EAAAirVenture OshkoshShopJoin

Gravity vs Pump Assisted Fuel Systems

Posted By:
Dan Morehouse
#1 Posted: 2/13/2010 23:43:58



I’d like to get feedback about the differences between a gravity fuel system and one assisted by a pump or two. Here is my particular situation:

Background and Tests

I am building a Sonex (VFR only) with a 120 hp 3300 Jabiru engine. The fuel tank is under the glare shield. Most people run a strictly gravity system and remove the mechanical fuel pump. For my tests, I’ve elevated the engine 20 degrees and put 2 gallons of fuel in the tank. This leaves only 6” between the level of gas in the tank and the level of the intake manifold. The fuel exits the hose at the intake manifold level in each test. I have a finger strainer in the tank and a gascolator (with a fine mesh filter) at the low spot on the hot side of the firewall. 3/8” hose has been used throughout. All unnecessary 45 and 90degree fittings have been removed.

Flow tests through this system were 20 GPH. WOT fuel flow for the 3300 has been reported to be 10-12 GPH. So there’s plenty enough reserve to meet the 150% rule stated in the FAR. But this is without installing a fuel flow robber…er, I mean fuel flow sensor. I get only 8.5 GPH when this is added. I will repeat the tests when a larger, less restrictive fuel flow sensor arrives. I expect this will improve flow rates, but doubt they will increase to15-18GPH, the minimum I’d need for 150%.


Reports from Others

There are several people flying with 3300’s in their Sonexes using a gravity fuel system like I’ve described, INCLUDING the flow sensor. Most say they have never experienced a problem. One person reported a performance drop when he added a fuel flow sensor. He has since removed the sensor.


Other people have added the mechanical pump to the system. Some have added the mechanical pump by itself, but most have added an electric backup pump (usually a Facet) to the system as well as a fuel pressure sender.  The mechanical pump is reported to put out ~5 psi and the Facets only 2-4 psi.


If the mechanical pump system is added to my flow sensor equipped system, it seems I would want an electric backup in case of mechanical pump failure. Most people with a system like this have the electric pump plumbed in series with the mechanical pump, turning on the electrical pump only for takeoffs and landings.


There are also people that have chosen to add either a mechanical or electrical pump to an already adequate gravity system. They had tested more than150% of their WOT fuel flow rate by gravity alone and chose to add a backup pump to increase the systems dependability. For example, there is one Kitfox owner I know of that installed an electric Facet backup pump. He turns it on only for suspected fuel starvation emergencies.



Assuming that fuel flow tests through the larger fuel flow sensor gives me 10 GPH (~80-100% ‘takeoff’ flow rate) here are my choices:


  1. Remove the flow sensor. Gravity only.
  2. Remove the flow sensor and add either the mechanical or electric pump, assuming the 150% rate can still be obtained through a failed or turned off pump.
  3. Keep the flow sensor and add both pumps plus a pressure sender. The mechanical pump runs continuously, electric pump for backup.


Option 1, gravity with no fuel flow sensor, would be the simplest. But I wouldn’t have the advantage of the data provided by a flow sensor.


Option 2, one pump only, would be the next up the complexity scale. Adding either pump would not be a major undertaking. And each pump weighs about a pound. Not trivial, but not like an extra battery either. Choosing one pump over the other is not obvious to me.


Option 3, the whole 9 yards, is the most complicated, but still easily doable. It adds 2.5-3.5 pounds and ~$100 to the gravity only system.



Which of these systems is the most dependable? A gravity only system doesn’t have a pump to fail, leak, or squirt gas. But a system with pumps might provide the extra pressure needed to overcome a clogged filter, a plugged vent, or a tank with only a few ounces remaining. And I would have fuel flow data.


I look forward to hearing what your insight is.





David Darnell
#2 Posted: 2/14/2010 00:28:03

  Well, in my limited experience, have never seen gravity completely fail. Can't say the same for fuel pumps. About the only time I'd be worried about gravity is if you do lots of negative g and inverted flying.

Ron Wanttaja
#3 Posted: 2/14/2010 11:19:38

 I'm a huge fan of simplicity...the more simple a system is, the less there is to fail.  If you need fuel pumps to maintain fuel flow, you're more vulnerable.  A pump is going to be of limited help in any sort of flow-blockage situation.  With the addition of backup pumps, you've just made the pilot's job more complex, too... seventy feet up, heading for the trees at the end of the runway, is no time to be fiddling with an "emergency only" backup fuel pump.

Is it possible to install a different sensor without so much restriction?

The upshot is, gravity always works.  Well, ALMOST always... :-)



Ron Wanttaja
Dan Morehouse
#4 Posted: 2/14/2010 12:00:12

Thanks, Ron and David.

I'll be getting a less restrictive fuel flow sensor tomorrow and will do some tests with it. 

I can see the logic in 'simple is better', gravity never fails.

But I may find out that I'll end up needing a pump system, even if flow to the intake manifold is 150% of take off GPH. I'll be using a Rotec TBI, which is very similar to an Ellison. It's been in production for less than a year. People generally are very happy with it, but there isn't much data on it yet. 

Both Ellison and Rotec state a minimum of 0.5 PSI. To get that much pressure in a static system, I would need 20" of head. Well, I only have 6". I'm already way below the stated minimum, and that doesn't begin to take into account the substantial drop in pressure from friction losses once the fuel begins to flow. I should hear back from Rotec shortly about this, but I really doubt they would say 6" of head would be adequate. 

So my plan would be to run gravity to the gascolator, a couple inches above the right lower corner of the firewall. NPT nipple from the gasco to a Facet boost pump in series, then another nipple to the fuel flow sensor, then a 90 fitting into a 16" section of hose up to the mechanical pump. I've read where people with a system like this get good readings through their fuel flow sensor when the boost pump is off. Turning it on effects the readings a little, but since it's on only for takeoffs and landings, they find that acceptable.

There are a few that plumb parallel around the boost pump. This would be more complicated for me to do and I don't know I would gain anything from it. The Fact pump I am considering is the 40104, which has no check valve or 'positive shut off' (anti-siphon?) valve. 

So maybe now the question is, do I plumb parallel or in series? FYI, I will not have a recirculating pump. 

And since I will not have a gravity system, I would think avoiding 45 and 90 fittings like the plague is no longer necessary. I'd end up putting replacing 2 gently curved hose sections with 2 straighter hose sections with 45 and 90 fittings on their ends. This would cut in half the amount of hose in the hot section.

Take care,


Maurice Caudill
Young Eagles Pilot or VolunteerHomebuilder or Craftsman
#5 Posted: 2/17/2010 17:27:52

Hey Dan,

I have two friends with 3300's and I thought the fuel burn was more like 5-7 g/p/h.

I have another friend with the 100 horse Subaru conversion in a Kitfox and he has gravity (high wing)  with a Facet pump in line, and he only uses it on take off.  That's a 5 g/p/h engine.  His concern is the tail drager attitude on the ground and during climb out.  His head is probably less than a foot because the twin carbs. are on top of the engine.

Dan Morehouse
#6 Posted: 2/17/2010 19:10:14

Good to know, Maurice :-)

5-7 gph is for cruise. Jab lists 9.25 or something for 'takeoff' and I've seen most report 10-12 max.


Maurice Caudill
Young Eagles Pilot or VolunteerHomebuilder or Craftsman
#7 Posted: 2/18/2010 23:46:29

I talked to two 3300 owners today and one gets 5 gal/hr in his Zenith 601 and the other has changed the jets in the Bing carb. to get 4 gal/hr in a Europa.

John Loram
Young Eagles Pilot or VolunteerHomebuilder or Craftsman
#8 Posted: 7/13/2010 13:58:16 Modified: 7/13/2010 14:09:12


I'm building a Q2 with a Jabiru 3300 and my situation is very similar to yours. The Q2 has a main tank from which you pump to a header tank. The header tank provides about a 6" head to the carby inlet when the a/c is in a 20 degree climb.

My interest in a gravity feed installation is modivated by simplicity.  I know that my BMW motorcycle uses the same Bing Carbys as the Jabiru and the head is the same.

When I ran my fuel flow tests got the following results delivering the 'final gallon' from the header:

1)  Finger Strainer to 1/4" fuel cutoff valve, 5/16" fuel hose to carby. 90 seconds = 40 gallons/hr

2) Finger Strainer to 1/4" fuel cutoff valve, 5/16" hose to fram G12 fuel filter, 5/16" hose to carby: 143 second = 25 gallons/hr

3) Finger Strainer to 1/4" fuel cutoff valve, 5/16" hose to fram G12 fuel filter, 5/16" hose to Floscan 231 sensor, 1/4 fuel hose to carby inlet: 235 seconds = 15.3 gallons/hr.

So far so good.

Now I hook up the fuel hose to the carby inlet and remove the float bowl. I run a final measurement of the fuel collected to see how much fuel I get through the system including the float bowl inlet valve: 470 seconds = 7.6 gallons/hr (not enough!)

I removed the flowscan sensor and the fuel filter and ran the test again. It still took 8 minutes to deliver the final gallon throught the float bowl valve seat. That is where the real limitation is. 

I contacted Bing in Germany, and learned that they have five different float bowl valve seats for the the model 94 carby. A 1.5mm seat for fuel pump based systems, and 2.5, 3.2, 3.7 and 4.0mm float bowl valve seats for gravity fed systems. They suggested that I rerun my tests with the 3.2 and 3.7mm valve seats.

I'm waiting for them to arrive.

A side note: The early Jabiru 3300 engines were shipped with model 94 carbys with the 2.5 mm float bowl valve seats and users had problem with carby flooding because the valve was unable to block flow from the engine driven pump. For the last several years the 1.5mm valve seat has been shipped. Float bowl valve seats are available from the Bing Agency here in the USA.





Dan Morehouse
#9 Posted: 7/13/2010 15:05:15

Hey, John :-)

I'm finishing up my Sonex build... that is, if you consider another year or two to be 'finishing up'. In the slower, frustrating times over the last few months, I started looking at what I might want to build next. I decided having had no experience in metal airplane building hasn't been a major hurdle, so why not build a composite next, since I don't have any experience in that either. 

The Quickie series of aircraft look good. If I don't go the whole nine yards and build a Cozy IV, I'll consider a Q2. So keep me posted on how this turns out. 

I ended up going with the mechanical pump of the Jab and a Facet boost pump. I couldn't get enough flow through the fuel flow sensor to make me happy. Others run the exact same system gravity feed and report no problems. But none of these airplanes had documented successful flow testing meeting the 50% above max takeoff flow. There's a discussion I started on the Jabiru Engines Yahoo group. Yahoo has screwwy search results, but you might try looking at message 28998. Rotec told me I'd need a pump for the Sonex, which makes sense from what I'd learned.

So I ended up going tank to finger strainer to gascolator to boost pump (in series) to flow sensor to mechanical pump then a Y to a pressure sender and TBI. The pump was the Facet 40104. I don't recall the exact numbers, but it met the 25% above takeoff flow (lower standard for pumps than gravity) without tons to spare when I did the test. 

If you go with a pump, definitely search the Jab archives. Some have had problems with the boost pump flooding the Bing. As I recall, these cases were all with the 40105. Don't think there had been any reports of people using the 40104, possibly because it may have just started production relatively recently. This has slightly lower numbers than the 40105. People suggested getting the smaller one to help prevent the flooding.