Tubing thickness and fatigue cracking in exhaust systems 1

[yoshimitsuspeed]

In another thread we are discussing ideal materials for exhaust systems and in that thread tubing thickness came up but a lot of responses are saying that in regards to thermal fatigue cracking thinner tubing would be better. This seems backwards to me but I want more information and I thought maybe this section may have some engineers who could go into this theory in a little more detail. Naturally I would prefer everyone just told me I am right but if I am wrong I want to have a much better understanding of why.

For any non petrolheads the exhaust gasses can range from 400C to probably at least 1000C seeing those highest temps at full throttle high RPM and those kind of temps in very high performance motors. And of course cooling to ambient every time the motor is shut off.
Cracks almost always start at weld joints and most of the time it is where the tubing hits a thicker or more rigid component like a bracket, bung, flange or collector. Some flanges also stay a good bit cooler which affect things as well. For example the flange that bolts to the head will stay much closer to the temp of the motor. Flanges downstream will heat up more but will take much longer to heat up than the tubing naturally. Add a turbo into the mix and it will be somewhere in between. The turbine side will get very hot but it has a lot of thermal mass and some heat drawn away through the turbo.

It seems to me that thicker tubing would generally be better because it will be stronger, more cross sectional cross area to spread forces over a greater area and closer in strength properties to the thicker and more rigid components it's attached to.

Here are some examples of typical failures.

https://www.miataturbo.net/attachme...73d1309839857-cracked-ebay-manifold-crack.jpg

http://img.photobucket.com/albums/v209/tomt64/Tomei010.jpg

https://goo.gl/images/ZagIrm

If you saw failures like this would you tend to want to go thinner or thicker tubing?

 

[racookpe1978]

Are you asking about internal combustion engines exhaust pipes, catalyst converter connections, or gas turbine engine exhausts? Or turbo-charger connections to the inlets and exhausts of ICE engine blocks? Or super-charger connections?

 

[yoshimitsuspeed]

I guess I forgot to specify but primarily in regards to turbo manifolds and NA headers on ICE where the heat is highest and thermal fatigue most likely. Tubular welded.

 

[Compositepro]

When temperature changes, metals expand and contract. You cannot prevent this. If you make a part "stronger" with more thickness, then the stresses increase as well and the part will still crack at its weakest point. In most cases the simpler and far more effective solution is to make the parts flexible so that the movement will not create any high stresses. For example, a flat sheet of metal will warp when heated unevenly. Corrugated sheet metal will not warp visibly because it can expand locally without creating a great deal of stress in surrounding areas.

 

[TGS4]

To expand upon what Compositepro said, thermal stresses are a result of compatibility, not equilibrium. So, making things more flexible (as permitted by your other loads) will reduce the thermal stresses. In the case of thermal stresses, the old adage of "when in doubt, make it stout", is the exact opposite thing to do.

 

[MikeHalloran]

One obvious next step is to put a short bellows in each pipe.
There are downsides;
extra pressure drop for every convolution,
and the best bellows last only a year or so in exhaust systems,
and you have to support both ends of the bellows.

Obvious next step after that is to make all the pipes longer and thinner.
Downside is some extra pressure drop, and finding space for all the extra pipe.

Everything is a compromise.



Mike Halloran
Pembroke Pines, FL, USA

 

[yoshimitsuspeed]

Unfortunately with something like a turbo manifold making things flexible isn't an option.You are constrained by the head flange connecting multiple cylinders at one end and the collector and flange at the other. Both fairly rigid and both heating and cooling at much different rates than the tubing. And the thinner the tubing the more extreme the temp differentials between those components would be.
If you have 8" of curved runner between the head flange and the collector heating and cooling at much faster rates than the flanges thinner pipe will flex and bend more wheras thicker pipe would stay closer to the temp of the flanges and also impart more movement into the thicker stronger components like the collector and move the whole assembly more instead of the tubing flexing more right at the weld joints right?

I see how thinner tubing would put less stress on an area but it also has less cross sectional area and therefore less strength to resist that stress. So I am still failing to grasp how thinner and weaker would be better especially when welded to much thicker more rigid components.

 

[racookpe1978]

Cheer up. Gas turbines have been trying to solve this problem since the early Pegasus VTOL aircraft of the mid-50's, the ME262 and Gloster Meteor fighters, and the first Pratt and Whitney gas turbines. Today's 250 Megawatt GT power plants still are fighting the problem of a rigid flange (heavy, fixed in position to a moving/vibrating engine) is being forced to connect to a far lighter, very flexible exhaust pipe that expands and contracts significantly more than the heavyweight GT compressor body and and exhaust turbine casing.

Try these:
Separate the functions: Use an interior flow separator and an exterior (potentially) convoluted pressure barrier around that flow barrier.
Use as lightweight and as flexible a pressure barrier as possible so the heat stress loads as it expands and contracts are as low as possible.
Avoid stiff, straight sections. Let the tube flex and bend.
Try to avoid sudden changes in cross-section areas - the heavy, stiff sections will tear away from the lighter weight more flexible sections.

 

[yoshimitsuspeed]

Mike I was thinking about bellows when compositepro mentioned Corrugated sheet metal but like you said bellows don't seem to last very long and are pretty guaranteed to be the place that fails if there is a failure which further makes me question the theory that thinner and more flexible is better.
When you say making pipes longer and thinner, I can see how on longer piping like say a long runner header having thinner tubing being more flexible could allow movement over a long enough section of tubing could allow stress to be applied over enough area that it would be less likely to over stress a certain area. But I still can't help but feel like there would be advantages to thicker material like it staying a more consistent temperature across the whole length instead of being much hotter further from flanges or changing in temp much more rapidly. For example it's pretty rare for full throttle acceleration to last more than 20 seconds and usually far less than that. So on your more typical say 10 second full throttle pull if the center section of thin wall tubing got up to 700C and thick wall tubing got up to 500C while the flanges remained closer to say 200C wouldn't the latter be significantly better?
Then as Racookpe mentioned "Avoid sudden changes in cross section areas. Wouldn't it be better to keep that change as small as possible? Wouldn't it be better to go from a 3/8 flange to 1/8 tubing instead of .040 tubing?
Maybe if I present more specific examples you guys can help me understand better where stresses would be applied and how thinner tubing would be better.

http://cdn3.volusion.com/pa9ja.37rfp/v/vspfiles/photos/1046002-2.jpg?1486109685

Starting with a long runner header. Now for typical fatigue anchoring this header firmly at the collector would reduce vibration and forces from the exhaust system moving it around. But anchoring it hard would increase thermal stress and fatigue.
So for example 1 let's say that the header is not anchored but able to float freely, or maybe anchored on a rubber mount or bowed metal allowing it to move pretty freely.
Thicker tubing would bend and flex less. It would move the collector end more and transfer more movement into the rest of the exhaust system. A downstream flex pipe, rubber exhaust mounts etc. But the manifold would stay cooler and a more consistent temp. So wouldn't thicker tubing apply less overall stress to the header and more movement in the rest of the exhaust system?

Now let's say the header is mounted hard to the block. I always try to avoid this. I like to use a bowed thin hangar that allows movement, but let's just say it's hard mounted to the block. In a case like this I could see how the tubing being able to bend more freely across the entire length would put less stress on the header flange and collector but there is also less cross sectional area where they meet, a much more drastic transition in area from the pipe to the weld to the flange, and the tubing would be changing temp much more and much more rapidly. Thicker pipe would stay cooler and more even in temp, and being stiffer would transfer more force into the collector, flanges and other components that can handle more force. If we think in terms of localized stress and let's say the head flange is 200C and the thin wall tubing 20mm away is 500C and 40mm away 700C vs thicker material being 300C and 500C shouldn't that help reduce localized stresses near the welds?

So especially if it's allowed to flex and move a little at one end I still can't see how the thicker tubing would be harmful.

Next if longer more flexible tubing is actually better if you have to go short is there a point at which thicker would be better trying to get the whole thing to move as a single body?
This is our turbo manifold.

https://scontent.fapa1-2.fna.fbcdn....=9310bfd596a9ad02690d9664067eaae7&oe=5AA7F784

In our application longer runners are not possible. The lengths of tubing between flanges is short and the length of tubing between welds is extremely short. We have been making them for years without one reported failure. I use 11 gauge tubing. I like the fact that it's easy to weld and also easy to make welds that are very close to the cross sectional area of the tubing as opposed to big high welds which I always felt would be more likely to create a failure point. When I am making the downpipe I make a flexible hangar mount to the block that helps with some of the weight, vibration and exhaust system flexing but I know some customers have run them for years with only the flexible chassis mount supporting the rest of the exhaust system and the turbo hanging off the manifold with no issues. Over those years I have seen hundreds of pictures of other thin wall manifolds cracking. I don't know about the manufacturing of all of them. Many and most common are the cheap Chinese manifolds and such which could be more of an issue with metalurgy or QC or something but any way you cut it in my experience thin wall manifolds have a much higher reputation of cracking just like in the pics above.

I am also curious how this ties into cast manifolds. Naturally the cast iron or steel will behave a little different and due to it's nature I'm sure thin wall cast would be a horrible idea but thick wall cast manifolds usually last hundreds of thousands of miles before cracking. If thinner is better then why do some thin wall welded manifolds fail after 10k miles while a cast manifold may just outlive the car?

 

[racookpe1978]

You have found the problem.

A thick-wall combustion chamber casting IS a pressure vessel subject to thousands of high-pressure cycles each minute as each combustion chamber is vented, the valve(s) closed, repressurized by the upward-moving piston, rapidly heated and suddenly over-pressurized by the fuel-air mixture and then by the combustion flame front, then vented again when the valve opens. The heating cycle is rapid, but not instantaneous, as the engine starts and the cooling water-glycol mixture and oil begins flowing. These dynamic pressure forces are much, much greater than the bolted stress-strains at the connections to the (quickly-heated, always-hot-under-operation, but not high-pressure) exhaust manifolds and pipes.

So, the heating and vibration cyclical forces of the combustion casting are near-trivial compared to the rapid pressure cycles inside the piston chambers.

If the car's (gas turbine's) exhaust could be economically built "battleship tough, battleship thick-walled" then those components (and the supercharger,turbo-charger components) would also be strong enough not to crack. The steel alloy would face the same forces, but they would be "smoothed" over a very thick wall, and the localized strains and stresses would be low enough so no cracking would develop. If enough space permitted, the curves and gradual transitions from flange to flange would allow movement as well.

Weight, cost, and space prohibit a battleship-type exhaust wall thickness. (Heck, even in tank and battleship designs, the designers are always fighting weight, volume, casting cost, fabrication cost, crane limits, ground traction and ground penetration forces in mud and grass, flow resistance in water, payload capacity, fuel capacity, etc. "Ain't nuthin' in life be free" even when the government is paying. )

So a very-thin-walled exhaust manifold and exhaust pipe is always chosen, but that must be bolted to the far more rigid combustion (piston) casting. Hence, my comment about the transition flange point: You're fighting the need for a flange thickness for enough bolted strength - which, if great enough to clamp the exhaust manifold/exhaust pipe firmly will restrain the flexible pipe itself and cause localized stress at the transition - or, if flexible enough to allow the pipe to move, will allow the bolted-gasket connection to leak and separate.

In piping, we'd use a raised face weld neck flange and butt weld joint to make a smoother transition for high pressure steam pipe connections that are subjected to thermal stress changes, pressure stress changes, and long-life fatigue stresses. But that space, weight and welding fabrication expense can't be tolerated in an economical car design. Today's 3D casting techniques could improve things by checking the 3D stress-strain model for high stress points, designing them out with transitions or mechanical slip connections to the piping network, then fabricating the intricate net of final parts by additive manufacturing. But figure the part-by-part cost and time, and you're back into race car technology and cost. Where long-time fatigue can be ignored if the exhaust only needs to last 110% of one race duration.

Then you throw the exhaust manifold away and make a new one.

 

[MikeHalloran]

I learned about casting from a Boy Scout merit badge booklet.
They cover a lot of territory in very little space.
Much of it is instructions to patternmakers.

You won't find sharp internal corners on a professionally made casting.
Also on your turbo header, but the radius and its tangent transitions are a product of the person doing the welding, so that skill is critical to your success.
I also >maybe< see a wall thickness transition at the engine flange to tube joint, where the first section of tube welded to the flange per se is/may be of thicker wall than more distal tube sections.
... or maybe it's just an artifact of the photo, but it's a possibility to consider.
Of course, wall thickness transitions are easily arranged in a casting.

For the long runner header, thin and straight-ish tubes would be a bad idea, as they would fight each other for dominance over temperature changes. Gently curved tubes could co-operate to some extent, and hence be made somewhat thinner without issue.

In Phil Irving's book about motorcycle engineering, he devotes quite a lot of space explaining how the Greeves motorcycle evolved to have a 'frontbone' in its frame, and how it could have been better, if only the designers understood 'stiffness' in a quantitative way, instead of as an imaginary concept. Simply put, for a multi-element structure like a header, if you analyze each tube as if it were independent, e.g. apply an arbitrary load and calculate the resulting deflection, and compare the spring rates of the individual members, then you will find that if one element is vastly stiffer than its mates, it will carry most of whatever load is applied.



Mike Halloran
Pembroke Pines, FL, USA

 

[yoshimitsuspeed]

All of the tubing is the same material. The bends do pull and stretch the metal enough on this thick wall tubing that especially on the OD a piece cut in the middle of the bend will not line up perfectly with a straight piece of tube or another cut bend at a different angle.

I am curious if the thinner is always better crowd is still of the same opinion for all three of the examples.
I would also like to hear more from them in regards to attaching thin wall material to thicker flanges and the concepts of heating and cooling that I mentioned.
If I am understanding him correctly it seems like racookpe1978 is at least partially onboard with thicker is better if done right and in a situation like this with so many other limiting factors, space, budget and so on.

As far as our turbo manifold I would be hesitant to change things because 1 it works and we haven't had any issues with failures and 2 even at a slight increase in weight and cost the time it saves and ease of welding it quickly makes it hard to consider changing but is anyone here looking at that specific design and confident that thinner tubing would in fact be better, last longer or be more resistant to cracking?

 

[davefitz]

The cracking at the location of sharp thickness variation can be addressed by revising the the gradient of thickness such that the slope in thickness variation is less than 18 degrees. Additionally, welds usually have an inherent crack , caused by the welder's removing the welding rod from the workpiece. Such cracks have a stress concentration factor of about 5 , and additional variations in ductility.

There is a simple correlation between wall thickness, rate of change of temperature, material thermal diffusivity, and the imposed thermal stress. Refer to EN 12952-3 annex C. Simply put, the thermal stress will be directly proportional to the square of the wall thickness.

"...when logic, and proportion, have fallen, sloppy dead..." Grace Slick