Twin Scroll Turbo and Headers
 

OK, I have read about a twin scroll turbo and why it is an advantage many times and never quite got it but I think I have had a brake through moment.

First, a 4 stroke motor must turn 720 degrees to complete a full cycle of every cylinder.

Second, a stock 930 cam is open for 200 deg of exhaust duration, an SC 220 deg, and a C2 230deg.

Thus, in 720 deg of motor rotation there is an exhaust valve open at least 1200 degs of the time! That is , 40% of the time, or more, tw

That means that 100% of the time there is a 40% or more over lap where two exhaust valves are open at the same time with a 6 cylinder motor!!!

On a connected exhaust manifold, any time there is overlapping of exhaust events, pressure will be elevated in the exhaust manifold and this will inhibit the exhaust cycle of the closing valve.

Now the fun stuff:

If we split the motor to into what is principally two 3 cylinder even fire motors things can improve dramatically.

With a split system having three cylinders on each side we then have an exhaust valve open 600 deg out of 720 degrees of motor rotation.

This gets us no over lap of to exhaust events!

Now we can get more exhaust out of each cylinder for less contamination of the incoming charge, it will run cooler, and we can even run more timing for a further increase in motor efficiency (VE) at the same boost levels.

Potentially important stuff with significant opportunities to improve our performance.

Cool!

I suspect that if we identify a proper sized split scroll turbo and divorce both sides from each other we will see an improvement in performance and I now understand better why.

Of course a fully split system would be the best. That being a twin turbo system.

A split turbine turbo on an equal length would also be a best of the best system and easer for most of us to obtain.

Following this further and working toward divorcing the two sides of a 911 motor could generate several new products opportunities.

By just modifying one's old B&B style headers one would be able to fully eliminate the 'overlapping of exhaust events' for a potentially solid bump in performance.

Even the stock J pipe 930 system. If we replaced just the J pipe with separate tubes and waste gates to the split turbine turbo we would also be eliminating the overlapping exhaust events issue.

I suspect even on a stock 3.2 NA Carrera converting from a joined to a split system, including muffler, could make for a bump in performance.

A twin scroll is a twin turbo with a common compressor. The power pulses from each bank are both transferring exhaust gas momentum to turbines linked on a common shaft but at different times so the spool-up is more linear and slightly faster. This benefits the compressor on the same shaft.

Flieger,

Agreed.

The benefits told are it improves spool and also that it extends the turbine's efficiency up top.

I had read that many times but did not really feel like I got it other than it was a mechanical advantage at the turbine.

Not that it also might make for in increase in VE.

Splitting the two banks on a flat 6 seems like it make a difference in clearing the cylinders over a joined system.

Keith this is what we did with my headers. If you take this line of reason one step further and add equal length primaries and secondaries you not only alleviate overlap but line up the pulses resulting in a tuned system. This style of header makes great power as we have seen, and a beautiful sound. You can utilize the benefits using a conventional turbo as well. The key is to have the plenum divide completely flush with the turbo flange. Many headers stop short of this.

http://forums.pelicanparts.com/uploa...1275451061.jpg

Brian,

I keep learning but here is what I believe at this juncture:

A common manifold equal length system by its self can not stop cylinders from seeing exhaust pulse overlap on a 6 cylinder. --- It is mathematically impossible.


Equal length seems a solid improvement.

A fully split system seems a solid improvement.

Equal length with twin scroll turbine would seem even better.

Subject to there being an ideal split scroll turbo for one's goal and application.


Only if there there are divided wast gates and the turbo is a true split scroll design.

No you do not have to divide everything to see some benefit, there are benefits to using any part of this design with a conventional turbo.

OPTIMAL is a different story and we have done custom headers that use it all. That includes equal length primary and secondary, dedicated waste gate to each bank, split plenum flush with the turbo flange, and the twin scroll turbo. Essentially you have an equal length twin turbo system that comes together at the turbine outlet.

Your system is in the best position to make a split scroll equal length happen.

It would be interesting to see a controlled test comparing a non equal length split and an equal length split system.

Though I do not see anyone going to the effort.

However, Turbo new car makers seem to be going to split systems on most there 4 & 6 cylinder cars.

A complete split system and true equal length tubes per cylinder are separate designs and each has unique merits and influence on power delivery.
You can compare equal length vs unequal and split vs merged and then extrapolate equal split vs non-equal split. I'm dead certain that equal split will yield the best results. How much better depends on many things.

Brian,

As you just said, equal and split are two distinct and different systems and I agree.

We both have now stated that combining both technology's has a high expectation of being synergistic. I to firmly accept this.

However to me, though it may be a good idea, adding the division between the secondaries just at the turbo as you pictured above has nothing to do with any of the benefits associated with a 'split scroll' system.

I can see that it might positively benefit reducing turbulence at the collector.

The attributes associated with a split scroll system are 1) separate intersection of the exhaust pressure's at the turbine wheel and 2) no common connection to that point. One might try associating just one of these attributes alone in the hope of benifiting. However, it probably would not be prudent to try to guess or 'exrapolate' if one might have any positive effect without the other.


We understand you make great headers, we all know you make a great 'trully' equal length system, and thanks to your efforts I think most here should know you have the flexability to make them in to a split system to. We are lucky to have you to supply such and I hope you are very successful so they will keep being available.

However, that is not the question but thank you for reminders.

The question is -- Twin Scroll, better or not, by how much, and 'why' is it better?

Twin scroll is best...period, where possible.

A fully split AND equal length exhaust system is the only (feasible) method for harvesting maximum exhaust pulse energy to improve efficiency and spool performance (there are theoretical methods for achieving this otherwise, which I won't bore you with). Just like wrapping your exhaust to insulate it and keep more heat energy (and minimize exhaust gas contraction and thus keep gas volume at its highest) in the exhaust tubes for best performance, maintaining the sine wave integrity of the exhaust pulse with a minimum of wave clipping (which destructively misapplies this energy in directions you don't want it), gets more energy to the turbine, as opposed to wasting in places of no or (negative) benefit to our purposes.

I have read a lot of technical discussions and documentation on these points, which is where I draw the above high points from. The Subaru and Mitsubishi enthusiast sites are full of deep and technical to light and anecdotal information on twin scroll; I'll post links here as I dig them. For now, here's a brain dump of "the state of the union of twin scroll" today:

There are numerous permutations of turbos that can be described as "twin scroll", all of which have different levels of 'discreteness' in terms of how clean they keep the exhaust pulses in pursuit of extracting the "free" additional energy in the exhaust stream. As a result, by designing the exhaust system to harvest unmuddled exhaust pulse energy to drive the turbine more efficicently, you get quicker spool.

This requires an equal length exhaust system that mates the cylinders that fire at even crankshaft degree spacing from one another. The Porsche 911 and Japanese 4 cylinder pocket rockets are perfect for twin scroll turbos in this regard. On the flip side, you will never get a twin scroll system to work on a 5 cylinder Audi (without some truly insane engineering wizardry), nor will you be able to use two twin scroll turbos with any positive effect on a 930TT. A V8TT on the other hand, would be able to.

The next step beyond simply keeping the exhaust discrete until it hits the flange are turbo designs that divide the exhaust stream past the flange, then comingle exhaust gases just before the turbine - maybe a few centimeters before the turbine. The longer the gas signatures stay discrete, the less energy is wasted.

Taken even further, the latest twin scroll designs not only have a divided housing, but actually have completely segragated cavities which ensure that the exhaust streams stay competely separate until the gases are actually in contact with the turbine wheels. The latest Mitsubishi and IHI turbos literally keep the exhaust streams separate until just a few thousandths of an inch before the vanes of the turbine wheel, so in reality aren't ever in contact until after they have crashed into the turbine, exactly where you want all available energy to be expressed (see the Mitsubishi Evo VII/IX/X or Subaru WRX STi).

The final (theoretical) permutation would be any flavor of the above twin scroll housing with variable geometry, whether single vane, dual, linked vanes (one for each scroll path) or even (although I have never seen anything like this) dual vanes with independent pitch control.

Net-net, more energy expressed in the direstion you want to apply force is always a good thing. This is what a twin scroll turbo designer's ultimate goal is.

SmileWavy

Outstanding.

Excellent info Kenik!

Also I am thinking, as long as there are three or less cylinders per split turbo feed, the cylinders on said feed are equally spaced, and the exhaust cycle duration is less than (720/3) 240 degs -- we also get an increase in Volumetric Efficiency in the cylinders because of the lack of overlaping common exhaust events. That is all the other exhaust valves will be closed anytime one is open.

Double whammy!!

The best split system is probably a twin turbo system.

As you notes imply, all this is just an exercise if we do not have good Split Scroll turbos to choose from???

Any referance info is appricated.

Thx!!!

My pleasure to chip in. I spent a looooong time researching turbo tech for my upcoming 2.3L turbo build. In the end, I went with a Mitsubishi Evo IX TD05-16G with 10.5cm hotside for all of the reasons above. Of course I had Chad Block and Blouch modify it into custom 650 cfm, 400+ HP capable beast that hits 1 BAR @ 2500 RPM on a 2.3L motor and blows 1.5 BAR+ without boost taper to 8000 RPM. :D

I agree with you that Brian's headers are by far the best 911 Turbo header on the market. Once my short block is done, these will be purchased.

Yes, TTs are a great alternative if you can't twin scroll. I mapped TTs for my motor and the boost curves and transient response curves I extrapolated from existing data looked identical. Draw your own conclusions... smilewavy

Keith, I wasn't saying these things because I like to type. The reason I took the time to share my experiences is to show that there are benefits to using even partial bank separation, and the benefits increase the closer you get to complete separation. This helps in understanding fully the benefits of a true split scroll system. These are results that I have witnessed in the evolution of my header system. Below is a picture of my Generation 1 flange. You will note it does not have a split volute. With that change alone I saw a difference in both how the engine ran and sounded indicating increasing efficiency.
The things Kenik is talking about take this evolution to the extreme, and you see increases in performance with each step.

http://forums.pelicanparts.com/uploa...1275619732.jpg

I am not saying it dose not work.

If your sound test shows a difference we know it has an effect.

If the dyno proves it is a positive effect that is excellent.

I suspect two things going on.

First the header cross section restricts down from the secondary tubes to the turbo buy almost 2 to 1 .

Second, the secondaries are in effect at almost a right angle to each other as they approach the area into the the turbo.

The dam should help reduce turbulance and help compensate for both of these factors.

On top of this, it would seem the 'bridge' wast gate conection will already have functioned as a 'cross over' and had a significant equalization effect to balanced each sides pressure waves.

Thus, I do think the 'cross over' and 'dam' are probably both positive features.

However, I do not see how we can attribute any of the benefits of a 'split scroll' system to them.

Just my unexpert opinion.

The best.

Anything that crosses over the gasses is detrimental. For the wastegate, the crossover design is less than ideal as anything that acts as a pressure equalizer is detracting from pulse integrity. The best solution is to either use two separate, independent wastegates or use a turbo that has an internal wastegate. Gases should ideally never comingle anywhere before the compressor wheel.

Brian's system is a true equal length and not a twin scroll system.

Thus, I am guessing the cross over might be a positive on this type of a system.

A cross over connection probably helps to further reduce the peaks of the amplitude of pressure changes inside the headers. This is mostly the same thing an equal length system dose.

The net effect being lower back pressure and even more orderly exhaust.

Anything that sooths the highs and lows should let it better flow, with lower restriction, less noise, and less back pressure.

Said lower back pressure's further aids cylinder filling and increases VE.

I could be mistaken but this is what I am believing at this point in my education.

The only reason it isn't full twin scroll is the crossover pipe and a non-divided flange. Ben and I have chatted and when I am ready to buy, we'll be working together to omit the crossover and fit the pipes to my divided flange. Then the header system will be truly twin scroll.

Excellent!

Assuming a good twin scroll turbo being available, I would do the same thing!

Another thought.

We spend a fair amount of effort on the primary tube collector. Seem to not be to concerned with the collector into the turbo, and almost ignore the collector that splits exhaust to feed the WG's! Seems that on some turbo installs that run smaller turbine to compressor ratios, the WG circuit in some cases might be approaching the volume flow of the turbo.

Back to better understanding Twin Scroll.

Twin Scroll is said to be most effective on a 4 cylinder motor, there is some benefit on a 6 cylinder, and on an 8 cylinder it is not worth the effort (even on an even fire).

As an example, if we run a cam that has say 230 deg duration like a C2 cam:

Two cylinders of a 4 cylinder motor will have an exhaust valves open for 460 of 720 deg of rotation which leaves a fair amount of time for the pressure to drop between exhaust cycles and exit the header.

This would give us a lower average header pressure than if all cylinders were merged.

On a 6 we would have 3 cylinders per side with an exhaust valve open 690 of 720 deg. The time for each exhaust event to exit the header before the next is greatly reduced and the benefit lessened.

By the time we get to a V8 and have to put 4 cylinders together. We end up with exhaust valves open 920 deg of 720 deg of crank rotation. Now we an overlap of exhaust events with all exhaust cycles and much of its advantage has been lost.


Per above, all exhaust valves closed as follows:

4 cylinder, 36% of the time.

6 cylinder, 4% of the time.

8 cylinder, 0%. In fact we now have two exhaust valves open 28% of the time per side.


Cool stuff!!!

Now I better understand why a 4 cylinder sees more benefit with a split system.

Taking this ideology, why not to look putting 4-cyl car all exhaust in one single scroll turbo and then there are no time where turbo feels any pulses, like you have twin scroll for 8-cyl car?

I found this and is explains it better than I.

Found this here: Twin-Scroll Turbo System Design Explained

I found the very same good write up.

It looks I may go for twin scroll GT35 for my upcoming modification. I do plan to use 993 headers to have nice heater boxes. Only question is what size of waste gates shall I install?

For a car where you want heat that sounds like a potentially great idea. I think Brian (RL8) and Ben (M&K mufflers) can make that happen.

Of course equal length would be better however, it seems many of the new car makers are going for the twin Scroll without the equal part. Thus, draw your own conclusions.

It should also be a very low volume system and spool nice and fast.

Not an expert but I think two 38mm Tial's has a bit more capacity than a stock 930 WG and should work well.

That article is terrible, and the writer is using a bunch of terms that he/she does not understand.

For a twin scroll system to work, each pipe must be kept separate until they reach the turbine inlet flange in order to conserve the momentum of the pulse. This is more important than equal length. You can not have ANY wastegate pipes. When they are closed they provide volume for the pulses to expand into. The ‘secondaries’ of Brian’s headers will do the same thing, so that system is not useful. With no wastegate pipes, an integral wastegate turbo must be used.

The part about the scavenging and overlap in the article is incorrect. They have never run this engine. I know for a fact that this engine runs very little overlap, except for when they want to force exhaust INTO the cylinder for EGR purposes.

This is a picture of the exhaust manifold. It is somewhat misleading. The individual pipes are actually in layers, so the cutaway makes it look like the ports dump into a single volume. They don’t. There is an individual pipe from each cylinder all the way to the flange. Beautiful.

http://forums.pelicanparts.com/uploa...1275789976.jpg

Here is the dirty secret about divided systems. ALMOST ALL BENEFIT IS LOST WHEN THE WASTEGATES ARE FULLY OPEN. This is not a maximum power enhancing approach, it improves the low end, but the maximum power is about the same.

+1

If you do a Google search on a phrase of the article, "twin-scroll design is better pressure distribution in the exhaust ports and more efficient delivery of exhaust gas energy to the turbocharger’s turbine", for instance, you find it plastered all over the internet. Had a difficult time even finding the author who lays claim to the info. It guess if you plaster enough crap everywhere, people start believing it as fact. Ah, the power of the internet. ;)

Dave I get what you are saying and may use that technique in the future. I to was wondering where it came from and thought the artical I referanced was the first place said quote was used.


SS,

I am still learning.

It is not the best writing but thought it better than mine if you labored through it.

I will look for some more reference but that article reflects the major points I am thinking are the keys to a split system.

The big one being, it keeps there from being two exhaust valves from being open at the same time which can reduce pressure at the only open exhaust valve on that bank. In turn this makes for better cylinder filling.

The other is it creates an advantage at the turbine by how the exhaust relates to the wheel.

They get a little deep saying we can increase over lap.

My take away from that I might put like this: A C2 cam has a small amount of over lap. A 930 has none. I am guessing he is saying that if there is less pressure in the cylinder when the exhaust valve closes, they can run a little bit more overlap to a benefit. Overlap if used right should increase HP for two reasons. Longer duration and cleaner cylinder fill.

I do not think comparison to scavenging on a normally aspirated race motor made it any clearer though I get his point.


I love that picture of the BMW 6 posted.

What I see is two banks to a split turbinee. Each side has log style manifolds inside an encasement not unlike our heat exchangers to be more thermally efficient, be easer to build, and weigh less than a cast iron system.

They seem to have sized the turbine such that they can control boost by only bleeding off one bank.

All cool stuff.

What am I missing or getting wrong please.

Thanks if you can help.

Just thought, I suspect the BMW has active cam timing and can vary overlap by rpm and load.

Might be it can work with the lower residual cylinder pressures that come with a divided turbine system and get more aggressive with valve overlap at times.

I can not see how BMW or anyone could have made a lower volume manifold. Further, putting the WG in the turbo eliminates the volume of the WG circuit.

(Maybe a VAT)

I now understand why an equal length header has more even cylinder temps. More even pressures when the exhaust valves closes. This also allows a motor to be run closer to the optimum timing and AFR with out fear of having a cylinder or two detonating.

Speedy Squirrel & Dave,

You seemd to think the source I referanced above was junk.

In my effort to learn more about the new BMW motor that hits full TQ by 1200rpm (and near full boost). I found this:

See:TwinPower: Modified Shows How a Twin-Scroll System Works - BMW Monitor

He seems to think a bit more favorably of the quote and aritical. He went on and said the following that helped me further.

I am not sure if the N55 (twin-scroll) uses this but the naturally aspirated inline 6 uses Valvetronic and double-VANOS. VANOS is the varaible valve timing and Valvetronic is the variable valve lift system which eliminates the throttle.

Gee, you searched the internet found someone who agrees with you. I'm happy for you. :rolleyes:

Dave,

Once again you only seem intent on insearting yourself with emotional response's only to do this provoking thing!

Flieger,

You are right. After my comments above I did some more research and BMW's new motor variant is truly state of the art.

It is there first time combining variable valve timing, direct injection and turbo charging and some other trick technology's like controllable and variable oil pump delivery. It is able to run at over 10/1 CR on top of running a turbo.

In all, there efforts resulted in over 300# TQ by 1500rpm (apx 200rpm earlier than there twin turbo 6) and about a 7% increase in fuel economy.

Basically this little gem makes full boost by 1500rpm and pulls a 7000rpm red line.

It also makes more TQ and it makes it sooner than Mercedes E class diesel.

Cool stuff and it is interesting they did this by leaving the twin turbo behind for a divided turbine set up with a log exhaust system.

Dave, is the issue simply that only one reference was cited and that you need hard, empirical evidence or that you think twin scroll technology is without merit?

I think we'll all agree that anything that allows an engine to optimize intake and exhausg velocity gas velocity throughout the rev range is goodness. While I haven't read any PhD these or industry papers on twin scroll technology and variable valve train management, the overwhelming practice of these technologies by massively capitalized companies who make their money by selling optimized solutions tells the story.

The newest turbo engines from the big boys are all going twin scroll on single turbos with variable valve systems. Having talked to camgrinder at length on the effect of lobe spacing, overlap and limiting exhaust duration/lift low in the RPM band to increase boost response was an eye opening conversation.

The best minds with the deepest pockets betting the farm on these technologies are enough for me, especially since my ability to dissect the math isn't there. I accept E=MC^2 not because I can dissect the equation, but because I can understand the intuitive bases for this and also see that nuclear weapons work. I accept these automotive technologies for the same reasons.

The Mitsubishi Evo X and Bimmer turbos are the equivalent of 'the bomb' in the automotive context.

That BMW engine is a bit too hot to run at the track IIRC. Nice technology still.

Is a twin-scroll turbo system more efficient that having two separate compressor/turbine assemblies? I want to isolate the power variable here- how much waste energy is recovered. I realize that a flat-6 packages a separate system quite nicely while an inline engine would like a twin-scroll.

That is mainly an issue with the N54 two-turbo engine in the 335i. They do not put much of an oil cooler on it and at the track the temps get up to 250+F and goes into limp home mode. Dinan puts a bigger oil cooler and ducting on and makes a nice track engine.

Let me take a moment to explain the above technologies and the benefits they impart on the turbo motor and why:

Direct Injection:

• The 997 DFI injectors operate at 1750 psi. If MFI injectors used to be the holy grail in terms of atomization, injectors that operate at 5 times that pressure are a wonder.
• Additionally, injection into the combustion chamber means that the fuel's full evaporative capacity is absorbed by the combustion gases, providing better cooling of the combustion process, as well as quenching rogue combustion elements.
• With these injectors ability to control full directional trim control and withy the ability to fire more than once per cycle based on engine state, means that the fidelity of controlling warm up, knock and other variables is downright unbelievable in terms of tuning flexibility.

Variable valve timing:

• Controlling overlap by varying lobe spacing is huge: you can go from zero overlap at idle to race overlap at high RPM. At idle, you can build pressure extremely quickly as you can optimize the intake tract's VE to the needs of an engine in terms of air requirements at level that is functionally infinitely variable. This is great for emissions and amazing for power production. No more compromises on overlap means that you can have your cake and eat it.
• Controlling lift is even more valuable for turbos. By lift limiting the exhaust valve at low RPMs in relation to the intake valve, you can create a massive pressure differential on the exhaust side that creates significant dividends in the integrity of the exhaust wave heading toward the turbo. This translates into ridiculously low boost thresholds. You can basically use a full race cam, but make it pretend to be a cam that if static, would be useless above 3000 RPM, as it would choke off top end power for anything but a low revving diesel.

As far as the "log exhaust" goes:

This is an educated guess, but follow the reasoning: With variable cam timing and lift, at low RPM, the effects of a cam tuned for no-overlap and low lift can provide perfect exhaust wave resolution, then open things up higher in the RPM band for excellent HP production. In an engine like a 911 that has to compromise on a given, fixed cam timing spec, this resolution is not possible - any cam that would exhibit a good exhaust signature at low revs would run out of steam very quickly without being able to increase lift and overlap.

Thus, the reasoning is that at a low number of relative exhaust events over time (RPM), coupled with no overlap and high intake velocity (due to lower valve lift at low engine speed) allows the log type of exhaust to operate as well as a discrete, equal length system. BMW can make incredible low RPM power with a log style header that would not be possible in a fixed cam motor and by the time that RPMs rise, to make full power, the benefits of equal length exhaust tuning are gone (due to gas volume overcoming any synergistic effect that could be derived from pulse tuning/scavenging).

And that's the latest hail mary from an armchair ME wannabe... :D

You guys really need to test drive a BMW 335i to appreciate it.

I've had a variety of high-powered cars in my life and I don't impress easily, but every time I get in my wife's 335xi it impresses me, sewing machine smooth, almost instant torque, revs clean to redline. Damn fine bit of engineering!

Thanks for your input. That's exactly what I was asserting. To use the BMW engine as an example of an efficient twin scrolling exhaust system, grossly over simplifies what BMW accomplished. It's an "apples and oranges" scenario. The BMW engine has direct injection, a more efficient combustion chamber design, variable valve timing, higher compression, PLUS twin scrolling exhaust manifold/turbochargers. It's remarkable the torque this engine produces below 2000RPM...my former boss has one in his 1 Series and that thing is a rocket!!!

But let's keep in mind that engine is producing 100 horsepower per liter and most of us are already beyond that. Twin scrolling will produce more efficiency at low engine speeds, but there is no advantage to twin scrolling when an engine achieves higher engine speeds (above 3000RPM, for instance).