09/02/2003

Homogenous Combustion

Using Rogowski "Elements of Internal-Combustion Engines" ( a complete scan of
which is available in the fangle base ), let us lay out the basis for the use
of ignition accelerant or Cetane Improvers.

Chapter 4 - The Fuel Air Cycle Approximation.  

Notice first that it is called an approximation - a SWAG that will be close to
reality but not quite there.  Not a "Computer Model" or the result of a
desktop dyno.  Approximation means close - close enough to have been used for
designing and building engines since Hottel in 1936 laid out charts on the
data.  The Computer stuff means expensive - we need more funds to cover our
star wars galaxies additiction.

"Assumptions.  In the constant-volume fuel-air cycle the following assumptions
are made for the sake of simplicity:

1. The fuel is completely vaporized and perfectly mixed with the air.
2. Burning takes place instantaneously at top center (constant volume).
3. There is no heat exchange between the gases and the cylinder walls.
4. Compression and expansion are frictionless as well as adiabatic."

The normal methodology was to lay out this cycle based on a whole bunch of
data, and then compare it to the real engine to get a fudge factor - typically
around 85 percent or so.  Once compensated - actual performance was within a
couple of percent of theory.  Try that desktop dyno!!

The closer the engine comes to meeting the assumptions and quantities, the
more reliable and accurate the approximation becomes.

Homogeneous pre-mixed flames such as from propane, both will run leaner and
make more efficiency and the engine will be closer to the approximation -
Assumption 1 rules here.  Nice little nasty about propane - its high octane
about 105-110 depending on the test and who's doing it - but its also a mild
ignition accelerant in that the flame speed goes up faster than the engine
speed so that by 2500 rpm or so, the lead needed to fire it is shorter by at
least 10 degrees.  Plus, propane will burn out to 75% excess air or more.
Tossing a modest say 5% into the mixture will make the fuel burn much faster
under boost WITHOUT increasing detonation, and will ignite lean areas of the
charge much quicker than gasoline only.  Even a tad should be good for 10% or
more effective power.

A quick side note about Assumption 3 and 4.  From the Tacom Studies of diesel
transportation engines in the early 80's, a number of adiabatic engines were
built and tested.  Power increased modestly - between 5-10% - but
understandably from the changes.  The head valves ports and pistons were
ceramically coated - essentially eliminating heat transfer.  The cylinders
were uncooled, and since some cases, the rings were eliminated and the
cylinders were gas sealed.  As they come closer to an adiabatic engine, the
fudge factor got more accurate.  These engines became closer to the
approximations the closer the assumptions were meet.

Which now takes us to Assumption two.  Constant volume combustion at TDC.

Adjusting the timing, we can bring the pressure peak ( non detonating ) to the
close vicinity of TDC and get maximum bmep and minimum bsfc.  This is normal,
effective, and been done since Moby Dick was a minnow.  But, since the entire
pressure curve is being adjusted, it is still subject to time issues with
light load and low pressure.

With the use of an ignition accelerant such as propane or Cetane Improvers,
the time for the combustion event to occurs is significantly shortened.  With
the enflamement and burning times significantly shortened, more pressure
closer to TDC will be generated and thus more of the fuel will be burned at
conditions closer to the approximation and therefore efficiency will be
increased.

In an engine where the timing is fixed, seriously shortening the burn time
works well.  Ideally, the timing would be retarded to place the pressure peak
in the close vicinity of TDC.

Life is good - peak pressure locked at tdc and short combustion beginning to
approach constant-volume.  

Kerchunk - what be dat noise boss?  Its the sound of knock.  Oh shucks oh dear
- what do we do now?

At some low compression pressure - around 7.3 to one for Iso-octane, we reach
the critical compression ratio for the fuel and the mixture.  We get knock in
the end gases.  

The usual suspect method of dealing with it is to reduce compression pressure
in the end gases - by retarding the timing.  With a timing retard, the
duration of the combustion event does not change - it simply starts later.
The peak pressure then occurs later, the end gases do not knock - life is
better and Mean Gringo Bob spends time at the Rio sucking O Dohls and lying to
"little" Anna Belle about his prefences in her nail art hiding the fact that
he really wants them ripping chucks of flesh out of his back as they climb the
mountain.

Why does retarding work?  The piston in a real engine is not stationary at TDC
for the time needed for constant volume combustion.  It begins to move
downward and expanding the volume.  The volume of the chamber doubles in
expanding the first clearance volume, thus potentially halving the pressure.
It only takes a small expansion to stop detontation of the end gas - a couple
of degrees will add significant volume.

The normal method of tuning for light knock is an absolute conformation of
this.  Peak pressure is retarded off TDC just enough to reduce the knock to a
light level and peak power is measured.

There is reference to peak pressure being around 10 to 20 degrees ATDC in a
sweet spot mentioned in mucho literature - but no generally explanation as to
why.  

This point is the reduced compression point that just allows the end gas to
detonate lightly.  Since this is based on the fuel, the mixture, the
temperature and the time, its not a constant point - but a range where because
of geometry, if we place the peak pressure, max power will be made.

GM's approach to this issue is the testing advance.  GM will start with an
advance and increase it to a limit and then reset to a stored value and try
again.  If it detects a knock, it backs off, records and tries again.  Over at
period of time, the computer knows approximately where the engine will knock,
and adjusts timing accordingly.

I have a friend who has a 91 Corvette that he runs on 87 octane.  Its not his
"hot rod" its his puss wagon.  He trolls for chiguitas in it.  The GM computer
in it has "tuned" itself to his fuel and the engine is spiffy fine.  He's
happy when the right seat gets filled, the cars happy and the 20 to 25 cents a
gallon he saves is worth it to him.

For those with the knock bump type control algorithm - on GM I know for sure,
adding cetane improver will not affect knock under load because of self
adjustment.  

I am not yet sure that a retarded cetane enhanced combustion under knocking
load would make more milage than an unretarded regular fuel.  Need
investigation - let me borrow your GM dually for the summer and I'll give you
an answer.

Hopefully by next month - pickling vinegar injection....   Giggle.

----------------------------

I do.  I am looking at it.  Initial flame development 1940's bomb studies.
Flame in a chamber studies - 1940's Naca with open flat l-head flat slice
chamber - Naca Papers has the original data particularly with the 400,000
frames per second camera stuff. The engine for the flame choice is the NACA
high speed camera engine developed for the SOLE purpose of recording
detonation waves.

Jaundice must be applied to his flame speed data from 400 rpm to 1200 rpm.
Significant later studies.  You understand your idling chevy motor with a cam
is running faster than this data peak rpm.

Then every one gets wild about spark plug placement.  The data is for a side
valve engine with no quench and a slice head.  Since every head since the late
20's is a quenched ricardo style - either by royalty or by similarity, this
data was totally meaningless even at the time the book was published.

Then, he determines on page thirty 

"Patterson (1.833)( 1966 still pre transistor much less computer ignition )
found that with a well mixed charge the following variables have  no
measurable influence on cyclic variation: Spark Energy, Spark Voltage Rise
time, spark plug gap size and electrode configuration, fraction of residual
gas.  On the other hand, all investigators have found significant effects due
to the fuel air ratio and to the character and degrees of turbulence in the
cylinder before ignition."

Stop right there.  Variations in the mixture and turbulence within the
chamber.  Excluding Independent Runner configurations, a common manifold with
fluid flow routinely exceeds 5 to 10 percent variation on mixture.  With
leaded fuel drizzle properly taken care of, they still are not much better.
Fuel air varies radically in the manifolds on older Taylor Era leaded fuel
carbed engines.

Next, since EPA, the malefactors have been building highly turbulent chambers
with repeatability to meet the lowest residual gas traces.  This has meant
that the F/A must be nearly perfect from idle to max forever.  And there must
be sufficient turbulence to actually make the charger burn repeatedly and
reliably and WITHOUT cyclic variations.  This is done down to the Yugo level
and would be beyond the belief of 1950 engineers.

Again on page 30

"Consider a turbulent mixture made up of vortices, as illustrated in Fig 1-9.
If the spark occurs at the center of a vortex a, the flames must spread out
without the aid of turbulence until it reaches the vortex boundary.  On the
other hand, ignition at the vortex boundary b will immediately aid the spread
of the flame because of the shearing action there incountered.  Now suppose
the whole charge moves past the ignition point, as in a swirling motion of the
cylinder contents.  If the spark has a finite duration, as is usual, it will
follow the dotted path a-c with relationship to the vortices and will soon
encounter vortex boundaries.

This theory seems to explain the following observed facts:

1. Multiple ignition points reduce cyclic variation.
2. Cyclic variation is also reduced by increased engine speed or by decreased
inlet valve area.  These variable lead to smaller and more intense vortices
and to more rapid motion of the gas.
3. Cyclic variation is markedly reduced by a tangentially oriented swirl, such
as that from a shrouded inlet valve.  The same valve arranged so that the
gasses enter radially instead of tangentially gives much greater cyclic
variation.  The explanation is probably to be found in the results a-c type of
relative motion indicated in Fig 1-19."

Bingo - what the world have massively worked on since the time of this book.
Swirl, turbulence and fuel air mixture distribution.  

Other works will demonstrate the variableness of the spark ignition cycle.
Heck even Heywood may have it right.

Oh BTW, the breaker point ignition system at the time he was doing his
research was the ne plus ultra system.  Transistors were invented half way
through his career and fuddle duddying around at the time this book was
written and way too small and fragile and weak to even consider for an engine.


On Wed, 3 Sep 2003 18:15:31 -0400, you wrote:

> >I guess I'm disagreeing with everyone today.
> >
> >Bob, I can't give exact quotes since I don't own a copy of Taylor (I've
> >only been able to find one that I can borrow), but most of the actual
> >engine studies we're talking about were done on a single cylinder engine
> >specifically built to test this with a CR of 7.5-8.5:1 and setup so that
> >he could change the intake pressure to test it's effects on all this,
> >taking most of the variables that you mention out of the picture.  The
> >one remaining variable is what was used for the ignition (I don't know
> >and don't remember it being mentioned) and I don't believe that it would
> >be incredibly short sighted to us a 'sloppy breaker point' ignition with
> >tons of spark scatter since he was basing all of his observations off of
> >the ignition event.
> >
> >For that matter, most of the tests that were published by him on this
> >were not engine studies but "bomb" studies where the chamber was a
> >static size and everything was based off the ignition event, so none of
> >this makes any difference, there was no "timing," spark scatter or any
> >other issues to mess up the results.  The chamber was just pressurized
> >to whatever pressure he wanted with whatever mixture he wanted and the
> >spark was set off when everything was "perfect."
> >
> >