Sharing: Alpha-N tune

[f8.]

Hello all,

I would like to share my current tune status, just to depict something familiar in a slightly different way, also to highlight (if that needed any further stressing!) the power of datalogging.

Engine is a 1.5L inline 4, naturally aspirated, stock cams. ECU is in TPS based mode/alpha-n.

Results below are from datalogs with vertical axis being air temp and horizontal being Lambda. Red line is average target lambda in that RPM and load range.

[SportsCarRacer]

Nice MATLAB plots!

This is how model-based cals are developed in the OEM world to highlight interdependent variables, significant variables, and to verify the model function/surface is a true representation in all domains.

There are some awesome regression tools in the MATLAB Engine Calibration Toobox that can fit the engine aircharge to a surface, with a DoE mapping procedure to minimise the number of mapping points required (ie: not "full-factorial" mapping at every speed load point!!). For example, on a modern 2.3L IntakeVVt NA engine that would take more than 10,000 discrete points to optimise it at all speed/load/spark/cam points, using the regression tool & DoE it can be simplied to be VERY effective with less than 400 mapping points....and without the need for strict emission compliance (ie: race car), under 100-150 would be all that's required! Done in a few days of automated dyno rather than 4-6months of mapping......
What is the purpose of your project...is it for a research project....did you manually manipulate the data into Matlab or did you write a script to take the motec log file (or export from I2 into csv, then manipulate into MATLAB i expect would be most efficent).

BTW, it shows how "ineffective" that Alpha-N is at modelling the true aircharge of the engine...especially without lambda feedback as a minimum! As such, alpha-N is a largley forgotten way of modelling aircharge in modern engines (except in the aftermarket!)....in some high-end race systems (Bxxxh!), alpha-N is used as a feed-forward for base aircharge, to calcualte fuel mass required, then useing wideband lambda feedback, precisely calibrated for transport delay at each speed and mass flow point (and turned off in the "blow-through" periods at high-load and predominantly low speed points caused by large charge bypass with big overlaps).


Look forward to more of your shared data....if you ran speed-density (MAP) you would clearly see the linear relatinship odf iof MAP to actual engine aircharge( air ingested)...it's extremely linear from low loads up to almost full thrrottle/full rpm ...but some slight "turn-downs at each each of the line due to non-linear flow losses.

T.

[Aurélien]

Nice to know that i'm not the only one to use that mbctoolbox !

Did you build your own models for calculating lookup table values based on measured data ?
Or maybe you're just using mbc to help visualise how input values affect output values ?

If you've more to share, i'm very interested in as i'm spending lot of time to adapt that tool for performance tuning and aftermarket ecu tuning !