GMVE calculators

[DSteck]

I was having issues getting Bluecat's software to behave with C7 stuff (mostly issues with importing from 2.25). As a result, I set out to make my own deal... This is the end result. I have configured two Excel files for use on NA as well as a 1.8 PR (pressure ratio) setup. Why 1.8? Because the axis labels ended up nice and clean with that, and realistically, it should extrapolate past that point pretty closely.

How to use it:

There are two sheets.

COEFFICIENTS CALC
In this sheet, you past your VE table in the light blue area. The yellow area spits out your coefficients. Copy these back into your tune.

TABLE CALC
In this sheet, you can paste your coefficients in the light blue area, and it will generate the table with the appropriate axis labels. It's a good idea to copy the coefficients from the other sheet into this sheet and grab the table that's generated to edit.

I've attached to this post a couple of HPT files that run the custom operating system from Gen IV. The purpose of these is for ad hoc editing of the tables. Each table is set up with axis labels to mimic the pressure ratio found in the Excel files. You will put your table into here and apply your errors, then do whatever smoothing you want and copy it back into the Excel file. Every time you generate new coefficients, you should update this with your newly generated table (using the TABLE CALC tab).

This is not perfect, it uses predefined labels that I have picked, and doesn't necessarily try to blend zone edges. It'll get you pretty close though (and looking at GM's stock tables, you can see that it isn't a precision thing).

You will need to define custom histograms in your scanner that match the axis labels used in these files. You may also want to define a custom PID which is Pressure Ratio (should just be [PID.51]/[PID.11]).

Also, due to limitations on file sizes on HPT's forum, you will have to rename the files to .xlsx for them to work. If this doesn't work in Google Docs, well, I don't know what to tell you.

[mowton]

Let me be the first to say thanks and YOU ROCK.....I am not worthy for sure!

Ed M

[Zero4488]

Great job Dave!

[jimbos'ss]

Subd

[schpenxel]

Sub'd

[Wnts2Go10O]

tag to see where this goes.

[DSteck]

tag to see where this goes.

I'm waiting for the "aw man, you gots ta unlock it so I can see the equations, gibs muh dat!". Lol.

[Road]

"Aw man" y'all way more advanced than me I am still trying to figure how to put a extension at the end of a file lol

[schpenxel]

I'm waiting for the "aw man, you gots ta unlock it so I can see the equations, gibs muh dat!". Lol.

I was surprised to see that hasn't been said yet..

And we both know Excel PW's are not very secure.. but they probably are enough to keep the average joe blow that wouldn't understand the equations anyways from mucking with it

[redhardsupra]

I'm waiting for the "aw man, you gots ta unlock it so I can see the equations, gibs muh dat!". Lol.

We already got the maths for it, we'e had it since 2007, nothing to uncover there. but it would be nice to verify.

[DSteck]

Ok, take stock values, generate a table, and throw the table in. It'll put the same values back out.

The guise of vetting it won't work. Lol. You want to know how it's done... =index(linest(y,x,1,1),1) etc. There.

[redhardsupra]

'throw the table in'...which table? 'the' implies like i supposed to know which one you're talking about.

Generating surfaces out of coefficients is just addition and subtraction, that's not the problem.

Making the surfaces 'work together' as you transition from one zone to another is a much bigger problem.
Working with stock coefficients is nice cuz you know they're going to give you somewhat nice surfaces. Surfaces (and thus coefficients) backcalculated from fuel consumption are nowhere near as nice and easy. So then you need to revert to robust statistics using nondeterministic iterative methods, combined with some data cleanup before the fitting. At that moment Excel is out, unless you really want to implement a lot of hardcore stats in VBA.

So you really want to decide early at which moment you want to stop with this project.

The real fun with this sort of stuff is dealing with cams that will simply not allow generation of data in large portions of the rectangular zones. you might have data that just barely covers a corner of a zone, which usually yields an utterly insane surface for the rest of the zone, and who knows how well it's gonna blend with the other surfaces in neighbouring zones. So then you decided to optimize the zone boundaries, to catch as much data as possible and segregate the largly empty zones out. Except then every time you calculate something, you have to recalculate up to 8 neighbour zones, as the borders moved, so now you're not sure if you want to solve it as an optimization problem with one objective value and boatload of the contraints, or simultaneously solve 9 zones at the same time somehow make up a metric that would do a good job assessing of how the zones blend together...

This is NOT an easy problem. Fun, but it will drive you nuts. This is as much of a promise as it is a warning

[DSteck]

I invite you to make your own. You are quick to criticize and complain when people don't crack open their methods. You are exactly who I expected to whine about it being locked... Just like with the Ford to GM data transposition.

[redhardsupra]

I have, ages ago, that's why I'm telling you where the problems are going to show up. I'm not pulling them out of thin air. Oh yea, and it was all open source

[DSteck]

Must have missed that thread where you put out anything to convert from a table into the polynomial coefficients.

[redhardsupra]

https://github.com/marcinpohl/vebiasoptimizer last commit date is right about 7yrs ago

[DSteck]

So a MATLAB based system inaccessible by most for outdated applications. Got it.

[jarrah]

Did u ever try with simulink.....i think it can be mapped in simulink

[Redline MS]

Did u ever try with simulink.....i think it can be mapped in simulink

You have access to Simulink?

[jarrah]

yes I do, I did 6DOF module for UAV's in the paste while doing mechatronics and aerospace engineering.