Human Factors: A watered-down test flight

It’s like the old light bulb joke: How many airline pilots does it take to sump a fuel tank in a general aviation airplane?

Apparently, the answer is four, because three failed to do it correctly and ended up crash-landing in a North Carolina field because of it.

It all started when a pair of airline pilots journeyed from Florida to North Carolina to check out an airplane they were looking to purchase, a PA-32R Lance, an IFR certified six-seat, retractable gear, high performance hot rod.

Being meticulous professionals, the pair spent the first two hours of their day-long visit reviewing the airplane’s paperwork to ensure proper airworthiness, and talking to the airplane’s owner about its service history.

Satisfied, it was then time for the test flight.

Both of the prospective buyers had Lance time, but it was a loooooong time ago, at the very dawn of their professional careers. As young commercial pilots, they had worked together flying Lances for a 135 outfit in Alaska.

One of the pair, in his narrative report to the National Transportation Safety Board (NTSB) after the accident, said “neither one of us had flown a single engine piston aircraft in many years.”

In fact, it had been 25 years since either pilot had set foot in an airplane like a Lance.

But to their credit, they recognized this weakness and asked the owner to secure a qualified CFI to re-introduce them to the type.

The Pilots

One of the two prospective buyers sat left seat for the first test hop and was PIC.

The CFI who was to check them out sat right seat, and the other prospective buyer sat in the third row in the back, taking one of the forward-facing seats. The Lance has a club-style seating arrangement in the back, with the second row facing rearward for a more communal, interactive passenger experience.

The PIC was a 54-year-old male who held ATP, CFI, and Flight Engineer tickets, along with multiple type ratings. He listed himself as an occupational pilot, with more than 24,000 hours in his logbook. The PIC actually had 645 hours in make and model but, as already mentioned, that was more than two decades ago. He had a current First Class medical, and his last flight review was about two weeks before the accident. In a Boeing 757.

The CFI in the right seat was a 57-year-old male from Pennsylvania. He also held an ATP, making him an airline pilot for the purposes of our joke, but he clearly never used his ATP to work for the airlines. He was on a Second Class medical, and listed his occupation as “other.” His total logbook time was a hair over 3,000 hours, suggesting he was not a full-time CFI. He only had 23 hours in make and model, a fraction of the PIC’s hours, and none of those in the last 90 days.

We don’t know much about the pilot in the back seat beyond the fact that he was a 53-year-old male, as the NTSB doesn’t collect pilot information on passengers.

But from the written statement of the PIC, we know he was a longtime friend, colleague, and fellow airline pilot, who would have been a co-owner, had they bought the airplane.

The Flight

After an initial preflight, the trio had the aircraft towed to the pumps, although it’s not clear why the airplane was towed rather than taxied.

At the pumps, 10 gallons of 100LL were added to each side, bringing the fuel load up to about 38 gallons total.

After fueling, the PIC’s report to the NTSB says that the three pilots performed “a thorough preflight” and “sumped the fuel tanks thoroughly.”

They also did a weight and balance at this point, and then ran through all the pre-start checklists.

The engine “started perfectly on the first attempt,” one of the pilots reported.

Can’t you just see the prospective buyers? Oh, yeah, we found THE airplane.

Next came a slow taxi, a good 15 minutes long, as the PIC reported he was “still getting used to smaller aircraft again.”

This adjustment also extends to the first leg of the flight.

After an uneventful takeoff, the PIC ends up flying a self-confessed “significantly” wide pattern, and the CFI instructs the PIC to close it up next time. They do one landing and power up to go around for another.

As they climb out, the CFI reminds the pilot that they’ve been on the right tank since engine start and the PIC says he’ll switch tanks as soon as they level off.

On the downwind leg, he switches on the fuel pump, changes to the left tank, and verifies that fuel pressure remains good after moving the selector. He then focuses “on a closer downwind leg.”

Or he does until the sudden silence up front demands a shift in his focus.

The Final Moments

In the PIC’s words, midfield downwind for Runway 17, “the engine very abruptly stopped running.” The prop is windmilling, and there’s zero power.

He works the throttle, but nothing happens. He snaps the fuel back to the right tank. No joy.

He turns in toward the intersecting Runway 23 and establishes best glide. The CFI is trying to re-start the engine with no luck.

The PIC reported that “it quickly became clear that we weren’t going to make the airport.”

He heads for an open field — one bordered on three sides by deep irrigation ditches, while the CFI makes the “Mayday Mayday Mayday” call.

The PIC holds the flaps and gear in reserve, deploying them “a couple hundred feet” above the ground.

The Accident

The off-airport touchdown is actually pretty sweet, “firm but steady,” and just past one of the ditches. The aircraft is not damaged and slowing down nicely — just not slowing down quite fast enough.

The next ditch is filling the windscreen dead ahead. The CFI shouts “ditch,” the PIC slams on the brakes, the tires slide in the dead winter grass, the airplane slows and…Thump!

The nose gear drops into the ditch — which is fully three feet deep — and the Lance belly flops across the scar in the landscape.

One blade of the three-bladed Hartzell constant speed prop is bent, the engine mount is bent, the empennage is bent, and the left entry door is bent and jammed. Yeah, the airplane suffered a bad case of the bends.

The three pilots secure the cabin and evacuate, with no injuries save, perhaps, to their pride.

Actually, the pride injury happens much later when they learn what caused the engine to fail.

The NTSB

Because after the crash, investigators drain a full three quarts — QUARTS — of water from the left fuel tank — the one that stopped the engine.

Additionally, they drain a pint of water from the fuel lines. There was no water in the right tank, the tank used for the taxi, run-up, and first circuit of the pattern.

How on earth could this have happened?

Upon interviewing various people associated with the airplane, the NTSB investigators learned that the Lance had been sitting for 11 months before the accident flight, and for “several” of those months was parked outside on the ramp.

That said, the left wing fuel cap appeared to have a good seal, leaving the entry source of the water a mystery.

Naturally the NTSB throws the PIC under the bus with a finding of “inadequate preflight inspection, which resulted in a total loss of engine power due to fuel contamination.”

Analysis & Discussion

But was the preflight truly inadequate?

In point of fact, according to the narrative, there were two preflights: An initial one at the airplane before it was towed and a more detailed one after fueling.

But clearly, the left tank had a lot of water in it.

So how did three ATP pilots, all of whom were also CFIs, collectively do a pair of preflights and NOT sump the fuel tanks properly, missing nearly a gallon of water in the left fuel tank?

There are a couple of things to consider here, and the first one is expectation bias.

As pilots, we don’t really expect to have a gallon of water in a fuel tank. Usually there’s no water at all. And, if there is any water, it’s generally a small amount. So we are expecting to see that little mercury-like bubble in the bottom of the fuel strainer.

Why didn’t the pilots recognize that the sample was clear, not blue? I don’t know.

But from experience, in some lighting, the 100LL blue looks nearly clear. Perhaps they were so focused on looking for the tell-tale split between fuel and water that they didn’t register the fact that the sample was pure water.

Or maybe, just maybe, there wasn’t any water in the sample when they sumped.

How could that be? The water was there before takeoff, we know that. It didn’t just magically appear in the pattern.

While it’s not clear if they sumped on the pre-fueling initial preflight, we do know that they refueled the airplane before they “thoroughly” sumped the tanks, which actually isn’t a bad procedure to ensure that there’s no contamination in the fuel you just purchased.

But there’s a problem. All that new fuel gushing into the tank “stirs up” any existing water that might already be in the tank. It causes it to blend with the fuel, at least for a time.

How much time? AC00-34A, which dates from the 1970s but is still active, states that the “settling time” for water in avgas is 15 minutes per foot of tank depth.

So how many inches deep are Lance’s fuel tanks — or any other GA airplane’s for that matter?

I’ve got no idea, and couldn’t find a number, but the “universal” Fuelhawk dipstick-style fuel gauges come in 11-inch, 16-inch, and 26-inch varieties, suggesting that most GA fuel tanks range between 7 and 22 inches deep (allowing four inches “on top” to hold the straw comfortably).

Fifteen minutes per foot works out to a settling rate of about a minute and a quarter per inch. That, in turn, gives us a GA fuel tank settling range — rounding up the seconds for safety — of between nine and 28 minutes. The deeper the tank, the longer the settling time.

Needless to say, sumping right after fueling is nearly worthless unless your airplane has the world’s shallowest fuel tanks, in which case, I suspect that water would be the least of your problems.

Which is not to advocate that we shouldn’t sump after fueling.

Other sorts of contamination, such as dirt, rust particles, etc., will fall out more quickly, so sumping after fueling — even with an immediate departure — is still warranted (although I’m not that confident that fresh particles would quickly find themselves to the sump drain). Still, it does no harm.

But a better order of battle would be: Fuel, hit the head, sump, hit the sky.

Failing that, changing fuel tanks after taxi and before run-up might be one last worthwhile tactic. Especially in this case, with the extended taxi, the water would have surely come out of the solution, or at least enough of it to stop the engine on the ground in the run-up area, instead of in the air in the pattern.

The Takeaway

There are at least five takeaways from this accident that I think we can all benefit from.

First, if an airplane has been parked for a time — especially out on a ramp — sumping before fueling is warranted, even if you are not yet doing the full preflight.

Second, be alert to expectation bias when sumping fuel. The human mind learns to expect what normally happens, to the point that we sometimes are blinded to contrary input.

Third, know that the act of fueling will stir up water — either water already in the tank or water in the fuel supply.

Fourth, know your tank depth so that you have at least a baseline settling time to work with, and consider the timing of your other pre-departure activities so that you give any water in the fuel time to settle.

And fifth, if you don’t already, change your fuel selector at some point during your ground ops before takeoff.

Because while a long, cool drink of water might be good for pilots, it’s not so great for airplane engines.

The Numbers

Want to read more? Download the NTSB’s final report here or view the items on docket here.