water injection and then some (long!)

[Chris Conlon]

Hi everyone,

The following is all sliced together from various posts. I'm mainly
replying to Barry, but have quoted many other people too in here. It's
long and long winded... consider yourselves warned. :)


> > Water flow at 50% of the fuel flow rate (mass/mass) is the number I
> > keep hearing, not to say that other numbers can't work.
>
> This is what I am not sure of.  Seems to me the kits I have seen are much
> much lower than 50%.  I am wondering what range should be considered usable.
> I read one study that worked with 5-15% based on mass.  Also the nozzles
> that come with a number of the kits I have seen would be limited to more in
> the 10-15% range, but I dunno.  This is something I need to determine.  I'd
> prefer to not use trial and error to get into the ballpark.  Do you recall
> where you got the 50% number?

Ok, I have to apologize for using lazy science in this one. Basically I
have just seen a bunch of people running 25%-50% water:fuel on various
engines, and similar numbers in papers that people have quoted (but which
I have not read). I fully believe Bruce's value of 50% max, and apologize
for being imprecise and unreferenced. ;)

Yes a bunch of commercial kits seem to run way less than that. Another
imprecise, unreferenced impression is that the quantity of water is not
very critical, and can vary over a wide-ish range. I plan to use trial
and error and EGT to get into the ballpark. Sorry, I know this is kinda
vague.


> I am thinking an el cheapo PIC 16C84 will do the trick since there is not
> much work to be done and a very small amount of code to write.  I will need
> ADC for the MAP but I could use a chip for that if I get lazy.
> 
> Next on the gadget list is to come up with a good algorithm for an
> intelligent self-learning boost controller coupled with an intelligent
> blow-off valve.  Then I could tie the water injection controller into the

At least some of the PICs have 8 bit A/D, IIRC. I was going to use one
before I went 68hc11.

<cranky rant>
As for the boost control: Have a small dedicated air pump and a solenoid
valve to actuate the wastegate. Keep the wastegate FULLY CLOSED at ALL
TIMES except when the turbo nears overspeed. (Yes you need a turbine RPM
measuring setup.) For boost control, get a fly-by-wire throttle, put it
*before* the turbo, and have the throttle follow the driver's foot except
when boost is too high. (Thus putting boost control where it should be
IMHO, but without burdening the driver.)
</cranky rant>


> > If non intercooled, and your looking to help cool the intake air charge,
> > then I'd go with the finer the mist the better.
>
> Okay.  I am intercooled using factory air-air exchangers so they are "okay"
> but will become soaked withing 20-30 mins of hard use like on a road course.
> Maximum peak boost will be between 21-25 psi with 21+ psi sustained.

If I might suggest, stop for a moment and ask why exactly you're using water
injection? It sounds like the answer is "to run higher boost without
detonation". Cooling the intake charge is great, but not at the expense of
*displacing* large amounts of that intake charge with inert steam. Ideally
you want the water to enter the cylinder just under the boiling point,
but still as liquid. This is because the main power increase due to water
injection (assuming constant boost) is not due to cooler charge but due
to the large increase in cylinder pressure as a small volume of water
flashes into a large volume of steam. Also the anti-detonation effect is
largely due to the heat of vaporization. (Not entirely - in humid air the
water is already fully vaporized, but even though there is no heat of
vaporization to overcome, there is still some slight anti-detonant effect.)

Now you may notice that here I took my supposed answer to why *you* want
to run water injection, and sort of ignored it anyway. This is because it
can give power increases even on NA engines, by making use of more of the
heat of combustion, while still having the anti-detonation effect.

Water injection on a boosted motor is IMHO a great thing. It lets you run
more boost (if you want), but even if you don't it still lets you extract
more power from the motor. Essentially it's recovering heat energy that
would have gone into high EGTs, and turning it into hot steam which does
extra work pushing on your pistons. *And* while displacing only a small %
of the intake charge, as long as the water stays as liquid. This is why
Greg is so insistent that the water be injected near the valve, and timed
with the valve opening. Water evaporating outside the cylinder = big lose.
Think of it as accidental EGR.  I can't manage to build that kind of setup
just yet; my approach is ceramic coatings on the intake ports and valves.
Hopefully this will at least help keep more of the fuel and water in the
liquid phase til they pass the intake valve.


> If the droplets are small enough however, won't this effectively be the same
> deal?  Clearly we don't want puddling, and we need "vapor" (small airborn
> droplets) that can take heat from the air charge to allow more fuel to be
> dumped in (but not so much that the water displaces required fuel).

In case I was unclear above, IMHO a slight err on the side of larger
droplets and too many droplets will probably net you more power than too
fine a mist or too little water (which will more easily flash to vapor).
Generally you'll have enough heat of combustion to vaporize everything
and then some. One of ERL's suggestions, which I like, is monitoring
EGTs.


> >A last thought on evaporation.  The evaporation rate of water is very low
> >below its boiling point 100c stp.
>
> Suggest reading a steam table. It all depends on temperature and PRESSURE.

Water and/or methanol will produce significant vapor pressure well below
boiling. Temperature is a measurement of *average* molecular energy, the
peak of a bell curve. This is why a swamp cooler can cool air below
ambient, why wind chill happens, and so forth.

> Simply atomizing water will not cause it to change from a liquid to a
> vapor - it takes 100c STP to accomplish that.

Nope. And atomizing it will greatly speed the process. Finely divided
materials dissolve and react more quickly. Consider how gasoline mist
burns compared to a pool of liquid gasoline.

...

Now if water/methanol formed a more stable azeotrope, in some ratio,
that might give a higher heat of vaporization than either one alone.
I think it's the other way round, though. Does anyone know offhand,
if you mix a bunch of water and methanol, do they get hotter or
colder? If they get colder, the heat of vap should be higher.
Otherwise the addition of methanol is mainly as antifreeze.



> Yippee.  I understand more or less.  A side benefit of doing this is that it
> reduces NOx emmissions.  Neat huh.

Ok, now you're in for it. You came close enough to one of my pet ideas
to trigger off another rant. Faint hearts, bail out now. ;)

The main reason water inj reduces detonation is that the high heat of
vap sucks up "excess" heat energy, and thus reduces the formation of
various oxygen radical species. These radicals are essential for flame
propagation, but if the mixture has a high enough energy density, it
can cross over into detonation. Lowered combustion temps reduce the
formation of O radicals. O radicals and NOx are intimately related,
since the same conditions, lean mixtures and hot mixtures, favor the
formation of both. In fact NOx are formed mainly through O radicals
bumping into N2. When the mixture is rich, O radicals (and NOx) are
more likely to "bump into" some fuel molecules, and keep the fire
going. When there's lots of spare O2, heat and N2, lots of "leftover"
O radicals end up bumping into N2, which then hangs around. Oppositely,
rich mixtures tend to "sop up" O radicals more aggressively, which
keeps the mixture burn rate from accelerating enough to detonate.

Water vapor ("humidity") also slightly reduces detonation. IMO this is
in part because water is a modest spin trap, and can convert some O
radicals into slightly less reactive species. This has the effect of
slowing combustion slightly, or rather, preventing it from making the
1000x speedup it would need to make to turn into detonation. It's no
accident that many of your higher octane fuels (most things with an
aryl group for instance) are decent spin traps as well.

...

One of my pet ideas for when I hit the lottery is a *high* compression
NA engine, with full ceramic coatings and almost full-time water
injection, that runs slightly lean mixtures, on pump gas. And produces
very low emissions even without a cat.

The thinking went like:
High compression - to get a higher efficency
Ceramic coatings - ditto, plus keeping the fuel and water from boiling
  off in the intake ports.
Water injection - above maybe 25% throttle, to prevent detonation and
  reduce NOx, which would otherwise be sky high.
Lean mixture - to reduce CO and HC, which would be sent sky high by
  the water injection.

High compression (+ coatings) would also help get the lean, wet mix
hot enough to ignite reliably.

Can all this be made to work? I dunno... comments welcomed.


> If you were "whacky", you might consider adding steam after the throttle at
> *** PART THROTTLE ***.   This would greatly reduce the pumping work and
> recover some waste heat.  At less than wot, this would not cause a loss of

Hmm I'll have to think about this. :)  My idea has been focused on WOT
power, not part throttle BSFC, but I'm not against upping BSFC if it
doesn't hurt max power.


Ok, it's way too late, as should be evident from the foregoing. Thank
you all for your kind indulgence.

   Chris C.