S54 Header Design talk

[George Hill]

I'm not 100% sure on E46 fit yet. I need to look at that closely next. I have an e46 chassis here I can scan but no way to correctly get the S54 placed into the right spot (its a 330i with an m54) Bry5on shared a scan with me a while ago that I think includes some E46 chassis in it. I might be able to align that with my E46 scan and get the engine placed correctly. Also I was under an E46 looking at space last week and I think there is actually less room on the E46 than the E36 which I was surprised by because the E46 chassis is wider. If it's just the engine support arm causing an issue then the designed based on the p54 has plenty of room to move primary number 4 away from the arm. Not so much on the other design without really throwing off the lengths.

I believe the part numbers for the front right motor mount arms and the crossmembers are both identical between the S54 and M54 cars.

• You can definitely use the M54 engine support arm to line up the S54 engine as they are the same on the RH side and dimensionally the same on the LH (S54 has a mount for the P/S res, plenum and SMG mount).
• The engine bays are fairly different you definitely have way less room in the E46 vs the E36 in the area where you #4 & #6 are. The Kromers will absolutely not fit as is in an E46 without cutting sheet metal, let alone the issue of the A/C lines in that space as well.
• On the E46 pretty much anything 1" or so vertically of the inside frame rail is off limits for a street car. So on your Kromer version your height is likely fine, but I would wonder about the width of #4 specifically.
• The engine support arm is massively larger on the E46, which is why I have a custom arm for the Kromers. Additionally it places the engine in a longitudinally different spot that the E36. *You can use E36 arms on an E46 chassis but it moves the engine back 1" (IIRC) and vice versa)

As far as how close to equal goes. This is mm of delta from the average for each of the two designs. P54 style is the red line and kromer kraftish layout is the blue line. Right now its at about a 50mm spread which I think is pretty good

I’m targeting +/-<3 mm, so +/-25 seems pretty bad! What mean length are you going for, and have you settled on a half-way step or something like a 2/3, etc? I was looking at the T50 and it appears the step is super late, more than 2/3 in.

I don't have any science to back it up, but I agree with Bry5on that a 50mm spread seems huge especially given all the work to get to this point plus then having a final product that can be mass produced. On the other end 3mm is wild, lol.

[3staxontheradio]

Getting them exactly equal and keeping the merges in line becomes increasingly difficult. If your don't care about merges being in line then you are basically just needing to make up for the differrence in 3 port positions. If the merges are in line with each other the primaries have to find the full length of 6 ports.

Basically primary #1 has to be as straight and direct as possible to keep its length as short as possible and since the merge is further back it becomes the outlier. I can probably bring it down by 10-20mm by tweaking the primary to make sure it's as straight as possible and also adjusting the position and angle of the merge. I could also scoot the merge back 10mm move some of the inequality into the collector lengths.

For perspective I measured 80mm of spread In the kromer Kraft and way more In every other header.

But also: my understanding is that geometric length at the centerline is not 1:1 with what will happen with pulse timing through bends. #1 being the straightest but also the longest ought to be less significant than if it was twisty.

[George Hill]

These have a 50mm (or so) spread and I was really trying to figure out how to tighten them up without making crazy bends. Do you have any data showing affects of different length primaries, cylinder to cylinder? I'd love to see something like that vs just what "seems" best to my brain with limited knowledge in the science of it.

[Bry5on]

You should model the factory headers, they’re within about +/-1-2mm from my models.

I don’t understand how pulse timing should vary much due to the bends. If the density in the pipe is largely constant and you’ve got smooth bends with a short wavelength (effectively a pulse) and relatively large pipe, then your wave propagation is just a function of temperature in the pipe varying the speed of sound. Maybe I’m just dense here but it seems that this is likely to be a third or fourth order effect, not a first order.

[Bry5on]

These have a 50mm (or so) spread and I was really trying to figure out how to tighten them up without making crazy bends. Do you have any data showing affects of different length primaries, cylinder to cylinder? I'd love to see something like that vs just what "seems" best to my brain with limited knowledge in the science of it.

It won’t sound as smooth! Maybe I’m over indexing on sound here.

[George Hill]

It won’t sound as smooth! Maybe I’m over indexing on sound here.

Sound is the only reason I put the Kromers on my car, because they destroyed the torque curve, lol. ***Edit, so sound is very much a driving force here.

But as I have learned more, I think most of the sound of them comes from the megaphone

[digger]

These have a 50mm (or so) spread and I was really trying to figure out how to tighten them up without making crazy bends. Do you have any data showing affects of different length primaries, cylinder to cylinder? I'd love to see something like that vs just what "seems" best to my brain with limited knowledge in the science of it.

posted earlier but

This compares the SSV1 with its variable primaries to the same pipe sizes but made equal length (lengths are equal to the shortest of the primaries and the longest)


SS lengths below

https://nam3forum.com/forums/forum/m...601#post326601

i wouldn't get hung up on equal length, about all you will achieve is that they will be all precisely off by the same amount. It behaves like the average length and this engine is not some narrow 1500 wide rpm band drag engine where it is possible to work out the optimum within a small tolerance. If it requires putting an additional bend or convolution or something to get 2" more centreline length then it might hurts pumping loss and attenuates the wave more than it gains. Like everything there isn't optimum anything as there isn't one criteria so tradeoffs to be made

[Bry5on]

Here's a spline set that's straight into a merge and 580.5 +/-.4mm in length - beat my target
**note that these are still 37mm ID

[tlow98]

Lovely progress!

[Bry5on]

Easy to add steps too. Pictured here is a +4mm ID step at 250mm, within a few millimeters of the OEM header length. Lots of room to optimize bend radii and everything right now is already less tight than the factory headers. Good practice if nothing else.

[Sharocks]

This is an interesting engineering exercise but I do believe the solutions that exist for this chassis at this moment in time, exist for a reason.

V1's have been tried and true and don't cost 5k.

It would be cool to see something 3D metal printed and run to see how it performs but the amount that would cost vs. how many people would actually buy it makes this a very bad business idea lmao

[George Hill]

posted earlier but

Sorry, I should have clarified, I meant some real world back to back testing or something along those lines. I appreciate the flow graph you posted, but I was wondering what changes in a dynamic engine running situation.

[Obioban]

This is an interesting engineering exercise but I do believe the solutions that exist for this chassis at this moment in time, exist for a reason.

V1's have been tried and true and don't cost 5k.

It would be cool to see something 3D metal printed and run to see how it performs but the amount that would cost vs. how many people would actually buy it makes this a very bad business idea lmao

I suspect there will be a lot of V1s for sale when this project is concluded-- including mine.

[George Hill]

i wouldn't get hung up on equal length, about all you will achieve is that they will be all precisely off by the same amount. It behaves like the average length and this engine is not some narrow 1500 wide rpm band drag engine where it is possible to work out the optimum within a small tolerance. If it requires putting an additional bend or convolution or something to get 2" more centreline length then it might hurts pumping loss and attenuates the wave more than it gains. Like everything there isn't optimum anything as there isn't one criteria so tradeoffs to be made

And that's a totally fair thought process too, like most on this thread/forum we like to overthink things.

I do appreciate the idea that if they are all the same they are optimized for a narrower RPM band vs being irregular in length which then makes individual cylinders optimized for different RPMs. I wonder how that affects individual cylinder tuning?

With that in mind in your opinion how much would you say is an acceptable length difference?

[3staxontheradio]

Easy to add steps too. Pictured here is a +4mm ID step at 250mm, within a few millimeters of the OEM header length. Lots of room to optimize bend radii and everything right now is already less tight than the factory headers. Good practice if nothing else.

That looks great but it's already longer than the target. Are you able to pull out 100mm and keep the equal lengths? I think we should design shorter so we can easily add length and tune it as needed. It's very easy to continue bundled parallel primaries.

Also a few small things: you probably need to space your primaries in their cluster more so that you can get a 15 degree merge angle. I think tight now your merge would end up very narrow. Maybe I'm just visualizing that wrong and you am still get a clean 15 degree merge?
I have been designing around a 50mm OD to give room for different engine configurations. Cams, stroker kits, etc. just making extra space so it's as flexible as possible.

I did another revision on mine before bed last night and I got the delta much smaller but cylinder 1 is still the issue. It's down to around 25mm total but the merges are about 25mm out of sync now. I'm also not really sweating it because I don't think you will be able to hear that difference.