How to interpret bolt preload relaxation test data

A preload relaxation test is useful only if it shows how much clamp force remains under conditions relevant to the joint, and whether that force remains above the minimum needed for its intended function. A small percentage loss can still be unacceptable when the initial preload is marginal; a larger loss may be tolerable when the joint retains sufficient force with an appropriate allowance for uncertainty.

Start by reading preload relaxation test data as a time-dependent measurement, not as a single product rating. The initial reference point, temperature history, joint stack, loading method and measurement system all affect the reported result. Before comparing two fastening systems, establish whether their curves describe the same physical conditions and whether the measured force can be related to the force that matters in service.

Establish the reference points before calculating loss

Preload loss is commonly expressed as a percentage of an initial measured force:

Preload loss (%) = [(F₀ − F_t) / F₀] × 100

Here, F₀ is the reference force and F_t is the measured force at the stated elapsed time. Retained preload is F_t / F₀, expressed as a percentage. Both figures depend on when F₀ was recorded.

Consider an illustrative test in which the measured force is 100 kN at the chosen reference time and 88 kN after the specified hold period. The reported loss is 12%, and the retained force is 88 kN. If the assembly fell from 110 kN to 100 kN before that reference reading, the reported 12% does not include the earlier reduction. Relative to the first 110 kN reading, the loss to 88 kN would be 20%.

Neither calculation is automatically wrong. They answer different questions. The first describes retention after the selected reference point; the second includes more of the post-tightening change. A report should identify when tightening ended, when the reference force was taken and whether the test included a re-tightening or settling procedure.

Also check whether the endpoint is a force measured continuously under test conditions, a reading after the assembly has returned to room temperature, or a value estimated from another measurement. These are not interchangeable endpoints. An endpoint force may be essential for a design check, while a normalized percentage helps compare curves only when the baselines and test procedures are aligned.

Read the curve in stages

The shape of the force-versus-time curve helps distinguish an initial adjustment of the assembly from continued time-dependent loss. It does not, by itself, prove the underlying mechanism.

A rapid reduction immediately after tightening can reflect seating and embedment at threads, bearing surfaces and joint interfaces. Surface roughness, coatings and local contact stresses influence how much the parts settle. If measurements begin only after this period, the curve may appear stable while omitting a meaningful part of the total loss.

A slower decline during a hold period may be associated with time-dependent deformation in the clamped materials, coatings, washers or other components. In a joint containing a polymer layer, gasket or other compliant material, the joint stack can have a substantial effect on the force history. The result should therefore be treated as a property of the tested assembly under the stated conditions, not as an intrinsic relaxation value for the bolt alone.

Look for changes in slope as well as total loss. A curve that is flattening during a constant-temperature hold supports a different assessment from one that is still falling at the end of the test. A late downward trend does not establish that the force will continue at the same rate, but it does weaken any claim that the assembly has reached a stable level. Extending the hold or testing at additional durations may be more informative than extrapolating the final segment.

Step changes deserve separate investigation. They can correspond to a temperature change, adjustment of the rig, a change in data acquisition or a physical event in the assembly. A graph without annotations can make these causes look alike. Before attributing a sudden drop to joint behaviour, compare it with the test log and any independent force or displacement measurements.

Separate thermal effects from permanent loss

Temperature affects both the assembly and the measurement system. As a joint heats or cools, the bolt and clamped parts can expand by different amounts. Their stiffnesses and thermal expansion determine the direction and magnitude of the resulting force change. A force increase during heating is therefore possible; it should not automatically be treated as improved retention.

For an elevated-temperature test, inspect the force at three points where available: before heating, during the temperature hold and after cooling to a defined reference temperature. The high-temperature reading describes the force under that hot condition. The post-cooling reading helps assess the residual change after much of the reversible thermal effect has been removed. Both can matter if the joint must perform while hot and retain clamp force after a thermal cycle.

Record the actual joint temperature, not just the chamber setpoint. A large assembly may heat unevenly or take time to reach the target condition. Temperature gradients can also affect a load cell or an instrumented bolt differently from the joint itself. Without a temperature trace and measurement locations, it is difficult to tell whether a changing force curve reflects material behaviour, thermal mismatch or instrument response.

Comparisons become particularly weak when one result comes from an isothermal hold and another from repeated thermal cycling. Cycling can introduce changing contact conditions and relative movement that a steady hold does not reproduce. A test should match the operating temperature range, exposure duration and number of cycles relevant to the decision being made.

Check what the instrument actually measured

Direct force measurement is valuable, but every method has a measurement boundary. A load cell placed in the stack measures force through its own load path and can alter the stiffness or geometry of the assembly. An instrumented bolt may measure strain from which axial force is inferred; calibration, temperature compensation and strain distribution affect that inference. The evaluator should understand how the device is installed and whether the instrumented assembly remains representative of the intended joint.

Ultrasonic bolt elongation can provide an estimate of axial force when the relationship between elongation and force is established for the bolt and measurement conditions. Temperature and access to measurement surfaces matter. In all methods, resolution, calibration, zero stability and drift become especially important when the claimed loss is small.

Torque should be treated with particular care. Installation torque is influenced by friction at the threads and bearing surfaces, so it is not a direct measurement of achieved preload unless a suitable relationship has been established for the specific conditions. A residual breakaway or check torque is likewise not a direct measurement of remaining clamp force. Torque-based observations can be useful in an inspection procedure, but they do not replace a validated force measurement for interpreting relaxation.

For close comparisons, ask for the measurement uncertainty and the way it was evaluated across the temperature and force range. If two reported losses differ by less than the uncertainty of the respective measurements, the data alone may not support a ranking. A single trace also says little about assembly-to-assembly variation; replicate tests are needed to assess whether an observed difference is repeatable.

Compare assemblies on the conditions that drive the result

A meaningful comparison starts with a shared test matrix. Bolt grade or nominal size alone is not enough. At a minimum, examine the following details:

• Joint stack: materials, thicknesses, surface condition, coatings, washers, gasket or seal layers, and the number of interfaces.
• Fastener and assembly: geometry, grip length, thread engagement, lubrication, tightening procedure and achieved initial force.
• Exposure: temperature history, hold duration, thermal cycling, external loading and environmental conditions where relevant.
• Measurement: sensor type and location, calibration, sampling frequency, reference time and reported uncertainty.
• Results: individual force-time traces, endpoint forces, loss percentages, replicate count and observed variation.

Initial force deserves particular attention. Two tests can report the same percentage loss while ending at materially different clamp forces. Conversely, two assemblies can retain the same endpoint force but start at different loads. Differences in initial loading can also change contact pressure and the extent of early seating, complicating any attribution of performance to the fastener itself.

External loading is another boundary. A static relaxation hold does not automatically represent a joint exposed to vibration, repeated shear or fluctuating tensile loads. Those service conditions may require separate testing or analysis. Similarly, a laboratory stack with polished steel plates should not be assumed to represent an assembly containing coated surfaces, a compressible layer or a different bearing arrangement.

Connect the result to an acceptance decision

A relaxation percentage has no universal pass or fail meaning. The relevant question is whether the joint retains sufficient clamp force throughout the required exposure, including the most demanding point in the temperature and load history. Depending on the application, the required force may be set by resistance to separation or slip, sealing performance, electrical contact, fatigue considerations or another defined function.

Compare the measured force with that requirement only after accounting for differences between the test assembly and the design case. The assessment should also consider the expected range of installed preload, test variability and measurement uncertainty. A result just above the minimum in a single specimen leaves less evidence of margin than a set of representative tests that maintains adequate force across the relevant range.

Where a contract, design specification or referenced standard defines a test method or acceptance criterion, use its prescribed specimen preparation, timing and reporting rules. Standards can govern different aspects of fastening or joint testing; the designation alone does not establish that two reports used equivalent methods or that a particular retention value is acceptable for every application. Check the applicable edition and the specific clauses invoked by the project.

The strongest conclusion is often conditional: under a defined initial preload, stack configuration and temperature history, the measured force stayed above a specified requirement for the observed duration. If the curve is still declining, the installation range is unknown or the test omits a critical service condition, the next step is to resolve that gap before treating the endpoint as evidence of long-term joint integrity.