PDA

View Full Version : Turbo Exhaust Design



bernmc
17-03-2007, 04:30 PM
Since getting my dirty great 3inch fujitsiwotsit system I've been thinking about turbo exhaust design.

A long time ago I made some enquiries about a full twin pipe system - separate pipes from each turbo back - and also a 2.5 vs 3inch system. The exhaust bods I spoke to (who seemed knowledgeable) told me that the twin system would definitely lose power, and that 3 inch would be pointless until I was in the 350bhp region. Something about this has never really rung true, but it's only now that I've been reading up on exhaust design that I've worked out why.

The problem was that the bunch I spoke to (in common with most exhaust shops in then uk I suspect) are probably very experienced with exhausts for Normally Aspirated engines, but have very little knowledge when it comes to turbo systems. It turns out that the design considerations are in fact very different.

Exhaust flow is enormously complex, but I've tried to simplify things to get my tiny little mind around it. Each 'pulse' of exhaust gas from a cylinder travels down the pipe as a wave. In a NA engine, you want all those pulses to line up nicely, one behind the other at perfectly spaced intervals. The situation becomes even more complex when those pulses exit the pipe, as they send a reciprocal wave traveling back up the pipe. Ideally, you want those reciprocal pulses to be perfectly spaced between the pulses coming down the pipe. Next, for a given pressure, the velocity of flow in the exhaust is greater through a narrower pipe versus a larger pipe. NA exhausts use this fact to improve 'scavenging' of the exhaust gasses: Get a pulse flowing down the pipe at high velocity, and it 'sucks' the pulse behind it along. The downside is that the narrower the pipe, the greater the backpressure there is in the pipe.

So 'tuning' a NA exhaust involves getting all the pipes from each cylinder to join up after equal lengths - to keep those pulses in a nice neat row - and finding the optimum diameter pipe from then on to get the right balance between flow velocity and backpressure. Too big a pipe loses velocity, and greatly reduces the exhaust efficiency.

So all those pumped up novas with huge exhausts are almost certailnly down on power compared to the std setup /lol

For turbo systems, however, you have to throw all that theory out of the window.

In a turbo car, the important thing is boost. Boost is generated by the turbocharger. The 'shaft power' (the energy generated by the turbine) of the turbo is determined by the pressure difference between the turbo turbine inlet and outlet - you want as high a pressure as possible going into the turbine, and as little pressure as possible on the outlet. A bit like running down a hill - the higher the top of the hill, the steeper the slope, the faster you run. Any backpressure after the turbo reduces the pressure ratio (makes the slope gentler). and reduces shaft power. Enough backpressure can even stall the turbo completely.

So for the exhaust header - the bit before the turbo - you want to get the exhaust velocity as high as possible as quickly as possible, so you actually want short, narrow tubes leading into the turbo. You will gain far more by doing this to increase turbine inlet pressure than by messing about with tuned lengths etc. The faster the gas flow, the quicker to turbo will spool up. A well designed trubo header will have pipes considerably narrower than a NA car with considerably less power.

Obviously, you don't want ot go so narrow that you 'choke' the engine, particularly at higher revs.

After the turbo, you want as little backpressure as possible - everything else is virtually irrelevant. The bigger (cross section), shorter, straighter, the better. I've read one quote from a garret engineer who says 'stick a hoover on the end of the exhaust if you can' :)

In terms of diameter of turbo systems, it seems that 2.5" is good for up to 250 BHP - beyond that you need 3". If you're making more than 400BHP, even 3" is on the small side... :o

Other things to avoid are:
sharp bends (tight radius)
compression bends - use mandrel bends
sudden changes in diameter - avoid all together, or make them gradual

The optimal shape for the pipe from the turbo outlet to the desired exhaust diameter is a graduall cone shape with wall angles of 7-12 degrees. (Actually, a hyperbolic increase (like a trumpet) is supposed to be better, but gains for the added difficulty in construction would be minimal).

Naturally, you have to play all these 'desirables' against what is practically possible and cost effective.

So downpipes that look like this:

uploaded/1552/1174145136.jpg

May look like the dogs danglies, but are probably an expensive waste of time - they're based on NA reuqirements. You're better off having a smooth-walled pipe with the minimum bends. That tight radius bend in the rear pipe will generate an enormous amount of back pressure, as will the disrupted airflow over all those welds.

The cat is the one thing I haven't mentioned yet. Again, it's essential to have as liitle flow restriction as possible through the cat (and any mufflers/backboxes etc), so a free flowing 'sports' cat is essential - otherwise you're wasting your money just doing the exhaust.

So there you go. Bigger is better.

I'm not an expert, so I'm happy to be corrected...

ralliart-type-s
17-03-2007, 08:23 PM
I've often wondered if the two banks having their own separate exhausts would be viable and by the sounds of it 2.5" per bank would equate to a 5" overall bore( for the engine). if this was the case you would need 400bhp plus, so would you just have the two exhausts made out of 2" for 300-350bhp?? and instead of silencing it, have a small sports cat per bank?? this would make the system environmentally legal and still free flowing cos less gas is trying to get through them.

Nick Mann
17-03-2007, 08:50 PM
Doubling the diameter is more than doubling the volume. Be careful! Two 2.5" diameter pipes is no-where near one 5" pipe!

The interesting thing with our cars about a true twin pipe system would be the lamda sensor. The ecu uses info from that to tune the car at low throttle, AFAIK. If you split the pipe, how do you get a decent measurement? If one bank leans out, and the sensor is on the other bank, you'd be in trouble, wouldn't you?

Lots of good info there, though Bern.

bradc
17-03-2007, 08:51 PM
Bern - great post mate :D

With the stock ecu it is designed to pick up a reading from the oxygen sensor in the exhaust just before the cat. The only way you can go to a complete two pipe system is if you leave the O2 sensor in one cylinder bank, but have a pair of wideband AFR gauges in the cabin so you can monitor and check that both banks are fine. As for the size, the outlet of the turbos is about 50mm, or 2 inches, depending if you're a backwards bastard or not :)

A good solution sounds like a pair of 2" pipes from the turbos down to the rear of the car doesn't it?

WRONG!

For a start if you have a 2" pipe and bend it 90 degrees you will have as much resistance as 1 metre of straight pipe (something a physics teacher told me in 2001) Next thing, even a perfectly straight piece of pipe has some resistance. Try breathing through a normal length straw, it is rather difficult. Then try cutting the straw down to only 1" long, you will find it is a lot easier to breath.

Hence my reccomendation is for a pair of 2.5" pipes right from the turbos the whole length of the car, or a single 3.5" pipe coming together after a pair of 2.5" downpipes. Ralliart, you're slightly off with your calculations because the area of a circle is pi x radius^2, or in the case of what we're talking about:

2.00" = 12.56 sq in
2.25" = 15.90 sq in
2.50" = 19.63 sq in
3.00" = 28.27 sq in
3.50" = 38.48 sq in

Something interesting to consider is that the stock system is only 2.25", which is hardly bigger than a single exhaust outlet from one turbo, hence you really really want to go bigger.

richy rich
17-03-2007, 08:54 PM
Hmm interesting but , why is it then a 600+ bhp car has only 2.5" and there is no drop in performance .

3" and 2.5" can both flow a hell of a lot of gases , but all other parts of what you are correct , design , bends , shape , angles etc all play a part .

Nutter_john pretending to Richy Rich

Dan_G
17-03-2007, 08:57 PM
by the sounds of it 2.5" per bank would equate to a 5" overall bore( for the engine).

think of it in terms of twice the area not diameter... Area = Pi X Radius^2...

diameter 2.5"= 63.5mm

then Pi x 31.75^2=3167mm^2 and you have two so thats 6334mm^2 total area

then work that back Radius = Square root of [6334/pi] = 44.91mm

diameter approx 2 x 45 = 90mm = 3.5" overall bore to maintain overall area of two 2.5" pipes.

[EDIT] was beaten too it :P... thats what I get for fetching a beer whilst writing a reply [EDIT]

bradc
17-03-2007, 09:03 PM
Richy/John/Dave/whoever :P A 2.5" system will be able to flow a huge amount of exhaust gases, and may be able to support 600hp, but I'm willing to bet that whatever that car is, if they changed to a 3" system power would go up.

Wodjno
17-03-2007, 09:14 PM
Doubling the diameter is more than doubling the volume. Be careful! Two 2.5" diameter pipes is no-where near one 5" pipe!

The interesting thing with our cars about a true twin pipe system would be the lamda sensor. The ecu uses info from that to tune the car at low throttle, AFAIK. If you split the pipe, how do you get a decent measurement? If one bank leans out, and the sensor is on the other bank, you'd be in trouble, wouldn't you?

Lots of good info there, though Bern.

Interesting Interestingness :speechles

But with the system how it is now, with both banks into 1 exhaust ! And 1 02 sensor ! You are not getting individual readings from both banks or individual readings from each cylinder.. If 2 Cylinders started running Mega rich, when 1 or 2 cylinders were already running lean.. Then the 02 sensor would see the richness and lean all the cylinders out a little to bring the 02 back into line.. So now we check the AFR's and they look spot on ! But the 2 cylinders that were alredy running lean are now running dangerously lean and have a good chance of overheating and failing.. The current way our AFR's are monitored and altered are only done on an average basis.. We might have a perfect AFR trace on the Rolling road.. But in reality we've almost got rich cylinders and lean cylinders :thinking:

The only true way of having a completley efficient burning engine. Is to have individual 02 sensors for each cylinder and individual control over each injector..

bradc
17-03-2007, 09:20 PM
read my suggestion wodj, have the stock O2 connection on one exhaust, and have a pair of widebands as well so you can monitor how different each bank is.

MPBVr4
17-03-2007, 09:39 PM
Just been playing with a few numbers

A 2.5 litre engine running at 6000 rpm will draw 7500 litres/min of free air on the basis that the engine draws air every second stroke. The difference in velocity between two sizes is quite significant.

If the exhaust gas was at ambient temp (which isn't/pan ) the velocity would be
2 1/2" - 39.78 m/sec
3" - 27.47 m/sec

At 400oC it would be
2 1/2" - 91 m/sec
3" - 62 m/sec

A 3" exhaust will give a 32% reduction in velocity hence 54% reduction in
backpressure from any restriction in the system - I think:inquisiti

I really must get out more/pan :pimp2:

bradc
17-03-2007, 09:51 PM
A normal exhaust temp for a turbo car would be 700-900C under full throttle.

ralliart-type-s
17-03-2007, 10:01 PM
I must be reeeeeeeeely thick cos all those calcs almost made me stare outta the window!! lol. The job seems to be a no go then,could you have a collector at the front which would house the sensor while still keeping to two systems separate?? a sort of manhole as it were.

Kieran
17-03-2007, 10:09 PM
I must be reeeeeeeeely thick cos all those calcs almost made me stare outta the window!! lol. The job seems to be a no go then,could you have a collector at the front which would house the sensor while still keeping to two systems separate?? a sort of manhole as it were.

I guess you could use an 'X' pipe. This links the two banks but that's no better than the standard system - not sure how accurate the readings would be either.:inquisiti

MPBVr4
17-03-2007, 10:11 PM
A normal exhaust temp for a turbo car would be 700-900C under full throttle.

At the down pipes may be but not from the "cat or de-cat" back surely?:inquisiti

ralliart-type-s
17-03-2007, 10:21 PM
I s'pose the temp depends wot grade pipe is used and where you take the measurement. Different tubing will dissipate heat at different rates. Is someone gona use pi again??? I'm scared!!

bradc
17-03-2007, 10:26 PM
Kieran, yes you could, and I intend to have a cross over pipe in my system, but not for the purpose of measuring at all, I want to try it without a join at all first of all and see what it sounds like. If I don't like the sound, then I will try a cross over.

Malc, I dunno, I was referring to what I've seen on EGT gauges which are mounted as close to the engine as possible, most of the time before the turbo

Kenneth
17-03-2007, 11:53 PM
Good post bern, but I have to say that I have been saying that exact same thing for ages now.

I dont agree with your bit about the tuned length pipes being a waste of time though... I dont think you have taken into acount the exaust disruption when the 2 exhaust gasses meet. the idea is to get the exhaust gasses to meet with similar velocity and temps etc.

zentac
18-03-2007, 12:38 AM
Just for the record, I was running somewhere near 400bhp with 100bhp nos on top, on a 2.5" exhaust. I am changing this to a very short 3"

ralliart-type-s
18-03-2007, 12:45 AM
Hey zentac, is yours the recent banzai featured fto??

Wodjno
18-03-2007, 12:51 AM
Just for the record, I was running somewhere near 400bhp with 100bhp nos on top, on a 2.5" exhaust. I am changing this to a very short 3"

And the reasoning ???

fredfour4
18-03-2007, 09:59 AM
I was told that for a twin pipe system (twin turbo) you would need a balancer pipe. This then makes the one of the pipes longer. The balancer pipe is just U shaped and evens out the pulse before both pipes exit into the silencer.

If you look at Maserati BiTurbo engines, V6 & V8, there pipe work has this balancer pipe.

Is is only on Maserati I have seen this.

bernmc
18-03-2007, 04:38 PM
Hmm interesting but , why is it then a 600+ bhp car has only 2.5" and there is no drop in performance .
... no drop in performance as compared to what? You can only make a statement like that if you've run the same (turbo NB) engine with a 2.5 inch vs a 3 inch exhaust.

The exhaust sizes I've quoted are not my own - they're suggested by a chap called Jay Kavanaugh who is a turbo systems engineer for Garrett, so I'm inclined to believe he knows what he's talking about.


Good post bern, but I have to say that I have been saying that exact same thing for ages now.
Yes, but I say it better ;)


I dont agree with your bit about the tuned length pipes being a waste of time though... I dont think you have taken into acount the exaust disruption when the 2 exhaust gasses meet. the idea is to get the exhaust gasses to meet with similar velocity and temps

I agree with you to some extent, but from what I've read, the scavenging gains are small compared to simply getting the boost up as quickly as possible with short pipes - ie high velocity somewhat turbulent flow gives a boost gain that outweighs a tuned length header with better scavenging with a lower velocity flow due to the increased length. Due to space constraints, I think it's usually quite difficult to get proper tuned lengths that don't bend and twist excessively, again negating any advantage gained by scavenging.

I suspect you're probably better off spending the time/money sorting out the wastegate flow instead...?




bradc][/b] A normal exhaust temp for a turbo car would be 700-900C under full throttle.
At the down pipes may be but not from the "cat or de-cat" back surely?:inquisiti
I don't know what the temps are, but they actually increase just after the cat - the catalytic reaction is exothermic.

Since fitting my shiny new exhaust, I've noticed the following:

Pickup from low revs is much quicker - suggesting that the turbo is spooling up quicker, which fits with the theory above. The car feels so much more responsive now.
I have less boost spiking - With the std system, I was running 80kPa (0.8bar) and getting boost spikes just beyond 110kPa. I was hitting fuel cut occasionally at full throttle in 2nd/3rd. Now I've been able to turn up the BC's gain, increased the boost to 90kPa, and spiking to just over 100kPa with no fuel cut.It would be interesting to see what has happenend to the boost curve - which is why I'm tempted to run the car at the dyno day.

Kenneth
18-03-2007, 09:41 PM
Bern, how does the increased length reduce velocity?

There is not much out there that relates to twin turbo setups, mostly its singles.

I couldn't say for sure how much difference it makes all I know is the tuned length ones work very very well on my car. They also make the car sound quieter and nicer :)

stuartturbo
18-03-2007, 09:55 PM
Bern, how does the increased length reduce velocity?


the incraesed length will reduces velocity due to friction on the walls of the exhaust pressure will also increase
Bernoulli equation?

test example
2 buckets at the same height with a length off hose on the bottom 1 at say a meter and 1 at say 10cm, the water has to go through the same size hole but the short hose will flow faster

bernmc
19-03-2007, 09:25 AM
Bern, how does the increased length reduce velocity?

There is not much out there that relates to twin turbo setups, mostly its singles.

I couldn't say for sure how much difference it makes all I know is the tuned length ones work very very well on my car. They also make the car sound quieter and nicer :)

Have you got any pics of your headers, Ken? I'd be interested to see what they look like.

It's a pity we don't have a club member with a dyno who'd let us put a car on the rollers, and then work through a couple of variations in exhaust components to see what effect they each have...

Kenneth
19-03-2007, 09:32 AM
Have you got any pics of your headers, Ken? I'd be interested to see what they look like.

It's a pity we don't have a club member with a dyno who'd let us put a car on the rollers, and then work through a couple of variations in exhaust components to see what effect they each have...

They look exactly like the pic you posted in first post of this thread!

bernmc
19-03-2007, 09:47 AM
They look exactly like the pic you posted in first post of this thread!

A crossed wires! Header I have been referring to is cylinder head to turbo (ie pre turbo). the bit you're talking about is the downpipe - but I know they're sometimes called 'lower headers'.

Thing is, once the exhaust gasses hit the turbo, much of the pulsatile nature of the flow gets mashed up in the turbo, so the theory of equal length post turbo is much less relevant. Again, it's gain vs cost/effort I suppose.

Eurospec
19-03-2007, 10:19 AM
I've had this niggling thing for a while now about exhaust size. My main pipe is 3", my downpipes are 2 x 2.5 inch.

It all works, but the gas coming out the back on full chat is enough to blow over a stack of insulation blocks positioned 3 feet from the exhaust tip (Dont ask, lol!).

The maths says the gas must accelerate where the two streams merge, and it certainly comes out the back like a freakin cannon!

I cant help thinking a 3.5 inch main pipe would help, but there are people out there making more power than me on a 3" pipe. The downpipe is fine.

You certainly want the gas to have the ability to expand through the turbine, so anything you can do to alow this to happen is going to help. At the same time as that you also dont want the gas to slow down too much post turbo.

Cheers,

Ben.

Kenneth
19-03-2007, 09:55 PM
A crossed wires! Header I have been referring to is cylinder head to turbo (ie pre turbo). the bit you're talking about is the downpipe - but I know they're sometimes called 'lower headers'.

Thing is, once the exhaust gasses hit the turbo, much of the pulsatile nature of the flow gets mashed up in the turbo, so the theory of equal length post turbo is much less relevant. Again, it's gain vs cost/effort I suppose.

Ah, but I was never talking about headers... And your original post never brought them up... So I don't know how you got confused ;) lol

I dont know if my down pipes are equal length, or whether it is just done to increase the length of the back down pipe. From what I understand it is better to have them merge further down to reduce the turbulence on the turbine.

probbo
20-03-2007, 09:30 PM
Don't forget reduction in exhaust gas flow means less backpressure which can give higher bhp but in turn usually means a reduction in torque