Wind turbine flange bolts: what do we know between inspections?

A connection has a history. Do we retain it?

One question I keep coming back to is what we actually know about a wind-turbine bolted connection between scheduled inspections. We may have an installation record, a maintenance report and a list of bolts checked on a particular day. But can we compare the condition of the same connection across those visits, with enough confidence to distinguish a real change from a change in how we measured it?

That is the question I want to explore here. My interest is in building a useful connection history: a defensible reference condition, repeatable observations and a clear route from an unusual observation to an engineering investigation. This is a proposed workflow, not a claim that a particular instrument has already solved wind-turbine integrity.

The useful lesson in an older bolting example

Spector’s application note on steel studs and bolts describes magnetic assessment of connections in power equipment, including high-pressure preheater studs. It reports a distribution interpreted as uneven tightening, with deviations from the average of −37% to +43%; after selected studs were adjusted under instrument control, the reported deviations were within 15%. The note also discusses access where enclosed cap nuts prevent end-face measurements. [1]

These are Spector’s reported findings, not independently verified BROT measurements. The published note does not establish uncertainty, independent preload confirmation or longitudinal performance for a wind-turbine flange. Its transferable idea is to examine a connection as a population rather than treating each reading as an isolated result.

Rinat V. Zagidulin and Timur R. Zagidulin’s related bolting work appears in the ECNDT 2014 proceedings. The proceedings listing confirms the publication and authorship; it does not establish turbine performance. [2]

Installation evidence and condition evidence answer different questions

A record of the tightening operation is valuable. It describes what was done, under an agreed procedure, at a particular time. A subsequent measurement asks another question: what can we establish about the connection now?

Phares, Lee, Brockman and Rooney’s 2016 study of structural-bolt verification reviews several ways of approaching this problem, including ultrasonic, strain-based and magnetic techniques. It also cites the Zagidulins’ bolting work. Its scope is structural steel connections, not validation of a wind application. [3]

For a wind flange, I would keep three questions separate: was the joint installed according to the specified procedure; is there evidence of a change that requires investigation; and is the connection acceptable for continued service? One measurement does not automatically answer all three. The applicable manufacturer’s procedure and responsible engineer still govern acceptance and intervention.

A population map needs more than colours

It is tempting to present a ring of bolts as a red, amber and green picture. Before those colours mean anything, I would ask what the comparison actually represents.

Bolts should be grouped by relevant material, geometry, connection position and measurement access. The record should retain a stable bolt identifier, raw readings, measurement locations, instrument configuration and the operating conditions that could influence interpretation. Replaced bolts and maintenance interventions should remain visible in the history.

For an exploratory RMF workflow, I would initially describe the output as a comparative screening indication. A bolt that differs from comparable neighbours deserves attention; it does not automatically have a defect, a known preload error or a known remaining life.

The average is not an acceptance criterion either. A whole population could share an undesirable condition. I would want an independently characterised reference, alongside population comparisons, before treating an apparently uniform map as reassuring.

Repeatability comes before trend interpretation

A baseline becomes useful only when we know how much the result changes without a meaningful change in the connection. My first validation exercise would therefore be deliberately modest: repeat the observations on the same bolts, remove and reposition the probe, involve a second operator and retain the raw results.

I would then compare that repeatability with controlled changes confirmed by an independent reference method. The test needs to distinguish the contribution of measurement position, material differences, magnetic history and operating conditions from the change we actually want to observe.

Only after that would I treat a change between inspection dates as evidence of a developing trend. Even then, two visits do not reconstruct everything that happened between them. Repeated screening can improve the recorded history; it cannot provide continuous knowledge of an unmonitored connection.

Screening should earn its place beside conventional inspection

The role I would test for RMF is specific: comparative stress-state or population screening, abnormal-zone prioritisation, a repeatable baseline and history, then targeted NDT or engineering investigation. The question is whether this helps the team decide where to look more closely, with a measurable improvement over its existing workflow.

A useful trial should investigate some unflagged locations as well as flagged ones. Otherwise, it may tell us how often an indication is confirmed while telling us little about important conditions that screening missed. The confirmation method must be suitable for the particular question—preload assessment, crack investigation and joint behaviour are different tasks.

Spector’s 2015 diagnostic paper is relevant here. Its abstract describes different possible causes of a high magnetic indication and the use of additional examinations to investigate the cause. [4] A screening result is a starting point for diagnosis, not a substitute for it.

The evidence needed before predicting life

If the ambition is eventually to support a residual-life decision, the evidence burden changes substantially. I would want longitudinal observations tied to a defined connection configuration, independently verified condition changes, relevant loading information and documented outcomes. Any model would need a clear domain of use, uncertainty and prospective testing on connections outside its development dataset.

An attractive trend line is insufficient. So is a laboratory response to loading. Neither alone demonstrates when a field connection will become unsafe. At present, I would make no claim of validated remaining-life prediction, guaranteed defect detection or universal absolute-stress measurement for this proposed wind workflow.

The first milestone is smaller and useful in its own right: show that a repeatable comparison can identify changes worth investigating, and that the investigation helps an engineer make a better decision.

The question for the next workshop

I would start with a single, bounded question: can we repeatedly distinguish independently characterised differences within one representative flange-bolt population, using a procedure that another operator can reproduce?

If you work with wind-turbine connections, I would welcome your perspective on the reference measurement and repeatability test you would require before trusting such a history. That discussion is more valuable than another promise about predictive maintenance.

ENGINEERING DISCUSSION

What would you require from the baseline?

Share a technical observation or challenge an assumption. Comments are reviewed before publication. Your display name and comment will be public if approved; your email stays private. Please omit confidential projects and customer details.

Register for free technical workshop invitations