Turbo bolts keep breaking

Op can you post pictures of failed parts? Would help alot more than everyone guessing with opinions.


But I think you said you were running 41 psi of boost. So am I safe to say, its not a stock tune? So is it a stock turbo, or aftermarket turbo.

Does this thing belch black smoke?

 

This doesnt quite jive with my knowledge of fasteners. Bolts are known to stretch but while there is a certain amount of springing back to original, it will never go ALL the way back, rebuilding my car engine, the procedure was to torque to spec plus 1/4 turn, let sit an hr, then loosen and tighten to spec again.

Also yes, going red hot would soften, but 600 is nowhere close to that temp, thats near the tempering point for steel..

 

A chart similar to this one would be useful for fastener selection and sizing. The elastic range is the never exceed limit which if exceeded will permanently change the shape of the metal, which is known as the plastic range. It's a simple matter to prove the integrity of the bolt with the specs provided.

Now that's just part of the story. While steel and ferrous metals are remarkably elastic, there are fatigue limits which are often exceeded. So fatigue is where the bolt is stretched within its elastic range many, many times until it finally fails. That is not shown on that simple chart but here's a sample of a fatigue limit chart. It'll get a little more complex from here.

This chart basically indicates the metal will fail given enough load change cycles even at relatively low loads.

Let's dig in a little more about this concept of preload. Picture a balance, like a scale, like the one Lady Liberty holds. Where you can put 10 lbs on the left tray and 10 lbs on the right tray so the scale would balance. So now picture there's 10 lbs on the left and on the right we put 2 lbs but the scale goes all the way left. Now we keep changing the weight from 0 to 7 lbs over and over again, but the scale never moves no matter how many times I change the weight on the right tray. It doesn't move because the weight on the right is not near enough to the weight on the left tray to move it. It may move a tiny little bit if we put 9.5lbs on it but will really move at 10lbs and move a bunch at 11 lbs. The same thing would happen if you put a spring on the left holding the tray down with 10 lbs of strain. Now imagine I put a 5 lb weight on the left and kept changing the weight on the right. Every time the right weight exceeded 5 lbs the scale would flop over and when less it would flop back. If it were a spring it would stretch and contract over and over again. If I do that a million times, the spring will break.

The turbo flange is subject to weight, inertia, thermal expansion, and vibratory loads. The bolts are sized to provide more pressure due to the tension of the stretched bolts than the total forces on the flange. A bolts stretched with tension works exactly like a very strong spring. So properly tensioned bolts do not stretch and contract repeatedly in operation. It is the heavy weight (or spring) on the left of the balance. This eliminates the fatigue problem because there are no cycles at all. If the tension pressure is less than the load pressure, there will be millions of cycles where the bolt length extends and contracts in excess of the fatigue limits.

Engineers typically limit bolt tension to 75% of the elastic limit, and limit the forces applied to 75% or 80% of the tension provided by that. So there's a bit of wiggle room. If you have a fatigue failure it is almost certainly because of insufficient (<80%) bolt tension applied when it was installed.

Again, indicated torque has very little to do with how much bolt tension is applied as errors are common. Opportunities for errors include; excessive friction, static friction measurement error, incorrect fastener material, reuse of annealed metal, or any combination.

 

Okay I'll work to make it jibe.
The spring back effect is well defined, measured, proven, analyzed ad nauseum for all materials. The fasteners you installed in your car engine live in an oil bath which is not a corrosive environment. These are high carbon steel bolts. Iron, steel and carbon steel materials all exhibit a characteristic called anelastic deformation. This deformation requires time to occur. It works to your advantage because it increases the tensile strength of the bolt in the direction of the applied stress as Carbon atoms migrate to favored locations provided by the stress. This effect is not permanent, but it is very slow to recover. The elastic and anelastic strains are both reversible which means your bolts will eventually return to their original length. This is why they didn't just stretch them out at the factory then ship them to you. If you do not perform the technique you described, the preload on the bolt will reduce over time and likely cause the bolt to suffer fatigue failure. This is known as "stress relaxation" or an "elastic aftereffect"; I prefer the prior term because the second term really should be "anelastic aftereffect".

But... the turbo bolts are not high carbon steel but rather high nickel, cobalt and chromium steel. These alloys do not have the anelastic characteristic.

I'm not certain about what temp these bolts would glow red but I'd guess about 1500F. Regardless, if you heat a hardened metal until it glows red, the atoms are definitely moving vigorously. If you allow it to cool in air, the typical result is carbon migration to the center so it'll be more malleable and softer. No matter what happens, harder or softer, the resulting material will not meet specification. Perhaps I'm not clear on what "hit with torch" means. I did assume red hot, so I'm open there for an error. These bolts tolerate high temps extremely well so anything near 1000F should have no effect.

That issue is outside the point. I was answering a question regarding the idea to pre-stretch the bolts on the bench. This will not be effective because it is the strain on the bolt while installed which must exceed the forces applied in operation that is critical. If not, fatigue failure is certain.

 

In-elastic

And the higher temps you are referring to are to do with the alloy state changes, martensitic and austentitic. With inconel/high nickel theres sure to be another word for that

Sorry, i was inexact, i thought i had said, hit with torch to 600degrees f, and i only said "hit with torch". Point being, i was thinking of essentially retempering (presumably because thats what i would assume the turbo does, running between 300 and 900 degrees).

Meh, was a thought

 

While I tend to agree with the science of the subject to answer the OP , when I went through this fiasco (actually I think it happened 5 times before I called my friend the old Cat mechanic) I read the Cat manual and followed procedure, Nevrsieze and all. I searched the internet high and low for hours in my spare time, as I prefer to solve my own problems. I had no specimen to analyze because it is a nut and bolt joint so when it comes apart, it falls out. All I could find was "use Cat bolts and then some mention of proper torqing procedures" Well, I was doing all that to the best of my ability, including using a torque adapter and the associated calculations to make sure things were even. I use a torque wrench alot more than most people I know. I was taught from a young age that as long as you are in a proven range of torque for a fastener that where you are using multiples that having them all even is just as important, or more, than an exact number. Anyways, for the OP or anyone else searching for an answer that goes a step or two beyond the obvious, see my original reply. It isn't that hard and it absolutely made the difference for me.