Acceleration enrichment: wall wetting

It's all about the code!
Post Reply
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Acceleration enrichment: wall wetting

Post by AndreyB »

I was just surprised to learn that @ uses current rusEfi wall wetting and it's doing him some good?!

I think we all have to ask @ to share some data and wisdom.

https://rusefi.com/forum/viewtopic.php?f=5&t=979 is the older thread where some of this was started with some data already lost :(
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
mck1117
running engine in first post
running engine in first post
Posts: 1493
Joined: Mon Jan 30, 2017 2:05 am
Location: Seattle-ish

Re: Acceleration enrichment: wall wetting

Post by mck1117 »

First my settings in the tune:

The only things of interest are the 0.14 and 0.3 in the middle. All other AE is disabled.

Image

Confirmed via the log that the only thing contributing to AE is wall wetting.

And the log:

Image

This is about 15 seconds worth of log, during a 2-3-4 shift at low-ish RPM, but near wide open throttle. Worst case scenario for AE!

The yellow trace on the top graph, AfrRatio, is a ratio of target vs. actual. If the engine is lean, it goes below 1, and rich above 1. Graph is scaled to +-20%. I'm using this because my actual AFR target varies from 14.7 all the way down to around 11.5. This swing causes the AFR signal to look like the ECU is doing a poor job, when in fact the setpoint is moving by nearly 30%.

There's a bit of a spike rich-then-lean during the shifts, but it's not bad (results in a nice burble out of the exhaust :lol:), and resolves once back on the throttle. Some of this is caused by the phase shift between the AFR target and measured AFR, but some of it is real.
Last edited by mck1117 on Sun Jan 06, 2019 12:41 pm, edited 1 time in total.
mck1117
running engine in first post
running engine in first post
Posts: 1493
Joined: Mon Jan 30, 2017 2:05 am
Location: Seattle-ish

Re: Acceleration enrichment: wall wetting

Post by mck1117 »

Tune attached
Attachments

[The extension msq has been deactivated and can no longer be displayed.]

mck1117
running engine in first post
running engine in first post
Posts: 1493
Joined: Mon Jan 30, 2017 2:05 am
Location: Seattle-ish

Re: Acceleration enrichment: wall wetting

Post by mck1117 »

Log attached
Attachments
2019-01-01_19_modified.msl
(3.08 MiB) Downloaded 627 times
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Acceleration enrichment: wall wetting

Post by AndreyB »

Thank you! This is now part of first rusefi overview pages! https://github.com/rusefi/rusefi_documentation/tree/master/overview
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: Acceleration enrichment: wall wetting

Post by AndreyB »

http://nefariousmotorsports.com/forum/index.php?topic=15018.0title= is making some cool documentation available so we can compare with what we came up with:

FB ESUK 9.70 Description of functions
The transition compensation determines the required increase or decrease in injection quantity for a load change to allow for compensation of wall film effects.
The transition compensation consists of 2 parts:
- long-term share rkukl_w
serves to compensate for slow wall film effects. Derived from the predetermined load signal rlp_w and the rpm
Calculation is carried out synchronously with the segment. (segmentsynchron- no idea what segment refers to)
- Short-term portion rkukk_w
serves to compensate for quick wall film effects. Derived from the predicted load signal rlp_w and the rpm.
Calculation is carried out synchronously with the segment.
The wall film characteristic value WFRL contains the amount of fuel that is stored in the wall film at the respective load. If the load changes during a segment by the value drlp_w = rlp_w (s) - rlp_w (s-1), the wall film difference dwf_w = WFRL (rlp_w (s)) - WFRL (rlp_w (s-1)) must be injected as an additional amount of fuel in order to build up the wall film.
This additional amount must be distributed over time in such a way that lambda remains constant during the transition. For this purpose, the excess amount, using the distribution factor aukk is split into a short-term and a long-term portion and added to the corresponding short-term (sdwfk_w) and long-term (sdwfl_w) storage (assuming storage ECU memory parameter and not to actual fuel on walls)
In each segment the fuel amount sdwfl_w * ZBALM (when accelerating) or sdwfl_w * ZVALM (when decelerating) are decucted from long-term storage and is added to the long-term component rkukl_w of the total value rkukkl_w of the UK. In the following iteration the long-term memory will be reduced by the same amount rkukl_w, provided rkukl_w has been used with at least one ti-initialization or ti was switched off (B_redkl = 1).
The derating factor ZBALM or ZVALM is dependent on the engine temperature, since the evaporation of the wall film is less significant when the engine is cold.
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
mk e
Posts: 486
Joined: Tue Dec 06, 2016 7:32 pm

Re: Acceleration enrichment: wall wetting

Post by mk e »

That looks a lot like what is described in the motec M1 manual and what I kind of semicopied into my ecu. If you down load the motec software there is a good writeup of how their system works in the help section. Attached is a writeup of what I did, its not tested on a running engine but seemed to do what I expected in simulations. All my fuel calculations are by required mass, so those are the units carried into the wall wetting, lots of other approaches that would get to a similar place. Motec uses a fast and slow time constant, I used a table to handle decay so its more continuous as the main difference in what I wrote vs theirs iirc, its been a few years now. I can share the "code" if you like, but its not actually code its a "model" written in the higher level enginelab language, but pretty easy to follow....and I'm sure written in the hardest, most awkward way possible because I'm just not very clever with things like this most of the time.
Attachments
Wall Wetting.doc
(112 KiB) Downloaded 176 times
mk e
Posts: 486
Joined: Tue Dec 06, 2016 7:32 pm

Re: Acceleration enrichment: wall wetting

Post by mk e »

I found these screen shots from the motec system ....I think from the manual, don't recall. They held constant rpm and did a step MAP change before and after tuning the wall wetting, again iirc. You see the lambda move about 20-25% before and 5-10% after so it works pretty well.
Before
M1 film beforeCapture.JPG
M1 film beforeCapture.JPG (51.03 KiB) Viewed 4620 times
After
M1 film AfterCapture.JPG
M1 film AfterCapture.JPG (64.48 KiB) Viewed 4620 times
Post Reply