Remap ....Predicting MAF readings.

[Relic]

I've been pretty quiet on here lately as I have been studying the derv ecu and what makes it tick.
I'd like to remap it myself at some point when I know what I am doing.
But before that day comes, I need to predict what happens with the engine as the weather changes.

Ive collected many datalogs over the past couple fo years.
So I know what the MAF readings are when the air pressure and temperature changes.
Thats the easy part....the hard part is trying to predict what will happen.

Garrett have their formula for this....
Terminology | Turbobygarrett
..under corrected airflow.

Their is also the DIN and SAE correction factors for power and torque used on dynos.

None of which really seemed to fit.
So I went my own way.
1. download the honda map.
2. identify necessary tables like, turbo, lambda and driver wish.
3. digitise the data from the garrett compressor map I got for our turbo.

Chuck all that in a pot....stir well and voila!
I finally managed to do it.
All I do now is plug in the air temp and air pressure.
The tables spit out what the MAF reading will be at all rpm.
I compare to my datalogs at those temps and pressure.

Garrett uses 13.92psi and 85f as the datum for their compressor maps.
So I plotted the Honda data onto this map at that temp and pressure.
I also show how much power that would produce using the boost and lambda maps vs the driver wish map at full throttle.
FMIC pdf.
[NB its into surge at low rpm. However the garrett pressure and temperature mean the density of the air is at the extreme and will rarely be as close to the surge line if at all]

I have compared the predictor against many FMIC datalogs and its pretty much back on.
BUT I thought I would test an extreme of -3c and 1050mbar climate in the last winter.
This was actually with the Honda intercooler in place rather than the FMIC.
So I thought it would be interesting to see what would happen if I compared them under the same conditions.
SMIC pdf.

[coste1l]

mick..... well done! im just sorry that my understanding isnt good enough to offer constructive debate if you say its right i believe you, it goes without saying

[Relic]

Not easy to digest I guess :/

The reason I did all this is because if you use the garrett formula for any datalog you should end up with identical MAF readings after correction.

So I have taken the 3 datalogs used at varying weather conditions...then simply applied garretts correction.
Me thinks there is a wee problem

Lots of turbo with a short shelf life I think ....why am I laughing

[Ben GT]

Good work (I think)

May as well be in Arabic for me though!

[Relic]

Thanks ....I think

Air Temperature (Air Temp) - 60°F
Barometric Pressure (Baro) - 14.7 psi
Engine air consumption (Actual Flow) = 50 lb/min
Corrected Flow= Actual Flow SQR([Air Temp+460]/545)/ Baro/13.95
Corrected Flow= 50*SQR([60+460]/545)/ (14.7/13.95) = 46.3 lb/min

Corrected Flow= Actual Flow ([Air Temp+460]/545)/ Baro/13.95

I modified the garrett formula so it just uses air density ratio instead...
Remember the -3 temp was with SMIC and the other two are with FMIC.

[Syx]

Well I must say I'm very impressed with the ability to outdo Garrett here Mick Looking at the graphs I know which formula I'd rather be using...

Before you posted your corrected formula I was somewhat puzzled by how the SMIC and FMIC were giving similar MAF readings after correction, when we know that isn't the case. What you've got in the latest PDF is what I would expect to see!

[Relic]

I'll pull apart about 10 logs for each at different temps if I can find them.
Just to be sure.

Using the density ratio isnt perfect as it doesnt take into account the higher manifold temps.
The Turbo/manifold MAP is fixed...but the air pressure changes the actual boost.
You end up closer to the Garrett figures using 14.92psi and 85f ...more by luck than judgement I think.

Anyway....hopefully see more clearly tommorrow.

http://www.not2fast.com/turbo/glossa...orDutyCycle=85
This guy appears to be closest to the truth omce you get your head around what he has done.

[coste1l]

So it took you a while to get your head around it.... And for the rest of us?

Sent from my GT-N7000 using Tapatalk 2

[Waxdee]

its 6.10am. and after a 12hour shift at work youv just made my head explode. just saying

[Yassian]

Nice work Relic! Keep it up!

[Relic]

OK ...
I gathered 8 raw datalogs for the FMIC with varying air temp and pressure.
Each datalog consists of 6 runs one after another.
These have then been sorted by RPM.
This way you can see the MAF reading noise and variation between runs for each datalog.

I then applied 3 alternative corrections.
1. Atmospheric pressure only.
2. Garrett correction.
3. Atmospheric density.

If the correction generally works, then all of the MAF readings should line up after correction.

These are each shown in the attached...

[Syx]

Can you do the same with your correction rather than Garretts?

[Relic]

Originally Posted by Waxdee

its 6.10am. and after a 12hour shift at work youv just made my head explode. just saying

Few Beers will cure that

[Relic]

Originally Posted by Syx

Can you do the same with your correction rather than Garretts?

I have something different planned for that.
I will generate the model and do it for each of the 8 datalogs.
Didnt create the spreadsheet to allow multiple temps/pressure so I have to do them 1 by 1.

Will take a while for this one .....be interesting to see how it turns out (especially the 1050mbar and the 21c logs) and where the model may need tweaking.

[Relic]

Originally Posted by coste1l

So it took you a while to get your head around it.... And for the rest of us?

Sent from my GT-N7000 using Tapatalk 2

You know how I like to mentally stimulate you all

[Relic]

Thinking about it ...I better check requested boost vs actual boost beforehand.
No Guarantee they are the same and the model uses requested boost.

[coste1l]

Yeah i was thinking that myself.....

Sent from my GT-N7000 using Tapatalk 2

[Relic]

Originally Posted by coste1l

Yeah i was thinking that myself.....

Sent from my GT-N7000 using Tapatalk 2

Well I couldnt be bothered to do that too at the moment.
So....

1. I have used the requested boost demanded by the ECU map (bad idea).
2. I have use 85% efficiency for the FMIC throughout.
3. I have used the compressor efficiency from the islands on the compressor map.
4. I have used a rough VE vs RPM that approximates the FMIC from the datalogs.
5. 0.5psi loss peak induction & 3psi loss peak FMIC.

Everything else is automagic.

[Relic]

Well guess I should have checked the actual boost.

Using this model with a 3psi loss or 2 psi loss the turbo would be in a surge condition at that boost request.

See attached 3psi and 2psi pdf for how the boost request looks on the compressor map.

So I changed the boost request to be OK at 3psi loss on the compressor map.....at the ragged edge in fact.

I then plotted the ACTUAL boost from all logs along with requested boost and the FIXED request boost.....



..which ties in pretty close to the driver wish.
Basically there is just enough boost, to provide just enough air (at the required AFR) for the Driver wish upto 1750rpm.
So either surge is killing the boost or boost is being cut by some other limiter.
Probably air temp related by the look of it.

[coste1l]

The limiter isnt subtle is it... You have surge then overkill as it reigns it back in. You need to find out what triggers the cut as you surge and then map to 0.01 below that point. Im very tired so apologies if i thought i understood something

Sent from my GT-N7000 using Tapatalk 2