Preload and Fatigue in Slewing Ring and Final Drive Bolted Joints

2. The joint, not the bolt

2.1 What the fastener actually does

The intuitive reading of a bolted slewing ring — that each bolt carries a share of the load in tension — is wrong in a way that matters. The bolts' function is to compress the mating flanges together hard enough that the interface never opens. Once that condition holds, external load is carried by the clamped interface, and the bolt sees only a small increment of additional tension. The bolt is a clamp, and the joint is the structure.

Everything downstream follows from this. It is why preload is specified rather than merely "tight"; why a joint that has lost preload fails by fastener fatigue rather than by overload; why bolt fatigue strength is often not the governing calculation while clamp force nearly always is; and why a slewing ring joint that has begun to open can destroy its bolts in a small number of operating hours after years of untroubled service.

2.2 Load path in a slewing ring mounting

A slewing ring on an excavator carries an axial load from the machine's weight, a radial load from digging reaction, and — dominantly — a tipping moment from boom reach and load. The tipping moment is reacted across the bolt circle as a distribution of interface pressure: compression increases on the side toward which the machine is tipping and decreases on the opposite side. Bolts on the unloading side are the ones at risk, because it is there that interface pressure approaches zero and the clamped joint approaches separation.

The distribution is not uniform and it rotates continuously as the upper structure slews. Every bolt in the circle therefore passes through the maximum-tension condition once per revolution of the house relative to the load direction, which makes this a fatigue problem by construction. A machine that trenches in one orientation all day loads a few bolts many times; a machine that slews continuously loads all of them.

3. Preload and the joint diagram

3.1 Bolt and member stiffness

When a bolt is tightened it stretches and the clamped members compress. Both behave as springs: the bolt with stiffness k_b, the clamped members with stiffness k_c. Because the members are a large volume of metal in compression and the bolt is a slender rod in tension, k_c is typically three to six times k_b in a steel joint of ordinary proportions.

When an external tensile load F_ext is subsequently applied across the joint, it is shared between stretching the bolt further and relieving the compression in the members, in proportion to their stiffnesses. The share taken by the bolt is the load factor of VDI 2230 Part 1¹

so the bolt tension rises by only Φ·F_ext while the interface clamp force falls by (1 − Φ)·F_ext. For the stiffness ratios above, Φ falls in the range 0.15 to 0.25. A bolt in a properly preloaded joint therefore experiences roughly a fifth of the external load range it appears to be carrying — and this, rather than any property of the fastener material, is the principal reason preloaded joints survive fatigue loading at all.

3.2 Separation

The protection holds only while the interface remains in compression. Once the clamp-force reduction (1 − Φ)·F_ext reaches the preload at that location — that is, once F_ext passes the preload divided by (1 − Φ) — the interface opens and the load path changes discontinuously: the bolt is now the only element bridging the gap and Φ jumps to 1.0. From that point the bolt takes all of any further increase in load rather than Φ of it — four to seven times the closed-joint rate for the stiffness ratios above — so its alternating stress climbs steeply with every increment of load beyond separation, into a regime where fatigue life is short.

This is the single most important behavior of the joint, and it explains the characteristic field observation that slewing ring bolts fail in groups and suddenly. The joint operates for years with the bolts nearly static; preload is lost slowly by the mechanisms of section 5; at some point the least-preloaded bolt's local interface begins to open at peak load; that bolt now carries the whole alternating load and fails; the load it was carrying redistributes to its neighbors, which open sooner; and the failure propagates around the arc.

3.3 Choosing preload

Preload is therefore specified high — commonly 70 to 90 % of the fastener's proof load, the proof load itself being defined by property class in ISO 898-1³ — for two independent reasons. It maximizes the margin before separation, and it maximizes the friction available at the interface to resist transverse slip. The upper limit is set by the need to leave headroom for the tension increment Φ·F_ext, for the scatter of the tightening method, and for any thermal differential, without reaching yield.

The corollary is that a slewing ring joint is not conservatively tightened by using a lower torque. Under-tightening does not make the joint gentler; it moves it closer to separation and therefore closer to the fatigue regime it is designed to avoid.

4. Achieving preload

4.1 Torque control and where the torque goes

The overwhelmingly common method is torque control, in which preload is inferred from applied torque through the relation

where d is the nominal thread diameter and K is the nut factor, an empirical lumped coefficient determined experimentally under ISO 16047⁵. The relation's weakness is that K is not a property of the thread geometry but of the friction conditions at two sliding interfaces.

The energy accounting is instructive. Of the torque applied to a typical fastener, only on the order of 10 to 15 % does useful work stretching the bolt. Roughly 50 % is consumed overcoming friction under the turning head or nut face, and roughly 35 % overcoming friction in the threads. It follows that preload is governed by friction coefficients that are not measured, and that a modest change in either one produces a large change in the delivered clamp force.

4.2 The lubrication trap

The single most consequential error in field practice on these joints follows directly from Table 2. Applying the dry torque figure to a lubricated fastener roughly halves K and therefore raises delivered preload by something approaching a factor of two, which for a target of 75 % of proof load takes the bolt past yield. Applying a lubricated figure to a dry fastener leaves the joint substantially under-preloaded and near separation. The torque specification and the friction condition are a matched pair, and neither is meaningful without the other.

This is why manufacturers' slewing ring bolting instructions state the lubricant or the coating alongside the torque, and why a bolt kit supplied with a specified coating is not interchangeable with a bare fastener of the same grade and dimension at the same torque.

4.3 Methods with lower scatter

Torque-plus-angle control tightens to a modest seating torque and then turns a specified further angle, which relates preload to bolt extension geometrically and largely removes the thread friction term. Yield or gradient control monitors the torque–angle slope and stops at the onset of plasticity, giving the tightest scatter available from a powered tool but consuming most of the fastener's reserve — such bolts are single-use by definition. Direct elongation measurement, by micrometer or ultrasonically, measures the quantity of interest rather than a proxy. Hydraulic tensioning stretches the bolt axially and runs the nut down against the flange, eliminating torsion in the bolt entirely; it is standard on very large bolt circles.

5. Loss of preload

5.1 Embedment

Real surfaces contact only at their asperities. Under the very high contact pressures at the bolt head, nut face and thread flanks, those asperities flatten in the first hours of service. The total settlement is small — typically a few micrometers per interface, with more at a thread interface than at a flat one — but preload is proportional to bolt extension, and bolt extension in a short-grip joint is itself only a few tens of micrometers. A settlement of ten micrometers in a joint whose bolt was stretched by fifty costs sixteen per cent of the preload at Φ = 0.20: the compressed members spring back by Φ of the settlement, so the bolt's extension falls by only (1 − Φ) of it, and the full twenty per cent applies only if the members were rigid.

Embedment loss is therefore acutely dependent on grip length. Short bolts through thin flanges lose a large fraction of their preload; the same absolute settlement in a long bolt is a small fraction. This is why long bolts, or the addition of a resilient element, are a legitimate design response to a joint that will not hold preload, and it is the technical basis for the practice of re-torquing after a running-in period.

5.2 Relaxation and creep

Distinct from embedment, relaxation is the time-dependent loss of stress at constant strain in the fastener and members. In steel joints at ambient and moderate temperatures it is small. It becomes significant where a soft element is included in the clamp path — a coating of appreciable thickness, a gasket, a painted interface — and this is the reason slewing ring mounting faces are specified as bare or thinly coated and why paint is excluded from the clamped area.

5.3 Transverse self-loosening

The mechanism identified by Junker is the dominant cause of rotational loosening and it is driven by transverse, not axial, load. When the clamped members slip relative to one another — even by a few micrometers — the friction that holds the thread helix against back-driving is momentarily relieved, and the bolt rotates loose by a small increment. Repeated cycles unwind it progressively.

The key point for design is that the remedy is not a locking device but the elimination of transverse slip: sufficient clamp force that the interface friction exceeds the transverse load throughout the duty cycle. Thread-locking compounds, wedge-locking washers and prevailing-torque nuts are secondary measures that limit the consequence of slip; they do not prevent the mechanism. A slewing ring joint that is loosening is telling you that its interface is slipping, and therefore that its preload is already inadequate.

5.4 Thermal differential

Where the fastener and the clamped members have different coefficients of thermal expansion, or reach different temperatures, preload changes with temperature. In an all-steel slewing ring joint this effect is minor. It becomes relevant in mixed-material mountings and in joints that see large thermal excursions, and it is one of the terms the preload headroom of section 3.3 must accommodate.

6. Fatigue of the fastener

6.1 Where cracks start

Bolt fatigue cracks initiate overwhelmingly at the root of the first thread engaged in the nut or tapped hole, the thread geometry and its tolerances being those of ISO 965-1¹⁰. The reason is load transfer: the first engaged thread carries a disproportionate share of the total thread load — on the order of a third in a standard nut — because the bolt is stretching while the nut is compressing, so the two thread helices progressively lose register along the engagement. That load concentration coincides with the geometric stress concentration of the thread root.

Secondary initiation sites are the head-to-shank fillet and the thread run-out, both of which are geometric transitions. Nut designs that equalize thread load — tapered or relieved first threads, tension nuts — move the initiation site and raise fatigue strength, and are found on the most heavily loaded joints.

6.2 Rolled threads

Threads formed by rolling rather than cutting have a fatigue strength substantially higher than cut threads, for two reasons that act together: the rolling process leaves compressive residual stress at the thread root, which opposes crack opening, and it produces a grain flow that follows the thread contour rather than being severed by it. The improvement is large enough that rolled threads are effectively universal on high-grade structural fasteners.

The order of operations matters. Threads rolled after heat treatment retain more of the beneficial residual stress than threads rolled before it, since the heat treatment would otherwise relieve much of what rolling produced. This is one of several respects in which a fastener of the correct grade marking is not necessarily equivalent to the fastener the joint was designed around.

6.3 Eccentric loading

The joint diagram of section 3 assumes the external load is applied concentrically with the bolt axis. In a slewing ring mounting it is not: load reaches the bolt through a flange that bends, so the bolt is loaded eccentrically and sees a bending component superimposed on tension. Bending raises the peak stress on one side of the thread root without raising the mean, which is precisely the combination fatigue is sensitive to.

The magnitude depends on flange stiffness, which returns the analysis to the structure. A thicker, stiffer mounting flange reduces the bending component; a flange that is locally compliant — over a cut-out, at the end of a stiffener, adjacent to a weld — increases it. Slewing ring bolt failures are commonly found clustered at such features rather than distributed uniformly around the circle.

7. Features specific to slewing ring mountings

7.1 Mounting surface flatness

A slewing ring is a comparatively slender ring, and it is not stiff enough to correct the surface it is bolted to. If the mounting structure is not flat, tightening the bolts pulls the ring into the shape of the structure, and that shape is transferred directly to the raceway. Two consequences follow: the raceway acquires a waviness that redistributes rolling element load, in the manner described in DYCO-TP-105, and the bolts in the high spots carry more preload than those in the low spots, so the joint begins life with a preload distribution it was not designed for.

This is why bearing manufacturers specify mounting surface flatness — as a total deviation over the circumference with a limit on the rate of change per unit arc — and why that specification is checked before installation rather than assumed. It is a measurement of the machine, not of the bearing, and it is the one most often omitted.

7.2 Bolt grade

Property class 10.9 in the sense of ISO 898-1³ is the common choice for slewing ring mountings, with 12.9 used where the required clamp force cannot be achieved in the available bolt size. Higher is not automatically better. As tensile strength rises, susceptibility to hydrogen embrittlement and to stress corrosion cracking rises with it, and the notch sensitivity of the material increases so that the fatigue benefit of the higher strength is not fully realized at the thread root. Class 12.9 fasteners in particular require care in coating selection and processing.

7.3 Hydrogen embrittlement

Electroplating processes introduce hydrogen into high-strength steel, where it diffuses to regions of triaxial tensile stress — thread roots in a preloaded bolt — and causes delayed brittle fracture, typically within hours to days of tightening. Electroplated coating systems are specified in ISO 4042⁷, which sets the post-plating bake; the established verification is the preloading test of ISO 15330⁶ on production fasteners. The characteristic signature is a bolt that fractures with no plastic deformation some time after assembly and without having been loaded in service, and it is distinguishable from fatigue by the absence of beach marks on the fracture face.

Hot-dip galvanized coatings, specified for fasteners in ISO 10684⁹ and generally in ISO 1461⁸, avoid the electrolytic hydrogen route and are common on large fasteners for this reason, at the cost of a different and thicker coating whose effect on the nut factor must be accounted for.

7.4 Re-use

Whether a slewing ring bolt may be re-used has a specific technical answer rather than a general one. A bolt tightened by yield or gradient control has by definition been taken into plasticity and is single-use. A bolt tightened elastically to a torque specification may in principle be re-used if it has not yielded, its thread and bearing surfaces are undamaged, its coating is intact and its length is unchanged — but the last of these is the difficult one to establish in the field, since a permanent extension of a fraction of a millimeter indicates yielding and is not detectable without a measurement against the original length.

The practical position taken in most manufacturers' instructions is that slewing ring bolts are replaced as a set at each removal. The reasoning is not conservatism for its own sake: the bolts have all seen the same history, a set replaced as a set has a single known friction condition, and the failure mode of section 3.2 is a progressive one in which the weakest bolt determines the outcome. Mixing new and used fasteners in one circle creates exactly the non-uniform preload distribution that starts the sequence.

The same logic accounts for bolt kits being supplied as a matched set with a stated coating, nuts to the matching property class of ISO 898-2⁴ where used, and a torque specification tied to that coating — the joint is the unit of replacement, not the fastener. Mechanical property verification of the fasteners themselves follows ASTM F606¹³.

8. What the scatter does to the risk

Sections 3 and 4 give a nominal separation load and a set of published scatter bands, and they are almost always read separately: the joint is designed against the nominal, and the scatter is treated as a tolerance to be lived with. Propagating one through the other is arithmetic rather than research, but it is not usually done, and the result changes how the tightening method should be chosen.

Take a bolt preloaded to a 100 kN target in a joint with Φ = 0.20. The separation condition of section 3.2 puts the nominal knee at 125 kN of external load, and that figure is identical whichever method of section 4 was used, because they all aim at the same target. What differs is the spread about it. Modeling preload as normal about the target with the Table 2 band at ±2σ, the probability that a particular bolt has already opened at a given load follows directly.

Two things follow. The first is that a nominal separation load is a poor summary of a joint tightened by torque alone: at 110 kN — twelve per cent inside the nominal knee, which would ordinarily be read as comfortable — about one bolt in five is already open if the fasteners were run down dry, against roughly seven in ten thousand under yield control. The joint has not been overloaded; it has been under-tightened, and the shortfall was invisible because the torque wrench read correctly.

The second is that this compounds around a bolt circle. Section 3.2 describes a progressive failure in which the least-preloaded bolt opens first and sheds its share onto its neighbors. A circle of forty dry-torqued bolts at that load contains, on this model, roughly eight already-open joints to start the sequence from, and the sequence only needs one. That is the argument for angle or yield control on a slewing ring, and it is not an argument about average clamp force, which the methods share.

The calculation is a model and is offered as one. It assumes normality, treats the published band as ±2σ, takes a single Φ for every bolt, and ignores the load redistribution it then goes on to discuss. It reports no measurement. What it does do is turn two separately published numbers into a statement about risk, and it can be reproduced or refuted from the relations in this paper alone.

9. Practice

Five points carry most of the practical weight.

Check flatness before fitting. The mounting structure's flatness is part of the bearing's operating geometry and is not recoverable afterwards.

Match the torque figure to the friction condition. A torque value without a stated coating or lubricant is incomplete, and the error it produces is roughly a factor of two in either direction.

Tighten in a controlled sequence, with tools verified under ISO 6789¹¹. A crossing or star pattern in two or three passes, rather than one pass around the circle, because each bolt tightened alters the preload in its neighbors through the flange it shares with them.

Re-torque after running in. Embedment loss occurs early and is largely complete within the first hours; a re-check at the interval the manufacturer specifies recovers it.

Treat loosening as a preload symptom. A joint that is backing off is slipping transversely, and adding a locking device addresses the symptom while leaving the cause — inadequate clamp force — in place.

10. Summary

The bolts in a slewing ring or final drive mounting are not load-carrying members in the ordinary sense; they are the means by which a clamped interface is created, and that interface carries the load. While the interface remains in compression the fastener sees only fifteen to twenty-five per cent of the applied external load, and its fatigue strength is often not the governing calculation; once it separates the fastener sees all of it, and fatigue becomes the dominant mode. Preload is consequently specified at seventy to ninety per cent of proof load, is achieved by methods whose scatter ranges from a few per cent to more than thirty, and is lost over time to embedment, relaxation and transverse slip. The features particular to a large bolt circle — a tipping moment that cycles every bolt once per slew, a flange whose compliance loads the bolts eccentrically, a mounting surface whose flatness is transferred directly into the raceway, and coating processes that can embrittle high-grade steel — all compound rather than offset one another. The resulting practice, of replacing bolts as a matched kit with a stated coating and a torque figure tied to it, follows from the physics of a progressive failure in which the least-preloaded fastener determines the fate of the circle.