Summary

• Vibration loosens bolts mainly through small transverse (sideways) slip in the joint, not through axial vibration along the bolt axis.
• The Junker test reproduces that mechanism on a bench and measures how quickly clamp force decays.
• Split (helical spring) lock washers perform essentially like plain flat washers in this test. They do not prevent self-loosening.
• Devices that do well either block rotation directly (wedge-locking washers, prevailing-torque nuts, thread-locking adhesives) or remove the cause by preventing slip altogether.
• The test is severe and deliberately forces slip. Use it to rank devices under identical conditions, not as a direct prediction of service life.

Why bolts lose clamp force

A bolted joint works because the bolt is stretched like a stiff spring, and the resulting preload clamps the parts together. Anything that reduces that preload weakens the joint: friction no longer resists shear loads, the bolt sees larger stress cycles and fatigues sooner, and the joint can eventually separate. Preload loss comes in two broad forms:

• Non-rotational loss. The nut and bolt do not turn, but clamp force falls anyway. Typical causes are embedment of surface asperities and coatings, gasket or paint creep, and thermal effects. This is a settling problem, addressed mainly by good joint design, adequate preload, and retightening where appropriate.
• Rotational loss (self-loosening). The nut turns backward relative to the bolt a little at a time until preload is gone and, in the worst case, the nut falls off.

Locking devices are sold to stop the second kind. The question is which ones actually do.

Why a tight nut stays put, until it doesn’t

A standard threaded fastener is self-locking in the static sense. The helix angle of an M10 coarse thread is about 3°, while the friction angle for typical thread friction coefficients of 0.1 to 0.2 is roughly 6° to 11°. Because the friction angle exceeds the helix angle, preload alone cannot turn the nut backward, even though the thread flanks push on it with a component that tries to unscrew it.

In the 1960s, Gerhard Junker of Bosch investigated what upsets that balance and published his results in 1969. His central finding was that axial vibration is a relatively weak cause of loosening, whereas transverse relative displacement of the clamped parts is a strong one. When the sideways motion is large enough to cause slip at the nut face (or bolt head) and in the threads, friction momentarily stops holding the nut in place. The unwinding torque generated by the preload is then free to rotate the nut slightly backward. Each cycle produces only a tiny rotation, but because the bolt is stiff, even a fraction of a degree of rotation removes a meaningful amount of preload. Over hundreds of cycles, the clamp force can fall to zero.

Two practical implications follow:

• There is a threshold. If the joint is clamped hard enough that no slip occurs, this loosening mechanism does not operate.
• Beyond the threshold, loosening can be fast. Small per-cycle rotations accumulate quickly at typical vibration frequencies.

How the Junker test works

The Junker test (standardized in DIN 65151, with similar methods in other standards and company specifications) is a bench test built around that mechanism. A test fastener, with the nut and whatever locking device is under evaluation, clamps a stationary part and a movable part together. A load-sensing washer or strain-gauged bolt monitors preload. An eccentric or hydraulic drive shuttles the movable part sideways at a fixed amplitude and frequency, often in the low tens of hertz, forcing the joint to slip on every cycle. The machine records preload against cycle count.

To make results comparable, the test fixes bolt size and property class, grip length, surface finish and coating, initial preload, displacement amplitude, frequency, and number of cycles. Pass criteria vary by specification. A common form is a minimum fraction of initial preload retained after a set number of cycles, or the absence of nut rotation. Aerospace practice also uses shaker-based vibration tests (for example NASM 1312-7), which load the fastener through inertial vibration rather than imposed displacement.

What the test shows

The table below summarizes typical qualitative outcomes reported in the literature and in manufacturer and independent test data. Exact numbers depend heavily on test parameters.

Device	Typical Junker outcome	Why
Plain nut and flat washer (baseline)	Rapid, often complete preload loss once slip occurs	Nothing resists rotation after friction is overcome
Split (helical spring) lock washer	Essentially the same as the baseline	Flattened by preload; spring force is negligible
Toothed (star) washers	Little or no improvement in structural joints	Small bite area; teeth embed and bite is not sustained
Conical spring (Belleville-type) washers	Preload held against settlement, but slip-driven loosening is not stopped alone	Adds elasticity, not locking
Serrated flange bolts and nuts	Variable; helps at modest slip, unreliable at large amplitude	Relies on friction and surface bite
Wedge-locking washer pairs	Strong results; preload typically retained	Cam angle exceeds thread helix angle
Prevailing-torque nuts	Nut retained; some preload loss possible	Friction torque independent of preload
Anaerobic thread-locking adhesive	Good when properly applied and cured	Bonds the threads against rotation
Preventing slip (preload, dowels, shoulder bolts)	Removes the mechanism	Eliminates the cause rather than treating the symptom

Split lock washers

The helical spring washer is intended to keep spring force on the joint and to dig its sharp edges into the mating surfaces. In practice, the preload of a properly tightened bolt is many times the force needed to flatten the washer. At full preload it is effectively a flat ring with a gap in it. It provides no meaningful spring reserve and no resistance to rotation. For an M10 class 8.8 bolt, assembly preload is on the order of 20 to 30 kN, while the washer flattens at a small fraction of that. Junker-type tests and independent studies generally find decay curves nearly identical to those with no lock washer at all, and standard references on bolted joints, including Bickford’s text and NASA’s Fastener Design Manual, reach similar conclusions. Split washers persist largely out of habit and from an era of low-preload joints in soft materials.

Wedge-locking washers

These come as a pair. The mating faces carry cam ramps whose angle is larger than the thread helix angle, and the outer faces are radially serrated to grip the nut or head and the clamped surface. If the nut tries to rotate backward, the washers must slide over their cams, which lifts the nut by more than the thread would advance and therefore stretches the bolt. The preload itself resists loosening, so the lock works by geometry rather than by friction. In Junker tests these washers typically hold preload with only a small initial loss attributable to embedment. Caveats: much of the published data comes from manufacturers, the serrations must actually engage (very hard or heavily coated surfaces can prevent that), and the pair must be installed with the cam faces mating correctly.

Prevailing-torque nuts

Nylon-insert and all-metal deformed-thread nuts add a friction torque that does not depend on preload. They tend to keep the nut on the bolt even after substantial slip, but they can still lose a noticeable amount of clamp force. They suit applications where retention matters more than precise preload. Reuse limits and temperature limits (especially for nylon inserts) apply.

Thread-locking adhesives

Anaerobic adhesives bond the threads and resist rotation. They perform well at moderate slip amplitudes when the threads are clean and the adhesive has fully cured. Performance depends on strength grade, thread gap, temperature, and oil contamination, and large repeated slip can crack the adhesive film.

Conical spring and serrated-flange solutions

Belleville-type and serrated conical spring washers store elastic energy and so compensate for settlement, which makes them useful against non-rotational loss. They do not by themselves stop a nut from backing off under transverse slip. Serrated flange fasteners rely on surface bite and friction, which can be adequate at modest slip but degrade as amplitude rises.

Limits of the test

The Junker test is useful precisely because it is harsh, but that is also why it needs careful interpretation.

• It forces slip. A real joint that is well clamped and loaded mainly in tension may never slip, so a device that fails the test may be perfectly adequate in that service.
• It is a single-fastener test with specific stiffness. Grip length, joint stiffness, and bolt compliance change results significantly.
• It is sensitive to parameters. Amplitude, preload, surface finish, coating, and lubrication all matter. Only compare results obtained under identical conditions.
• It omits other mechanisms. Thermal cycling, fatigue, and long-term relaxation are not captured unless added deliberately.
• A pass at one amplitude is not a pass at all amplitudes. Ask at what displacement the device was tested.

When a supplier quotes Junker results, request the full report: standard or procedure used, amplitude, frequency, cycle count, preload, bolt and plate materials and coatings, number of repeat tests, and scatter.

Practical guidance

1. Ask first whether the joint will slip. If it carries significant transverse load, consider dowels, shoulder bolts, or locating features so the bolts do not carry shear through friction alone.
2. Maximize usable preload. Higher clamp force raises the slip threshold. Tighten with a controlled method and stay within the bolt’s yield margin.
3. Improve the grip. Longer, more compliant bolts tolerate embedment better, and hard, flat, clean mating surfaces reduce settlement.
4. Match the device to the failure consequence. For critical joints, use wedge-locking washers or properly applied adhesives, and add positive locking such as safety wire or castellated nuts with cotter pins where loss of the fastener is unacceptable. Remember that positive locking keeps the fastener in place but does not preserve preload.
5. Do not stack devices without testing. Combinations can interfere with one another, for example adhesive on threads combined with wedge-locking washers.
6. Use lock washers for what they are actually good at. Toothed washers are valuable for electrical bonding and grounding because they bite through coatings. Conical spring washers help compensate for settlement.
7. Validate on your own joint. Run a Junker or equivalent test on the actual fastener, materials, and coatings you will use rather than relying on generic data.

Conclusion

The Junker test overturned a comfortable assumption: that a lock washer under the nut makes a joint vibration-proof. The evidence shows that split lock washers add little or nothing once a bolt is properly preloaded, because they neither keep meaningful spring force nor resist rotation. Loosening is driven by transverse slip, so the most reliable strategies either prevent that slip or use devices that physically resist rotation. Treat the test as a ranking tool and as a prompt to understand how your joint actually moves, and choose locking methods based on that understanding rather than on tradition.

References and further reading

• Junker, G. H., New Criteria for Self-Loosening of Fasteners Under Vibration, SAE Technical Paper 690055, 1969.
• DIN 65151, transverse-vibration (Junker) test method for locking of threaded fasteners.
• Pai, N. G. and Hess, D. P., Experimental study of loosening of threaded fasteners due to dynamic shear loads, Journal of Sound and Vibration, 253(3), 2002.
• Bickford, J. H., Introduction to the Design and Behavior of Bolted Joints, CRC Press.
• Barrett, R. T., Fastener Design Manual, NASA Reference Publication 1228, 1990.