Position and placement
 

I agree with most of the preceeding statements. My intention was to have the drip step well back from the turbo inlet. I would be very reluctant to place it right up against the compressor impeller as it could create all sorts of strange effects on the compressor map, moving the surge line around and so on.

This is the mental image I formed. The injection nozzle is placed .5 -.8 meters back from the compressor inlet. Some reasonable distance down stream from that, you flare out the inlet pipe some. This slows airflow speed and encourages any large dropplets to drop out of the flow. After a short distance you began to taper back in, creating a venturi. Liquid water would tend to collect at the point the intake tract begans to converge again. The liquid would be forced to move "uphill" on the converging section. If air speed is low enough at that point it would be impossible for the liquid dropplets to run on the surface. If you place a small lip near the end of this converging section, you would stop any dropplets that try to climb the "ramp" and as they get pushed up over the lip they should get ripped into a fine mist again by the high speed airflow at the throat of the converging section.

A secondary possibility would be to place a small drain at the bottom of the wide point to allow any liquid water to drain away. This may require some form of active extraction, or simply a gravity drain with a vacuum check valve would suffice. For active extraction you could just plumb a simple windshield washer pump with its suction to this low point and its outlet back to the water reservoir or suction side of the WI pump.

For reference, Oakos Automotive on their web site ( http://www.oakos.com/wrx/installs/AEMCAI/index.htm ) did a measured air pressure drop test on a WRX. The stock intake created a 26" of water pressure drop at max flow, the high flow AEM intake reduced that to about 12" of water pressure drop. Any drain system would need to have a check valve so it did not create a vacuum leak post MAF due to the intake pressure drop but allowed the collected liquid to drain away.

My current setup (old version) had the spray injected from the outside of a horizontal 90 degree elbow, with the spray at right angles to the air flow. When I pulled the turbo out that I photographed earlier, I could see clear evidence of liquid water flow on the inside of the intake tract. There were slight but unmistakable markings left by the evaporating water as it ran on the surface. The water mark basically went downward from the jet turning down stream and then followed the bottom center of the intake tract.

In considering the physics of what was happening I remembered some old info I had picked up years ago about water flow in a stream as it enters a turn. The fast moving flow moves to the outside bank and then due to drag with the outside river bank the current turns and moves down toward the river bottom then comes back upward on the inside of the turn toward the surface. In the case of air flow in a duct you would have this same hydraulic flow in the full 3 dimensions of the tube, with the air scrubbing hard against the outside center of the duct at the point of the turn then splitting and 1/2 going up and 1/2 going down (on a horizontal bend) and curling back toward the inside of the bend a few diameters down stream. This sort of flow would carry most of the water dropplets down the bottom center of the duct. In the faint water marks on the inside of my duct I could see exactly this same type of movement. I suspect a large fraction of the liquid drop out occured during the turn on and turn off period of the spray event when atomization is not good, not to mention some after drip that I imagine every nozzle would have.

The most simple solution if you have the room would be to create a P trap intake like on a sink drain where the injection occurs on the downward section of the P and then any liquid water would collect at the bottom as it could not climb the vertical leg of the trap. Air flow would likely quickly evaporate any water collected but it would be a bad thing if a large quantity of liquid were to collect there allowing a water slug to be pulled up the intake during a hard "yump" or other violent change in direction.

I also agree that Pre-compressor injection is best reserved for use only during high airflow situations where the benefits it provides would be best used. My 10 psi turn on was too low and as mentioned above I will move that turn on point up toward peak boost where I need to maximise the compressor effeciency.

FWIW

Larry

Your diagrams still seem to show the water spraying onto the whole face of the turbine. I assume that the water erosion could be avoided by ensuring that the water didn't impinge on the tips of the blades, could this be achieved by having a narrow jet on the center line very close to the turbine? I mean within an inch or two. I'm envisaging something like a pitot tube inserted into the air intake close to the turbine.

On another tack, to stop droplets formed on the walls from hitting the tips of the blades it may be possible to pinch an idea from a Porsche crank scraper. Imagine the air intake as a vertical tube with the turbine at the bottom. Step the wall in just above the turbine, and shape the wall so it forms a gutter. Angle the gutter so it forms a spiral rather than a circle. At the 'lowest' point, put a rib taking the gutter out to the middle of the turbine, still with the gutter profile so that water has to run out to the end of the gutter and can't just drip off the sides. Assuming the air speed is high I think aerodynamic forces would be quite high compared to gravity so the water might be pursuaded to go 'uphill' towards the axis of the turbine (if you see what I mean). From point of view of air flow it would look like a straightening vane with relative little drag.

Just a thought, what do you make of it?

Oh, by the way, can anyone put any numbers to the turbine tip speed and the water droplet speed at the jet? I'm wondering whether there's any mileage in tangential injection ...

http://home.pchome.com.tw/personal/r...essorWheel.jpg

On second look, I don't think I buy this being water damage. What is your filtration? Did you run w/o a filter ever? Lose a nozzle screen ever?

i am no expert, but this looks like grit damage to me. Sorry, just that I have to question if we are chasing our tail.

Larry, you have the same turbo, yes? You had nothing like this, correct?

greenv8s

I am not able to envision what you have described.

But, for it's simplicity, I like the center-mast stream, or narrow cone (preferred) idea. But again, anything located close to the blades will be an airflow disruption, and I think a good rule is to keep at least 2 diameters away from it. Then pray your install never get's loose! For many, there is a bend in the intake close to the blades (considered bad design) and that may limit how close you can get anyway.

If only someone can come up with an easily serviced coaxial alignment. The wall nozzle has that advantage. Someone can do it, to make this idea practical.

I've been using K&N cone filter for years, and never run without it.

I do hope, too, this is not caused by the injection, but now I hesitate.

And I remember Larry had a similar picture here somewhere. The damage was not so severe, though.



Michael,

Why don't you try it yourself & see how it goes? :twisted:
Maybe you'll have better luck :wink:

My pictures
 

My compressor erosion pictures are on page three of this thread. My damage is less severe and I KNOW that some of my damage was due to dirt ingestion during a couple unfiltered drag strip runs.

Yes it is likely it can, you just discribed the setup used by one guy that has been running pre-turbo spray for years. He injects directly on the compressor shaft nut with a narrow jet. The rapidly spinning shaft nut simply blasts the water stream into a fine cloud. I suspect the effect is not dissimilar to a 1000 psi impingment fog nozzle. The advantage of that system is as you say little or no chance to wet the intake tract walls, and the water is ejected tangientially outward inline with the compressor blades so little relative speed difference should exist. This is also very similar to the configuration used in WWII combat aircraft where they injected both fuel and WI "into the eye of the compressor".

The problem with that setup is it is not easily setup on many systems as the available space near the compressor inlet is quite limited in many cars, and service (ie to change jet size) would be quite difficult.

One of the things that may have limited my compressor damage is that I set the system up with a constantly pressurized system and the flow was controlled with a solenoid only about 4 inchs from the jet, so after drip would have been essentially eliminated, and rise time to full spray pressure at the jet should have been very short. Mine was also a continuous spray rather than a modulated spray as you would get with the HSV turning the spray on and off for folks using that style of system.

I also had a manual test spray button that at times I would briefly test the spray at low rpm to confirm my system was working. Based on this discussion, I now consider that to be a bad practice and will avoid it in the future, unless the engine is at a fairly high air demand, so that intake velocity would be likely to keep the plume from heavily wetting the intake walls.

Larry

Would this method work?

http://www.aquamist.co.uk/forum/sleeve3.gif

Nice Richard,

Would it "splice" into the conduit? Can the plumbing be routed within the support structure?

Haven't thought that far yet. It wil be quiet a simple fit, just slice it between the inlet hose and turbo flange, possibly with a few slots on the ring so it can be tightened up.

Do you think it could be used on different size ducting?

For universality (all diameters) and ease of maintenance:

how about a device that can be inserted in a standard size hole drilled into the conduit? Insert, and then have a band clamp of sorts on the opposite side that would pull it tight around the periphery, for a seal. Would also minimize drag interference I think.