MAF and SD (was re: Re: New member intro)

[Dale Ulan]

> Speed density systems that I know of include a barometric pressure sensor.  I
> don't really know who the audience is here yet, so if my comments are to 
> generalized or simplified I apologize.

Actually, the GM 1991 pickup truck (6801 processor, TBI, hydraulic tranny,350)
uses an estimate of barometric pressure, calculated when the throttle is
quite open at low engine speeds. I could get the actual numbers used...

Speed-density usually measures the absolute pressure in the intake, so
the inlet side of the engine is already compensated for. The backpressure
from the exhaust side changes, though, so there are correction tables for
the EGR flow rate (MAP vs BARO) and volumtric efficiency (BARO only). The
base VE is from MAP and RPM, as expected. Baro also affects the idle
control actions. This is what I gather from reading the code, anyways.

>|Geez, I hate differential equations :-).
...
Engine controllers generally solve a lot of them at once. Normally in the
form of difference equations (discrete time). I simple GM controller I
mentioned solves at least 15 of them constantly, and another couple when
the engine is knocking.

> SAE 900616 - Mean Value Modelling of Spark Ignition Engines, by Hendricks
>              and Sorenson, 1990.

There are more really good SAE papers. I don't have the numbers with me,
but Hendricks and Sorenson also did a few followup papers, covering
using the Mean Value Engine Model (MVEM) embedded into the control
algorithms rather than the traditional linear SISO control methods with
'tweaks' done in the older GM code that I ripped apart.
Also, open and closed loop observers are discussed. These are basically
an engine model operating in real time, using the sensors to tweak the
model. This allows the model to predict the 'true' value at any given
time, rather than the lagging sensor value.
These are covered in papers ranging from 1991 through 1993. In 1994,
Hendricks (and co-authors) cover theoretical advantages and disadvantages
of time vs crank angle based sampling systems.

Pretty cool stuff. I don't know what the current state of development
of any of the manufacturer's software, but I know the stuff I ripped
through was kind of ugly. It was obviously written by a couple of
assembler gurus, and is the 'classical' semi-static control algorithm
designs. Fortunately, the 'perturbation' tweaks appear to have been
designed not too badly, but this makes things a bit more complicated.
I would think that an MVEM based controller would actually be simpler
in some respects than a 'standard' one. Maybe my opinions are wrong???

-Dale