Cylinder head swirl and flow data.

[Alex22]

I finished up my swirl meter the other day and have been porting and testing since. I am using a Superflow SF600 flowbench with a clay radius leading into the intake port and an 80mm bore adapter. The swirl meter is mounted in the bottom of the adapter. My baseline test is a stock AHU head. All that I did to the head was clean all the gunk out of the port, glass bead the port then lap the factory valve job to be sure it was still sealing. The injector and glow plug holes are blocked.

The reason that it says <300 RPM at .050 lift is because at that low if swirl is present it is not enough to move the meter but at the same time I cannot say that there is no swirl. Since I'm not done with the testing the other flow and swirl numbers are not worth posting. When I have some results worth posting I will put them up. My goal is to improve the airflow and swirl, or at the least to improve the flow without decreasing the swirl.

After some porting I flowed the head again at 28" and then again at 42", when I converted the numbers back to 28" they were within 1 or 2 CFM, at least the port doesn't have a huge problem with flow stability.

I have not been able to find any other published swirl numbers for TDIs, has anybody else ever found any or had some done and would like to share? The only published swirl numbers I have found are for BB Chevy's and SB Fords there isn't much in common for that data to be useful here.


~Alex

 

[Fix_Until_Broke]

Cool - Thanks for sharing your data.

It's interesting that the swirl suddenly drops so much at lifts >0.500" with no increase in flow. Also it seams that there's little to gain going over ~0.350" lift.

Would you mind sharing some details on your swirl meter - What it is, now it works, etc?

Just thinking out loud here - What if you repeated your baseline testing without the swirl meter installed? Would you get the same CFM numbers?

Thanks again for sharing! Keep up the good work.

 

[bhodgkiss]

does the AHU head you're using have the 7mm or 8mm valves (stems) please?
currently porting an AFN head with 8mm valves and will swap to 7mm if there's a measureable difference (less flow restriction under the valve head).
also if you could post any pics of the porting you're doing that'd be appreciated.
what lift does your standard AHU cam give please?

 

[TDIMeister]

Great work!

 

[chapelhill]

Some great data, looking forward to seeing the effect of porting on those numbers, and I too would like to know a bit more about the swirl meter. Does it just measure in one axis?

Looking at the inlet port I think it is shaped to promote flow entry to the cylinder as close to the cylinder wall to give a higher tangential component and also a tornado type rotation around the valve from the swirl dam.

I don't know if one is more important than the other, but for max power I presume more swirl of whatever kind will give faster and more complete combustion.

I visualize the flow in the combustion chamber during an injection cycle as flow around a doughnut which comes from the tangential velocity to the cylinder wall and then flow like wrapping the doughnut with string through the hole in the centre back out over the outside and back in.
http://forums.tdiclub.com/showpost.php?p=3245122&postcount=1
I saw on the pictures of the ported PD head where the injector mounting more is more intrusive that the porting had broken through to the bore, but I don't think that may be an issue, I need to go back and have a look at my scrap head.
http://forums.tdiclub.com/showpost.php?p=3245122&postcount=1

Would you be expecting to be able to impove both swirl and flow with porting?

I think the lower swirl rates are inevitable at the higher lifts, but this is being measured at significantly lower pressure differences than will actually exist at max power.

Many thanks for sharing!

Regards
Chapelhill

 

[mk1-83]

Nice work great!! have you also flow numbers of the exhaust side ?
the exhaust port is the bottelneck of the tdi cilhead.

 

[foxracer1]

Awesome can't wait to see after porting numbers! or even larger valves!

 

[MethylEster]

Alex,

Can you clarify the 24" and 48" units of measure please.

I'm assuming they are inches of Hg but you now how assuming things goes.

Thanks.

 

[TDIMeister]

The standard is 28 inches of water.

 

[Fix_Until_Broke]

Just for a point of reference, water is ~1/2 PSI per foot so the pressure difference during the flow bench testing is approximately 1-1/3 PSI

 

[Alex22]

It's interesting that the swirl suddenly drops so much at lifts >0.500" with no increase in flow. Also it seams that there's little to gain going over ~0.350" lift.

Would you mind sharing some details on your swirl meter - What it is, now it works, etc?

Just thinking out loud here - What if you repeated your baseline testing without the swirl meter installed? Would you get the same CFM numbers?

It is more difficult to improve low lift numbers than the high lift numbers because the same percent gain will equate to a lower flow gain at the lowers than at the higher lifts, also low lifts have more to do with the valve seat profiles and valve dimensions. I still have a few tricks up my sleeves though.

The basics of the swirl meter are the DC tachometer and the 4 bladed paddle wheel to pick up the swirl and turn the tach.

Since the swirl meter is mounted to my bore adapter it is not easily removed and reattached. I was able to compare the stock head flow numbers with the swirl meter to some tests that I had done a year or so ago on stock heads and it is within a few CFM and all 3 tests have the flow back up at .350 to .400 lift. When all is said and done I plan on removing the meter and doing a test. I don't see it as a problem since there is such a small reduction in flow are in the bore through the meter.

does the AHU head you're using have the 7mm or 8mm valves (stems) please?
currently porting an AFN head with 8mm valves and will swap to 7mm if there's a measureable difference (less flow restriction under the valve head).
also if you could post any pics of the porting you're doing that'd be appreciated.
what lift does your standard AHU cam give please?

From what I have seen and heard all US market TDIs have had the 7mm valve stems, even though the parts books list them as 8mm valve stem. I don't have any 8mm stems around for testing but I would say that the 7mm valves flow better. I don't have the stock cam in front of me at the moment but IIRC it is about .350 inch lift, maybe another member can confirm that before I can measure it tomorrow.

Some great data, looking forward to seeing the effect of porting on those numbers, and I too would like to know a bit more about the swirl meter. Does it just measure in one axis?

Looking at the inlet port I think it is shaped to promote flow entry to the cylinder as close to the cylinder wall to give a higher tangential component and also a tornado type rotation around the valve from the swirl dam.

I don't know if one is more important than the other, but for max power I presume more swirl of whatever kind will give faster and more complete combustion.

I visualize the flow in the combustion chamber during an injection cycle as flow around a doughnut which comes from the tangential velocity to the cylinder wall and then flow like wrapping the doughnut with string through the hole in the centre back out over the outside and back in.

I saw on the pictures of the ported PD head where the injector mounting more is more intrusive that the porting had broken through to the bore, but I don't think that may be an issue, I need to go back and have a look at my scrap head.
http://forums.tdiclub.com/showpost.php?p=3245122&postcount=1

Would you be expecting to be able to impove both swirl and flow with porting?

I think the lower swirl rates are inevitable at the higher lifts, but this is being measured at significantly lower pressure differences than will actually exist at max power.

Many thanks for sharing!

Regards
Chapelhill

The paddles on the swirl meter measure radial swirl (tornado like motion) at the cylinder bore walls as opposed to tumble (I'm not really sure how tumble would be measured.) Tumble is flow that goes directly over the short turn radius or over top of the valve and into the bore, which is most likely what is happening at .050 lift. To put tumble into perspective imagine rolling a barrel down the intake port and over the short turn radius, the barrel would be tumbling as it enters the bore.

The goal is to improve swirl and flow, but swirling air doesn't flow through the head very well so its a challenge to improve both.

Nice work great!! have you also flow numbers of the exhaust side ?
the exhaust port is the bottelneck of the tdi cilhead.

I have some exhaust numbers from testing that I had done last year but I forgot to print out the flow reports for the two. Exhaust porting is pretty straight forward on these heads so I'm consternating my efforts towards a better understanding swirl.

Alex,

Can you clarify the 24" and 48" units of measure please.

I'm assuming they are inches of Hg but you now how assuming things goes.

Thanks.

TDI Meister and Fix_Until_broke are correct, One of the standards for flow data in the automotive world is 28" of water (1psi.) Typically flow at 10" is used for power calculations and flow benches are calibrated at 25" of water.

I wish I had a flow bench that could flow a head at 42" of mercury. I did hear of one shop that had a flow bench that was powered by a big block chevy so they cold test at very high test pressures.


-----
The swirl and the flow are like the north and south sides of the magnet, they just aren't happy together, from what I've seen so far whatever improves one will decrease the other, time to find this port's sweet spot where I can balance the two out and make some improvements. The swirl is WAY more sensitive than the flow so I can only make small changes between tests.... this is gonna take a while.

~Alex

 

[Whitbread]

Good job sir! I wish I had more time to play around on a flow bench . On the SF600 I played around with a while back, I was able to pick up a 22-24% gain on both ports after porting. They sure are a fun toy!

While I did not measure swirl, even when you remove as much of the swirl dam as you can, it's still very much a swirl port. I may not be an expert on the topic at hand from an engineering side, but from my experience a "race" ported head on a street tdi has no ill side effects from removing as much as possible of the swirl dams.

To test this on somewhat of a "control" vehicle (compared to the frakensteins I usually build), I offered to race port a head with a Colt S2 cam for my friends 99.5 with a vnt15, 10mm and pp764s. Just got it running last week and so far its runs very strong, gets great mpg, and no weird smoke or throttle response issues at low rpm.

 

[dieseleux]

Oh yes... Good job!
One area for improvement is the edge of the head around the valve, this is the place where the valve spends the most time when it rises.



Dieseleux

 

[MethylEster]

Just for a point of reference, water is ~1/2 PSI per foot so the pressure difference during the flow bench testing is approximately 1-1/3 PSI

For those of us who are un-initiated in this world of flow testing, how does one then take the gains seen at this approximately 1-1/3 psi delta and extrapolate those gains at a real world pressure difference of 28 - 30 psi ?

How do you know that those positive changes at the lower delta P actually mean something measurable at the higher boost pressures?

Are the Reynolds numbers that similar between the two different flow regimes?

 

[MarkoP]

Many in this tread have mentioned that running engine has much bigger depression than what you can archieve with flowbech.

Can you please elaborate what you have calculated or measured being stock intake ports depression at 4000rpm.
What is the intake velocity at same RPM at smallest cross section position?

28-30psi, I do not think so.

If you have data to backup your ideas it would be really interesting to see.

 

[chapelhill]

Many in this tread have mentioned that running engine has much bigger depression than what you can archieve with flowbech..

MarkoP, Thankyou for turning my head upside down, I see what you are geeting at. I was certainly guilty of having presumed the larger pressure difference, but having re-thought it quickly if we were to map the cylinder pressure againts the MAP through the induction cycle he prssure difference would have to be much smaller in comparison to MAP other wise the flow would increase to balance it out. What the MAP under high boost does is to increases the density of the gas flowing not the speeds or the relative presurre differential so the bench test probably still stands as a good measure of change to performance.

Does anyone know of some real life data for the rpm's we are inetersted in tha we could see?

I presume the pressure difference will be negative at inlet valve opening quickly reducing to a small minumum then increasing to a maximum at just after Mid way down and then reducing again up until Inlet valve closes.

 

[Fix_Until_Broke]

I'm no expert in this field by any means - Just thinking about the subject at hand.

For turbulent incompressible flow the reduced equation is:

Q = A x Cd x sqrt(P1-P2)

Q = Volume flow rate
A = Area of the hole
Cd = Combined Discharge Coefficent of the area A - how efficient the hole is
P1-P2 = Pressure difference across the area A

The gas IS compressible and I'm sure that there is laminar flow at very low valve lifts, but for the most part to get the idea across the above simplified equation works well.

To increase the flow we can increase the area (open the valves more), improve the Cd (porting, valve angles, etc) or increase the pressure differential (more boost).

The pressure differential across the valve for a given RPM will be proportional to the boost pressure. The cam timing and rod ratio will determine the basic differential for a given RPM - given the high compression and very close valve to pistion clearances at TDC, the piston is on it's way down before the intake valve opens so there may be a higher than boost pressure dP at the beginning of the intake stroke. Yes, there will be some interdependance as a better flowing head at the same port pressure will fill the cylinder better creating less dP across the valve since the piston rate of displacement is constant, but again, generally speaking, more dP will yield more flow. The only way to really figure this out is to have in cylinder pressure measurement and compare it to port pressure and measured flow.

How well it will all extrapolate from ~1+ PSI to 30 PSI is a little bit of a moot point as flow bench data is stead state, whereas nothing in the actual dP or flow rate in the engine is steady state. The flow bench is just a way to compare relative gains/losses in steady state repeatable conditions.

Just thinking outside the box here....maybe one could take a long block, drive it with an electric motor and have a huge (55 gallon drum size) supply plenum to the intake manifold that has a mass airflow meter on the inlet of the plenum which will give an average mass airflow rate. You could pressurize the plenum to see the effects of boost pressure on mass airflow as well as adjust cam timing, duration, lift as well as rod ratio, etc with RPM.

Sorry Alex22 - Didn't mean to hijack your thread here. A before/after test with/without the swirl meter and comparing CFM's is a good idea to see how much of an effect the swirl meter has on the flow rate.

Keep up the good work!

 

[Alex22]

The main reason that I am flowing the head at 28" of depression is that it is one of the industry standards and I will have something to compare flow results to, that and the DC tachometer is only rated for 8000 RPM and I would rather not find out how long it can be driven above its maximum speed. Another thing to keep in mind is that the SF600 draws almost 40 amps on a 220 volt system and it is only capable of about a 1.5 change in pressure. The amount of power required to test at 30psi is just unreasonable for 99% of the people out there.

A running engine is a dynamic system and the flowbench is a static testing machine, but that doesn't mean that information gained on a flow bench is not useful. The flowbench allows the operator to not only measure how much air is moving through the port per unit time and in my case the swirl in the chamber, an experienced operator can make observations based on sound and test probes. All of that information combined will give a very good picture of how the air is moving through the port. From there problem areas can be identified and modified and the head can be flowed again to test the head porter's work.

Here are some results from the last 2 days of porting.

From the way things are looking the finished port design may have some weld added to certain areas and have some very funky shapes, but I'm nowhere near finished with the testing so that plan could change any day. The results posted are from repeated tests on the same port. Some of the grinding that I did reduced swirl or flow and since there is no eraser on the back of the die grinder and welding takes too long those areas are still in the port. Once I get a good idea of what adding or removing material from each area of the port does and how they work together as a system I will apply all of what I have learned on a fresh port... then probably more grinding and on to another port...

What do you make of these results and do you think there is any to know just how much swirl the port needs and how much can be sacrificed for flow?

No exhaust numbers at the moment.

~Alex

 

[TDIMeister]

Testing at 30 PSI pressure drop is non-sensical. Flow becomes choked at a critical pressure ratio of 0.528 for air; the absolute values of the pressures do not matter; the ratio of the drop does. Keeping the measurements at 28" water with inlet pressure being ambient keeps results consistent with almost universal practice, so keep that up Alex.

Edit: pictures added. The critical pressure ratio comes from (2/(κ+1))^(κ/(κ-1)) -- some may know κ as γ (gamma) instead.

 

[TDIMeister]

There is a clear trade-off between swirl/tumble and flow. The more flow you want, the more the swirl/flow must be compromised. No free lunch.

 

[MarkoP]

Testing at 30 PSI pressure drop is non-sensical. Flow becomes choked at a critical pressure ratio of 0.528 for air; the absolute values of the pressures do not matter; the ratio of the drop does.

Would 0.528atm be approximately 970fps or 0.87mach?
..I am not so good in math.
That would mean ~153VE% potential which you do not see often with real life engines and I think there is some lower values to aim for when porting heads.

There is a clear trade-off between swirl/tumble and flow. The more flow you want, the more the swirl/flow must be compromised. No free lunch.

Your the chief in here, but I have to disagree.
I have seen 80cfm/sq.in of valve area flow numbers from ported TDI head which is decent number when comparing any of 8v gasser production heads.
My swirl meter is still un-attached, but I assume that because smoke levels reduce under whole RPM band also flow quality must be better.
If not, then I would think that more smoke should be seen as long as RPM is high enough to get benefit of increased flow.
Also when larger valves are used, low lift flow is increased which reduces low rpm output even more.. but gives some to higher RPM band.

 

[TDIMeister]

Would 0.528atm be approximately 970fps or 0.87mach?

Not 0.528 atm, just 0.528 P/Po. By definition this is when the flow approaches M=1 at the smallest cross-section.

Your the chief in here, but I have to disagree.

Well then, here's a perfect opportunity to prove me wrong and vindicate yourself. Get a before-after of flow AND swirl tested by Alex (or any engine consulting company like Lotus Engineering, Ricardo, AVL, FEV, etc., or any half-decent University engineering department). Ruben might even be able to hook you up...

My swirl meter is still un-attached

Then attach it or talk to Alex. Then talk to us.

 

[TDIMeister]

BTW, the graph shows flow and swirl/tumble coefficients at full lift. It doesn't make any attempt to explain characteristics at low lift. Of course, larger valves and better port shaping will improve flow and probably not affect charge motion too much at the same time if done correctly. But the trends -- if not the absolute numbers -- as shown in the graph will be undisputable. It's nothing personal It has nothing to do with how good of a porter you are, it's physics.

Flow coefficients are normalised to piston area. Swirl coefficient is also normalised as the ratio of the circumferential velocity component to the axial velocity component of the swirling flow.

 

[TDIMeister]

For your edification, you don't have to take my word for it. Here's the bibliography:

Title: Modern engine technology: from A to Z / [edited by] Richard van Basshuysen and Fred Schäfer, editors.
Translation: Lexikon Motorentechnik
Publisher: Warrendale, PA: SAE International, c2007.
Subject (s): Internal combustion engines - Dictionaries
Internal combustion engines - Encyclopedias
Description: xix, 1047 p. : Ill.
Other (s) Author (s) / Title (s): Van Basshuysen, Richard, 1932 -
Schafer, Fred, 1948 -
ISBN: 9780768017052

Title: Internal combustion engine handbook: basics, components, systems, and prospects / edited by Richard van Basshuysen and Fred Schaefer.
Translation: Handbuch Verbrennungsmotor
Publisher: Warrendale, PA: SAE International, 2004.
Subject (s): Internal combustion engines
Description: xxxix, 811 p. : Ill. (Some col.)
Other (s) Author (s) / Title (s): Van Basshuysen, Richard, 1932 -
Schafer, Fred, 1948 -
ISBN: 0768011396

 

[MarkoP]

Not 0.528 atm, just 0.528 P/Po. By definition this is when the flow approaches M=1 at the smallest cross-section.


Well then, here's a perfect opportunity to prove me wrong and vindicate yourself. Get a before-after of flow AND swirl tested by Alex (or any engine consulting company like Lotus Engineering, Ricardo, AVL, FEV, etc., or any half-decent University engineering department). Ruben might even be able to hook you up...

Then attach it or talk to Alex. Then talk to us.

How much 0.528PR is from atmospheric?
I calculated that 1Mach should be close to 0.7atm.

Real life results do not justice bragging rights?
No, really, do you think that by incresing flow @ 2000-2500rpm by reducing swirl could yield to reduced smoke?
My believing is that added air helps from the point when engine starts to starve on air.
Increased low lift flow typically reduces power under to at least torque peak of N/A engine.. on forced engines torque peak is more or less "artificial" depending how turbo is sized.

There are several reaons why I did not attach swirl meter.
One was obviously that I am lazy =)
One was the fact that there are different type of swirls, low port runner would make huge swirl, but not that type of swirl you could use with TDi engine.
When I started porting TDI heads I noticed at some point that they did have unique shapes compared to typical heads they liked, so I went after what they wanted from flow and sound vise.. I know you enjoy reading this

If you did not notice Alex has increased flow and swirl on some lifts.
You can fix the seat area / do some bowl porting and flow will go up.
If you do not touch swirl, then flow speed at that point is increased which will increase swirl.. not so hard to figure without engine consulting company =)

 

[MarkoP]

BTW, the graph shows flow and swirl/tumble coefficients at full lift. It doesn't make any attempt to explain characteristics at low lift. Of course, larger valves and better port shaping will improve flow and probably not affect charge motion too much at the same time if done correctly. But the trends -- if not the absolute numbers -- as shown in the graph will be undisputable. It's nothing personal It has nothing to do with how good of a porter you are, it's physics.

Flow coefficients are normalised to piston area. Swirl coefficient is also normalised as the ratio of the circumferential velocity component to the axial velocity component of the swirling flow.

And that is absolutely correct Sir, the straighter the port, more it has opportunities to flow.
I was just refering to you previous statement and the fact that your the respected authority in here so I tried to straighten that a bit as its not that black and white:

There is a clear trade-off between swirl/tumble and flow. The more flow you want, the more the swirl/flow must be compromised. No free lunch.

 

[TDIMeister]

Another excellent flow bench thread:
http://speedtalk.com/forum/viewtopic.php?p=83856&sid=3fe13da6c51b11381c3c2ac64713b4b8

Schematic of Tippelmann test rig and flow coefficient normalisation equations:

 

[TDIMeister]

I was just refering to you previous statement and the fact that your the respected authority in here so I tried to straighten that a bit as its not that black and white:

I see, ok.

 

[TDIMeister]

How much 0.528PR is from atmospheric?
I calculated that 1Mach should be close to 0.7atm.

I think you're using Bernoulli's equation for that. Bernoulli only applies for incompressible flows; beyond Ma≈0.3 the flow can no longer be considered incompressible, so you must use the long equation I gave a few posts up for isentropic flows. When you go past Ma>1 and have a normal shockwave, both equations don't work (because the flow is not isentropic across the shockwave, but I digress... )

No, really, do you think that by incresing flow @ 2000-2500rpm by reducing swirl could yield to reduced smoke?
My believing is that added air helps from the point when engine starts to starve on air.

Yes you can. It is explained by other factors. You can have less smoke with more airflow but less swirl because you're simply running an effectively higher lambda. There's nothing really wrong with doing that, but there may be circumstances that you want to target lambda/smoke because it's a sign of how efficient the combustion is with the available air (air utilisation rate).

Most people here won't care about that, but say you're racing in an air-restricted rule regime, so you want to get the most performance (most fuel burn) out of a given amount of air you can get into the engine. Or, you want to reduce trapped air mass in the interest of targeting PCP if you're already at the bleeding edge.....

Increased low lift flow typically reduces power under to at least torque peak of N/A engine.. on forced engines torque peak is more or less "artificial" depending how turbo is sized.

Agreed.

There are several reaons why I did not attach swirl meter.
One was obviously that I am lazy =)
One was the fact that there are different type of swirls, low port runner would make huge swirl, but not that type of swirl you could use with TDi engine.
When I started porting TDI heads I noticed at some point that they did have unique shapes compared to typical heads they liked, so I went after what they wanted from flow and sound vise.. I know you enjoy reading this

I do! And if you find a way to have a great flowing head whilst keeping a healthy amount of the useful form of swirl, then you've hit upon a holy grail of Diesel performance tuning and I want to be in on it.

If you did not notice Alex has increased flow and swirl on some lifts.
You can fix the seat area / do some bowl porting and flow will go up.
If you do not touch swirl, then flow speed at that point is increased which will increase swirl.. not so hard to figure without engine consulting company =)

The numbers do show that what you describe is perfectly true, but the normalised numbers (because it's a ratio of the swirl-velocity divided by the axial velocity component of the total mass flow) also will not drastically change, which is my original point.

 

[MarkoP]

Another excellent flow bench thread:
http://speedtalk.com/forum/viewtopic.php?p=83856&sid=3fe13da6c51b11381c3c2ac64713b4b8

Schematic of Tippelmann test rig and flow coefficient normalisation equations:

That sort of swirl measuring device would be better, especially if honeycomb could be divided to 2 or 3 separate circles measuring force.

Thanks for bringing that thread up.
Nice to compare flow numbers to top of the line flowbench =)
My bench is calibrated to show ~same as flowbench what Ford is using, or actually mine reads 3% more, but as SF and other commercial benches show approximately same I can live with that.

My M20 flow numbers are somewhat higher than what marquis is showing, ~5-10% to 10mm, but then at 12mm I got 1% less.
There could be variance in seat profiles etc causing errors, so maby I should not compare like this.

He calculated flow speeds of 77m/s and I did measure 87m/s.
Usually measured flow speed is higher than calculated and I do not know at what flow/lift point he did calculation.
But if he used stock cam as baseline he probably calculated from 9mm lift where I have ~10-13cfm (7-9%) more flow.. that would match close enough.

How much do you calculate TDI engine would consume air at 4000rpm if boost pressure is excluded?
Or with given flow data from Alex how much depression there might be at same RPM?