Sliding Off The Runway In Hard Rain, Part 2

Miami Air 293 resting in the river after skidding off the runway. Photo credit: NTSB

The first part of this article series describes what led to the aircraft sliding off the runway into the river.

The NTSB conducted a major accident investigation of Miami Air 293, launching a full cohort of headquarters investigators. When they arrived at the scene, they found the airplane centerpoint about 100 ft. from the shore and 75 ft. from the non-frangible approach light pier in water 3-5 ft. deep. Parts of the engine cowlings, the radome and both main landing gear were separated from the airplane.

White landing gear track marks were visible on the pavement starting 1,592 ft. from the displaced threshold and continuing all the way to the end of the pavement. They zigzagged and were mostly to the right of the runway centerline.

The airplane was defueled, and 1,400 gal. of jet fuel was removed. After the airplane was lifted by crane and moved upriver to an open field, it was examined by investigators. It was then dismantled, and components were moved to more secure storage.

The airplane, manufactured in 2001, had accumulated 38,928 hr. of flight time and was maintained according to an FAA-approved continuous aircraft maintenance program (CAMP). In addition to the number-one engine thrust reverser, the right air-conditioning pack, and no pneumatic duct pressure, there were MELs on the WiFi system, the satcom and the extended twin-engine operations performance standards (ETOPS) status.

The payload of 136 passengers and their baggage amounted to just more than 35,000 lb. and the takeoff fuel was 28,200 lb. The planned fuel burn of 17,000 lb. put the landing weight at 143,100 lb., only 3,200 lb. under maximum landing weight. At that weight the landing reference speed (Vref) was 148 kt. and the no-wind approach speed was 153 kt.

The terminal aerodrome forecast (TAF) for KNIP for the 24-hr. period beginning at 7 p.m. was for wind 180 deg. at 4 kt., greater than 7-sm visibility, scattered clouds at 1,500 ft. AGL, with a broken ceiling of CBs at 3,000, 10,000 and 25,000 ft. There were thunderstorms forecast to be in the vicinity, and temporary conditions to be expected were winds gusting up to 20 kt. with 3-mi. visibility in moderate rain showers and mist.

The actual weather observation at the time of the accident was marginal visual flight rules (VFR) with heavy rain and thunderstorms. Precipitation of 0.53 in. was recorded at the airport in the 20 min. before the airplane landed. In the same 20 min., 714 lightning flashes were recorded nearby, one of which was just 1,637 ft. north of the airplane’s track.

The captain said in an interview that the weather conditions did not require the crew to run performance calculations before landing. The airport winds given to him by the controller were 340 deg. at 4 kt., and that was within the limitations of the airplane, so he continued. He said approach control did not provide any braking reports or runway conditions, and he never lost sight of the runway or approach lights during the approach. There was no wind-shear alert, and the radar showed the weather was to the right of the airport. The increase in rainfall on short final surprised him, and he did not think to go around, even though the EGPWS “don’t sink” aural warning sounded.

The 55-year-old captain had recorded 7,500 flight hours, of which 2,204 were in the Boeing 737. He held Boeing 727, Saab SF-340 and Swearingen SA-227 type ratings on his airline transport certificate as well the 737. He had been employed at Miami Air for 11 years and had upgraded to captain on the 737 in 2015. He had been a ground instructor, flight instructor, simulator instructor and line check airman at the airline before being appointed to the APD position, which allowed him to conduct FAA certification evaluations. He was originally from Argentina.

The Miami Air chief pilot described the accident pilot as an “excellent pilot” who was very knowledgeable. The vice president of flight operations described him as “one of their best,” and very thorough. Both managers were high-time former U.S. Air Force pilots who had flown at Miami Air before assuming their positions. The FAA’s aircrew program manager (APM) for Miami Air, who was a 20,000-hr. former Air Force pilot and United Airlines captain, said he had observed the accident captain on his most recent proficiency check (PC), and he had performed very satisfactorily. The principal operations inspector (POI) said there was no reason to require the crew to undergo a recertification check ride, commonly called a 709 ride, because they had received a retraining event from the company and were proficient.

The FO, who was 47 and also from Argentina, had accumulated 7,528 flight hours before coming to Miami Air in late 2018. He had been a freelance flight instructor in the Miami area. He had not flown a regular flight with the captain before the accident trip, but he had received his 737 type rating ride from him with an FAA inspector present.

An extensive aircraft performance study was done. Highly detailed flight data recorder (FDR) information, runway surface examination and precise examination of the landing witness marks provided the basis for one of the most-detailed examinations of actual braking action on a contaminated runway that I have ever seen. The report says that the white tire marks, the lateral load forces as the airplane skidded sideways, and the very low coefficient of friction calculated for the tires during the landing indicate the airplane experienced viscous hydroplaning.

Shortly after touchdown, the computed wheel braking friction coefficient was about 0.05. It varied between 0.04 and 0.09 for the remainder of the landing roll. This is similar to landing on ice. The friction coefficient of an unbraked tire is about 0.02.

The performance specialist calculated the airplane’s hydroplaning speed to be 110 kt. The ground speed of the airplane at touchdown was 180 kt. The runway was ungrooved and rainfall rates at the time of landing were two to eight times higher than the “heavy rain” rate normally assumed. Portions of the runway were likely to be very close to or over the 3-mm water depth that defines a “flooded” runway.

The NTSB asked Boeing to use a simulator model to determine how much the sideways drift of the airplane could reduce the braking coefficient of friction. Their simulation showed that an 8-deg. drift on a wet runway would reduce the braking coefficient by 24.4%.

The performance study determined that if the runway had actually been only “wet,” as defined by the FAA’s runway condition assessment matrix (a code 5), the airplane would have stopped on the pavement with 17% of the distance remaining, even with its excessive approach speed, tailwind and delayed speed brake deployment. On the other hand, with the conditions that actually existed (approximately code 2), the airplane could not have stopped on the runway even if it had been on target speed, in the touchdown zone and with calm winds.



Despite an extensive investigation, the NTSB members did not meet in a “sunshine” board meeting to deliberate, and a full “blue cover” report was not issued. By my count, there were 65 factual documents totaling 1,583 pages in the accident docket, but it was all condensed into a 32-page summary final report with no recommendations.

The safety board’s probable cause was “An extreme loss of braking friction due to heavy rain and the water depth on the ungrooved runway, which resulted in viscous hydroplaning. Contributing to the accident was the operator’s inadequate guidance for evaluating runway braking conditions and conducting en route landing distance assessments. Contributing to the continuation of an unstabilized approach were (1) the captain’s plan continuation bias and increased workload due to the weather and performing check airman duties and (2) the first officer’s lack of experience.”

The weather forecast, while accurate enough, did not communicate the true hazard. If the crew had done a landing distance assessment, it would have confirmed their intention to land. The captain flew an unstabilized approach and the FO did not call out his deviations, but these mistakes did not cause the accident. There was plenty of evidence that there would be a contaminated runway but no way to know just how contaminated. It was only the flooding of the runway caused by the sudden downpour that ruined their day.

There’s no rule that says you should simply not conduct an approach when you know there’s a thunderstorm moving across the airfield, but maybe there should be. I have held and even diverted in this situation. However, this captain was pressing for reasons of his own, and those reasons were in vain.

I suspect that on a less fraught, less lengthy and less stressful day these two pilots would’ve sensed that things were going badly, and they would have broken off the approach before it was too late. Fatigue and stress have a way of dulling the senses and the judgment, and even good pilots are not immune.

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.