MPG optimised header ideas

[Madact]

Originally Posted by Nigel_S

There are no explosions involved, the valves open smoothly on a smoothly curved cam, initially allowing a tiny flow which smoothly increases before being smoothly closed again. The exhaust gasses should flow smoothly out the ports and down the pipes. Even the burning of the fuel isn't an explosion as many people think, it starts burning at the spark and the flame spreads out smoothly through the piston giving a smooth increase in pressure.

If things start exploding then things are going wrong, that is what detonation is and that destroys engines.

The use of the word "explosion" was hyperbole / non-technical-definition there (sorry for the imprecision) - the rest still applies though. Exhaust gas flow is one of those problems for which you don't need to consider laminar flow conditions at all.

Originally Posted by Nigel_S

The main job of the header is to transport the exhaust gasses, they will flow easier through nice straight pipes, putting extra curves in to get equal lengths seems very questionable to me.

Yes, it's definitely questionable.

Extra curves are going to introduce extra resistance that's a given (this is also true of adding extra length). On the other hand, extra length is going to change the resonance and dynamic properties of the system.

The actual question, of course, is whether the tradeoff is worth it - and I definitely don't know the answer to that in this case .

I do know that for many, many racing and 'high performance' applications, equal length headers are worth quite a few extra bends even in cases where they have the option of running 'zoomies', 'shorties' etc. - I don't imagine Nascar and F1 teams would put that much effort into something that didn't offer an advantage. Whether it's worthwhile for the other end of the spectrum is, indeed, another question.

As with any engineering tradeoff, any given amount of extra bend, bend tightness & length will be (in very crude terms) "worth" a certain amount of advantage from dynamic properties. How much exactly? Well, I reckon I can manage double length with a pipe layout with only an extra 210 to 240 degrees (depending), some of which can be at a fairly 'lazy' radius. A 'ram horn' adds about 180 to 240 of extra curves depending on the layout.

Some people claim that for raw "performance", the tradeoff between building a ram horn and having slightly shorter pipes definitely is worth it (e.g. "the k-tuned" company - see the writeup here:http://www.k-tuned.com/blog/products/ram-header/). Of course, their shorter headers may simply be the "wrong" length, as they're working with K-series (rear exh. ports) swaps, so their other headers might just be tuned for detrimental resonance - I'd be more convinced of course if they ran a comparison with an untuned but cleanly built 'shorty' tube manifold, of course if the shorty came out better than their standard 4-1 and 4-2-1 they'd have a bit of egg on their faces not to mention that a shorty doesn't look as sexy as their other offerings..

---

I strongly suspect I'll have to build, dyno, tune, and road-test two versions to answer the question in my own mind. On the upside, at least one of the versions should be better than the stock manifold...

[Nigel_S]

Originally Posted by Madact

The use of the word "explosion" was hyperbole / non-technical-definition there (sorry for the imprecision) - the rest still applies though. Exhaust gas flow is one of those problems for which you don't need to consider laminar flow conditions at all.
...

I realised that you didn't mean a real explosion, but still, it doesn't help to think about it in that way because it should operate smoothly,

The flow in the pipe should not be turbulent, maybe "laminar" is not the best description for a tube but there is certainly a boundary layer against the edge and the main flow goes down the centre, unless the pipe is curved in which case it will be off centre and you need to worry about keeping the flow attached to the inside of the curve - it is laminar flow.

I dug out this diagram:



EVO - exhaust valve opening takes place during the power stroke, long before the end of the stroke, red line is the pressure in the exhaust port which rises smoothly and then starts to drop again just before the end of the power stroke because by the end of the power stroke most of the exhaust gas has left the cylinder. Dark blue is cylinder pressure. Half way between BDC (end of power stroke, start of exhaust stroke) and TDC (end of exhaust stroke) the cylinder pressure reaches atmospheric despite the fact that the piston is rising so should be compressing the gasses left in the cylinder, the inertia of the exhaust gasses in the manifold have sucked the gasses out of the cylinder leaving the piston with almost no work to do, they continue to drop so that by IVO (inlet valve opening) there is a strong vacuum in the cylinder to pull the new fuel air mixture into the cylinder even though the piston is still on it's way up. Cyan line is inlet port pressure which meets the exhaust port pressure before the end of the exhaust stroke and then there are a few waves on the exhust port caused by wave reflections in the exhaust manifold one of which is timed to give a little extra suction just before the exhaust valve closes - those little waves are the effect of the wave timing, the creation of the large vacuum is the result of suction from the exhaust gas inertia and venturi effect between header pipes. Remember that EVO, IVO and EVC are events that take time, not instant events as indicated by the lines on the graph.

[Madact]

Originally Posted by Nigel_S

The flow in the pipe should not be turbulent, maybe "laminar" is not the best description for a tube but there is certainly a boundary layer against the edge and the main flow goes down the centre, unless the pipe is curved in which case it will be off centre and you need to worry about keeping the flow attached to the inside of the curve - it is laminar flow.

Turbulent flow will still have a boundary layer, and will still have the potential flow separation problem, you're right, but it's still turbulent flow - the velocity profile is quite different between laminar flow and turbulent:



Obviously laminar is preferable as the overall rate is higher, so you don't want to muck it up if you have it - but going past the edge of the valve, the valve guide, and taking a ~90 degree turn out of the head ensures that the flow is turbulent to start with.

Even if you were to start with a smooth laminar flow, it will transition to turbulent after a given length of pipe (dependent on the Reynolds Number) anyway (quite soon, for the flow rate and cross section of an exhaust pipe), and once you go turbulent you don't go back. Now, the *amount* of turbulence can still change, but if you start out with turbulent flow, you don't have to worry about a (slightly) rough surface finish or gentle curves, because you're no longer concerned with avoiding a premature laminar-turbulent transition.

[Madact]

OK, got PipeMax... I've got to say, it's pretty impressive - enter stock valve specs, cam timings, volumetric efficiency etc., select "shorty header + muffler", and the predicted torque and HP values seem to hit the stock engine specs pretty darn close which is good for confidence.

Anyhow, the length values seem to agree very much with previous calculations, using the tri-y street+muffler setting. Having done more calculations for a couple of different operating conditions though, I think the magic number may be 27" instead of 32".

Why is this? Well, the "standard" header formulae are based around the kind of exhaust gas temperatures you'd expect to see while racing, or near the end of a WOT dyno run (around 1485F, the default in PipeMax, gives numbers that agree with other formulae I've seen). Lower EGT brings the calculated lengths down, and 'cruising' EGT at part throttle & closed loop ECS control will a bit lower than 'race track'/dyno EGT... also, given the (future) head swap, other considerations are that exhaust gas recirculation also brings down EGT, while lean combustion raises it - not sure which will be dominant. If anyone a civic HX with an installed EGT sensor is reading this, some input would be great

So, picking some numbers out of the air (and this thread what are typical gasoline exhaust temperatures? - Diesel Forum - TheDieselStop.com), I've also run calculations for EGT=1000F at low RPMs and EGT=1200F in the mid range.

The really, really cool thing here is that it seems that Physics has thrown us a bone here, and the exhaust system will 'tune itself' to higher RPMs if you're stomping the accelarator for extended periods. Meaning when it's tuned for cruising at low RPM, it's also, by lucky coincidence, not too far out for being thrashed around at WOT in the high RPMs...

The other interesting thing here, is that the recommended primary diameters are a bit smaller than I was previously thinking from other formulae... based on this, it looks like a 1.25" OD primary may be a better choice than the 1.5" (all diameters below are O.D., by the way)

----------
Many numbers follow
----------

"Low range" calculations:

Vtec-e "low cam":
Setting "3100rpm" max HP at 90% VE, at 1:9.7 compression:
59.4 to 62.7", "operating range" 1000-3500 rpm
* 3100rpm is guess for the max HP RPM, based on promotional material, theoretical/illustrative graphs for the D15Z7, and staring at the bumps on a bunch of D16Y5 dyno graphs from the internet.

Same RPMs for non-VTEC:
Setting "3100rpm" max HP at 85% VE, 1:9.4 compression:
61.9 to 65.2", "operating range" 1100-3600 rpm
* Cam specs appear to change only the pipe diameters & predicted performance here, not the pipe design, but as I understand it, non-VTEC has lower volumetric efficiency as the cams can't be so well optimised.
*** And with EGT set to 1000F:
53.2 to 56.5

"Mid range" calculations, based on peak HP at 6400 (design spec - D16Y8 quotes 6800rpm max HP, but that's past the VTEC switchover):

Standard cam, using D16Y8 specs:
Setting "6400rpm" max HP at 98% VE, 1:9.7 compression:
25.6 to 28.8", "operating range" 4400-6900 rpm, 1st segment diameter 1.212 (or 1.412 for "higher RPM power, possible TQ loss")

D16Y4 guess, using D16Y8 standard cam specs because I couldn't find anything for the Y4:
Setting "6400rpm" max HP at 95% VE, 1:9.4 compression:
26.2 to 29.5", "operating range" 4400-6900 rpm, 1st segment diameter 1.2 (or 1.4 for "higher RPM power, possible TQ loss")
*** And with EGT set to 1200F:
24.4 to 27.6"

"Mid range" calculations, based on peak HP at 5800 (de-tuning peak power for wider torque):

Standard cam, using D16Y8 specs:
Setting "5800rpm" max HP at 98% VE, 1:9.7 compression:
28.5 to 31.8", "operating range" 3800-6300 rpm, 1st segment diameter 1.157 (or 1.357 for "higher RPM power, possible TQ loss")

D16Y4 guess, using D16Y8 standard cam specs because I couldn't find anything for the Y4:
Setting "5800rpm" max HP at 95% VE, 1:9.4 compression:
29.3 to 32.6", "operating range" 3800-6300 rpm, 1st segment diameter 1.145 (or 1.354 for "higher RPM power, possible TQ loss")
*** And with EGT set to 1200F:
27.2 to 30.5"

"High range" calculations:

Standard cam, using D16Y8 specs:
Setting "6800rpm" max HP at 102% VE, 1:9.7 compression:
23.2 to 26.5", "operating range" 4400-6900 rpm, 1st segment diameter 1.404 (or 1.604 for "higher RPM power, possible TQ loss")
At "7000rpm" max HP, same settings - as I have no data for the D15Z7 high cam:
22.5 to 25.7" 1st segment diameter 1.423 (or 1.623 for "higher RPM power, possible TQ loss")

[Nigel_S]

Originally Posted by Madact

OK, got PipeMax...

Does it suggest the pairing of the pipes in the tri-y header?

Seems to me that for a 32" to end of secondary header you need 1 pared with 4 but for a 32" to end of primary you want 1 pared with 3, but I've never noticed that written or implemented...

"I think the magic number may be 27" instead of 32"." - mine have a range from 26.5 to 31.5, measured from the valve to end of secondary. (different engine)

[Madact]

Originally Posted by Nigel_S

Does it suggest the pairing of the pipes in the tri-y header?

Seems to me that for a 32" to end of secondary header you need 1 pared with 4 but for a 32" to end of primary you want 1 pared with 3, but I've never noticed that written or implemented...

That's a very good question. I think the double-length version will have the same cylinder pairing as the equivalent single-length design, but I'm not entirely sure, and it can't hurt to ask...

One of the interesting things from PipeMax is that the lengths calculated for the "3-step" header (small primary -> large primary -> secondary in length ratios) look very similar to the "hybrid tri-y" designs I posted earlier in this thread, the small primary:large primary:secondary length ratio is 2:1:1... and they all seem to use sequential pairing instead of even-spaced pairing. On a side note, pairing 1 with 3 and 2 with 4, is equivalent to 1 with 2 and 3 with 4 - in either case, each secondary pipe gets two pulses in a row.

Originally Posted by Nigel_S

"I think the magic number may be 27" instead of 32"." - mine have a range from 26.5 to 31.5, measured from the valve to end of secondary. (different engine)

Indeed - magic number for the specific engine, obviously, give or take an inch or so. I say 27" because it's in the overlap point between the "EGT corrected" calculations given a bit of uncertainty in valve timing, for the low to mid RPM ranges, and it's also darn close to the acceptable range for the high-rpm, high-EGT VTEC cam results. I believe my calculations should apply fairly well to any of the Honda D16Y engines though, if they're running stock cams - minor variation in cam timings didn't seem to change the recommended length by more than about an inch, but aggressive aftermarket cams would definitely change things - not that that should be an issue for most people on this forum .

[Madact]

Just found a 1D research simulation code which is open source: https://code.google.com/p/icesym/

Paper which I believe refers to the experimental validation of this code: E. J. Lopez and N. M. Nigro,
"Validation of a 0D/1D computational code for the design of several kind of internal combustion engines"
and another paper using this in a 3D / 1D coupled simulation:
Ezequiel J. Lopez, Norberto M. Nigro, "Computational simulation of in-cylinder flows in internal combustion engines by means of the coupling of zero-/one-dimensional and multidimensional codes"

Will have to see if I can get this up and running.

[freebeard]

Your '“k-tuned” headers' link needs touching up.

This all sounds reasonable, 4-2-1 would definitely be the way to go. Except I would think the exhaust port should be matched. A round pipe on an oval port will induce turbulence.

Ceramic coating would be a preservative. It may/may not be more effective for heat than wrapping with a thermal tape.

For after your catalystic converter:


If you look at vascular structures like capillaries, you see structures like this:

http://www.fractal.org/Fractal-Research-and-Products/Fractal-Tissue-Engineering.htm

The 'pipes' don't have a constant cross-section in the vicinity of the branch. At the very least, the branches should be a lower case y not an upper case Y.

[Madact]

Originally Posted by freebeard

This all sounds reasonable, 4-2-1 would definitely be the way to go. Except I would think the exhaust port should be matched. A round pipe on an oval port will induce turbulence.

Agreed, and I found an excellent series of how-to videos on the actual header fabrication on the youtube channel of stainless headers mfg. inc. - the fourth video in particular covers dollying out the primary tube to match the ports for a smooth transition.

Custom Headers Part(4 of 6)

The 'fun' part though - the PipeMax program is predicting optimal primary sizes of 1.25" OD at the outside (28.6mm ID), with 'possible torque loss' at around 1.375" OD (31.7mm ID). The ports themselves (inner measurements) are 28mm x 39mm. If I had a converging taper there, it would have to fit within about 75mm length, which seems like a bit much when going down from 39mm to 28.6mm in one dimension (the other dimension would stay the same). On the other hand, using the 1.375" OD means I won't choke the engine too much (for street use) even with the VTEC wild cam, and a 'double length' design may reclaim lost torque.

Originally Posted by freebeard

Ceramic coating would be a preservative. It may/may not be more effective for heat than wrapping with a thermal tape.

I've seen good reports, relatively low measured surface temperatures (below WOT crazy-turbo-boost scenarios) in particular are quite convincing as to the effectiveness - if the heat isn't coming out thermodynamics says it must be staying in it will depend somewhat on how much it costs, of course.

"For after your catalystic converter:"

LOL ... mind you, a logarithmic horn does have very good atmospheric coupling ... "cataclysmic converter" perhaps?

Originally Posted by freebeard

If you look at vascular structures like capillaries, you see structures like this:
[...]
The 'pipes' don't have a constant cross-section in the vicinity of the branch. At the very least, the branches should be a lower case y not an upper case Y.

I'd be very cautious about extrapolating across scales that much - the flow regimes will be very, very different. Analogies can definitely be drawn from nature, but only if the Reynolds Number is very similar. For example, a supersonic aeroplane wing, a bird wing and a fruit-fly wing are operating across wildly different flow regimes, and therefore have to obey completely different 'design rules'. On the other hand, bird wings, model aircraft wings, and light aircraft are in a similar regime and follow common 'rules', or various insect wings are operating in a similar regime and so follow common 'rules'.

In design generally, "Form follows function". In fluid dynamics, "Form follows Reynolds Number".

[Madact]

OK, so here's the specs for two variants, the (sane) "single length 2-step tri-y" and the (questionable) "double length 3-step tri-y".

--- Megaphone Collector Specs ---( Diffuser or Diverging Cone Shape )---
*** Note: this will common to both designs, and matches the "Best HP/TQ Tuned Collector Length" from the mid-range calculations at realistic EGT. To save space, the last 4"-6" would be flex tube with an interlock liner.
Diameter= 1.5 taper to 2.5 Length= 16 inches

--- 2-Step "Single length" TRI-Y Pipe Specs ---

1st Y-Segment 1st Step Dia. inches= 1.375 Length= 13.5
2nd Y-Segment 3rd Step Dia. inches= 1.625 Length= 13.5
Combined Total of all Segment Lengths = 27 inches long
Including megaphone and 2" for cat. flange, -3" for exhaust runners inside the head = 42"

--- 3-Step "Double length" TRI-Y Pipe Specs ---

1st Y-Segment 1st Step Dia. inches= 1.375 Length= 27
1st Y-Segment 2nd Step Dia. inches= 1.5 Length= 13.5
2nd Y-Segment 3rd Step Dia. inches= 1.625 Length= 13.5
Combined Total of all Segment Lengths = 54 inches long
Including megaphone and 2" for cat. flange, -3" for exhaust runners inside the head = 69"

And how, might you ask, do you fit all that between the radiator and the engine? Here's one way :

(Note: Y merges are marked / tied together with the green tape. The 'pipe' is 15mm plastic-foam "gap filler" as used for sealing building cracks, in case you were wondering - actual diameter more like 16mm)






Madness? ... THIS! IS! SPARTAAAAAA!

But more seriously, using the aforementioned ram horn header as an "as far as you can push it" scenario:
* The ram horn header has all the ports taking a sharp turn (looks like 1D) straight out of the port, while this only has a 1.5D turns starting 4" from the port.
* Cylinders 2 and 3 have a similar amount of total bend at a similar radius to the ram horn header.
* Cylinders 3 and 4 have a bit more total bend than the ram horn, but can use much more 'swoopy' curves.

So, another question comes to mind... can a ram horn header only get away with this much bend because the 'ram horn' itself is a constant, steady turn? Or is having straight sections between the bends a good thing? In either case, next thing I'll try is a more traditional rams horn with primary Y merges just below the flange, and extra length from swoopyness below - but that's for another day.

-----
EDIT:

And I just realised I put the primary merges at 27" from the exhaust valve, rather than 40.5" - whoops! Just imagine them paralleling thier respective partners a little further before merging. And that the lower section might be better if it spiralled (like the Mazda header) rather than taking an S-bend. That's what comes of doing modelling after midnight I suppose ...