Port WI build thread
 

So there have been some good discussions in several threads here regarding nozzle placement, and single nozzle versus port injection WI. Long story short, I strongly suspect water distribution between the cylinders led to the detonation of my #1 cylinder to the point where I had to pull the motor and replace it.

Tired of screwing around with this stuff I decided to spend a little extra money and build a port injection setup. It uses four 2gpm DO external nozzles, and three 2 into 1 DO fittings to split the flow. There are four DO 90 deg elbows on top of each of the nozzles. FWIW I like all of the DO fittings; the elbows swivel even when everything is tight, and the flow splitters are lightweight. They all have the push-lock style tube inserts. I am also adding an Aquamist high speed valve (HSV) that will be PWM controlled by the Adaptronic.

I drilled and tapped a hole in each of the four runners below the upper-to-lower intake manifold flange and mounted the nozzles there. On a related note I completely gutted the upper half of the manifold, and opened up the lower and removed the VICS system. More about that here:
https://www.miataturbo.net/forum/t38809/

Gutted upper:
http://lh3.ggpht.com/_jh2HiMm4nKQ/Sq...0/P9030193.JPG
Ported lower, with no VICS
http://lh5.ggpht.com/_jh2HiMm4nKQ/Sq...0/P9030194.JPG

Now for some initial pictures. The engine is out of the car right now and on an engine stand. The first photo shows a shot of the back of the intake manifold. Notice how the nozzles and fittings are below a horizontal plane tangent with the top of the intake manifold. The engine was more or less level when I took the picture. In short, the nozzles won't hit the hood.

http://lh4.ggpht.com/_jh2HiMm4nKQ/S2...0/P1310162.JPG

This photo shows the location of the four nozzles from below, looking up at the intake manifold.
http://lh3.ggpht.com/_jh2HiMm4nKQ/S2...0/P1310164.JPG

Lastly this shows the tubing 'manifold' that splits the inlet into four flows. From right to left, and in the direction of flow, it is a 1-into-2-into-4 setup. Before someone asks, note that the flow rates (and hence flow velocities) are very, very low, and the difference in the length of the tubing will almost certainly not create enough pressure drop to significantly imbalance the distribution. We are talking about 2gph (or .00056 gallons per second) through each of the approx. 5mm i.d. hoses. With the DO fittings this took all of about 10 minutes to build once the nozzles were installed.

http://lh3.ggpht.com/_jh2HiMm4nKQ/S2...0/P1310163.JPG

Ideally I would have spot-faced the point where the nozzles go into the manifold so the o-rings would have a better sealing surface. I decided not to do this for a couple of reasons:
- One, the manifold is pretty thin here and I did not want to give up any thickness to a spot face. There are only three or four whole threads in the tapped hole as it is.
- Two, I think it will seal fine anyway, and I put a little gasket sealant on the threads as insurance. If it leaks I'll beef things up by welding tapped bungs on the intake. But if this works, then everyone will know that welding bungs onto the manifold is not required. Not everyone, like I, have access to a nice TIG for aluminum welding (a buddy of mine has one).

More to follow as I get things going.

It looks beefy but the controll factor of it is what i wonder about. you are looking at about 340 or so whp on this setup ultimatley?

perfect nozzle placement imo

thats about how my DP nitrous was (before i was turbo of course)

Attachment 200813


one thought though, have you considered any possibility of uneven distribution due to different length tubing coming from the distribution blocks?
Or do you think that the difference will be insignificant?

Finally, nice to see someone taking this subject seriously.

Yeah tops. I think I need to turn it down a bit and keep it under 300 since it is a DD. I am growing weary of damaged engines.

Unless I am missing something I think the control will be good. I have a few options with the Adaptronic and the HSV. For example, I could scale the PWM DC based on boost. Or I could scale the DC with the fuel injector duty cycle, and have it come on at a set boost threshold. In other words the fuel and water could be held at an approximately fixed ratio. IIRC this is somewhat ideal.

But if you think I am overlooking something though let me know. Input is welcome.

Nice setup!

Seriously? If you are, so much for my preemptive answer. Read my post again :)

ahhh, that was a fail on my part, haha

I still have the holes in my manifold from the nitrous and Ive considered going to DP W/I route with them. Im looking forward to seeing how it works out for you.

Scaling WI to fuel PW is pretty much ideal. All the data I've seen shows that for a constant ratio of water to fuel, the anti-knock index is reasonably linear.

The one thing I'd be cautious of is the flow rate through the HSV. It has a fairly narrow orifice. At least, the old one did.

Curiously, it looks like Aquamist has redesigned their whole product line. Not much useful tech into at the website, but I see a new controller, a new "Fast Acting Valve", a switch to ShurFlow style pumps, etc.


More info:
Waterinjection :: View topic - Aquamist HFS-3 system for 2010 ..... "Q and A"

http://www.howertonengineering.com/Aquamist_hfs3.html

http://www.howertonengineering.com/Aquamist_hfs2.html

http://www.aquamist.co.uk/HFS6/HFS-6-3w.pdf

Yep that is what I remember and it sounds reasonable.

I should probably measure the flow rates at the nozzles and look at the atomization. You are right though, even just the fittings to get fluid in and out of the valve have very small internal diameters. They are still large though compared to the nozzle orifices.

I had not seen that yet. You are right the new valve looks completely different. It does retain one feature of the old valve; it is still expensive.

The only thing that scares me about port injection is one of the ports clogging and you never knowing it til that cylinder blows. A flow meter won't show one port clogged.

The fittings don't concern me. They're huge compared to a nozzle.

One thing that I do wonder about is whether, in a system using an HSV and many nozzles, the relatively great length of tubing between the HSV and the nozzles might act as a dampener. Actually, I can see good and bad here. It'd smooth out pulsations from the valve, but might also decrease the accuracy with which you can control flow. I've noted, for instance, that after I de-power a pump which is connected to a nozzle with maybe 3 or 4 feet of tubing, the nozzle will continue to spray for perhaps a second.

Bench testing of this configuration will be a good thing.


Yeah, scares the willies out of me, too. One HSV and one flowmeter per nozzle would cover you there. What was that about a single one being expensive?

Yeah I have wondered the same thing for the same reasons. I'll try to bench test it with the HSV. However I do not have a function generator or something like that to trigger the HSV and simulate the PWM at the frequency the Adaptronic will use. Any suggestions on a cheap approach? It has the same impedance as a high-impedance injector.

So I talked about this in another thread on a related discussion. Here is what I am thinking. This setup is essentially somewhat redundant. The nozzles are partially spraying into their respective runner, but also spraying into the now completely open and large plenum formed by the gutted upper intake manifold. In fact, it is probably on the order of a 50%/50% distribution. The upper part of the 'spray cone' is going into the plenum volume, and the lower half is going into the runner volume. Thus, if a nozzle fails, the runners adjacent to it should provide a fair amount of WI into the runner with the failed nozzle. They only draw one at a time after all, loosely speaking.

In other words, there should always be a good supply of mist in the upper intake volume that is available to all of the cylinders. That provides some redundancy in case one valve clogs. Mist will still make it into that runner. This is one advantage of positioning the nozzles where I did, rather than down at the other end of the runners close to the head.

This whole redundancy scheme was my main motivation for gutting the manifold. Improved flow, which is debated elsewhere, was somewhat a secondary consideration.

Whoa Aquamist switched to the Shurflo piglet ??!!

I don't think it's a Shurflo brand, but yeah, they're finally eating crow on that issue. IIRC, they phased in that pump over a year ago on an entry level system. I didn't realize they'd switched over to it completely until yesterday.

That is a big deal since IIRC they made a big point of talking about how superior their pumps were over the shurflo type.

Exactly.
Are the newer pumps cheaper?
Is the problem of the old one reliability?

I can't believe they switched over to the same cheap pump the ebay kits use, when they could be using one of these:
https://www.miataturbo.net/forum/t20899/

The Shurflo pumps are optimized for the wrong thing. They are high volume high pressure pumps; a WI pump can be a lot smaller because they are LOW volume high pressure. The frakkin pump needs to be no bigger than a *&^ fuel pump.

Sheesh.

I disagree. Each nozzle has a given flow limit and a relatively known flow curve vs pressure, even though it may not be published. Assuming the flow sensor provides actual feedback in Pulses Per Liter you would be immediately able to tell if there is a change in output from a single clogged nozzle. Your controller just needs to be "smart".

When choosing a part for a kit, do you use something that is proven and cheap or something that is more expensive and unproven? They are using an aquaflo pump, its pretty much the standard for all water injection pumps on the market.

auqaflo pump? Is that a Shurflo pump?

My beef with the Shurflo is its sheer SIZE.
Why couldn't they have developed something better that is only a bit pricier than the Shurflo? They went thru the trouble of designing their older pump...

Sorry, meant Aquatec. AQUATEC WATER SYSTEMS INC. Precision Diaphragm Pumps, Flow Control Components, Water Pumps, Engineering Expertise I agree they are big, but from a design and engineering perspective its a simple choice. It works, its proven, its effective. Sure you might not like the size, but 99.9% of customers don't care. Why spend 2x as much to import or custom make a pump?

How are you handling failsafes?

So flow monitoring (Neo) and failsafes (Sav).

With PWM the flow would have to be monitored and compared to the PWM input via a lookup table, or a curve fit equation, or something like that, with a deadband for normal variation/error. It would have the usual tradeoff between sensitivity and false-alarms. I would like to do that but I do not think the Adaptronic is set up to read in an input (flow meter) and compare it to an output (PWM duty cycle), at least not to that level of sophistication. I also do not want to add another controller.

Failsafes. So from this previous thread...
https://www.miataturbo.net/forum/t38494/
- I'm already implementing #4, that is using multiple nozzles that are sharing a common plenum.
- I will probably still do #2, that is, use a switch at the outlet of the pump to monitor pump pressure and add fuel or retard timing if the WI is commanded on and the switch does not change state.

#4 protects somewhat against a clogged nozzle, but not against a dead pump, dry tank, or a catastrophic leak. #2 is the opposite. It protects against a dead pump, dry tank, or a leak, but does not protect against a clog. The two are complimentary. It is not a foolproof scheme of course, but it is relatively cheap and easy to implement.

Also, as extra insurance, I could dyno tune with the WI off, then add the WI on top of that as extra protection.

So, seriously, any EEs or Techs out there got an idea what I can use for a cheap PWM generator to bench test the HSV? I would think a waveform generator would do it as long as it can produce 12V on a 12 ohm load? I could get one from work, maybe, but do not want to take a chance on damaging something expensive if I can get something cheap on my own.

You can buy old function generators on ebay for like $40.

Can I run them at a fixed frequency with a square wave, and vary the width (time) of the pulse? Pardon the ignorance but function generators are one piece of bench top equipment I have not used before, at least not enough to remember anything about the specifics.

What about doing something really klugy, like using a waveform generator on a PC (like some freeware app) and using the headphone output to drive a power amp, and adjust amplitude (gain, volume, whatever) until I hit the right 12V-14V peak voltage? An audio amp is made for about the same resistive load, and is capable of higher peak voltages than what I need, and can provide enough power. Crazy like a fox, or just crazy? I have all the crap I need on hand to put that together.

Yes, a function generator is designed to do EXACTLY what you just asked. Freq, Duty, wave type, and amplitude can all be set independently. You can get an old school one with dials for cheap. Def something to have around when designing/messing with analog and digital circuits. I recommend just buying an older used one, they last forever and its nice to have one on your bench.

Smaller PUMPS:

$55
Chris Coffee Service - Ulka Vibratory Pumps EAX5 52 Watt

$64
120V Ulka Vibratory Pump - Espresso Machine Vibe Pump - Espresso Parts

Pump curves:
http://images.google.com/imgres?imgu...%3D42%26um%3D1

Neither says anything about flow rate or compatibility with methanol.

http://www.ulka.it/admin/moduli/m003...g_ULKA_web.pdf

Some datasheetage. Says suitable primarily for water, my guess is that it doesn't have viton or buna-n orings. Another issue is that for the pressures you want >=1Mbar-ish it maxes at like 200-250 cc/min. So for this application and power levels it might be fine. Over that you need more than 1 pump (another shortfall of the old aquamist pump). Also you have no built in method of maintaining a constant line pressure, you would need some sort of super high pressure regulator with a tank return.

So at 4 bars pressure the E5 will deliver 400 cc/min.
That's about 20% the fuel flow rate at 300 hp / 11:1 AFR.

What water flow rate do you need and at what pressure?

All mainstream WI kits operate at at least 150PSI in order to get good atomization. 4bar is like 60PSI. Subtract boost pressure from that and you have actual pressure. That assumes that you have a high speed valve. If you use a check valve then you can subract that opening pressure as well. So 4bar is suddenly not adequate. So your small pump becomes an issue for the following reasons. A single pump is likely not adequate for high HP, the flow rate at high pressure is poor, you must have a special high voltage power supply, you must devise a custom pressure regulator/fluid return, and possibly/probably incompatible with methanol.

Suddenly the aquatec/shurflo pump becomes attractive because it has all these issues solved.

Plus I already have one :D It takes up more space than I like but it is in place and operational.

More failsafe thoughts. If I am unable to set up #2 above in logic on the Adaptronic, or for those who do not have an Adaptronic or the like, there is another simpler option. Not a failsafe, but a warning light. Tap off the pump power (at the relay) and send it to a pressure switch at the pump outlet. Wire 12V to the common lug and wire the normally closed (NC) side to one terminal of a dash mounted warning light. Ground the other light terminal.

When the pump is powered on, the dash light will light up and remain lit unless the outlet pressure breaks the NC circuit. It could be kind of annoying though. It would come on for a brief instant every time the WI operates.

The same scheme could be used to trigger the alternate map input on the MS, or on the Adaptronic for that matter.

Pressure switch and a small accumulator.

Why do you need a fixed pressure anyway? It's got a flow curve, your nozzles have a flow curve, they meet somewhere. Apply power to pump when you need flow. Adjust flow with your HSV.

And, without a pressure regulator, if 1 of your 4 nozzles clogs, the increased pump pressure will force a bit more flow from the other 3, helping your failsafe.

Why do you need super atomization? Even with less atomization, the water will still absorb heat during combustion.

Aren't those Shurflo piglets are only good for 60 psi?

You know I thought about that, though I am not sure that is true. I think the flow rate is so low, that the shurflow pump is effectively dead-heading or running at max pressure most or all of the time anyway, so a clogged nozzle would not create that much more pressure in the system. Thus there would not be an appreciable rise in flow at the other nozzles. Right?

One way to know for sure, test it. Along those lines, I found this shareware PWM generator by the way. It is free for 30 days.
Order - PWM Generator - The PWM Generator can generate PWM signals in real time.You can save 4 memory presets and the software supports multiple sound cards.

Maybe I can hook that up to an audio amp and I'll have my function generator. I sure would be pissed though if I fried the HSV screwing around with this.

yeah the pressure will pulse in a predictable range unless you experience a nozzle clog. In your position to limit your water usage you should probly scale your pwn usage at the 40% + point becouse since you are IC'ed etc you would just be wasting water spraying at anything less.

Smaller droplets = more surface area. More surface area = better vaporization.

Pretty much all nozzles you see for sale for WI are rated at 100PSI. Apply less pressure than that and you get less flow and larger droplets. The point of the HSV is obviously to control flow. You want to maintain a constant pressure so that you have a consistent droplet size and therefore consistent vaporization of the fluid into the air. Progressive systems that do not employ a HSV control output through pump speed. They are not the best because they have inconsistent pressure (low pressure at low output). The best system would constantly have high pressure, even at low flow, and control flow with a high speed valve like the aquamist kits.

The entire point of a good system is performance and consistency. DIY is all well and good, but you still want to achieve the same results.


If you think that increased pressure isn't beneficial I will go ahead and say that you are COMPLETELY wrong. There is a reason that F1 fuel injectors operate at 4+ MPa (aka 40+ Bar). They realized long ago that super small droplets increase their detonation resistance and power substantially.

the sureflow pumps are stock 140 psi my snowperformance stuff was rated at 70 psi. Any way you cut it though you can just do the math to figure out your flow per any given pressure situation. just like we do with our injectors.

1) Why wouldn't the water give up its latent heat of vaporization eventually, whether it's before it enters the cylinders from fine atomization, or from the heat of compression if the droplets are large? Detonation typically happens due to auto-ignition of end gases; by that time even larger droplets will have vaporized, cooling the charge. Do you have any links to papers that say otherwise?

2) The big Shurflo pump will not change its outlet pressure much if 1 of 4 nozzles clog. The smaller pumps will; thus if you size the nozzles for proper flow with a smaller pump, and 1 clogs, the pressure will rise and the water delivered by the other 3 will rise.

3) How fast does the HSV pulse? How far is the HSV from your nozzles? How much compliance do the lines after the HSV have? I think your lines are somewhat flexible, between the HSV and the nozzles. Thus any rapid pulsing will be lopass filtered - i.e. the nozzles will not show pulsed flow. The nozzles will not see pulse pressure behind it. Instead, the nozzles will see some constant, reduced *pressure* when the HSV is pulsing. There goes your fine atomization. There's a reason injector pintles open right at the tip and not upstream.

How fast does your HSV pulse relative to the injector pulses and are they sync'ed? If they are sync'ed, how does the dynamic distribution work out even if the nozzle has pulsing output? The air entering the cylinders is pulsed.

I think you have the right idea having 4 nozzles, but the nozzles will all have some relatively constant flow proportional to the HSV duty cycle.

Having said that, even with effectively variable pressure driving the nozzles, I would think you'd want at least 15 psi head; if you had a max of 60 psi available, a 4:1 ratio, that only gives you 2:1 available ratio of flow.

Looks like the Shurflo pumps are regulated at 70 psi. .. ??

What he said.

they are fully adjustable from 0-140 psi with some going up to 200 psi

I measured mine a few months ago. It's making about 150 PSI, and has been quite happy with this for years.

Actually, yes: http://ntrs.nasa.gov/archive/nasa/ca...1993093245.pdf

Summary: this is a NACA paper which documents the anti-knock effect of atomized (but not vaporized) water in the combustion end-zone. In this particular study, they've cut right to the chase and are injecting water directly into the combustion chamber late in the compression cycle. In other words, the idea that water injection suppresses knock by evaporatively cooling the intake air during its trip through the intake tract is refuted by this study.

Obviously, good atomization is still required.




Not really regulated, per se. They have a simple on/off switch which opens the circuit when pressure rises above a certain level. The activation point is generally adjustable via a screw in the top of the pump head.

If injecting air late in the compression cycle works, why wouldn't poor atomization in the intake tract work too, given that those large droplets would still vaporize in the compression cycle?

Agreed, but be careful here about all-or-nothing reasoning. The pulsing will be dampened by the tubing, sure. But I doubt it will end up being completely damped to the point where it is an average, drastically reduced pressure. Even when I had a single, much larger nozzle on a much longer run of tubing, it was clear that the mist at the nozzle was still pulsing strongly in response to the pump cycling on and off. Anyway, testing will tell. At the worst, there could be some minimum duty cycle below which atomization is not OK, and so I just set the onset duty cycle above that value. Do I really need to flow down to 1.532% duty cycle? Nope. Starting with 20% would be fine, and it is still much better than an on-off system.

No the valve is run at a set frequency with PWM. It is not synced to the inlet pulse. I doubt this is significant, especially since the duty cycle of the spray is essentially 100%. It is the same deal as with batch fired ignition. Batch firing works fine and almost exactly as well as sequential at high injector duty cycles (max power) because the injector is open for almost the entire cycle of the cylinder anyway. Plus the open plenum in the upper will act as a mist accumulator and further negate the need for inlet pulse sync. Strictly speaking I see your point, but I think it is a negligible problem.

This is true regardless if it is a single or multi nozzle system. Flow will be proportional to the HSV cycle, that is the whole point. The HSV is not intended to shut the nozzle flow off completely between pulses. It is just a PWM flow controller that can operate at a high enough frequency to be useful for this application.

All these points are not unique to a multi nozzle setup.

Empirical data shows that better atomization controls detonation. In fact development of good port fuel injection is one of the reasons compression ratios could be raised again in the late 80s-early 90s, after plummeting in the 70s. All else being equal, the finer atomization in the cylinder made the engine less detonation prone, and compression ratios could be raised.

Also, things are happening very quickly on the compression stroke. To illustrate by extremes, think of it this way. If all of the droplets were incredibly atomized to the point where they were molecules, then evaporation is instant (it already happened). If atomization was horrible, and you ended up with one 'big' water drop in the cylinder, then it is at least possible that there is not enough time during the compression stroke to completely evaporate that drop and distribute it evenly around the cylinder.

Speaking of distribution, that is important too. Smaller, finely atomized fluids are going to distribute more evenly around the cylinder.

Anyway, this is all academic. It has been shown to be important like Joe pointed out, and with years and years of good test data (the ultimate indicator), that atomization is important. I am not sure why I am even bothering to spend this much time writing about it honestly :)

Answer:
Ok, moving on.


Yeah, I'd not want to pin it down too hard, but definitely one contributing factor (along with better mixture control, more stable and accurate ignition control, refinements in combustion chamber design, etc.)


As much as I hate to disagree with you Jason (you seem like the kind of guy who does his homework, so to speak,) I just don't like the "big droplets are just as good as small droplets" theory. I have seen no data at all to support this observation, I have seen considerable anecdotal evidence to refute it, and it seems highly counterintuitive to me. Large droplets are more likely to come out of suspension as they impact the walls of the intake manifold, ports, valves, etc. Yeah, they will eventually vaporize, but that's not what we want- we want finely atomized water present inside the combustion chamber during the compression cycle in the 50-60° leading up to the ignition event. If the water has already vaporized by that point, it's not being used to its fullest potential.


In the '80s, the OEMs went from big carburetors on crappy manifolds to TBI fuel injectors on crappy manifolds.

In the '90s, the OEMs moved the injectors out of the manifold and into the head, vastly reducing the proclivity of the mixture to become inhomogeneous and virtually eliminating the potential for uneven distribution from one cylinder to the next.

In the '10s, the OEMs are moving the injectors directly into the combustion chamber.

The OEMs may be onto something here.


There are loads and loads of good scientific material on this subject, both contemporary and historical. The paper I linked to is just one of many that I have. I'd been meaning for quite some time to do a sort of thesis on the subject, though to be honest I sort of lost my motivation when I found out that CA puts OBD-II cars on the rollers. (It's a very long story.)

I still intend to do this... At some point.




Because that is the Way of the Internet.

I don't understand why you are putting all this effort into safety when you have no idea how your gutted manifokd flows air. You could easily be lean in a cylinder.

If you mean uneven air charge, maybe.

Some have gone before me with the gutted manifold and as far as I know they are not having lean cylinder issues. The only debate seemed to be whether or not the flow gains were worthwhile. Of course, they could have a problem and not know it. I think the only way to know for sure would be to run individual runner EGT gauges right at the exhaust ports, right? I think it would be too hot for individual O2 sensors. Either that or do an old-school WOT ignition/fuel cut on the dyno (with no WI) and compare the spark plugs.

At any rate, I think the open plenum into individual runner design is pretty similar to many other OEM designs. It is also similar to several of the successful one-off manifolds I have seen here, albeit they have shorter runners. Plus I would think the larger plenum volume would tend to even out the distribution, not make it worse? The air should stagnate quite a bit in the plenum and help even out the pressure gradient across the volume. Maybe there could be some resonant effects...

Oh yeah +1 to what Joe said above. Not trying to be difficult, just saying the same thing Joe did. Improved atomization has been shown to improve combustion and reduce detonation/knock.

Yes, you'd only know with individual EGT sensors. Just because other people have done it and aren't noticing problems doesn't mean that there a) aren't problems and b) won't be problems down the road. They are also probably running less boost and boosting much mroe infrequently than you plan to. That means problems could occur much more rapidly.

I understand that a pressurized plenum will flow differently than one on an NA engine but you are still taking pretty big risks IMO. There are some threads here with computer modeled flow (I forget the name of the program) and manifolds that look pretty darn good flow like crap in the simulators. I don't know how much work mazda put into making their OEM manifolds flow evenly, but they probably at least sorta tried right?

Nyways, instead of putting a have-at-ye-with-a-dremel- manifold on your nice shiny built engine I would take the time/money to find a properly flow bench tested manifold.

All good points. However I think it would be prudent to have THIS manifold flow tested before deciding it is inadequate (and dropping big bucks and/or big time on a custom-fabricated, fully analyzed and tested manifold). I still suspect this might be overthinking things a bit. Do you know of example of where someone did something similar to this and damaged an engine from flow imbalance?

Anyway, let's get back to the WI discussion.

For the failsafes, could you use some kind of (NC) switched one-way valve instead of flow meters?

I'm not current on low flow rate meters, but for the kind of flow we're talking about I don't know how quickly they'd react and let you know a nozzle has taken a dump.

I'm thinking of a switch that is wired off the pump signal wire, that grounds a big warning light if both: A.) pump is on and B.) check valve is closed. You could use these instead of the regular check valves (which you need anyhow).

Just throwing that out there...open to riducule. I had drawn a diagram of it, but can't upload at work.

Hang on, that's for fuel. Does the paper you posted (I haven't read it yet, it's huge), show that to be true for water injection?

I think Joe's paper was in regards to water injection? Joe I think you linked to the wrong paper BTW :)

At any rate, from what I remember from a discussion I had with an R&D Gasoline Engine Engineer buddy of mine, the same thing applies to WI as well. It has to do with propagation speed of the flame front and evenness of the combustion. The WI well suspended in the charge effectively cools the oncoming flame front and thus slows down the flame front propagation speed, and the better suspended it is, the better/faster it absorbs energy from the flame front. I think it is due to the increase in surface area with the finer droplets.

BTW, IIRC, I do not think the WI droplets are all evaporated by the compression stroke. Neither is the fuel for that matter. Either it is not hot enough, or the increase in pressure lowers the evap temp, or there is not enough time for evaporation, or a combo of these.

When we start talking about really detailed gas engine combustion physics/behavior though, I am getting out of my area of Engineering expertise. My buddy knows a lot more than I do since it is HIS job. I pick his brain for info he can share with me (non-proprietary) on a regular basis.

Anyway, go find a reputable paper/test that says it is not true. Until then, let's assume atomization is important. Please.

I think I am following you here. So by check valve, you are talking about a on-off switch that is triggered by flow? If it was adjustable and repeatable that could work nicely for an on-off system.

I am also not aware of any small, low cost, accurate, low flow meters that could work for this, at least not last time I looked (albeit briefly). There are of course fantastic flow meters out there that will measure low flow rates with ridiculous accuracy, but they are not cheap. You can dig around here to see what I mean.
Flow and Level

I find it believable that finer atomization is better, but I want to question it.
Seeing that there are so many papers on WI, there might be something that actually tests it.

And read what I wrote about HSV + long hoses = reduced pressure at nozzle, and atomization goes out the window.

I don't see how thats possible. The restriction that increases pressure is the nozzle, not the valve.

The valve pulses on and off. Between the valve and the nozzles is rubbery expandable hose. If that hose has enough compliance, the hose will absorb pressure pulses, and at the nozzles, all you have is some average, lower pressure isntead of the pulsed pressure you want in order to get pulsed flow.

Note how Joe mentioned that when he turns off the flow, it takes a second for the flow at the nozzle to fade away.

The way to prevent this is to have some short, rigid piping from the HSV to the nozzles.

Try it - pull one nozzle out and run the HSV at say 50% duty cycle at say 5 Hz and observe the flow.
Come back and tell me who's yo daddy! ;)

Assuming Aquamist uses the same polypropylene(polyethylene?poly-something) hose that everyone else uses I doubt it expands outward much at all. It is rated to something like 450psi working pressure, and over 1000 burst. It is pretty rigid, it just has good bending properties so that it doesn't kink. The reason water continues to come out is because the water in the hose after the valve will drain out from suction and/or gravity, pressure equalizing to atmospheric... So granted, you will have lower pressure after the valve shuts off. When using a true HSV that cycles at something like 200Hz though, I highly doubt that the pressure will drop considerably between cycles. In the end though even Aq recommends keeping the length of hose between the valve and nozzle as short as possible.

I already read it, and responded. Go back and read THAT. Simply put, no, and even if it is, which is is not, so what? And, it is not unique to my setup. And I am not running the valve at only 5 Hz so that is pointless. I might as well try your experiment at 0.5 Hz, in which case I agree, yes, it will have low flow at some point.

So I had my R&D gas engine Engineer buddy take a look at the manifold and he does not see any obvious problems with the gutting that would cause a flow imbalance between cylinders.

Good news on the manifold. I'd just go for it, despite the naysayers. I forget what you're using for knock protection...consider trying Jeff's new device?

On the one-way(check) valve switch...you've got the right idea, I've just yet to find one in 12V. Tried Grainger and McMaster...need to look around more. If we find one I think I'm going to order 20 of them as they could be applied to ANY setup. You'd just need 4, and a lot of wiring, for yours.