Splashdown in the Caribbean: Investigating The Accident

Placencia Airport. Image credit: Google Maps

Note: The first part of this article lays out the lead up to this accident.

The Belize Department of Civil Aviation (BDCA) conducted the investigation of the Cessna 208 Caravan that collided with a vehicle on takeoff and ditched in Belize. After retrieving the aircraft from the water, two investigators took photos; documented the damage; and interviewed the crew, passengers and crash site witnesses.

The cargo pod was damaged, the engine exhaust was compressed and a landing gear fairing was damaged. The propeller blades were also damaged, and the entire structure of the airplane had water damage. The automobile’s roof was crushed on its right side and its windshield was shattered.

The 50-year-old pilot held a commercial pilot license (CPL) issued by the Belize BDCA and had logged a total of 19,040 hr. of flight time, 12,092 of which were in the Caravan. His most-recent instrument and proficiency check was two months prior to the accident. The right-seat observer had been penciled in on the manifest as a first officer, but the BDCA considered him unqualified and did not obtain his pilot information.

Investigators counted and obtained the weights for all nine regular passengers, the right-seat observer and the pilot. They speculated that the airplane might have been up to 250 lb. overweight but did not clearly ascertain this.

The airplane was delivered to Tropic Air in October 2005. Its total flight time was not stated, but a Hobbs meter time of 2,106.1 hrs. was given. The Pratt & Whitney Canada (P&WC) PT6A-114A turboprop engine was installed in January 2016 and had apparently logged 1,895.9 hr. since that time. (The report used the term “time since repair” rather than “time since installation.”) The engine’s total time of operation was 8,157.1 hr. The overhauled Hartzell propeller had 2,536.4 hr. since it was installed.

There were 25 engine logbook entries since the end of January 2017, and six were for hot starts, overtemps or delayed acceleration. Cracks were found in the inner starter vane ring in March. The engine was “preserved” for 30 days between June and July to conduct a P&WC service bulletin, but there was another overtemp in August. On Sept. 28, a month and a half before the accident, the engine failed to achieve full torque on takeoff.

I asked a pilot/mechanic who is highly experienced with PT6 engines about the airplane’s history, and he said it was very concerning. Start temperatures within 100C of maximum indicate a need to fix the problem and should not be accepted as normal. A healthy engine should be able to idle at 52% Ng, and setting the idle up from 58% to 64%, as was done on June 12, is an indication of degraded efficiency. The highest temperatures are often seen at starting, and short flights with frequent starts are hardest on the engine. He also said pilot bad habits and disregard of indications can be a huge factor in engine life.

The report did not specifically say when the engine was last overhauled. Normal time between overhauls (TBO) on the PT6A-114A is 3,600 hr., but according to PT6 Nation, Caravan engine TBOs can be extended to 8,000 hr. if the engine has an aircraft data acquisition system (ADAS) and is on a P&WC trend monitoring program. Operation in highly corrosive environments is said to reduce the benefit of TBO extensions. It appears that the engine was past its extended TBO, which would be consistent with the hot starts and other problems.

Investigators removed the engine and sent it to P&WC, where it was examined under the supervision of an investigator from the Transportation Safety Board of Canada. The teardown report found considerable corrosion but did not distinguish between pre-existing corrosion and corrosion due to the saltwater immersion of the engine after the accident. P&WC’s report said that other than saltwater contamination they found no form of damage to the engine. Their only comment on the power turbine section was that it was covered by salt deposits. The fuel control unit was not tested.

The airplane’s ADAS box was removed, placed in fresh water, and sent to the manufacturer for analysis. The ADAS data provided details of the airplane’s location, speed, altitude and certain engine parameters. The Caravan was equipped with a Garmin G1000, but data from electronic devices other than the ADAS was not obtained, and the airplane was not equipped with a flight data recorder (FDR) or cockpit voice recorder (CVR).

The BDCA report stated that the airplane struck the car when it was 28 ft. past the end of the runway and at a height of 5.5 ft. above the surface. The main landing gear would have been lower than the point where the altitude was measured by the ADAS. An excerpt of the ADAS flight data log was included in the report. A close examination of that data shows that the engine torque began to decrease when the airplane was only 4 ft. in the air and that interstage turbine temperature (ITT) and high-pressure rotor speed (Ng) began to decline within 3 sec. after liftoff. By the time the airplane was 11 ft. in the air, and 4 sec. after liftoff, torque had fallen from the takeoff value of 1,800 to 400 ft.-lb.

The ADAS data showed the maximum torque the accident engine obtained on takeoff was 1,800 ft.-lb. My PT6 expert said that engine should have developed 1,875 ft.-lb. at sea level with ambient temperatures less than 42C. The engine was making about 641 hp, not the 675 hp it would at full power.

The existence of a main public highway so close to the departure end of Runway 07 is an obvious hazard. Traffic signs warning of low-flying aircraft are posted, and a crossing barrier is in place to stop traffic during an aircraft takeoff or landing. Operation of the barriers is a responsibility of aerodrome management, but at Placencia that responsibility was delegated to the operators, Tropic Air and Maya Island Air. In practice, the operators did not send anyone out to operate the barriers and there have been multiple close calls between aircraft and automobiles.

The report provided only minimal information about Tropic Air’s management. According to the company’s website, Tropic Air has been in business since 1979 and now carries about 300,000 passengers a year and operates over 200 flights a day. It also has a “Tropic Air Pilot Program” that accepts low-time pilots into a flight-time-building scheme. Pilots with 200 hr. and a commercial certificate pay a fee to act as first officer.

 

Conclusions and Comments

The BDCA concluded there were three probable causes of the accident. They were:

  • • “The driver of the vehicle failed to adhere to traffic warning signs regarding low-flying aircraft and drove his vehicle directly into the departure path of an aircraft.
  • • “The pilot flew the aircraft at a dangerously low altitude over the road and did not properly assess the risk at hand, which was a vehicle advancing into the aircraft’s departure path that could cause a collision.
  • • “The pilot did not take collision avoidance (evasive) measures in a timely manner.”

The BDCA, which is a regulatory agency, issued several recommendations. These included better training of pilots for ditching, fines for the accident pilot, random checks of life preservers on aircraft, and the need for operators to update their emergency response plan. In addition, they recommended that the Ministry of Transport increase fines for “violations around Placencia Airport” and that the Belize Airport Authority do a study to displace the Placencia runway or consider a different airport site.

There is just no rationale for having an active runway so close to a major highway. There were multiple precursor close calls warning that this accident would happen, and yet neither Tropic Air nor Belize aviation authorities took any effective steps to prevent it. This was the classic “accident waiting to happen.”

The investigators placed the responsibility for the accident squarely on the pilot, with a minor role also going to the automobile driver. The pilot may very well have erred, but a simple “pilot error” conclusion doesn’t do much to prevent another similar accident. The pilot did not do himself any favors by refusing to take a drug and alcohol test or a medical exam. He also failed to assist passengers in the evacuation and implied that his observer was at fault for lacking competence.

It would have been appropriate for investigators to question the observer and the passengers to find out if they noticed any impairment of the pilot. It would have been useful to interview pilots, managers and others who knew the pilot to find out if he had a substance abuse problem. That could have identified a real problem that could be addressed and fixed. Of course, when the agency has the power to levy or recommend fines, people who could be helpful tend to disappear or become forgetful.

It doesn’t make sense to me that a highly experienced pilot would deliberately fly the airplane at such a low height across the highway if he had any choice. The report says the ADAS shows the airplane remained on the runway for 13.33 sec. after he had achieved a rotation speed of 65 kt. and that he had 872 ft. remaining in which to abort the takeoff. They say he should have been at 50 ft. crossing the highway. Yet the pilot stated that he rotated the nose to 2 deg. above normal takeoff pitch and the stall warning sounded. What would account for these conflicting statements?

Could the airplane’s performance have been less than book figures? The engine output was less than full rated power. Was the takeoff weight greater than calculated, or were there aerodynamic drag or lift issues? The information I could find on C208B performance stated that flaps should be set to 20 deg. for takeoff and that the rotate speed should be 70 to 75 kt., not 65. I did not find any documentation of the airplane’s flap position in the wreckage. An over-rotation and stall warning would be consistent with a no-flap takeoff. The pilot’s true error may have been in failing to set flaps for takeoff.

The investigation did not explain how the engine shutdown happened. There is no evidence the prop or engine components struck the car. I think the abrupt drop in fuel flow 2 sec. after takeoff was probably caused by a failure in the fuel supply, and that coincides with the collision. Nonetheless, the engine was sick. Given the company’s history of engine failures, I would suggest the BDCA and Tropic Air take a very hard look at their engine operation and maintenance practices.

I think Tropic Air would benefit from a real-time engine monitoring program. Lightweight data recorders and flight data management software are available and would be appropriate in this operation.

I would like to thank Capt. Paul Rule for his expert advice on the PT6 engine.

Roger Cox

A former military, corporate and airline pilot, Roger Cox was also a senior investigator at the NTSB. He writes about aviation safety issues.