exhaust design

[triathlonboy]

I have read on this site about people making their primary lengths adjustable for optimisation on the dyno, but what about secondary lengths? Is this important, and have people tried it?

We are designing our system to have adjustable primary lengths but I was wondering what the concensus was for secondary?

[Matt Ahl]

They are important that they are there, but have little effect compared to other aspects of the exhaust.

[rjwoods77]

call burns stainless. They rock. Save youi a bunch of time too.

[-]

Although Burns Stainless has quality good, they tend to be much more expensive then some other places out there for items such as bent and straight tube.

[Charlie]

They do have an effect, it is less of one than primaries, especially if you run a 4-1 collector. Exhaust outlet length matters too, don't forget to test with the muffler you plan to run.

To me, adjustable just means cut a longer pipe shorter http://fsae.com/groupee_common/emoticons/icon_wink.gif If you are going to do some physical testing it's a very easy thing to try some different outlet lengths, somewhat harder to do secondaries (if it's not your outlet) but the best way to find out how your system works is testing. And only you know what you want your power curve to look like.

Rob are you saying that Burn's will tell you what your secondary length is supposed to be so you don't have to spend time thinking about it? Wow...

[BeaverGuy]

Burns will help you with sizing your exhaust. Last year there were a few emails between our engine team captain and Burns about exhaust lengths and diameters. However, when you have the ability to actually test different lengths that far outweights anyones recomendations.

[triathlonboy]

Cheers guys, will do.

[Bazanaius]

Reviving an old thread here, but I'm designing our exhasut this year and I'm trying to ascertain the best way to figure out what's going on in a practical sense, given our budjet is unfortunately rpetty much non-existent.
There's all the theory behind exhausts but at the end of the day, it invariably ends up with various lengths of tube and a hacksaw. Reading Smith's Intake and Exhaust Design, he advocates the use of pressure tappings etc. This seems expensive and time consuming (although I'm sure very useful).
Do people mostly use dyno's and an experimental approach, do you simulate in some software before trying actual lengths, do you take pressure readings? Are there other methods I'm missing?
Basically, I'd like to get an idea of what people use to develop their exhaust designs - how good have you found your process, does it work etc?

any thoughts and ideas welcome,

cheers,

Baz

FBR

[exFSAE]

Wave reflection tuning has worked pretty well in my experience.

Figure out where you want peak torque, design header length around that, and build.

The biggest thing regardless of what exhaust you come up with is to have it tuned.

[Bazanaius]

Hey exFSAE thanks for the quick reply.

FYI, we're running an R6 engine, and we'll be using a 4-1 collector. So if I understand you, you suggest we design our header length for the peak torque (yep, happy with that), get our 4-1 collector, then tune for the final length from there?

Could you give me a brief description of wave reflection tuning - I'm not familiar with the term, and a brief google hasn't turned up anything?

thanks again

Baz

[exFSAE]

I believe there's been some discussion in years previous here as to exhaust tuning.. by Helmholtz resonator methods (which I call BS on).. impedance transform methods.. and wave reflection methods. The last one makes the most sense to me. Here's as I understand it, and was explained to me by a well-respected and very successful race engine builder..

At exhaust valve opening (EVO), the cylinder still has a lot of pressure built up after the power stroke, and "blows down." This rapid release of pressure, starting say at 0.050" EV lift, creates a positive pressure wave. This wave travels down the header at the speed of sound in exhaust gas at that temperature (not to be confused with speed of sound in air at room temp!). When the wave reaches a change in medium (in this case a larger diameter orifice, as in a 4-1 collector), in accordance with wave theory a fraction of the wave is transmitted and keeps goin down the exhaust pipe, and a fraction is reflected back up the header, with negative sign.

So now you've got a negative pressure wave comin back up the header. The idea is to set the header length such that the reflected wave gets back to the exhaust port just before it's about to close (so say again, 0.050" EV lift before close). The pressure differential accross the valve helps scavenge the last bit of exhaust gas out of the cylinder, and increases engine VE. You will have boosted torque here.

For a 4-1 collector, as in an I4 engine, or half an F1 V8 bank, you have a large rapid change in medium at the junction. Good wave reflection. Very peaky.

With a 4-2-1 collector as in more of a street car setup, you have slightly softer reflection at the primary junction but you get more tuning points (reflection points) and a broader, more driveable, "autocross" torque band IMO. The effect gets less and less the further down the pipe you get, until you get the weakest reflection from the actual end of the pipe to atmosphere.

You can try fooling around with intake runner length tuning, as we did on our 03, 04, and 05 car.. and had almost no result. Same exhaust setup more or less on every car.. a short-header 4-1. You'd have no torque up until like 10000rpm and then it would blast on like a lightswitch. While the result was fun, and the first time you drove the car at WOT and got in the powerband it was like having an orgasm.. it wasn't the most driveable setup.

After that we played around with exhaust change.. with a longer primary, 4-2-1 setup. The change was amazing. Torque started to come on much lower, like 7000rpm or so and stayed pretty constant and then tapered off in the high revs. Area under the torque curve was probably the same, but it was much smoother to drive.

But that's just my thoughts and experience. Take it for what it's worth. I'm more of a tire/suspension guy.

[Bazanaius]

As always, additional opinions (especially from people who have more experience) are welcome! Your reasoning makes sense and is based on experience - I shall chat to the other guys in the engine team and bring up your thoughts. our main reasons for selection are packaging (although I think this is less marginal than we first expected) and a few sources implying it mgith garner more power. Obviously a more driveable car is preferable to more power so it's definitely worth reconsidering.
The wave tuning method you describe is indeed familiar (particularly having read Smith), I've just never heard it called that before :-)

[mangel83]

You could get in touch with the people at Ricardo. They support the event and will give you few licenses of WAVE and limited support time if you put their logo on your car.

Ricardo (http://www.ricardo.com)

[Bazanaius]

haha chances! I'm working for them over the summer... I'll get in touch.

cheers mangel :-)

[TMichaels]

Hi Folks,
we used both setups over the years, 4-2-1 and 4-1. There is indeed a difference in torque developement, but as long as you choose your drivetrain and gearbox ratios right, it is not really a problem that the usable rpm band of the 4-1 is a bit shorter. We had no problems with a peaky powerband, I think that does only depend on your whole setup and your dyno work. I think it depends mostly on your package which configuration you should use. You can get good results with both setups.

Regards,

Tobi

[Wesley]

Last year we ran a 4-2-1 header for that very reason - more torque peaks. But at the same time, for every peak you create with resonance tuning, you create a depression, and when tuning with math you need to remember that.

This year we switched back to a 4-1 for a few reasons, weight (less collectors, less tubing,) packaging, cost, and ease of manufacture. On top of that, given the RPM range we were tuning for, WAVE showed a 4-1 as being quieter than a 4-2-1.

We spent a lot of time last year correlating a WAVE model of our F4i with loads of dyno data (no pun intended) and I worked extensively in WAVE to iterate a power curve that increased high-end power without losing any power from last years above 4 grand. The goal was to have the car pull all the way to redline at 14,000 and not fall off. I succeeded in this, but with vastly different geometry that we had last year: larger and longer runners, shorter 4-1 header, and a slightly smaller plenum.

Intake runner geometry does make a bit of difference - both for the sake of wave resonance tuning and inertial ramming effect. A longer runner will give you more of a momemtum ram effect, but it will take longer to get the flow started, so if you tune the length to create a positive pressure at the valve at IVO you can regain some of that loss.

One thing that is largely overlooked by a lot of teams is plenum shape and volume. The more open area a cylinder has to draw from, the less pumping losses you will generate (by having to draw more air in through the restrictor, increasing wall friction, etc. Simply, the more air is available around the runner, the easier it is to draw it in. In this case, shooting for as spherical a drawing volume as possible, and elevating the runner inlets off of the plenum floor.

You do lose some inertial ram effect compared to the methods used by those teams without much of a plenum, but a larger plenum also dampens pressure pulses through the restrictor, reducing wave interference and allowing a smoother flow through the restrictor. Any time you can minimize pressure waves acting on the restrictor lessens the chance you'll develop a standing wave or interference that impedes flow.

On top of all this, almost any intake design will get you good power with a proper tune - and I believe that is the biggest source of power loss than any 2% gain you'll make with geometry changes. Our engine guy last year (and temporary Chief Engineer this year) developed and is still developing an auto-tune program that takes DAQ AFR data and modifies the ECU map accordingly - it has made every car run smoother faster. If you have the choice to forgo a month of intake design for a month of engine tuning and drive time, DO IT!

Unless you run ridiculously dimensioned induction and exhaust systems, you'll be competitive if it is well tuned.

[TMichaels]

Wesley, I agree to your post, but you forgot to mention that an increased plenum volume will affect the throttle response of the engine.

Regards,

Tobi

[Wesley]

Tobi,

True, it is a tradeoff. Though for most "reasonably sized" intakes, it's not a big deal. We've gone up to 2.5L without there being a lag noticeable to our best drivers. You can fix that by tipping into the throttle a little earlier though...

[Mikey Antonakakis]

Bringing this back again...

Last year, I found myself being the engine tuner for our team. I did not design the intake and exhaust, and no one else on the team last year designed them either. It was designed by a guy from two years ago who quit before last year. I am now in charge of intake, exhaust, tuning, acquiring a dyno, etc.

So what I learned about last year's design is that there is a HUGE drop in VE at 5000 rpm. About 30%. I know this because I spent several hours trying to perfect our AFR at full throttle on a dyno. I COULD NOT get it very good at 5000 rpm because of this huge drop in VE and because we had a 16X16 fuel map. I think we're still running rich there.
So over the summer I was skimming Design and Simulation of Four-Stroke Engines, and towards the back he was describing characteristics of different muffler types. All but basically a straight-pipe (no noise deadening, basically like an F1 exhaust) created a significant decrease in torque at 5000 rpm in the engine he was using in the example. This got me thinking that maybe our stock bike muffler was the cause of this decrease in torque. Any ideas? I know that it is probably just coincidence, but this loss of torque had a noticeable affect on acceleration... If we didn't keep the engine over 5000 rpm on a launch (which was tough considering our about 11:1 final drive ratio in first gear with 20.5-inch tires--first gear topped out around 50 mph), we would have to wait for what seemed like an eternity for it to climb a few hundred rpms higher into an area of the fuel map where it was about to get a decent AFR. I want to avoid this this year, because we still aren't running an ECU with a very high-resolution map (in fact, probably lower, we're doing Megasquirt most likely).
Oh, and for reference, we ran an F4i with a tapered-log-style plenum attached to equal length (not sure what length)intake runners, and not-really-equal-length exhaust primaries into a 4-2-1 collector where the three 2-1 collectors were directly attached, with a very long tube connecting the collector to the muffler (a baffled-thing with three tubes inside or something).

EDIT:
here are some pictures of last year's setup:
http://www.columbia.edu/cu/sae/car/18.JPG

http://www.columbia.edu/cu/sae/car/11.JPG{/IMG]

Here's the best shot I can find of our intake:

[img]http://photos-c.ak.facebook.com/photos-ak-sf2p/v286/238/5/124249/n124249_34829466_1755.jpg

[Wesley]

Mikey,

We ran into almost the very same problem when we first developed our intake and exhaust package for '08.

We had a similar problem, we had a "wall" where under about 6500 RPM, it would lug badly and then wake up and take off. It was especially bad during skidpad testing, because that transition RPM was right at the brink of breakaway speed, so you'd stomp on the gas to get the revs up, it'd start to step out, you'd let off, and it'd fall back into the twilight zone (as I've just taken to calling it)

It turned out to be a timing issue. In fact, most everything I've discovered in regards to revving issues is timing related. We had timing too retarded by misaligning our trigger wheel, so it was off 15 degrees or so. It manifested in free-rev at less than 50% throttle by revving up to 6000 RPM then dropping back to 4000 in a repeated quick oscillation.

That intake looks like it's good enough to at least make power. Any design that's not ridiculous will make 60HP on a 4-cyl, and if it isn't, it's a tuning issue.

What ECU are you running?

[Mikey Antonakakis]

Last year we were running the PE ECU. This year we'll be running Megasquirt, not sure whether we'll be doing I or II, because we'd like to do MS/Extra, and I've heard that Megasquirt I/Extra has had a lot more development than MS II/Extra. If it were up to me, we would be running a more expensive ECU, but we're pretty badly limited by budget.

I had at least triple-checked our timing wheel last year. PE has a setup guide for the F4i and they include a picture that is very straightforward on how the wheel should be positioned. We may have been off by a few degrees, but that would be something that would have been easily fixed with an adjustment in the ignition table. I'm almost certain timing wasn't the issue. Here is a picture of our best dyno pull, we made plenty of power, I like to think of our 6000 rpm threshold as "Hyperspace," because it was so incredible when we first experienced it (with a shoddy tune). This pull was done with an AFR curve as flat as we could make it at full throttle. Just for reference, I'll also post a chart of our injector duration at full throttle that got us the flattest AFR curve.

Okay, so I lied, it's more like a 20% drop in VE, but possibly more depending on the AFR at those points.
http://i7.photobucket.com/albums/y294/hhspunter/CIMG2677.jpg

Here is our fuel map:
http://i7.photobucket.com/albums/y294/hhspunter/2008FuelMap.jpg
The 2D chart at the bottom is full throttle pulse width. We didn't have time to really tune part throttle, I just did the best estimating I could, and we made one half-throttle pull.

[Mikey Antonakakis]

OOOO another thing I just thought of. The guy that DID design the intake on last year's car at least wrote up a design proposal. At least with simple CFD analysis, the 3D model had as much as a 13% difference in mass flow between runners. That, coupled with a pretty much untuned exhaust, coupled with a big muffler, may explain it? We ran 4-2-1, so if t.... crap I just lost my train of thought... oh well.

Point is, we are going to be running a more symmetrical intake this year.
I know this is mostly off topic, but for an intake plenum where the runners meet the plenum in a "square" layout instead of a "linear" layout, what effect, if any, does the distance between runners have? Also, how does the bottom surface of the plenum affect things? Like if it were a flat surface with the runners extending into the plenum, or if it were a smooth transition from runners to plenum... If no one wants to answer this question, that is totally fine, I'm just being lazy now.

[Wesley]

We've been running a PE for 3 years now. For what we do with it, we have no need for more complexity. It's cheap, it's easy, and it works.


A CFD-only designed intake is the least useful. This is where I've disagreed with judges in the past, who ask if I've flowbenched the intake.

CFD and flowbenching gives you steady state numbers, which have little bearing on the actual conditions within an engine. A pressure wave simulation, like WAVE, will give you way better results, and then you can refine your geometry with a CFD program. We had a CFD-designed intake in '05 and I think we made 35HP. Now I won't claim that it was a good CFD design, but...

Are you talking about a front or back entry intake instead of a side-entry intake? (square vs. linear)

Both can be successful, but to maintain even cylinder distribution with a side entry you have to design your taper well.

A rear entry intake is much easier to get even distribution.

There are many schools of thought on which intake-runner transition is best, just take a look around competition. Ours is the first that you mentioned - the runners are elevated off of the plenum floor, because it increases the total available drawing spherical volume around the runner inlet. It lessens the inertial ram effect than of a smooth plenum to runner transition, but an elevated bellmouthed inlet decreases acoustic impedence making pressure-wave tuning more effective.

Renault built an FSAE engine claiming 105HP or something and they recommended something around 1" in every direction from the runner diameter be open plenum, with 2 or so to the roof of the plenum.

However, some very successful schools simply have a transition from the plenum to the runners in a smooth path, maximizing inertial ram effect.

It's a tradeoff. Just pick one, then be able to justify it.

Here's a version of our 2008 map and accompanying power graph - this isn't our final one, so ignore the bumpy bits, but it's the right shape.

http://photos-e.ak.facebook.com/photos-ak-snc1/v350/181/98/9619773/n9619773_38070444_1103.jpg

http://photos-773.ll.facebook.com/photos-ll-snc1/v274/181/98/9619773/n9619773_37284567_9168.jpg

[Mikey Antonakakis]

I'm with you on the steady-state thing being mostly useless. I plan on doing a square layout, probably with the air filter sitting under the rollhoop somewhere. I've got one runner configuration modelled up in Pro|E, which was a big pain. I wasn't even really trying to get equal length runners, just to get the general shape of the bends in the runners. They currently have no spacing between them, which is a more complicated design as far as the bends go. I did this so it would be easier to go back later and change to add spacing between them (~1.5" between runners would make all of the bends coplanar). Needless to say, just by freehanding it, not really trying to get equal length, the first attempt they were within 1/16" of each other, so I lucked out.

And good explanation of the tradeoff there, again, very helpful.

If I may, I still know very little about the subject, but your ignition map looks pretty similar to ours. But I think both of us may be missing a key aspect of the shape, and that is that as load increases, you have to decrease timing. At least that is what I gathered from megamanual. Their maps have as much as like 35 degrees advance at "idle" conditions, ie low load low RPM, and it drops to about 15 degrees as you snap open the throttle, ie low speed high load. At high load, it increases with RPM until advance maxes out at high load around 3000 rpm (for street cars). Then as you decrease load at high RPM, you INCREASE timing. Something I was not expecting. I would copy paste, but I don't want to keep going and get yelled at for going off topic.

Thanks for the help.

Oh, PS--- how do you like your dyno? I'm assuming you have the Land and Sea setup?

[Mikey Antonakakis]

Also, and more info or links about this Renault engine? That is pretty darn impressive.

[Wesley]

Alas, no links. I just got info from a previous member of the team, and never saw the paper.

But also, remember that street cars aren't restricted, and don't suffer from increasingly worse cylinder filling (and thus slower flame speed) requiring the increase in timing at the high end.

But yes, there's much room for improvement in our timing map.

[PBnG]

Originally posted by Mikey Antonakakis: Okay, so I lied, it's more like a 20% drop in VE, but possibly more depending on the AFR at those points.

It appears the drop in VE is about 33% according to your map. However the torque drops only about 21%.

[Mikey Antonakakis]

Originally posted by PBnG:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Mikey Antonakakis: Okay, so I lied, it's more like a 20% drop in VE, but possibly more depending on the AFR at those points.

It appears the drop in VE is about 33% according to your map. However the torque drops only about 21%. </div></BLOCKQUOTE>

Yeah, that's more of an issue with not being able to make the fuel curve match the VE curve, because our resolution is one point per thousand RPM. Neither curve accurately portrays the VE. I have a feeling the torque at 4000 and 6000 rpm would increase with better resolution. ButI think it's safer to go with the torque curve to estimate VE, because with our low resolution, I had to exaggerate the pulse width at 4k, 5k, and 6k to get the best AFR between 4,5, and 6k.