Indepth study of WAI injection systems...
 

It is my attempt of explaining how different type of systems work and its advantage and dis-advantage, based on current systems offered. It is a my view and findings, please do chime in and discuss.

1) Single-stage
2) Two-stage
3) Progressive Pump Speed system (PPS)
4) PWM Valve controlled system
5) System that will integrate well with third party controllers.
6) Direct port



The single stage WAI (water alcohol injection), as the heading implies, is not as basic as most people expects. It some cases, it will out perform a two-dimension progressive system. Please do not underestimate it. I will try to explain briefly why after the next few paragraphs.

Having a single trigger point and a fixed flow rate, one will get to know its effect on your engine very quickly. Due to its consistent repeatability, it is very easy to tune. This type of system is normally set to start spray in the peak torque region, where the engine is most likely to knock.

As the RPM climbs, the ratio of water to mass air tends to decrease. This may not be a bad thing because the tendency to knock is also lessen as the wastegate starts to open and prevent the boost pressure from increasing further. The volumetric efficiency of the engine also decreases as RPM climbs, breathing in less air. This also has the effect of reducing the engine's tendency to knock, demand of WAI flow is less. Unfortunately some engines do require continuous WAI flow at higher RPM due to heat build up through friction and turbo efficiency.

A 2-D pump speed system based on manifold pressure is a little bit tricky to tune compared to the single stage system. The user has to set the start and finish pressure points, those points are sometimes set at a considerable distance apart. Matching those operation points in a 3-D environment such as RPM/Boost ramp (nonlinear) is quite difficult. We will be discussing it in more details later.

FOR:
1) Low cost, simple and dependable.
2) Easy to tune
3) Very effective on a stock factory set up with a few pounds of boost extra.

AGAINST:
1) Dynamic operating range is narrow, may not be as effective on a high RPM knock suppression.
2) For high power/ high % alcohol applications, considerable fuel has to be taken out (boost clamp) to make the afr tolerable. Some sort of failsafe mechanism is necessary to prevent engine destruction when the WAI fails to delivery the correct flow.


At present, adding a second manifold pressure switch to active an additional solenoid valve at a higher manifold pressure is the definition of a 2-stage system.
This arrangement gives the system greater flexibility as well as extending the flow range. It addresses the problem associated with the single stage system, too much flow at the start and not enough when RPM climbs beyond the wastegate setting.

As the system is based on boost trigger, it still won?t address the RPM related flow. For a turbo charge engine, the most significant active regions are the boost ramping stage and engine?s maximum torque range. A two-stage system fits these two regions nicely, allowing the some form of cooling demand during the ramp-up stage. The second stage provides the in-cylinder cooling and knock suppression as the engine is under the most stress or highest BMEP (Brake Mean Effective Pressure).

FOR:
1) Relatively low cost to give mark improvement to the single-stage system.
2) Provides well defined triggering points during the boost cycle.
3) Minimising the under/over flow problem.

AGAINST:
1) Trigger points requires some time to set up.
2) Triggering points may differ on each gear if you have a fast spool up turbo
3) Require a bit more care during tuning

END
----------------------------------------------------------------------------------------------------------------------------------------------------------------

Does the pump speed controller perform better than a two-stage system, you are about to find out.

Changing pump speed merely put more pressure behind a nozzle, hence more flow. This type of system is commonly known as a progressive system (pump-speed).

Let us examine how much a M5 nozzle will flow between 40psi to 160psi. According the chart below (Published by Hago, a well know US oil heater nozzle manufacturer), the flow starts from 200cc/min and ends at 400cc/min., when pressure is increased from 40psi to 160 psi.



Almost all PWM pump controller on the market uses Shurflo pump, designed to operate between 0-150psi. The heart of the system is an electronic motor speed controller, vary the speed according to a sensor. It could be a MAP sensor, a MAF or any sensors that read engine load. It is normally a 2-dimensional system. A manifold-pressure type system does not take into account of any RPM change.

A swirling type atomising nozzle requires a head pressure of at least 30psi to produce a decent mist. Droplet size is very important to the inlet cooling ability and even cylinder distribution. Let say the system pressure starts at 40psi (as shown on the chart) and ends at 160psi. One can assume you will get a 4x flow range? In practice, not so, according to the chart, you will only get a flow change from 200cc/min to 400cc.min (see M5) instead of 200cc/min to 800cc/min. Flow/pressure obeys the square-root law.


Being "progressive" implies a reasonable dynamic range between start and finish. How progressive? Almost no one ever questions this. Most people just assumes it covers all the flow requirement between 10psi to 20 psi of boost once the range-dials are set on their pump speed controller. In practice, you cannot expect the same M5 nozzle will serve a wider operating range between 5-25psi by merely changing the dial, the range is governed by the law of physics and not a technically advanced motor speed controller.

If one would want to delve deeper into the subject, as the title demands. So the subject will continue?

Just to recap, good dynamic range (pressure/flow span) is the main factor one should expect from a "progressive" WIA system. Let see what a 150psi system can really offer. We shall take into account of the effect of manifold pressure, inline checkvalve as well as minimum pressure for a good atomised spray.

For example:
1) Manifold pressure start: 10psi
2) Manifold pressure ends: 20psi
3) Inline checkvalve crack pressure: 30psi
4) Minimum pressure of the atomising nozzle: 30psi.

When the system starts: it will instantly see an initial back-pressure of 60psi and a final back-pressure of 70psi (extra manifold pressure). The actual dynamic pressure range is now from 60psi to 110psi. The system can now only manage a 35% change in flow, far from one would imagine a 150psi pump system should perform.



There are other factors that could also affect the performance of the 2-D progressive pump system. It may be a subject for a later discussion, depending on the interest of the readers. Chart below is a predicted performance of a progressive system compared to a single and two-stage system. I hope there will be people chiming in to add to this. At first glance, it doesn't appear there is a distinct advantage for adding a progressive controller. Adding a bigger nozzle doesn't alter the dynamic range, it just shifts the whole curve higher.



FOR:
1) Easy to set the start and end point.
2) Some correlation between manifold pressure and flow
3) Cost effective.

AGAINST:
1) Limited dynamic range, system becomes less effective after wastegate pressure. (see addendum)
2) Extra cost can easily be spent on a higher performance two-stage system with greater dynamic range .
3) If the 2-D system is used to replace high % fuel with alcohol, re-mapping the 3-D fuel map will be very difficult due to the wide dynamic flow range demanded by the engine.
4) Pulsing due to demand switch ~20psi ripple. (some system by-pass this switch, but risking system pressure beyond design limits). May require further explaination
5) Response time due to inertia of a rotating - laggy (start) and over-run (stop). May require further explaination

END
-----------------------------------------------------------------------------------------------------------------------------------------------------------

These systems require a stable system pressure, normally held between 100-125psi. An inline solenoid valve and a PWM controller that modulates the opening and shutting time to meter flow. Before getting too deeply into the subject, note that there are two types of inline solenoid valves on offer.



This type of system resembles the modern automotive fuel injection system. The system can also be controlled by a third party EMS with a spare PWM channel. Delivery rate can either be mapped or mirroring the fuel injector duty cycle. The latter makes tuning very simple.

The valve behaves similar to an on/off gated button on a garden hose. The longer the gate is opened, the more the flow (duration). Alternative, rapid opening/shutting the gate per second (frequency) also control the flow. The common EMS uses duration for load change and frequency for RPM change. The dynamic flow range is extremely wide, 100:1 is normal.

A WAI valve should closely match the closing and shutting characteristic of a fuel injector. This is important for fuel flow mirroring algorithm since the modern EMS has a correction stage to compensate the opening delay and shutting delay.

This type of valve resembles the action of a rotary water tap. As more current is applied to the valve coil, the valve opens more. It is a very nice way to control flow.
There are a few minor problems associated with this type of valve: Atomisation at low flow and lift variations (hystersis of the magnetic circuit), approximate +/- 10-15% flow deviations.

All in all, it will deliver liquid well compared to the PPS system. There are some similarities between the two. The nozzle tip pressure is directly proportional to the flow. This is because the proportional valve acts like a variable restrictor upstream of the nozzle tip. Resulting in: restricted flow = low pressure. Low pressure = low atomisation.

NOTE: This type of valve has a typical open time of 4ms and closing time of 4ms at its designed voltage. It is too slow to be used as a true PWM valve. Open/close speed can be increased by over-voltage pulses. The design of spider valve is also vastly different from the PWM valve.


Summary:
It is important to know some basic facts between the Proportional valve and PWM valve systems before choosing this type of system.

Here is an illustration of the difference in construction of the two valves, made by Clippard. Notice the proportional lift has a stiffer spring rate than the PWM valve, enabling the PWM valve to perofrm full on//full off a great deal easier.


Because the way the PWM system meters its flow based on a simple pulse width, it is very accurate. Further precision can be increased by introduce a suitable RRFPR to maintain Manifold pressure against water pressure. It is also possible to factor in a small duty cycle increase to the valve relative to boost increase.

Final consideration: If you are planning in future to create your own MAP via a third part system - only the "PWM-valve" can be driven directly by the ECU, matching the principle of a modern "fuel injection system" in every respect. Warning, before rushing off making your own system, Clippard valve is only rated up to 100psi, even with the smallest orifice version. The larger orifice type can only sustain 25psi. Multivalve is needed for flows over 500cc/min.

Very nice write-up...love the detail...{thumbup}

Almost all progressive systems use a checkvalve between the pump output and nozzle for the reasons listed below:

Positive effects (well documented):
1. Retain some pressure in the line to compensate the next injection event. A 20psi loaded checkvalve will keep 20 psi of pressure in the line after injection.
2. Stop water being siphoned into the engine if the water jet is installed in the vacuum side of the manifold.
3. Prevent emptying the entire tank into the inlet tract if the tank location is higher than the jet (gravity fed) or the car is parked on an incline.
4. Stop some dribble after an injection event. Even when the power of the pump is switch off, the inertia of the rotating mass keeps the pump running for a second or so.

Negative effects (less well documented):
5. The presence of a checkvalve has a very significant impact of the dynamic range of a 150psi progressive pump speed system. A 20 psi checkvalve inline will instantly drop the 150 PPS system down to a 130psi span.

6. A normal nozzle requires ~30psi for produce a decent atomised spray. An inline 20psi checkvalve means the pump has to produce 50psi to produce a decent spray.

7. Let say the PPS system’s starting point is at 12psi boost, the system will now require 62psi to produce a decent spray. Some vendor will tell you a check valve will not impede flow once it is opened, true. But it will tax the pressure heavily where the PPS system relies heavily upon.

8. When the PPS system arrives at 24psi boost (end point) manifold pressure, the dynamic pressure is now further taxed. Before not too long later the dynamic range of a PPS system is now from 62psi to 126psi – translate it to flow: 176cc/min to 326cc/min, a mere 84% percent increase.

9. I have taken some data from a reputable PPS system manufacturer, a 150-psi 60W Shurflo flow pump has the follow characteristics:

M2 nozzle with 1/8 ID hose 180psi
M3 nozzle with 1/8 ID hose 160psi
M5 nozzle with 1/8 ID hose 135psi
M14 nozzle with 1/8 ID hose 105psi

From the figures above, the pump is only capable of sustaining 135psi system pressure on a M5 nozzle. Despite claiming the pump is capable of flowing one 3-4 litre per minute, conveniently missing the pressure parameter. The above PPS system maker is the only one that published these figures in public – thumbs up for them.

Summary:
Some PPS system makers are now offering an inline solenoid upgrade so that the dynamic range is improved by a good margin. Not sure why they didn’t include it in the kit at the beginning. Many PPS systems run their pumps up to 200-300psi - something I have questioned Shurflo, they said NO, NO and NO - no "ifs" or "buts". They are looking into a higher pressure pump but not yet done and will not be released until they are comfortable with it. I have a meeting with the Shurflo's director of engineering at our works two months ago during his UK visit - he confirmed the imaginary 200psi+ pump (8000 series).



It appeared Shurflo is the main supplier of the majority of water/alcohol injection systems makers. So it is more appropriate to take a good look into them first. You will be amazed to know how many pump variations Shurflo offers over the standard "off the shelve" configuration. I was told it has over 300+ "custom" configuration is still active.

Pump motor:
Currently, there are three frame sizes: short, medium and long stack, covering three power rating of 60W. 100W and 150W.



Pump cam angle:
Cam angle dictates and governs the final specifications of pump's flow rate, pressure. Shurflo offers: 2, 2.5, 3.0, 3.5 or possibly more profiles. Depending on the application, WAI makers can select the most suitable cam for their system.

At first glance, using the highest lift cam will produce the most flow and pressure. But if this cam is matched with a small motor, it will cause undue stress on the motor winding. This is very similar to going up a steep incline on high gear where the car's engine and gearbox is being stressed.

On the other hand, a low cam angle will produce less pressure and flow. Some PPS manufacturers prefers using the low angle cam because of the following:
1. Less stress to the diaphragm - long term mechanical reliability
2. Less stress to the motor - lower running temperature.
3. With in flow range up to a 600-800cc/min or so, there is ample pressure generated to hold the system at 150psi.
4. A much smoother control range from a PPS controller. High cam lobe tends to ramp up too much pressure with the same duty cycle applied.
5. The pressure spike is also much smaller, this allows the peak pressure closer to the "150psi demand switch". Overshooting will cause the infamous "pulsing" often associated with a PPS system.
note: a system advertised as 150psi@3 litre/min will NOT out-perform a 150psi@1 litre/min system. Often, the latter is a much better system. As far as the raw material cost difference is concerned, there is NONE. In these days, hypes rules the market.

By-Pass pump (no demand switch):
Shurflo Offers a huge range of internal by-pass valve to overcome the necessity of using a "demand switch" . Excess pressure is being "by-passed" internally by a set of spring load poppet valves (x3).

This type of configuration is more suited for PWM valve system rather the PPS system. The pump is switched on prior just before injection on and attains a "steady" line pressure continuously through out the entire delivery cycle.

Only a very small numbers of PWM valve WAI manufacturer uses this set up.


Regulating pump pressure steady (with on-demand switch):
Other than employing the bypass valves, the water pressure can be limited by using an "on-demand" switch. This method is simple, every time the pump hits the "set pressure" of the on-demand switch, the 12V feed to the pump is interrupted. Shurflo recommends this method should only be applied to application where the usage is intermittent and not cyclic use.

Against the recommendation, there are WAI systems on the market employ this technique to control water pressure. Most PPS systems only hit this pressure at peak injection pressure intermittently so long term damage to the pump is not severe. Pressure spikes of 20psi+ exists.

If this method is being used to regulate water pressure on a PWN valve system, it is a completely different story. The "intermiittent" use now becomes " cyclic" usage. This will create long term problem on the switch as well as the pumping mechanism. Water pressure will also suffer from pressure spikes, sometimes as much as 20+psi spike. Using a low hysteresis switch may reduce the ripple but the other half of the problem still remains.


to be continued...

this is excellent Richard, thanks a ton.

........ patiently waits for the PWM valve controlled system write-up.........

I've seen 3 wai companies pushing injection pumps outside the safe operating limits of 150psig/35psig

2 shurflos: one to 250psig and another to 220psig

1 flojet: from 35psig to ~100psig (and they blame failure of pump because customer use a "wrong" kind of VP methanol) :eek:
http://www.flojet.com/files/rlf_series_f100-196.pdf

This is just wrong! {thumbdwn}

Are the shurflo pump a positive displacement pump? I have never had one apart to look at it and I dont work with diaphragm pumps.

Excelent thanks for the info.

Not quite, but close.

If the diaphragm is made of metal instead of rubber on cam, it will be a positive displacement pump.

I have to question your logic on item No. 6. A spring loaded in-line check valve containing a 20# spring will require 20# to open. In this point we agree, but once opened the downstream line pressure will see the same pressure as the upstream side of the check valve. This of course does not take into account the pressure drop across the valve (Cv), pressure losses due to frictional losses of the pipe (f) or pressure drop across the spray nozzle itself (Cd). There are other factors to be sure (bends, fittings, etc.), but you do not add the spring pressure and the desired nozzle pressure to determine your required pump pressure.

that is correct. A 20psi check valve will not adversely effect your pressure.

I'm on the same page as Richard. As long as there is fluid flowing, the check valve will cause a 20# drop.

What do you base your opinion on? My information is based on 37 years in the engineering business. I size piping systems, valves and pumps on a regular basis. I would refer you to Crane Flow of Fluids Through Valves, Fittings and Pipe manual so that you may see the error of your decision. As I stated earlier, there will be a pressure drop, but depending on the orifice size of the valve it can be less than 1# and as high as 3-4#.
I find most all of what Richard has written to be very informative and accurate, but on this one case, he is quite wrong.

Thanks for chiming in, I share your input. It is a difficult one to explain.

The easiest way to determine if there is pressure drop or after the checkvalve is setting up a test rig - a video to record the test and post it here.

I have all the equipment at our works and If enough people want this done, I am more than happy to set it up.

Before we go further, do you think the the following set up is acceptable?

I can only do this with air as I can generate accurate and stable air pressure much easier than water.

Pressure source => gauge => checkvalve=> gauge => variable restrictor (jet replacement)

Back to your theory, if the pressure is the same between the two chambers (before CV and after CV) the CV should close?

The way you describe the progressive controller for the turbo is interesting. A turbo spools up and reaches a set psi, then the rpms keep rising as the boost stays the same. Technically the injection system sprays the same amount of water/meth/whatever after that set boost level and at any RPM. Correct?

What about a centrifugal supercharger? I am about to put on a progressive kit like this, but my blower starts building boost at about 3500-4000rpm and reaches 10-11psi by redline. There is no boost plateau so the injection should keep increasing up until I shift. Should I set the controller to peak the injection pressure at a lower rpm than redline or should I set it at my maximum boost so that the power keeps building throughout the rpm range?

Lots of stuff to think about, definitely subscribing to this thread.

^You will see a pressure drop across the valve. You will not see a 20psi drop across the valve, there will be a pressure drop across the valve just a very small amount. You also have to consider flow through the valve not just the pressure.

This is very an interest aspect for the boost related PPS system. It looks the PPS system suit your application perfectly. I think it will also suit the positive displacement SC well on the surface.

The Centrifugal surcharge obeys the square law - boost is not strictly linear to RPM rise - hence it rarely see much boost at low RPM.

If we discussing the WAI flow matching the power curve, it will also need match the fuel demand. For this to work well, we have to take account of the specific fuel consumption of the set up. More fuel is needed to supply the added load imposed by the supercharger - If you base the WIA flow relative to boost and and not fuel flow, you need to make some adjustment on your fuel map to compensate the non-linearity between the two flows - fuel=3D and PPS=2D.

It will not going be a simple "bolt-on and go" affair - but pretty close.

Each different type of medium that you flow through a specific type of conduit will record a different reading at the measured point. Viscosity of the medium and frictional factors of the specific material of the conduit will contribute to the variance in your collected data. Gasses, obviously, will not give you the same values as liquids, but for the test you propose, air would give a relative recordable value.
First of all it is not theory, the information I passed along is based on engineering constants. As to your question, yes if the pressure was equal on both sides of the check valve the spring would not open. But, that is not the case in an injection system. You have a constant pressure on the upstream side of the valve and when that meets or exceeds the spring value, naturally the valve opens. The difference is that you have an open orifice (the spray nozzle) that is very much open. Because of this, the pressure passes through the valve to the nozzle and into your device (or I/C pipe in our case). Petro/Chemical plants all over the world use similar set-ups for chemical injection in cooling towers, lube oil systems etc. The more critical the system the more sophisticated the pressure control device becomes.

If the pressure drop is small than 20psi, the valve should close by definition?

I am be able to explain why you think this is the case. If you have any open ended checkvalve, you will not see any pressure at all.

Mass flow generates pressure = keeps the valve open - agreed 100%.

But if you have a restriction downstream (water jet), the flow will no longer able to keep the valve open, 20psi differential remains true.

No, because you have a constant upstream psi value.

Let me clarify here. These numbers are just for explanation purposes. The checkvalve has a cracking pressure of 20 psi. So the pump will discharge 20 psi and open the valve. There will be a small pressure drop across the valve due to flow restrictions, lets say 1 psi. So the outlet side will have 19psi going to the nozzle. Now the nozzle creates a restriction which will create back pressure on the valve, but the pump discharge pressure will increase as well. Pump are rated at a certain psi and flowrate.

Every bend, valve, fitting, or anything else installed in the line will create pressure loss. The size of the hose or pipe will also effect the pressure of the pump. The loss usually will not be noticeable on a gauge though.

I understand basic fluid theory and know how to apply it, I just suck at explaining it.

isnt this easy to prove if the checkvalve is a source of restriction...

constant pressure source - 125psig

A - pressure source/pump -> nozzle

B - pressure source/pump -> checkvalve -> nozzle

the difference in volume output of injectant will indicate if the CV is a source of restriction

A - B = 0, no restriction
A - B >0, there is restriction from checkvalve

or is this too simplified?

This is getting to become a good discussion. I think I'd better set up the simple test and than we can discuss the result.

It looks pretty 50/50 for the time being. I will do it with air first and if time permit, I set up with water with a 22psi checkvalve.

why is it that everyone has to throw their pedigree around when a different opinion is posted? {thumbup}

I've only been in the chemical industry (chemical engineer) for 20 yrs, graduated in 87, so you have me on experience. Yes, I have a Crane manual right behind me, in my book case, haven't cracked it open for probably 10 yrs or so. Haven't needed to. :beer:

If you need an engineering type example..... say you have a boiler that makes 150# steam, and you let down 150# steam across a pressure regulator (pretty darn close to the ball/spring design Richard showed) to control another header at 50#, you're saying that regulator is only going to take a 1-4# drop? Last time I checked, that's 100#'s. The cross sectional area for flow will change based on flow demands, but the pressure drop across the regulator remains constant at 100#'s.

It is difficult to set this up with pump with pulses, even with damper etc. The checkvalve react with peak pressure rather average pressure. I normally test prototype with air than water. I will get myself with a pressure vessel so I can get pressurise water with little or no ripple.

Your test is simple and will get result. Didn't you once set something up similar a months or twos ago with a digital weighing machine. For I could remember, the test with two checkvalves flows less than one with one checkvalve. It will be interesting to do the same with no checkvalve at all and plot the three - three points are very useful.

Do you still have the link?

I have no experience in those industries. I just keep to simple logic, sometime I guessed right and sometime I guessed wrong. All we need is a simple set up I am about to do and if slowcar has the video clips, it also helps on the practical side.

If A(CV) - B(no CV)=0 than I will correct the text and apologise. {thumbup}

i think the hosting site erased those files because they were huge.

i'll redo the 3 experiment today after work:

A - pump -> nozzle
B - pump -> checkvalve -> nozzle
C - pump -> checkvalve<->checkvalve -> nozzle

Pump = pulsing on-demand shurflo
Time for injectant collection = 60secs
Checkvalve = swagelok 25psig crack

:beer:

do you have the plot of flow vs. pressure for that nozzle? If so, you can get exactly what the dp is for the checkvalve.

i'll be using Hago nozzle
http://www.hagonozzles.com/documents...M%20&%20MW.pdf

nozzle = M5

In no way did I intend to belittle you or to throw around my "pedigree". If you are in engineering then you know that you must substantiate any statement or finding. The flaw in your thinking comes from trying to equate a regulator with a spring check. The regulator sole purpose, as you stated, is to reduce downstream pressure and hold it at a constant. A spring check serves two functions a) as a backflow preventer and b) to limit low pressure pulses from traveling downstream. Once opened (or cracked) the upstream pressure can rise to any set value and the resultant downstream pressure will be equal to that pressure minus the sum of pressure drop across the valve and line losses.
Richard, because I'm anal by nature and wanted to get a second opinion on your testing methods, I asked our Sr. Process Engineer what would be the most accurate testing device and set-up. Here is what we concluded:

Air as your medium would be acceptable (not totally accurate, but acceptable).
You will need a regulated air source. (compressor w/ regulator).
All connected piping should be hard piped for repeatability (any bends or kinks will change your readings).
Install one (1) pressure gauge upstream of the check.
Install a centrally located check valve with a known spring pressure.
Hard pipe to a spray nozzle installed in a larger diameter sealed tube (say 2-3 ft. long).
Install a pressure gauge in the end of the sealed tube.

To test, merely turn on your air source with a regulated starting pressure below the spring pressure. Open the regulator up untill you read 30# on the gauge in the sealed tube. At that point, read the pressure on the gauge in the upstream gauge. Subtract the downstream gauge from the one upstream and this will give you your pressure drop for the total system.

the upstream and downstream pressure across a spring loaded ball check can never get closer than the spring rating, or else the ball will check. In this case, it's 20#'s. The pressure drop across the checkvalve can be more than 20#'s I agree, but never less, or the ball is gonna check! It's gonna open/close just enough to cause a 20# pressure drop until flow exceeds the systems capabilities to control at 20#'s, then it can be more.


No worries about belittling me, I love a good debate. I've been wrong before, I'm sure I'll be wrong some time soon again. Just not this time {thumbup}

This is a bad example to use. I know for a fact that ball check valves are not used to regulate steam pressure. Instead diaphragm operated valves are used, either spring adjusted pressure or pnuematic controlled. The regulator valve is not set at one simple point and left there, the valve has to be able to adjust to varying pressures and demands.
A ball and spring type valve can be used to regulate pressure in a system, but it regulates the presure up stream of the valve not down stream. Where this is commonly used is in hydraulic control systems. It operates just like a MBC. You adjust a nut on top to control the spring pressure and when it is exceeded the valve opens to relieve pressure, once the line pressure upstream of the valve is less than the spring pressure the ball reseats. Some cars use this same way to control oil pressure from the oil pump.

100% correct!

You can use this method to see if the valve is a source of restriction, however sometimes the amount of restriction may be less than 1 psid. I work with numerous fluid systems and seeing a restriction from one point to another is not always simple as it sounds. I have piping systems that are 4" in dia and numerous valves and bemds ar in the system as well as flanges. If you read the pressure in one point of the system and read the pressure at another point in the system you will possibly get the same reading. Most gages do not read accurate enough. The best way to conduct this test would be to use a gage calibration unit, however these are pretty expensive and hard to come by. Even gages in engineering plants are allowed to be off by a certain %.

No pedigree here. I am a highschool flunky. I am a Chief in the Navy that operates boilers for a living. No classroom or fancy college books just real life experience. I have to figure out what is wrong when the Engineers screw everything up. :lol:

all you engineers!! :rolleyes:

:lol:

I am not an Engineer I am a Steam Plant Operator.

O.K. you and I were doing allright until this statement. Let's examine your "expertise".
If we were talking about a sealed system, regardless of volume, you are correct in that once both sides of the valve reached equilibrium, the valve would "check". Wherein our case we have an open nozzle, the downstream side will never be equal to the upstream side and, therefore, the valve will never close. At least not until the upstream pressure falls below 20#. You refer to the spring pressure as "pressure drop". Wrong! Pressure drop is a function of frictional losses as a medium passes through an orifice or restriction. It is certainly not a function of the spring value. The valve will open when the pressure exerted on the ball reaches and exceeds 20#. The valve opening and closing has nothing to do with "pressure drop". You say you have no need to open your Crane, but if you've forgotten the difference between line pressure, pressure drop and flow coefficients for valves perhaps it's time to "crack" the books.

Whoosh,

I have outline the test set up, please pass it on to you Sr Engineer and confirm is a good enough for the a rough indication before I spend more time on a more refined test rig.

http://www.aquamist.co.uk/forum/gallery/EVO/test.gif

As a matter of interest, what is his view on this topic?

it's going to boil down to the checkvalve design. If it's a straight through design like Richard has shown, where the back of the ball sees the pressure of the fluid downstream of the seat, it's gonna cause a pressure drop equal to spring pressure.

I see where you guys are coming from, and agree with you, IF you're refering to a design other than the way it's drawn. But if it's a check valve like Richard has drawn, it will check at the spring pressure and cause that same amount of pressure drop.

Picture a spring loaded diaphragm valve (simple regulator with no reference) with the backside of the diaphragm vented to atmosphere. Then yes, the upstream and downstream pressure across the valve can be very close, all that matters is the pressure pushing up against the diaphragm and in the pipe, and pressure drop will be minimal. This is similar to a fuel pressure regulator, only the reference for the backside of that diaphragm is manifold pressure, not atmospere. But that's not what is drawn. Follow me? Like someone already said, I suck at explaining this.

Slowcar, any idea what the design of the check valve you're testing is?

I will draw one, give me 30 minutes

we're still doing allright, don't get bent. :lol:

Look at the picture Richard posted. The orfice size around the ball/seat will change based on flow. More flow, the orifice gets bigger, less flow and the orifice gets smaller. Yes, the orifice is causing the pressure drop, not the spring, but the spring is in charge of the orifice size. So yes, the spring pressure can be said to be in "control" of the pressure drop. I think you're thinking of a typical piping check valve, which isn't what Richard has represented.

security at work is tight like a solenoid valve...i cant upload anything to the web

the checkvalve is an in-line swagelok CP series - will have a x-section pic and pressure Vs. flow diagram posted when i get home

The downstream pressure gauge needs to be past the "variable restrictor". I'm assuming that the variable restrictor equates to spray nozzle. This is why I suggested that the nozzle be fitted into a sealed pipe that would replicate an upper I/C pipe. By having the gauge at this point it will give you a true value of the required pump pressure required to achieve a desired nozzle pressure and, hence, a proper flow pattern at the nozzle face. If you duplicate what you have drawn and include the changes I have suggested, we both are certain you will get the results you are looking for.
It's actually a "her". I have discussed everything I have posted here with her. By nature and by company mandate, each persons work is QAQC'd prior to going out of the office. I gotta tell ya...she's on my side on this one.

Abner, I have found your videos, I will render them to a more a manageable file size and put up a viewable link on our web space. :)

its been awhile...
does it cover all 3 cases?

A - pump -> nozzle
B - pump -> checkvalve -> nozzle
C - pump -> checkvalve<->checkvalve -> nozzle

The two check valves I posted is the most commonest valve used on the WI community.

Whoosh, any objection using them for my test?

Abner will post up the section drawing of his Swagelok check valve - also commonly used on the water injection application - Please comment when drawing becomes available.

OK from looking at the picture above you will have a pressure drop due to the "orifice" of the valve. This creates a "venturi" effect which will cause a pressure drop. How great the drop is unknown with out extensive formulas that I no longer remember.

I can only find cv and no-cv. I think it is good enough - 27MB each !!! :D

Abner is this the check valve you use? sorry file is too big to upload.

should I do the test or not?

Seemed to have some mind changing since the drawings regarding my first statement of pressure drop. :confused:

Do the test. I want real world results here not just some paper and formulas.

Wonderful! That's an industry standard. If you are using a 1/4" npt valve it has a Cv of .35. If you tell me what the low rate wil be in gpm, I'll tell you what the pressure drop across the valve will be. On Wednesday because I'm done with calculating today!
The formula is ^P=(Q/Cv)(Q/Cv) p/62.4
Do the test!

i'm been sitting infront of the monitor the whole day. Itching to do to the back to do some experiments :p

I guess post the set of videos from awhile back and i'll do another set of experiment, this time i'll just take pics

i have a tabulated table from swagelok posted

Abner,

I have found all three!!! {thumbup} {thumbup}

{thumbup}

:( ...no experiments to do

OK, abner, I lost them again. :lol: :lol: :lol:

here are the pic/data for the checkvalve that Richard will be posting the videos on

http://img1.putfile.com/thumb/2/5002075614.jpg

X-sectional view

http://img2.putfile.com/thumb/9/24618581715.jpg

Pressure drop Vs. flow table

http://img2.putfile.com/thumb/9/24618581865.jpg
http://img2.putfile.com/thumb/9/24621471367.jpg

it's hard to tell what's what in the diagrams of the check valves posted.

Here are a couple of pics of a checkvalve in the water line to the intercooler sprayer on an 03 Evo. You can clearly see the ball on one side, and the spring on the other. I hooked up a little rig I have with a bicycle pump, vacuum tubing, tee and pressure gauge. The spring is a 3# spring.

http://i124.photobucket.com/albums/p...Picture091.jpg
http://i124.photobucket.com/albums/p...Picture092.jpg
http://i124.photobucket.com/albums/p...Picture093.jpg

This is the type check valve I see in Richards diagram in question with the 20# check valve. And I still say this type of check valve will cause a pressure drop equal to the spring pressure. The differential pressure across the ball and seat has to equal 3#'s to get check valve to crack open, and it maintains a 3# drop even after it's open, at various pumping rates. The differential has to equal 3#'s to keep it open. So if you have no pressure on the backside or downstream of the ball, it opens at 3#'s. If you had say 10#'s of pressure on the backside of the ball, I'd have to pump 13#'s of pressure with the bicycle pump to get it to crack open and stay open and flow.

Are you guys trying to tell me once the seat is broken and the valve cracks open, the pressure drop will be dependent upon some Cv value alone????? Nope, it's going to be the pressure drop caused by the orifice (Cv) PLUS the spring pressure. At low flows, it'll be damn near 3#'s, at higher flows, it could be considerably more. But never less than 3#'s.

If this is the type of check valve for the test, by all means run the test.

It is 1 am Wednesday morning for me. This thing keep me wide awake - interesting stuff.

I will run the test in 7 hours time when I return to my office - I will take my video camera. If the results looks positive one way or the other, no need to do more. But if the result is murky, I will set up a test in a more controlled environment.

Anyone is take bets?

Very simple check valve design, good engineering.

I can see you point very clearly. the ball valve will create a flow paths relative to the upstream pressure. CV only applied to mega flow in relation to the ID.

My original statement is balanced on a knife edge until tomorrow. :confused:

uh, didnt you guys forget that there is a pressure on the downstream side of the checkvalve, generated by the turbo? i.e. Air pressure.

...or did you account for that already? I might have missed it in the throwing around of formulas...:rolleyes:

(im still studying to be a mech. engineer :D)

VIDEOS (Microsoft IE only):

Here are abner's Check valve test - please be patient, video files are about 5MB


NO CHECK VALVE (see video)
Download (1.78MB): Click here

ONE CHECK VALVES (see video)
Download (4.40MB): Click here

TWO CHECK VALVES (see video)
Download (3.86MB): Click here

The final flow quantity should give you some clues on the effect of the CV.

{thumbup}

See post #3, a chart with CV accounted for turbo pressure.

^yea, i saw that, but i didnt see it anywhere in the other guys formulas. im sure they prolly did get it, as they know more than me.

me = still in school
them = out practicing what im learning...lol

although the starting volume with no checkvalve seems a bit low for an M5 at 125 psig, the drop in flow is what's significant. After looking at all the backup data Abner provided for the nozzle, and check valves, it's obvious that each check valve is adding close to 25#'s of pressure drop (and they have 25# springs, imagine that) between the pump and the nozzle.

Just like Richard had originally stated. You can sleep well tonight Richard knowing your were spot on to begin with. :beer:

Abner, thanks for going to the trouble of setting up the test and running it!

I'm going to make this real simple for you...please go borrow someones Whitey/Swagelok catalog, page through the little tabs until you come to one that says "Check Valves" on it, go to page 3 for the inline CP Type valves and find the chart at the middle of the page. See where it has 4CP in the left column? Put your finger there and go across the columns until you come to the column that gives a value for Cv. See how the manufacturer of the valve provides you with this number so that you can determine...pressure drop through the valve!!!! Damn, ain't that sumpin'? Slap me naked and hide my clothes! The people that make the valves actually give you the tools you need to size the valve so that you can get the downstream pressure your system requires. Ain't America great?

So what causes the pressure drop, the spring or obstruction to flow? The latter would play more on dynamic pressure I think

A rod bolt may snap in two in either case - kinda like waiting for an earthquake

I'll slap you silly, but I won't slap you naked {thumbup}

The test Abner did shows the check valve adds a significant amount of pressure drop, and reduced flow at the nozzle. And it's not due to the Cv of the valve by itself. You have the flow measurements you requested, now go do your calculations and confirm this. It's the spring!!!!!!! lol

Notice how the graph in the swagelock catalog starts at 25 psi pressure drop at 0 flow, and only goes up from there? Ain't it sumpin? The spring basically resets the starting point to 25 psi worht of pressure drop, and as flow increases from 0, so does the pressure drop. And the pressure drop increse as flow increases can be calculated by the Cv, but the spring resets the baseline to 25 psi pressure drop minimum to begin with.

You failed to recognize in the video that the pump is also a demand pump so it cycles on and off so yes the check valves are closing and opening. The argument was never made on volume of flow, the check valve will restrict flow. The argument was will the check valve decrease pressure by the amount of the spring. Volume and pressure are 2 different things are not to be used in a comparison like this. The video did prove that check valves will restrict flow rate though..

now imagine progressive pump speed system using a on-demand pump + checkvalve....:crap:

if you look at the flow data Abner generated at the nozzle in the videos, and then compare that flow rate to the published data for that nozzle on flow vs. pressure, which Abner also posted, it's easy to see that each check valve accounts for about 25 psi or so of reduced pressure at the nozzle to account for the flow reduction recorded in the videos. The WHOLE argument is about volume of flow, and what impact will a check valve have on the volume of flow at the nozzle. The only way to reduce the flow at the nozzle is to decrease the pressure, and two of you say a check valve won't decrease the pressure onec open, but by about 1-4#'s or so, depending upon Cv of the valve.. Abner's test of check valves with 25# springs shows a corresponding flow reduction at the nozzle associated with about 25#'s less pressure. And this was at flow rates of 125 ml/min with two check valves, a very low flow that the Cv won't calculate out to be a 25# drop. With this style check valve, you most definitely have to add the spring pressure plus the pressure drop due to flow to get total pressure drop across the check valve. You can't simply say once the valve cracks open, the spring no longer matters and doesn't figure into the equation.

Yes, flow and pressure are two different things, but intimately related to each other. Change one, and the other must change also. You can't separate the two.

I don't use a check valve I use a solenoid for that exact reason.

Finally, you've caught ...some...snap. I absolutely agree with you that the spring is there to hold the initial pressure on the upstream side of the valve to whatever the spring value is. But that is not called pressure drop. The friction of the fluid as it travels through the orifice (i.e. the inside restrictions in the valve) for a specific flow rate is called pressure drop. The entire basis for my original disagreement with Richard was not flow, but the required pressure to maintain a 30# pressure at the nozzle for proper atomization. As flow (gpm if you will) increase so does the pressure drop across the valve. Just as it does for fluid traveling along the inside of some type of conduit (pipe, tubing, etc). Can you see that there is a greater pressure drop in a pipe that is 1 mile long as a opposed to a pipe 1 foot long? The spring in this case acts as a backpressure regulator to maintain a minimum of "X" amount of pressure. As soon as you overcome that resistance and the ball is unseated, the downstream side of the valve become pressurized to a value that is equal to the upstream side minus pressure losses due to valves, pipe and fittings. Now...go spank yourself you silly Chem E.!

I define pressure drop as the pressure before the check valve minus the pressure after the check valve. Everything missing is pressure drop, no? Whether it is caused by piping design, elbows, friction, restrictions, turds, springs or balls. Any pressure missing from the inlet to the check valve to the exit of the check valve is pressure drop. When the ball is unseated, there will still be less pressure downstream of the valve (compared to the valve inlet) that is equal to spring pressure PLUS pumping losses. But to say that once the ball is unseated, the spring is totally out of the equation just isn't right.
We can agree to disagree. Abner's test has the original answer we were looking for.

For whatever the reason that caused the flow drop, the spring-loaded checkvalve has a definite influence on the outcome.

Call it what you will... pressure or flow. I just happened to illustrate the effect of CV in "pressure" units to simply the chart as the crack pressure of CV was specified in pressure.

I will post the video tonight and should see everything in the form of "PRESSURE" or pressure drop.

The case for the CV should be closed after this. :beer: :beer: :beer:

I thought some pumps have an internal gate valve when not running.

hey, you are about to start the whole debate again, this time regarding the valve inside the pump :lol: :lol:

I am moving on to the pump design next - will make additional info on post #3 after the posting the video.

uh oh....

:lol:

Richard, I was under the impression that (concerning my first post) that you were trying to determine the correct amount of pump "pressure" to properly atomoze the fluid at the spray nozzle. That desired "pressure" was 30psi. Is that correct? If that is true, let me pose this question/scenario. If the spring pressure is 25psi, if you were to put 26psi at the inlet face of the valve, does anyone truly believe there will only be 1psi of pressure downstream? If that is what everyone believes, I'm out of this thread!

Now, to respond to Abner's experiment. What he performed was a flow test for volume of water with "no" restrictions in the line, with "one" restriction and finally with two restrictions. He did not perform a pressure test. If volume is your only concern, you can increase the volume by simply increasing the line size as a 1/4" line is very restrictive and imparts a high coefficient of friction on anything passing through it. It is entirely possible to have a large volume of water at the nozzle without sufficient pressure to properly atomize the fluid.

To respond to the gentleman student of engineering, you are absolutely correct that the formula I provided did not take into account the I/C pipe internal pressure as the turbo builds pressure in that pipe. We're still trying to get past the check valve let alone the complexity of an everchanging pump head pressure. One dragon at a time please.

Finally, Mr. dubbleugly01, your definition of pressure drop is spot on. Your error is in thinking that the pressure downstream of the valve will not try to equalize once the valve is opened.

whoosh:you are making mole hill into mountain...

Case 1: pump -> nozzle
pressure-flow/atomization from hago chart:
http://www.hagonozzles.com/documents...M%20&%20MW.pdf

Case 2: pump -> checkvalve -> nozzle
pressure-flow/atomization ( between checkvalve -> nozzle ) from hago chart:
http://www.hagonozzles.com/documents...M%20&%20MW.pdf

Case 3: pump -> checkvalve<->checkvalve -> nozzle
pressure-flow/atomization ( between checkvalve<->checkvalve -> nozzle ) from hago chart:
http://www.hagonozzles.com/documents...M%20&%20MW.pdf

I'm sorry you feel that way. I was merely trying to help obtain true calculated data. There is much you are assuming. Out!

so... if you're video is measuring in grams... you're talking a difference of 60 cc/min in volumetric flow between using a checkvalve and no checkvalve? (Approximately)

Datalogs from some of the systems would be good too

1) I was try to convey a nozzle requires 30psi to produce a decent atomised spray. Hago (nozzle maker) don't even list droplet size below 40psi on their site.

2) 26psi - 25psi =1psi yes this is what I am saying. For the same reason, 100psi pump with a 50psi checkvalve - nozzle will see 50psi.

I think you want to explain yourself an bit better on Abner's video. A combination of pressure and flow after the checkvalve has yielded a lesser flow - why? please explain this again.

I may need your help on this if possible, I believe you have a great logging device. Can it take external 0-5V signal?

si o no?

CV pressure drop test set up
 

Here is the set up of the day to perform the Checkvalve pressure drop test. A video will be available to view or download as soon as I finish rendering it (4MB)

seeing is believing...
 

Links dont work.

please try again

No, this is where we agree to disagree. The pressure will not equalize unless the check valve leaks like a sieve and does not function as designed.

I keep tellin ya, it's in the spring!!!!! That dang spring is causin pressure drop :lol: I know you can't comprehend that, it's ok, you believe what you believe, I believe what I know.

now that's irrefutable. :mitsu:

dubbleugly01,

I am glad this is over, would like hear Mr Whoosh's view on the result of the test.

Flow drop by a check valve: irrefutable!
Pressure drop by a check valve: irrefutable!

Glad I have no check valves
Although I have never used the channel yet, yes you can set it up to take other external sources

Not sure the video is entirely precise. Question: since the first test w/o cv has no way of stoping the line from emptying out in the end, how is this variable taken into account? From the end results I'm seeing a ~ 23% decrease in flow by adding 1 cv, and ~ 53% decrease by adding 2. This is hard to believe. The test needs to ensure that when the time is up the water is not allowed to drain from the tube into the beaker when there is no cv, b/c the cv will stop this flow when it is in place. While that may add to some skew, it is still undeniable that there is some drop in flow. ...but ~ 25% per cv?!? a bit hard to believe. lol sorry, I'm no fluid dynamicist either... just a lowly ME.

edit: I'd also really like the pressure drop test to be conducted with water. While regulation may not be as easy... it can't be much harder to just hook up pressure transducers/gauges before and after a cv with a real WI nozzle acting as the "adjustable flow regualator". Air is much more compressible than water and acts very differently. There was never a question of whether there is a dP across the cv, but in order to quantify it and rule out spring pressure, you have to actually use the medium that we are interested in.

edit: sorry for one last question, but what was the cracking pressure for the cv(s) used in the video test? Mine is 40psi, and I would like to know similar it is. Thanks.

The proper thing to do here is concede. Your test was well done and irrefutable. Please understand that at no time did I not state that there would be pressure drop and loss of flow across the valve. My error is in the psi drop caused by the spring across the valve.

dubbleugly01 please forgive my doubting your expertise.

Upon further pressing my Sr. Process Engineer, she stated that in practice, the pressure drops are added up across the entire system. In other words, if the spray nozzle will see 28psi in the I/C pipe, the nozzle needs 30psi to properly atomize, there are 5psi in line and fitting losses and the spring has a 8psi value...then the total system would require 71psi to see 30 psi downstream of the spray nozzle. She further stated that that will be a conservative, but safe, method to establish minimum pressure. "Better to error on the safe side she said". I thanked her and then later shot her for leading me astray.

Whoosh, you are a gentleman.

I was in two minds about this, the urge to set up a test was so compelling to say the least!

I will continue on uttering the topic - please keep me honest - I was lucky this time.