Why M6 Screws Fail - White Paper

[FSM]

Why M6 Screws Keep Breaking in Foiling

A Systemic Fastener and Torque Specification Failure

Executive Summary

Across the hydrofoil industry, riders routinely experience broken or seized M6 screws, most commonly at the mast-to-fuselage connection. These failures are often attributed to improper installation or user error.

In reality, the failures arise from a system-level mismatch between:

• Torque specifications

• Fastener material behavior

• Thread lubrication practices

• Structural packaging constraints

• Inconsistent hardware quality

Compounding the problem, most brands do not provide guidance on thread lubrication at all, creating two opposing and equally damaging failure paths.

1. The Real Foil Fastener System (As Built and Used)

Typical foil hardware systems share the following characteristics:

• Fastener size: M6

• Fastener material: Stainless steel or titanium

• Female threads: 316 stainless inserts

• Structure: Carbon fiber mast and fuselage

• Environment: Saltwater, cyclic loading

• Assembly: Frequent disassembly

• Thread engagement: Short by necessity (weight and space constraints)

• Lubrication practice: Unspecified by most brands

This is not a generic bolted joint, yet it is treated as one in documentation.

2. The Industry Silence on Lubrication

Experienced users widely understand that anti-seize is essential in stainless-on-stainless marine assemblies. However:

• Most foil brands do not recommend any thread lubricant

• Installation manuals frequently omit lubrication entirely

• Torque values are provided with no stated thread condition

This silence leads to two predictable outcomes:

Failure Path A: Dry Threads

• Stainless-on-stainless galling

• Progressive seizing during tightening

• Cold welding

• Broken screws during removal

• Seized remnants in inserts

Failure Path B: Lubricated Threads (Without Torque Adjustment)

• Users apply Tef-Gel correctly to prevent seizing

• Torque values remain unchanged

• Clamp load and torsional stress increase dramatically

• Sudden screw fracture during tightening

Both paths end in failure — and neither is user error.

3. Short Thread Engagement Is a Constraint, Not a Defect

Foil hardware necessarily uses:

• Short threaded inserts

• Engagement typically ~1–1.5× diameter

This is a deliberate design tradeoff driven by:

• Weight

• Limited laminate thickness

• Structural packaging

• Hydrodynamic requirements

Importantly:

• Insert pull-out is rare

• Thread stripping is rare

The dominant failure mode remains torsional fracture of the fastener, indicating excessive applied torque.

4. The Torque Specification Problem

Many foil brands specify 10 Nm for M6 fasteners.

This value typically originates from generic torque tables that assume:

• Dry threads

• Carbon steel fasteners

• No galling risk

• No anti-seize

These assumptions are invalid for:

• Stainless fasteners

• Marine environments

• Carbon fiber structures

• Repeated assembly cycles

5. Lubrication Fundamentally Changes Torque-to-Load Behavior

Torque is a friction-dependent proxy for clamp load.

When Tef-Gel or similar anti-seize is used:

• Thread friction decreases by approximately 30–50%

• Clamp load increases for the same applied torque

• Torsional stress in the fastener rises sharply

For A4-70 / A4-80 M6 stainless screws:

• Torsional yielding typically begins around 5–6 Nm when lubricated

• Sudden fracture becomes likely above this range

Thus, a torque that may survive dry installation becomes destructive when lubrication is added.

6. Titanium Fasteners Increase Risk in This System

Titanium fasteners are often perceived as superior, but in this application they:

• Gall aggressively against stainless inserts

• Have poor thermal conductivity

• Are sensitive to micro-stopping during tightening

• Fail suddenly with minimal warning

• Are frequently sourced without certification

In M6 size, titanium provides less tolerance, not more.

7. The Aftermarket Hardware Quality Gap

Many riders unknowingly install uncertified fasteners:

• Unknown alloy and heat treatment

• No proof-load testing

• Inconsistent thread geometry

• Decorative markings without certification

Torque specifications implicitly assume certified hardware, but this is rarely stated.

8. Why Mast-to-Fuselage Joints Fail So Often

This joint combines:

• Small fastener diameter

• Long screws with short engagement

• Stainless-on-stainless threads

• Carbon fiber conductivity

• High bending moments

• Frequent assembly cycles

With unclear lubrication guidance and excessive torque, failure becomes inevitable.

9. What Actually Works in Practice

Field experience consistently supports the following:

• Fastener: Certified A4-70 or A4-80 stainless steel

• Lubrication: Tef-Gel or equivalent anti-seize

• Torque:

  • 4–5 Nm typical

  • 6 Nm absolute upper limit

• Technique: Smooth, continuous tightening without pause

This combination:

• Prevents galling

• Avoids torsional overload

• Provides sufficient clamp load

• Greatly reduces failures

10. Why This Persists Industry-Wide

• Lubrication guidance is omitted to avoid liability complexity

• Torque values are inherited rather than revalidated

• Titanium is marketed as premium

• Hardware quality is assumed

• Acknowledging over-torque implies past guidance was flawed

Conclusion

The persistent failure of M6 fasteners in foiling is not due to rider ignorance or misuse.

It is the predictable result of:

• Silence on thread lubrication

• Over-torqueing lubricated stainless fasteners

• Inappropriate fastener material choices

• Generic torque tables applied to a constrained system

Until lubrication and torque are specified together, failures will continue.

Final takeaway

If lubrication is not specified, torque is meaningless.

If torque is not reduced, lubricatin becomes destructive.

ChatGPT helped me….but I did work with ChatGPT providing corrections until it got it right.

[silas]

Please please please keep the AI slop out of this forum. If you have to, provide a link to the output and a clear note that the content is AI generated, and focus on the bits you find relevant or want to have a discussion about. Those that want to can jump in to the AI content, those of us who don’t can skip it without having to scroll through walls of text.

As with most AI, there are some valid basic points such as torque delta between dry and lubed threads. But overall this is, in my view, another useless steaming pile of trash that in itself fails to move the discussion forward or help end users manage their gear better.

[sailandfish]

Its actually not useless info. There is a lot of good information presented. Though I’m not sure this is actually a problem that needs solving.

I think over torquing is probably problem 1. Followed by leaving stuff assembled leading to crevice corrosion breaks.

But in general stainless on stainless galling is an under reported issue. I have given up on educating people about it after rigging for a sailing event. The skipper and boat owner was really cranking down on cap shroud turnbuckles (which were stainless on stainless), putting all of his strength into it. I suggested he may destroy them, and should a. Use never seize or b. Just realize he’s strong and cranking down on small hardware. Anyway, I got called an idiot and verbally attacked.

I make a living looking at people’s broken marine shit, and when necessary explaining to the courts. I have no need to loose friends by trying to educate the gen pop.

[Hdip]

@silas

Please please please keep the AI slop out of this forum.

But @Erik uses ai to read this forum. So won’t it be funnier to have an ai giving him answers from another ai?

[FSM]

The big misunderstanding that Titanium is a better screw is a biggie many still don’t get.

[Erik]

Hahahaa. I don’t. Should we ban ai posts? I’d be in favor of it. It’s just using the forum for its info anyways ….

[Hdip]

I don’t especially care one way or the other, but I did just see this image and it seems appropriate.

[silas]

I think my preference would be to try guidelines first, if that isn’t hopelessly naive. I think how @matt dealt with it referencing AI input over in the Tubercles thread is a pretty good way to still let AI fuel some interesting discussion without having it take over. Doing nothing is going to result in people ChatGPTing at each to try and support whatever their preferred position is on a topic.

[silas]

As I said, there are correct and accurate nuggets. But almost no manufacturer provides torque specs. The whole yack about the joint closure is just rubbish. Overall, it is objectively wrong, irrelevant due to incorrect context, or inaccurate over a large percentage of the word soup. This is my definition of low quality and useless. Great that you have a different standard. The fact remains that you are really reinforcing my point: it is with responses like yours that really bring value through interpretation, reflection, and inclusion of relevant experience from parallel applications. As others have shown on this very forum, it is eminently doable to reap the benefit of getting a discussion going using AI, without just cutting and dumping the full output.

[Erik]

Agreed. I think linking to or copying parts of an llm post is ok as long as it’s then contextualized from the user.

[sailandfish]

You’ve responded to me, I’m not Ai, and Dwight, the OP is also not Ai. He clearly stated he used Ai, but the stuff he posted is generally not inaccurate information. He is trying to educate.

You don’t want to be educated, and you’ve also failed your reading comprehension test. Did you use Ai in your response? Are you the Ai?

Since you have not contributed anything technical or informative, what is the purpose of your attack on the OP? Just bored? Enjoy riding with a gap between your mast and fuse.

I know for us in FL conditions have sucked a big one lately. Hopefully we get a break soon and can go foiling in real life so we don’t need to entertain ourselves by attacking a fellow foiler.

[EmeryThrill]

I’m in the no-AI camp-on this one … my frustration is not that people use AI, it’s that they relinquish every responsibility when they use AI, vs. using it as a research source or using it as a proofreader.

We are poorly served by a 1-sentence prompt and a 3-page output thrown over the fence into this forum. But if someone wrote a 3 paragraph thought on their own experience and expertise (e.g. how to lubricate their and properly torque their screws) then had ChatGPT check for typos before they post, I don’t think anyone would mind.

I’ve had this happen multiple times at work, where someone sends both the real work, as well as some Gemini or ChatGPT output and expects everyone to also read 12 pages of crap they haven’t properly sorted through, and then expects someone else to merge these two work products.

For example, at minimum … why not format this better, single line spacing in line with how everyone uses this forum? And why not have some sort of wrapper/intro: “Hey guys, I keep breaking my screws and recently lost my Lift 130x in shore break due to this … so I’ve done some research and….” This is just a tad too lazy, IMHO.

[KDW]

I don’t think you need ChatGPT to write pages about information that you could explain in a few paragraphs. This much embellishment with such little substance tends to contain a lot of errors or repetition and I don’t think is valuable to a forum. If I want ChatGPT’s thoughts I can ask it myself.

Many issues I see with bolted connections actually come from people under-torquing/ inadequate preload for the load, compromised, corroded or dirty threads, not replacing after stretching, and occasionally over torquing leading to galling or cold welding then torsional failure (titanium m6).

Anything with an Alu thread needs some kind of grease but more importantly should be loosened or disassembled after use. Bolt threads will accumulate aluminum oxide and should be cleaned periodically.

When using stainless bolts, use only high quality 316 stainless. These are way harder to over torque than titanium or 304.

On a butt-style joint (f-one KT duotone afs Mikeslab etc.) movement in the joint will lead to unintended bending or shear load of the bolt eventually leading to failure. Under-torquing these connections will make them a lot weaker. it is important in these cases for the manufacturer to engineer enough safety margin to minimize issues.

I would avoid titanium bolts on butt-style connections. (Except for baseplate).

Foot style connections don’t need much torque at all as they only rely on the bolt to hold the pieces vertically.

in any connection, if you come in from a session with loose bolts, I would recommend replacing them as they are probably stretched.

Higher grade black oxide bolts can last a few weeks before needing replacement. If you are breaking bolts you can always order a bunch and replace as needed.

If you lubricate, don’t forget the bolt head mating surface. Lots of friction there

[Kurtis]

I deal with fastener’s, breaking strength, pull out strength, and types alloys regularly for building components in exterior finishes.

Its typical to see fasteners manufacturer changed from the fastener that was originally tested with an assembly due to availability. But, the value generally need to exceed those of the tested fastener.

My opinion is that M6 just doesn’t leave enough safety factor for our use in this case. If it was a one time assembly, great. But its easy to break an M6 on the first use if you aren’t careful. But after a couple of load cycles, its too easy.

I’ll stick with M8, even if that means a little more drag due to extra meat.

[FSM]

It’s sad to me to see so much hate for ChatGPT. I already knew everything I posted when using it, but thought to myself, maybe people who don’t know I spent my entire career as a mechanical engineer, would believe ChatGPT more.

We can’t all know it all, about everything in the world. Accountants, lawyers, dishwashers, etc, not knowing engineering.

Far fewer people use AI, than do.

One great example of AI use for me, was something that could have saved my wife’s life. Her follow up monitoring of cancer, didn’t seem right to me. So I gave ChatGPT her pathology report. It said her monitoring was wrong for the type of cancer she had. So I got a second opinion from the Moffitt Cancer Center in Tampa. Moffitt agreed with ChatGPT 100%. Her local doctor was an idiot. The wife is doing fine.

[TooMuchEpoxy]

Tensile strength is diameter^2 and bending strength is diameter^4 ThankYouForComingToMyTedTalk

[Speedialer]

Thanks for the post. It confirms my experience as a longstanding m6/tefgel/light torquer without any incidents after 300+ sessions on the same mast.

[kaimoana]

I don’t come here to read what a computor thinks, I come here to engage with other humans. For me it’s all about the stoke rather than the “correct information”

[Paul]

I see this forum as a beach after sunset filled with foiling enthusiasts, wrapping up their gear, hanging out and sharing tips, tricks and stories.
If somebody pulled out their phone and started reading a LLM, I’d leave.
They have their uses, but I don’t think social groups are the right place.

[Matt]

Agree. It’s justifiable backlash for the way you shared, not the what. It’s just lazy word vomit that you posted, too long, too inhuman, not enough context, zero effort in a preamble? TLDR instantly when I see the AI formatting without context.

It’s a developing best practice and dumping noise is clearly poison to a forum like this.

To be productive: lead in with “Why are you sharing this?” “Who should read it?” “Why did you need AI to help?”, “What did you ask it?” “Why did you ask it that?” “Why does AI need to mansplain this” etc

In your case, you have thoughts+experience, AI could tighten up the ideas by generally, use it to share and refine your thoughts, speak to the thing and ask it to lightly edit it for coherences, not to be drafting the entire text.