RANS-Hine S-10 N104AH Analysis of Accident Occurring on October 15, 1994 Alison Hine
The aircraft was built by me, Alison Hine, over a period of 3 1/2 years between 1988 and 1991. It is equipped with a Rotax 65 hp water-cooled two-cylinder, two-stroke engine. Engine instrumentation includes an EGT guage on each cylinder, a water temperature guage, a water pressure guage, a fuel pressure gauge, and a tachometer. Fuel level is observed by looking through the wing roots at the sides of the fuel tanks, which are made of fiberglas. A series of markings on the tanks indicate the contents in gallons, both in level flight and on the ground.
The aircraft had about 260 hours on the airframe, and 170 hours on the engine at the time of the crash. Much of the flight time was hard aerobatics, included inverted flight and negative-G maneuvers. It had been flown on long cross-country flights of as much as 700 miles round trip, and had flown at gross weight, in steep climbs, with a full range of fuel loadings from full fuel to minimum fuel.
With one exception, no fuel system anamolies were ever observed. The one anomoly that was observed occurred after a maintenance operation on the right aileron linkage, which required the removal of a fairing on the right wing. This fairing also contained a breather for the right wing tank, and upon reassembly, the breather inlet was positioned slightly differently than before. During the first flight after this maintenance, fuel was drawn almost entirely from the left tank, even though the fuel valves for both wing tanks were open. After landing and adjusting the breather to its original position, the problem was eliminated, and fuel was drawn evenly from both tanks.
One engine problem had been encountered with the original engine, which was a high time rebuild. It appeared to suffer a crankshaft failure, probably because the life of the crankshaft had been exceeded; the engine manufacturer had not established a crankshaft lifespan at the time of the rebuild. This engine was replaced at approximately 92 airframe hours, and the current engine, a zero-time rebuild, was installed at that time. Prior to the crash, no problems had been encountered with the new engine, except for an occasion of lead fouling of the spark plugs after using 100LL aviation gasoline. Thereafter, high octane auto gas was used, and the lead fouling did not recur.
An annual condition inspection had been completed the week before the crash, followed by a shakedown flight. During the condition inspection, maintenence was performed on certain fuel system components. The squeeze bulb in the header tank breather line was replaced (see description of the fuel system in the appendix). The gascolator bowl was removed, the screen removed and cleaned, and the screen and bowl were replaced. The aileron linkage fairings were removed for inspection of the aileron linkages, and then replaced; the wing tank breather inlets may not have been returned to exactly their original positions in these fairings.
The shakedown flight was conducted by me, the builder/pilot, with no passengers, and approximately 3 1/2 gallons of fuel in each wing tank. During the shakedown flight, which included aerobatics and inverted flight, all systems functioned normally, with normal engine temperatures and fuel pressure. During and after the flight, the header tank remained completely full of fuel.
Description of Crash
I piloted the aircraft, and was accompanied by a passenger who weighed 220 pounds. To keep the weight of the aircraft within its design gross weight, I fueled the aircraft with approximately 2 1/2 gallons of fuel in each wing tank. Otherwise the configuration was not changed from the shakedown flight.
During the takeoff roll, and shortly after liftoff, I observed the engine temperature and pressure guages, and all were normal. Wanting to achieve maximum altitude as quickly as possible due to the aircraft's near-gross-weight condition, I did not lower the nose to look for traffic, as I usually do. At approximately 700 feet, the engine stopped abruptly, without prior warning. The propeller stopped immediately and did not rock back, indicating that the engine had siezed. With no suitable landing spaces in the wooded area ahead, I immediately executed a 180 degree turn and attempted to return to the runway. Previous testing had shown me that I could complete such a turn in the altitude available. However, insufficient altitude remained to reach the runway, and the aircraft contacted trees just short of the runway, and then the ground, sustaining substantial damage to the right wing, tail, main gear, and fuselage.
Post-Crash Observations
The fuel line to the forward carburetor was found to have air in it. Approximately 1 1/2 inches of this clear fuel line had no fuel. This "air bubble" was adjacent to the carburetor, indicating that, at the time of stoppage, fuel delivery to this carburetor had probably ceased.
The foward piston was found to be melted, particularly in the area adjacent to the center of the exhaust port and also opposite this on the intake side. This caused the engine seizure. This type of melting indicates a lean mixture was fed to the forward cylinder for a short time (a few seconds) prior to to the engine failure. A lean mixture was almost certainly the cause of failure. (See discussion of seizure causes and signatures in the Appendix.)
The entire engine and fuel system was intact. The header tank was completely full of fuel. The wing tanks still contained 2 2/2 gallons each. The gascolator was full of fuel, and the fuel filter was about half full (this is normal). No air bubbles were observed anywhere in any fuel lines except adjacent to the foreward carburetor, as noted above.
Disassmbly and testing of the fuel system components, including the fuel pump, pulse line, fuel lines, and forward carburetor, revealed no blockages, leaks, or other abnormalities or malfunctions.
Possible Cause of Crash
It is evident that the engine failed because of inadequate fuel supply to the forward carburetor, and that the fuel supply declined gradually in such a way that the forward cylinder ran lean for long enough to produce a seizure.
I hypothesize that the fuel supply declined because the fuel pump was unable to supply adequate fuel pressure for the conditions. Since the fuel pump does not appear to be defective or worn out, this indicates that some other factor changed which caused its capacity to be exceeded. It seems likely that some change which occurred during the recent condition inspection was the cause, despite the success of the solo shakedown flight afterwards.
During the previous 260 hours of flight, I noticed that during operations when fuel in the wing tanks was low (i.e. below about /2 gallons per side), the level of fuel in the header tank, which normally ran about 1/2 inch below the top of the tank, would gradually decline to as much as two inches below the top of the tank. Observation of the relationship of the wing tanks to the header tank indicates that is is probably because the top of the fuel in the wing tanks was actually below the top of the header tank during level flight. Note that during climb conditions, particularly with relatively low fuel, this effect would become even more pronounced, since the level of fuel in the wing tanks would now be appreciably below the level in the header tank.
Analysis of the fuel system (see appendix) indicates that entry of air into the header tank (necessary to allow the fuel level in it to decline) during level flight could occur only through the header tank vent line, and only if the one-way valves in the squeeze bulb in this line were not preventing backwards flow of air from the right wing tank to the header tank. I believe that this was in fact the case, and that the original squeeze bulb was, during "resting" conditions (i.e., when it was not being squeezed) "loose" enough to allow air to flow into the header tank, allowing the level of fuel in the header tank to decline as the level of the fuel in the wing tanks declined.
During the condition inspection, the original squeeze bulb, which had been in the airframe for the entire 260 hours to that point, was replaced. After this, during both the shakedown flight and the subsequent flight in which the accident occurred, the header tank always remained completely full, not even having the 1/2 inch of air at the top which had previously been typical. This implies that the new squeeze bulb had better valves, and these valves sealed at all times, not allowing air into the header tank during level flight or during climb.
A side effect of allowing air into the header tank, which the original squeeze bulb had done, was that the engine-driven fuel pump was always, in effect drawing fuel only from the header tank. In low fuel and/or climb conditions, when the fuel level in the wing tanks was inadeqate for gravity feed to refill the header, the header tank level simply declined.
When the new, more effective squeeze bulb was installed, the header tank, now effectively ventless, became essentially a fat section of fuel line. The fuel pump now had to draw through the header tank directly from the wing tanks. In level flight, with half-full wing tanks, this made no real difference, since gravity feed from the wing tanks assisted flow through the header. However, in a steep climb, the fuel pump now had to overcome a greater head height (forward carb to top of fuel in wing tank, instead of forward carb to top of fuel in header tank). It also had to overcome significantly greater friction due to the additional length of fuel line between the header tank and the wing tanks.
Discussions with an experienced Rotax engine rebuilder, with extensive experience with these engines and fuel pumps, indicated that it is possible that these conditions did in fact exceed the capacity of the pulse-driven Mikuni pump, although the only way to verify this would be to reproduce the conditions in which the failure occurred.
This did not cause a problem during the shakedown flight for several reasons: first, with only one person aboard, the angle of attack was slightly less, and secondly, when I lowered the nose to look for traffic, this relieved negative pressure in the header tank and fuel lines since gravity could assist feed to the fuel pump while I had the nose lowered. In addition, the extra gallon of fuel in each wing tank further relieved the load on the fuel pump by decreasing the head height which the pump had to overcome.
A contributing factor to the overloading of the fuel pump may have been the position of the wing tank breather inlets in the aileron linkage fairings. Examination of these after the crash revealed that both, although facing forward, were not facing directly forward but were at about a 20 degree angle. This is similar to the positioning which occurred during the previous experience, when fuel failed to flow from the tank which had a slightly mispositioned breather inlet. It is possible that a reduction of positive air pressure on these breathers and therefore in the wing tanks occurred, which would have increased the load on the fuel pump. However, both wings were damaged in the area adjacent to these breather inlets, so it is impossible to be certain that the inlets were positioned at this angle before the crash.
Recommendations for Preventing a Recurrence
I intend to make three alterations to address the problem of potentially overloading the engine-driven pump:
1. Add an electrically-driven auxilliary fuel pump, in parallel with the engine-driven pump, to be used during critical situations such as takeoff and climb.
2. Add a fuel valve in parallel with the squeeze bulb in the header tank breather line. This valve would remain open at all times except when the squeeze bulb was being operated to clear the breather line. This would ensure positive venting of the header tank during steep climbs with low fuel, when it is necessary for air to enter the header tank to avoid forcing the pumps to draw all the way from the wing tanks.
3. Replace the current wing tank breather inlets, which are simply flexible plastic fuel line projecting from a hole in the aileron linkage fairing, with vents made of rigid tubing, bent and positively anchored so that they can be assured to always face directly into the airstream. The existing flexible line would attach to these breather inlets inside the fairing.
After rebuilding the airplane, and making these modifications, I intend to conduct ground tests, followed by extensive flight tests at extremes of fuel and passenger load, and angle of attack (from an airport with a very long runway, such as Pease) to ensure that the fuel system functions correctly in all conditions.
I feel it may be prudent to notify other builders and pilots of this aircraft type, and its single-seat derivative, the RANS S-9, of the possibility of similar failures in their aircraft. Appendix
Causes of Piston Seizure in Rotax Water-cooled Two-cycle Engines
Three common causes of piston seizures are encountered in these engines: inadequate oil in the fuel mixture, engine overheating (ie coolant overtemperature or loss), and lean fuel/air mixture. Each of these failures has a characteristic "signature", or pattern of melting of the piston. Inadequate oil in the fuel mixture results in melting of the piston around its entire circumference. Overheating results in melting of the piston at its "corners" (in other words, what would be the corners if the piston were square). Lean fuel/air mixture results in melting of the piston in the center of the exhaust port, and sometimes opposite this, adjacent to the intake transfer ports.
Effects of Lean Fuel/Air Mixture in Two-cycle Engines
In four-cycle engines, each cylinder fires on alternating upstrokes, which allows a relatively long time for heat from the combustion to be dissipated through the head, cylinder walls, and piston. A four-cycle gasoline engine with adequate cooling can be run very near or at the optimum fuel/air mixture ratio of 15:1 (known as "stochiometric").
In two-cycle engines, each cylinder fires on every upstroke of the piston, reducing the time available for heat dissipation between firings by half. In these engines, in order to keep the temperature of the piston below the melting point of aluminum, the engine must be run with a fuel/air mixture considerably richer than stochiometric. The evaporation of the excess fuel cools the combustion chamber and keeps the temperature of the piston below its melting point. If, for some reason, the mixture is allowed to approach stochiometric, the engine will continue to run (indeed, it may produce more power) but the temperatures in the combustion chamber will rise quickly until the piston melts and the engine seizes. This can occur in a very short time, sometimes only a few seconds.
S-10 Fuel System Design
The RANS S-10, when equipped for negative-G aerobatics (as is N104AH), has two 6 1/2 gallon tanks, one in each wing, and a 1 gallon header tank in the cockpit. The S-10 is a midwing design, which places the wing tanks slightly above the header tank when the aircraft is in a level attitude. The bottom of the header tank is about inches below the bottom of the wing tanks in level flight.
The header tank uses a flop tube to maintain fuel feed while inverted or performing negative-G maneuvers. In level, positive-G flight, the wing tanks, which are mounted in the wing roots, gravity feed into the bottom of the header tank through lines which run from the aft portion of the floor of the wing tanks forward and down to the bottom of the header tank.
The header tank has a breather line, which exits the top of the header tank, wraps around the header tank, and runs to the top of the right-side wing tank. This line is intended to allow air to vent from the header tank after inverted flight, allowing the header tank to refill by gravity feed after being depleted during inverted flight. This line contains a squeeze bulb, oriented to push fuel away from the header tank into the wing tank. This is used after inverted flight, to clear the line of fuel if necessary, to allow fuel to flow into the header tank. The squeeze bulb, by the nature of its design, contains a pair of one-way valves that would be expected to prevent the flow of fuel or air from the wing tank through the breather line into the header tank.
The wing tanks are vented by a breather line exiting the top of the tanks just aft of the fuel filler necks. This line, made of flexible clear plastic fuel line, is routed outboard to a hole in a blister-shaped fairing which covers part of the wing end of the aileron linkage. The hole is in the front of the fairing, and the breather line projects through this hole forward into the airstream, thus acting as an inlet. This breather inlet is held in position in the hole by friction.
The fuel pump is the standard high-capacity round Mikuni dual-output pump, as supplied by Rotax for this engine. This pump is driven by a pulse line from the aft crankcase. In N104AH, it draws fuel through a gascolator and a clear plastic fuel filter, both mounted on the firewall. The dual outputs feed the carburetors directly, although a sensor for an electric Westach fuel pressure guage is inserted in the line to the forward carburetor.