GMVE decomposition

[Ryan]

My first post here; I hope it's a good one. I have come upon a wall and no searching has uncovered anything else, so hopefully it is not just ancient information I have failed to dig up. :)

A little background:
I came across one of RedHardSupra's posts on here in regards to the lack of attribution of charge temperature in SD tuning in addition to some of his posts about temperature modeling.

Looking at the EFILive documentation for the Main VE tables (and RHS's Speed Density document), the GMVE dimensions are [ m * T / P ], resulting in an airmass when the sensor inputs are applied to their appropriate values.

With some manipulation, we can get:
Temperature = GMVE * Pressure / airmass

I also had the thought that, using a wideband, we can also (roughly) calculate the airflow (or airmass, accordingly) using the calculated injector flow rates and wideband output.

Alternatively, since I have been unable to drive around and get some calculated injector flow rates, I have assumed that the flow rates will stay roughly constant across the RPM range.

So with some conversion between the air/fuel masses and rate conversions, I have come up with:
GMVE = (120 * AFR * IFR * Temperature) / (RPM * MAP * cyl)
and
Temperature = (GMVE * RPM * MAP * cyl) / (120 * AFR * IFR)

Now, assuming that the target AFR before any modifiers is 14.63 (a major assumption, but I don't think it significantly changes the results), I used the data I have from the stock calibrations to come up with an array of what would be (or what I thought would be) "Temperature" from the GMVE and IFR across the MAP/RPM axes.

I made a quick spreadsheet in Excel to mash all of the numbers together and return some kind of plot of the value for temperature from MAP/RPM, and got a strange looking (exponential) curve (see attached Excel spreadsheet).

My questions: are there any other values that go into the calculation for the GMVE dimension (unitless values or some function)? The surface I have plotted for stock values seem to indicate some kind of natural exponential curve. Also, what does these numbers represent? They don't appear to be temperature values as they are unreasonably high ( > 2000 K).

I've attached the spreadsheet I used as well as an image of the resulting plot (for anyone that can't use Excel but is interested). I don't think I have made any kind of algebraic or formula-related mistakes, and seeing as how the values which I have assumed (i.e. constant stoich AFR) would vary linearly or approximately linearly where as the plot is obviously nonlinear, I don't think I have made any unreasonable assumptions. Please forgive the sloppy methods of referencing RPM and MAP, I am definitely no Excel god, and didn't intend on anyone else seeing the spreadsheet. :) Sorry for the novel!

[redhardsupra]

that's a one helluva first post ;)
you got a good idea, but unfortunatelly getting airmass from fuel consumption is way too unreliable to use it as a source for computations (not that it ever stopped any AutoVE using folk...)


here's how i think about it:
there's multiple ways to calculate Cylinder Air Mass (CAM):
from MAF: CAM1=120*MAF/(RPM*CYL)
from FUEL: CAM2= IFR*IPW*AFR
from GMVE table: CAM3=GMVE*MAP/TEMP

so to tune a GMVE table, you gotta make sure that the two CAM's in question are equal:
CAM2=CAM3
IFR*IPW*AFR=GMVE*MAP/TEMP
so:
TEMP=GMVE*MAP/(IFR*IPW*AFR)

here's the first problem i noticed with your post: you forgot to include IPW in your calculatation of airmass from fuel. this is why your final graph looks funny.

there are multiple practical problems:
IFR is a function of fuel pressure and injector flow characteristics. fuel pressure fluctuates quite a bit, so to assume that IFRreal==IFRfromIFRtable is less than optimal. It is the goal, but it's a hard one to achieve.
IPW is also a troublesome variable, as we dont know if that's the pulse width before or after all the modifiers that can apply (short pulse adder, voltage offset, there's some other ECT/IAT adders too if i recall corectly).
so the whole IPW*IFR bit is very very sketchy.
also, with the formula for temp, how do you know if the errors stem from GMVE, MAP, IFR, IPW or AFR? GMVE is a grid of numbers, interpolated in an unknown manner, MAP is a sensor so there might be calibration/wear issues, IFR and IPW i discussed before, and AFR from WB often has problems with not being up to temp, airleaks, dying sensors, wrong placement in the exhaust... all these errors are gonna compoud quickly.

even then, what is this gonna give you? TEMP by itself is kinda useless, the useful stuff is the parameters determining the TEMP from IAT, ECT, BIAS, and FILTER tables. if you start trying to get TEMP you think is correct by adjusting GMVE, then you have an inverse of people trying to tune GMVE by fooling TEMP with resistors soldered to an IAT sensor. neither approach is any good.

great attempt tho, your excel approach is actually very Matlab'ish, nothing wrong with that ;)

[joecar]

Ryan, welcome to the forum, very good 1st post... :cheers:

Good thread, good reading... :cheers:

[Ryan]

Thanks for the replies. :cheers:

I wondered if assuming the IFR from the table in the tune definition might have been a bit off; the only reason I thought it might be valid was that because the "calculated injector flow rate" PID showed values approximately equal to the flow rates at their respective MAP pressures. I didn't even think to consider the pulse widths.


even then, what is this gonna give you? TEMP by itself is kinda useless, the useful stuff is the parameters determining the TEMP from IAT, ECT, BIAS, and FILTER tables. if you start trying to get TEMP you think is correct by adjusting GMVE, then you have an inverse of people trying to tune GMVE by fooling TEMP with resistors soldered to an IAT sensor. neither approach is any good.

My main interest was trying to see if there was any useful information that could be revealed through the reversal of the GMVE unit. I had thought that there might be some kind of coefficient relating the temperature of the charge to its density (or maybe some kind of low reference temperature, 0K or 0C?), allowing for the stock VE table to properly fuel at stock engine at various temperatures. Multiplying a temperature into the value for GMVE didn't seem particularly useful unless there was some coefficient for density multiplication "built in" to the GMVE value itself (with unit K^-1).

We 98s are a little disadvantaged in that we don't get both BIAS and FILTER tables for temperature (only BIAS). It seems as though you have come up with a good model for the way the parameters determine intake charge.


there's multiple ways to calculate Cylinder Air Mass (CAM):
from MAF: CAM1=120*MAF/(RPM*CYL)
from FUEL: CAM2= IFR*IPW*AFR
from GMVE table: CAM3=GMVE*MAP/TEMP

so to tune a GMVE table, you gotta make sure that the two CAM's in question are equal:
CAM2=CAM3
IFR*IPW*AFR=GMVE*MAP/TEMP
so:
TEMP=GMVE*MAP/(IFR*IPW*AFR)

How accurate can we believe the MAF to be? There seems to be two schools of thought here: one that says it needs to be tuned after modifying the engine, and one that says it should be left alone as its stock calibration will be accurate regardless of what is taking place around it.

I am inclined to believe that it would be accurate regardless of other engine modifications as long as the flow was not extremely unsteady, and that tampering with the stock calibration is not the correct way to achieve desired fueling.

If accuracy of the MAF (and MAP) were established, it seems that equating CAM1 and CAM3 above would provide better results than CAM2 and CAM3, if nothing else than because we would have successfully eliminated some sensor error. Maybe I can make friends with some of my professors so they will let me play around in our fluids lab for a little while. ;)

[redhardsupra]

Well, you gotta use the IFR values from the tables if you dont have a FP sensor hooked up. so you just gotta think of it as an estimator and take it for what it is. generally speaking tho, the more info you get, the better results you'll get at the end.

When it comes to undertstanding the GMVE by looking at it's unit, it's a very good idea. this is exactly how I came to understand it.
I think of it as a more generalized efficiency. It is not just volumetric anymore, it also has temp and pressure in there, so it's more of 'volumetric efficiency at given conditions.' it isn't just about the breathing capability based on volume, but how well it reacts to pressure and temp.

As for different model/years, so far I have identified 4 distinct airmass models. they're all different, and create different shape of estimated temperature curves in time, which would explain the different shapes of VE tables, despite similar engines.

As for MAF being accurate, you gotta recalibrate it every time you change something in the intake tract. Since just about every fbody/vette i ever worked on had some sort of CAI/lid/filter/bellows just about all of them need a recalibration. even if it was stock, there is a reason why GM went to a hybrid MAF/SD model for airmass estimation. MAF has resolution issues at low airflows.

more later, gotta jet. good stuff tho, keep 'em comin'!