304 vs 316 in an exhaust manifold aplication 1

[chad86tsi]

For the record, I am NOT a metalurgist, just a hobbiest that tries to understand things with research and "good" advice.

I have been fabricating automotive turbo exhaust manifolds for some time using .065" wall 304 and 321 ,and some inconel 625 tube, and now similar models using sch 10 pipe of 304 and 316 grades.

Question is, am I really making a better product using 316 vs the 304 on the pipe aplication? the exhaust temps peak at 1800*F for periods not likely to exceed 15 seconds, but normal operational temps may be as high as 1400*F sustained. this product is subjected to vibration stresses, weight of the turbo and exhast, and the normal motion of the engine.

I can't totaly remove these stresses, so I need the manifold to be capable of handeling them. With the tubing, I've had several fail using the 304, I belive it's just too thin and the tubes reach a temp nearly that of the exhaust stream, even during the transient peaks. They generaly fail in the tubing, or near the welds, but not directly in the welded areas. It appears to be a funcion of vibration and tube movement (which I can't avoid) so I belive it's jsut weakened metal structure, and carburization, hence the use of 321 (at 2-3 times the cost), or inconel when I can afford it (aprox 10 times the cost).

I tired adding bracing, but this just bound the flanges into one position so the tube elongation just caused thermal expansion stresses instead (tube movement).

Now I'm focusing on using pipe instead of tube, it's more cost effective, easier to find, and the added thickness of the walls yileds greater strenght all by itself. I've been told by an ehaust manifold fabricator that I can go back to using 304 in this pipe aplication, but I'm leary of using it after the failed tube units I had before. Do you think I should continue to use 316 for it's margionaly better thermal properties at a cost premium of ~20%?

The pipe body will be less likely to reach the temp of the exhaust stream just due to the added thickness and poor conductivity of stanless in general, but I can see it reaching pipe tems of up to 1600*F occasionaly.

 

[EdStainless]

Chad, Email me please.
tks ed

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[LarryW1]

Would it be possible to use a flexible corrugated tube in your application? Here's a company that makes them with 321 stainless:

http://www.dmeexpansionjoints.com/exhaust-flex-connectors.htm

 

[hgo]

Chad,

Looking at your pictures. I think that you could do more for your product and riliability by improving your welding technique. You need to argon purge and use more heat and get 100% penetration. Keep your tungsten clean and make sure that you have a good shield flow, pre, post, and during your weld. Your welds look cold (width to build-up of filler metal) and appear to have little penetration. The argon internal purge will help with consistancy on the inside where you can build rough oxides that cause the joint to be brittle. This is a cheap step in the right direction. Also, .030 to .045 t-321 will work for very high pressures on your turbo for paved racing. I have not used them on off-road (dirt) racing.

 

[chad86tsi]

I use solarflux type B, it goes on as a paste on the back side of the weld (inside the tube in this case) and hardens into a glass/ceramic like substance when welded. It shields the molten weld pool on the back of the weld from any/all air, and the glass/ceramic substance supports the weld from sagging.

I understand your cirticizm, it does look like a cold weld. I have purposly added *slightly* more filler than required to the weld pool to add strenght to the welded area, just incase. I have done some failure annalisys on these welds with this technique, kinda crude but I did this by placing them in a vice and bending back and forth, tearinng(shear) and cuting. The welded area is stronger than the parent metal due to the added thinckness. The HAZ seems to be the weakest point.

The penetration is more than adequate. this is only .065" thick, and it is the thickness I have a great deal of experinace with (probably 50-60 hours continious weld time), it doesn't take much to penetrate, infact I often have to be careful not to over-penetrate.

thanks for the input, I have not had weld failures, it's mostly been parent-metal failures from inadequate metal temp range (this the purpose of this thread), and excessive warpage after repeated thermal cycles.

Thanks for your input, any/all advice is welcome.

 

[MikeHalloran]

You might try skeletonizing the head flange, by getting it cut with a CNC water jet or laser (not as expensive as doing it with a jigsaw), say into four sturdy flanges separated by slightly flexible S-shaped bars. The reduced mass should make it easier to weld the tubes in.



Mike Halloran
Pembroke Pines, FL, USA

 

[chad86tsi]

I actuay do something very similar to your suggestion, I just had not completed that manifld in the picture. I fully complete the manifold, and then cut the space between each runner on the flange with a plasma, leaving just 1/2"-3/4" un-cut. that remianing uncut portion retains the position between the runners, but the long cuts allow the ports to move a little, and return with out building up a lot of stress in the flange.

It also allows the head studs to pull it back straight when it cools. When not cut, the flange tends to bend like a banana pulling out the #1 and #4 studs. The falnge material is the cause of this, not the runners. By cutting the "webs", the flange nolonger builds up stress and it is only moved by the runners, which are far easier to bend and store less stress.

If I do the cuts at any time before completion, the subsequent welds invarriably move the flanges, stainless is pretty good about moivng things for you when heated

I just started doing this 1 1/2 years ago, and you are right, it did make a difference.

 

[legend4life]

Chad,

Could you post a picture of the skelatonized flanges or cuts you make to them?

 

[EdStainless]

Chad, Don't get carried away building the welds up. You don't want them stronger than the tubes. The welds are probably a bit stronger, so they don't need to be thicker. You really want everything to distort together. Don't make any part heavier or stiffer than is absolutely needed.

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[chad86tsi]

Here is a pipe version with the flange plasma cut. this is the "log style" one, I also make an equal lenght one using the same material and methods.

 

[legend4life]

Chad,

Thank you for the picture. What material and thickness do you utilize in the flange? Do you feel the small uncut portion is strong enough to hold it together reliably under expansion, and how long have you had these hold up under hard street usage?

 

[chad86tsi]

Some of your message is cut-off so I can't read the whole queston, I'll answer what I can read...

I have about 15 in use with this flange cut design, none have reported any problems with it. The end of the cut is clearly a stress focal point where cracking could occur, but the flange is held in place by the manofld studs, and it's not subjeted to much vibratoin at that point. the only reall stresses it sees is thermal expanson/contraction, which is limited by the clamping force of the studs. Idealy I'd drill holes at the terminus of the cuts to make a nice radiused end for stress riser reduction, but I'm lazy

It is 1018 cold rolled. I chose cold rolled as it's surface is far easier to true after fabrication. I clamp it down to a jig, and after welding there is little or no distortion, just a quick run over the belt sander to reveal highspots. If they exist, the belt takes them off quickly and easiliy.

I'd like to use stanless, but with my limited machine capacity, the flanges are 2-3 times easier to fabricate in 1018 with what tools I have. the raw material is also 2-3 times less expensive.