MPG optimised header ideas

[Madact]

You could be right - I was thinking scavenging might be a motivation though simply because they've gone to the trouble of a tri-y design, but the merge with the "holey plate" looks a heck of a lot cheaper to fabricate than a conventional merge. You can see the pipes coming in have been simply cut at right angles and inserted...

[Nigel_S]

Originally Posted by Madact

they've gone to the trouble of a tri-y design

I think that was normal for that generation of engines, don't know much about Toyotas but Rovers of the same generation certainly had tri-y headers as standard even in the cheapest models, they only disappeared with the introduction of environmentally friendly pre-cats in the manifolds which caused significantly more fuel to be burned and thus more CO2 released into the atmosphere! Now to upgrade we have the choice of an aftermarket tri-y header or an ancient standard one from last century which performs almost identically!

[freebeard]

In the 1980s there was an available 4-into-1 header for VWs that had a cast section at the merge with a very specific shape, that was presumably better that just jamming the tubes together. I can't remember the name, or a relevant search term.

So why as I posting this? Frustration.

[freebeard]

Pretty much. Thanks. I bookmarked it. That will save me having to go through banker boxes full of magazines from the 80s.

[Madact]

Yet more confirmation that 1.25" is the right choice for the primaries: Exhaust Header Design Comments

Same formulas as in the early posts of this thread, but makes a specific point to say "Note that 1-1/2” pipe is large enough for a 273” motor with max torque @ 4000 RPM." ... for reference, a 273" has 559cc per cylinder compared with a D16's 399cc.

On a side note, does anyone have an idea how much dyno time (and cost) would be involved in doing a full BSFC map for several throttle positions? I have an annoying feeling the answer is 'a few hours' and 'pricey' :-/ but I'm thinking A-B-A testing at cruise won't really show up header differences very well w.r.t. torque curves etc., and I'm not sure A-B-A testing going up hill would be very easy to get consistent (doubly so as I have no cruise control). Not to mention that A-B-A testing on a header is a fair bit of effort and waiting for the thing to cool down enough would put a bit of time between runs...

[Madact]

Originally Posted by Old Tele man

Something like these people sell?

http://www.spdexhaust.com/pdfs/02-11...Collectors.pdf

On a side note, a quick google search for exhaust merge collector or exhaust merge spike will net you more suppliers, manufacturers, jigs and how-to guides than you could poke a stick at.

Another interesting merge style is used by "Banks Power", the apparently very patented "power pickle", which is much like a conventional merge spike but looks fatter, presumably reducing cross-sectional area in the merge and thus keeping velocity up:

Call me cynical, but patenting "X but fatter" seems a lot like patenting "X but on a mobile device" as seems to be the vogue for software and user interfaces these days - there may be something more to it, of course, and I suppose novelty is always harder to judge in hindsight (the old 'anyone could have done that' vs 'did anyone do that?' argument). Not that merge *spike* design is that useful to me designing a tri-y, but there should be a few good ways to control cross-sectional area through the merge, using either internal elements ('merge spike' in a 2-way merge anyone?) or wall profiles (probably - but not necessarily - bulges along the edges where the tubes 'meet' along the merge) ...

[Nigel_S]

Originally Posted by Madact

('merge spike' in a 2-way merge anyone?) or wall profiles (probably - but not necessarily - bulges along the edges where the tubes 'meet' along the merge) ...

Due to the inertia of the gasses, I doubt the shape of the spike makes much difference, your probably more interested in gathering the flow up again after passing the other tube. As you noted with the Toyota merge, the primaries in OEM headers tend to be just cut square and inserted into the merge but some do go to a bit of effort after the join, often creating a cone on the start of the secondary rather than just having a straight cut pipe. The shape of the end of the primary will have some effect on reflected pressure waves traveling back down the primary, possibly a straight cut pipe is actually the best for creating those reflected pressure waves. Shaping the start of the secondary to control the expansion into the wider secondary can create a bit of extra suction in the area of the merge. I've not seen much written on this subject, it seems most designs are the result of a good idea followed by testing and normally limited testing due to limited budgets so it may be worth a look at more OEM exhausts as they do tend to have decent testing budgets and the better ones clearly do get things like pipe diameters for each section correct.

[Nigel_S]

Originally Posted by Madact

but I'm thinking A-B-A testing at cruise won't really show up header differences very well w.r.t. torque curves etc.

I managed to draw this graph showing %torque increase of my tri-Y header over the standard header with pre-cat in the manifold just by timing the cars acceleration to each 250rpm over a few runs before and after using a camera in video mode - the timing is accurate to the video frame rate of 1/30th second which is good enough:



I didn't have a way of measuring fuel flow so that was as far as I got but if you have a flat piece of road that you can play on then I think a DIY approach is likely to give a more satisfying result than a dyno. Obviously you can do more than one throttle position simply by putting a stop at half throttle or whatever and that probably tells you all you need to know. What you are really interested in is how much does it improve your MPG in normal driving conditions and that can't be measured on a dyno.

[Madact]

Not a bad thought. With an MPGuino (which I have, but not yet installed), a tachometer and a vacuum guage (both on the to-do list), you could get a pretty decent approximation for BSFC. The torque measurement might not be perfectly precise, but if done on a very level stretch of road (ridiculous long stretches of dead flat road are kind of a speciality around here... well, within 50km of where I live, anyhow) with a warmed up car in several different gears and several throttle positions, something usefulcould probably be done with the data. Enough to do A-B testing, anyhow (still not keen on A-B-A for a header )

[Nigel_S]

Originally Posted by Madact

The torque measurement might not be perfectly precise, but if done on a very level stretch of road (ridiculous long stretches of dead flat road are kind of a speciality around here... well, within 50km of where I live, anyhow) with a warmed up car in several different gears and several throttle positions, something usefulcould probably be done with the data. Enough to do A-B testing, anyhow (still not keen on A-B-A for a header )

You need to do it on the same piece of road since invisible humps and dips do affect the results a fair bit!

I put the car in 3rd gear, dropped down to 1000 rpm, floored the throttle as I passed a marked point and kept it floored to 7250 rpm. humps didn't really matter since the graph is %increase so unless there was so much extra power than the car reached the hump at a different point in the graph it would still be correct. I did attempt to remove the drag losses by subtracting a coastdown run but that can't be measured on the same piece of road as it takes a very long time to drop from 3rd gear 7250 rpm speed to 3rd gear 1000 rpm speed so I just subtracted an averaged and smoothed version.

The graph is % increase, not actual values measured in a particular unit since I had no way of calibrating it to get specific units. The graph is accurate, the only real problem with it is that it doesn't show if the peaks are due to a good new header or an average new header but a poor old header, I found it quite hard to judge that from the data, partly because the road wasn't perfectly flat even though it looked flat.

To get the torque figure you need to calculate the change in kinetic energy needed to change from one speed to the next - an amount that increases with speed, and the time taken to make that change.