Cylinder head clamping and bore distortion

[Speedy Squirrel]

I read through the cylinder head stud discussion and everyone seemed to miss the issue of bore distortion.

Most modern water cooled engines use aluminum blocks and heads. The head is supported by the cylinder AND the block walls. You can see how this works in the water cooled Porsche engine in the photo below. The cylinders are free standing (open deck), but there are also block walls to support the cylinder head. Further, being water cooled, the block, heads, cylinder and bolts are all maintained at nearly the same temperature by the cooling fluid.



Our immortal beloved air cooled engine on the other hand carries the full clamp load on the cylinders.



Additionally, the distribution of the cooling air is such that the side facing the fan (intake side) gets more direct airflow than the exhaust side, so the temperatures are different across the various components.

A water cooled engine can handle very high clamp load because of the block support. It uses a multi-layer steel head gasket and torque to yield head bolts (not reusable).
The classic air cooled engine has only the cylinder to support the clamp load. It occasionally uses some type of sealing ring/gasket, and the head studs are reusable.
These differences change the requirements for the head fasteners.

The diagram below shows a typical stress-strain curve. Stress is proportional to tension, and strain is proportional to stretch. Young’s modulus is a measure of elastic behavior. In this region the fastener will return to its original length when un-torqued. The yield point is where the fastener takes a 0.2% stretch. The ultimate tensile strength (UTS) is the maximum tension that the fastener can maintain. After this point, the fastener will keep stretching and stretching, it will noticeably “neck down” and finally break. If you have ever over-torqued a bolt, this is where it starts to get easier and easier to torque, and then breaks.



For the water cooled engine, the structure supports enough clamping force that the head is always clamped firmly to the block and cylinder. It can never be allowed to separate under any condition. If it does, the coolant leaks out. To get the most out of the fastener the torque turn (torque to yield) method is used. This method takes the stress (tensile load) to the blue dot. You have to be careful though. If you go over the UTS it will fail. For this reason, the final torque is given as a number of degrees of turn in order to eliminate the uncertainty of thread and bolt or nut friction associated with using torque. Because we have moved out of the elastic region (over the yield strength point) the bolt is permanently stretched and cannot be reused.
Water cooled engine cylinders deform under from the clamping load. One method of correcting this deformation is to finish the cylinders with a simulated clamp load (torque plate).
Our air cooled engine, with only the cylinder as the structure also can’t take that kind of load without deforming. We have to be careful with how much load we apply to the cylinder in order to keep it from bending out of shape. The sketch below shows in a greatly exaggerated way how our cylinders deform when clamped. The point is, we can’t just increase our load higher and higher without consequences.



To deal with this problem Porsche only tightened the fastener in the elastic range (red dot shown above, below the yield strength point). The expansion of the cylinder supplies the rest of the clamping load, though the stud still remains in the elastic range. This is not a perfect arrangement, as the cylinder at a lower temperature has less clamping load than at a higher temperature. This is where the various sealing concepts came into play, in an effort to stop oil leakage and combustion gas leakage when the cylinders are not very hot
This whole scheme first fell apart with the magnesium block. On shutdown, the steel studs do not expand as much as the aluminum cylinders. The expansion of the cylinders added too much pulling force on the hot magnesium case and the studs gradually pulled out.
The engine bore size was also getting bigger and bigger. This caused the cylinder walls to get thinner and thinner. At 95mm the bores were getting too deformed with steel studs.
The solution to both of these problems was to find a material for the fastener with a thermal expansion coefficient that more closely matched aluminum.

Dilavar fit that requirment reaspnably well, but had its famous problems of stress corrosion failure. Various coatings were tried, with uneven results. Finally, they decided to roll the entire surface of the stud, and used a rolled thread form. Here is a great animation of that process.

https://youtu.be/rwArBBcUNr4

Rolling the thread on the surface crushes all the exposed grain boundaries closed, which prevents salt from getting into the grain structure and corroding the alloy.

This has proven to be outstandingly successful. Now you can have the benefit of low cylinder distortion, and reliability, even at high power levels.

[jmc1313]

I must say that was very interesting. Thanks for posting it.
John

[winders]

Originally Posted by Speedy Squirrel

This has proven to be outstandingly successful. Now you can have the benefit of low cylinder distortion, and reliability, even at high power levels.

This post eloquently explains why so many builders of high performance air-cooled Porsche engines use the 993 Twin Turbo head studs! They work and work well!

[KNIGHTRACE]

Now that the subject is brought up, a lot of people use a straight thread tap to clean up their threads in the case and loose a significant amount of strength. The factory uses rolled threads for strength in the later motors maybe all I am not sure about 2.7's and smaller. I made a rolled thread cleaner that keeps the studs tight on installation. It is a simple motor with a lot of complicated science.

[Henry Schmidt]

Originally Posted by KNIGHTRACE

Now that the subject is brought up, a lot of people use a straight thread tap to clean up their threads in the case and loose a significant amount of strength. The factory uses rolled threads for strength in the later motors maybe all I am not sure about 2.7's and smaller. I made a rolled thread cleaner that keeps the studs tight on installation. It is a simple motor with a lot of complicated science.

The recommended thread forming tap for this application is (1.50x10 PD-6 6-N).
R&N part number is 62653.



As for the theoretical information posted, the engineering is correct. The application is flawed.
The data does not consider real world applications.
It misses perhaps the most important modifier to the data. That being, inconsistent temperatures applied to each individual stud.
Every stud on a 911 based engine functions in a different environment. The more a material is affected by heat, the greater the inconsistency. If the temperature is different, the expansion is different. Different expansion rates on the same cylinder will produce inconsistent clamping forces. The next question: "what is the correct clamping force"?
Another question that can only be answered by real world experience.
My experience tells me that the clamping force of the 993 Dilivar stud is sub-par.
Every engine we disassemble that has 993 Dilivar show signs of cylinder to head movement.
The solution, a stud that offers controlled expansion greater that common steel yet a clamping consistency superior to that of Dilivar.

[Harpo]

Thanks Speedy Squirrel good explanation. I design bolted joints for a living and the automotive industry had lots of problems with SCC, HE and LME in the 80's. Ultimately we limited our HRC to 38 and now only use dip spin zinc rich coatings that give us 1000 hours or better of corrosion resistance. While most of these bolted joint applications are straight torque specifications many are torque + angle. None are torque to yield

I agree that the rolling may offer some resistance to premature failures listed above but I have never seen enough data to fully support that theory. Another approach is lead patenting which is used by the bridge industry and appears to also be helpful. Lastly are austempered (12.8 & 14.8) fasteners and I have been very successful implementing them over the past decade with no known failures that I'm aware of.

[3rd_gear_Ted]

Does torqueing the camshaft carrier to the heads first allow the clamping of the heads to the cylinders in a more even manner ???
Or should the carrier be installed after the heads???

The only dumb question is the one not asked, so I'm asking

[dtxscott]

Originally Posted by 3rd_gear_Ted

Does torqueing the camshaft carrier to the heads first allow the clamping of the heads to the cylinders in a more even manner ???
Or should the carrier be installed after the heads???

The only dumb question is the one not asked, so I'm asking

Check out Henry's comment in post #2 here based on my noob question.

https://forums.pelicanparts.com/911-engine-rebuilding-forum/1083982-questions-cylinder-base-gasket-assembly.html

[Speedy Squirrel]

Originally Posted by 3rd_gear_Ted

Does torqueing the camshaft carrier to the heads first allow the clamping of the heads to the cylinders in a more even manner ???
Or should the carrier be installed after the heads???

The only dumb question is the one not asked, so I'm asking

Not dumb. The cam carrier fits to the heads with the alignment dowels. Tightening it down all the way with the bolts, or just finger tight wouldn’t make any difference on the head being torqued evenly, in my opinion. I don’t think the carrier is stiff enough or attached well enough to the heads to change the torque appreciably.

[Speedy Squirrel]

I see you are local to the D. Maybe we can meet up at an event. Not sure if the Belle Isle races are happening this summer. Probably not.

Originally Posted by Harpo

Thanks Speedy Squirrel good explanation. I design bolted joints for a living and the automotive industry had lots of problems with SCC, HE and LME in the 80's. Ultimately we limited our HRC to 38 and now only use dip spin zinc rich coatings that give us 1000 hours or better of corrosion resistance. While most of these bolted joint applications are straight torque specifications many are torque + angle. None are torque to yield

I agree that the rolling may offer some resistance to premature failures listed above but I have never seen enough data to fully support that theory. Another approach is lead patenting which is used by the bridge industry and appears to also be helpful. Lastly are austempered (12.8 & 14.8) fasteners and I have been very successful implementing them over the past decade with no known failures that I'm aware of.

[Speedy Squirrel]

I know exactly how much clamping force is required

Calculate BMEP from target power
Calculate Wiebe function from BMEP
Calculate peak cylinder pressure from Wiebe function
Multiple peak pressure by cylinder head area to get force.
Multiple force by 1.5 for dynamic factor and divide by 4 to get required force per stud.

You are the one who doesn’t know, so you overclamp, way over, which is not good.

Originally Posted by Henry Schmidt

The recommended thread forming tap for this application is (1.50x10 PD-6 6-N).
R&N part number is 62653.



As for the theoretical information posted, the engineering is correct. The application is flawed.
The data does not consider real world applications.
It misses perhaps the most important modifier to the data. That being, inconsistent temperatures applied to each individual stud.
Every stud on a 911 based engine functions in a different environment. The more a material is affected by heat, the greater the inconsistency. If the temperature is different, the expansion is different. Different expansion rates on the same cylinder will produce inconsistent clamping forces. The next question: "what is the correct clamping force"?
Another question that can only be answered by real world experience.
My experience tells me that the clamping force of the 993 Dilivar stud is sub-par.
Every engine we disassemble that has 993 Dilivar show signs of cylinder to head movement.
The solution, a stud that offers controlled expansion greater that common steel yet a clamping consistency superior to that of Dilivar.

[Henry Schmidt]

Originally Posted by Speedy Squirrel

I know exactly how much clamping force is required

Calculate BMEP from target power
Calculate Wiebe function from BMEP
Calculate peak cylinder pressure from Wiebe function
Multiple peak pressure by cylinder head area to get force.
Multiple force by 1.5 for dynamic factor and divide by 4 to get required force per stud.

You are the one who doesn’t know, so you overclamp, way over, which is not good.

You're right. So do us all a favor and fill in the equation......I'll wait........

What is Dilavar? What is the chemical makeup of Dilavar?
Not even Mezger knew the "true" expansion rate.
Calculating expansion rate without knowing the chemical properties is just guessing.
Tell me oh squirrely one: what is the actual expansion rate of Dilavar.
Even Mezger is quoted as saying "Dilavar is a new steel alloy—strong yet flexible. Dilavar studs are theoretically able to keep pace with the longitudinal “growth” of the engine”.
You state as a fact that I'm "overclamping" something?. By how much?
What is the proper clamping force (real number) for every Porsche air cooled cylinder.
At what temperature does Dilavar function best?
Mezger stated that the studs in his 917 had to stay hot to operate properly. That is why they coated the 917 Dilavar with resin and fiberglass. Studs pulling threads in 917 cases @ 550+ horse power, was the reason for Dilavar. I have never seen a statement by any Porsche engineer stating cylinder distortion as a reason for Dilavar.
If there is a popper temperature, how can you claim with any certainty that all the studs in a 911 engine operate at the same temperature? answer, they don't.
If the studs don't heat evenly, clamping force will be uneven. What do you think uneven clamping forces will do to cylinder distortion?

[Speedy Squirrel]

Bring a snack. It’s going to be a long wait. You need to do your own engineering.

Originally Posted by Henry Schmidt

You're right. So do us all a favor and fill in the equation......I'll wait........

[Jeff Alton]

Speedy,

In your original post, you forgot to mention the watercooled motors in your pictures have head bolts that thread into an iron crank girdle that is held inside the aluminum block... Not sure if that affects your thoughts or theories..... They are open deck, unlike the later 09 and on 9A1 motors which are closed deck. Either way, we have yet to see a Porsche watercooled engine with a multilayer head gasket fail at the gasket. Is it the gasket, the stud or a good combination of chosen materials.... ? These motors fail other cylinder rated issues (bore scoring) which may of may not be attributable to temp related issues...

Also, as someone who is clearly knowledgeable in engine dynamics and engineering, whey do you use a Pseudonym?

Cheers

[Henry Schmidt]

Originally Posted by Speedy Squirrel

Bring a snack. It’s going to be a long wait. You need to do your own engineering.

We get it, you have all the answers but you're keeping them a secret....

I'm not an engineer. I'm a Porsche 911 engine builder with almost four decades of experience.
What I do is evaluate real world scenarios and to do so you must ask the pertinent questions. I notice you avoid these questions like a child avoid logic.
It's not your fault, you just don't know what you just don't know.

Insignificant amounts of cylinder distortion happens no matter what head stud is used but that is no reason to surrender the cylinder to head joint consistency.
That is why Aaron alluded to the fact that Mahle believes .004" bore variance is within tolerance.
Years ago, I found a new 100mm Mahle cylinder that was .004" out of round. When I contacted Mahle Motorsport the engineer I spoke with said that was with tolerance. When I voice my concerns he suggested I "prang" the cylinder. "What?" ....His response "just hit it with a mallet". I sent the set back for an exchange.

Even Porsche race engineers understood that Dilavar could not hold the heads in place under grueling racing scenarios. That's why Ni-Resist ring joints (flame rings) made their appearance on almost every air cooled turbos. Flame rings were intended to maintain the head to cylinder joint when the Dilavar head studs failed to provide adequate clamping force.

[winders]

Originally Posted by Henry Schmidt

Even Porsche race engineers understood that Dilavar could not hold the heads in place under grueling racing scenarios. That's why Ni-Resist ring joints (flame rings) made their appearance on almost every air cooled turbos. Flame rings were intended to maintain the head to cylinder joint when the Dilavar head studs failed to provide adequate clamping force.

Uhm, I am not an engineer but I am pretty sure that the Ni-Resists rings don't do any good if the adequate clamping force is not maintained.....

[Ollies930]

Based on my engine analyzer program the peak cylinder pressure of a 3.2l 935 at 1.35 bar would happen around 6500rpm and would be approx 2056psi. With a 95mm bore, the effective pressure on the studs would therefore be 11sqin(95mmbore) x 2056psi x 1.5 / 4 = 8481 lbs clamping force per stud. To achieve 8481 lbs clamping force on a 10mm stud, it would need to be torqued to approx 45 ftlbs, using ARP lubricant. Since these cars had the dial of death, I am sure that they would see even more boost on occassion, which would require even more clamping force. I wonder how much bore distortion that would cause. Probably not nearly as much as we think.

BTW, who thinks that these were Dilavar studs?
ROFL

[reclino]

Point of fact. The chemistry of a metal and to a minor deg the heat treatment causes the characteristic CTE of that metal. But CTE is measured in a lab and is "easy" to measure, so the CTE of Dilavar can be known. In most metals CTE is relatively linear, but if the application is critical, simple measurements in the temp range of interest gives you a result. It's not magic or a mystery, it's just simple measurements. What temp is a head stud at in any.postion, at any RPM, at any airflow, at any engine load, or ambient temp, this is where having a high expansion stud would add yet another variable to the calculations, so many variables that I would agree with the logic behind Henry's argument.... You could calculate till the cows come me home and it's not going to give an full answer. Testing on an instrumented engine, or years of experience and trial and error are they way to solve this problem.

On the recommendation of a very well known engine builder, I bought studs from Henry.
The SHOCKING truth here is this same engine builder recommend 993 studs to winders..... There is often more than one solution.

[Henry Schmidt]

Originally Posted by winders

Uhm, I am not an engineer but I am pretty sure that the Ni-Resists rings don't do any good if the adequate clamping force is not maintained.....

Of course they do. Anyone who has disassembled a performance turbo that ran increased boost with Ni-Resist rings has seen the effects. The cylinder sealing surface between the bore and the inside the ring is quite often coated with carbon. The only way for this to happen is for the head to lift.
Carbon inside the ring with little or no carbon outside the ring indicates the the flame ring is functioning as designed. It also shows that the head stud is not supplying adequate clamping force.
As comical as the claim about rolled threads reducing surface failure, the real reason for rolled all-thread is increased shaft diameter economically. This increased diameter (calculated at 9.1mm vs 7.6mm) increases clamping force over the original Dilatard stud.

[dannobee]

2056 psi might be a bit optimistic for one of our old turds. The Honda 1.5L F-1 engine with 2.5 bar boost had measured peak pressures of just under 167 bar or 2422 psi. And that's with 20 years of technology over the 935.

But if you want to really spike up the cylinder pressure to REALLY high pressures, let it detonate. Over 6000 psi aren't uncommon. I'd have to believe that that's THE main reason that heads rattle around on the big boosted Porsche's.

On another note, race diesels can run incredibly high cylinder pressures. With compound turbos with inter and after coolers, 200psi of boost is common. This is what happens when head studs become the least of your problems. Excuse the language.

https://www.youtube.com/watch?v=eOQBPwKlsoY