Injecting prior to turbo comp' impellers
 

Hi there,

I've recently installed a charge-cooler system to my car, running ERL's 1s water injection system. I used to have an fmic with the water jet located on the outlet end-tank, which worked well, but because of the short pipe-run with the C.C (approx 8" from turbo to C.C inlet, approx 12" from C.C outlet to plenum), I'm undecided where to locate the jet now?

As I don't really want to insert the jet on the C.C outlet pipe, as I think this would be too near to the plenum (agree/disagree?), and it isn't really possible to have the jet on the C.C inlet pipe/turbo compressor outlet, I was wondering if it would be ok to have the jet/mist firing into the turbo compressor intake air-stream; ie, between the air-filter & compressor housing.

Do you think this idea is feasible? Has it been tried & tested before? (with good results?)

If so, which jet size would be a good starting point, and at what boost threshold level should the jet/pump be activating?

Apologies for all the questions/long-winded post.

Just for reference, the turbo being used is of the T28 variety running @ 25psi, and I have 0.4, 0.5, 0.6 and 0.7mm water jets to hand.

Thanks.

Mart

Depending on your plenum and intake runner lengths 8-12" may be sufficient to get a good uniform water injection/air charge mixture.

There are a few posts around about injecting before the turbo compressor/impeller. Historically it has not been good for the blades. Some think improved materials along with improved jet atomization may have negated some of the prior experience. I think in general though you will not find many people willing to risk injecting in front of their turbo.

Sorry to reawaken this, but I just want to put the record straight on this.

Pre-compressor injection offers a host of benefits. In a nutshell, injecting water (and ideally a few other miscible fluids of high specific heat capacity) pushes the compression from adiabatic to near isothermal. This is way more efficient (up to 30%). So you can reach the same boost for considerably less exhaust flow.

Yes atomisation is critical, but the technology is there.

I have been spraying into the turbo for more than a year WITHOUT any problems . I am using an 0.9 nozzle and a system pressure of 100 psi , this produses a spray with very fine atomisation.

Can you expand more on this mate?
I find it interesting... :)

Quickly. Normal compression is adiabatic, which to a first order means that as the gas is compressed, it gets hot. Now this heat is one of the biggest problems with forced induction, for 2 reasons. Firstly you have to get rid of the heat, and secondly you need to take power out the turbine shaft to perform the heating.( heat is work and work is heat).

Now with the right level of water injection, the heat is removed before it builds up, pushing the compression closer to isothermal (not all the way, but closer). In round terms this is about 30% more efficient (less exhaust gas required for the same boost, or more boost at the same exhaust flow).

problem 1 is that water on its own, whilst having a very high latent heat of vapourisation, doesn't have that good a saturation partial pressure. You can only put so much in before the air is at 100%RH. Above this the water will not cool the air until you compress it in the engine. However, if you add a second fluid, say methanol, Dalton proved with his law of partial pressures that the methanol doesn't know that the air is saturated with water vapour and goes on to vapourise as well. Add a 3rd, such as acetone (all available chemicals) and you get a 3rd tranch of cooling.

The net result is that you may be able to cool the air to slightly below ambient with the right mix. So your compressor becomes more efficient, you can throw away the intercooler, increasing your flow, and in some cases significantly improve the flow from turbo to inlet.

If you leave the intercooler in, you are generally warming the air back up again, so you have to take the plunge and remove it to gain the best benefit.

Its good, very very good.

Bill

Bill,

your theory sounds very exciting (to me, at least)

So to recoup, you're saying that injecting before the compressor actually increases the efficiency of the compressor - leading to less heat in the charge as it leaves the compressor.

You're also saying that injecting a water/methanol mix at that stage would be far better than just pure water, right?

I always thought that such a setup would decrease the intercooler's efficiency, compensating most of the benefits. You agree with this, to the point where you see the intercooler as a thermal liability

So what sort of injection volumes are we talking here?

spray point
 

Yes spraying before the compressor impeller has its place.

As stated above, it improves the effeciency of the compressor. Many people tend ot look at WI from an engine centric point of view. If they cannot infer a direct benefit to the engine they assume it is a bad idea.

You are working with a complex system of mechanical devices that interact with each other in many ways. Even though on first blush injecting infront of the compressor or between the compressor and the intercooler might appear to be less effecient you need to account for ALL the interactions. In many cases we simply don't have enough information to predict the results so frequently experimentation will give you better data in a matter of minutes, than all the incomplete computer simulations you can afford.

Injection in front of the compressor accomplishes several things. A turbocharger is a constant pressure variable volume "DYNAMIC" compressor.

A turbocharger only knows 2 important properties of the gas it is compressing. The density of the gas at the compressor inlet and the pressure ratio it is operating at, which is determined by the rotor rpm and the gas density. If you increase the pressure or reduce the temperature at the inlet you will modify both of those parameters. In both cases (increased inlet pressure, or lower inlet temperature) you increase the apparent density of the gas passing through the compressor. At a given rotor rpm with a given gas density you will flow a very specific volume of gas and it will be compressed to a specific pressure ratio on exit. That is what the compressor map is based on. If you change the inlet conditions (gas density) you in effect slide the compressor map left and right. This is the "corrected flow" of the turbocharger.

By injecting water/alcohol ahead of the compressor two things happen. You cool the inlet air substantially, this in effect moves your true operating point to the left on the compressor map. (in most cases for max performance this is a good thing, although on some turbocharger conditions it can cause compressor surge.)

You also change the pressure temperature profile inside the compressor wheel itself. You probably actually change the shape of the compressor map. As the gas moves outward and is compressed, heat that would have gone into heat and increased pressure is absorbed by the WI mist and so the compressor has less work to do since it is no longer fighting this temperature driven pressure increase, it can achieve more mass flow at that pressure ratio. The cooling should also modify the speed of sound in the gas and the mach number of the compressor blade tips should also change. This should change the choke flow characteristics of the compressor but I don't have the information to comment in detail on that.

Net result is, you increase the mass flow through the compressor --- in effect you make it act like it is bigger than under normal conditions.

During WWII this was the way ADI (anti detonation injection --- the common term in the aircraft world for WI ) was set up on military aircraft in most cases. The water and the fuel was injected into the eye of the centrifugal supercharger.

Errosion of the compressor blades is not a problem if steps are taken to ensure the mist is very fine when it arrives at the compressor inlet so that it follows the airflow and does not imping on the blades with a high differential speed.

The ideal is to get drop size down as close to 10 microns as possible but due to the brief periods of use and intermittent nature of most WI systems, in reality you can live with larger drop sizes in real world systems.

If you have ever ridden a bicycle or motor cycle in a rain storm you know how sharp the impact of a large water dropplet can be, but a fog or drizzle will not cause the same painful experience because the droplets are small enough they are strongly influenced by the direction of flow of the air stream around you and impact with much less velocity and obviously also have lower momentum.

For people in hot dry climates that lose a lot of turbocharger performance in hot weather, pre-turbo injection should be looked at.

In regard to the effects on the compressor mass flow, maximum results appear to occur with mist flow of about 3% - 10% of the air mass, so you will likely need to inject additional WI near the throttle body to reach maximum detonation suppression and best power.

Larry

Re: spray point
 

If you mean that the ambient temp air flow is affected, I wouldn't expect that to be much. Say ambient is 20C, how much lower can it get by injecting mist of water that's 30C+ (the bottle is in the engine bay!). I suspect that after the pump the water is even warmer.
I would have thought that this 'efficiency gain' is the effect of water droplets inside the compressor blades, as they try to squeeze the air.
Am I right?
This one actually
fascinating :!: ...

cooling
 

Actually it can be very substantial. The cooling is by evaporation, and typically will exceed 10 - 20 deg C . The initial temp of the water is of little consequence. The cooling due to evaporation (latent heat of evaporation) is very much larger than the latent heat of the liquid water.
As you can see below the evaporation of a gram of water will absorb 542 x the heat energy required to lower that same quantity of liquid water one degree in temperature. That means that even if the water were nearly at boiling temps when injected, evaporative cooling would reduce the air flow temp below ambient temperatures long before all the water changed to vapor.

Specific heat capacity of liquid water - 4.187 kJ/kgK
Latent heat of evaporation - 2,270 kJ/kg

Since the evaporation of the water and the alcohol will essentially stop when the air becomes saturated (ie. 100% humidity) there is a practical limit to the cooling or around 20-30 deg C for alcohol water mixes, and in real systems you seldom get much more than about 15-20 deg C.

For every 5 deg C you drop the inlet air temp you will increase mass flow by about 1% due to density increases.


Only in a very crude sense. Your thinking in a mechanical piston pushes on air sort of way, but what happens inside the compressor impeller occurs at a molecular level.

Think of it this way If you could freeze frame time, and stop what was happening inside the impeller while its spinning at 120,000 rpm. Each impeller passage between adjacent pairs of compressor blades contains a wedge shaped parcel of air. When spinning at 120,000 rpm the air is subject to huge centrifugal forces as it moves away from the hub of the impeller and toward the rim of the compressor. The trapped air would like very much to be slung out of the impeller but like a crowd at a stadium after a match it simply cannot all get out as fast as it would like. As a result it stacks up (compresses) as it gets near the exit. In this process a lot of internal friction occurs. The air near the tips of the compressor might be moveing near the speed of sound at maximum flow, this heating makes the air try to expand. This increases the pressure which fights the outward movement of the air. Eventually a balance is achieved between the centrifugal forces trying to throw the air out of the impeller and the pressure build up due to the compression and the pressure build up due to the heating. The addition of the water mist removes a very large fraction of the pressure gain due to heating. As a result more air can exit the impeller over a given period of time, and more of the pressure gain is real compression rather than waste heat. The net result is a more isothermic compression which is always more effecient than an adiabatic compression.

Larry

Re: spray point
 

another 2 minor points:
1. a 2 litre engine at 1 bar boost and 5Krpm will be consuming around
10.000 litres of air per minute, or 10cubic metres of air, which is about 12.2 kg air/minute.
10% of that equates to 1.2kg water /min, that's quite a lot isn't it? With the aquamist pump and nozzles, it will have to be closer to 3% in practice.
Would this be enough?

2. Wouldn't the extreme centrifugal forces in the compressor force the tiny water droplets to become liquid again?

Re: spray point
 

A 2 liter engine consumes 2 liters every TWO revolutions (720 degrees) so assuming 100%VE, 100% intercooling, it would consume 5,000 liters/minute.

Adrian~

Re: spray point
 

you think you got me there on the two revolutions thing? :twisted:
Well you didn't, because I mention 1 bar boost pressure, which multiplies it by two.

Nice try though. :wink:

stuff
 

Four cycle engines only intake stroke every other revolution for each cylinder ;)

JohnA -- actually its not all that much for two reasons. First your not on full boost for more than just a few seconds at a time unless your making a landspeed record run on the salt flats.

Also 1.2 kg/min is only about 20 cc/sec or about 2/3's of a shot glass of fluid, divided among the engines cylinders. On a 4 cylinder 2 liter engine with 500 cc/min injector running static will inject 2000 cc/min of fuel

The turbo and the air flow currents and heating through the rotor actually tend to shred and tear the dropplets apart and cause nearly explosive evaporation, very little liquid water exits the rotor.

If you want to have a base reference on how much water an engine can ingest during WWII Pratt and Whitney ran what they called flood tests on their aircraft engines where they turned up the water injection to the point that liquid water was pouring out the exhaust ports and the engines still ran with no problems. That level of injectant reduced the engine power from a max of 2000-2800 hp down to about 600 hp but there were no problems.

As far as the limits of the pump and injectors you are probably correct. That is why I am running a 100 psi Shurflo pump with a capacity of 1.4 gal/min ( 5.3 L/min) and a 5 Gal / hr nozzle ( 315 cc/min) on a rather small 13G turbocharger. The 13G only flows about 25 lb/min of air (11.45 kg/min) which comes out very close to 3 % of air flow for the spray flow.

I will probably be adding nozzles here soon.

Larry

Re: stuff
 

...and 4-cylinder, 4-stroke engines (inline) have one cylinder in intake stroke every half revolution...the end result is the same: every 2 rpm 2litres of air have been inhaled (assuming 100% VE of course)

But I'd like to believe that the level of discussion here is a bit higher :cool:

Yeah, but when calculating (roughly ofcourse) the max water flow, we have to assume sustained full throttle.
And Volumetric Efficiency won't be 100% of course, especially at full-power revs.

explosive evaporation ... interesting. Didn't know it even exists! :lol:

It's OK for people with traditional superchargers to say that injecting all this water before the blower is fine. But I had reservations about centrifugal compressors. If the centrifugal forces don't mess up the mist, then that's one less headache.

So such a setup wouldn't be boost-driven like std water injection, would it?
To realise the gains of near iso-thermal compression we'd want to inject water from zero boost, yes?

Do you think that's safe then?
Unfortunately my aquamist setup is not boost-sensitive, it's rather on/off.
:cry:

details
 

Very little of the mist survives the path through the compressor in liquid form. In the discharge you have much cooler air at very high relative humidity but almost no water droplets.

Literally 10's of thousands of military aircraft engines in the period 1939 - through the early 1950's used before the compressor injection with no problems at all. It is also widely used in the tractor pulls, and on diesel trucks, CART race cars I understand, used a pre-compressor injection of some of their fuel for the same purpose, a handful of production cars and trucks have also used the system.

All you have to do is avoid large water droplets or a solid stream of water from impacting the impeller.

Depends on how you are defining boost driven --- some systems use a pressurized water reservoir so that boost pressure provides the motive force for injection. These systems are best for trucks pulling heavy loads up mountain passes where they stay on boost for extended periods of time. A presurize reservoir system will not react quick enough in my opinion for a high performance application.

Other systems use a pump supplied pressurized system with a valve of some sort that opens to allow spray at a certain boost pressure.

You generally do not want a continous on system (except for an rv or truck as mentioned above). On my system I have a pressure switch which activates at about 8 psi to turn on the spray. It is a simple on/off WI system. It will spray anytime boost is over the 8 psi threshold I have currently set.

As far as computing max water flow, your correct to establish a limiting case you would figure at max air flow which would typically be at max power rpm and a true engine VE of about .85-.88 for most engines.

However for detonation control your most critical flow rate is in the midrange max torque rpm range, say 3000-5500 rpm. That is where the engine has max VE and by definition you have max torque because you have max cylinder pressures. This is the RPM range where a sudden loss of WI can be fatal to the engine.

If you look at a manifold pressure log of a typical turbocharged engine when the turbo begins to make serious boost you go from near normal atmospheric pressure to a significant boost pressure >6 psi for example in just a fraction of a second ( ie only a 1000 rpm change in engine speed at WOT).

On my WRX on a 3rd gear WOT blast, I go from -1.2 psig manifold pressure at cruise, to 13.5 psig manifold pressure in 0.922 seconds when I just stab the throttle wide open, on the shift to 4th I go from 10.4 psig to -8.4 psig manifold pressure in 0.234 seconds as I close the throttle for the shift, and then jump to +5.9 psig .235 seconds later as I go back to WOT and another 235 milliseconds later (the time resolution of this log), I am backup to a manifold pressure of 11.9 psig. (this is all at 5800 ft altitude).

Total time under load in Low gear is about 1.5 seconds, 2rd gear was 3.828 seconds, 3rd gear 2.359 seconds, and 4th gear only lasts a couple of seconds before I'm over 100 mph.

So in a highway acceleration type situation, you would seldom see more than 7 - 8 seconds of flow, with brief interruptions of about .5 second every 2 - 3 seconds, so your actual duty cycle for the spray activation will be something like 80% or so. In my case I have an accumulator in line with the solenoid so the system pressure changes very little over this short interval.

If you are looking for detonation supression you need it most in the midrange rpms near the engines torque peak. That is when the engine is most likely to detonate under high engine load. You want the WI to kick in just below the rpm range and boost pressure that you can first experience detonation under heavy load. If you turn it on early you cool the exhaust gasses just as the turbo needs hot EGT's to help it spool up. So in that case you delay the WI spray as long as you safely can.

If your looking for maximum efficiency of the compressor, you want the spray to come on at a boost pressure when the compressor is just moving out of its maximum effeciency island on the compressor map. What you are doing is artificially stretching the compressor map to the right so the compressor has a wider island of efficiency.

In my case I will eventually be using a compound system with some spray pre-turbo for the purpose of maximising the compressors efficiency, and if I need it, a secondary jet spraying pre-throttle body to control any detonation that is not supressed by the pre-turbo injection.

Larry

On advantage of injecting pre-compressor is that, should the WI fail, it would take several seconds for all the water to evaporate off the walls of all the tubing before the engine. This was noted by the folks at Linkoping University when doing experiments with WI and an Ionization Gap Sensor feedback ignition system.

If you have a very very good ECU on the engine a WI failure would be not much worse than running into a bad tank of fuel and the ECU should eliminate the knocking within one or two engine rotations. Most engines are designed to be able to handle rapid changes in octane even if they like higher better.

EGT might be another problem, but if you feel the car suddenly getting slower it doesn't take a rocket scientist to know something might be amiss and worth investigating.

Adrian~

Re: details
 

That's easier to implement actually, I'll probably give it a try. Just Tee the existing WI line to another nozzle just before the compressor. The aquamist pump should keep atomisation at decent levels, eh?

Re: spray point
 

Technically the engine consumes 5,000 L/min but at 29.4 psi instead of 14.7. :wink: But fair enough, I didn't notice the 1 bar bit.

Adrian~

Re: spray point
 

The exact psi figure depends on atmospheric pressure conditions, doesn't it? And assuming sea-level is a bit presumptious. :smile:

Tell you what - why don't we end this futile excercise in bandwidth wastage? :wink:

You guys make my head hurt!

Observations from a mechanic that drag races are that injecting prior to the turbo produces positive effects in every instance. The ones I've seen that say injecting prior to the turbo creats pitting, are also the ones that run their cars with volocity stacks and NO air filter.

Kent

I'm into trying this out. I've now read what I can on the pre-compressor experiments and it looks like if the injection is not continuous and droplet size is as small as possible, you'll have minimal wear on the impeller wheel.

I'll have the Element Hyday in my STI in about a week or two. I can program it to open a valve at 5000 or 6000 RPM at 15 psi of boost and let the Aquamist take over from there.

From my reading on "wet compression" or "fogging" as decribed in the turbine engine lingo it looks like 2-3% water(alcohol etc):air is about all you need or want to shift toward isothermal compression. Any thoughts out there on how much to inject?

I'm primarily interested in try to "stretch" out the compressor map at high RPM/boost when my stock turbo is starting to wheeze.

How about droplet size from an Aquamist jet? Hotrod mentioned 10 uM. That is pretty small on the absolute edge of visual perception of a single dot. The smallest cells are 10 uM and you can only see them if you know what to look for.

Some sort of atomizer like is used in "cold" humidifiers might be a better solution. They use a small apeture combined with some sort of high frequency osillation to break up the water. Or maybe an ultra sonic tip of some kind.

drop size
 

The 10 micron drop size would be ideal, but appears to not really be necessary in a practical real world system.

Most common spray mist nozzles can achieve sprays with drop sizes down near 50 microns, plus you have evaporation that takes place as the mist moves down the induction path, which reduces the maximum drop size.

I just finished a turbocharger swap, and I took a good look at the compressor impeller on my old turbo which I had pre-compressor injection on for several months at about the 3% rate. (4 GPH nozzle max rated air flow for the compressor at sea level is about 25 lbs/min. Here at altitude of 5800 ft it is probably about 20 lbs/min.)

If the nozzle was spraying at its rated 4 GPH that would be about 252 cc/min or about .55 lb/min which works out to 2.7% of air mass at max flow. My turn on point is 8 psi so at the turn-on air flow, the mass fraction would be higher, probably near 5%.

On casual examination you could see no evidence of compressor blade errosion, just the normal discoloration you see after the turbo has been in use for some 37 thousand miles. On very careful examination under high magnification (about 10x), you can just see a small bit of roughness on the outermost 1 - 2 mm of the compressor impeller blades at the very leading edge.

I will need to do a similar examination of another turbo of the same design that was never exposed to water injection to determine if this is normal wear. This engine (like most Subaru WRX engines) has a crankcase breather inlet in the inlet tract a few inches ahead of the turbo inlet. Under high boost, these engines can blow noticiable amounts of oil mist into the intake. This over time results in a build up of a very thin layer of oil and dust "crud" on the inside of the inlet pipe and I'm sure that from time to time bits of this crud, and small oil dropplets gets suspended into the intake air stream under high air flow conditions.

I also on 2 occasions ran this engine with no air filter for a couple of drag strip passes to determine how much influence air filter resistance had on performance, so there was most likely a small amount of dust ingestion from those experiments.

Lastly when I first fabricated the system I assembled the prototype with a 35 psi pump which would not have given best misting behavior by the nozzles. The last couple months was with the system running at 100 psi max pump pressure which should have given a much finer mist.

Larry

Thanks for the data.

I'm thinking I will inject about 6 inches after the MAF sensor, the water will travel about another foot to the turbo. I'll look into getting a couple of thermocouples to measure temp post turbo and post intercooler. I'll have to rely on data logging to look at pressure change rates and increases +/- pre-turbo WI.

I shoot for 2-3% injection rate and see how it goes. I'll have the pre throttle body injection going too. By the end of month I should have some experience and data.

Excellent thread, the most compelling read on a forum ive had in ages.

This is my pre-turbo injection nozzle, half way between air mass meter and compressor.

One thing I can say for certain is that the (aquamist) pump is louder now that it runs both nozzles.


http://homepage.ntlworld.com/johnnya..._nozzle_lg.jpg

Aquamist 2d installed on 2 ltr engine tuned to 250 - 260 bhp @ flywheel.

Two questions,

Ive decided to inject at the throttle body and before the turbo, what size jets will i need ?

Intercooler becomes interheater ! Ive had the stock intercooler replaced with a larger one, im a bit nervous of having no intercooler, what about when pressure is below the injection trigger level (9 psi) or if the water injection fails ?

I still have the stock intercooler, which i believe has a lower pressure drop.

I would not worry about the intercooler becoming a heater just yet. Prove it's heating and then worry about it.

2-3% of air mass injection pre-turbo is just an educated guess. I read some turbine engine papers that indicated 2% was best and hotrod figured he was injecting 2.7%. Turbine engines are not the best model for our little turbos, much higher compression, much larger surface area. I'd love more input on what percentage to inject and droplet size if anyone has these data. Maybe Richard knows some mavericks.

2-3% is alot to inject. Generally WI injection is expressed as percentage of fuel injection 2.5% of air mass would correspond to 25% at 10:1 AFR, or 33% at an AFR of 12.5:1. That is a lot of water to move through the turbo, more than my pump and reservoir can flow.

I will start with something more modest. I have two 0.5 mm jets at the throttle body, shunting to a third 0.5, 0.4, or 0.3 mm will give me a range of about 125 cc/min, 90 cc/min, and 60 cc/min, respectively. At its highest, that corresponds to about 0.5% water:air mass with 1% water:air mass left over for injection at the throttle body. So, I would be injecting only about 20% of the amount that hotrod injects.

That may be too little, but I'm not sure anyone knows. I can monitor pressure and temperature post-turbo and see if temp continues to drop and pressure increase with added water.

Commentary on this set up is most welcome.

Another up-shot to injecting prior to the compressor is for those people who run greater than 30 psi of boost. Above Pressure Ratios of 3:1 the air-charge is heated enough at the aluminum compressor begins to weaken from the heat.

Holset has been working on their cast Titanium compressors to solve that problem, but water injection might also help to keep the compressor wheel sufficiently cool.

Just a thought. It's probably already been discussed anyway.

Here's Holset's page on "high pressure ratio compressors": http://www.holset.co.uk/files/2_5_1_...compressor.php

Adrian~

Inlet tract Simulator
 

I found this link to a simulator from JohnA's web site, it has some features that simulate water injection before the TC or IC and after the IC.

I thought it would be interesting if i used the data i had from a RR to see the effects of WI. I entered the variables to the best of my ability, the models behavour is quite realistic. RR 230 bhp / 222 lbft @ 5371 on a cold March morning. I can only guess at VE.

The link below has all the data saved to it, which means you can see what i mean when you change the WI settings:

[admin - have to make the link's font smaller to enable the page wrap to function]


http://not2fast.wryday.com/turbo/glo...orDutyCycle=85



Is interesting to note that a the model show that injecting 400cc before the TC, which is about 3.5% of the air liquid ratio, shows that the IC now only cools the charge by 1 C. The IC is at the threshold of becoming an Inter heater.

Injecting 320 cc, which is about 2.9% of the air liquid ratio, (and the limit of the pump with 1 injector) shows the IC still being useful cooling from 31 C to 19 C.

Probably shouldnt read to much in to this, but it would indicate that to inject at the manifold, a seperate pump would be desirable, in which case will the FIA 2 run two pumps and HSV's ? I remember that the MF2 can !

Your views....

work sheet
 

That is an interesting work sheet on the effects of WI, but you need to read the cautions about the model. I did a lot of playing with it and at reasonable amounts of injectant, it seems to give very good information.

It does not do any "bounds checking". If you put absurdly high amounts of water in, it while happily compute the cooling from total evaporation even though the amount of water is far beyond what is necessary to reach 100% humiditiy. Sometimes giving unrealistic air charge temps as a result.

So like any tool its no better than the base assumptions of the model. It does however, clearly indicate the basis for the assumption that post intercooler injection is the best "from the perspective of intake air cooling per gram of water injected".

It has no way of accounting for things like the change in timing of the pressure peak in the cylinder with different amounts of water and how that would effect the power output of a specific engine design.

The pre-turbo injection totally discounts any changes in the turbocharger compressor effeciency and any secondary feedbacks like, higher mass flow, lower turbo discharge temp, or lower exhaust gas back pressure that might result from that increase in effeciency.

Of course all those changes would be turbo type specific and would require a ton of code to cover all the popular turbochargers.

Larry

I too have played with that calculator. And for me the best thing about it is getting a handle on how the different components of intake, IC, turbo, AFR, and WI cooperate to make hp. I agree with hotrod the model used in the calculator is too simple for our stuff and does not account for the saturation limit of dissolved water. The 100% "humidity" is an upper bound of WI, but both ambient air saturation and temperature affect this bound. Ideally we'd like to have a system that never reached the 100% saturation limit at any point in the air path (the bound is going to change, being highest just after compression, and probably the lowest while inside the intercooler". We dont' want precipitation at any point (e.g rain fall, fog formation), that will lead to water build up and all sort of "unintended" effects.

One calculation that I think would be worth performing is to figure the amount of heat produced at the highest operating RPM of your turbocharger and then calculate using the heat of evaporation for water +/- % methanol the amount of fluid needed to bring the compressor to isothermal compression. Assume that you are getting "explosive evaporation" as hotrod suggested.

Instead of using a compressor map with "predicted efficiencies" and a PR, if one were to actually measure pressure and temperature pre and post turbo, the energy going into heat should fall out of the ideal gas law. Now we divide this energy loss by the energy gain of evaporation, and we'll get a percentage of water/methanol injection. It's a ball park ideal figure, but it could be informative. Anyone have thermocouples in said places to take the measurements, and a boost gauge?

Ok, I'm out on a limb here, but I did some thermodynamic calculations on water absorbtion of heat during compression at the turbo impeller. The calculation is in 4 parts and relates to my WRX STI 2.5 L turbo car.

Part 1 Assumptions and definitions

At 15-16 psi the engine is consuming ~250-275 g air/sec
Temperature of compression at 60% turbo efficiency, produces ~100 degrees C change at 7000 RPM
Heat of evaportation 540 calories/ gram of water
1 calorie = 1 degree C/ gram/sec at one atmosphere
1 calorie = 4.184 joules
Specific heat of nitrogen gas 0.25 calories/degree C (air is mostly nitrogen)


Part 2 Heat change under compression and amount of water to restore starting temperature


For 1 degree change in water temp we get 1 calorie*grams/sec.

For 100 C change (increase) of air (nitrogen) of 1 gram = 25 calories to heat 1g of air 1 degree C

540 calories/gram of water evaporated, heat of evaporation

Too cool 250 g air / sec heated 100 C costs 250*25 or 6250 calories / sec

6250 calories/ 540 calories/g water is 11.6 g water / sec = 11.6 ml water/ sec or 695 ml of water / min

Thus, 695 ml/min of water brings the compressor to isothermal compression with 100 C change in temp
during compression.


Part 3 Engine's consumption of air at 7000 RPM in a 2.5L engine


For my 2.5 L engine at one atmosphere and 7000 RPM, that's 3500 cylinder fillings per min,
or 2.5L * 3500, or 8750 L / min * 1000 ml/L =

8.8 million ml of air / min

At 15 psi of boost the amount of air is double or 17.5 million ml / min


695 ml/min cools 17.5 million ml/min air 100C or

0.004% (by volume) water to air ratio = isothermal compression with 100C temperature change


Part 4 Aquamist pre-turbo injection effects with my set up


100% stated "efficiency" of a turbo compressor is not isothermal, but adiabatic compression.

With Aquamist and a 0.5 mm jet I can inject about 125 ml / min, that is about 18% of 695 ml / min.

The turbo is operating at ~60% adiabatic efficiency at 7000 RPM, the 695 ml/sec water to brings it
to isothermal compression which corresponds to about to 118% efficiency at a pressure ratio of 2,
a difference of 58%.

18% (my 125 ml/min) of 58% is 10.5%

60% efficiency of turbo + 10.5% cooling from water = 70.5% efficiency of compression adiabatic.

250 ml / min water injection gives twice that or 81% compressor efficiency.

So with just 125 ml/min WI, compressor efficiency is near 70% at redline, quite a feat!

It's like having a 20% larger turbo.

I don't know the actual change in temperature during compression at 7000 RPM on my '04 Subaru STI with
the stock turbo. 100C is a reasonble guess, but it could be as high as 150C if the compressor efficiency
is lower than the predicted 60%. Thus, these numbers are only a rough estimate. While isothermal
compression would be nice to achieve, I can be happy with just extending the efficiency of the turbo
beyond it's normal bounds, and make up for the adiabatic heat by injecting more water after
the intercooler. Under this scenario maximum total water injection is 15% of fuel mass, very reasonable.

my experience
 

That fits well with my experience.

The cliffs notes version is:

As you make the compression more isothermal, the true operating point of the compressor moves to the left on the compressor map. Since most performance applications push the compressor off the compressor map on the right side as they try to squeeze the last bit of performance out of the turbo this is in most cases a good thing.

Here at high altitude, the already small TD04L-13G which comes stock on the WRX is not just operating off the compressor map, it is on the next page.

When I started with pre-compressor WI, it made the turbo much more responsive in the mid range, and improved the max airflow a small but noticable amount. The mid range rpm onset of boost became so agressive I had to modify my boost controller settings to keep the throttle from becoming an ON-OFF switch.

Larry

Thanks Larry for the "real world" experience.

It's just as I would like, move the compressor effectiveness back (left) into a more efficient island of the compressor map. I'll compress not necessarily more air, but compress it more easily with less intoduction of heat.

What effect the water will have on impeller speed questionable. While theoretically more viscous, the water mist may evaporate so quickly at the impeller to essentially behave as a gas, albeit a more dense gas. I believe it will slow the impeller to some degree, another benefit that I think Saabtuner mentioned.

I'm going to try to get WI volume up to 20% of fuel, such that when shunting 1/3 to the turbo we sending a bit more water/methanol through the turbo, around 0.7% by air mass. Or I'll shunt a larger percentage than 1/3. I'll have to see what works best.

We'll see how it works around mid-Oct.

I've measured this jet at well over 200cc/min, almost 240cc actually with the engine running.
That was static, with no boost (or vacuum) in front of the nozzle.
If it's pre-compressor, then maybe the flow could be even higher (I'm not sure about this without testing further)

I wish I can contribute more but it is getting beyond my level.

It might slow the impeller slightly ... or it might speed it up. The reason I think either is possible is that whether it slows or speeds up will depend on exactly where the water is atomized. If atomized near the tips of the impeller blades (as I suspect it would be) the over-all pressure drop across the turbo would be reduced at a given mass-airflow, and that reduction could allow the compressor to speed up much more quickly. I think Larry said something about it turning the throttle into an "on/off switch", which would go along with those lines perhaps?

Also I mentioned earlier that it would cool the impeller wheel slightly, which when boosting past 2 bar can be quite beneficial.

Adrian~

It's interesting that you would say that Saabtuner. I've reading about air density and humidity. Surprisingly humid air is often stated as less dense. Let's take a different look at this, not from a throttle/turbo response view, but an efficiency point of view.

At the same temperature, water+air should be more dense than dry air alone--a simple more molecules/cubic area argument. I think the reason for the humid air being seen as less less dense than dry air is a misunderstanding of temperature, hot air being significantly less dense than cold air--hence the hot air ballon--and hot air holding more water--higher saturation point (the mid-west is summer).

If we picture a droplet entering the center of the impeller, it is immediately whisked centrifugally outward and "backward," evaporating along the way, perhaps explosively. Locally, liquid water is being transformed into gas, along with the air being compressed. On a purely local level there are more molecules in a gaseous state at the end of the compression, but since the liquid evaporated heat was absorbed.

If we imagine a model were droplet size reduces exponentially as it traverses the impeller to expeller length, at the beginning the density of the air with or without the droplet is nearly the same, however, at the end of the traverse (notably the point of most leverage) the air + droplet gas will be much more dense since the gaseous contribution of water will be huge.

I think what we will see by injecting water is that there is a slowing of the impeller wheel that results in the same compression of air at that slower speed plus the added benefit of reduced temperature of that compressed air. It my case, turning water at 5000 RPM, we may shift the impeller speed back to where it was at 4000 (and 70% efficiency), as we increase RPM on up to 6000, 7000 while increasing the water injection correponding, we can try to stay at the same 4000 RPM, 70% efficiency range.

This is one way to look at: view it as creating a static impeller speed of high adiabatic efficiency (e.g. the 4000 RPM equivalent impeller speed, say 120,000 impeller RPM) by injecting more water to slow the impeller down. But we don't want to do that, we want the impeller to continue it's upward climb in speed so that it can compress more air. The reality is most likely in middle, slowing the impeller some (while cooling the air that is compressed), but allowing for more compression with increasing speed (but still cooling the air compression). There is no way density of water + air is going to completely overcome the torque of the exhaust side.

Now, coming back to the rapid throttle/turbo response view. The quickness of the throttle is going to be mainly a function of the rate of air and fuel entering the engine. Early in the turbo's speed range, slowing the turbo's ascent in speed will be a disadvantage, prior to peak compression, in fact the impeller is an impediment to air flow if the wheel is not turning at all. Once a decent efficiency is reached though, any further cooling of the air being compressed will result in more air in the cylinder for every turn of the impeller, thus the rate of air will be faster at that instant, and the rate will increase with increasing water injection linearly. I think that is where you could see an increase in throttle response over no WI pre-turbo, but it would depend on maintaining the temperature difference as the compressor continues to spool.

I can't believe we stumped Richarl Lamb with this stuff. To his credit, it is more turbo theory than WI theory. We are all learning something I guess.

Thanks to John A for the measurement. I'll have three 0.5 mm jets, so I don't think I'll achieve his volume as it is distibuted between the jets unless I add another pump or greatly increase the pressure.

In the end, it's going to come down to testing it out. :wink:

I think there may also be some mixing here of the term "pressure". Static pressure and flow pressure (or total pressure) are somewhat different things.

"evaporating along the way, perhaps explosively. Locally, liquid water is being transformed into gas, along with the air being compressed."

When thinking of this problem remember that the turbocharger's compressor wheel only accellerates the gas. The compression takes place in the diffuser scroll in the housing. Most of the evaporation/boiling should take place towards the tips of the impeller as that is where the static pressure should be lowest (though velocity pressure quite high). Thank the Bernoulli principle for that one.

While I'm not sure if the water's evaporation would increase or decrease the density of the gas, I think it's fairly safe to say that the evaporation of water in the compressor would significantly change the flow through the compressor.

The reason I say that is that on the forward facing edges of the blades there is much higher static pressure than on the backwards facing edges. Since more water will evaporate in lower pressure, the low pressure regions may either get MUCH lower, or get much LESS lower, depending on what effect the water has on static pressure when it evaporates. In either case, it would dramatically change the flow field through the impeller.

Adrian~