WPR19FA022 · Piper PA28 236

About 45 minutes into the cross-country flight, the student pilot passenger reported to air traffic control that the pilot was having a heart attack and that he was going to take control of the airplane and divert to an airport about 20 miles away; however, another airport was much closer, and controllers received no further communications from the pilot. The student then reported over the nearby airport’s common traffic advisory frequency that he would attempt to divert there. GPS flight track data indicated that, a short time later, the airplane began to fly in an erratic manner, climbing and descending over the area of the diversion airport. The airplane subsequently collided with terrain in a wings-level attitude.

All four occupants sustained fatal injuries, and toxicology testing indicated that all had high levels of carboxyhemoglobin, which is caused by the inhalation of carbon monoxide (CO), an odorless, tasteless, colorless, nonirritating gas formed by hydrocarbon combustion. Levels in both pilots and one passenger where high enough to have caused confusion, seizures, or loss of consciousness. There was no evidence the pilot had a heart attack, so his symptoms were likely due to the CO poisoning.

The engine was equipped with three identical exhaust mufflers. Examination of the aft muffler revealed a crack that allowed engine exhaust gases to enter the cabin through the cabin heating system.

The muffler’s crack developed as a result of corrosion and thinning of the muffler wall. The crack and small perforations in the muffler wall were likely present at the time of the last inspection, which occurred shortly before the accident; however, due to damage from impact, it could not be determined if the extent of cracking was readily visible at the time of inspection. Also, the crack could have opened just before the accident flight due to an engine backfire that occurred during startup.

Use of a pressure check during an inspection could help identify cracks and small perforations that can be obscured by oxides and deposits during a visual inspection; however, the manufacturer's service manual only recommended a pressure check when a visual inspection could not be accomplished.

The manufacturer and Federal Aviation Administration (FAA) both recommended that the mufflers be replaced at or near 1,000 hours time in service (TIS). Although maintenance records revealed that three mufflers were replaced over the previous 27 years, the records did not indicate if the cracked muffler was one of those that had been replaced. Review of the available records indicated that the cracked muffler likely had at least 752 hours TIS, but due to missing records, it could have had over 1,033 hours TIS. Additionally, one of the mufflers was replaced at 549 hours TIS, well before the manufacturer's recommended interval, and likely an indication of premature failure.

The hazard of CO poisoning via leaks in reciprocating engine exhaust systems that are used to provide cabin heat has long been known to the industry. In 2004, the NTSB issued a safety recommendation to the FAA to require the installation of CO detectors in all single-engine airplanes such as the accident type.

In response, the FAA undertook extensive research on the detection and prevention of CO exposure in general aviation aircraft, created a technical standard order specifying minimum performance standards for CO detectors, and recommended, but did not require, that all operators install CO detectors.

The FAA concluded that the primary method to prevent CO contamination in the cabin is through proper inspection and maintenance of mufflers and exhaust system components and that CO detectors are a secondary method of preventing CO exposure. The FAA further stated that since a lack of a CO detector alone is not unsafe, installing a CO detector would not correct an unsafe condition.

Because the FAA did not require installation of CO detectors, the safety recommendation was classified "Closed – Unacceptable Action."

Verbatim NTSB analysis and probable cause from the NTSB dataset