What's new
Van's Air Force

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

Thoughts on EFI fuel redundancy

htx9a

Member
I'm planning for an auto conversion with EFI that requires around 45 PSI to the engine-attached pump and I'm wondering what is typically done in terms of redundancy. With the bottom of the wing tanks at the same level as the bottom of the fuselage, there's no way to gravity feed into a common header so it's a bit awkward.

The current plan is to use dual transfer pumps (Facet piezoelectric or similar) to push fuel from the left or right tank into a small header tank, and then two in-header auto pumps to get to the 45 PSI.

The header would initially have air that would need to be purged, so I've added a small orifice to return air/fuel to the source wing tank via the duplex valve. Each pump has a check valve to prevent backflow when only one pump per pair is on. All pumps would be on during takeoff/landing.

There is also an orifice at the output of the header pumps to prevent over-pressure from the EFI rail after engine shutdown.

Takeoff/landing power draw with all pumps would be about 8 A and half that at cruise.

Am I over-complicating this too much, or have others done similar configurations? Is there an in-line 45 PSI pump that can pull 12"+ that won't require a large amount of circulating fuel and current draw? Getting rid of the transfer pumps would be ideal.

fuel_system.png
 
Necessary?

I am running an EFII system in my -10. Full flow return to tanks, two Walbro racing pumps in a manifold. No extra pumps or header tank. System automatically reverts to second pump at loss of fuel pressure or it can be manually selected. Works great.
 
I ran a system as depicted in the graphic for about 10 years, worked fine but was heavier and more complicated than required.

These days, we usually dispense with the header tank and low pressure pumps. Just mount the pumps low on the floor, avoid 90 degree fittings prior to the pumps and it works fine. Most electric pumps have a pressure relief valve internally so the orifice isn't required either.
 
Similar to these? https://walbrofuelpumps.com/walbro-gsl393-fuel-pump.html
Looking at the chart, that's around 5A draw (not bad) for 45 PSI but also at 42 GPH which is an order of magnitude more than the engine consumes.

Do these pumps have an integrated pressure regulator? If I'm understanding correctly, "full flow return" means you're circulating 30+ GPH and the resulting pressure is due to the diameter and length of fuel line in the circuit. Does the high flow rate have to do with vapor-lock, or is that a carbureted engine issue?

Is it uncommon to not circulate fuel and instead vary pump RPM to regulate pressure? I'm reading that recent automotive pumps operate in this way but they all seem to be in-tank, which lead to the addition of a header tank in my diagram.

Sorry for all the questions, I'm just trying to figure out why things are done they way they are, and if I want to deviate.
 
Similar to these? https://walbrofuelpumps.com/walbro-gsl393-fuel-pump.html
Looking at the chart, that's around 5A draw (not bad) for 45 PSI but also at 42 GPH which is an order of magnitude more than the engine consumes.

Do these pumps have an integrated pressure regulator? If I'm understanding correctly, "full flow return" means you're circulating 30+ GPH and the resulting pressure is due to the diameter and length of fuel line in the circuit. Does the high flow rate have to do with vapor-lock, or is that a carbureted engine issue?

Is it uncommon to not circulate fuel and instead vary pump RPM to regulate pressure? I'm reading that recent automotive pumps operate in this way but they all seem to be in-tank, which lead to the addition of a header tank in my diagram.

Sorry for all the questions, I'm just trying to figure out why things are done they way they are, and if I want to deviate.

No integrated regulator. My EFII system uses an aeromotive pressure regulator set to about 35 psi static. You may want to look at drawing #3 in this link; it is the setup without a header tank. Drawing #2 is setup with a header tank.

https://flyefii.com/media/System32-Installation-Manual-rev-6-19.pdf
 
I doubt you want a variable frequency drive for your fuel pump. It’s one more thing to FEA. Not just the VFD, every system related device like sensors for input, etc. then you’d need to determine fail-safe operational modes. The list grows and grows. My advice would mirror most here, i think. Keep it simple. Stick to proven systems as much as possible. Listen to people like Ross.
 
In the EFI systems we are talking about here, fuel pressure is held at a constant differential above MAP by varying the return fuel volume at the regulator. The pumps put out a constant volume at X voltage.

Walbro only approve PWM on a couple of their pumps and specifically recommend not use PWM on most of their older designs.
 
Ideally the fuel pumps will push quite a bit more fuel than is needed with the excess going back to the tank via a pressure regulator that holds a fixed backpressure on the system and allows the rest to return to the tank. The Borla regulator as sold by SDS is very good for this, with a pressure tap on it to reference the fuel pressure to the manifold pressure. With this the fuel pressure actually rises and falls as the manifold pressure rises and falls with throttle use, with the result that the pressure delta across the injector face is always the same no matter how much power you are making.

Remember that the pumps are cooled by the same fuel they are pumping - so you want to return the fuel to the tank and not the suction inlet of the pumps. You don't want to keep recircing fuel in a loop gaining heat each time it passes through the pump.
 
Pressure regulators

It seems both pressure regulators sold by EFI and EFII work well and I am extremely pleased with my system. Their is some caution that needs to be looked into. Make sure you have independent breakers on each fuel pump. Make sure the cut-off pressure is programmed correctly (switches over pumps if the system goes below X) and make sure you can supply electrons with 100% or close reliability. (2 busses, 2 batteries, 2 alternators, that truly are independent) and can feed the alternate bus if required. They both are an amazing system. I can tune my system to +/- 10 F EGT differences between cylinders at the full range of operating MP's. I think that must improve fouling issues and improve long term maintenance on my engine. I even have pulled the breaker (On the ground) at 24 square for the primary fuel pump and the engine never missed a beat. ONE issue that can occur is with pumps that pull these kinds of flow rates the slightest leak in the suction side (air) produces very small bubbles that vibrate the pressure regulators seat (And causes deformation of the seat) that reduces fuel pressure. The motor seems to run fine even 10 psi lower than set-point but eventually this will hurt the reliability. (Or at some point will). A leak on the suction side does not produce the tell tail blue stain or even smell as nothing is leaking outside the system. The minimal head pressure when the system is off is not enough to cause a leak that you can see or smell. (Ask me how I know) What I did to eventually find a leak was install a clear line after the fuel rail and visually watch for bubbles or place the hose in a open container and see if their are any bubbles. Once I saw the bubbles I was able to troubleshoot the problem. Again very nice systems but some due diligence required.
 
What about the engine attached pump needing 45 psi? So I'm guessing that its not a mechanical, but a secondary electric pump mounted on the engine. You could get a Romec mechanical pump and probably crank the pressure to 45 psi---but whether is takes a special drive is out of my knowledge---there was a RG17980J (?) that was used on a IO360---according to a Lycoming SB.

I'm betting that you arent using a mechanical pump

Tom
 
Definitely message received on the keep-it-simple, redundant pumps + electrical systems advice. I'm a software and electrical engineer who designs electric vehicle inverters so I'm very familiar with what is possible, but there is merit in the simplicity of a mechanical switch and a DC motor driven pump with mechanical bypass regulator.

I mentioned earlier this is an auto conversion, and being direct-injected it actually does have a mechanical engine-mounted ~2000 PSI pump that likes inlet pressure in the 40-45 PSI range.

As a compromise, instead of attempting to regulate pressure, on one pump I could use a DC-DC buck converter at a fixed set-point. The switch to the "eco" pump would have a backup position that bypasses the DC-DC and just sends full battery voltage (with a second fuse if the DC-DC was the source of failure). Pump #2 just gets full voltage (on power rail #2 with fuse).

Part of the reason that these external pumps need to circulate so much fuel is they are wasting 60%+ of the power. If they could be run more slowly, a much lower flow rate would be sufficient for cooling.
 
The under appreciated BPR

Definitely message received on the keep-it-simple, redundant pumps + electrical systems advice. I'm a software and electrical engineer who designs electric vehicle inverters so I'm very familiar with what is possible, but there is merit in the simplicity of a mechanical switch and a DC motor driven pump with mechanical bypass regulator.

I mentioned earlier this is an auto conversion, and being direct-injected it actually does have a mechanical engine-mounted ~2000 PSI pump that likes inlet pressure in the 40-45 PSI range.

As a compromise, instead of attempting to regulate pressure, on one pump I could use a DC-DC buck converter at a fixed set-point. The switch to the "eco" pump would have a backup position that bypasses the DC-DC and just sends full battery voltage (with a second fuse if the DC-DC was the source of failure). Pump #2 just gets full voltage (on power rail #2 with fuse).

Part of the reason that these external pumps need to circulate so much fuel is they are wasting 60%+ of the power. If they could be run more slowly, a much lower flow rate would be sufficient for cooling.

Was going to start a new thread here to avoid the drift but the topic swung to this so I think it’s appropriate.

BTW, &**&@! you smart contributors. I don’t often sleep for any extended periods of time and when I read something interesting, it keeps bumping around in my skull and makes things worse. The FWF section of this forum is good for that as I’m trying to learn. I anticipated the direction of this thread and the OP just beat me to it. Big disclaimer; I’ve never designed a gasoline system but I have designed liquid fuel and natural gas fuel systems for turbines. They all play in the same area of physics.

To the OP; yes, I hate waste and inefficiency as it appears you do. Sometimes there’s no easy way around it. Would be great to eliminate or greatly minimize the fuel return flow as it’s wasted energy. There’s always going to be some with a positive displacement pump. 35GPM @ 35psig (static) takes 0.7 theoretic HP. Pump inefficiencies, frictional losses increase that obviously.

The role of the Back Pressure Regulator is probably decently understood by most. The operating conditions maybe less so. It can lead a very harsh life. The flow through the BPR is obviously less full system flow, most of the operating time. The excess flow has to be “let down” (pressure reduced) between the valve seat/plug for return to the tank. Despite the normally reduced flow, the local velocities here are even higher due to the small flow area between the aforementioned valve seat/plug. This results in a very low (local) static pressure. The conditions can let the liquid then “flash” to vapor then cavitate (collapse when phasing back to liquid) when the Ps is recovered downstream of the BPR plug/seat. It is extreme violence on a micro scale. The valve materials can erode very quickly if this happens rendering the BPR unable to do it job and/or liberating DOD into your fuel system. This condition has to be avoided under all operating conditions.

The result = higher return flows are needed to keep the BPR "less closed", to keep the local velocities low enough, which keeps the local static P up, to keep the liquid from flashing to gas, all from full return flow to full power return flow (35gph versus ~ 15 through the BPR).

The BPR is probably an under-appreciated piece of engineering, IMO. Bless those under appreciated design engineers. Never apply such a device outside of it’s OEM rating.

The pump inlet pressure requirements stated by the OP is called NPSHr (Net Pressure Suction Head required). This is to keep that pump from cavitating. The same type of damage can occur. The arrangement you’ve conveyed so far is NOT redundant like the legacy fuel systems on aircraft. Those are fully redundant, series pumps with fail-safe bypass modes. Please keep asking questions. The system and related subject matter is probably a little more complicated and complex than it appears.

Tried to shamelessly copy some pix of valve and pump cavitation damage from the net. Dodn't see one for a PD pump so used a centrifugal one instead. This are pix from industrial grade equipment with margins far beyond aircraft components. Build safe, Sir.
 

Attachments

  • valve-cavitation-damage.jpg
    valve-cavitation-damage.jpg
    18.6 KB · Views: 188
  • pump cav.jpg
    pump cav.jpg
    125.7 KB · Views: 146
Last edited:
There’s always going to be some with a positive displacement pump. 35GPM @ 35psig (static) takes 0.7 theoretic HP. Pump inefficiencies, frictional losses increase that obviously.... 35 GPM flow is already flat screaming in 3/8” tubing; around 100 mph.

You're off a bit, methinks. We're talking a 35 gallons-per-hour system, not gallons-per-minute ;)
 
You're off a bit, methinks. We're talking a 35 gallons-per-hour system, not gallons-per-minute ;)

Corrected. Been awake since 0230 and really feeling it while trying to use my head during my lunch break. I don't get to engineer much anymore. Throw my idiocy on top of that and bad stuff happens. Aren't you supposed to keep track of units during calcs to avoid dumb stuff like this? Guess I've never learned. Thanks.
 
Last edited:
I mentioned earlier this is an auto conversion, and being direct-injected it actually does have a mechanical engine-mounted ~2000 PSI pump that likes inlet pressure in the 40-45 PSI range.

It will be interesting to see how this works out with direct injection.
 
fuel_system_EFI.png


I let the previous post about cavitation-induced pitting/erosion simmer in my brain for a bit and came up with a thought. If normal operation for the BPR is to be mostly-open, then what about using an appropriately-sized orifice downstream that allows a bit of return flow while also maintaining a low drop across the regulator, say 10 PSI delta while the orifice sees full pump pressure.

In the non-eco mode the pump(s) would need a higher-flow escape so we can add a pressure-relief valve in parallel set to around 50 PSI.
 
…and

It is amazing that the forum “experts” aren’t criticizing this system more for being unnecessary and complicated with multiple potential failures points.

Just see the threads on electrically dependent architecture…
 
I let the previous post about cavitation-induced pitting/erosion simmer in my brain for a bit and came up with a thought. If normal operation for the BPR is to be mostly-open, then what about using an appropriately-sized orifice downstream that allows a bit of return flow while also maintaining a low drop across the regulator, say 10 PSI delta while the orifice sees full pump pressure.

In the non-eco mode the pump(s) would need a higher-flow escape so we can add a pressure-relief valve in parallel set to around 50 PSI.


Valve % open/close is a very relative thing. effective flow area is a combo of eth aforementioned and various, associated component sizes. As previously stated, it's more complex than that. If the conditions were consistent, I could size an appropriate orifice in a couple of minutes. It's not appropriate, IMO. See below.

It is amazing that the forum “experts” aren’t criticizing this system more for being unnecessary and complicated with multiple potential failures points.

Just see the threads on electrically dependent architecture…


The entire approach is worrisome to me. I hinted at it in an earlier post. All “established” aircraft fuel systems have a fair degree of system redundancy; even the designs that are many decades old. The schematic here, as I interpret it, (can only see on my phone versus Chrome browser for some reason) is adding unnecessary complexity/failure modes while not addressing the most critical component.

The engine driven main fuel pump is a highly strained device (direct injection here). At It is operating under some very harsh conditions. Extreme pressures associated with contacting surfaces in true positive displacement pumps (citation needed, assumptions made on pump application) and the relatively very low lubricity of gasoline are a recipe for a high failure rate. I inquired/verified the suspected failure rate of such pumps in the automotive world with someone here, off-line. The story isn’t good. If required to use aviation fuel (probably not the OP’s plan) versus MoGas would make things worse. Throw in the fact that the application to aviation probably won’t be validated outside of flight time, the approach gets more alarming to me.

Tons of other factors to consider.

Off my soapbox.
 
Last edited:
Back
Top