How much swirl is too much swirl

[hoffman900]

We had swirl, tumble, and wet flow test equipment available as part of Dodge's Nascar effort from 2000 to 2012.
We measured swirl as rotational torque with a honeycomb insert and a rotary strain gauge. Tumble was measured on another fixture with a perforated plate and three load cells. Data from the three load cells gave us total combined tumble torque, and more importantly, the force location defined as distance from the bore center and clocking in degrees.
We found that the swirl and tumble combined define the helix of the incoming air column in a two-valve chamber.
As Mummert stated, the swirl tends to jump up at low lift, lag behind in the mid-lift, and ramp up again at high lift. The tumble curve tends to do the same. The tumble force location (we called it "moment arm") moved closer to the bore center as the intake valve opened, and moved away from the bore center rapidly at high lift, when the swirl was spiking.
Our conclusion was we needed a smooth, linear increase in the swirl as the valve opened; not a set value.
To manage the swirl, we worked with chamber containment of the flow cone, runner trajectory, and some slight steering with the fin behind the guide. The flow must stay attached to the short-turn; swirl and tumble go out of control if there is separation. The modern steep seat and top angles really helped.
Reverse swirl fins behind the intake guides can improve burn uniformity, BSFC, and help prevent the swirl from spiking at high lift.
If you can do some wet flow testing on your cylinder head, you will be mesmerized by what you see. It will tie in with your swirl and tumble observations.

Bryan

What did you guys see by focusing on tumble after the swirl intensity was contained? Improved BSFC? Did you guys every get as far as getting the tumble portion modeled in a CFD program with a modeled piston top? How did you guys go about managing tumble short of moving the entire intake port angle?

Thanks

The swirl control was tied in with keeping the tumble moment arm value low. I don't recall the BSFC being improved much, but the engines seemed to better utilize the .894 net intake valve lift we were running at the time.

We didn't see any CFD modeling on our race engines back in 2004, but later I saw a simulation on a production engine that showed a dished piston allowed better tumble continuation after BDC. We ended up running a small chamber and a shallow spherical dish in our race engines; as many others had.

Tumble management boiled down to chamber containment, and runner short-turn/ roofline height. Twist in the intake port floor can have some affect, too.

Thanks, Bryan.

There are some good papers on tumble in engines, the self-published ones from Honda (free) are interesting. There is also a SAE one during the design process of the "new Hemi" about mixture motion / tumble. Interesting comment about being able to use the valve lift, I'll have to think about that for a minute...


Here is what Randy Gillis (formerly of JE Pistons, now at Racetec. Also ‘piston_guy’ on here) said eons ago here about dish shape.

I did the first "tapered" or conical dish pistons for (then) Busch series engine builder Frank Leeson of Bill Davis Racing . He approached me with the idea and we made some test parts for him. Several dimensions were changed and those changes had a very clear affect on performance. One critical aspect was the width of the "perimeter squish band". Frank and I morphed the piston into the "spherical radius" from the conical due to the need for increased negative volume. After a couple of weeks I was contacted by Bob Fisher ( then of Ernie Elliott Inc) who was building engines for Bill. I gave both of them a 1 year exclusive on the design and development. Both did extensive back to back ( spherical to mirror image dish) testing . The spherical required at least two degrees less timing and always made a significant improvement to torque with a smaller improvement in HP. We figured combustion efficiency was responsible for that. There was also a stability condition. the feeling was the load was focused in the center and not offset by the mirror image dish. After the year was up , I offered the design to a west coast Craftsman truck ( then) engine builder. He was extremely skeptical and wanted no part of this "dumb design". I offered to ( and did) make two sets of equal pistons weight , rings, skirt profile, dish volume , etc . except for the dish design. The deal was to test the "conventional" design he was using first and then ( while still on the dyno)) pull it down , change the pistons and use the same used rings , and test it again. The result was a 10 hp 13 ft lb increase in power with 2*s less timing required. We didn't "invent" the concept , it was already out there on Hondas and other imports, we just adapted it to the V8 engine. There would eventually be a few cases where results were neutral as far as power increase but we did the design on all kinds of pistons. None made LESS power. I still use it today

Warp speed should remember when the concept hit and how soon it became available from every piston supplier. He might even have some comments on it.

I suspect a good part of what they were seeing was keeping tumble active longer due to the dish design.

Your work, Randy's comments, etc. are pushing 20+ years old now, so certainly not new concepts in the racing 2 valve V8 arena at the top end of things, and certainly not in the OEM / racing 4 valve world (though they seem to have figured it out a lot better in the last 15 years). I don't think enough builders beyond the top pro world pay attention to it.

I'll have to dig up Honda's tumble meter design. They have an illustration in one of their papers and they were using it to validate their CFD models.

[hoffman900]

Here is a screen shot from a Honda white paper:

Research on Combustion Improvement Techniques by Intake Valve Offset and Squish Effect

Reported in this paper are the technologies for improvement of combustion efficiency by applying two simple methods to a single cylinder, 110 cm3 displacement, four-stroke, two-valve gasoline engine.
In the first attempt, we tried to improve combustion efficiency by increasing tumble of the air-fuel mixture flow. To increase tumble, we devised an offset intake valve design in which a part of the intake valve was located outside of the cylinder bore. With this offset intake valve configuration, a part of the inlet port perimeter was blocked causing disturbance of air-fuel mixture flow along the cylinder wall that resulted in a strong turbulence. The increased turbulence permitted lean burn at an air-fuel ratio leaner by two points, reducing Indicated Specific Fuel Consumption by 4.8% from that of the base engine. With the intake valve shifted outwards against the cylinder bore, the spacing next to the exhaust valve increased, allowing the intake valve diameter to be enlarged to compensate for the deterioration of the maximum power.
In the second attempt, we tried to improve combustion efficiency by increasing the reversed squish flow of the air-fuel mixture. As the means to increase reversed squish flow, we employed a slant-parallel squish configuration. With the application of this squish arrangement, the margin against knocking generation was enhanced and the compression ratio was increased to 9.5 from the original 9.0 while reducing the Indicated Specific Fuel Consumption by 2.6%. The offset intake valve design coupled with the high compression ratio produced by the slant-parallel squish design lowered the Indicated Specific Fuel Consumption by 6.0% compared to the base engine.

This test engine is a hemi design.



Another from:
Development of New 3.5 L V6 Turbocharged Gasoline Engine for New NSX

The concept for Honda’s new NSX was to balance power performance achieving a high level of dynamic performance with a level of environmental performance meeting the demands of the times. A 3.5 L V6 turbocharged gasoline engine with a new frame was developed to realize this concept. In order to balance high power and high thermal efficiency, combustion was optimized through the use of in-cylinder direct injection and high-tumble ports, and the engine applies a twin turbocharger system featuring independent turbochargers positioned on the left and right banks and a fuel supply system combining in-cylinder direct injection and port injection. The powertrain was reduced in size and weight and given a low center of gravity in order to enhance vehicle dynamic performance. In addition, in response to the higher power of the engine, Fe spray-coated cylinder bores, a dry sump lubrication system, a three-piece water jacket, a swing arm valve train, and reduction chain drive were applied in order to help ensure cooling performance and frame reliability suited to the demands of sports driving.
As a result, the new engine achieves a maximum power of 373 kW (500 HP) and a maximum torque of 550 Nm across a wide range from 2000 rpm to 6000 rpm, in addition to complying with the European Euro6b and US ULEV125 regulations.

This one shows a similar tumble curve like you mentioned in the 2 valve NASCAR heads.

Also many more papers here: https://www.hondarandd.jp/index.php (sign up for free)

The SAE paper for the new Hemi:
Multidimensional Optimization of In-Cylinder Tumble Motion for the New Chrysler Hemi

The current is an investigation of the effects of charge motion, namely tumble, on the burn characteristics of the new Chrysler Hemi SI engine. In order to reduce prototyping, several combustion system designs were evaluated; some of which were eliminated prior to design inception solely based on CFD simulations. The effects of piston top and number of spark plugs were studied throughout the conceptual stage with the AVL-FIRE CFD code. It has been concluded that large-scale, persistent and coherent tumbling flow structures are essential to charge motion augmentation at ignition only if such structures are decimated right before ignition. Piston top had a detrimental effect on tumbling charge motion as the piston approaches the TDC. When compared to single spark plug operation, dual spark plug reflected considerable improvement on burn characteristics and engine performance as a consequence. The CFD simulations demonstrated good correlation with early dynamometer data. Finally, given the complex characteristics of rotating intake flows, it is conceivable to design an optimum charge motion level pertinent to specific combustion system architecture.

https://www.sae.org/publications/techni ... 2-01-1732/

This is a bit old now (2002), so I would love to see this done with modern CFD analysis and engineering understanding as CFD technology then was a bit, course.

[Bryan Maloney]

We had swirl, tumble, and wet flow test equipment available as part of Dodge's Nascar effort from 2000 to 2012.
We measured swirl as rotational torque with a honeycomb insert and a rotary strain gauge. Tumble was measured on another fixture with a perforated plate and three load cells. Data from the three load cells gave us total combined tumble torque, and more importantly, the force location defined as distance from the bore center and clocking in degrees.
We found that the swirl and tumble combined define the helix of the incoming air column in a two-valve chamber.
As Mummert stated, the swirl tends to jump up at low lift, lag behind in the mid-lift, and ramp up again at high lift. The tumble curve tends to do the same. The tumble force location (we called it "moment arm") moved closer to the bore center as the intake valve opened, and moved away from the bore center rapidly at high lift, when the swirl was spiking.
Our conclusion was we needed a smooth, linear increase in the swirl as the valve opened; not a set value.
To manage the swirl, we worked with chamber containment of the flow cone, runner trajectory, and some slight steering with the fin behind the guide. The flow must stay attached to the short-turn; swirl and tumble go out of control if there is separation. The modern steep seat and top angles really helped.
Reverse swirl fins behind the intake guides can improve burn uniformity, BSFC, and help prevent the swirl from spiking at high lift.
If you can do some wet flow testing on your cylinder head, you will be mesmerized by what you see. It will tie in with your swirl and tumble observations.

Bryan

What did you guys see by focusing on tumble after the swirl intensity was contained? Improved BSFC? Did you guys every get as far as getting the tumble portion modeled in a CFD program with a modeled piston top? How did you guys go about managing tumble short of moving the entire intake port angle?

Thanks

Also, how did you guys target a tumble value? Here is what LoganD said about tumble. At the time he worked for AVL. He works for Ilmor now:

Well, I guess it depends on what you consider "old". You have to be careful because an engine released in 2005 had the combustion system designed at least 5 years earlier, so there's a time lag. Basically any engine with the combustion system designed in the last 15 years is going to have very high tumble due to shallow valve angles. So that would basically be any production engine after 2010. In the late 90's and very early 2000's high end CFD was still so expensive that it was being used sparingly and time-dependent CFD was virtually impossible with automotive development budgets. It was also pretty impractical from a time perspective, what used to take 6 months to run now takes 2 weeks and can be done on computers that are 1/100th the cost you would have paid 20 years ago.

I guess what I'm saying is that the drastic increase in speed and drastic reduction in cost for high end CFD has caused us to change a lot of our previous perceptions about engine design. The new Mercedes M139 is a perfect example, 20 years ago if someone were going to design a 2.0 turbo 4-cyl to make over 400 hp on pump fuel they wouldn't make it heavily undersquare (83x92) with almost no valve angle. That engine revs to 7400 RPM and makes 370 lb-ft from 121 ci.

It makes a lot of sense if you think about it. If you're knock limited and not airflow limited, which most is the case for most heavily turbocharged engines, it makes sense to design the entire engine around reducing knock instead of just trying to make it flow more. This is where the aftermarket is behind, a Coyote or LS head doesn't need more flow to make 1000 hp on pump fuel, it needs more knock resistance.

This thinking spilled over into naturally aspirated engine design. They now give the engine just enough airflow to make the RPM/power target, and then they design the rest of the engine around maximizing efficiency and cylinder pressure. That's why the new GT3 engine has 13.5:1 compression and pretty shallow valve angles for an engine that revs to 9000 RPM. They spent a great deal of time making the large bore engine very knock resistant, and that's a hard thing to do.

To piggy back on my other post, have you worked with tumble specifically on hemi's? What have you found if so.

Sorry for all the questions, but I'm starved (and I think the forum is too) for a good conversation on this topic

There may be some difference in tumble terminology. With the Mercedes example you listed, they were altering the tumble characteristics by redesigning the head and altering the valve angle and placement. We could not do this on our race heads. We could only perform port and chamber reshaping to the existing design.

Simply labeling the motion "high" or "low" tumble can be misleading. We rated the tumble motion by how far off of the bore center the force center was, at 2/3 of the stroke length down the bore
We called this the "moment arm" length.
A port with a low moment arm would have the charge more closely following the center of the piston down the bore.

If the charge was directed across the chamber, the moment arm value would be higher. We considered this to be higher tumble. We only aimed for this on the hybrid engine, although I suspected this strategy would be best for GDI engines.

There were four or five tumble rigs produced. Chrysler kept one and the rest were distributed to the race teams. We did not spend any time testing hemi heads on ours. Chrysler was focused on getting CFD data on new production engine designs.

Hopefully the original poster, Gobrdgo, hasn't chucked his Pontiac heads into the dumpster, and said to hell with all of this.

[hoffman900]

Simply labeling the motion "high" or "low" tumble can be misleading. We rated the tumble motion by how far off of the bore center the force center was, at 2/3 of the stroke length down the bore
We called this the "moment arm" length.
A port with a low moment arm would have the charge more closely following the center of the piston down the bore.

Thanks, Brian. Yeah, a bit of an information dump and straying off topic. Also thanks for clarifying this. I know what a momentum arm is, but clarifying its relation to bore center and how far down the stroke length made it much clearer in regards to the information you shared.

Certainly wouldn’t mind any more NASCAR era lessons you found in regards to combustion efficiency.

[maxracesoftware]

Bryan , have you ever checked out #292 SBC Turbo Heads on your Swirl Meter ,
i know its very old low Tech Heads , but maybe you did already ??

the #292 Turbo Heads had a lot of in Cylinder Swirl induced by Port direction angle -to- Cylinder Bore and low Short Turn Apex height .
In Ported form , they would "Peg" my Quadrant Scientific Digital Swirl Torque Meter when i tested near 28" inches on my SF-600 Bench ,
which was a hassle , as i was normally Flow testing Heads at 36" Test Pressure and upwards and converting Flow Numbers back to 28" inches

i broke the strain-gage trying to test past 28" inches,
and sent it back to Oz , he fixed it , later it broke a 2nd time , i was able to fix it again ,
it still works , but i haven't used it in many years .

[tjs44]

he sent me a text this AM.Said you guys are WAY over his head but still listening.Tom

[BLSTIC]

We didn't see any CFD modeling on our race engines back in 2004, but later I saw a simulation on a production engine that showed a dished piston allowed better tumble continuation after BDC. We ended up running a small chamber and a shallow spherical dish in our race engines; as many others had.

Tumble management boiled down to chamber containment, and runner short-turn/ roofline height. Twist in the intake port floor can have some affect, too.

I've seen that in a Volkswagon (I think) paper. Also another one somewhere that showed tumble breakdown in time at various piston positions. It showed tumble lasting longer at BDC on this particular square engine. I came to the conclusion that undersquare engines were on average more 'square' than square or oversquare engines, and should have longer lasting tumble as a result.

Certainly my current favourite engine (the above mentioned Mercedes M139) has dish pistons, tiny valves spread apart so far they are almost certainly bore shrouded at low lift, a lot of tumble, is undersquare, and at 9:1 compression isn't *that* low for the 30psi boost it runs on pump gas while having the additional backpressure related stress of noise and emissions requirements... No squish though

[hoffman900]

he sent me a text this AM.Said you guys are WAY over his head but still listening.Tom

I think to start with "what is too much and what isn't", he needs to make sure the mass air flow demands of what he is working on are being met. PipeMax is a great tool for that.

Swirl and tumble gets pretty complicated when you factor in the moving piston, topography of the piston / chamber, changing densities, etc.

Bryan,

It would seem to me the end goal of pursuing the maximization of tumble and preservation of it as the piston approaches TDC with the NASCAR applications (despite the geometric limitations) would be to find improved fuel mileage, correct?

I think this is one thing that gets lost in most drag race engine builders (the best of which are very smart people - not saying this as a slight to them) pursuit is that they can keep adding fuel, and in many instances aren't limited by a spec fuel either. With most working on applications that are air mass limited, gaining cylinder flow / quality of the flow is where the gains are found.

As an aside:
https://www.motorsportmagazine.com/arti ... s-possible

What’s most important when you’re trying to make an engine work effectively on part throttle?

Internally, it’s basically the gas exchange [the cycle of supplying fresh air and removing exhaust gases] which you have to set up correctly. That’s port design, turbulence inside the cylinder and so on.

Everything has to work together. If you have a port that gives good flow figures that doesn’t necessarily mean you’ll have better combustion. Let’s say the combustion could be bad at low throttle or at low engine speed, even if the port flows well, so it’s all a bit of a compromise and a combination.

Since Michelin returned to MotoGP the back tyre has become a crucial part of stopping the bike, so negative torque is very important. So when you design an engine do you think about getting the bike into corners as well as out of them?

You need negative torque to stop the bike, that’s clear. What we’ve learned is that you need an engine that burns consistently at very low throttle openings. If you close the throttle completely the engine will shut off and there’s a fine line where you can still keep the engine burning and also deliver some negative torque for engine braking. For the rider it’s very important to have a consistent feeling in this area of corner entry – because if the engine shuts off and cuts in again it’s a disaster.

Obviously the application and rules dictate design. While Kurt Treib (also former BMW F1 engineer) doesn't say tumble specifically, the pursuit of lean burn quality / cycle-to-cycle variation is accomplished through increased cylinder turbulence, which is the same strategies played by OEM's for fuel economy and performance efficiencies. From a road racing car application, there is even opportunity to tweak engine braking in regards to corner entry, especially if the differential isn't as tuneable as it could be.

[Ken_Parkman]

And 2 phase flow

[tjs44]

He is building a street 455 with a 240-248 cam @50.Really good heads usually are about 260@600.He is having fun doing them with the use of a flo bench.Just trying to learn.Tom

[hoffman900]

He is building a street 455 with a 240-248 cam @50.Really good heads usually are about 260@600.He is having fun doing them with the use of a flo bench.Just trying to learn.Tom

I hope he is having fun learning too!

I think the answer to his question of what is too much is hard to answer, but mapping something like a LS or a Vortec head and mimicking their swirl intensity and shape isn’t probably a bad place to start, as long as he has the mass airflow needed.

[Bryan Maloney]

Bryan , have you ever checked out #292 SBC Turbo Heads on your Swirl Meter ,
i know its very old low Tech Heads , but maybe you did already ??

the #292 Turbo Heads had a lot of in Cylinder Swirl induced by Port direction angle -to- Cylinder Bore and low Short Turn Apex height .
In Ported form , they would "Peg" my Quadrant Scientific Digital Swirl Torque Meter when i tested near 28" inches on my SF-600 Bench ,
which was a hassle , as i was normally Flow testing Heads at 36" Test Pressure and upwards and converting Flow Numbers back to 28" inches

i broke the strain-gage trying to test past 28" inches,
and sent it back to Oz , he fixed it , later it broke a 2nd time , i was able to fix it again ,
it still works , but i haven't used it in many years .

The low, tipped short turn of the #292 may be a contributor to the extreme swirl.
I wish I could test a #292 port against an iron bowtie with a taller short turn. I suspect the iron bowtie would have better tumble characteristics. If I recall, the last version bowtie had the "carbon crunch" area in the chamber filled in.
Unfortunately, our swirl and tumble equipment returned to Chrysler when they pulled out of NASCAR, and the wet flow rig I built at Arringtons was dismantled when they shut down.
I'll try to find photos and start a thread about the wet flow testing setup. It was quite a learning curve.

[Bryan Maloney]

he sent me a text this AM.Said you guys are WAY over his head but still listening.Tom

I think to start with "what is too much and what isn't", he needs to make sure the mass air flow demands of what he is working on are being met. PipeMax is a great tool for that.

Swirl and tumble gets pretty complicated when you factor in the moving piston, topography of the piston / chamber, changing densities, etc.

Bryan,

It would seem to me the end goal of pursuing the maximization of tumble and preservation of it as the piston approaches TDC with the NASCAR applications (despite the geometric limitations) would be to find improved fuel mileage, correct?

I think this is one thing that gets lost in most drag race engine builders (the best of which are very smart people - not saying this as a slight to them) pursuit is that they can keep adding fuel, and in many instances aren't limited by a spec fuel either. With most working on applications that are air mass limited, gaining cylinder flow / quality of the flow is where the gains are found.

As an aside:
https://www.motorsportmagazine.com/arti ... s-possible

What’s most important when you’re trying to make an engine work effectively on part throttle?

Internally, it’s basically the gas exchange [the cycle of supplying fresh air and removing exhaust gases] which you have to set up correctly. That’s port design, turbulence inside the cylinder and so on.

Everything has to work together. If you have a port that gives good flow figures that doesn’t necessarily mean you’ll have better combustion. Let’s say the combustion could be bad at low throttle or at low engine speed, even if the port flows well, so it’s all a bit of a compromise and a combination.

Since Michelin returned to MotoGP the back tyre has become a crucial part of stopping the bike, so negative torque is very important. So when you design an engine do you think about getting the bike into corners as well as out of them?

You need negative torque to stop the bike, that’s clear. What we’ve learned is that you need an engine that burns consistently at very low throttle openings. If you close the throttle completely the engine will shut off and there’s a fine line where you can still keep the engine burning and also deliver some negative torque for engine braking. For the rider it’s very important to have a consistent feeling in this area of corner entry – because if the engine shuts off and cuts in again it’s a disaster.

Obviously the application and rules dictate design. While Kurt Treib (also former BMW F1 engineer) doesn't say tumble specifically, the pursuit of lean burn quality / cycle-to-cycle variation is accomplished through increased cylinder turbulence, which is the same strategies played by OEM's for fuel economy and performance efficiencies. From a road racing car application, there is even opportunity to tweak engine braking in regards to corner entry, especially if the differential isn't as tuneable as it could be.

Maximizing the fuel mileage was of huge importance to the NASCAR program. After seeing fuel buildup and clumping during wet flow testing, keeping the mixture homogeneous in the cylinder was our goal.
Cylinder heads had evolved to where the quality of airflow was important as quantity. If the program would have continued, I would have concentrated on the wet flow optimization.

[hysteric]

Cylinder heads had evolved to where the quality of airflow was important as quantity. If the program would have continued, I would have concentrated on the wet flow optimization.

Interesting you say that as what the fuel is doing is always overlooked for raw airflow.

[Mummert]

I like the wet flow testing. The biggest thing we had a hard time with was filming. Trying to get the proper amount of light and camera angle and experimenting with length of bore and so on played big role, not mention figuring out what we were seeing was a whole other thing.