NefMoto

Hello gents and ladies.

Lately something with regards to the torque->load and load->torque conversion has been bugging me, specifically, how for a given cylinder filling (load), torque will increase. I've noticed this behavior in two completely different ECUs, both in the 2.7t ME7.1 and my 3.0 ford running something called GreenOak

For example, the stock M-box 2.7 KFMIOP, with units being load %, RPM, and relative torque %. This is of course a turbocharged engine, so many of you will be familiar with the limitations of it's x-axis :D :
(https://files.catbox.moe/edrfc0.png)
And then the stock ford 3.0 file (AU7A-14C204-XJ), with units being load in %, RPM, and absolute torque in newton meters. This is a naturally aspirated engine, so it will theoretically only see maybe 0.95 load:
(https://files.catbox.moe/92ky87.png)

As you can see, for a given load, torque increases with RPM. Does anybody know why this behavior exists? My theory is that given the same amount of air and fuel (and presumably same amount of explosion), there are less thermal losses to the piston/block/cylinder head at higher RPM since there is less time to do so, but I have nothing to back this up.

If my theory stands true, then of course these tables would need to be modified to reflect the bore, stroke, materials, running temperature, and fuel composition to be completely accurate, although in practice the only thing that's crucial to the running of the engine is that your load+RPM->torque and torque+RPM->load tables are inverses of each other.

Would love to hear some thoughts as to why our OEM calibrators made these tables like this.

This is inner tq without friction losses!

Friction torque is measured first, it is filled into maps by spinning the engine with an electric motor at different rpm's and temperatures.
Then the torque output of the engine is measured at the crankshaft for a given load with ideal ignition angle.
Then the friction torque is subtracted from this output and what remains goes into this table.

What the physical reason is, no idea, but considering the miniscule difference I wouldn't put too much emphasis on it.
Consider also that the lower end of the table is just filled with fairly random values, there's no way to get 190 load into the engine at 440 rpm.

I feel this is because there is a huge disconnect between the overeducated people who write the engine model routines and the people tasked with doing the calibrations :)

That's a great point I hadn't thought about! This would make sense too, since on the ford ECU, driver requested torque (the equivalent of KFPED) will go negative at lower throttle input+higher RPM, which would only be possible if the ECU can factor in friction; in my case I have yet to figure out where in software exactly this torque loss is described.

On my stage 3 car I've scaled the KFMIOP X-axis to support up to 240 load while leaving the data mostly alone, and scaled the KFMIRL data so that it matches the now adjusted KFMIOP, and the end result feels very smooth, much better than when I simply had the last 3 columns of KFMIRL scaled up. Now, being that ME7 uses relative torque which is now scaled up, any other reference to torque such as the engine friction losses and accessory losses are no longer accurate. As far as I can tell, mds_w would describe torque loss internal to the engine and mdverb would describe torque losses from accessories (alternator and a/c load?).

Has anybody ever really messed with these after scaling KFMIOP/KFMIRL? I can certainly see how engine loss torque might not be so noticeable, while accessory load such as the compressor clutch engaging can have a dramatic effect such as at light throttle cruising.