Lessons Not Learned From The Past (Part 3)
Guest Editorial written by David H. Marion – In Part 1 I described the tragic events that occurred near St. Thomas, U. S. Virgin Islands on September 2, 1978 when Antilles Air Boats Flight 941 (Grumman G-21A Goose N7777V, c/n B-111) experienced an engine failure and crashed into the sea, killing the pilot and three passengers. In Part 2 the focus was on the subsequent FAA and NTSB investigations, their findings, and the lessons that should have been learned long before that crash happened. Now in Part 3, my attention jumps forward almost 30 years to the even more tragic events of December 19, 2005 when Chalk’s Ocean Airways Flight 101 (Grumman G-73T Turbo Mallard N2969, c/n J-27) which lost its entire right wing shortly after take-off from the Miami Seaplane Base (X44) on Watson Island, killing all 18 passengers and 2 crew on board.
My inspiration for starting this series of articles in the first place was an episode of Air Disasters on the Smithsonian Channel which focused on the Chalk’s crash (Season 2, Ep. 1 “Cracks in the System” which first aired June 7, 2012.) I was a little bit put off by some of the television production details. For example, in their reenactment of a mechanic installing “doubler” patches on the lower wing skin, he used non-structural “pop” rivets! And in the scene where he was “resealing” the right-side “wet wing” fuel tank, they showed him using a grease gun to pump lubricating grease all over the seams of the fuel tank! It was blatantly obvious to a “real” mechanic! So, I thought that I ought to write something maybe a bit more analytical for Seaplanemagazine.com.
In Parts 1 and 2, I started with the Antilles Air Boats crash in 1978 because right off the bat, the NTSB report indicated that there were a lot of similarities to the much later 2005 Chalk’s crash in terms of maintenance factors that directly contributed to the tragic chain of events. Most aircraft accidents are caused by a series of faults, failures, missteps, or mistakes the avoidance of any one of which may have prevented the crash. It seemed just as obvious to me that there had been lessons apparently not learned in spite of the preponderance of evidence supporting those lessons and time in which to learn them. That to me made the later, more deadly Chalk’s crash that much more tragic and unnecessary.
The Mallard’s wing is comprised of three major structural sections, a center section and two outer wing panels. The center section extends from Wing Station (WS) 125 Left to WS 125 Right – which means that it is 250 inches or 20 feet, 10 inches wide. The sides of the fuselage intersect with and attach to the wing at WS 34 Left and Right (note: at its widest point, the fuselage is 72 inches or exactly 6 feet wide) Inside each wing section, there are two main spars, one forward and one aft. The space in between the two spars outboard of WS 34 all the way out to WS 125 is sealed and used as the main fuel tank on each side and together they hold a total of 380 US Gallons of fuel. (The fuel tanks of some earlier Mallards extend out only to WS 107 and they hold only 330 USG while later ones use 180 gallon rubber bladders on each side instead of “wet wings.”)
Any seaplane and especially those that operate in salt water environments have issues with corrosion. Any airplane of any kind (not just seaplanes) that has “wet wing” fuel tanks eventually has issues with leaks. These two maintenance factors intersected in such a way on the G-73 Mallard series that in May 1963, when the type was barely 15 years old, Grumman issued Customer Service Bulletin no. 89. In it, Grumman acknowledged that corrosion is the single “greatest maintenance problem” for seaplanes. They said that the area of the lower wing skins in the vicinity of the fuel tanks on the Mallard were of particular concern. Also specifically that corrosion was to be suspected if there was ever any “chronic leakage of fuel, looseness of sealant inside the fuel tanks, or blisters, swelling, or visible corrosion on the skin near the fuel tanks.”
As it turns out, the accident airplane (N2969, model G-73T c/n J-27) had a long history of “chronic leakage of fuel.” It had been built by Grumman in May 1947. Chalk’s had bought it in 1980 and had Frakes install the turbine engine conversion (STC no. SA2323WE) in July 1981. At the time of its crash, the airplane’s records show that it had 31,226 flight hours (Time in Service) and 39,743 total cycles. Note: a “cycle” is the combined set of one take-off, one flight, and one landing and it is usually more important in regard to pressurized aircraft, which the Mallard is not.
According to the final NTSB report (AAR-07-04) on the Chalk’s accident, the maintenance records for the accident airplane (N2969) showed that even just in the few months prior to the crash, it was written up on pilot’s squawk sheets for recurring fuel leaks three times in August 2005 and three more times in September 2005. The maintenance work cards for the aircraft also showed a total of eight times that fuel leaks were supposedly repaired between August 2005 and December 2005.
Its maintenance logbooks also showed numerous related “major repairs” to the fuel tanks and lower wing skins dating back to January 1995. The NTSB also found that there were no company work records of any kind for the airplane at all during the period from January 2000 until January 2001 – even though there should have been something corresponding to a FAA Form 337 in the airplane’s logbooks which was dated July 6, 2000 and which documented a “major repair” to the same critical portion of the right wing. That particular repair had been made at WS 50.5 of the rearmost of three Z-stringers (so named because of its cross-section profile) which stiffen the lower skin inside the “wet wing” fuel tank in between the front and rear wing spars.
Additionally, there were other significant major repairs to the wing structure going back as far as April 13, 1992 when the upper cap strip of the right wing rear spar was repaired at the wing root at WS 34. Besides that, there was another major repair during which one external “doubler” patch and three more internal “doubler” patches were installed on the lower skin of the inboard right wing fuel tank. Troublingly, there was no supposedly “required” Form 337 for those doubler patches, although the workmanship seemed to match a similar repair to the LEFT wing from May 6, 1992. The maintenance supervisor claimed that it “must have been done prior to Chalk’s getting the airplane” in 1980 and that he had never directed any of his mechanics to make such a repair. Nevertheless, based on relative amounts of corrosion on and around the repair parts in question, the NTSB decided that it actually must have been done around the same time as the July 2000 repair to the adjacent internal Z-stringer at WS 50.5.
The four doubler patches seemed to have been installed in order to deal with a crack that had been “stop-drilled” three times but which nevertheless continued to propagate forward from the vicinity of the fuel drain near the aft, inboard corner of the right fuel tank skin panel, past the rear Z-stringer to the middle Z-stringer. It was apparently first stop-drilled when it was 7 inches long, then again at 9 inches, and the third time at 16 inches before the doubler patches were installed.
Whenever or by whomever the doubler patches were installed, they seem to have done as much harm as good. The investigators found that areas of corroded skin between the doubler patches had been filled in with non-structural “hard green” sealant instead of being more properly cut out and replaced. They also found that the aforementioned “workmanship” of the repairs was sorely lacking. Holes were double-drilled and/or elongated, oversize rivets of types not approved were installed, and edge distance and rivet spacing standards were not met. Some rivets as well were anchored not in good metal but in the hard green sealant.
It’s not important in the grand scheme of things, but the “stop-drilling” of the critical crack was another “reenactment” on the Air Disasters episode that completely missed the mark. The “mechanic” used what looked like a ¼ inch or larger drill bit when it reality you are supposed to use a very much smaller drill bit to stop-drill a crack. He also used a cordless-electric drill in the vicinity of a fuel tank! More importantly, stop-drilling is not considered a sufficient remedy for cracks longer than just an inch or two; cracks longer than that, especially in structural components, are supposed to be properly repaired by means of structural patches or replacement of the damaged parts with new.
In any case, the Chalk’s maintenance supervisor also denied any knowledge of when the initially 7-inch, then 9-inch, and finally 16-inch long crack in the wing skin had been stop-drilled or by whom. Even so, he should have been aware of and concerned about its very existence. Even though it was not visible to the flight crew during pre-flight inspections (because it was hidden by an aerodynamic fairing blending the wing into the fuselage at the wing root) it was still readily apparent to mechanics conducting proper inspections of the aircraft during which those fairings would have been removed.
Post-crash metallurgical analysis by the FAA and NTSB determined that the right wing of the doomed Mallard essentially unzipped itself from the airplane under normal flight loads. The likely scenario is that stress cracks originally started in the vicinity of a slosh hole in the internal, rear Z-stringer in the right fuel tank. (Note: the slosh hole prevents fuel from being trapped behind the internal framework and allows heavier water contamination in the fuel to move down to the drain at the lowest point in the fuel tank.)
The evidence suggested that the rear Z-stringer had cracked through and failed possibly several years prior to the crash. Its failure transferred more load to the skin itself, causing the problematically long crack to propagate forward to the middle stringer and aft toward the fuel drain and rear spar. Numerous, tiny, slow-growth, old, fatigue cracks were found in the rear spar lower cap and the middle Z-stringer lower flange, each of which had double-drilled fastener holes in them. Both areas also showed evidence of long-term rubbing or chafing and crack propagation as the wing was subjected to normal flight load cycles and flexing over time.
At a critical point when too much of the wing’s structural integrity and load-carrying capability had been compromised, the rest of the rear spar, the forward Z-stringer, and the whole of the front spar all failed in relatively quick succession causing the wing to separate from the airplane. As a pilot myself, I think that this scenario falls into the category of “worst nightmare” because there is absolutely nothing the flight crew can do about it or to maintain control of the airplane when something like this happens.
In the final Part 4, I will address how Chalk’s handled the maintenance on its small fleet of very old Mallard seaplanes and how they should have handled it differently than they did.
Dave Marion is the Technical Content Editor at Seaplanemagazine.com. As A&P and IA with more than 30 years of experience in aircraft maintenance, he is also a Commercial Pilot with Airplane, Single & Multi-Engine Land, and Instrument ratings. He has a BA from Colgate University in 1984 and also graduated cum laude from Embry-Riddle Aeronautical University (DAB) with a BS in Aviation Technology in 1990. He can be reached along with all of the editors via E-Mail: [email protected]
Editorial Note: We apologize for the somewhat low quality picture, we would have preferred to show you a beautiful flyby picture taken by Jose Fernandez. Mr. Fernandez authorized the use of his image, however the following stipulation was added: “as long as the article is positive about Chalk’s and not a critique about its maintenance record or accident of that particular plane”.