An analysis of the fuel system in the BD-4
Author: Lance Schlichter
Background
One of the constant discussions about the BD-4 revolves around a stoppage of fuel flow on takeoff. If you read Roger Mellima’s newsletters over the years, many incidents and “fixes” have been discussed. At the builders meetings each year at Oshkosh the subject comes up. In 1998 and 1999 I listened to the discussion at Oshkosh about the fuel/header tank problem with great interest. I know from personal experience that a 1946 BC-12D Taylorcraft with a clogged fuel shutoff valve has enough gasoline in the float chamber, gasolator and fuel lines to start up, taxi out, do a run up and take off to an altitude of 100 feet before the engine quits cold. Fortunately, I was able to land on a parallel dirt runway, with no damage. Not wanting to go through that again, I set out to study the BD-4 fuel system and resolve in my own mind what was happening.
Methodology
My original occupation, before my demotion to management, was a test engineer. Over the years I have shown a real knack for breaking things, and the company I worked for decided that if I couldn’t break something, it was probably safe to ship to the customer. With that in mind, I built a full-scale mockup of my BD-4 fuel system, using a small end wing panel, 2x2 lumber and plastic lines. I duplicated the entire right side of the cockpit including all the braces. The wing panel was one of the 2 inch wide ones used at the ends of the wing. I glued a piece of Plexiglas over the open side so I could observe the interior as I filled it with colored water. The outlets were in the same location as the prints, and the lines were the same size only made out of clear plastic.
Experimental Runs
Run Series #1
Run number one was made with the entire system empty. The tank was set in the tricycle gear configuration i.e. the door sill was level, and the tank was set at a 3 degree angle of attack. As I poured colored water in the filler, the two lines from the tank to the instrument panel tee promptly air locked. In other words, no fuel flowed from the tank to the tee. Just to be sure, I emptied the system and repeated the test. This time, the fuel lines filled up. Closer observation shows that (in the tricycle configuration) the rear fuel outlet is slightly higher than the front outlet. As fuel is poured into the tank, fuel flows down the front line while air is vented back into the tank through the rear line. This loop bleed stops the moment the rear outlet is covered. In other words, how well the loop fills with fuel is highly dependent on the rate that the tank is filled, and the attitude of the plane. I tried a number of different runs and fill rates. I was never able to reliably prevent the lines from air locking.
Run Series #2
This series of runs was made in the tail dragger configuration. The mockup was rotated until the doorsill was at 10 degrees, and the resulting wing angle of attack was at 13 degrees. In this configuration, both the tank outlets are at the same level, and the lines air locked about 95% of the time. Also, if you are building a tail dragger, you should know that there is a huge (about 20% of the volume) air bubble in the top of the tank in front of the spar. If you build a tail dragger, put the filler caps in front of the spar and you will pick up about 20% more fuel per bay.
Run Series #3
In this series of runs I played with the routing of the two fuel lines. I moved the two lines from the tank to the tee around. I raised them. I ran then straight down for a ways, then to the tee. Many people run one line down the front door post, and another one down the rear doorpost. I tried this configuration in both the tail dragger configuration and the tricycle configuration. Nothing changed. As soon as the two outlets in the tank were covered with liquid, the lines air locked.
Run Series #4.
The purpose of this series was to determine what effect unporting the front outlet would have. A lot of the discussion at Oshkosh and in the newsletters revolved around the effect of unporting the front outlet during takeoff. First I calculated the resultant angle that would result in a 1 g takeoff (Ray Wards Rocket). During a 1 g takeoff acceleration, the liquid in the tank will see 1 g back, and 1 g (gravity) down. The resultant fuel level will be at a 45-degree angle. So I filled the tank to about ¼ full (the fuel lines promptly air locked) then rotated the whole assembly to a 45-degree nose up attitude. In sequence, the front outlet unported, the rear fuel line bled and filled, and fuel started flowing in the system. Changing the flow rate had an interesting effect. As the flow rate slowed down, the level of fuel in the front (unported) line rose. As the flow rate increased, the fuel level in the front line went down until it was sucking air bubbles into the main line. This is due to the venturi effect. The flow from the back line created a small vacuum at the tee, and pulled air or liquid down the front line. Note that this whole effect is highly dependent on the location of the tee. If the tee is located per the print on the corner of the instrument panel, both lines unport at the tee itself at about 50 degrees nose up. In other words, if you have a big engine, partially full tanks, pull 1 g acceleration during takeoff, and yank the nose up 10 degrees you have probably unported both your lines at the tee and sucked a bunch of air into your system. If you have a hot rod, move the tee down to the floor level and take off with full tanks. Better yet, build a mockup and test this yourself.
Run Series #5
The purpose of this series was to determine how to bleed the airlock out of the front-outlet-tee-rear-outlet loop. The most obvious way was to open the gasolator and drain the lines. The results were somewhat unexpected. If the cap was on the tank, and the tank vent was blocked, nothing came out the lines. When the vent was unblocked, the lines would slowly fill with fuel. One line would fill fairly quickly, and the other line would trap air for quite a while. As a minimum, I had to drain at least a quart of liquid out of the gasolator before both lines were reliably filled. Next, I duplicated the vibration of an aircraft by shaking, rather unscientifically, the mockup. As I shook it, air bubbles worked themselves out of the lines, and after a while the lines filled.
Run Series #6
The purpose of this series was to investigate the effect of a header tank. The first run was with the two outlet lines split one down the front doorpost, and the rear outlet down the rear doorpost, both were brought together at the tee, then into a 1 gallon header tank mounted under the instrument panel. Starting with a totally dry system, the header tank only filled if the wing tank outlet loop didn’t airlock. If the tank outlet loop air locked, the header tank filled verrrrry slooooowly as air vented slowly out the header tank vent. If the test started with the header tank full, and the lines and wing tanks empty, as soon as the wing tank outlets covered the lines air locked. My working conclusion is that a header tank doesn’t fix the basic air-locking problem. It simply fly’s the airplane until vibrations have time to work the airlock out of the lines and fuel starts to flow normally in the system. The second conclusion is that filling the header tank is a separate task. Don’t think that simply putting fuel into an empty system and letting it dribble into the header tank will solve your problems.
The Hewes Mod.
A fellow named Hewes built a BD-4 he called the Virginian Patriot, and then published a book with a number of mods he had made. One of the mods was to put a “L” shaped tube in the wing tank outlets so he scavenged more fuel and reduced the unused fuel left in the tank. I mocked this change up, and found that it caused the lines to airlock faster. In other words, both ports were covered with fuel sooner, and air locked sooner.
Fixes
So how do we stop the air locking of the lines? Or to rephrase the question-Why doesn’t Cessna and Piper have this problem? With that thought in mind, I got the fuel schematics for a Cessna 182 and a Piper Tripacer. They both have front and rear outlet tanks, near duplicates of the BD-4, and they had two different and equally clever solutions. The 182 schematic showed a tee in the lines that feeds to an outlet in the top of the tank. This line lets the loop bleed out before the tank is full. Piper routes one of their fuel lines down the front door post and the other down the rear doorpost, then joins them at a tee under the center of the instrument panel. They put a gasolator under the center of the fuselage. Sampling this gasolator automatically bleeds the tank outlet loop. I mocked both of these fixes up on the mockup, and they worked equally well. Teeing either line to a third outlet high on the tank (the Cessna fix) vented the loop and filled it with fuel every time. Putting a gasolator under the doorsill (the Piper fix) worked equally well for those that route one line down the front doorpost and the other down the rear doorpost. Both worked equally well with the Hewes mod.
Conclusions
| 1. | Air in the lines from the wing tanks to the instrument panel tee will stop the fuel flow at the normal aircraft fuel flow rates. If the tank has been run empty, there will be an airlock in the gas lines. Period. When you fill the tank, gravity will force the fuel down one line, through the tee, and the air back out the other line where the air vents into the tank. The problem is, this takes time. Before the air is all out, fuel covers the second port, and air is trapped in the lines. When the tank is level, like the nose wheel configuration, the fuel goes down the front fuel port and line, and the air goes up the back line and out the back fuel port, until fuel covers the back port. With the tank 10 degrees nose up, like the tail dragger configuration, it's a race condition. Both fuel ports are at about the same level, so they get covered with fuel at the same time. In about 50 attempts, I was never able to reliably fill the tanks without an airlock in the lines. |
| 2. | In all cases, there was air in the lines from the tee to the valve that switches tanks. If I drained the system, then refilled it, the air from the tee to the gasolator had no way to vent. It could only be vented by disconnecting the line at the gasolator and letting fuel flow out. The dinky little sampling outlet on the gasolator simply wouldn’t flow enough liquid to clear the lines. |
| 3. | Clearing the air out of the lines by using the fuel pump suction is problematic. How good is your fuel pump at moving air? Is it self priming? |
| 4. | Header tanks are a real problem. The problem is, the header tank vent must dump into a neutral pressure area. If it dumps into a high-pressure area and you ever run the header tank dry, air is forced down the header vent line, then back up the fuel line into the tank. You will never clear the airlock out of the lines. My plane came with a header tank. The fuel inlet from the wing tanks is about 2 inches above the bottom of the tank, and the outlet to the gasolator is in the very bottom. It is possible that you could switch tanks after the wing inlet unports, and before the header tank went dry (it would have about 2 inches of fuel in the bottom), and blow the fuel back up the line into the wing tank if the vent pressure was high enough. If the header tank vent dumps into a low pressure area, it'll suck the airlock out, however it'll also suck all the fuel out of your wing tanks. This can be partially solved by venting the header back into the corresponding wing tank. But then when you switch tanks, you have to switch vents. Alternatively, you can design a system with a single interconnected venting scheme that ties both tanks and the header together, but I’m not interested it that mess. |
| 5. | There was quit a bit of air in the lines. By disconnecting the line at the gasolator, at full flow, it takes about 10 seconds to clear the system. In the plane, the fuel pump will have to pump all this air out through the fuel injectors or float chamber. I have no idea how long this takes. I suspect it will scare the hell out of you while it's happening. |
| 6. | Unporting the front outlet didn’t stop the fuel flow. However--- and this is a biggie, if things slosh enough to unport the front outlet, you are dangerously close to unporting the tee and that will lead to air locked line between the tee and the tank switch valve in the center of the instrument panel, and that’s a real problem. Putting the tee closer to the floor increases the margin of safety. |
Recommendations
| 1. | If the tank has been emptied, disconnect the fuel line at the gasolator and drain about a quart of fuel out. (your individual plane may take more, say a gallon or so). (This assumes the gasolator is in the engine compartment just before the carburetor, otherwise you have to work out the solution yourself.) Do it for both tanks. This seemed to clear out the airlock in both the front and back lines. Anything less often left air in one line or the other. The problem is, one line would clear, and the other wouldn't. Again, it takes time for the air to work through. I really don't think blowing in the tank works for both lines. It probably clears one line, but not the other. Reading some of the reports, I tend to think people forget to do it to both tanks. I’ll bet that some people have bled one tank, and then taken off on the other unbled tank |
| 2. | Unless you are know for certain where it vents, under all flight conditions and weather conditions, avoid header tanks. If it vents into a high pressure area, and you have an airlock in the fuel lines I described above, when you run a tank dry you will never restart the fuel flow to your engine. You'll land with fuel in the wing tank, and due to the airlock, you will never be able to refill the header tank and use it. |
| 3. | The best solution is to vent the lines like the Cessna system. Put a tee in either the front or rear line, then route a third line back to the top of the corresponding tank. It worked every time. |
| 4. | The second best solution is a modified Piper system. In this case you’ll need two more gasolators, one under each door. |
| 5. | Then there is the Roger Mellema solution. Roger recommended that you have a takeoff tank and a cruise tank. He never let his takeoff tank go dry, but often ran the cruise tank dry. My guess is that the normal vibrations of flight worked the air out of the lines to the cruise tank before it was needed again. |
| 6. | If you have a hot rod, be real careful about taking off with partially full tanks. The combination of acceleration, gravity and angle of attack can conspire to unport your fuel lines at the instrument panel tee and stop your fuel flow. |
Lance Schlichter
You can download this article as Microsoft Word document: bd4_fuel_schlichter.doc