M20J GAMI spread issue (flow divider?)

[jaylw314]

19 hours ago, jkarch said:

Kortopates,Shadrach,  Carousam, CaptRJM, and TeeJayEvans, Thank you all for that information.  I will rerun the GAMI test per the Savvy Test Profile, and will also find a way to get the valve lift test done.  It does seem I have a hard time getting #3 and #4 to peak without roughness, unless I'm looking at the secondary peak, but it's also possible I'm not moving as slowly as the Savvy Test Profile suggests.  Regarding the cleaning in an ultrasonic bath, that would be worth doing especially for the fuel lines.  I think the injectors themselves are pretty clean but I might as well do that.  As for the uneven flow, that may only be part of the story, and there may be a combination of fuel issues and air issues, including valves.  It's strange,because the engine is extremely smooth and runs very well while rich, and it's also strange that the flows are not even.  I did call a company that deals with fuel metering today and found out that I should check the fuel servo as well to make sure the air intake part doesn't have any fuel staining in it, so I guess I have my work cut out for me.  I'll post as I get through these hurtles.  In the meantime, the engine does run smoothly, I just don't know what the story is with the actual GAMI spread.

 

-J

FWIW, I thought I had a hard time getting a consistent GAMI spread on my IO-360 A3B6.  After getting one mag IRAN'd, I've not had problems.

I think the problem was getting a false second peak.  The is presumably caused if one spark is not strong enough to reliably ignite the leaned fuel-air mix.  The EGT increases (like during your runup mag check), but since the spark may not fail to ignite the mixture every cycle, the EGT rise is gradual.  In my case, I believe this was starting to happen BEFORE one of my cylinders reached peak EGT.  This would look like 2 cylinders would peak as I leaned the mixture, but the other two EGT's would continue to climb as I leaned the mixture further and further.  They would only start dropping when the engine started running rough as the other spark plug started failing to ignite, by which point the other cylinders were more than 80^o F past peak (which equated to a GAMI spread of about 1.5 GPH).  Other times I tried it, I could find a GAMI spread of 0.5.  The culprit was the inconsistent artifactual rise in EGT in two cylinders.

After I get the mag IRAN'd, I've since determined my GAMI spread is more like 0.5 GPH.  Now I am able to reliably get past peak without the artifactual peak.  I can still get it past 30-40^oF LOP (presumably because my other spark is getting weak), but at least now I can tell the difference.

Anyway, long story short, double-check your ignition performance, especially since you're reporting the engine runs smoothly even LOP.  It's hard to imagine an engine with grossly uneven fuel flows would run smoothly with the engine leaned out.

[carusoam]

+1 for normal psychology rules... (everything works as planned)

One of the things to look for during run-up.... things not working as planned

When testing the magnetos, be looking at the JPI.  

Each time you select a single mag, each of the EGTs will rise uniformly.  Allow for a few seconds to see them all rise evenly...

when you select both mags, the EGTs will all descend evenly, allow a few seconds to see it...

If a Mag isn’t working or one is timed differently than the other... it usually shows up during the run-up.

Best regards,

-a-

[jkarch]

So I flew this evening and got the following data: 8500 MSL, I ran this test at around 21" and 2500 RPM,  0C, almost full throttle. (Pulled out a crack)

At considerably past LOP for all cylinders, I was around 8.4 GPH.  I could have gone as rich as 8.7 as my peak was 8.9 for the richest cylinders.  This is where I tested the mag drops in flight.

It seems to be a 100-120F climb or so in EGT for all cylinders on the left and right mags.

Peak EGT happens for all cylinders in the high 1400s , 1470-1498F.

Cylinders 3 and 4 peak close to 9.8GPH, cylinders 1 and 2 peak close to 8.9 GPH however there is an extremely sharp rolloff from 8.9 whereby the slightest vernier pull from that point on drops the fuel flow closer to 8.6 and below.  50 ROP is 10.3 and that's where I typically have operated, unless I occasionally operate around 9.5 between both peaks.  Note that these are the peaks where I found big roll-offs. There are certainly points where the EGTs level off for a bit, but these were actual measured peak temps.  Also note that cyls 3 and 4 were ~50(40-60 perhaps) LOP at the point when 1 & 2 peaked.

I kept the boost pump on for the test.

Engine runs very smoothly and power is good. I think my FF is probably off by 10%(it's showing more fuel burnt than actually burned. ) So I'd say I'm getting 0.8 spread, however per John Paul I'm maxed out on available options in terms of richness on the leanest cylinder.  I think the valve lift test makes sense, in addition to a borescope and checking out the brass nut covering the fuel servo as fuel stainage behind means I should overhaul the fuel servo according to Mike's Fuel Metering Service of Oklahoma.

 

Everyone I speak to says "just fly it..." Especially those who have flown in it with me, and my mechanic says it runs great and tells me I'm wasting money and brain cells thinking about it. I've even taken him flying and did the test with him.  It certainly runs smoothly...Thoughts?  Many mechanics I know would be itching for a chance to tear it apart for cash, but this guy doesn't see any issue.   I videoed the whole thing but it's 15 minutes long. Cyl temps were from 348, 335, 289, and 329.  Note the 289F cylinder is a secondary CHT gauge, the temps were more in the 330 range.  OAT was close to 0C at 8500 feet on an extremely dry day. 

Mag-wise I had the mags overhauled 200-250 hours ago and had an overhaul exchange mag.  Harnesses are original however, and I've had #1 harnesses repaired (leads replaced) in the past to fix radio noise issues.  

 

 

 

 

 

Edited December 6, 2018 by jkarch

[jkarch]

An update on the testing.  I flew across the country to the East Coast and stopped at a Mooney shop where my friend works to have a look at the issue.  It appears that my #1 and #2  fuel injection lines were changed in 2007 with lines that have an OD of 0.140, while the #3 and #4 lines have an OD of 0.120. I need to have the IDs verified but if the ID of the larger fuel injector lines is indeed larger than the ID of the smaller lines, could that account for #1&2 being richer (bigger tubes) than 3&4 (smaller tubes )and the spread being hard to fix?

[N201MKTurbo]

1 hour ago, jkarch said:

An update on the testing.  I flew across the country to the East Coast and stopped at a Mooney shop where my friend works to have a look at the issue.  It appears that my #1 and #2  fuel injection lines were changed in 2007 with lines that have an OD of 0.140, while the #3 and #4 lines have an OD of 0.120. I need to have the IDs verified but if the ID of the larger fuel injector lines is indeed larger than the ID of the smaller lines, could that account for #1&2 being richer (bigger tubes) than 3&4 (smaller tubes )and the spread being hard to fix?

Yes, it could.

[jkarch]

Thanks N201MKTurbo! Yeah I think we have a smoking gun. In the meantime after running the engine, the mechanic said it was a really smooth running engine and that I’m splitting hairs trying to fix this- he advised I wait till Annual to change out the rear lines. So it’s time for another cross country flight!

[aaronk25]

If that isn’t it, all I can offer is after much time invested on my M20j I pulled the gamis out. At full throttle John Paul was getting erratic results where different ones would peak at different times. He chaulked it up to a air leak in the motor. Not the case, same problem after overhaul. Other issue was blue staining around the injectors.

If a person looks close at a gami the area where the fuel comes out has a concave bowl shape where the lycoming injector is flat. At full throttle on my plane and some other Js it appears, we’re getting localized pressure inside the intake tubes which is same as outside atmospheric pressure and some fuel was actually spitting out the hole in the side of the gami injectors. In my case we think the concave shape of the gami face catches the very brief but present localized high pressure air created by the intake valve momentarily snapping shut just downstream of the injector.

The lycoming ones don’t do this. If we pulled the throttle back .5” of MP we could tune the gamis perfectly. That just seemed like a joke to me so we pulled the gamis and switched around the lyc injectors and got the spread to .3. Close enough. So I believe gami has some more homework to do on lyc planes with low restriction intakes such as the M20j.


Sent from my iPhone using Tapatalk Pro

[jkarch]

2 hours ago, aaronk25 said:

If that isn’t it, all I can offer is after much time invested on my M20j I pulled the gamis out. At full throttle John Paul was getting erratic results where different ones would peak at different times. He chaulked it up to a air leak in the motor. Not the case, same problem after overhaul. Other issue was blue staining around the injectors.

If a person looks close at a gami the area where the fuel comes out has a concave bowl shape where the lycoming injector is flat. At full throttle on my plane and some other Js it appears, we’re getting localized pressure inside the intake tubes which is same as outside atmospheric pressure and some fuel was actually spitting out the hole in the side of the gami injectors. In my case we think the concave shape of the gami face catches the very brief but present localized high pressure air created by the intake valve momentarily snapping shut just downstream of the injector.

The lycoming ones don’t do this. If we pulled the throttle back .5” of MP we could tune the gamis perfectly. That just seemed like a joke to me so we pulled the gamis and switched around the lyc injectors and got the spread to .3. Close enough. So I believe gami has some more homework to do on lyc planes with low restriction intakes such as the M20j.


Sent from my iPhone using Tapatalk Pro

Aaronk25 I did actually notice my spread went down at a reduced power setting. How many M20s have had that turbulence issue?  I did notice the injectors look different than the stock ones on the inside.   I think I found the smoking gun which I will know better once the ID is checked, but it stands to reason that the flow rates between 1&2 and 3&4 will be different at higher fuel flows (not as much at lower flows) due to the diameter difference.  I will know more for sure once my mechanic checks the IDs but John Paul of GAMI esssentially said this was a smoking gun. I’ve ordered four new injection lines of the same diameter so I can restart this process. Interestingly enough I also noticed my engine has the Ney Nozzle Mod which is made to help with (but not eliminate) the potential for Cam corrosion.

Edited December 30, 2018 by jkarch

[carusoam]

Change in dimension of the fuel lines can be an issue...

It is also a sign of odd parts being used... check the logs for proper part numbers...

Realistically, the fuel injector nozzles are the source of a calibrated pressure drop, larger than the fuel lines...

Measuring your Gami spread will directly tell you if the two fuel lines are causing a flow issue.

 

Some mis-guidance on Ney nozzles.... cam and follower oxidation are A problem of engines that don’t fly often enough... the Ney nozzles can’t do anything to help this out.

PP thoughts only, not a mechanic.

Best regards,

-a-

[PT20J]

Things don’t add up here. Engine roughness is caused by the cylinders delivering unequal power. Conversely, if it runs smoothly, especially LOP, the cylinder power, and thus the fuel flows, are balanced. 

The lines from the flow divider to the injectors are much larger than the injector nozzle and are not a factor unless severely clogged. It is possible for the valve in the flow divider to be damaged or hang up and cause unequal flows, but this should cause roughness or shut down issues.

It seems your engine operates normally, but you are concerned about what your instruments (EGT & FF) are telling you. Perhaps it’s an instrumentation issue. You could eliminate one source of confusion by cleaning and re-installing the stock injectors and rerunning the bottle test. 

Skip

[Shadrach]

I don’t wish to be the bearer of bad news, but I don’t think the size of your injector line matters. I have seen more than one Lycoming with different size lines injector lines. Having a slightly larger ID should not affect flow from injectors. As Anthony explained above, the size of the injector orrifice and pressure determine flow.  Think about a sink faucet, it does not matter if it a 1/2” or 3/4” water line, flow is determined by pressure and the degree to which the faucet opened. 

[N201MKTurbo]

5 hours ago, Shadrach said:

I don’t wish to be the bearer of bad news, but I don’t think the size of your injector line matters. I have seen more than one Lycoming with different size lines injector lines. Having a slightly larger ID should not affect flow from injectors. As Anthony explained above, the size of the injector orrifice and pressure determine flow.  Think about a sink faucet, it does not matter if it a 1/2” or 3/4” water line, flow is determined by pressure and the degree to which the faucet opened. 

It absolutely matters. The question is how much does it matter. We don't have to argue about it, It can be calculated. I don't know the IDs of the different lines, if we knew this we could run the numbers. The orifice in the injectors is ~0.038 diameter and ~0.125 length. 

Remember the poster's engine runs smooth. he is looking for a very subtle difference. He is a victim of the engine analyzer. Without the analyzer there wouldn't be a problem and everybody would fly along happily. 

[Shadrach]

2 hours ago, N201MKTurbo said:

It absolutely matters. The question is how much does it matter. We don't have to argue about it, It can be calculated. I don't know the IDs of the different lines, if we knew this we could run the numbers. The orifice in the injectors is ~0.038 diameter and ~0.125 length. 

Remember the poster's engine runs smooth. he is looking for a very subtle difference. He is a victim of the engine analyzer. Without the analyzer there wouldn't be a problem and everybody would fly along happily. 

No, he's not a victim of the data. You and I both know that an engine will run acceptably smooth with relatively lousy fuel distribution as long as all cylinders are somewhere in the 200 ROP to peak range.   I'm sure he would like access to the entire mixture spectrum and I can't imagine that much of a possibility with a spread of 1gph.

[Shadrach]

2 hours ago, N201MKTurbo said:

It absolutely matters. The question is how much does it matter. We don't have to argue about it, It can be calculated. I don't know the IDs of the different lines, if we knew this we could run the numbers. The orifice in the injectors is ~0.038 diameter and ~0.125 length. 

Remember the poster's engine runs smooth. he is looking for a very subtle difference. He is a victim of the engine analyzer. Without the analyzer there wouldn't be a problem and everybody would fly along happily. 

Can you post the formula you would use to make such a calculation?

[jkarch]

Does the Hazen-Williams equation apply to 100LL as it does to water?  I used an online calculator to calculate head loss and found with a pipe or 0.08” vs .06” at 10GPH the loss differential is 0.9PSI and at 5GPH the loss differential is 0.2 PSI. That applies to water though.  I found this on engineeringtoolbox.com.

 

John Paul of GAMI said the different line IDs can absolutely be the cause of my troubles. Another thing to notice is that cyls 3&4 are about 70 F LOP at 1&2 peak,all are up in the 1400s even when 1&2 peak. CHTs are all in range as well, in the 330-360 range this time of year at altitude (13K ft).

And yes, without pesky engine monitors I would be on my happy way. Sometimes too much information can be problematic.  The smaller 3&4 tubes do seem kind of ratty and convoluted compared to the 1&2.

-J

 

Edited December 31, 2018 by jkarch

[Shadrach]

15 minutes ago, jkarch said:

Does the Hazen-Williams equation apply to 100LL as it does to water?  I used an online calculator to calculate head loss and found with a pipe or 0.08” vs .06” at 10GPH the loss differential is 0.9PSI and at 5GPH the loss differential is 0.2 PSI. That applies to water though.  I found this on engineeringtoolbox.com.

 

John Paul of GAMI said the different line IDs can absolutely be the cause of my troubles. Another thing to notice is that cyls 3&4 are about 70 F LOP at 1&2 peak,all are up in the 1400s even when 1&2 peak.

And yes, without pesky engine monitors I would be on my happy way. Sometimes too much information can be problematic.  The smaller 3&4 tubes do seem kind of ratty and convoluted compared to the 1&2.

-J

 

Well then, I hope it solves your problem.  I look forward to the results as soon as you post them.  Hazen-Williams is used to determine friction losses over comparatively large distances (irrigation pipes, municipal water supply etc).  It has to do with discharge pressure at the open end of the pipe.  If pressure drop due to the minuscule differences in ID were an issue then surely differences in length would be an issue as well.  Are all of your injector lines the same length?

[jkarch]

Shadrach, Thanks for the note—in the case of this io-360 the spider is about 4 inches longer for  2&4. One interesting thing to note is that the equation is sensitive to 1/4 diameter so it is very sensitive to diameter changes. But this equation only works for water and neglects viscosity, etc.   I hope it fixes it too: I will keep you all posted once I get the information/ new lines installed.  Thanks again all! -J

[Shadrach]

13 minutes ago, jkarch said:

Shadrach, Thanks for the note—in the case of this io-360 the spider is about 4 inches longer for  2&4. One interesting thing to note is that the equation is sensitive to 1/4 diameter so it is very sensitive to diameter changes. But this equation only works for water and neglects viscosity, etc.   I hope it fixes it too: I will keep you all posted once I get the information/ new lines installed.  Thanks again all! -J

I suspect the equation works for water because Friction coefficient of water differs from other liquids. Well you may not be able to do the math for avgas, I would guess that the principal is the same. However I have to say that while I’m rooting for your success, I don’t think the size of the fuel lines is the problem.

Edited December 31, 2018 by Shadrach

[jkarch]

Let’s definitely hope it fixes it- the alternative is ugly! 

-J

[PT20J]

Fuel injection systems and carburetors are based on a couple of principles:

1) The flow rate across an orifice is a function of the pressure change across it.

2) In a pipe having multiple restrictions along it's length, the flow rate is determined by the smallest orifice.

The fuel flow rate to the engine is determined by the main jet. The fuel injector measures airflow and controls the pressure across the main jet to match the fuel flow to the air flow. Equal distribution to the cylinders is determined by the nozzles. The rest is just plumbing.

Bendix RSA fuel injectors are based on design principles from Bendix-Stromberg pressure carburetors used on the big radials. A great explanation of how these work (written by the man that designed them) is found in Aircraft Carburetion by Robert H. Thorner.

[Shadrach]

52 minutes ago, PT20J said:

2) In a pipe having multiple restrictions along it's length, the flow rate is determined by the smallest orifice.

You've eloquently stated what I've been unable to for the last several posts...

[N201MKTurbo]

1 hour ago, PT20J said:

 

2) In a pipe having multiple restrictions along it's length, the flow rate is determined by the smallest orifice. 

That is not true. The flow through a pipe will be determined by the sum of all the restrictions. The smallest will be the dominant one, but they all contribute.

I will find the exact formula when I get off work, or recover from my pending hangover. 

It is something like F= P/C * V where flow = pressure/ C * viscosity where C is the total restriction to flow and is calculated in smooth pipes as the diameter and length of individual sections.

[jkarch]

54 minutes ago, N201MKTurbo said:

That is not true. The flow through a pipe will be determined by the sum of all the restrictions. The smallest will be the dominant one, but they all contribute.

I will find the exact formula when I get off work, or recover from my pending hangover. 

It is something like F= P/C * V where flow = pressure/ C * viscosity where C is the total restriction to flow and is calculated in smooth pipes as the diameter and length of individual sections.

N201MKTurbo it certainly makes sense that any source of resistance along the way affects flow the way in electronics a 100 ohm resistor affects voltage drop in series with a 1K resistor. Since the flow nozzles are passively controlled devices as opposed to regulators they work in an open loop fashion. That means any upstream effects aren’t compensated for. The GAMIs are calibrated assuming straight tubes of the same size, and have grades that theoretically provide .2 to .4 gph of spread per grade. So certain things such as length or light bends can be nulled out with injectors, but extra flow restrictions might exceed the calibrated range.  Happy New Year all by the way!

Edited December 31, 2018 by jkarch

[carusoam]

A few things of interest....

• length, resistance 
• ID, resistance
• Viscosity of fuel, resistance
• Delta P, driving force
• speed, laminar vs. turbulent flow...
• Streamlined constrictions...
• selecting a flow model, Newtonian, or non-Newtonian...
• calculate stress rate and strain rate at the flow wall, viscosity = stress rate / strain rate
• Then simplify it all to be useable... Flow rate is a function of viscosity and flow geometry... similar to the V=IR proposed above...
• expect that the viscosity is low enough that turbulent flow can be assumed. The flow can be modeled like pressure flow of water in pipes...
• Keep in mind we see actual turbulent flow adding challenges caused by elbows in the line, before and after the FF sensors...
• Anyone want to calculate a Reynolds number for the fuel flowing in our distrubution system?
• All of this leads to an answer to ‘is Hazen-Williams applicable’....

When designing and selecting the parts, it is best to have everything to be low resistance until the injectors. This allows the injectors to be the tuning mechanism for flow....

Changing Anything, opens the owner up to questions of ‘is this OK?’

Realistically, the entire system’s flow gets tuned by using the gami spread.  NOT, just the tuning of the fuel injectors...

Once the system is tuned... changing anything with parts that are not like-for-like, will change the results...

This discussion gives us some insight why and how the FI system uses pressure that is five times higher than the carbureted system...

Chemical engineering ideas only, not just a PP today.  Still not a mechanic. 

Go MS!

Best regards,

-a-

[m20kmooney]

46 minutes ago, carusoam said:

selecting a flow model, Newtonian, or non-Newtonian...

The viscosity of Newtonian fluids is constant at a given temperature. Avgas is a Newtonian fluid. A non-newtonian flow model does not apply. 

Edited December 31, 2018 by m20kmooney