Turbo Compressor Maps - Effect of Ported Shrouds and "The Tornado"

[The Radius Kid]

TTipe - Just to clarify, *all* of the flow through the exhaust is highly turbulent. We're talking about a gaseous mixture moving at high speed, high pressure, and high temperature... very high Reynolds number and thus quite turbulent. Cylindrical pipe flow goes turbulent at Re ~ (o) 1000, and these flow numbers are likely (o) 100,000.

The sensor is likely introducing some slight restrictions in the exhaust flow, but I'm really not sure that it's a first-order effect. Bison's experience shows that it's not significant. It's also probably subject to lots of other factors, so the results would vary from car to car.

mgmshar - Thanks for an interesting read. Makes sense to me that a fixed "tornado" device wouldn't likely result in significant gains across the entire range of operating conditions, but may have benefits under certain applications (part throttle, etc). I appreciate the discussion.

True enough.
I always work from the premise that total flow in a system is due to the sum of all restrictions in the system.
With that said,reducing or removing those restrictions [within reason] can't be a bad thing,according to the laws of physics.

 

[TTipe]

TTipe - Just to clarify, *all* of the flow through the exhaust is highly turbulent. We're talking about a gaseous mixture moving at high speed, high pressure, and high temperature... very high Reynolds number and thus quite turbulent. Cylindrical pipe flow goes turbulent at Re ~ (o) 1000, and these flow numbers are likely (o) 100,000.

The sensor is likely introducing some slight restrictions in the exhaust flow, but I'm really not sure that it's a first-order effect. Bison's experience shows that it's not significant. It's also probably subject to lots of other factors, so the results would vary from car to car.

mgmshar - Thanks for an interesting read. Makes sense to me that a fixed "tornado" device wouldn't likely result in significant gains across the entire range of operating conditions, but may have benefits under certain applications (part throttle, etc). I appreciate the discussion.

I agree that the reynolds number is very high and that there is so much variability from car to car that comparisons are next to impossible. You can say that small "directionally correct" changes add up, at least based on a Bernoulli approach. For example the Nippondenso 234-1001 narrow band o2 is diametrically smaller the the "OE" sensor and does the same job.The exhaust pulses coming through the exhaust system are not all evenly spaced and going through a necked down passage with an object in the way. Isn't restriction through a system cummulative? I would like to model this using GT Power and CFD.

 

[TTipe]

Id think the MAF pipe with the filter is better than nothing at all since it will help straighten the air out as it enters the pipe. That compressor may be a bad example though. You really need to lean on these things as hard as possible to see small differences. Just the machining of the cover could have an effect on the compressors efficiency. Ive noticed a lot of 49's have ill fitting components. Ive see nthe stock back plate used with the standard back height T04e 60 compressor. It sits up above the back plate. Ive also seen billet back plates with this also. Ive seen billet plates with a slight taper to them too. If youre not going to run the proper back plate or machine the compressor wheel so its sitting at the correct height you are probably beating your head in trying to milk one for all its worth.

I have a modified Ta61 in my car ( that the Radius Kid put a lot of time into-thought and machining) at present and I'm going try to get some relative numbers to see what has improved.The butt dyno says "definite improvements but the way to really show improvement is to have "before mods" TEMP IN and TEMP OUT and "after mods" TEMP IN and TEMP OUT".This will allow an adiabatic efficiency calculation.Also I need to get "rate of flow/unit of time" in and out before and after mods.This will provide the information required to map the turbo with these mods. The butt dyno has identified more power and less turbo noise.I'm bringing back a powerlogger since I have not set up my fast yet.

 

[Jonasterg]

I agree that the reynolds number is very high and that there is so much variability from car to car that comparisons are next to impossible. You can say that small "directionally correct" changes add up, at least based on a Bernoulli approach. For example the Nippondenso 234-1001 narrow band o2 is diametrically smaller the the "OE" sensor and does the same job.The exhaust pulses coming through the exhaust system are not all evenly spaced and going through a necked down passage with an object in the way. Isn't restriction through a system cummulative? I would like to model this using GT Power and CFD.

Good points, with a little bit of clarification. The diameter of the obstruction is important, but if both are cylindrical and both are similar sizes then there will be more variability due to external factors than the obstruction itself. The restrictions aren't cumulative in a linear fashion, though more restrictions will impact flow rate. I also agree with Radius Kid that fewer obstructions should be better than more obstructions. My argument is simply that this *may not* be a leading order effect for our cars.

It would be a fun CFD exercise (my thesis was CFD w/ Immersed Boundary Method, I do hydrocode work for the Navy now) but to accurately model this type of pipe flow would be computationally expensive. We've got a few different generations of scales, heat transfer can't be ignored, flow is rotational and an inviscid assumption may not be valid (not sure, I don't do much work at these regimes). If you have access to some idle CPU*hours, I'd be happy to lend a hand with the experiment.

Sorry for being pedantic; you bring up some good ideas and I'm just trying to clarify a bit.

 

[bison]

Good points, with a little bit of clarification. The diameter of the obstruction is important, but if both are cylindrical and both are similar sizes then there will be more variability due to external factors than the obstruction itself. The restrictions aren't cumulative in a linear fashion, though more restrictions will impact flow rate. I also agree with Radius Kid that fewer obstructions should be better than more obstructions. My argument is simply that this *may not* be a leading order effect for our cars.

It would be a fun CFD exercise (my thesis was CFD w/ Immersed Boundary Method, I do hydrocode work for the Navy now) but to accurately model this type of pipe flow would be computationally expensive. We've got a few different generations of scales, heat transfer can't be ignored, flow is rotational and an inviscid assumption may not be valid (not sure, I don't do much work at these regimes). If you have access to some idle CPU*hours, I'd be happy to lend a hand with the experiment.

Sorry for being pedantic; you bring up some good ideas and I'm just trying to clarify a bit.

There are so many things to consider besides the pipe(s) themselves. There's a window of acceptable operation where you will not lose power on a turbocharged engine like you would with an NA engine. I've seen drops in back pressure that didn't increase power much till I ran a higher pressure ratio. If you look at a 7.5:1 engine vs a 9:1 engine the obstruction is the exhaust pressure more so on the 7.5:1 engine than the 9:1 even with everything else the same. Its still easier to push all the inert gas out of the cylinder than to increase the turbine flow rate if you want a responsive setup
and still make power. The CR will raise mass efficiency quite a bit itself. Cam profile and timing can further complement this.

 

[The Radius Kid]

Good points, with a little bit of clarification. The diameter of the obstruction is important, but if both are cylindrical and both are similar sizes then there will be more variability due to external factors than the obstruction itself. The restrictions aren't cumulative in a linear fashion, though more restrictions will impact flow rate. I also agree with Radius Kid that fewer obstructions should be better than more obstructions. My argument is simply that this *may not* be a leading order effect for our cars.

It would be a fun CFD exercise (my thesis was CFD w/ Immersed Boundary Method, I do hydrocode work for the Navy now) but to accurately model this type of pipe flow would be computationally expensive. We've got a few different generations of scales, heat transfer can't be ignored, flow is rotational and an inviscid assumption may not be valid (not sure, I don't do much work at these regimes). If you have access to some idle CPU*hours, I'd be happy to lend a hand with the experiment.

Sorry for being pedantic; you bring up some good ideas and I'm just trying to clarify a bit.

That would be one VERY interesting set of computations to run.
There's a whole load of variables that would affect the numbers .... and since we're dealing with gases and not liquids ...... oh boy !!!!!!

 

[mgmshar]

This would be extremely hard to model in CFD. The primary reason is because of the fact that the flow is not "constant flow", but basically a series of 6 pulses for every 2 revolutions of the crankshaft. It would be neat to try to CFD the problem, but as stated above, it would be a nasty problem. Besides, does anyone have a 3D model of the stock GN manifolds handy? How about a 3D scanner to make a model? You could spend weeks on this.

Maybe a little engineering common sense is in order. First, with the bends, corners, welds, joints, etc. plus the high Reynolds number in the exhaust system, the flow is going to be turbulent. Laminar flow in anything involving an automotive engine is pretty rare (one glaring exception - the flow of the air going through the screens in the MAF - look at the extremes the MAF designer had to go through to ensure laminar flow through that thing!).

Does the oxygen sensor in the exhaust pipe cause a flow restriction? You bet it does. Jeez, when you take your turbo off, just look at it - the sensor probably takes away 5-10% or more of the cross-sectional area of the pipe (depending on which one you have). However, remember that when this system was designed, non-heated oxygen sensors were used. The engine systems guys had to put the sensor in that location to get a combined reading from both cylinder banks and to keep the sensor as hot as possible. The question is this - in the grand scheme of things, how much of a restriction is it, really? Well, the biggest restriction by far in our exhaust systems is the turbine of the turbo. A "rule of thumb" is that the pressure in the exhaust manifolds is roughly 2X the boost pressure (very general, but in the ball park). At 25psig of boost, there is probably about 50 psig or more of back pressure in the exhaust manifolds for my TA49. So, there is something like a 50 psi pressure drop across the turbine of the turbo. How many psi pressure drop occurs when the exhaust flows past the oxygen sensor? Not nearly that much!

Now, if you have a really large turbo with a free-flowing turbine, then the effect of the oxygen sensor at the turbine inlet is probably significant enough to consider removing it. Actually, just the restriction of necking all of the exhaust flow down to a single ~2-inch opening at the end of the passenger-side manifold is probably a factor when running a large, free-flowing turbine. The presence of the oxygen sensor would just make this a bit worse.

I'd be interested to see some back-to-back runs with and without the oxygen sensor. I'd love to see that with a car with a small turbo (TA49 or similar) and a big turbo (the 700 hp variety). My "engineering estimate" is that for the small turbo, removing the oxygen sensor won't help anything much (spool-up, maximum boost, hp, etc.), but for the bigger turbo, it might help a little.

Then again, I could be wrong... However, my thinking is that on most cars, there are other areas that will yield more bang for the buck.

Anyway, interesting discussion, all. But I'm still not putting one of those Tornados in my intake pipe!

 

[TTipe]

Good points, with a little bit of clarification. The diameter of the obstruction is important, but if both are cylindrical and both are similar sizes then there will be more variability due to external factors than the obstruction itself. The restrictions aren't cumulative in a linear fashion, though more restrictions will impact flow rate. I also agree with Radius Kid that fewer obstructions should be better than more obstructions. My argument is simply that this *may not* be a leading order effect for our cars.

It would be a fun CFD exercise (my thesis was CFD w/ Immersed Boundary Method, I do hydrocode work for the Navy now) but to accurately model this type of pipe flow would be computationally expensive. We've got a few different generations of scales, heat transfer can't be ignored, flow is rotational and an inviscid assumption may not be valid (not sure, I don't do much work at these regimes). If you have access to some idle CPU*hours, I'd be happy to lend a hand with the experiment.

Sorry for being pedantic; you bring up some good ideas and I'm just trying to clarify a bit.

I worked on the tune a bit and the turbo has to be more efficient given the improvements already made.1) The torque converter in the car is a 3000 rpm stall n/l.Prior to any turbo enhancements the foot stall of the converter was 3000 rpm.I bought the unit from Jack and he said foot stall was 3k.I realize that instrument error, temperature,state of tune etc contribute to a poor conclusion. Once I started evaluating these changes (internal to the turbo) the foot stall dropped from near 3000 rpm to about 2780 rpm.Ambient air temp was about 78deg F.I have no baro data. Next I worked the tune slightly and the foot stall dropped to 2600 rpm at 1 psi.Ambient air was about 70 deg F.Again I have no baro data. I got my son to watch the gages with me since I had assumed I had taken data incorrectly and his readings agreed with mine. Just blipping the throttle and letting off (while in park) the wheel is whining when dropping rpm like "it has more air /gulp but its momentum is fighting the gas energy captured in the compressor wheel pockets along with a reduction of gas energy when lettting off the throttle. When in drive it builds so easy and fast its sick. The streets are wet and my scanmaster bit the dust (I don't have my fast in yet) otherwise I would be out conducting "unscientific" experiments.I'm using an LS1 maf, tomka hose from the maf to the turbo bell. I have a longer bell on the turbo to help straighten the air for a proper presentation to the turbo. My exhaust is closed ( 3in straight shot) and I'm running 93 octane that is old. This description is not founded on properly acquired scientific data with a proper design of experiment , but something positive is going on.

 

[TTipe]

This would be extremely hard to model in CFD. The primary reason is because of the fact that the flow is not "constant flow", but basically a series of 6 pulses for every 2 revolutions of the crankshaft. It would be neat to try to CFD the problem, but as stated above, it would be a nasty problem. Besides, does anyone have a 3D model of the stock GN manifolds handy? How about a 3D scanner to make a model? You could spend weeks on this.

Maybe a little engineering common sense is in order. First, with the bends, corners, welds, joints, etc. plus the high Reynolds number in the exhaust system, the flow is going to be turbulent. Laminar flow in anything involving an automotive engine is pretty rare (one glaring exception - the flow of the air going through the screens in the MAF - look at the extremes the MAF designer had to go through to ensure laminar flow through that thing!).

Does the oxygen sensor in the exhaust pipe cause a flow restriction? You bet it does. Jeez, when you take your turbo off, just look at it - the sensor probably takes away 5-10% or more of the cross-sectional area of the pipe (depending on which one you have). However, remember that when this system was designed, non-heated oxygen sensors were used. The engine systems guys had to put the sensor in that location to get a combined reading from both cylinder banks and to keep the sensor as hot as possible. The question is this - in the grand scheme of things, how much of a restriction is it, really? Well, the biggest restriction by far in our exhaust systems is the turbine of the turbo. A "rule of thumb" is that the pressure in the exhaust manifolds is roughly 2X the boost pressure (very general, but in the ball park). At 25psig of boost, there is probably about 50 psig or more of back pressure in the exhaust manifolds for my TA49. So, there is something like a 50 psi pressure drop across the turbine of the turbo. How many psi pressure drop occurs when the exhaust flows past the oxygen sensor? Not nearly that much!

Now, if you have a really large turbo with a free-flowing turbine, then the effect of the oxygen sensor at the turbine inlet is probably significant enough to consider removing it. Actually, just the restriction of necking all of the exhaust flow down to a single ~2-inch opening at the end of the passenger-side manifold is probably a factor when running a large, free-flowing turbine. The presence of the oxygen sensor would just make this a bit worse.

I'd be interested to see some back-to-back runs with and without the oxygen sensor. I'd love to see that with a car with a small turbo (TA49 or similar) and a big turbo (the 700 hp variety). My "engineering estimate" is that for the small turbo, removing the oxygen sensor won't help anything much (spool-up, maximum boost, hp, etc.), but for the bigger turbo, it might help a little.

Then again, I could be wrong... However, my thinking is that on most cars, there are other areas that will yield more bang for the buck.

Anyway, interesting discussion, all. But I'm still not putting one of those Tornados in my intake pipe!

The temperature difference between the exhaust valve and the turbo exducer (~300 deg F) is a large portion of the total energy (from exh manif press, 2x the boost pressure) being changed to mechanical energy. The pressure drop across the o2 is not a player as is the increase in turbulence which reduces the gas fit / gulp of the turbine blade as it exposes to the slot in the turbine housing volute. Years ago I measured flow across a stock maf with the cone vs no cone on a certified flow bench (at GM) and the difference was significant (I can't remember now ). The screens were used to straighten the air so the cone could force the air across the film to read the entire air speed bandwidth. I used to run with no cone and no screens. Parts or the rpm band could be read but a lot of the band was driven by a TPS vs RPM vs LV8 (I think) to reduce pressure drop across the maf to help the compressor.The o2 was a forced position as you said to create o2 cross counts to say "its in feedback". Oh by the way who can take the derivative of a cusp??

 

[The Radius Kid]

The pressure drop across the o2 is not a player as is the increase in turbulence which reduces the gas fit / gulp of the turbine blade as it exposes to the slot in the turbine housing volute.

Good thing it isn't the twiddle of the kadiddle or the Newton of the high falutin' Canutin.

 

[Jonasterg]

Good thing it isn't the twiddle of the kadiddle or the Newton of the high falutin' Canutin.

Whoa whoa whoa... when it's the kadiddle you really need to be careful!

 

[TTipe]

This would be extremely hard to model in CFD. The primary reason is because of the fact that the flow is not "constant flow", but basically a series of 6 pulses for every 2 revolutions of the crankshaft. It would be neat to try to CFD the problem, but as stated above, it would be a nasty problem. Besides, does anyone have a 3D model of the stock GN manifolds handy? How about a 3D scanner to make a model? You could spend weeks on this.

Maybe a little engineering common sense is in order. First, with the bends, corners, welds, joints, etc. plus the high Reynolds number in the exhaust system, the flow is going to be turbulent. Laminar flow in anything involving an automotive engine is pretty rare (one glaring exception - the flow of the air going through the screens in the MAF - look at the extremes the MAF designer had to go through to ensure laminar flow through that thing!).

Does the oxygen sensor in the exhaust pipe cause a flow restriction? You bet it does. Jeez, when you take your turbo off, just look at it - the sensor probably takes away 5-10% or more of the cross-sectional area of the pipe (depending on which one you have). However, remember that when this system was designed, non-heated oxygen sensors were used. The engine systems guys had to put the sensor in that location to get a combined reading from both cylinder banks and to keep the sensor as hot as possible. The question is this - in the grand scheme of things, how much of a restriction is it, really? Well, the biggest restriction by far in our exhaust systems is the turbine of the turbo. A "rule of thumb" is that the pressure in the exhaust manifolds is roughly 2X the boost pressure (very general, but in the ball park). At 25psig of boost, there is probably about 50 psig or more of back pressure in the exhaust manifolds for my TA49. So, there is something like a 50 psi pressure drop across the turbine of the turbo. How many psi pressure drop occurs when the exhaust flows past the oxygen sensor? Not nearly that much!

Now, if you have a really large turbo with a free-flowing turbine, then the effect of the oxygen sensor at the turbine inlet is probably significant enough to consider removing it. Actually, just the restriction of necking all of the exhaust flow down to a single ~2-inch opening at the end of the passenger-side manifold is probably a factor when running a large, free-flowing turbine. The presence of the oxygen sensor would just make this a bit worse.

I'd be interested to see some back-to-back runs with and without the oxygen sensor. I'd love to see that with a car with a small turbo (TA49 or similar) and a big turbo (the 700 hp variety). My "engineering estimate" is that for the small turbo, removing the oxygen sensor won't help anything much (spool-up, maximum boost, hp, etc.), but for the bigger turbo, it might help a little.

Then again, I could be wrong... However, my thinking is that on most cars, there are other areas that will yield more bang for the buck.

Anyway, interesting discussion, all. But I'm still not putting one of those Tornados in my intake pipe!

Why can't you CFD worst case pulse through the exhaust tract.The worst case would be the longest run (cylinder 1). That would be a quick and dirty way to establish a trend.Multiples could only get worse.Cylinders 3 and 4 will have issues due to the angle of approach to the log. If it was me I would model cylinders 1 and 3.

 

[mgmshar]

Why can't you CFD worst case pulse through the exhaust tract.The worst case would be the longest run (cylinder 1). That would be a quick and dirty way to establish a trend.Multiples could only get worse.

Do you have a 3D model of the driver side manifold, cross-over pipe, passenger side manifold, and turbine housing handy? If so, all the power to you

Anything can be modeled, but the question that always needs to be asked is whether or not it's worth the time and money. If you have the time and resources, go for it and post the results here. Would love to see it.

Mike
Sent from my HTC Droid Incredible using Turbo Buick

 

[The Radius Kid]

Whoa whoa whoa... when it's the kadiddle you really need to be careful!

Ain't that the truth.

 

[TTipe]

Do you have a 3D model of the driver side manifold, cross-over pipe, passenger side manifold, and turbine housing handy? If so, all the power to you

Anything can be modeled, but the question that always needs to be asked is whether or not it's worth the time and money. If you have the time and resources, go for it and post the results here. Would love to see it.

Mike
Sent from my HTC Droid Incredible using Turbo Buick

I have nothing that is considerred a solid , however some adiabatic efficiency computations are possible.

 

[RmvBfrFlght]

Thanks for posting your results - very interesting. I guess the Tornado doesn't make that much of a difference in your combo - definitely the restrictive nature of the turbine has a lot to do with that. I have never seen a Tornado in person - can you estimate the angle of the blades on the Tornado relative to the air flow direction? The above article used blade angles of 60, 70 , and 80 degrees (pretty extreme angles), so maybe the Tornado has shallower blade angles that don't affect compressor efficiency as badly.

Doing a bit of work on the Black Car today so snapped a pic or two. This is looking into the end from the turbo side. Appears to be a 3" unit cut down to fit a 2-3/4" pipe. Blade angle is 12*. Blades are rather long, but didn't measure the length.

This is in front of a TA62 with a few inches between it and the inlet bell. My understanding is that part throttle boost was undriveable without it.



RemoveBeforeFlight

 

[TTipe]

Doing a bit of work on the Black Car today so snapped a pic or two. This is looking into the end from the turbo side. Appears to be a 3" unit cut down to fit a 2-3/4" pipe. Blade angle is 12*. Blades are rather long, but didn't measure the length.

This is in front of a TA62 with a few inches between it and the inlet bell. My understanding is that part throttle boost was undriveable without it.

View attachment 167577View attachment 167578

RemoveBeforeFlight

Sounds like your induction plumbing is not helping to stabilize the air. In a stock car, if the turn in front of the inlet bell is a sharp angle along with a drop and a covoluted tube.The air is not going to be very staight so you won't pack the compressor well. Do you have any pics of your air intake plumbing?

 

[Jonasterg]

Why can't you CFD worst case pulse through the exhaust tract.The worst case would be the longest run (cylinder 1). That would be a quick and dirty way to establish a trend.Multiples could only get worse.Cylinders 3 and 4 will have issues due to the angle of approach to the log. If it was me I would model cylinders 1 and 3.

Not just the structural model, but the inlet and outlet boundary conditions are going to be a very tough modeling construct. There aren't many BCs that can model such a variable inflow and remain stable to the timescales we'd need. Also to reduce the influence of the BCs (source of error) you'd need to move both the inlet and outlet fairly far away from the area we're studying.

Also, like I mentioned before, in addition to a complex CFD analysis there's a nontrivial heat transfer analysis to do.

It sounds like a fun exercise but we're talking about a few months just to generate the models, then a few hundred thousand CPU*hrs for the analyses (RANS capable CFD code at least... LES would be more accurate but much more expensive).

 

[RmvBfrFlght]

Sounds like your induction plumbing is not helping to stabilize the air. In a stock car, if the turn in front of the inlet bell is a sharp angle along with a drop and a covoluted tube.The air is not going to be very staight so you won't pack the compressor well. Do you have any pics of your air intake plumbing?

The intake plumbing is short and straight, no MAF either. Look at the picture that is straight into the tornado. You can see the inside of the K&N cone filter. The filter is about a foot long.

RemoveBeforeFlight

 

[The Radius Kid]

Doing a bit of work on the Black Car today so snapped a pic or two. This is looking into the end from the turbo side. Appears to be a 3" unit cut down to fit a 2-3/4" pipe. Blade angle is 12*. Blades are rather long, but didn't measure the length.

This is in front of a TA62 with a few inches between it and the inlet bell. My understanding is that part throttle boost was undriveable without it.

View attachment 167577View attachment 167578

RemoveBeforeFlight

Interesting.
The turbine engineer I used to talk with told me it was best to just straighten the air into the inlet.
That was my plan from the outset,so it seems we were in agreement.
Turbines do tend to like directed air into the vanes.
It makes them more efficient.
The trick is finding the vane angle and shape that the wheel likes at a given wheel speed/flow rate.
A shrouded wheel never hurts either,but we won't be using those.