Announcement

Collapse
No announcement yet.

S54 Header Design talk

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Ok, here's what AI did for the longer 9* collector - a little funky and it also extends the runner by 20mm to ~610mm
    Click image for larger version  Name:	Screenshot 2026-11-12 at 10.28.00 PM.png Views:	0 Size:	849.2 KB ID:	615269
    Click image for larger version

Name:	Screenshot 2026-11-12 at 10.37.24 PM.png
Views:	141
Size:	1.09 MB
ID:	615273

    And one of the odder steps that's mid-bend:
    Click image for larger version  Name:	Screenshot 2026-11-12 at 10.25.10 PM.png Views:	0 Size:	747.7 KB ID:	615270

    Getting there:
    Click image for larger version  Name:	Screenshot 2026-11-12 at 10.31.42 PM.png Views:	0 Size:	685.2 KB ID:	615272 Click image for larger version  Name:	Screenshot 2026-11-12 at 10.31.52 PM.png Views:	0 Size:	702.5 KB ID:	615271
    Last edited by Bry5on; 11-12-2026, 10:37 PM.
    ‘02 332iT / 6 | ‘70 Jaguar XJ6 electric conversion

    Comment


      Originally posted by 3staxontheradio View Post
      Altaran how long are the primaries and what are the diameters?
      Primaries are 571.6 to 572.9. I gave the AI a goal to keep it within 2mm. Appearantly thats the tolerance for F1 headers. Weather this is going to be 100% true to length after welding is another topic ^^

      Here's a comparison of the variance in each step length:

      | AD | Length |
      |------|----------------|
      | 44.5 | 148.9–149.3 mm |
      | 47.6 | 136.5–137.5 mm |
      | 50.8 | 148.7–149.5 mm |
      | 54 | 136.6–137.3 mm |​

      BTW the sizes are aligned with what I was able to find on mandrel bent inconel pipes, they would still be made to order but at least its known that the diameters are being manufactured and the bends are manufacturable.
      Stainless steel tubing is much more available, so it would be easier to find the closest available piping there.

      44.5 with 0,7mm tubing would be bigger than stock ports. Here, also it is easier to make the header work with larger piping in the space available and then downsize for stock heads. Naturally you will need head work for big power, but these long tube headers should still give a torque benefit. That and I'm still hoping for the exotic sound lol
      2002 E46 M3 TiAg Coupé >> full tracktool conversion @m346gt
      2000 986 Boxster S >> ice cream getter
      Past: E46 330Ci, 944S2, 996 C4S

      Comment


        We're really living in the future.

        2005 IR/IR M3 Coupe
        2012 LMB/Black 128i
        100 Series Land Cruiser

        Comment


          We've got ourselves a fresh collector design. Any thoughts on whether I should replace this with a shorter converging section, neck down to ~58mm or so, then diverge back up to 63.5? Similar to the vibrant design top left.
          Click image for larger version  Name:	collector overview.png Views:	0 Size:	427.5 KB ID:	615319
          Click image for larger version  Name:	section analysis.png Views:	0 Size:	234.7 KB ID:	615318
          Click image for larger version  Name:	Screenshot 2026-11-13 at 8.30.36 AM.png Views:	0 Size:	717.9 KB ID:	615320 Click image for larger version  Name:	Screenshot 2026-11-13 at 8.37.42 AM.png Views:	0 Size:	743.3 KB ID:	615321 Click image for larger version  Name:	Screenshot 2026-11-13 at 8.36.35 AM.png Views:	0 Size:	692.3 KB ID:	615322
          ‘02 332iT / 6 | ‘70 Jaguar XJ6 electric conversion

          Comment


            Originally posted by Obioban View Post
            We're really living in the future.
            amen, wild and impressive

            Comment


              It’s getting even more wild. The bots talk to each other to analyze peak scavenging figures based on my actual tune map.
              Click image for larger version  Name:	IMG_1123.png Views:	0 Size:	246.4 KB ID:	615350
              ‘02 332iT / 6 | ‘70 Jaguar XJ6 electric conversion

              Comment


                we're going to need cam variants :P

                2005 IR/IR M3 Coupe
                2012 LMB/Black 128i
                100 Series Land Cruiser

                Comment


                  Gotta love the bots planning for the Monday morning grind


                  Sent from my iPhone using Tapatalk

                  Comment


                    Originally posted by Bry5on View Post
                    It’s getting even more wild. The bots talk to each other to analyze peak scavenging figures based on my actual tune map.
                    Click image for larger version Name:	IMG_1123.png Views:	0 Size:	246.4 KB ID:	615350

                    Massive caveat here- Primary length tuning is 100 percent EGT dependent and unless there is an EGT input not shown Grok can't possibly know that with precision. Personally I would rather have a first design slightly too long and risk peak scavenging RPM being being 7500 instead of 7800 than accidentally building a set of headers that don't hit peak power until 8400 rpm where it will be useless. It would be a shame to build the primaries too short and then get a dyno result that makes all of this look like a waste of time. Personally I would much rather end up in a situation where a second iteration got made after being sure it was doing something useful. FWIW I was using 700c from your logs which is also just about dead on for the predicted EGT from pipemax.

                    Also on merges: Those look like they are more shallow than they need to be? 10.5 is sort of the "ideal" but 15 is the norm for packaging reasons and just about all off the shelf merge collectors are at a 15 degree angle. Also it would be best to minimize bending of the merges for them to equally scavenge across all the cylinders. If you look at the examples the Vibrant collector uses a small choke and a taper back up to collector diameter. The collector choke doesn't see steady state flow from all 3 primaries at once so it doesn't need to be the full collector diameter and shrinking diameter at a choke before tapering back up helps keep velocity high across the merge where it can scavenge between cylinders. Also all of these diameters need to be scaled to the engine output. For stock or mild cam engine I would think keeping the collector at the stock diameter for at least some harmonic length of collector tuning would be best. I had 1800mm of collector length (second collector harmonic) for my car but that's obviously going to require fully custom exhaust (which is what i"m doing on my car right now).

                    If anyone has a test part they want me to print I'm happy to print it and try to fit it. I have a few 3kg PLA rolls ready to go and car sitting with no headers in it.

                    Comment


                      Originally posted by Obioban View Post
                      we're going to need cam variants :P
                      OH man, would love to get a set of cams optimized for these headers and mullet tune
                      | 06 6 Speed Vert - JB on Cinnamon|Beisan|ACL RB's|ST Sway Bar|​3.91|Bi-LED|​ TMS CSL Box|Megan headers|H.T.E Tune|Bryson Air Inlet|​550 Injectors|OEM Section 1, 2, 3|200 Cell Cats|OEM SSK|Bilstein B12|712 yellow tag




                      Comment


                        Originally posted by 3staxontheradio View Post


                        Massive caveat here- Primary length tuning is 100 percent EGT dependent and unless there is an EGT input not shown Grok can't possibly know that with precision. Personally I would rather have a first design slightly too long and risk peak scavenging RPM being being 7500 instead of 7800 than accidentally building a set of headers that don't hit peak power until 8400 rpm where it will be useless. It would be a shame to build the primaries too short and then get a dyno result that makes all of this look like a waste of time. Personally I would much rather end up in a situation where a second iteration got made after being sure it was doing something useful. FWIW I was using 700c from your logs which is also just about dead on for the predicted EGT from pipemax.

                        Also on merges: Those look like they are more shallow than they need to be? 10.5 is sort of the "ideal" but 15 is the norm for packaging reasons and just about all off the shelf merge collectors are at a 15 degree angle. Also it would be best to minimize bending of the merges for them to equally scavenge across all the cylinders. If you look at the examples the Vibrant collector uses a small choke and a taper back up to collector diameter. The collector choke doesn't see steady state flow from all 3 primaries at once so it doesn't need to be the full collector diameter and shrinking diameter at a choke before tapering back up helps keep velocity high across the merge where it can scavenge between cylinders. Also all of these diameters need to be scaled to the engine output. For stock or mild cam engine I would think keeping the collector at the stock diameter for at least some harmonic length of collector tuning would be best. I had 1800mm of collector length (second collector harmonic) for my car but that's obviously going to require fully custom exhaust (which is what i"m doing on my car right now).

                        If anyone has a test part they want me to print I'm happy to print it and try to fit it. I have a few 3kg PLA rolls ready to go and car sitting with no headers in it.
                        This is apparently a pretty straight 9* angle. Pretty easy to do a choke and make it steeper from here. With a printed choke we can also make it a smoother transition vs a sharp expansion. I don’t know if flow separation is super relevant here but that sharp transition is awful for flow separation. And I disagree that the collector choke doesn’t see steady state. Regardless, what do you think is an ideal internal diameter for the choke from your research? I’ll try to package this all on Monday since it looks like there’s plenty of room.

                        I also agree on runner length, erroring long is better so you can at least see the peak in a dyno or fueling log. If you include the exhaust port and the internal of the collector to where the pipes meet their adjacent friends, that puts these actually around 630mm + port length, so definitely on the longer side and pushing into the midrange solidly under 7k rpm (which would probably be my personal preference honestly). Easier to make them shorter than longer too. That RPM estimate was using “Bell’s formula” and not a wave model. Just a high level empirical estimate. Wave model to come once everything is more dialed in but at present it’s predicting 5700rpm with the current collector and an 80mm port length (710mm total acoustic) which I forgot to feed to the AI brain earlier.

                        I think ~7300 is a good target at 700C, which would bump peak scavenging under a track condition up closer to something like 7600 due to the higher exhaust gas temps. Maybe I’ll play with a shorter version next once the collector is done to bring that back up. It should be easier now that I’ve cleaned up all my keepouts and nesting requirements.
                        ‘02 332iT / 6 | ‘70 Jaguar XJ6 electric conversion

                        Comment


                          As far as choke sizing goes I'll ask burns what they think

                          Comment


                            Originally posted by Obioban View Post
                            we're going to need cam variants :P
                            I did ask AI to do a comparison of the headers on stock cams and stock engine, they still work with similar. As far as the research i did came out with, the longer primary design and a proper collector still makes it work on stock engines with stock cams. The higher the cam opening times go (especially with overlaps between intake and exhaust opening), the more important is a proper collector with a spike and to have a proper intake.

                            Another big difference would be if there was head work done, in this case especially larger exhaust ports as it impacts primary inner diameter.
                            2002 E46 M3 TiAg Coupé >> full tracktool conversion @m346gt
                            2000 986 Boxster S >> ice cream getter
                            Past: E46 330Ci, 944S2, 996 C4S

                            Comment


                              Something for everyone working on designs to consider that became more apparent when I was printing plastic version: Try slicing your models into pieces that fit into a 350mm bounding cube and then figure out if there is weld access, especially at the merge. When primaries are separate it is very easy to access every joint. The merge isn't accessible in the middle. With closed loop fuel control we need zero leaks, especially at the collector and 3d printed stainless. I'm thinking double slip merges are the way to go but that requires tolerances that might be hard to hit with DMLS stainless. I think using a last 2-3" of primary that is an off the shelf straight tube is a good way to limit tolerance stack to just the merge part. Burns sells double slips in 1 7/8 for $37 each. Not cheap when you need 6 of them. It would be ideal if there was a way to cut these up so that the merge joints could be fully welded. I keeping welds accessible on bank 2 is possible but on bank 1 the primaries tightly bundled and parallel and I haven't been able to make it work.

                              Comment


                                Originally posted by 3staxontheradio View Post
                                Something for everyone working on designs to consider that became more apparent when I was printing plastic version: Try slicing your models into pieces that fit into a 350mm bounding cube and then figure out if there is weld access, especially at the merge. When primaries are separate it is very easy to access every joint. The merge isn't accessible in the middle. With closed loop fuel control we need zero leaks, especially at the collector and 3d printed stainless. I'm thinking double slip merges are the way to go but that requires tolerances that might be hard to hit with DMLS stainless. I think using a last 2-3" of primary that is an off the shelf straight tube is a good way to limit tolerance stack to just the merge part. Burns sells double slips in 1 7/8 for $37 each. Not cheap when you need 6 of them. It would be ideal if there was a way to cut these up so that the merge joints could be fully welded. I keeping welds accessible on bank 2 is possible but on bank 1 the primaries tightly bundled and parallel and I haven't been able to make it work.
                                I think there’s a way. I’ve been considering splitting the collector at the choke to give some extra space for runner printing and also allow the same file to be compatible with either stock or 63.5mm section 1. I really don’t think slip joints are the way to go - we can print male/female clocking features and design for a nice butt joint which works super well with stainless.
                                ‘02 332iT / 6 | ‘70 Jaguar XJ6 electric conversion

                                Comment

                                Working...
                                X