N5474DBeech H35 · 1958

On December 27, 2020, at 1812 Pacific standard time, a Beech 35 airplane, N5474D, sustained substantial damage when it was involved in an accident near Santa Ynez, California. The pilot and two passengers sustained minor injuries. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 personal flight.

The pilot estimated that he departed with about 32 gallons of useable fuel. While en route to Lompoc, California, the pilot altered his course slightly to avoid inclement weather. The airplane passed 1 mile north of the Santa Ynez airport and when about 17 miles from Lompoc the pilot switched the fuel selector from the right main tank (which was indicating close to empty) to the left main tank (indicating slightly less than half full). As he manipulated the selector handle, the engine rpm decreased. Concerned that he was not able to get the handle into the detent, the pilot placed the selector back on the right tank.

The pilot elected to perform a 180-turn and land at Santa Ynez because it was closer in proximity. He attempted to troubleshoot the reduced rpm and despite his efforts, was unable to retore the engine power. The airplane could not maintain altitude and the pilot performed a forced landing short of runway 08. The pilot attempted to avoid a tree and collided into a fence.

A postaccident examination revealed that airplane’s fuel system was modified in accordance with the supplemental type certificate (STC) SA2660SW, when the engine was converted to an IO-407-N which is fuel injected. The fuel supply was held in 4 bladder-type cells; the main left and right tanks, located in the inboard wing (adjacent to the fuselage), had a total capacity of 20 gallons each, equating to 17 gallons of useable fuel. The auxiliary tanks, located in the mid-wing section (outboard of the wheel-wells), were 10-gallon fuel cells in the wings and connected to each respective main. The fuel selector had a left, right, auxiliary tanks, and off position and was located to the left of the left-forward seat. Both auxiliary cells were connected to a common port in the fuel selector valve, so that both fed simultaneously when the selector valve is set to auxiliary.

Individual electric sending-units were installed in all tanks, transmitting fuel quantity information to the sole cockpit gauge. The fuel level of either cell was designed to be reflected on the fuel gauge in the cockpit by the pilot switching the auxiliary fuel gage selector switch on the left subpanel to “AUX” or “MAIN” position and the “FUEL GAUGE” switch to the desired right or left position.

From the fuel selector, the fuel was plumed to the electric-driven fuel pump and then through the firewall to the engine-driven fuel pump (see Figure 1 below). From the fuel pump, the fuel was routed to the servo. The fuel then continued to the manifold (mounted on top of the engine), where it was divided to each cylinder through GAMI injectors. The fuel return line was located at the engine-driven fuel pump and was routed only to the left tank; at normal operating power, about 8-10 gph would be returned to the left tank (as opposed to the carbureted equipped airplane, which would return about 3 gph).

Figure 1: Fuel System Design

Investigators established continuity in the fuel system. A water supply was plumed to each fuel supply line at the wing roots and the line at the fuel pump was disconnect with a bucket placed under. When a position was set on the fuel selector with the electric boost pump activated, investigators observed fluid flow from the selected tank inlet through the system and exit out the bucket. The selector positions all corresponded to the desired tank(s) and the detents were felt to be strong. Disassembly of the selector revealed no anomalies. No debris was seen that was flushed from the system. The filter was disassembled and was clean with no debris noted. The fuel pump was disassembled and was found intact; the carbon vanes moved freely, and light scoring was observed on cavity walls. The fuel servo had both the mixture and throttle cables attached; no anomalies were found. The fuel manifold was disassembled, revealing a pliable diaphragm that was intact.

The fuel caps remained secure on the tanks. The right and left main-tank fuel bladders had wrinkles in the bottom and were collapsed inward (not taut against the wing structure) consistent with the attachments not being secured. The aft section of the left main bladder was bulged upward in the area the fuel-sender float was positioned.

Investigators filled the left-wing bladder with water and simulated the dihedral the wing would have when installed. With the quantity sensor wired back to the fuel gauge in the cockpit, eight gallons of fuel was added to the bladder. The fuel gauge was observed to show about half-full with a measured eight gallons inside. The sending unit float cleared the bulged portion of the fuel bladder by approximately 3 inches with 8 gallons of water in the bladder.

The external examination of the Continental Motors IO-470-N-3E1 revealed no evidence of a catastrophic failure. Investigators removed all cylinders' rocker box covers and noted a light oil film on the rocker arms and valve assemblies. The upper spark plugs were removed, and all were similar in appearance one another displaying a light gray coloration, consistent with normal operation; the No. 3 cylinder was slightly different with a whiter and rust color. The cylinders' combustion chambers were examined through the upper spark plug holes utilizing a lighted borescope. The combustion chambers remained mechanically undamaged and there was no evidence of foreign object ingestion. The valves and their respective seats appeared undamaged; the exhaust valves all had a red coloration. Investigators achieved manual rotation of the crankshaft by rotation of the propeller and observed spark at each lead. Thumb compression was established in all cylinders; the No. 3 cylinder initially did not have compression until the valve was staked several times. The upper rocker arm toe was not aligned with the valve end on several cylinders.

The JPI was sent to the Safety Board recorders laboratory and did not contain any non-volatile memory that was able to be recovered.