Water Injection Analysis

[banana_socks]

So after trying to setup my Water Injection system, and getting a little overloaded with conflicting information about the best setups, I tried to do some calculations to clear things up. In particular I wanted to know how much of a difference each control method might have. Most basic kits work on a simple on/off switch at a certain boost level. The basic progressive controllers ramp up pump voltage (hence pressure) based on boost. More complicated controllers can vary flowrate based on AFM or injector duty cycle.

At a basic level, it would be great to have the water rate proportional to the level of fuel or air. I'm basing this on a 3SGTE (as thats what I have), and I'm going to work from dyno charts on 3 states of tune:
-Stock 3SGTE
-Upgraded turbo @ 13.5psi
-Upgraded turbo @17psi, exhaust, cams and tune



Now we can approximate the amount of fuel needed. A rough rule is 5cc/min per horsepower, so multiplying the horsepower figure by 5 gives the total fuel flow in cc/min.



Most recommendations are for the amount of water to be 12.5% - 25% of total fuel (by weight). Adjusting for difference in density, it gives a target of 9.25% to 18.5% by volume.
Considering just the stock engine for now, we get minimum and maximum water flow rates.



If we were to use an 80cc/min nozzle to inject water, it would supply the recommended amount of water from about 3200rpm until 6000rpm. (Stock CT26 is useless beyond). A simple on/off switch would work fairly well.

Looking at the mild engine.



If we used a 100cc nozzle, it would be within the limits from 3500rpm until 6500rpm. However its towards the lower end of the ratio, and if we increased the nozzle size it would supply too much water in the lower part of the rev range. A progressive controller based on boost would actually work quite well here.

Finally looking at the modified engine.



Using a 140cc nozzle would give us the right ratio from 3500rpm until 7000rpm. However again we are towards the lower side of things. Using a larger nozzle could work, although even this engine comes on boost fairly soon so we might need the cooling early and a large nozzle would drown it. A progressive controller again seems like it would work very well, as the second half of the rev range has fairly static requirements.

Overall it seems like an on/off switch can be sized to cover most setups fairly well, and a progressive boost controller would work very well. However there are issues with reducing flow by controlling pump voltage: Lowering the voltage reduces the water pressure, but not linearly. Furthermore it can cause poor atomisation from the nozzle and the cooling effectiveness will be reduced. Finally they can be pretty expensive.

A final setup I'll consider is using a dual stage setup, with a secondary nozzle controlled by a solenoid and rpm switch. Consider the modified engine again, only this time with a 100cc primary and 75cc secondary triggered at 4500rpm.



We get coverage from 3200rpm, with a greater amount of water once we start making real power. Could easily use a 100cc or even 120cc secondary nozzle to get maximum cooling in the upper rev range. Importantly, you can implement this for about $100, much less than a controller.

This is a very simplistic way of looking at things, but probably better than some people come up with for WI setups. In retrospect I should have looked at some larger turbos, as all these spool up fairly quick. However I hope it gives an idea of how to size a water injection system, which you can base on your own dyno chart.

[aus jd 2703]

for all the stress and calculation etc i must ask how much are you saving compraed to a kit?
ive looked into this and i know aem are about $300 all the way to $800 for the HFS-4 aquamist kit which imo is perfect as it uses injector duty cycle.
im ll for diy but im just asking how much are you expecting to save as it may not be as much as you first thought....
and again u need to tune to maximise WI and that means you need failsafes... but above information is very good and your staged set up actually looks like a great idea.

[banana_socks]

It's not just about saving money. Even if you buy a top of the range kit, it still needs to be sized to your engine. You can blindly follow the manufacturers advice, but I prefer to check things myself. Did a little more today, based on the full progressive kits running from injector duty cycle. Perfect you say?

Using a slightly different dyno chart, but similar to the mild engine referenced above. I've skipped to the maximum and minimum water flow rates. Lets assume we want to inject exactly in between (18.25% by weight)



The maximum flow rate required is just over 150cc/min. We don't want to oversize the nozzle, so we'll go with 150cc. Therefore we want the pump flowing 100% between about 5000rpm - 6000rpm and lower either side.

If the controller reduces pump voltage linearly with respect to duty cycle (=fuel flow) it will give a voltage curve in the same shape as the previous charts. Max of 12V for simplicity, as most pump controllers drop it slightly.



Pump output pressure should be linearly related to voltage. Using a max of 150psi, we get the curve below.



Unfortunately the actual flow rate from the nozzle isn't linear related to pressure, but the square root of the difference. If we plot the actual flow rate against the target, you can see the limitation.



Its not terrible, but its hardly much better than a fixed rate nozzle.
Now if the controller actually takes these relationships into account, it is possible to match the flow rate exactly to the target. Alternatively you could use an ECU map to output a signal to achieve the same thing. Working out the pressure curve to achieve the target flow rate gives:



Although the flow rate might be perfect, the pressure becomes pretty pathetic in the lower rev range. At this point you may as well be dripping water into the intake, because its not going to atomise.
Here is where the dual stage setup has an advantage, as you're always running at maximum pressure.

So even if you're using injector duty cycle to measure load, varying flow rate by reducing pump pressure still has its limitations. There are a few variations that look to solve this, but 99% of kits I've seen for sale work this way.
I'm not really recommending anything, didn't know what the outcome would be until I ran these calcs. Just figure the more info the better.

[BLSTIC]

Have you thought of using an air atomising nozzle? Injecting WELL ATOMISED water before your turbo is unlikely to do any harm.

I say that because if you use one of these, the amount of water injected is proportional to boost pressure, and if you use a pulse-width modulated solenoid in the WATER supply line you can control it using one of these two kits Digital Pulse Adjuster or Independant Electronic Boost Controller to program in some RPM proportional control.

*EDIT*
Both of those kits use duty cycle as their only load input. IIRC the main difference is that the "Digital Pulse Adjuster" uses the same frequency (varying with rpm) as the input signal and has a higher resolution (128 points I think), where the boost controller uses a fixed output frequency and has two lower resolution (64 point) maps available for switching on the fly.

*/EDIT*

Also with the air atomising nozzle, if you throttle down the water supply while keeping the air supply constant the water is broken up even better...

Alternatively another approach is to spray finely atomised water across your intake opening. Aiming the spray correctly across the opening will give water induction that is roughly proportional to load. This approach is covered here Autospeed Water Injection System - Part 2

[TERRA Operative]

I would be inclined to inject as close to the intake runners as possible to minimise the chance of the water condensing in the pipework.