N945HS

On February 28, 2023, about 1249 eastern standard time, an experimental amateur-built Savannah S airplane, N945HS, was substantially damaged when it was involved in an accident near Trenton, South Carolina. The private pilot in the right seat was seriously injured and an airline transport pilot in the left seat received minor injuries. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 test flight.

The pilot in the right seat, who also owned and built the airplane, reported that they were in the traffic pattern for a local flight at Edgefield County Airport (6J6), Trenton, South Carolina. The left-seat pilot was at the controls and was turning from left base to final for runway 29 when the Y-stick “locked up.” The right-seat pilot joined him on the controls in an attempt to free the jam, but the Y-stick remained locked in both the pitch and roll axes. The airplane continued in a left, descending turn until it collided with trees about one mile short of the runway.

The wreckage came to rest inverted. The fuselage, wings, and empennage sustained substantial damage. The engine and propeller separated from the airframe during impact.

The right-seat pilot reported that the accident flight was the fifth test flight conducted in the airplane and the airplane’s total time was about 2.5 hours. The engine ran smoothly during the event. He also stated that the airplane was “very balanced” and the flight up to the flight control malfunction was uneventful. He believed that an 8.5 inch by 11-inch note pad that he used to record flight test data fell to the cockpit floor during the flight and lodged under the bottom of the Y-stick.

The left-seat pilot reported that he was mentoring the right seat pilot through the flight test phase of the build. He had logged about 7,500 hours of flight experience in a variety of airplanes. He also stated that he was an airframe and powerplant mechanic with 38 years of experience in both civilian and military aircraft. The accident flight was his first flight in the Savannah S. He carefully examined the airplane for about 45 minutes before the flight and felt that the pilot did a “really nice job” with its construction.

The left-seat pilot further stated that he was at the controls and was in the traffic pattern at 6J6 when, during the turn from left base to final, the Y-stick “locked up.” The airplane was about 400 ft above the ground in a 500-600 fpm descent at65 mph with a 3° glide path. The Y-stick was jammed in both pitch and roll axes. He yelled through the intercom, “The stick is jammed!” and the right seat pilot also grabbed the stick. They attempted to break the jam with force but were unsuccessful; the airplane collided with trees and then impacted the ground. He believed that the note pad that the right seat pilot was using during the flight jammed the stick. He continued by stating that the Y-stick-to-floor clearance, in his opinion, was a “gigantic design flaw” that “everybody missed at [the airframe kit manufacturer]” He questioned why the issue with the Y-stick had not come up before and why the kit manufacturer had not engineered a solution to it.

After recovery to the pilot’s workshop, the wreckage was examined by the NTSB investigator-in-charge (IIC) and an FAA inspector. The left wing had been removed by recovery personnel and the right wing separated from impact forces; however, the empennage remained attached. Flight control continuity was confirmed from the elevator and rudder to the cockpit Y-stick and the rudder pedals. No areas of binding or interference were found. The airplane was equipped with flaperons. A single Y-stick control, located between the cockpit seats, operated both the flaperons and elevator. The flaperon control system, located aft of the cockpit seats, was intact and all connections were secure with the exception of overload fractures to the outermost push-pull tube for each wing. The flaperon controls moved freely when the Y-stick was manipulated and there were no areas of binding or structure interference observed.

Further examination of the Y-stick control revealed that the smallest clearance between the lower portion of the stick and the cockpit floor occurred with the Y-stick approximately neutral (stick straight up). When the Y-stick was moved in the roll axis (left or right), the stick-to-floor clearance increased. Full left or full right Y-stick inputs resulted in the maximum stick-to-floor clearance. There was no device installed on the cockpit floor designed to prevent a foreign object from rolling or sliding under the stick (the kit manufacturer offered, as an option, a soft boot to cover the lower portion of the Y-stick).

The note pad that the pilots believed interfered with the Y-stick control was documented after the wreckage was recovered; however, it was not examined in detail during the initial wreckage examination. The right-seat pilot subsequently provided photos of the note pad after a request from the NTSB IIC. The 8.5 inch by 11-inch note pad consisted of several pages and a cardboard backing. There was no manufacturer’s information inscribed on the pad identifying its origin. The photos indicated an area of puncture damage that was consistent in size and shape with the bottom of the Y-stick and also appeared to penetrate the entire pad thickness.

The airplane kit designer and manufacturer, I.C.P. srl, provided data at the IIC’s request on Y-stick-to-floor clearances in various positions of the control. The data revealed that the smallest clearance occurred with roll input central (neutral) and pitch full nose down (forward). This scenario resulted in a clearance of 4.83mm between the bottom of the stick and the seat spar corner and 8.31mm between the bottom of the stick and the floor. The largest clearance occurred with full right roll input and full nose up pitch input. In this configuration, there was 31.4mm clearance between the bottom of the stick and the seat spar corner.

Due to the deformation damage to the cockpit floor, the pre-accident Y-stick clearance measurements of the accident airplane could not be determined.