Donations towards server fund so far this month.

 
£0.00 / £100.00 per month
Page:
Home > MS Trials & Testing > Megasquirt Port Injection Trial

matnrach

152 Posts
Member #: 1074
Advanced Member

Northamptonshire

I wonder if you had two injectors per port, one firing upstream, one downstream and calculated the transport delay (hence position relative to the valve) that it could still work with the current code.


matnrach

152 Posts
Member #: 1074
Advanced Member

Northamptonshire

I wonder if you had two injectors per port, one firing upstream, one downstream and calculated the transport delay (hence position relative to the valve) that it could still work with the current code.

Just thought that this could be difficult as we dont know the velocity of the jet or the direction of the travelling waves within the port at any one time.
Maybe easier just to try it!

Edited by matnrach on 26th Jan, 2008.


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 26th Jan, 2008 matnrach said:
I wonder if you had two injectors per port, one firing upstream, one downstream and calculated the transport delay (hence position relative to the valve) that it could still work with the current code.


The problem is that the ideal point of injection, in terms of cranks degrees, varies with the engine volumetric efficiency. So you would need some variable length runners.

I'm going to attempt to calculate the advance/retard curve for the siamesed code so that we have a reasonable starting point.

If you start with known parameters of port area and length, engine airflow based on rpm and cylinder capacity, you just need to estimate the engine VE to calculate the advance required to get the fuel in the outer cylinder just as the inlet valve is opening.

I just need a torque curve for my engine, then I can estimate the VE and go from there.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

ve would effect the pulse width , rpm and flow speed the point of inj .in my rarely expressed and ever so humble opinion . *wink*

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

I have based my calcs on Marcel's method as detailed here:

http://www.starchak.ca/efi/speed.htm

I've applied volumetric efficency to reduce the airflow and get the following results:

http://www.jaservices.co.uk/photos/Injecti...se%20Timing.pdf

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

yep the point is paul , the ve doesnt reflect the mass of air that is going into the cyl , it reflects the mass of air that is going into the cylinder,,,and being trapped there upon the valve shutting ... ,eg
if you have a load of air going out of the ex port in crossover ,then not all of it getting back to the cyl ,your calc will show one ve for the air flow ,and one different ve for the torque output ,this is called in analysis ,'''ve1'' and ''ve2''

another scenario ,is where you have a strong pulse wave tuning effect ,and the density of the gas moving through the port has a certain speed ,but a change in density ,this can really make a mess of calcs based on averages.

looking at the links to marcels stuff , i cant see a component that adresses the fact that the valve is only open fully a few degrees per inlet period and even then the port is not the size of the cross section , but the size of the valve opening ,and then not acting that due to the cross sectional area change as it flows past the valve ,making the port act a lot smaller ,and the rest of the opening sequence the valve is at less than full lift so effectively the port acts even smaller ,also the piston at tdc is not creating the draw it would at 90 degrees to tdc ,so the port again drops flow at tdc ,but then again inertia and port resonance can affect that to create more flow too !
i think the idea to rely upon the torque curve data is a better one ,but you need to get accurate data for that too ,remember that the long stroke engine creats a lot more friction than a modern engine with a shorter stroke ,so results have to take this into account ,the engine may be making a lot of torque ,but what gets to the rollers may be less , and have to worked back theoretically to asses the actual cylinderr torque , unfortunately ,the coast down method used on sun rr's will not measure the internal friction levels unless you want to take it up to 100mph ,and then whip the plugs out as it coasts down and let out the clutch in the test gear ,hey ,that sounds like a fun experiment !

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

This is a very complicated subject if you start to consider what is happening in too much detail. My intention is to try to get a base timing curve that will get the car working and then Sturgeo will fine tune it whist I'm driving it with my eyes fix firmly on the road ahead *wink*

I've backward calculated some VE numbers based on a typical power curve for an NA 1275:

http://www.jaservices.co.uk/photos/Airflow%20Calcs.pdf

I've then applied them to the spreadsheet above:

http://www.jaservices.co.uk/photos/Injecti...ming%20RevA.pdf

This is all at WOT, so I know need to consider the effects of vacuum.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

yep thats what i would do too paul , looking at the sheets , i may be drivelling but your not going to get the same bhp at 6k as at 5k if it flows the same amount of air ,due to the internal friction going up maybe 5 to 8 bhp .

hmmm yes vacuumwize its a intersting sit ,air speed i guess must drop as well as density the mass of the charge will be less so inertia lower ,the map for inj point will have to be 3d wont it revs versus map and degrees in the boxes ?oh and air temp compensation, i imaging this is old hat and all the people involved in this know all this , im just getting up to speed being laid up and nothing to do lol.

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM


alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France




On 28th Jan, 2008 Paul S said:


The problem is that the ideal point of injection, in terms of cranks degrees, varies with the engine volumetric efficiency. So you would need some variable length runners.


Paul, could you explain why the transport delay would depend on VE or MAP ?

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 28th Jan, 2008 alpa said:

Paul, could you explain why the transport delay would depend on VE or MAP ?


The above spreadsheets show how the transport delay may vary with VE.

I'm still thinking about the effects of MAP.

I'm inclined to think that as the pressure drops under vacuum, the volume increases and the speed goes up, so less advance required. I need to sleep on it.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 28th Jan, 2008 robert said:
im just getting up to speed being laid up and nothing to do lol.


Are you unwell? I'll try and keep you entertained.

I'm working, but have about 3/4 more weeks crunching numbers to get this catchment model built, hence, I need this distraction.

Other than test drive the car, I've only left the village about three times since Christmas!

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France




On 28th Jan, 2008 Paul S said:

On 28th Jan, 2008 alpa said:

Paul, could you explain why the transport delay would depend on VE or MAP ?


The above spreadsheets show how the transport delay may vary with VE.

I'm still thinking about the effects of MAP.

I'm inclined to think that as the pressure drops under vacuum, the volume increases and the speed goes up, so less advance required. I need to sleep on it.


I'm sorry, I've read all your documents. I don't see any explanation of why.
Why would the air speed change for different pressures ?
Do you then make connection between MAP and VE ?

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 28th Jan, 2008 Paul S said:

I'm still thinking about the effects of MAP.

I'm inclined to think that as the pressure drops under vacuum, the volume increases and the speed goes up, so less advance required. I need to sleep on it.


After further consideration, I do not think that MAP has a significant affect on the advance. Principally because MAP affects mass flow rather than volumetric flow.

It's a fact that VE affects the volumetric flow at WOT and the above calcs will be my starting point.

I think that maybe, my calculation of VE at 1000rpm is rather pessimistic, because I'm sure that I do not have that much advance at present and it is still injecting too early.

I'm hoping to generate an .msq file, flash in the siamesed code and give it a whirl at the weekend. As long as there is not too much snow!

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

!!!is this whole siamese code allready in existence then paul ?

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM


miniminor63

User Avatar

1849 Posts
Member #: 672
The oversills police

Oslo, Norway

yes it is. Jean finished it and made available for testing quite a while ago.


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 30th Jan, 2008 robert said:
!!!is this whole siamese code allready in existence then paul ?


Do try and keep up, robert.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

gosh ,the amount of times i ve heard that in the bedroom!
(but not , i should hasten to add ,from paul!)

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM


miniminor63

User Avatar

1849 Posts
Member #: 672
The oversills police

Oslo, Norway

lol!


alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France




On 30th Jan, 2008 Paul S said:

After further consideration, I do not think that MAP has a significant affect on the advance. Principally because MAP affects mass flow rather than volumetric flow.

It's a fact that VE affects the volumetric flow at WOT and the above calcs will be my starting point.


Yes Paul, more precisely VE affects especially in the NOT throttled mode because the bottleneck is on the engine side, not in the intake plenum/throttle. (At low RPM the NOT throttled mode is reached far before WOT). We know when we are in throttled mode, it's when MAP is lower then the ambient pressure.
In throttled mode VE affects the instant air speed, but its effect on the average speed is difficult to predict and I would not consider it.

My main problem now is to find how to weight outer injection time to get the same richness as in the inner. I don't think it's just a stupid multiplier, I believe it depends on the fuel film thickness and transport delay.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm


alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France

Yesterday I read some recent papers to check if Heywood's statements are still confirmed. It looks like they are still true, the fuel film dynamics can be described by a liquid fraction (MAP/engine temperature dependend) and evaporation time (tau, 0.5-1.5 seconds on MPI).
I believe the fuel extra of the inner cyl is the amount of fuel that has evaporate since the outer pulse (1.5 revolution duration) and inner pulse (0.5 revolution duration). So to get the correction we have to do:
<inj pulse> * fuel fraction * 2 * <rev duration> * <custom coeff> / <evaporation time>
It's not very precise as <inj pulse> should be different for both cyls but should work. I'm going to test that may be this WE, the time to implement the algorithm.
All modern correction algorithms work on an event (intake event) and not monotonic rate basis.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm


alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France

Yesterday I read some recent papers to check if Heywood's statements are still confirmed. It looks like they are still true, the fuel film dynamics can be described by a liquid fraction (MAP/engine temperature dependend) and evaporation time (tau, 0.5-1.5 seconds on MPI).
I believe the fuel extra of the inner cyl is the amount of fuel that has evaporate since the outer pulse (1.5 revolution duration) and inner pulse (0.5 revolution duration). So to get the correction we have to do:
<inj pulse> * <fuel fraction> * 2 * <rev duration> * <custom coeff> / <evaporation time>
It's not very precise as <inj pulse> should be different for both cyls but should work. I'm going to test that may be this WE, the time to implement the algorithm.
All modern correction algorithms work on an event (intake event) and not monotonic rate basis.

Edited by alpa on 1st Feb, 2008.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

I'm not convinced that evaporation is the problem, but you probably know a lot more than me about this subject, so best of luck.

I'm all excited because I've now got the siamesed code on the ECU, Megatune is talking to it and the .msq file is ready to drop in. The weather forecast for tomorrow is good although its going to be cold.

So watch this space. I'll start a new thread to record the results (if it works OK).

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."


TurboDave16V
Forum Mod

10979 Posts
Member #: 17
***16***

SouthPark, Colorado

Looking forwards to it!

On 17th Nov, 2014 Tom Fenton said:
Sorry to say My Herpes are no better


Ready to feel Ancient ??? This is 26 years old as of 2022 https://youtu.be/YQQokcoOzeY



alpa

520 Posts
Member #: 2093
Post Whore

Grenoble, France

It's going to snow here in Grenoble :(
Looking forward your tests.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm

Home > MS Trials & Testing > Megasquirt Port Injection Trial
Users viewing this thread: none. (+ 1 Guests) <- Prev  
To post messages you must be logged in!
Username: Password:
Page: