Head nuts coming loose
 

This is the 3rd time that I have had head nuts come loose. 2nd on a 3L and 1st on a 3.2.

The 3.2 was built by a renowned engine builder, so I need to think that it happens. When I told him I was getting oil seepage he knew right off the head nuts were coming loose.

The other 2 were steel studs in 930 case. No dilivar studs.

The studs are not coming out of the case. The nuts are just coming loose.

All studs were wire brushed clean and oiled along with the nuts and washers.

All studs were torqued to 26 ft lbs.

How common is this? Why is it happening?

Just re-torque them and go down the road? How often should they be checked?

Spec on a 3.2 is, off the top of memory, 15 ft# then 90 degrees swing which surpasses the 25 ft # for the 3.0
Bruce

15Nm or 11 ftlbs followed by 90 degrees for the 3.2

33Nm or 24 ftlbs for a 3.0 litre

The extra preload is possibly to prevent the head lifting due the extra force produced by the larger capacity and shouldn't really make much difference to the nuts coming loose.

I would think that the tightening strategy used for the 3.2 produces a preload increase of between 10 and 15% but should give a significant reduction in the scatter of the preload.

It is a tightening method I would prefer but I am not sure this is the cause of the problem.

I am not aware that the Workshop Manual recommends re-torqueing as this seems risky without checking that cams will still turn and that the cam carriers a still correctly aligned.

What lube was used on the threads? Were the nuts re-torqued after 500 miles?

Motor oil. No, not retorqued. Nowhere can I find that re-torquing is required. In fact, h ave read that it is not required.

Beginning to wonder about that premise.

Gordon,

I am interested to understand more about re-torqueing as this is something we never undertake and we have never had any problem.

When you re-torque do you take off the chain covers and check cam rotation or do you assume that there will be no change from the original build?

I would worry that re-torqueing could disturb the cam carriers but may be I am just over-cautious.

When I rebuilt my '71's motor many years ago, I was told by the experts at the time to re-torque my head nuts after about 500 miles. IIRC I backed off each nut just enough to make sure it was loose, less than a quarter of a turn, then tightened them to the correct torque. That was in 1985 and the motor is still running strong.

Wayne, in his book, recommends re-tightening the head nuts at 500 miles.

I would like the experts to chime in on this as I am in the process of building a 3,0 motor and am interested in the details of this procedure.

As would I, Gordon. Asked an expert about adjusting the valves and retorqueing the heads and was told it was not necessary. Clearly, that didn't work out too well. I can see the valves not needing adjustment.

If you read the original Volume 1 of the Workshop Manual it states very clearly that it is important to torque the head nuts with the cam carrier in situ and to rotate the cam to ensure that there is no distortion of the carrier whilst tightening the heads.

It also states that if the cam does bind to undo the nuts and change the tightening sequence until the cam is free to rotate.

The comment is made that 'there will be a tightening sequence that allows this to happen'.

There is no mention of re-torqueing.

If to re-torque you loosen the nuts then my concern is that any checking of the cam rotation will be rendered invalid and some distortion of the carrier could occur and with all the chains and rockers in place it would be hard to know if this had happened.

This raises some questions, an I am sad to say I don't have the answers

The first question concerns the cam carriers - are later 4 bearing carriers stiff enough that any deformation become irrelevant and it is only the early 3 bearing units that are an issue as they may not have the same level of integrity.

The second is whether backing off the nuts is a good thing. With some of the early Coventry Climax race Engines we build we do retorque but we don't loosen the nuts first.

It would be very helpful to gather more information.

it would be interesting to use some Dykem and see whats really turning?

4-bearing cam tower are not stiff enough. It is definately possible to have the cam bind during the torquing process.

My LHS did. Had to loosen the nuts and change the sequence. 100% possible.

I thought the guidance was to check the torque. Not loosen and retorque. The difference being you only try to tighten the nut and expect no motion, but could tolerate a small angular motion.

I also had an exhaust (Dilivar stud was not broken) nut loosen. Likely lubricated by rocker shaft leakage. Reinstalled and retorqued.

Unfortunately the damage was done. Head gasket was already damaged and did not survive another track day.

I thought the recommended lubrication was anti-seize. Not engine oil. Anti-seize should have a lower friction coefficient, hence a higher preload , and less chance of loosening.

Anti-seize compounds typically have a K factor of between 0.13 and 0.16.

There re some vey expensive nickel based compounds with values of 0.11 but they are normally for high temperature use and tend to be very expensive.

Engine oils seem to have K factors of about 0.12 to 0.15 at room temperature and this suggests that there is not much difference in either lubricant. Engine oils will not allow as many re-tightening cycles but for this application that may not be a consideration.

The manner in which the stud/nut is lubricated and how it is tightened may have a significant impact on behaviour.

A typical nut factor for a dry steel nut could be 0.5 or higher for stainless steels. These figures reduce significantly with lubrication but most quoted values assume that the faying surface of the nut/washer are also lubricated.

If only the thread is lubricated nut factors can increase significantly and then the preload could suffer quite badly.

I am not sure if the figures quoted by Porsche are for dry or lubricated nuts.

We tend to oil the studs and nuts but dry them prior to tightening.

The material used for the stud and the nut will also have an impact with Stainless Steels and Inconel being more sensitive to the manner in which they are lubricated and the way in which they are tightened and typically have higher K Factors.

I am confident that the way in which head nuts are lubricated and tightened will be relatively inconsistent with most individuals having their own preferred method.

It seems to me that if we were to use a stud/nut with either a specific coating or use a specific lubrication method we could eliminate some of the variability which may be the root cause of the problem.

There is also a question as whether it is too little preload allowing the nut to shake loose or if it is too much preload causing some stress relaxation in the thread of the case and hence loss of preload.

I can believe that if we measured the real preload in a range of studs we could see a 30% variation within a single engine and probably 50% if we measured a few engines.

Trying to determine what is happening could be tricky and I think needs some further investigation.

http://i197.photobucket.com/albums/a...psbtspodfp.jpg

Chris

Variation in cf is probably the strongest reason to go to angle of turn. Takes friction out and relies on the accurately cut threads to control stretch.

Would the 3.2 procedure be ok for the 3.0? I'd think it would be, since all the parts in the assembly are the same/similar.

When I installed ARP head studs in my 930 motor with red loctite on the case end threads I used ARP Ultra Torque Assembly Lubricant 100-9908 on the 12 point nut threads and the surface of the ARP forged steel washers where the nuts contact them.
That's what the ARP install directions said to do.
I'll check the torque on the nuts next time I do a valve adjustment.

I didn't like the little squeezed out mess of this stuff in there after tightening the head stud nuts so I sprayed a bunch of brake clean in there to rinse it off afterwards.
A paint spray gun filled with mineral spirits at 120psi adjusted for a conical spray pattern does a better job of rinsing it off and it's more economical than multiple cans of Brake Clean.

This is the thread lubricant I used and one small packet of it is a little more than enough for the 24 head studs and nuts on a 911 motor.

http://www.summitracing.com/parts/arp-100-9908?seid=srese1&gclid=CLSo8OzJwMcCFYoTHwodYfcD_w

It has been recommended by Porsche, I have read, that anti-seize should be used on the head nuts.
The concern I have with respect to retorquing is the starting friction. I'm not sure that is the correct term. I am referring to the force required to start the nut turning. Once the nut is turning it requires less torque to turn the nut than it did to start the nut turning.

That is why it has been suggested to back off the nut slightly then re-torque. Maybe this is unnecessary, and perhaps, I am over thinking it.

Interesting comments.

I would agree that a low initial torque followed by an angle will reduce the variation in preload quite significantly and can't see why there should be any difference between a 3.0 and a 3.2.

The preload will increase slightly but I would believe that this is well within the capacity of the engine case.

The use of ARP Ultra Torque will substantially affect the preload for a given torque and if you torque to the figure in the spec book then you will have a higher than standard preload, I am not sure how much of an issue this is but I would worry about magnesium cases.

The stiction between the nut and the washer will be quite significant and would equate to a significant percentage of the 11ftlbs used by the torque/angle method for a 'dry fastener'

If you have carefully lubricated the head nut with anti-seize then the stiction will be much lower and although it will still have some influence.

The more you think about tightening cylinder head nuts on a 911 the more you realise that is really is a 'black art'.

For engines without gaskets I am not sure if re-torqueing is needed and I am not sure if a CE ring qualifies.

Disclaimer: I'm certainly no engine builder nor claim to be, however:

About 1,000 miles after the top end rebuild of my 3.0 SC I assisted in having my valves adjusted and heads re-torqued by an indy whom I consider one of the best in the business that has built a LOT of 911 engines over the years. I had replaced the lower row of dilivar studs with steel, and for the procedure he started at the rear of the engine (chain end) and progressed forward re-torqueing the head nuts. All were loose between about 1/4 to 1/2 turn. What I don't remember was whether he did the top row or the bottom row first. A bit shocked as I had followed Waynes procedure of starting in the center and working outwards when the engine was built, I didn't question his procedure as I am more than confident in his abilities and experience.

What, in theory, would a re-torque accomplish? A torque check I understand.

So which side head studs are frequently coming loose... intake or exhaust?

Steve Weiner has mentioned that he has seen ARP studs not hold their preload. Perhaps the material and shank size has a role whether the studs hold their initial torque setting.

The shank size will affect the force that is developed due to expansion.

Dilavar will result in a relatively low force increase of around 400lbs per stud.

For the majority of steel and other studs the CoE is about the same level (11-12 ppm per degC) with Inconel being around 5% greater which for the sake of this argument we can neglect.

The diameter of the shank will have a very significant impact on the force developed by the expansion of the cylinder and it is convention to use reduced shank fasteners when expansion is significant.

The chances that any of the materials used for head studs would be prone to the effects of creep at the temperatures involved is extremely unlikely as the activation energy needed to allow the mechanisms that cause creep to occur would tend to be much higher than those experienced by 911 head studs.

ARP studs for 911 engines are made from Inconel 718 which is commonly used for Jet Engine components and is routinely used at temperatures up to 0.6 x Melting Point which is around 2400 degF so not much to worry about in a 911 motor.

I think ARP suggest a torque of 36 ftlbs and this is a relatively high preload compared to Porsche's recommendation and may be when combined with the force created by expansion does allow relaxation to occur in the case threads.

I can't imagine there is much else that can be of concern with regard to the metallurgy of the studs as Dilavar failures always seem to be brittle elastic in nature.

Here's a crazy thought.
After all the head nuts are torqued down good remove one nut at a time and spray all the thread lubricant off the stud and nut threads with brake clean. Then reinstall the nut with low yield loctite (not high yield red loctite) and torque it back down.

Maybe not the most kosher way of doing it but the nuts should stay put after that.

I have considered using an aerotight K Nut and I need to buy a pack to see what the variation in running torque is so that this could be added to the torque to tighten.

I will ask an old contact at Simmonds and see if there is a good method.

It seems that the determination of the 'running torque' to use a Kaynar prevailing torque nut is quite straightforward.

The only additional torque needed is the torque to cause the deformed part of the nut to expand and run along the thread.

Simmonds don't quote any values and always recommend that the user carries out their own testing to determine both the torque to run the nut and the scatter and then determine the torque to add to the value required to obtain the correct preload.

I have just ordered 50 x HM20 Kaynar nuts to match some new studs we have just designed.

These nuts are 12 point prevailing torque and are CAD plated and dry film lubricated with Moly disulphide.

We will also dry film lube the relevant and of the stud .

The plan is to determine the running torque and then to measure the preload in a test fixture which has a relatively precise load cell and to see if we can learn a little more.


http://i197.photobucket.com/albums/a...pspczsg1gx.jpg

We have been making some Ti studs for Mag cases for a while now and we have always made them with a socket head to allow ease of installation and we will continue with this feature on the new design.

We have also added a 'dog point' on the base of the stud to allow the stud to be tightened into the case without 'bottoming' into the last few threads which are never correctly formed. This feature is being used now on the majority of Indy and F1 engines and studs with this feature are also available for 996/997 Turbo engines.

This feature will allow all of the threads in the case to be uniformly loaded and should reduce any tendency to stress relaxation to occur.

It does also provide a second benefit and this concerns use of Loctite.

ARP suggest that studs are installed without thread locker or that they are torqued before the thread locker fully sets.

This is fairly tricky on a 911, although we have made a simple fixture to try to achieve this result.

By using a socket head on one end of the stud and a dog point on the other it will be relatively easy to install the stud to a modest torque of say 12 ftlbs and ensure that the thread locker is correctly applied and that the joint is stable.

Interesting.... So Loctite may not be completely stable over time?

I think is concerned with the fact that there if there is no load it could sit in the clearance and allow some movement but I am not really sure and I haven't found any good published data.

Never had a stud come loose, and never taken apart a motor with a loose stud... Dilivar, steel, ARP or otherwise...

Cheers

I am not sure anyone has suggested that studs loosen in terms of the case to stud interface but that doesn't mean that if thread locker is used incorrectly the resulting assembly is stable.

As any thread locker works by filling the gaps that occur when there is clearance it is possible that a thin film of thread locker could sit between threads. ARP's suggestion to torque any fasteners before the theadlocker has set could be to ensure that there is a metal to metal contact which will be more stable in an axial direction.

If thread clearance has been increased by poor cleaning techniques it is possible that some deformation of any thread locker that is present between the threads once they are loaded could occur.

This axial deformation may allow axial movement which could reduce preload. Only a very small and in the context of a stud virtually immeasurable axial displacement is needed.

The thread locker sitting in the gaps at the ends of the truncated threads will still operate to prevent loosening due to vibration or thermal cycling.

Installing studs that are torqued before the thread locker sets can only be a positive action and would eliminate another variable.

Sorry, wasn't clear. I meant, never seen a loose head stud NUT, ever.

We had a 3.0 SC with 100K miles about 3 months ago with significant oil leaks at the cylinder bases on 2 cylinders.

There were no broken studs but head nuts on both cylinders with the leaks were loose.

Both cylinders showed compression ratios of about 50psi. One had the complete CE ring missing and the other had a section of about 30% remaining.

The damned thing didn't even demonstrate a misfire and the only giveaway was the oil leak.

On an old engine I would suspect some relaxation of the threads in the engine case.

My concern is that although this issue may be rare it does occur and is generally unexplained, particularly on new builds.

Reducing variables seems to be a sensible idea.

Chris,

Was this a rebuilt motor, a motor that had never been apart or?? Was/is the issue related to assembly?

All 4 nuts loose, top or bottom or both?

Stock studs with Dilivar on bottom row? Which ones had loose nuts?

What sort of thermal stresses had it been exposed to? I realize we never really get the true story from the client... :)

Which failure mode occurred first, the CE ring or the nuts coming loose?

Reducing the variables is important, but understanding what causes them is equally important.

I realize our sample size is small. We see these motors daily and have rebuilt around 40 and inspected another 60 or so. But, every time we see significant leakage between cylinder and case, cylinder and head or a damaged mating surface (either cylinder or head) there is a broken stud to go with it.

I am not saying it doesn't happen (because clearly it does), just saying we are yet to see it.

Would love to know/understand a method or parts selection which would help improve or eliminate the issue.

Cheers

The engine was a 1982 (UK Spec) 3.0 SC.

It had never been apart but had been fitted with a pair of 40IDA3C carbs about 5 years before we first looked at the car.

The car is driven regularly including TrackDays when it is driven exceptionally hard but is has regular oil changes, is never thrashed until it is warm etc, etc.

We had the car in our workshop for routine maintenance and to fit a lightened flywheel and a set of headers.

We carried out a compression test as a matter of routine and found 2 cylinders which were well down.

Once we had dropped the engine and removed the fan shroud the oil leaks at thee cylinder bases were obvious and had clearly been leaking for some time.

We suspected broken Dilavar Studs.

It was surprising to find that there weren't any broken studs just loose nuts.

The upper nuts were 'finger tight' on the two cylinders but the nuts on the Dilavar studs seemed tighter but I couldn't state a torque figure.

The remnant of the CE Ring seemed to show that it had been burnt by combustion gases but the head and its interface were not too bad , again surprising.

We did refurb the sealing face on all the heads but didn't change any studs (No budget available!)

We cleaned all the leaks re-fitted the cylinders and torqued it all up and away it went.

For the last 2 years it has performed as well as ever, its compressions have been checked twice and are consistent.

I think the engine needs an overhaul and we have it booked in this winter and will see if we can find any other issues.

I can't say for sure but I think that the rings failed due to the heads loosening but have no proof.

I don't like the difference on CoE between the upper and lower studs as this must allow some odd forces to be applied to the interface of the joint and not something that IMHO is a sound practice.