1Z/AHU Volumetric Efficiency

[Digital Corpus]

Outputs are at the bottom of each sheet. I stuck with LibreOffice for the work since the current Office 2016 doesn't import CSV files.

Engine parameters:

• Stock camshaft
• PD130 Intake manifold
• Mildly ported head
• AFN exhaust manifold
• GTB1756VK turbo port matched to AFN manifold and adapter
• 2 feet of 2.5" downpipe with tapered-to-match hotside

Testing methodology:

1. For RPMs 900 to 2000 I used the ECU to ramp idle from 900 RPM to XXXX RPM as the engine warmed from 80 °C to 85 °C. Intervals of 250 RPM used
2. For RPMs 2250 to 4500 I use a 1" diameter wood pole, a dictionary, and driver's seat to hold the throttle at an appropriate point for the desired RPM. The engine was at least 85 °C before logging commenced and once at 250 RPM, intervals of 500 RPM were used.

The latter was utilized because the MSA15.5's software for a GH ECU doesn't like trying to idle over 2200 RPM. Due to human error and other physics, some continual adjustment was made using process 2 as RPM was difficult to settle. I removed 1 or 2 chunks of data that didn't average out over time along with the rest of the dataset for that RPM range if there was more than +/- 150 RPM fluctuations. I give a confidence of +/- 2% (so a score of 80% means 78%-82%) due to this, which seems reasonable.

Ice packs were used to cool down the fuel pump.

A minimum of 300 samples per RPM were utilized and averaged for the computations.

Current temperature and weather were pulled from observable history and linearly interpolated within reason for ambient air conditions.

Calculation order of operations:

• Determine current air density and factor in relative humidity.
• Determine 1 cylinder's mass of air when completely full with current density.
• Using barometer, MAP Actual, ambient air temp, and IAT, determine new RH % under compressed state.
• Determine post charge cooler air density from MAP Actual, IAT, & RH %.
• Determine 1 cylinder's mass of air under these new conditions.
• Divide MAF Actual by previous result to compute volumetric efficiency.

I used my handy graphing calculator, TI's Voyage 200, to handle the computation of density and RH % adjustments and [automated] unit conversions to reduce human error.

The following graph pulls the data from the aforementioned process and plots it from 900 RPM to 4500 RPM.

Data available here

Numbers for the TL;DL

RPM    900    1000    1250    1500    1750    2000    2250    2500    3000    3500    4000    4500
VE    86.9    87.2    88.6    90.8    92    90    88.6    86.8    84.7    82    80.1    74.4

Graph for the visual

Please feel free to check my work. Additional details available upon request. I do plan on doing this with the Colt Stage II camshaft in the near (1 or 2 months out at best) future.

 

[mk1-83]

nice work, now chance cam to nps 10mm lift and 35 ex and 38mm in valve and ported

 

[cruiserboy]

I had made this study as well but creating a dedicated measuring group with engine speed / MAP / IAT / MAF with increased datalogging frequency in VCDS

VNT mechanism impacts VE quite a lot as well as load. You should try to consider only stabilized points and draw a graph of VE against engine speed and MAP (recalculated at a fixed IAT of your choice)

 

[Digital Corpus]

I considered that, but when I looked at the data with a 25-sample running average, the points where the idle control was used to select a stable RPM had nearly as much variation in the MAF output as the ones I had to manually stabilize. The MSA15.5 software isn't as documented as the EDC15 so I've not been able to create a channel in one of the unused spots to need to log just one channel instead. However, sensor variation over time shows that mearly as an inconvenience.

I do want to re-run the same tests with the vanes fully open, but even with them mostly closed, I'm not expecting much of a difference unless it's high(er) RPM.

 

[TDIMeister]

Another great D_C thread!

VE is fundamentally independent of boost pressure and temperature, but it does depend on where one makes the reference, either ambient or upstream of the intake valve (for all intents and purposes intake manifold conditions). Calculated (corrected) VE at no boost will be essentially the same under boost.

In an engine with minimal valve overlap and conservative timings - not extremely late or extremely early timings, VE is also independent of vane position and EMP, as long as there isn't substantial trapped residual gas %.

I do have a ported, big-valve ALH head and 11mm race cam if someone wants to buy it and do the same VE tests.

 

[Digital Corpus]

Wouldn't a significant restriction at higher flow rates affect VE though? Such as vanes nearly closed at high RPM? The thinking being greater back pressure. If not, then I assume this is due to the back pressure being significantly less than the forces of the piston and gasses moving about the cylinders.

I appreciate the kind words.

 

[TDIMeister]

Wouldn't a significant restriction at higher flow rates affect VE though? Such as vanes nearly closed at high RPM? The thinking being greater back pressure. If not, then I assume this is due to the back pressure being significantly less than the forces of the piston and gasses moving about the cylinders.

Only true if there are significant valve overlap and trapped residual gas content - and yes, there tends to be more residual gas at higher RPM.

 

[GoFaster]

Significant exhaust manifold pressure will increase exhaust residual at the end of the exhaust stroke, but with a high compression ratio this is a small volume being carried through TDC into the intake stroke, and with essentially no valve overlap there is no back-flow that would otherwise make matters a lot worse. In other words, it would take a very large change in exhaust manifold back-pressure to make a relatively small change in VE.

Interesting experiment!

 

[Digital Corpus]

I want to refine 1500-2250, so I may run these tests again at 50 RPM intervals in the near future. Granted, that's where most of our non-extreme turbos are trying to spool. I was surprised to see values above 85% given the dogma on the forums. Crap, forgot to mention PD130 intake manifold. Editing...

Re-testing is really easy to do using the idle control...

 

[vanbcguy]

This is awesome data.
Also a place where cam changes may show themselves!

Sent from my XT1097 using Tapatalk

 

[All Stock]

I remember VE being a function of cylinder filling. NA motors at best were in the 85% range due to the fact that they had to pull in the air through vacuum created by the drop of the piston. That has increased into the 90's but its still a number based on how much air has actually filled the displacement of the cylinder. With any motor having forced induction that means you cram more volume of air into the displacement than atmosphere could fill at 100%. So VE should be 100-200 %.

So in truth to find the true volumetric efficiency of the TDI it has to be measured at zero boost or atmosphere. This means removing the intercooler and turbo as an intake restriction, as for the turbine side if it impacts pumping losses it to must be removed. What is then left is the motors true ability to fill its cylinders on its own ability to suck in air. Determining how much compared to its displacement will give you its true "Volumetric Efficiency" when compared to its calculated flow. That calculation has a lot of parameters factored in that has to be considered.

Keep in mind the nature of a forced induction motor capable of variable boost levels... has a variable efficiency just simply because of the varying ability to fill the cylinders.

 

[TDIMeister]

forgot to mention PD130 intake manifold.

Good for the sake of completeness and thoroughness of documentation, but I doubt it makes any difference that can be measured. Of course, you now have the method to do this and I would welcome to be proven wrong.

With any motor having forced induction that means you cram more volume of air into the displacement than atmosphere could fill at 100%. So VE should be 100-200 %.

So in truth to find the true volumetric efficiency of the TDI it has to be measured at zero boost or atmosphere......

Edit - D_C and GoFaster addressed this more tactfully than I did.

 

[Digital Corpus]

I remember VE being a function of cylinder filling. NA motors at best were in the 85% range due to the fact that they had to pull in the air through vacuum created by the drop of the piston. That has increased into the 90's but its still a number based on how much air has actually filled the displacement of the cylinder. With any motor having forced induction that means you cram more volume of air into the displacement than atmosphere could fill at 100%. So VE should be 100-200 %.

So in truth to find the true volumetric efficiency of the TDI it has to be measured at zero boost or atmosphere. This means removing the intercooler and turbo as an intake restriction, as for the turbine side if it impacts pumping losses it to must be removed. What is then left is the motors true ability to fill its cylinders on its own ability to suck in air. Determining how much compared to its displacement will give you its true "Volumetric Efficiency" when compared to its calculated flow. That calculation has a lot of parameters factored in that has to be considered.

Keep in mind the nature of a forced induction motor capable of variable boost levels... has a variable efficiency just simply because of the varying ability to fill the cylinders.

That would be mass efficiency, not *volumetric*.

I'm looking at the density of air before and after the turbo and comparing that with full cylinder volume. Though I keep the volume fixed in my calculations, the difference in density means I'd need greater volume to maintain the same mass of air going into the cylinder. By nature of having this a fixed value, the ratio between the two masses is equivalent to the volume exchange.

If I didn't explain that well enough, then I blame being not so wake from not sleeping well due to being sick.

 

[GoFaster]

I remember VE being a function of cylinder filling. NA motors at best were in the 85% range due to the fact that they had to pull in the air through vacuum created by the drop of the piston. That has increased into the 90's but its still a number based on how much air has actually filled the displacement of the cylinder. With any motor having forced induction that means you cram more volume of air into the displacement than atmosphere could fill at 100%. So VE should be 100-200 %.

So in truth to find the true volumetric efficiency of the TDI it has to be measured at zero boost or atmosphere. This means removing the intercooler and turbo as an intake restriction, as for the turbine side if it impacts pumping losses it to must be removed. What is then left is the motors true ability to fill its cylinders on its own ability to suck in air. Determining how much compared to its displacement will give you its true "Volumetric Efficiency" when compared to its calculated flow. That calculation has a lot of parameters factored in that has to be considered.

Keep in mind the nature of a forced induction motor capable of variable boost levels... has a variable efficiency just simply because of the varying ability to fill the cylinders.

You are thinking of the frame of reference being atmospheric pressure. For the purpose of evaluating camshafts and intake ports and the like, the frame of reference is the intake manifold pressure.

Also, it is quite possible for a naturally-aspirated engine to have a VE higher than 100% if pressure-wave and intake-ramming effects are taken advantage of. Normally this is only significant on a higher revving engine and in the upper portion of the rev range AND only if the intake ports are an appropriate size for the engine (not breathing through a straw, but also not oversized to the extent that flow velocities are too low). The piston going down creates a depression (partial vacuum relative to intake pressure) which is then reflected by a plenum back into the intake runner such that at the end of the intake stroke, airflow is still going into the cylinder and its momentum "overfills" the cylinder a little bit.

The same effects are present in a forced-induction intake manifold but simply scaled to the higher reference ("boost") pressure.

It's very hard to make a low-revving engine with mild cam timing take advantage of this.

 

[All Stock]

That would be mass efficiency, not *volumetric*.
.

Actually it is "Volumetric"

The Volume of actual air in a given space compared to the calculated volume is volumetric efficiency.

GoFaster has it right but the frame of reference isn't "just" atmospheric pressure. It could be more than that or whatever but it has to be a reference held constant. The problem is if its not atmospheric then its boost and you will see a higher volume of air placed into the given space. In addition air that is forced in can mask areas needing improvement. Bottom line...the motor is not naturally breathing. Get back to the basics, start building there and you will have a much better thing going on instead of masking it with pressures. Pressure changes everything.

Knowing the VE is huge because given a defined set of parameters you can measure the improvements or reductions any mods will make to the motors inherent efficiency. Add the turbo charger and what is a 5% NA gain through improved efficiency could be a 15 or 20% gain in a forced fed motor. In both cases that's huge.

So, I wasn't trying to be critical of your experiment, which I applaud, but to utilize the information in a way you can seriously benefit from it you have to strip it down of some of its variables to get comparable reference data which then can be used in a duplicated scenario to measure the impact of modifications. In fact if you can accurately do this you will have stumbled on the root gauging tool for every motor build in existence. Remember they are nothing more than an air pump that happens to have a combustion event that occurs during one of its strokes.

GoFaster was also right in resonance tuning that occurs over a very narrow rpm band in an intake manifold thereby allowing 100+ % VE. Former V10 NA F1 cars used reach as high as 150% through this tuning along with other things. In addition that tuning alone to a production car was said to be worth somewhere of 1-2lbs of boost.
Also If I recall correctly... the variable geometry intake tuning was called variorum tuning... just a 25 year old memory I thought Id share.

 

[temporaptor]

I'm not gonna pretend to understand most of the testing and things Michael does but he is part of the reason this Forum is soo great. Keep up the good work and pursuit of perfecting and pushing the performance tdi to the next level. Good job Michael!
Now just gotta get him into V10's and Bmw diesels lol

 

[TDIMeister]

Next step is to measure and display BSFC. Getting fuel flow is easier enough since the ECU already estimates it - needs to be calibrated though! - but getting power output is much more challenging. Ideally a simple strain gauge torque measurement somewhere + RPM or use something like G-Tech to get "power at the wheels".

 

[[486]]

Getting fuel flow is easier enough since the ECU already estimates it - needs to be calibrated though!

Been thinking about measuring it with graduated cylinders. did it not long ago on my megasquirt-ed gasoline motor and it's pretty easy to figure out stuff like injector dead time and flow rate

with a stepper/servo motor drive on the IP to get constant RPM and a pulse generator on the QA to run it at X volts for Y time, then measure fuel output in the cylinder, divide back through pump RPM and time, gets you actual IQ

Could populate a whole pump volt map this way with real accurate and proper numbers.

 

[Digital Corpus]

All Stock,
What you describe for force induction is not volumetric efficiency. Comparing density in a given volume is mathematically equivalent to VE, which is what I did. Our resident internal combustion engineer did no make any corrections to my methodology for this process. Forcing more air into a given volume with forced induction does not change the volume of air exchanged in the cylinder, only the mass of air in the cylinder. As stated later by Go_Faster, NA engines use valve overlap and tuned runners to increase VE, which is not the case in this application. A turbocharger just forces a greater mass of air into the same volume of space so you can burn more fuel. That efficiency at which that occurs is dependent upon the RPM of the engine.

I'm not gonna pretend to understand most of the testing and things Michael does but he is part of the reason this Forum is soo great. Keep up the good work and pursuit of perfecting and pushing the performance tdi to the next level. Good job Michael!
Now just gotta get him into V10's and Bmw diesels lol

Give me time Paul. This is 1/2 of the project and it took 10 hrs when I had the time. I'm sure I can do it in less and if you really want to do it on the TReg and beamer, we just need some time. Then again, I'm not tuning your cars for you cause I don't know how to .

Next step is to measure and display BSFC. Getting fuel flow is easier enough since the ECU already estimates it - needs to be calibrated though! - but getting power output is much more challenging. Ideally a simple strain gauge torque measurement somewhere + RPM or use something like G-Tech to get "power at the wheels".

Getting my fuel map properly adjusted is on the list of to-dos. The info about re-calculating flow with larger injectors doesn't jive with me since the nozzles are 2-stage pressure valves and the amount of fuel flow over the distance of the nozzle orifice shouldn't require *more* voltage for the same, low IQ. If that understanding is wrong, I'd like to know. Currently VCDS spits out at least a 0.3 V difference between the specified IQ vs voltage map and the corresponding IQ so I know its way off. New clutch means I can test that. Might have it installed in 24 hrs else it'll wait until Tuesday.

FUB demonstrated a live dyno on the ALH, but the B4's platform does have the engine mounted in a pendulum. A-arms are too mobile to get solid readings, I don't have weight distributions for the 3 engine mounts, and the axles spin too fast for me to ignore centripetal forces methinks. Suggestions?

Been thinking about measuring it with graduated cylinders. did it not long ago on my megasquirt-ed gasoline motor and it's pretty easy to figure out stuff like injector dead time and flow rate
with a stepper/servo motor drive on the IP to get constant RPM and a pulse generator on the QA to run it at X volts for Y time, then measure fuel output in the cylinder, divide back through pump RPM and time, gets you actual IQ
Could populate a whole pump volt map this way with real accurate and proper numbers.

It's not quite that simple. Injection pressure has an immense effect on the cavitation in the nozzle orifice which affects flow rather significantly.:

https://www.bsc.es/annual-report/2009/wikiar.bsc.es/index.php5/File_Imagen_Tirant.html

 

[[486]]

...It's not quite that simple...

You would be using the injectors for this, would even set up with the stock lines so that the length (and compressed volume) is taken into account. Probably easiest to set up on the bench, but it could be done on the car.

Hell, maybe spin it with the lathe. Only get a few different pump speeds, but you can interpolate the values in between.

 

[mk1-83]

Wait for few weeks I have bought a hartridge 700 ve pump test bench

https://www.youtube.com/watch?v=gcmSHR0jrYE

 

[All Stock]

My apologies for not understanding... I am confused..

Volume can be measured by the cubic foot. A cylinder only has so much volume.
A turbo is a compressor, it can compress air because its compressible.. In doing so It can force 10 cubit feet of air into 1 cubic feet of space. In that process it creates pressure by the compression. It has also increased the capacity "VE" of that space 10x. The amount of air that is in that space does in fact have a specific weight "mass" which is referenced at sea level. At higher altitudes it weighs less because there is less atmosphere and air above the proverbial scale.

So the question I am now wondering...what are we really even talking about??
So I went and reviewed a couple of white papers on VE and in conjunction with turbocharged applications along with a library I've collected over the years...
I am still convinced I am understanding this correctly as I have for the last 25 years because VE is still the same thing.

So I'll refrain from the conversation with this...

I really doesn't matter what any of the data is called. The question is whether or not you can accurately repeat the test (why I brought the issue up and in regards to having a variable compressor involved) thereby creating a baseline to which you can gauge the direction of changes a modification can make.

Years ago I built a flow bench... It would give very consistent w.c. measurements. However I can say with confidence that it wasn't calibrated to any SAE scale.. But it was consistent and therefore a great and accurate tool to see the difference whether positive or negative a change made against the repeatable point of reference.

I hope that this work creates the same scenario, as I have from the beginning... My point in chiming in was to show that for a true "VE" calculation the variable of the turbine is means you have to use adjusted numbers, which means even the slightest inconsistency mathematically could change the results exponentially. To get a solid repeatable baseline means you have to have as many constants as possible.

My intentions were to help the fine tune the tool, not hinder it... You really are doing good stuff! Best wishes on the project.

 

[turbobrick240]

Damn, 10 cubic units into 1 is nearly 150 psi of boost. Hope those clamps are good and tight.

 

[Digital Corpus]

density = mass / volume
mass = density * volume
volume = mass / density

If you compress 10 cubic units of a material into 1 cubic unit of space, you have a pressure ratio of 10:1. If you're at a high(er) altitude, this could be 120 psi. At STP, this would be ~147 psi. Your starting point has to be known.

That example doesn't change the final amount of space you're compressing the air into, such as a cylinder. This then means that volumetric efficiency is not changed simply because you increase the density of air inside the same volume even though it is "more" air. You have to define what "more" you're talking about.

The sensors on the car allow for measuring the mass of air before the turbocharger and the pressure and temperature of the air after the turbocharger. Thanks to weather services, we know what the general characteristics of the air entering the engine is like, i.e. 75 °F, 18% RH (relative humidity). From using appropriate psychrometric equations, the density of this air can be computed. Knowing the volume of the cylinder at BDC (bottom dead center) we can then know how much mass can fit inside the said cylinder. Follow so far?

Now, we do not have any sensors after the cylinder, i.e. in the exhaust stream that can give us mass & volume or pressure & temperature let along you'll have to remove the combustion products which requires a precise knowledge of the air composition, fuel composition, and other very complicated parameters of the cylinder. However, we know how much air should fit into the cylinder and we have parameters that allow us to measure how much air actually goes into the cylinder.

Because we know the mass of air being consumed by the engine and have pressure and temperature after the turbocharger, we can compute it's density before and after it. Since density is mass divided by volume, you can change the mass or the volume to change the density. As we don't know the volume of air that will go into the cylinder, but we know the mass of air that will go into the cylinder, we just have to run a few numbers to normalize the density before and after the turbo charger.

Normalization basically means we level the playing field. Even though we know the density of the air going into the turbochrager, we don't know the density of that air coming out. However, we know this density will be different than what it consumed. As such we cannot use the pre-turbo air density to know how much mass fits into the cylinder, but we have to use the post-turbo density. Humidity affects the density of air. Using appropriate psychrometrics, we can convert relative humidity to absolute, change the temperture of the air, and then determine relative humidity once again. Once we have temperature, pressure, and relative humidity again, we can then compute post-turbo air density and, by proxy, the mass of air that will go into the cylinder at post-turbo density.

Since we then keep the volume consistent between pre- and post- only the mass changes. Recall that you can change either the mass or the volume it occupies to change density. We normalized the density with a fixed volume so the ratio in mass going into the cylinder is equivalent to the ratio of volume of air being exchanged inside. In short, even though mass was measured, the resultant ratio is equivalent to the volume of air exchanged and this is then independent of the difference in pressure between the starting point and the ending point. As such, we measured volumetric efficiency and just because you have a forced induction engine, you don't have a change in volumetric efficiency.


I think that explains everything from point A to point B for this experiement. It's higher-up high school physics and/or 1st year college physics. Physics is the practical application of math, or an approximation thereof.

Side note: I took 2 samples of data at 2250 RPM too. The 1st set computes to 88-ish & VE which removes that plateau and allows the curve to continue naturally. I'll update the graph and data set within the next 24 hrs.

 

[Digital Corpus]

Wait for few weeks I have bought a hartridge 700 ve pump test bench

https://www.youtube.com/watch?v=gcmSHR0jrYE

FASTER!

Seriously. You have different pump cam plates and nozzles to toss in to test?

 

[mk1-83]

yes i have many nice camplates to test im even disgin a 13mm ve pump with dlc coating on the piston and im sell oem 12mm heads

and new nozzel .390 5 hole dlc needle

 

[Digital Corpus]

So my long desire was to have one Measuring Block able to log 4 pieces of data in one go and up my sample rate and accuracy of measuring V.E.. Well, happenstance would have it so:

HOWTO: Modification of VAG EDC16 measuring blocks, limits
EDC15, define your own log

The initial hex search revealed 2 pattern hits each in the couple of tune files I'm playing with; both in the same location, 0x02DE2 codeblock 2 and 0x3ADE2 for codeblock 1.

I compared it to my custom VCDS label file that more closely matches the language commonly used here and all is up to snuff.

In the MSA15.5 tunes for the Passat and similar ECUs, there are generally 4 measuring blocks that can be repurposed. The blank ones are 14, 17, and 25, and number 9 is for automatic transmissions. If you *need/want* more, number 18 contains your speed and a not-publicly-known bit, and number 24 is listed as Temperature Compensation in the Bently with an also not-publicly-known bit.

A more practical example of custom measuring blocks would be that of adding or modifying number 11 to include IQ instead of MAP Specified (my nomenclature) to aid in adjusting the N75 map without running 2 logs. Similarly substituting Coolant Temp for SOI Specified in number 4 for the MKIV platforms and the multiple, temp-based SOI maps.

Immediately after the maps mentioned (25 rows, 4 columns btw), are the data definitions of the ID numbers stored in the said maps. For the ECUs I'm looking at, I have 49 definitions for the maps are 49 rows and 4 columns each. I'm not quite at the point where I know how to interpret this map yet, but there look 2 or 3 open spots and room for a 50th sensor without overwriting other data.