What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Anatomy of a High-Pressure Electric Fuel Pump.

"ground-adjustible relief valve dialed to 24PSI"

In other words, a regulator. :)

If it doesn't have something in it to set pressure (typically a spring) which applies pressure to a movable valve, it won't maintain a set pressure. A fixed orifice will result in varying pressure as fuel consumption changes, with flow going up when you want it decreasing, and decreasing when you want it going up.
 
The way these work is that they strive to achieve the the Pressure commanded by the regulator. You read the chart by looking at the PSI set by the regulator
then observe the flow rate at that PSI (e.g. x GPH @ y PSI). The pump will keep trying to deliver the PSI set by the regulator, all the way down to a 0 flow.

Larry

I'll be clearer: 22psi at a full throttle, sea level condition of an O360 (16-17 gph) is not adequate. Yes, the engine will run, but the engine driven fuel pump will deliver in the high 20's psi at that flow rate.

The second part of my question was the apparently incorrect or incorrectly obtained data by the manufacturer. The graphs of 12 and 13.5 volts should not cross, unless there is some new mystery to DC motors and constant displacement pumps that I've not heard of.
 
I'll be clearer: 22psi at a full throttle, sea level condition of an O360 (16-17 gph) is not adequate. Yes, the engine will run, but the engine driven fuel pump will deliver in the high 20's psi at that flow rate.

The second part of my question was the apparently incorrect or incorrectly obtained data by the manufacturer. The graphs of 12 and 13.5 volts should not cross, unless there is some new mystery to DC motors and constant displacement pumps that I've not heard of.

Alex- I agree completely with everything you say here. With both pumps operating, which will be standard proceedure for TO&climb, I should have ample fuel flow with a huge surplus. I can also operate both pumps continuously anytime and for however long I wish, so would do so during aerobatics as an example. In fact, the only real reason for turning one off would be loss of the alternator, but I?ll probably run on one pump during cross country cruise just to lighten the load on the alternator a bit.

This installation will be analogous to dual mags- two for normal ops but one will get you by with a little judicious management. I would anticipate a similar rate of serious malfunction as well(quite rare). Just as with mag checks, runnups will include verfifcation that each pump is fully operational, and I would no more take off with one pump inop than I would ignore a bad mag check. The redundancy of this pair of pumps will also equal that of dual mags. Each pump will be powered through an on-off-on switch with direct, full-time access to either of the two essential loads busses, each of which will be hardwired to one of the twin batteries. Yes, you must turn off all pump and ignition switches as the essential loads busses will be ever-hot, but I have a good solution for insuring that.

The graph is from Walbro, and I agree that it does not make sense. I will be be doing extensive bench testing with Stoddard solvent before istalling this particular pump, and will report back with my own data. Thank you for your input!- Otis
 
Alex- I agree completely with everything you say here. With both pumps operating, which will be standard proceedure for TO&climb, I should have ample fuel flow with a huge surplus. I can also operate both pumps continuously anytime and for however long I wish, so would do so during aerobatics as an example. In fact, the only real reason for turning one off would be loss of the alternator, but I?ll probably run on one pump during cross country cruise just to lighten the load on the alternator a bit.

This installation will be analogous to dual mags- two for normal ops but one will get you by with a little judicious management. I would anticipate a similar rate of serious malfunction as well(quite rare). Just as with mag checks, runnups will include verfifcation that each pump is fully operational, and I would no more take off with one pump inop than I would ignore a bad mag check. The redundancy of this pair of pumps will also equal that of dual mags. Each pump will be powered through an on-off-on switch with direct, full-time access to either of the two essential loads busses, each of which will be hardwired to one of the twin batteries. Yes, you must turn off all pump and ignition switches as the essential loads busses will be ever-hot, but I have a good solution for insuring that.

The graph is from Walbro, and I agree that it does not make sense. I will be be doing extensive bench testing with Stoddard solvent before istalling this particular pump, and will report back with my own data. Thank you for your input!- Otis

My concern was the same as Alex's; that was why I posted my T/O numbers. This clarifies things.

Question: I presume the hexagonal cylinders on top of the pumps are the relief valves (or pressure regulators or..)? How does one set them? I'm familiar with external fuel pressure regulators from various automotive EFI projects over the years, for example: https://www.aeromotiveinc.com/product/universal-bypass-regulator/ but those remind me of a typical fixed-spring relief valve like one might see in industrial applications.

As for Airtex pumps, the E-2315 in my AFP pump assembly failed at about 650 hours of extremely intermittent use. I was surprised and disappointed. I have a Bosch external inline pump mounted inside the frame rail on my ground bound 4x4 that has been dead reliable for almost 20 years and 120,000 miles of abuse, mounted inside the frame rail and exposed (needlessly, I've been too lazy to move it) to a hot exhaust system and occasionally packed with mud (thus thermally insulated). Replaced with Walbro GSL393 which, as a bonus, is much quieter.
 
My concern was the same as Alex's; that was why I posted my T/O numbers. This clarifies things.

Question: I presume the hexagonal cylinders on top of the pumps are the relief valves (or pressure regulators or..)? How does one set them? I'm familiar with external fuel pressure regulators from various automotive EFI projects over the years, for example: https://www.aeromotiveinc.com/product/universal-bypass-regulator/ but those remind me of a typical fixed-spring relief valve like one might see in industrial applications.

As for Airtex pumps, the E-2315 in my AFP pump assembly failed at about 650 hours of extremely intermittent use. I was surprised and disappointed. I have a Bosch external inline pump mounted inside the frame rail on my ground bound 4x4 that has been dead reliable for almost 20 years and 120,000 miles of abuse, mounted inside the frame rail and exposed (needlessly, I've been too lazy to move it) to a hot exhaust system and occasionally packed with mud (thus thermally insulated). Replaced with Walbro GSL393 which, as a bonus, is much quieter.

The relief valves can be fine tuned to a degree by shimming to compress the spring slightly, or shimming the cap to reduce spring pressure. Altering the length of the body or the spring can do the same. Swapping out different springs can produce larger adjustments, but Airflow performance relief valves can be expectedto be pretty darn close to target without modification.

I was told today by a Walbro tech that their pumps are NOT, in fact, roller vane but geared, which could explain their higher reliability and their superior self-priming ability. Within a few weeks I will be doing some extensive bench testing on the GSL395 to determine its suitability for Bendix-type FI systems in a dual pump installation, and I will autopsy that pump to verify what I was told today, so at some point there will be an”anatomy a fuel pump, part II” thread.

Btw- the Walbro tech agreed that the published graphs for GSL395 performance are flawed. For example, he looked at the raw data for the 20PSI pressure point, and verified that at 12V flow would be 29GPH but over 40GPH AT 13.5V. The graphs indicate anout 27GPH at 12V and just 19GpH @ 13.5V, which is absurd!

I won’t be bench testing any short-loop recirculation setups because I will be installing fuel return lines to my tanks, but I will heat the Stoddard solvent to 100°F as part of the testing to determine effects of absolute fuel temp on flow dynamics. If I determine that the GSL 395 is inadequate, The obvious choice wil lbe the more powerful, higher pressure GSL393. My ultimate objective is to find the most efficient suitable pump for my needs, with emphasis on range under battery power. Obviously, energy used to push fuel through a relief valve is wasted, so using a 100+PSI pump pushing 40GPM, such as the Delphi obtained from AP or the walbro GSL393 to fuel my 24PSI, <10GPM requirement at most times is contra-indicated. The concern about the GSL395 isthat flow drops off pretty precipitously below 30PSI, so I’ll be focusing upon the 25PSI data point. The Walbro rep did tell me that, if I find performance satisfactory, there would be no harm in operating the GSL395 continuously at 25PSI.- Otis
 
Last edited:
I'll try to find time to cut open this bad 393 I have here and post a photo. I'm curious about what's inside too.
 
I'll try to find time to cut open this bad 393 I have here and post a photo. I'm curious about what's inside too.

Ross- That would be most helpful- I was one of those kids who revelled in taking things apart to see how they work(or don?t), and still find that technique hard to beat!-Otis
 
The relief valves can be fine tuned to a degree by shimming to compress the spring slightly, or shimming the cap to reduce spring pressure. Altering the length of the body or the spring can do the same. Swapping out different springs can produce larger adjustments, but Airflow performance relief valves can be expectedto be pretty darn close to target without modification.

I was told today by a Walbro tech that their pumps are NOT, in fact, roller vane but geared, which could explain their higher reliability and their superior self-priming ability. Within a few weeks I will be doing some extensive bench testing on the GSL395 to determine its suitability for Bendix-type FI systems in a dual pump installation, and I will autopsy that pump to verify what I was told today, so at some point there will be an?anatomy a fuel pump, part II? thread.

Btw- the Walbro tech agreed that the published graphs for GSL395 performance are flawed. For example, he looked at the raw data for the 20PSI pressure point, and verified that at 12V flow would be 29GPH but over 40GPH AT 13.5V. The graphs indicate anout 27GPH at 12V and just 19GpH @ 13.5V, which is absurd!

I won?t be bench testing any short-loop recirculation setups because I will be installing fuel return lines to my tanks, but I will heat the Stoddard solvent to 100?F as part of the testing to determine effects of absolute fuel temp on flow dynamics. If I determine that the GSL 395 is inadequate, The obvious choice wil lbe the more powerful, higher pressure GSL393. My ultimate objective is to find the most efficient suitable pump for my needs, with emphasis on range under battery power. Obviously, energy used to push fuel through a relief valve is wasted, so using a 100+PSI pump pushing 40GPM, such as the Delphi obtained from AP or the walbro GSL393 to fuel my 24PSI, <10GPM requirement at most times is contra-indicated. The concern about the GSL395 isthat flow drops off pretty precipitously below 30PSI, so I?ll be focusing upon the 25PSI data point. The Walbro rep did tell me that, if I find performance satisfactory, there would be no harm in operating the GSL395 continuously at 25PSI.- Otis

'geared' : Gerotor

Your statements about flow vs pressure seem to contradict reality. Given a non-changing supply voltage, a positive displacement pump's flow will not drop with reduced pressure; it will drop with increased pressure. As long as the pressure isn't above the pump's design point, the pump itself will flow at least the quantity of fuel shown in its ratings.

Charlie

http://www.pumpschool.com/principles/gerotor.asp
 
'geared' : Gerotor

Your statements about flow vs pressure seem to contradict reality. Given a non-changing supply voltage, a positive displacement pump's flow will not drop with reduced pressure; it will drop with increased pressure. As long as the pressure isn't above the pump's design point, the pump itself will flow at least the quantity of fuel shown in its ratings.

Charlie

http://www.pumpschool.com/principles/gerotor.asp

Charlie- I do not see anything I’ve suggested that contradicts what you are saying here, but I think I understand the confusion. I was referencing the erroneous nature of the performance chart published by Walbro for the GLS395. 30PSI is indeed where the flow falls to zero for this pump even at 13.5V, so my task is to insure that sufficient flow remains at 25PSI for my needs(with two operating in parallel for high-flow operations). If you go all the way down to 20PSI there is ample flow for my needs with even one pump, but I don’t want to push the Bendix- type fuel metering valve to its lower limit, so will target an operation pressure of 24PSI per AP instructions. Clearly, the GSL395, even two of them operating in parallel, would not suffice For any EFI system I am aware of.- Otis
 
Last edited:
"The concern about the GSL395 is that flow drops off pretty precipitously below 30PSI" and related statements are what I was referring to.

Are you trying to say that it drops off as pressure *increases*, approaching 30 psi? The wording sounds like you're saying the opposite, and is reinforced with the statement following that it 'would be ok to operate at 25 psi'.

FWIW, I suspect that your best path to lower power consumption at needed flow would be to move to in-tank turbine pumps. The ones I've researched seem to have significantly lower current requirements than gerotor or roller-vane pumps for the same flow/pressure, and have the added bonus of being easily controlled with PWM, which others have mentioned. With PWM, power required varies with actual flow/pressure, so low engine power would require less pump power.

Charlie
 
Walbro Autopsy

Ok, I found a few minutes today to get my pipe cutter out and slice open a Walbro GSL 393 pump returned by a customer a couple months ago.

I first checked to be sure it didn't pump and it didn't, as he had said. Motor ran nicely with no unusual noises but no pumpy. This is the first Walbro failure I've seen first hand after using and selling these for 20 years. (hundreds). I had another customer report 2 failures after only a few hours but he'd mounted the pumps vertically with the suction end facing up, which is a no-no.





The pump is quite well made inside and as Otis reported earlier, these are not roller vane types as I have read elsewhere but in fact a gerotor type which is probably a better design for run dry scenarios. The gears are hardened ferrous material riding inside the black housing which is slightly magnetic but appears to be some sort of graphite/ metallic composite material (just guessing).



You can see Walbro went to some lengths to balance the motor armature- nice touch for longevity. On the inlet side, they have a fine plastic mesh screen of large area to prevent trash from getting to the gears. Another nice touch so they don't rely on external filters or tank socks solely to keep debris out. Good to know.



There is a pressure relief valve (white plastic button in photo). I believe these are set to relieve at 110 psi.



It's solidly constructed and has long brushes. It's a nice design IMO. As I said before, I had one of these on our shop car for 19 years and maybe 5000 hours. No issues.

Now the strange thing here is, I couldn't find anything wrong inside this pump. The drive tangs from the motor shaft engages the inner gear holes, nothing was slipping on the shaft, all the other shims and gaskets were intact and can't spin as that stuff is all held in place with the two Torx screws. I'll have to take a better look when I get some spare time.

Not shown in the photos is the aluminum outer pump housing the magnet housing above slides into.

Always interesting to see what's inside a mechanical device.
 
Last edited:
I don't know if it's disappeared with the change in ownership, but both sales and technical pages on the old Walbro website were quite explicit on the GSL series being gerotor pumps.
 
I'll try to find time to cut open this bad 393 I have here and post a photo. I'm curious about what's inside too.

Ross- THANK YOU for the photos- they are most helpful! The reduced parts count alone argues for the georotor over roller-vane types, so I’m greatly encouraged.

I’m really looking forward to thorough bench-testing of the GSL 395, because it would yield 35-45% greater range on battery power than the 393, but that will be my #2 choice. I’m tied up with other projects for a few weeks but will report back on my findings. I’m almost tempted to install one of each, but that would probably preclude operating them simultaneously for fear of inadequate flow through the 395 for cooling. On the other hand, it would be pretty easy to lockout simultaneous operation with a pump selector switch.- Otis

Ps- I’ve decided to cut up one of the Delphi’s pictured in my original post and will add photos of that to the collection we are building here- O.H.
 
Last edited:
Another pump option and a new insight-

From what I can tell, the GSN Walbros are indeed the gold standard for in-line pumps for RV?s. Ross Farnham mentioned another Walbro model in a PM, the GSN 414, which looks even more promising than the 395 I?ve been considering as a solution to the most electrically-efficient choice for bendix-type FI aircraft. The performance chart I?ve seen for it, shown below, is clearly flawed as pump curves are never straight lines, but encouraging enough to have ordered one so I can bench test its performance. If the chart is even close, This pump holds the promise of well under 3 Amps continuous consumption at 25 PSI, which would really boost range after an alternator failure vs the 4.5-5Amps of the Delphi supplied by AP. It evidently also produces respectable flow at pressures that could even work with some EFI systems. I?ll be comparing performance of the Delphi, a Walbro GSN395, and the Walbro GSN414, measuring current and flow rates at 25, 30, and 40psi at three different fuel temps, and will publish the results in a new thread in a month or so. I?ll also disassemble all three because I plan on torturing them a bit as well, disqualifying them for installation on an airplane.
2v2EKmUsqxBELK5.jpg


The insight came from pondering the built-in relief valve Ross pointed out in the 393 above. I talked again to the tech advisor at TI Automotive. I was able to determine that all of the GSL?s have them, and their purpose is to prevent them from self-destructing in a no-flow condition when mounted in-line in vehicles without any means of returning fuel to the tank. They are set to the maximum rated pressure for each pump and simply shunt fuel internally back into the non-pressurized side of the pump section.

This is OK up to a point, but it would never be a good idea to rely on this for an airplane, which still require return lines like those coming off Ross?s regulators or off my external relief valves. They certainly would not tolerate operation with zero throughput for long without damage, and this relief valve would never actually open at all in a properly equipped aircraft. I suspect that any regular use of the built in relief valve would shorten pump life, as it would get hot in there very quickly at zero or near-zero throughput and the high-amperage that would be required to activate the relief valve.- Otis
 
The current draw would be so high at the pressure relief setting that I think you'd blow the 10-15 amp fuse/ breaker in a few seconds anyway. They are just there to prevent immediate destruction if flow was stopped for some reason. At some point, the motor would stall due to load. That wouldn't be a good thing either.
 
I installed one of the Airflow Performance pumps from the OP last year. The pump may be good, but the wire terminations that came on the pump were trash. One of the crimps failed after 50 hours flying time simply from loosening the pump to check the filter. The wire pulled out of the crimp. The 2nd wire pulled out with very little effort. The terminals were about the cheapest you can buy, WalMart quality. The crimps appear to have been made with pliers or something else entirely unsuited for the job, and with no attention to quality. My mistake for not checking them. Airflow Performance should simply leave the leads off the pumps. If you have the leads, throw them in the trash. I can't believe they used these on a $900 pump.
 
Last edited:
I installed one of the Airflow Performance pumps from the OP last year. The pump may be good, but the wire terminations that came on the pump were trash. One of the crimps failed after 50 hours flying time simply from loosening the pump to check the filter. The wire pulled out of the crimp. The 2nd wire pulled out with very little effort. The terminals were about the cheapest you can buy, WalMart quality. The crimps appear to have been made with pliers or something else entirely unsuited for the job, and with no attention to quality. My mistake for not checking them. Airflow Performance should simply leave the leads off the pumps. If you have the leads, throw them in the trash. I can't believe they used these on a $900 pump.

Something odd here. You buy it directly from AFP? Cut away the heat shrink to see the crimps? Got photos?

I'm pretty sure Don would want to investigate, and the rest of us are curious.
 
Back
Top