Tuning for water injection: fuel, ignition, and EGT
 

I'm starting this thread in the hopes of developing a Tuning Guide that could be a Sticky in this Forum.

Folklore has it that for now EGT is the best method of monitering the tuning with water injection. For us novices, some understanding of a tuning "algorithm" would be of benefit. I'm going to start with a straw-man approach, posing a method to use as framework for further development.

1) Tune the car for power without WI, no knock.
2) WI turns on at 3-5 psi boost.
3) Add water injection in the 10-25% range (ramping with boost is good).
4) Remove fuel until (an unobserved, but inferred) 12.5 AFR is reached, watching EGT.
5) EGTs should not rise above pre-WI temperature.
6) Advance ignition until knock is detected, watching EGT again.
7) If 12.5 AFR cannot be reached without knock or high EGT, stop reducing fuel.
8) Advance ignition or add more fuel to reach optimal power even at sub 12.5 AFR.

The parenthetical comment 'unobserved, but inferred' comes from the idea that with water injection on the oxygen sensor cannot be trusted. One may however subtract fuel until the ratio has reach 12.5 by simple subtraction of a percentage (e.g AFR 10:1 --> 12.5:1, subtract 25% of fuel at 10:1).

Now one thing I've notice in the various "calculators" I've browsed is that more power can be had with lower AFR than 12.5. While this is the fuel ratio of optimal power per gram of fuel, it may not in practice be the AFR of best power. So, I question the assertion that 12.5 AFR should necessarily be the starting goal for best power.

Also ignition advance is a moving target. The peak torque achieved during a combustion cycle varies across the RPM and load range (by only a few degrees probably). At some point more advance will become detrimental to power and especially at high RPM when valve timing will necessitate decreased advance.

None the less, in my steps 1-8 above, I have taken as a given that 12.5 AFR is the number to shoot for, and that ignition advance is used to achieve optimal torque once the "optimal" AFR has been reached.

This is a going to be a good thread. Tuning with water inejction is the least talked about topic. I will sticky it as soon as it gathers pace.

The steps listed is a very logical approach to the subject.

That sounds like a pretty good start to me! :D Of course, every car is going to be a little different.

For my DSM, I tune for knock and not EGTs so much. My car can have knock that can be tuned out without ever having significant changes in the EGTs. Of course 'if' the WI is turned off, the EGTs go through the roof!

I think another good thing to mention is trying different mixtures of Methanol and distilled water. I have found that I am able to lean out my fuel without knocking a bit more when I run less than 50% methanol. Maybe like 40/60 meth/water. I think this is because my knock is mostly a result of heat soak of the intercooler and so I need more water to do more cooling.

That's all I have, for now.

-Craig

My setup -> 2D kit, 0.5mm + 0.6mm dual nozzle setup, up to about 350 cc/min water injected (currently distilled water only, no alcohol) at redline 7500rpm. Water is about 25%-30% of fuel, injected pre blower (non intercooled Eaton M45 spun to 16K rpm). I am running 10 psi boost, water injections starts at 3 psi.

On the dyno for the final tune of my MX-5 we were tuning for 12.8:1 AFR (it did the max power there) indicated on my TechEdge WBO2 wideband with the water injection on just like with the water injection off.

I didn't notice the injected water (water 28%-30% of fuel) changing the wideband AFR readings

These talks of 12.5 AFR being indicated as 10:1 on the wideband O2 seem being bogus!

The EGT sensor is in one of the exhaust runners - as close to the head as possible (about 1 inch from the head).

My EGT were around 1200F (~ 650C) at 10 psi, 7500 engine rpm.

The intake air temperature sensor is in the intake manifold just in front of the runner of one on the cylinders (it measures the temperature of the air just before mixing with the fuel).

The intake air temperature at cruising on the dyno was 130F (~ 55C).

With water injection turned off at 10 psi of boost the air temperature was raising under boost to over 260F (over 130C).

With the water injection on the intake temperature raised only to 160F (~ 71C)

The ignition timing at 10 psi boost/7500 rpm was about 25-26 degrees advance

My power went up from 100 hp at the wheels (the stock power of a '94 Miata) to 190 hp at the wheels with the current setup (stock engine, never opened, 104K miles on it).

I am going racing this weekend :) Time to hug some curves

Next step is probably putting bigger water jets and increasing the water flow (probably above 30% water of fuel)

Re: Tuning for water injection: fuel, ignition, and EGT
 

couple of thoughts here :smile:
It's an important tool when you're tuning on the edge - but I wouldn't go as far as stating 'the best method'

Where did this come from? :shock:
That would only be true for non-intercooled engines mate, or else it's FAR too low
on well intercooled engines, over 10% could bog it down. Where did the 25% come from?
I've seen max power at around 14:1 with W.I., so we'd need to verify with others that the 12.5 figure stands in this case

Some observations
 

On turbocharged cars you want heat in the exhaust to get good spool up. In fact, some turners intentionally run lean AFR's at low boost to enhance the spool up of laggy large turbos. I would suggest the turn on point be as late as practical for that reason on turbocharged cars.

I would suggest you set the turn on point just a bit below the boost that under high load (with a given fuel) just begins to show mid-range rpm knock. Engines will typically knock first at rpm's near the torque peak rpm.

What I would, do is find a good steep hill, fill the gas tank and throw some crap in the back seat and make a series of pulls up the hill after you get your basic tune dialed in. Gradually increase boost until you just see signs of knock on a quality knock monitor like a Knock Link or on reading your plugs for signs of detonation. This becomes you ceiling for operation without WI. Then set your turn on point a bit below that boost setting, so you have a little cushion for extreme condions and any delays in the spray actually getting into the air stream.

(another way to determine your knock limited boost for the WI turn on point, based on high load, would be to intentionally run one grade poorer octane fuel than you typically run. Find the first knock limited boost point with that fuel, and set your WI turn on point at that boost level so you would be safe on a poor fuel grade one step lower in octane than you normally use)

Keep in mind your knock point will be lower in hot temps and on long duration high gear pulls, where you spend more time at high load. This is one reason dyno tunes frequently kill engines that are never double checked with a road tune. Unless the dyno can hold a load at a set rpm you can never see peak cylinder temps during a quick pull through the rpm range on most chassis dynos.


As far as timing, the folks at NACA always did their tests at MBT (Minimum Best Torque) timing. They did that by tuning for maximum torque than backing off the timing until torque dropped 1%. That guarantees your on the safe end of the torque/timing plateau. For example on a 400 ft/lb torque engine, you give up 4 ft/lbs of torque for a very large measure of safety.

As mentioned above the tuning for WI is not a single pass process but rather an iterative multi step process.

With that said, I would tune for best power without WI, and find your MBT timing under that condition. This should change very little as you add WI.

Find your best/latest safe turn on boost point.

The old school approach used by the Buick GN folks then involved an iterative process of adding water until the engine lost power, then add boost, or leaned fuel until it "wakes back up", then keep repeating that cycle until you learn what your engines tolerance level is. Usually by watching EGT's and knock indications.

All the NACA studies indicate that max power always occurs with WI at AFR mixtures leaner than max power without WI. WI at AFR richer than about 11:1 are a waste of time, as your drowning the engine in fuel and water. The ideal lean AFR will change with WI rate and mixture.

I would be slow to move away from the MBT timing, as over advanced timing can create huge cylinder pressures, with small net gains in power.

I would deal with timing as the last careful tweak to the step wise tuning process. As mentioned above your WI mixture will modify your burn speed and that in turn will slightly modify your MBT timing. So every so often you would double check to see your still near MBT, then go back to tuning via AFR, boost and WI mix.

It is interesting to note that engines of similar design, frequently cluster around a very small range of ideal ignition timings. To the point that many tuners believe you can get a given high performance engine family (say a small block chevy) very close to max performance with a widely recognized timing recipe.

Larry

HotRod (Larry) states the following:

As mentioned above the tuning for WI is not a single pass process but rather an iterative multi step process.

With that said, I would tune for best power without WI, and find your MBT timing under that condition. This should change very little as you add WI.

Find your best/latest safe turn on boost point.


I agree it's iterative as is most tuning. One question for me is: should the iteration be adjustment of the amount of water or the fuel. This brings me to the question of how much water to inject in general. There are all sorts of opinions in this area, most of which are stated with conviction.

How much water to inject

From what I've read less water is needed at low RPM under load, and the need for water increases with RPM. The same is true with boost, as boost increases (and consequently heat, abdiatic heat) so does the need for water. Makes sense.

Richard Lamb has an interesting graph of water requirement that can be seen in the Forced Induction thread: "Driving Aquamist 2c high speed valve with stand alone EMS."

Water is injected based on boost, RPM. I assume, as I think other's did in that thread that these parameters are fairly universal (hence the MF2). How the proportions on this graph are interpreted is important, but once a maximum rate is determined, ther remaining values are just relative to the maximum. The maximum injection rate is typically 10-25% of the fuel injection amount.

To begin tuning, I think a 15-20% maximum would be a good starting point. More can be injected if maximum brake torque (MBT) cannot be achieved with ignition timing.

Tuning for power with or without WI, will be targeted toward MBT. For a given engine, as Larry states, the MBT will be fairly consistent (cluster) in terms of ignition advance. An engine's MBT will differ according to its stroke length. MBT occurs near or at a point of greatest leverage on the crank when cylinder pressure does the most work. The speed with which the fuel burns will have impact on when ignition spark advance reaches maximum torque. I agree the MBT can and should be determined with WI off. With WI on, the ignition advance may be further advanced due a slower burning of the fuel.

Water Injection Point

The consensus would seem to be turn the water on as late as possible, which is sometime after boost onset, and prior to the onset of knock. This point will have to be determined empirically.

Knock onset, would be measured with ignition advance in the MBT range. In other words, pulling advance to suppress knock is counterproductive when tuning for power.

Air to Fuel Ratio

In Ed's paper (Charged Performance) he has a great graph that indicates that best fuel economy occurs an AFR or 14.5:1 and best power occurs at 12.5:1. Leaning or richening the mix above or below 12.5 leads to less power.

Please chime in on this. I've never seen someone lean out to 12.5 for maximum power. While maximum power per unit fuel may occur at 12.5:1, the maximum engine HP power may occur with a richer mixture of say 11.5 or 11:1.

O2 sensors
I have only heard that WI fouls up O2 sensor readings. By how much, I do not know. EGT has been held up as a means of tuning when O2 sensor readings are not reliably known. One way that I have heard of this working is to add WI, and let EGTs fall. Fuel is then subtracted, and the EGT's rise. Fuel is removed until EGTs reach their pre-WI level.

In reality, I imagine the O2 sensor readings are not very far off. Water is a byproduct of combustion, but not to the levels normally seen with WI. Partial pressure of water may or may not interfere with O2 detection.

Trimming WI
While starting with 15%-20% maximum maybe enough to get a good tune. Thinning the water out preserves your supply and extra water may lead to less power (bogging). I don't know how to determine when "enough" water has been injected, especially when I'm injecting post-IC and just prior to the throttle body--and no sensor, but power travels with the water past that point.

As far as I know there is a wide range of water that will produce good or best power, so trimming water back may not be a big concern.

Keeping the same format by b_boy, I like to add a few a little to each section for simplicity.

How much water to inject
Rather quoting a general figure of w/f ratio, it would be a good opportunity to look more deeper into the reason behind the suggestion. I have always lean on the side of using water injection as an in-cylinder coolant because the other alternatives are excess fuel. It is also quite simple to calculate the quantity need to perform such an task. Other cooling tasks such as inlet charge cooling is not as simple to calculate due to many variables. I hope we will reveal the way it can be calculated (it has already been mentioned in several threads), hope it wil be repeated on this section by previous contributors.

May be there is a thermodynamicist here would produce a model of how water react during the journey from injection point to exiting the exhaust pipe. Any offers?

Heat exchanging model per gram of water:
inlet cooling => induction stroke cooling => combustion stroke cooling => exiting the exhaust pipe.

Firstly I'd like to add that even small amounts of water injection significantly change MBT.

Now for the justification. :)

This stuff is all from the PDF document I submitted here several times which covers Ionization Gap Sensing in combination with feedback ignition timing and PPP control.

From page 100 of 207:

"Figure 3 A large part of the test cycle is displayed. The spark advance controller is shut off around cycle 100 and advance is held constant. The water spraying starts around cycle 250 which leads to increased PPP and decreased torque output. The spark advance controller is switched on at about cycle 400, controlling PPP back to MBT leading to increased output torque. The water spraying stops around cycle 550 and the parameters asymtotically go back to their initial conditions, when the water still in the system, e.g. deposited on the walls, decreases."

I need to resize this image to fit screen. Admin
http://www.aquamist.co.uk/forum/WaterInjection1.JPG


Now if I could just rig up my Saabs ECU to do that. :oops:

Anyhow, they were spraying a relatively small amount of water, but were running at part throttle, so it may have been a very high water/air ratio.

Nevertheless it should be noted that adding WI effectively moves the Peak Pressure Position (PPP) back about 5 degrees, at least on these experiments.

This is also the largest flaw I saw on the NACA studies. All of their conditions were at MBT without water injection. So of course, when the water moves the PPP back 5 degrees you can run more boost and less fuel. It's just like pulling timing 5 degrees.

Darn that flame development angle. :twisted:

http://www.aquamist.co.uk/forum//Mas...nedProfile.jpg

Adrian~

MBT
 

I think we are saying the same thing in different ways. Perhaps a difference in interpretation of certain words.

I don't think a 5 degree shift is all that big. I see a 3 -5 degree change in MBT as an expected change that is easily delt with.
That is why I included the following statement to periodically "tweak" the timing slightly.



In the NACA report E5E18 ( 1945) it shows that the shift in MBT timing is roughly proportional to the amount of ADI fluid injected, and depends to some extent on the mixture ( ie alcohol/water mixes had less need for additional ignition advance that pure water did.) There were also some differences noted depending on the location of the ADI injection point. (probably due to differences in the mass fraction that evaporated pre-intake valve vs in the cylinder)

So your rule of thumb for tuning would be:

As you increase WI rate, you will need to add a small increment of timing from perhaps 2-5 degrees, for each major step in injection rate, (ie going from a 10% to a 20% rate.

In the above NACA study they saw a change in ignition timing for MBT going from 29 deg -to- about 41 deg advance when going from no injection to 1:1 fuel/ADI fluid, when the ADI mix was 50:50 water ethanol.

This test series was run in 1945 where many of the other studies were run from 1938 - 1942 time frame. I understand that the ignition timing was not readily adjusted on many of the aircraft engines, so early studies focused on fixed timing operations.

The important aspect of the issue to me, is that it should be one of the last means used to bring the tune in. Too many people use it as the first thing they modify to solve knock problems.

There are some people that just keep dialing in advance and do not realize the dangers of overadvance in a high performance engine as the cylinder pressures can go orbital with just a couple degrees of excess advance.

Fuel air ratios also effect optimum ignition timing due to changes in burn speed. So by adding WI rate, and leaning the AFR you are introducing opposing effects. In the same study they found spark advance for peak power plots as a bath tub curve ( ie shaped like a large U ) when plotted against fuel air ratio. Minimum advance occured near AFR's of .08 - .095 (12.5:1 -- 10.5:1) with rapid increases in necessary ignition advance either richer or leaner than those mixes. At an AFR of .06 ( 16.6:1) the ignition advance needed was about 14 degrees more than the minimum.

Due to the effects of WI rate, mix and AFR on MBT, you actually have about 4 different ways to change effective ignition timing.


Larry

Indeed. Sorry, I hadn't meant to imply that any amount of injectant had the exact same effect.

I only meant to say that when you do the tuning without WI, you'll need to re-adjust the timing. 5 degrees forward probably is a pretty good rule of thumb. I'm not sure how much water they were injecting in this study as they were just spraying a fine mist at the exposed throttle plates. The study focused on the Ion Sensing feedback controller, rather than the WI itself.

Cyllinder pressures should not change as long as your PPP stays at the same crank angle. WI does not affect the overal pressure curve (Rapid Burn Angle), but rather just delays the flame development angle, and thus delays the combustion.

Apologies for the oversized picture. :oops:

Adrian~

Just summarizing the discussion so far on "effective" ignition timing:

1) lean mixture retards e-ignition
2) rich mixture retard e-ignition
3) injection of water e-retards ignition

MBT will be achieved by trimming (advancing) the static ignition timing according to the three variables. Is everybody in agreement in general?

Just like to iterate the above with more details:

1) lean mixture retards effective-ignition
2) rich mixture retard e-ignition
3) injection of water e-retards ignition

All above conditions will encourage detonation due to the slowing down of the frame front and the end gas has a greater opportunity to detonate. This effect is more apparent at low engine speed due to longer duration of the burnt. This is to assume the frame speed is pretty constant under boost.


1) lean mixture retards e-ignition
A/f ratio between 15-16+:1 Under these condition, detonation in almost imminent due to slow burn and reduced latent heat absorption from injected liquid. The frame-front temperature is also at its higest due to excess oxygen available to fuel the burnt, this will help igniting the end gas that causes detonation.


2) rich mixture retard e-ignition
A/f ratio of 10-11.9:1 Onset of detonation is partially suppressed by temperature reduction of excess fuel, the end-flame is cooled, but still exhibit the same threat of detonation tendencies.

Also like to add that the this method of suppressing detonation will lean to power loss due to some of the oxgen is used to produce Carbon monoxide (10% or more exiting the exhaust pipe) - full power conversion is not fully harvested inside the combustion chamber. Wasterful exercise, as far as I am concerned.

3) injection of water e-retards ignition
A/f ratio 12-15:1 At this region, the burnt rate is at its highest and power conversion is at maximum but at the same time promotes highest cylinder pressure and temperature. Detonation is again highly likely due to the other side of the burnt rate scale, fast burn effectively advances the set ignition timing. By adding water at this point would off set the early ignition and suppress the onset of detonation by ignition retard (niot as much as extent to excess fuel) and quench any abnormal burnt due to non-uniform distribution of a a/f fuel mixture.

The point I am try to make is- it may not be necessary to touch the ignition timing at all if you are running the ideal a/f ratio for your particular combustion design. Some expenses can be spared on paying out high-end third party "piggy-back" or "full-blown" ECUs to tune your water injection equipped engine. Tuning by water is less complicated than protraited on many SAE papers.

Most of my conclusion is based on some factual statement mentioned on this thread. I hope the discussion will continue and really analyse the myths and mysteries that surrounded the water injection concept for years.

ignition timing
 

I agree entirely. I believe that is one of the reason folks need to fool around with different water/alcohol ratios and injection rates. They are using those variables to arrive at an optimize point of peak combustion pressures, without changing the physical ignition advance.

That in turn leads to maximum mechanical effeciency and free power that would otherwise go out the tail pipe. One of the many ways WI allows both higher fuel economy and power production at the same time.

If you balance the water / alcohol ratio and water injection rate properly for your car, there should be no difference in ignition advance between the injected operation and when the engine is running without injection off boost.

Larry

SaabTuner,

Does your SAAB uses the same Ionization Gap sensor and are you able to read the ignition time when water is introduced? If it can, it will be great to know what the effect of water in reakl time. The Chart you shown has more or less confirming the effectiveness of the SAAB ECU.

It will be great to study the WI section a bit further as they just use a plabnt mister on the inlet and have no idea what ratio of water to air. Pity that we are not able to relate the chart against the w/a ratio.

So if your SAAB has such an ECU and just happen that you have the readout interface, we have some very accurate data.

The method for reading the ignition timing was first invented by the folks in that paper from LinkoPing University. That paper was written in 1999 after T7 (used in my car) was introduced. T8 could have it, but I doubt it.

Mine does have an ionization gap sensor, but it is used to detect knock, mis-fire, and cam phase, rather than as an ignition feedback sensor.

What I want to eventually do is modify it in a similar manner. That would be awesome to get some concrete numbers.

As for water injection slowing the flame rate, I don't believe that's true. I recal reading one of the papers on this forum, I believe it's still posted, which showed that, rather than retarding the burn rate, it retarded the flame development angle.

Slowing down the flame development angle has little effect on detonation because you can adjust the ignition timing to compensate with little or no effect on total cyllinder pressure, or rate of rise in cyllinder pressure.

It's just like a delay between when the flame kernel is ignited, and when it propogates through the combustion chamber generating heat.

The rapid burn angle is what will control detonation. A slow rapid burn angle, as produced with a bad A/F ratio, contributes to detonation.

Adrian~

ppp (peak pressure position) as you referred is a cylinder pressure plot against crank angle arriving at a peak value.

Are you implying that injecting Water will shift this PPP to the right hand side of the chart (delay). If this is the case, retarding the ignition timing would have the same effect as injecting water, I can agree to that to some extent. Torque change further confirmed this assumption.

I am interested on your view on what will happen to the ppp when water is injected after the igniton is advanced to re-align the ppp to the edge of detonation threshold. If this condition can be repeated many times until the there are so much water is being injected and the ingition is so far advanced but in the end the engine has no gain nor torque change? Does this sound correct?

------------------------------------------
ppp (peak pressure position) as you referred is a cylinder pressure plot against crank angle arriving at a peak value.

Are you implying that injecting Water will shift this PPP to the right hand side of the chart (delay). If this is the case, retarding the ignition timing would have the same effect as injecting water, I can agree to that to some extent. Torque change further confirmed this assumption.
--------------------------------------------

Yes! To be more precise, what I am referring to has to do with the latter of the two images I posted before. The second one is called the "Mass Fraction Burned" profile.

To completely understand what I mean, two concepts need to be quickly explained:

1. The Rapid Burn Angle: This is the "angle" or "rate" at which 80% of the charge burns. It starts at the crank angle at which 10% has burned, and ends when 90% has burned. Often the units for this are "milligrams/degree". In which case you take 80% of the total milligrams for that combustion, and divide that by the number of degrees it took to complete the "rapid burn".

2. The Flame Development Angle: This is the rate at which the flame begins to form. It is the amount of crank degrees which is needed to burn the first 10% of the charge. 10% of the milligrams/combustion is divided by that number of crank degrees to obtain the angle.

The flame development angle has little effect on detonation assuming you have the same PPP. This is primarily because it occurs next to the spark plug and flame kernel. Detonation usually occurs by the edge of the cyllinder wals at the end of the flame front ... or by the exhaust valves. In either case, the flame development angle has almost no effect on the processes which cause detonation.

Water Injection almost exclusively affects flame development angle. Once the Rapid Burn is taking place, the localised cooling of water is second order to the heat of the flame front. The flame temperature is well in excess of 2,000 degrees. It is the flame development that is sensitive to water ... indeed it is even altered by several degrees between 20% relative humidity and 80% relative humidity. Humidity is probably the largest disturbance to ignition timing on road cars as automobiles are usualy not equipped with humidity sensors, or any kind of feedback ignition system.

-----------------------------------------------
I am interested on your view on what will happen to the ppp when water is injected after the igniton is advanced to re-align the ppp to the edge of detonation threshold. If this condition can be repeated many times until the there are so much water is being injected and the ingition is so far advanced but in the end the engine has no gain nor torque change? Does this sound correct?
-----------------------------------------------

Possibly. But several the NACA studies were done injecting 60% as much water as air at A/F ratios as low as 9:1. So I think it would be difficult to drown the engine. You'd probably run into other problems first.

With water alone, the engine's "max output" seemed to drop richer than about 12:1. With Alcohol/water 70/30 it began to drop if the A/F ratio got richer than about 10.5:1. Remember that's still at a 60% ratio to the fuel.

In this report (812) the engine was non-intercooled, and ignition timing was set at 30 degrees BTDC. Because of the affect water has one Theta Flame, adding water was essentially like pulling back the timing ... so of course more power was available. That's one of the few problems I have with that particular study. Otherwise, the 70/30 Methanol/Water mixture allowed about 65% more torque at 12:1 A/F ratio compared to the maximum torque without any water at about 9:1 A/F ratio.

Also keep in account that the required fuel/water has a LOT to do with the thermodynamic loads of that particular engine. Engines which have well cooled internal parts tend to like leaner A/F ratios. This may be because it's estimated 40% of the cooling effect of surplus fuel is used cooling the combustion chamber, while 60% is used cooling the air-charge.

Apologies for the excessively long post!

Adrian~

I am begining to understand the published paper a bit more. The paper's main aim was to demonstrate the effectiveness of the controller using the "Ionization Gap Sensor" technique and not a study of the effect of water injection. Nevertheless, it did show the interaction of water in a combution process.

The paper also further demonstrated life after PPP. I think you might like post it here again. I too have the link of the university on can access their papers.

Based on the theory, practice and results of NACA and the above paper, it appears that water injection is a unique substance to controlling detonation without loosing power and torque as in the case of a "rich" a/f ratio.

In you next posting of the "life after ppp", if the chart show no delay/extension on "Mass Fraction Burned", it will clearly proof the water does not delay the burn-rate as a whole compare to rich a/f. Delaying the burn-rate promotes the onset of detonation.

We shall soon get to the point that we can throw some ideas how to tune an engine with water injection - b_boy's original question.

I think that we need to differentiate between non-intercooled engines and well-intercooled ones.

Ages ago I had a primitive W.I. setup on a non-intercooled turbo bike (before the compressor) and the engine would feel extremely happy no matter how much water I would inject (suck out of the nozzle more like)

I'm not sure that this would be the case in a well-intercooled setup

The 1945 documents all refer to non-intercooled engines, don't they?

PS
This is a very interesting set of threads, a far cry from the usual rubbish and slagging matches on other automotive forums.
Nice one :D

This is fascinating stuff!

I'm a biologist so my understanding can only go so far, but the physics is not so great that I can appreciate the content.

So, back to tuning.

HotRod suggests something that I've wondered for some time, that is, that fine tuning the water injection can permit an optimal balance between fuel, air , and power. This idea suggests a method of tuning quite different than is ordinarily taken. Whether the "angle", or the rate of burn, or the speed of the flame front is slowed by water injection is slowed, the effect with respect to tuning is the same: ignition needs to be retarded to achieve MBT.

We have a number effects happening in close succession: cooling of the charge, cooling of the cylinder, slowing of the rate/flame, and augmentation of the combustion itself. Most of these variables will affect ignition timing, not fuel. While WI permits more boost, thus more air, and a denser air, also more air, these WI effects are occuring pre-cylinder. The remainder are in cylinder effects.

Why is this important?

Well fueling is going to hover at a near constant ration of 12.5 to one. Thus from a tuning stand point fuel addition is constant value with respect to air mass. When tuning, fuel can be pulled back to this empirically derived minimum and left there.

Once fuel is set, the tuner has a choice: retard timing or reduce water injection. From the above discussion, I would say that reducing water is the first choice, finding a level that still permits 12.5 AFR and no knock. If the ignition advance (set by tuning with WI off) is still too little for peak power, advance can be further retarded to increase power.

In the end, a tuner will have achieved higher boost, cooler charge temp, elimination of knock, minimal water usage, and maximum power.

Now back to fuel

Is 12.5 AFR the goal for most power? I will suggest a way that this might be determined. If we use gas of differing octane, does the AFR of maximum power change? The octane rating everyone knows is a misnomer. It should be anti-knock rating, and the chemical composition of different octane fuels is different. While the maximum fuel efficiency occurs at around 14.5:1 AFR, I can see that "waiting" for the final stochiometric combustion products to form would be a detrimate to power production. The time associated with "waiting" for the final byproducts would probably hover close to a percentage of fuel left incompletely burned, no matter what the octane number.

Above is the perspective of engineer or scientist. In the real world of tuning another perspective holds. More fuel, means more power. While the 12.5:1 AFR may produce the most power per unit fuel, adding additional fuel may produce more engine power per unit time. My anecdotal observation is that no tuner tunes to greater than 12:1 AFR, usually 11.6-11.8:1 even with huge intercoolers and race gas.

Please comment on the difference.

Next I'd like to do another 1,2,3...tuning method revised to incorporate all of this discussion.

AFR and WI
 

As you can see below the relationship between rich / lean mixtures and power output is not a simple one.

If they had used different ignition timing they would have gotten much different curve shapes. My guess is the dip in power under WI in the lean mixture area is due to the (inappropriate) fixed 30 deg BTC ignition timing for the injection rate/AFR.
When they went lean enough they got enormous power output nearly 2x the non-WI max power.

NACA report 812 ( Feb 1944) They ran tests on several "internal coolants" and charted their imep vs Fuel Air ratio ( .08 FAR = 12.5:1 AFR )

One obvious lesson from these curves is there may be significant power available if the tuner is willing to explore the limits of the tune into areas that are normally considered unreasonable to even approach. Obviously proper test equipment is needed to prevent destructive knock during testing.

All these power outputs were limited by either presence of knock, they had reached max possible fuel flow, or the had reached max manifold pressure of 150 in/hg ( 75 psi boost) or preignition.

Larry

http://img.photobucket.com/albums/v1...CA812Graph.jpg

Staight Fuel = RED
Water Only = BLUE
Water/Meth = GREEN

y axis = Maximum Break Mean Effective Pressure (essentially torque)
x axis = Air/Fuel Ratio. (water/fuel ratio was constant at 60%)


Several things to note:

0. To convert Fuel/Air ratio to Air/Fuel ratio invert value.

1. As you can see, water and methanol make the best mixture, but this is partly due to the high blending octane of Methanol.

2. At leaner Air/Fuel ratios water can generate huge amounts more power when compared to just fuel. At leaner A/F ratios alcohol only adds to this a little.

3. Water Only in this case peaks at around 12:1 A/F ratio. Beyond that drowns the engine.

4. Meth/Water 70/30 peaks at around 11:1 A/F ratio. In fact, it produces around 60% more torque (and it's reasonable to assume that it would produce that much more horsepower) than just straight fuel. Indeed around 40% more than just water injection.

5. It's estimated that an engine which normally requires 100 octane at MBT, could be run on 80 octane with water/meth injection. Roughly 10 of those octane points are from the high blending octane of Methanol, but the other ten are from the combination of cooling, and retarded Peak Pressure Position.

6. I would say that very similar curves should appear even on an intercooled engine, as long as the water is completely vaporized before the combustion cycle begins. Even a poor water injection setup should accomplish that.

Adrian~

thanks
 

Thanks for the graphic, its much easier to see on a plot, but I don't have the means to host an image.

Larry

Re: thanks
 

Free Image hosting website :smile:

http://www.imagehdd.com/

And in case you need a good photoeditor for free ... do a google search on GIMP Photoeditor. :D

Adrian~

Well Adrian, what a great graph. I see you address some of the caveats here. I'm glad you posted it on NASIOC as well. Those guys are harsh on WI.

As I said on NASIOC, I find the difference in AFR optimum with WI and w/out very interesting. With it's 12.5:1 and without 9:1--an amazing difference. It really changes the way you think about tuning with WI--less fuel is the way to go, and 12.5 seems the magic number even with the crappy fuel that they were probably using in 1944.

Do they state the octane of the gas in that experiment?

I wish they had done the same experiment with less water than 60% of fuel, something not many are willing to haul around in our cars.

I also wonder if the experiment was performed at mulitple RPMs or just one?

Also, where was the water injected? Pre-turbo? Pre-throttle body? In cylinder? Me thinks it was pre-turbo.

The engine was a 7:1 compression single Cyllinder engine with Sodium filled Exhaust valves. RPM was 2500.

Inlet Air was injected by some mechanism. They had the temperature fixed at 250F degrees for the last graph.

I should have shown THIS graph first, as this graph is the same procedure, but done at 150F degrees, which is more indicative of an intercooled engine.

Same colors as before ...

http://www.imagehdd.com/d1y2004/3187...ntercooled.jpg

Enjoy. :D

Adrian~

Well Adrian, I'm tripping over myself in appreciation for these data. Such carefully controlled conditions go a long way to understanding the effects of WI. I'm a little ashamed that I too have not dug deeply into the ancient literature. As a biologist, I've read back into the 19th century for thoughts on animal and plant development, why not combustion.

It also makes me appreciate how many of the crucal data points were generated in the early era of the automobile, ones that are being repeated in the other data shown on PPP. It makes me wonder how much of the combustion dogma has been reproduced in modern engines. I suppose if radical changes to the dogma had been discovered they would have been widely distributed. But in an age of intellectual property and strict control of engineering info due to incredible competition in the auto industry I'm not sure how much useful data is published.

Your idolation of Saab and their incredibly sophisticated experiments only re-enforce my esteem for Saab. Saab and Mercedes have always pushed the envelop of auto engineering with other companies following their lead. Not to say that other companies have not made contributions, they have, but Saab and Mercedes seem to implement technology in production cars earlier than the rest.

I posted my thoughts on additional safeguards to running WI, that I wrote on the NASIOC board in response to your graph posts their, in the Avoiding Disaster forum on this WI board for further input.

It is interesting looking at this plot. It appeared to me that WI injection can make more power at richer a/f ratio than 12:1- at 9.5a/f ratio?

The alcohol plot is quite mis-leaning if it was described in text only. The engione virtually runs on alcohol rather than gasoline. The experiment is probably done on 2-valve per cylinder, long stroke enigne - octane demand/performace is that much greater is BMEP is "detonation threshold limited". However, it give a good indication of fuel quality against power.

Methanol has less than half the calorific value of gasoline, so in-cylinder cooling is doubled since twice the amount of liquid is being evaporated per unit of BHP generated. Due to the cooling and octane effect - antural way of making good power. Need a bigger tank.

I heard Indy car runs methanol and without interccoler and the turboed engine still have problems igniting the alcohol fuel due to excessive presence of liquid. I have also had the pleasure of meeting the people who run a ethanol fuelled Le Mans - problem was to stop the corrosion effect and the entire fueling system has to be washed out with gasoline after the race.

"The alcohol plot is quite mis-leaning if it was described in text only. The engione virtually runs on alcohol rather than gasoline. The experiment is probably done on 2-valve per cylinder, long stroke enigne - octane demand/performace is that much greater is BMEP is "detonation threshold limited". However, it give a good indication of fuel quality against power. "

I don't believe that is entirely true. Remember that the alcohol/water is injected at a fixed percentage relative to the fuel. If the fuel is leaned, so too is the alcohol. At a 22:1 A/F ratio, which is at the far left of the alcohol plot, the alcohol included A/F ratio is still only 16.5, and if you take into account the fact that alcohol is stoichiometric at less than half the A/F ratio gasoline is, it's realistically going to act like a 19:1 A/F ratio.

The engine in this experiment was running on AN-F-28 aviation fuel. The relative octane enhancement of the Methanol would depend on the octane of that fuel.

But whatever "octane enhancement" the Methanol provides should be constant across the air-fuel ratios. It would not enhance disproportionately more in the lean regions.

Adrian~

It is difficult to arise at a stoichmetric ratio betwen fuel and gasoline when the alcohol is at a fix ratio to fuel (it can be calculated).

At rich air fuel ratio, all liquid acted as a coolant when oxygen is fully consummed. But at lean air/fuel ratio, alcohol becomes a high octane fuel rather a coolant.

We need to define what portion of fuel is used for combustion across the plot, with the exception of water, air/fuel/methanol is still any unknown variable.

I suppose the details doesn't really matter as long as it produced a repeatable result.

Since this methanol was injected at essentially the same location as the fuel, and by the same method, it may as well have been mixed with the fuel.

Therefore I think it would be reasonable to use the "blending octane" listed for Methanol:

(Different from the straight octane. These are used when Methanol is blended with other fuels, which is essentially what is happening in this case.)

RON: 133
MON: 105

The blending octane for methanol should very accurately describe the effects of the methanol on octane given the method for injection.

The only variable now is the octane of the AN-F-28 Aviation fuel used in the study.

Also it should be noted that the consumption of methanol relative to fuel does not increase in the lean region, therefore if you have sufficient air you might find that running lean with a high meth/water injection level could allow for serious amounts of power without running a huge fuel tank for the methanol.

Adrian~

We are now getting into the realms of comparing power and octane valve, as stated on the (just about possible to read) - the chart is plotted against the resistance to knock. so if race gas is added to the chart, we will be able to another interesting curve. Keen to know the octane value of the fuel used on the test.

Most californian fuel is oxygenated, it doesn't seem to prodcue the same effect as the chart I wonder what happen to the knock resistance properties?

"Most californian fuel is oxygenated, it doesn't seem to prodcue the same effect as the chart I wonder what happen to the knock resistance properties?" -- Richard L

MTBE, which is what is used primarily in other states, is also an Oxygenate. California just has much lower octane in most places because the cost to refine 93 octane fuel at the VERY low Sulfur requirements of California would raise the price of fuel to beyond what most gas stations think the average consumer will buy.

So they makeup for the added cost of refining low sulfur by selling ultra low octane cheaper gasoline. Though you can buy 100 octane fuel at some pumps in CA, it just costs more than it does anywhere else in the US. (Except maybe Alaska or Hawaii.)

According to most tuners CA gas behaves like 89 octane everywhere else in the country. :roll: Sucks to live here.

Adrian~

AN-F-28 fuel
 

The Fuel: Spec AN- F- 28, is Graded as 100/130 using the aircraft PN system. The 130 corresponds to the rich power condition, and the 110 is the lean mixture max power rating.

For those who have never heard of the Aircraft PN system, it is an attempt to rank fuels that exceed the maximum octane of 120 that can be determined using the automotive octane tests. They extrapolate the "effective octane" of the fuels based on how much TEL would be needed to get a conventional gasoline to reach the same performance levels.

The Rich max power setting is the typical condition you would use for max take off power, using a rich mixture to prevent detonation under heavy load .

The lean max power setting would be the condition an aircraft would use for high altitude cruise, with lean fuel mixture to get maximum range.


I suspect if you looked at the specs for the California RFG gasolines you would find they are a "high sensitivity" blend. Sensitivity is the difference between the RON and the MON. RON has the most effect on engine run on at shutoff and low rpm knock, MON has most effect on high power high rpm condtion knock ( ie. exactly what most performance situations demand)

The R+M/2 AKI system used in the U.S. is simply the average between those two numbers but there (to my knowledge) is no requirement on how large a spread is allowable between the two. Most gasolines are blended with a sensitivity of 6 -10.

If your target is 91 octane you can get it with many different blends that average to that R+M/2 number. YOu could have a fuel of Ron 95, MON 87 and you would have an AKI of 91, and you could also get the same AKI with a blend that rated as RON 93, MON 89. Obviously the former would not be as suitable to use in a high performance car as the latter.

If your comparing fuels for high performance applications, pay attention to the MON numbers they are the most important.

Larry

Thanks, Larry.

Thanks for the information.

I am still puzzled why don't they add more alcohol to their fuel to boost the octane number to some where near the chart?

Too much alcohol and the Stoichiometric A/F ratio changes a great deal. Most cars are designed to be able to run on up to 20% Ethanol/Methanol blends ... but they have to make sure that the cars which are not designed for that can still buy gas.

Adrian~