05/27/03

Control Algorithms

In the pursuit of increased fuel milage, one must tune differently than for
increased power.  The easiest method to "know" when to optimize economy is a
cruise control engaged input.

The only purpose of a computer optimizing milage is to reduce Brake Specific
Fuel Consumption (BSFC) to the lowest number possible.  Acceleration is not
important and neither is power.

The engine at any given speed is required to produce a small subset of its
total power.  You can further guesstimate that it will be in the highest
economy gear for that speed. 

To optimize power, the ECU should be in a "closed loop" system.  Now before
the dweebdum mouth runs, what is a closed loop system.  It is by definition, a
control system that looks at one or more variables and optimizes itself
accordingly.  

In the all-defining automotive world where everything is what dweebdom 101
declares it to be, this has become associated with looking at an O2 sensor.
Unfortunately for the auto mental midget, this is not the case for the whole
rest of the world.  

What the rest of the world does extremely effectively is use control loops to
tune and optimize processes.  This is what needs to be done to minimize BSFC. 

The problem is "what variable do we optimize".  EGO?  Either its fixed because
we are forced to pay oral homage to the EPA, or where do we fix it at?  There
is some relationship between the mixture and BSFC - but it is highly variable
based on way to many things to be a control variable.  Spark advance - again -
too variable engine to engine.  Exhaust Gas Temperature - got a bobble such
that after the temperature drops to minimum - somewhere past that point as
mixture is leaned, maximum economy occurs and somewhat latter misfire.

In fact, most of what we know as conditions for BSFC maximization are "related
conditions" and we know by bringing them into certain values, we are
increasing BSFC - but we do not know how effective they are without a dyno or
other test.

The whole purpose is to optimize BSFC. Maximum BSFC = maximum economy or
milage.  By hand, we could use a milage gauge ala Davesworld for thunder. But
that is not practical for EFI with return paths etc.

Or is it?  We meter the fuel by varying the pulse width and fuel pressure.  By
knowing the rpm, the fuel pressure and pulse width, we could calculate the
fuel consumption to precision ridiculously and irrelevant.  Or we could use
them in a control loop.  

The load is constant per VSS or rpm or speed.  Tuning for minimum pulse width
at a given VSS would allow us to tune for optimum BSFC - regardless of which
variable we are tuning and the results would be correct and not dependent on
third party activity to verify.  We let the engine tune itself in cruise
control to optimum BSFC.

First control would be to reduce fuel pressure to the point that the pulse
width was at least 50% and close to 75%.  This increases the effectiveness of
both the fuel spray (within reason) and by broadening the pulse width, gives
us a better picture of each change.

Worst possible case is where you can't tweak the code.  A simple "add on"
system looking at VSS and pulse width and whatever else you choose and
controlling externally whatever you chose - such as R-Fluid or H2 generation,
could be used to minimize BSFC.

We talk all the time of the usefulness of a broadband EGO.  For those of us
who milage is a necessity, perhaps we could create a Pulse width meter to let
us minimize the average over time and instant pulse width to maximize milage.
And an mpg meter would not take much thinking to create - without any sensors
in the fuel system and being very close to correct.

Since I have a 93 Aerostar that must remain smogified, and I'd love much
better milage, I think I'll start working with the concept of an add on
computer.

But then - I must do an analopticitus check - cause if looking at pulse width
to control milage adders made sense - it would have been done years ago and
BMW would have claimed to invent it.