Reading the ATSB accident report on the Norfolk Island ditching I blogged about on Monday gave me a case of déjà vu about another eerily similar accident in 1970. Being a student of aircraft ditching and survival, I've known about this accident for years as the "National Airways" ditching near St. Croix. But I never researched the details until I chanced upon an excellent book on the subject: 35 Miles from Shore: The Ditching and Rescue of ALM Flight 980 by Emilio Corsetti III.
Like the Pel-Air accident, the ALM ditching was the result of a long string of errors, but the fundamental one was flying an airplane for a route for which it wasn't suited. ALMAntilliaanse Luchtvaart Maatschappij, the Dutch Antilles National carrierhad wet leased a DC-9 from Overseas National Airways to fly the route between New York and St. Martin, in the Leeward Islands, a direct distance of 1650 milescoincidentally the same distance the Pel-Air Westwind had to cover between Samoa and Norfolk Island.
At the time, Douglas had broken into the mid-haul market with the DC-9 and while it was suitable for flying city pairs up to 1000 miles apart with profitable load factors, it was never envisioned as a long-haul, oceanic airplane. Recognizing that the DC-9 could fly the route with the thinnest of margins if everything went right, the FAA approved the ALM/ONA routing if Bermuda was used as a waypoint, meeting the-then required one-hour drift-down limits in the event of an engine failure. Bermuda required a jink in the route that increased the flying distance, but it also offered a contingency fuel stop. But as is always the case, adding a fuel stop would mean the flight would lose money, so the crews tried to avoid it. ONA had also arranged with Douglas to add a fuel tank to the DC-9, but it never got done because the airline put it off until the end of high tourist season.
Another problem never addressed was significant inaccuracy in the DC-9's fuel gauge and totalizer systems for its two main and one aux tank. Douglas said the system shouldn't show errors greater than 800 pounds, but on the ONA DC-9, errors were in the plus or minus 2000-pound range. This was thought to be due to condensation on the fuel probes. In any case, the ONA airplanes continued to burn more fuel than the Douglas charts said they should. Just two weeks before the accident flight, an ONA airplane landed on St. Martin with minimum fuel, flown by the same captain who would later perform the ditching.
As Dom James did following the Pel-Air ditching, ONA's pilot, Balsey DeWitt underwent intense investigatory scrutiny about his fuel planning, especially given the fact that he had landed with barely 30 minutes of fuel two weeks before. Why, investigators asked, had he chosen FL290 when passing Bermuda rather than a higher, more fuel-efficient altitude? He later descended twice more looking for a turbulence-free altitude, increasing the fuel burn and encountering unfavorable winds.
What finally did ALM980 in was weather substantially worse than forecast. Apart from daily thunderstorms and occasional tropical storms and hurricanes, the Leewards enjoy mostly fair weather, so much so that the only instrument approaches in those days were non-precision VOR or NDBs. There simply wasn't much need for anything better. The forecast called for 2500 scattered to broken, 10,000 overcast and 6 miles in haze. Nothing in that forecast would raise concern for most instrument pilots with experience in the islands. What the forecast didn't say, says Corsetti's book, is that the haze was caused by a Saharan dust cloud, providing rich condensation nuclei for all that Caribbean moisture.
And that's exactly what happened. When ALM980 came in range of the St. Martin's tower, the controller reported 1000 broken, 5000 overcast and 2 to 3 miles in showers. It was marginal for an NDB approach, but legal. Moments later, San Juan Center advised ALM980 that St. Martin was below minimums and Captain DeWitt turned for a diversion to San Juan.
Then the tower called again, reporting a slight improvement in the weather to 1000 broken and 4 to 5 miles in rain, well above the 600-foot MDA and 2 miles needed for the NDB approach.
ALM980 turned back toward St. Martin, but DeWitt believed they would land with 4400 pounds of fuel, 100 pounds more than the required 4300. When the airplane arrived over the beacon for the procedure turn, the weather turned out to be far worse than reported and when it broke out, it was too close and too poorly aligned to try to land. The airport's short runway5200 feetleft no margin for error. After two circling attempts, the crew gave up and struck off for the alternate, St. Thomas, with 3800 pounds of fuel leftabout 38 minutes. Minutes later, St. Martin tower closed the airport due to poor visibility and low ceilings.
They never made St. Thomas, or St. Croix, which was deemed a little closer. With fuel dwindling, DeWitt realized he would have to ditch miles short of St. Croix. While all of the occupants of the Pel-Air survived, sadly, that wasn't the case for ALM980. After impact, the airplane remained on the surface for 10 minutes; 40 of the 63 people aboard survived. It's almost certain that more would have, but in one those seemingly innocuous twists on which survival can turn, the cabin PA wasn't working. The flight crew couldn't warn the cabin crew that touchdown was imminent. Although they used the ONA method of three chimes to warn the cabin, the ALM-trained cabin crew didn't know what three chimes meant; it was accustomed to only two. Many passengers were standing or were otherwise unsecured during the impact. The ditching occurred in late afternoon light in moderate seas and rain. In a stroke of good fortune, a Navy helicopter ship was anchored nearby and rescue efforts began quickly.
The post-flight investigation revealed some of the same problems James encountered off Norfolk Island. The procedure for handling liferafts was flawed; vests rode too high and blocked the survivors ears, funneling water into their faces. There were unconfirmed reports that the St. Martin tower operator was pressured to report better weather than actually existed. Witnesses varied on what weather existed at the time of the initial approaches, but some said it was as little as a quarter mile. The NTSB took some lumps, too. It was too slow in assembling a maintenance records group and during the ensuring delay, the airline destroyed the DC-9's records, hiding possible improper work done on the fuel probes, according to Corsetti.
On March 31, 1971, the NTSB issued its probable cause: fuel exhaustion due to repeated attempts to land at St. Martin until not enough fuel remained for flight to an alternate. Mis-reported weather was also a factor.
Some positives emerged from the accident. Captain Dewitt was at least cited for exceptional airmanship in ditching the aircraft in trying conditions. The NTSB also recommended eliminating the old automotive-style metal-to-fabric seatbelts and this was adopted. Today, you fly behind metal-to-metal seatbelt buckles as a result. Also, better communications were installed in parts of the Caribbean to improve flight handling.
In the end, both the Norfolk Island and ALM ditchings owe their origins to aircraft being pressed into missions they simply weren't able to fly with suitable safety margins. I suspect the management at Pel-Air had never heard of ALM980 and didn't realize they risked repeating it.