Intake Port Dynamic Pressure - High Speed Data on Kiggly Racing Drag car
Jan 18, 2015, 05:33 AM
#1
Intake Port Dynamic Pressure - High Speed Data on Kiggly Racing Drag car
While this isn't an EVO, the 2g DSM cylinder head and intake ports are just about identical in OEM form to the EVO aside from the injector location. All this applies directly to EVO's.
Over last winter I worked out a high speed data acquisition on my drag car, here are some combustion pressure and intake port pressure plots. Each channel is taken at 60,000Hz, which is fast enough to have a data point at every crank angle degree at 10,000rpm. I had a pile of questions I was looking to answer by generating this data and also a big pile of new questions have popped up since taking data at a couple dyno sessions. It has been a ton of programming and learning Octave to make some reasonable routines, but I have what I feel is a pretty good home-brew combustion system now and I can process the data quick enough to make decisions between dyno pulls. Ten seconds of data is about 25megs.
One early question I wanted to answer was what pressure do the intake valves really see during high rpm operation. My scenario is a bit unique as I'm running my own sheetmetal intake manifold and a 2g cylinder head, but it still gives a pretty good data point. The data below was gathered up at Force Engineering, Tyler and I had a blast putting this through the paces and learning a ton of new details. The pull below made approximately 740whp on a mustang dyno through the auto trans. It was on a #7 precision converter, which stalls at about 8300rpm at this power level.
Notes:
The green trace is cylinder pressure, this is a 5000psi sensor
The blue trace is intake port pressure, it is a 150psi sensor located about 50mm up the port
The red vertical lines are exhaust valve opening and closing events
The blue vertical lines are intake valve opening and closing events
This is 18psi and 6000rpm, dynamic pressure bounces around about 10psi and peaks at near 25psi at the back of the valve (40psi absolute). This rpm phases nicely with the intake cam advanced all the way and the pulse activity lines up nicely, but it is a little due to low port velocity.
This is 7500rpm and 40psi, dynamic pressure bounces around about 30psi and peaks at near 55psi at the back of the valve (70psi absolute). This RPM is about the best VE, at just over 110%. I can probably gain some by running a little more cam advance.
This is 8100rpm and 46psi, dynamic pressure bounces around about 40psi and peaks at near 70psi at the back of the valve (85psi absolute). This is peak torque and about 108% VE.
This is 9700rpm and 45psi, dynamic pressure bounces around about 50psi and peaks at near 75psi at the back of the valve (90psi absolute). It appears the header is tuned wrong for this rpm and not clearing the cylinder out well at exhaust valve close. Peak power is already past at about 8900rpm.
The point I was initially trying to get a handle on was how hard the port pulses were trying to pop the valves open. This proves it is WAY above boost pressure. Make sure you have good seat pressure as you start cranking things up. I'm going after exhaust port pressure next trip to the dyno, which I expect to be a good bit worse.
Thanks,
Kevin
Over last winter I worked out a high speed data acquisition on my drag car, here are some combustion pressure and intake port pressure plots. Each channel is taken at 60,000Hz, which is fast enough to have a data point at every crank angle degree at 10,000rpm. I had a pile of questions I was looking to answer by generating this data and also a big pile of new questions have popped up since taking data at a couple dyno sessions. It has been a ton of programming and learning Octave to make some reasonable routines, but I have what I feel is a pretty good home-brew combustion system now and I can process the data quick enough to make decisions between dyno pulls. Ten seconds of data is about 25megs.
One early question I wanted to answer was what pressure do the intake valves really see during high rpm operation. My scenario is a bit unique as I'm running my own sheetmetal intake manifold and a 2g cylinder head, but it still gives a pretty good data point. The data below was gathered up at Force Engineering, Tyler and I had a blast putting this through the paces and learning a ton of new details. The pull below made approximately 740whp on a mustang dyno through the auto trans. It was on a #7 precision converter, which stalls at about 8300rpm at this power level.
Notes:
The green trace is cylinder pressure, this is a 5000psi sensor
The blue trace is intake port pressure, it is a 150psi sensor located about 50mm up the port
The red vertical lines are exhaust valve opening and closing events
The blue vertical lines are intake valve opening and closing events
This is 18psi and 6000rpm, dynamic pressure bounces around about 10psi and peaks at near 25psi at the back of the valve (40psi absolute). This rpm phases nicely with the intake cam advanced all the way and the pulse activity lines up nicely, but it is a little due to low port velocity.
This is 7500rpm and 40psi, dynamic pressure bounces around about 30psi and peaks at near 55psi at the back of the valve (70psi absolute). This RPM is about the best VE, at just over 110%. I can probably gain some by running a little more cam advance.
This is 8100rpm and 46psi, dynamic pressure bounces around about 40psi and peaks at near 70psi at the back of the valve (85psi absolute). This is peak torque and about 108% VE.
This is 9700rpm and 45psi, dynamic pressure bounces around about 50psi and peaks at near 75psi at the back of the valve (90psi absolute). It appears the header is tuned wrong for this rpm and not clearing the cylinder out well at exhaust valve close. Peak power is already past at about 8900rpm.
The point I was initially trying to get a handle on was how hard the port pulses were trying to pop the valves open. This proves it is WAY above boost pressure. Make sure you have good seat pressure as you start cranking things up. I'm going after exhaust port pressure next trip to the dyno, which I expect to be a good bit worse.
Thanks,
Kevin
Jan 18, 2015, 06:46 AM
#2
Evolved Member
Nice of you to post up your finding/results and thank for sharing the info with the rest of us, it is very informative for those looking to do a seriously proper built as it gives one very good guidelines as to how to get close to the optimum V.E of their intended setup.
Achieving as close as one could, to a balanced intake port pulses pressure/exhaust port pulses pressure is the best volumetric efficiency of a setup.
Reaching twice or very very near to twice the boost pressure on intake port pulses pressure is the max an internal combustion engine can offer.
You are right exhaust pressure will be substantially lower.
Marios
Achieving as close as one could, to a balanced intake port pulses pressure/exhaust port pulses pressure is the best volumetric efficiency of a setup.
Reaching twice or very very near to twice the boost pressure on intake port pulses pressure is the max an internal combustion engine can offer.
You are right exhaust pressure will be substantially lower.
Marios
Jan 18, 2015, 07:13 AM
#3
Thanks for sharing this info. I see multiple traces for port pressure and cylinder pressure (looks like 3). Were these traces from different cylinders, or several readings from one cylinder?
How are you reading cylinder pressure? I guess you don't need a knock sensor anymore
How are you reading cylinder pressure? I guess you don't need a knock sensor anymore
Jan 18, 2015, 07:36 AM
#4
These plots are all 3 engine cycles in a row from the same cylinder. I only have sensors on cylinder #4, there is really no good place on the mitsu head to fit a pressure sensor to the chamber except coming in from the ends of the cylinder head. The cylinder pressure sensor is an Optrand 5000psi sensor directly in the chamber. To get the type of frequency response needed for knock you absolutely need to have the sensor in the chamber, otherwise the cavity to the sensor acts as a low-pass filter and cuts off all the high frequency response.
This is part of the data I'm still processing, but the knock sensor itself does a pretty poor job of showing knock. I figured most of the challenge would just be filtering out the mechanical noise (there is a ton), but the knock sensor signal sometimes doesn't even measurably increase when the cylinder pressure shows a large rise. Severity of knock is really easy to see on the pressure sensor.
This is part of the data I'm still processing, but the knock sensor itself does a pretty poor job of showing knock. I figured most of the challenge would just be filtering out the mechanical noise (there is a ton), but the knock sensor signal sometimes doesn't even measurably increase when the cylinder pressure shows a large rise. Severity of knock is really easy to see on the pressure sensor.
Trending Topics
Jan 18, 2015, 10:05 PM
#8
Are you talking about logged knock sensor activity through your system at 60kHz or just what the ecu is reporting?
Very cool to see this on a setup like yours. I've seen a decent amount of this type of stuff from OEM type papers but never on a setup running 40+psi and 9000+ RPM. Solid proof turbo motors are just as sensitive to pulse tuning as N/A motors.
Interesting to see you say your motor topped out at 110% VE. That's where I found my setup topped out as well, although I had targeted the cams and intake manifold to hit peak VE around 6000 RPM.
Very cool to see this on a setup like yours. I've seen a decent amount of this type of stuff from OEM type papers but never on a setup running 40+psi and 9000+ RPM. Solid proof turbo motors are just as sensitive to pulse tuning as N/A motors.
Interesting to see you say your motor topped out at 110% VE. That's where I found my setup topped out as well, although I had targeted the cams and intake manifold to hit peak VE around 6000 RPM.
Jan 19, 2015, 01:27 PM
#9
EvoM Guru
iTrader: (1)
Yes, you can see in all the logs the cylinder drops below port pressure leading up to the intake valve opening after the exhaust stroke.
Jan 19, 2015, 05:09 PM
#11
Wreckleford, you are correct on the pressure traces in this example, mis-timed pulses from running a manifold out of its range or trying to close valves when spikes are occurring could certainly lead to some issues. The vast majority of valvetrain issues people have and spring wear is almost always on the exhaust side, so will hopefully be some more interesting data. The point of this was the spikes are high enough they could be a liability. This example is a setup that works really well, so it isn't a great benchmark of things that can go wrong on pulse tuning.
03whitegsr, I'm talking about knock sensor signal logged at 60kHz and then post-processed in Octave. I've looked at a few different windows, filters, and data processing techniques and while there is a paired general trend with real pressure knock signal, it does not have a rock solid correlation. I can see the same knock sensor signal for knock events that are a factor of 3 more intense or worse. Knock sensors are cheap and can indicate knock to a better degree than anything even close to the same price range. None of it is an exact science, there is major variability cycle to cycle, particularly in knock. That is probably the biggest surprise I've had in this entire data acquisition setup and processing experience.
Peak VE in the maps are just under 120%, but some of that is blow-through the chamber and not trapped gases. What I'm actually reporting at 110% is packing efficiency or how much is left in the chamber compared to total engine airflow.
Aby, I'm seeing max pressures at about 3000psi with very low timing and an alcohol blended fuel (sorry, I'm not sharing the exact proportions). Burn rates on straight methanol and straight ethanol were identical within test repeatability. I am switching the car over the straight methanol for next season. At 9000rpm this is in the 8deg timing range as I was seeing some severe knock limitations. Knock gets scary from a parts damage point of view before it alters cylinder pressure in any meaningful way at these boost levels.
Jan 19, 2015, 08:02 PM
#14
These plots are all 3 engine cycles in a row from the same cylinder. I only have sensors on cylinder #4, there is really no good place on the mitsu head to fit a pressure sensor to the chamber except coming in from the ends of the cylinder head. The cylinder pressure sensor is an Optrand 5000psi sensor directly in the chamber. To get the type of frequency response needed for knock you absolutely need to have the sensor in the chamber, otherwise the cavity to the sensor acts as a low-pass filter and cuts off all the high frequency response.
This is part of the data I'm still processing, but the knock sensor itself does a pretty poor job of showing knock. I figured most of the challenge would just be filtering out the mechanical noise (there is a ton), but the knock sensor signal sometimes doesn't even measurably increase when the cylinder pressure shows a large rise. Severity of knock is really easy to see on the pressure sensor.
This is part of the data I'm still processing, but the knock sensor itself does a pretty poor job of showing knock. I figured most of the challenge would just be filtering out the mechanical noise (there is a ton), but the knock sensor signal sometimes doesn't even measurably increase when the cylinder pressure shows a large rise. Severity of knock is really easy to see on the pressure sensor.
http://www.kistler.com/us/en/applica...or-test-bench/
I am curious on why the runners pressure can exceed actual boost pressure. What is the lowest raster rate in ms could your acquisition tool measure? and how does the pulsation look at idle?
Thanks for sharing this data!
Edit: AVL carries those sensors as well.
https://www.avl.com/documents/10138/...7-4f34e948f95d
Last edited by detroit pistins; Jan 19, 2015 at 08:11 PM.
Jan 20, 2015, 09:59 AM
#15
Static pressure can exceed boost pressure because you are dealing with a dynamic system. Same reason an N/A motor can easily exceed 100% VE despite being feed by ambient pressure.
For clarity here, Kiggly is measuring static pressure (port wall pressure tap) in a dynamic system. In fluidynamics terms, this is not dynamic pressure that would have to be calculated from stagnation pressure and static pressure. Small technicality but it is worth noting before continuing on.
Detroit Pistins, keep in mind here that static pressure above boost pressure means the mass flow pulse is piling up and flow is slowing down at that point. Dynamic pressure (kinetic energy) is being changed into static pressure (potential energy) due to a flow restriction (back of the intake valve). This is why static pressure drops during the main portion of the intake cycle, air is accelerating as it fills the cylinder which means dynamic pressure increases but static pressure in the runner decreases (-360 to -200* crank position) as the valve starts closing, the air starts slowing down and static pressure builds as the air crashes into the back of the valve and the raising static pressure in the cylinder.
It's desirable to have a high static pressure as the valve closes to ensure the flow doesn't reverse as the piston travels up the bore. It's also desirable to have a low pressure during the main portion though as it likely means air velocity is high.
Greatest info thread on evom right here. Thank you kiggly.
For clarity here, Kiggly is measuring static pressure (port wall pressure tap) in a dynamic system. In fluidynamics terms, this is not dynamic pressure that would have to be calculated from stagnation pressure and static pressure. Small technicality but it is worth noting before continuing on.
Detroit Pistins, keep in mind here that static pressure above boost pressure means the mass flow pulse is piling up and flow is slowing down at that point. Dynamic pressure (kinetic energy) is being changed into static pressure (potential energy) due to a flow restriction (back of the intake valve). This is why static pressure drops during the main portion of the intake cycle, air is accelerating as it fills the cylinder which means dynamic pressure increases but static pressure in the runner decreases (-360 to -200* crank position) as the valve starts closing, the air starts slowing down and static pressure builds as the air crashes into the back of the valve and the raising static pressure in the cylinder.
It's desirable to have a high static pressure as the valve closes to ensure the flow doesn't reverse as the piston travels up the bore. It's also desirable to have a low pressure during the main portion though as it likely means air velocity is high.
Greatest info thread on evom right here. Thank you kiggly.
Last edited by 03whitegsr; Jan 20, 2015 at 11:07 AM.