified by Rockwell C to ASTM E18⁶. A specification that gives a hardness without naming the method and the limiting hardness for depth is incomplete, because the depth number changes materially with the limit chosen.

Typical published practice for slewing ring raceways is a surface hardness in the 55–62 HRC range with an effective depth of roughly 3–6 mm on medium and large rings, the depth chosen with reference to rolling element diameter and the resulting contact ellipse rather than to the ring size. The engineering rationale is Section 2.3: the depth to the limiting hardness must comfortably exceed the depth of peak subsurface shear stress, with margin for the tensile bending stress the case carries as it flexes over the softer core.

4. The soft zone

The raceway is a closed circle; the inductor is not. A scanning induction pass must begin somewhere and must return to that point, and where it returns, the trailing heat of the finish overlaps material that was hardened at the start. That overlap region is re-heated below the austenitizing temperature and is therefore tempered back, or is heated unevenly during the ramp-in and ramp-out of power and never fully transformed. Either way a short arc of the raceway ends up at reduced hardness — commonly falling toward core hardness, i.e. tens of HRC below the hardened track. The affected arc is typically of the order of 20–50 mm of circumference, though it depends on inductor geometry and scan control.

This is not a manufacturing defect and it is not eliminable by better process control; it is a topological consequence of hardening a closed ring with a moving heat source. It is therefore managed rather than removed. Two conventions follow. First, the soft zone is marked on the ring — most commonly a stamped "S", a punch mark, a plug, or a paint mark on the outside diameter or the top face. Second, in a single-row four-point contact ball bearing the balls are loaded through a radial filler hole closed by a plug, and that plug is itself a local interruption of one raceway. The filler plug and the soft zone are conventionally placed together, so a single marked position identifies both discontinuities.

Installation must then keep that marked position out of the principal load zone. On a machine with a defined working direction — an excavator boom, a crane jib at its normal working radius — the maximum-loaded rolling element sits on a predictable arc, and the soft zone is oriented away from it, commonly at 90° or more to that direction. Conventions differ between manufacturers and between machine types, and the correct orientation is the one given in that bearing's installation instruction, not a universal rule. What is universal is the failure consequence of ignoring it: the softer arc yields at a lower contact pressure, so the raceway brinells locally, the track deepens, tilt increases, load redistributes onto neighboring elements, and a progressive failure begins from a region that was never intended to be load-bearing at full duty.

5. The ring gear

Slew drive teeth are cut into the inner or outer ring of the same forging, and their hardening is a separate decision from the raceway's. Three treatments are in common use.

Quenched and tempered only. Teeth left at the base 250–320 HB. Simple, distortion-free, and adequate where the slew torque is modest and the duty cycle low. Rating is normally limited by tooth root bending and by flank wear rather than by pitting.

Flank-only induction hardening. The tooth flanks are hardened but the root fillet is not. This raises pitting resistance substantially, but it places the case/core boundary — and, critically, a region of residual tensile stress just beyond the hardened layer — close to the root fillet where bending stress is highest. Flank-only hardening can therefore improve surface durability while doing nothing for, or in unfavorable geometries slightly degrading, bending fatigue strength. This is a real and sometimes overlooked trade.

Contour hardening. The hardened layer follows the tooth profile around the root, putting compressive residual stress where the bending stress peaks. It gives the best combination of pitting and bending capacity available from induction, at the cost of a considerably more demanding inductor and process.

ISO 6336-2¹¹ and ISO 6336-3¹², with the material tables of ISO 6336-5¹³ (and equivalently ANSI/AGMA 2001-D04¹⁴ in the AGMA framework), quantify the hierarchy. The standards' MQ material grades place allowable contact stress for flame- or induction-hardened flanks materially below that for carburized gearing — on the order of 1 200 MPa against roughly 1 500 MPa — with through-hardened quenched-and-tempered steel lower again, and allowable root bending stress follows a similar ordering. The exact values depend on grade, hardness and the standard edition, so they should be read from the standard rather than quoted from memory.

Tooth-by-tooth induction hardening carries its own soft-zone problem: the process starts and stops, and where it does, one or two teeth are left unhardened or partly tempered. These are usually marked in the same way as the raceway soft zone, and the same principle applies — they belong outside the arc over which the pinion works hardest.

6. The bolted joint

Field experience across the sector consistently identifies bolting, not raceway fatigue, as a leading cause of slewing ring failures. The reason is structural: the overturning moment is reacted by the bolt circle as a couple, so on the tension side the bolts carry a large, fully reversing external load every time the machine slews or the boom swings through its arc.

6.1 Preload, load factor and separation

In a correctly preloaded flange the bolt does not see the external load. The joint members, being far stiffer than the bolt, absorb most of it by partial unloading of the clamped interface; the bolt sees only the fraction Φ, which for a stiff steel flange with the load introduced near the interface is commonly in the region of 0.1–0.3. This protection is conditional. The instant the preload is exhausted and the joint gaps, Φ effectively goes to 1 and the bolt sees the entire alternating load. There is no gradual degradation between the two states; the transition is abrupt, and it is the mechanism behind nearly all slew bolt fatigue failures.

6.2 What the bolt can actually endure

The arithmetic is stark. A grade 10.9 bolt to ISO 898-1¹⁵ has a minimum tensile strength of 1 040 MPa and a minimum 0.2 % proof stress of 940 MPa; it is typically preloaded to around 70 % of proof. Yet the endurable alternating stress amplitude of a rolled thread is very much smaller: VDI 2230 Part 1¹⁷ gives, for threads rolled before heat treatment, an endurance amplitude of σ_A = 0.85 (150/d + 45) MPa, which for an M24 bolt is roughly 44 MPa — under 5 % of proof stress. Threads rolled after heat treatment do better, but the order of magnitude stands. The practical conclusions are that (a) preload, not bolt strength, is what keeps the joint alive, and (b) the failure location is predictable: the first engaged thread or the thread runout, where the stress concentration is highest.

6.3 Preload scatter, embedment and progressive failure

Achieved preload is not the nominal preload. VDI 2230¹⁷ accounts for this with the tightening factor α_A, the ratio of maximum to minimum expected preload for a given method: broadly, torque wrenching sits at the high-scatter end and angle-, yield- or tension-controlled methods at the low-scatter end, because the dominant uncertainty in torque control is the friction coefficient — roughly 85–90 % of applied torque is consumed by thread and underhead friction, so a friction coefficient varying between 0.10 and 0.16 alone moves the achieved preload by tens of percent. Testing to ISO 16047¹⁸ is the way to pin the torque/clamp force relationship down for a given surface condition. Embedment of surface asperities then removes a further few percent of preload in the first hours of service, which is why a documented re-torque check after an initial period is standard practice.

Loss of preload in one bolt is not a local event. The moment must still be reacted, so the load that bolt no longer carries is redistributed to its neighbors, which are then closer to separation themselves. The result is a characteristic progressive or "zipper" failure that propagates around an arc of the bolt circle. A single fractured bolt found during inspection is therefore a system finding, not a component finding.

6.4 Mounting flatness and structural stiffness

Both the bearing's static capacity and its bolt loads assume the mating structures are flat and stiff enough to distribute load around the whole circumference. Slewing ring makers accordingly specify a flatness tolerance for the mounting faces, expressed as a permissible waviness over the circumference with a limited number of waves, and for rings in the 1–2 m range the permitted total deviation is typically a fraction of a millimeter. Bolting a ring down onto a face outside that tolerance forces the ring to conform, distorting the raceway, concentrating contact load on a few elements, and imposing bending on the bolts. It is an installation error that presents later as a bearing failure.

7. Tilt (rocking) measurement as a condition indicator

The accepted in-service condition check is a measurement of the change in axial clearance under a reversing moment — variously called tilt, rocking, or tipping clearance measurement. A dial indicator or displacement transducer is mounted to span the gap between the two rings at a defined radius and a marked circumferential position. The machine is then put through a defined maneuver that reverses the moment on the bearing (for example boom fully extended and loaded, then relieved), and the relative axial displacement is recorded.

What makes the measurement useful is that raceway wear, plastic track deepening and rolling-element wear all increase internal clearance, and clearance appears directly in the tilt reading. What limits it is equally important and is frequently under-appreciated:

• Only the change matters. A single absolute reading carries almost no information. The measurement must be baselined at commissioning, in the same machine configuration, at the same marked positions, with the same fixture. Wear limits published by manufacturers are diameter- and type-dependent curves of increase over the as-new value.
• The reading is not purely the bearing. It contains elastic deflection of the mating structures and bolt stretch. Repeating the exact machine configuration is what makes readings comparable.
• It is an average, not a survey. Localized damage — spalling at one point, or plastic deformation in a mis-oriented soft zone — may barely move a tilt reading taken elsewhere. Measurements at several marked positions around the ring are more informative than one, and a rising spread between positions is itself a signal.

Tilt measurement should therefore be read alongside other indicators: grease condition and ferrous debris content on relubrication, slew drive backlash and pinion tooth condition, bolt torque audit results, and any change in slew noise or torque. For diagnosis of what is actually found on a stripped raceway, the terminology and causal framework of ISO 15243³ should be used rather than ad hoc description. One damage mode deserves specific mention because slewing rings are unusually prone to it: these bearings oscillate over limited arcs and rarely complete continuous revolutions, so the same rolling elements sit on the same raceway positions for long periods under load. That is the classic precondition for false brinelling and fretting corrosion at the standstill positions, and it is why relubrication procedures for slewing rings generally require the bearing to be rotated while grease is applied, and why periodic full rotation is recommended even on machines whose duty does not otherwise demand it.

8. Limits of current practice

Three honest caveats close this review. First, as noted in Section 2.1, slewing ring rating is not standardized the way conventional bearing rating is; comparisons between suppliers' load curves require knowing the assumptions behind them. Second, the relationship between induction case depth and permissible contact pressure is established qualitatively — deeper case, more margin against case crushing — but the quantitative limit for a given geometry depends on the case/core hardness gradient, the residual stress field and the core strength, and published design rules for it vary between sources. Third, tilt measurement limits are empirical and manufacturer-specific; there is no general physical criterion that converts a measured clearance increase into remaining life. Where these uncertainties bear on a decision, the correct response is to obtain the specific bearing's documentation rather than to apply a generic rule.