I also found another interesting "formula" based calculator for header design here:
Pipe Sizing Calculator - this one is clearly a work in progress, but it's interesting because it implements a couple of alternative formulas from different sources for obtaining pipe lengths. All the methods which take into account RPM or mass flow (including the formula / calculator mentioned earlier) seem to be in fairly good agreeance on the basic parameters, so that's a positive sign
Some other good insights from comparing the various formulae:
The calculator asked for aggregate exhaust valve diameter (i.e. how big a single valve of the same area would be) which turns out to be 36.75mm. Various sources agree that if the cross sectional areas of the primary pipe, exhaust valve and exhaust runner (inside the head) are approximately equal, you shouldn't loose much/any power at high RPMs unless the primaries are very long (i.e. friction losses). 1.5" OD SS pipe (with the 1.5mm wall thickness that seems to be pretty standard for exhaust systems - not "steam pipe" thickness though) has an internal diameter of 35.1, which seems pretty close... taking into account the valve stem area, it's an area reduction of only ~3%. Based on this I'm comfortable choosing 1.5" OD for the (first section of) the primaries.
Another interesting thing, backed up by some further reading, is that there seem to be three schools of thought regarding primary vs. secondary pipe length with tri-y / 4-2-1 headers. One school prefers equal length primary and secondary (based on 1/2 the total length to the final collector), another other prefers 15" all the time, and the third uses a length proportional to volume * exhaust stroke time. Funnily enough, in the 5000-6000 rpm range, all these numbers fall between about 13.5" and 15"
, so they're pretty similar for 'conventional' applications, and the differences would probably only change things by a couple of 100rpm at most. For lower RPMs, though, they're very different - the methods that take account of engine speed all spit out 31-33", while other methods not dependent on RPM spit out 14-15".
Using some more accurate valve timings (turns out the ones I found earlier were for an aftermarket cam), all the different formulae give distance to the final collector / reflection point as being in the 62"-72" range at 2400rpm. Based on this, I'm happy with putting the final Y at around 60" with the cat straight after to provide secondary reflections.
Based on this, I think I'll keep my original plan of an anti-reversion step at 14-15" to capture the first rarification pulse reflection (especially useful at high RPM), and a merge at 32"... between harmonics and the fact the the flow-rate-based formula and 1/2 total length rule agree on ~32", it can't be too terrible
As for simulation software, I've taken a closer look at discussion and screenshots of pipeMax, it looks like 'simulation' consists of outputting predicted torque/power curves rather than any actual FEA or similar code, and it's unclear whether you can 'simulate' an arbitrary connection of pipes rather than the generated selection. Given the price of 'real' simulation software, the "try it and see" method actually works out cheaper provided I go through 3 or less iterations of the design, and even then that's assuming each iteration is scratch-built from new stainless steel
... and of course, getting the
exact measurements will likely only shift things by a couple of 100rpm, which only matters for a highly tuned system (high Q-value resonance) and I'm thinking mild UEL (by about 10%) would be better for this application anyway, as it should give a smoother response.