The error chain in action: Pilatus crash at Butte

According to Hemingway, a man goes bankrupt gradually, then suddenly. The same could be said of the way pilots crash airplanes: a series of small mistakes slowly build up until a final mistake suddenly ends the flight. A 2009 Pilatus PC-12 accident in Montana is tragic example, where a series of minor mistakes combined to create a fatal outcome for 14 people. There is a lesson here for all of us.

An imbalance

The flight began in Vacaville, CA (VCB), with the pilot and nine passengers on board. After a short hop to Oroville, CA (OVE) to pick up more passengers, the plane departed for Bozeman, MT (BZN) just after 11am PDT. Media accounts would later make a big deal out of the passenger load–six adults and seven children in an airplane with only 9 passenger seats–but this ended up having no significant impact on the flight.

PC-12 crash site in Butte
The Pilatus crashed into a graveyard just west of the airport in Butte.

That’s not to say that this decision was meaningless. Taking off 600 lbs. over maximum gross weight and with 4 passengers too many does suggest a casual disregard for the airplane’s limitations. While the Pilatus will haul a lot of weight, 600 lbs. is hardly a rounding error. This attitude may have come into play in a much more serious way, as we’ll see.

The flight progressed normally until about 22 minutes into the flight, when a fuel imbalance developed between the left and right tanks.  The Pilatus has a refreshingly simple fuel system, with two main tanks and one basic control–it’s either on or off. If a fuel imbalance occurs, the boost pump on the fuller tank automatically starts, forcing more fuel from that tank to the engine and solving the imbalance within a matter of minutes. There’s nothing for the pilot to do.

In this case, the system appears to have worked: the Central Advisory and Warning System (CAWS – a centralized annunciator panel in the Pilatus) recorded the right boost pump firing when a 70 lb. difference existed between the left and right tanks. The pump remained on for approximately four minutes, then went off. The fuel imbalance was solved, just the way the engineers drew it up.

About 1:18 into the flight, the opposite situation occurred, with the left wing holding more fuel than the right. This time, though, the imbalance was not fixed: the left boost pump fired and remained on. In fact, a few minutes after the left pump started, the right pump also came on, suggesting the left pump was not delivering enough fuel to the engine. Something was restricting the flow of fuel from the tanks to the engine.

The NTSB pinned the blame on ice in the fuel. The airplane was flying at 25,000 ft., where the outside air temperature was well below zero. As the fuel got colder, it turned to slush and prevented the left boost pump from working properly. The pilot’s first–but critical–mistake was not adding Prist (a Fuel System Icing Inhibitor) when he fueled the airplane. Prist lowers the freezing temperature of Jet A and is designed to solve this exact problem.

In fact, it seems the pilot never used Prist, at least as far as the NTSB could tell. Fueling records from FBOs showed no evidence of it being added on recent flights, and data recovered from the CAWS showed that the boost pumps had been operating multiple times over the previous few flights. On one 2007 trip, the boost pumps operated 260 times–a highly unusual situation. The likely reason for this would be ice in the fuel system, requiring the boost pumps to raise fuel pressure back to normal levels. This flight was a warning, one that was missed.

The pilot’s reaction

In cruise flight at 25,000 ft., the activation of a boost pump would be annunciated on the CAWS panel with a green light. This was not an emergency situation, but it was abnormal (especially both pumps at the same time), and it should have held the pilot’s attention. The Pilatus POH notes that if the fuel difference cannot be balanced, the pilot should “land as soon as possible.”

But the pilot flew on for almost 30 minutes before finally admitting there was a serious problem and diverting. While an imbalance of more than three bars on the fuel gauge is abnormal, the accident airplane showed a 15 bar differential between left and right when the turn toward Butte, MT (BTM) was made. It’s likely that the pilot, a professional hired by the airplane’s owner to fly his family, was focused on completing the trip with a minimum of hassle. Diverting, especially when fairly close to the destination, must have felt like defeat.

PC-12 fuel gauges
The pilot waited to divert until the fuel imbalance was a major problem.

The pilot had little choice, though. He was in trouble, and he seemingly knew it–busting two altitude assignments and turning for BTM before asking Air Traffic Control to change his flight plan. He was rattled.

As he flew on, the imbalance only got worse. Like most turbine airplanes, the pumps in the Pilatus deliver far more fuel to the engine than required for normal operation, and the excess fuel is returned to the tanks. But with the left tank basically iced up, the result was that the right boost pump was the only one working. In a circuitous way, it was essentially moving fuel from the right wing into the left wing.

By the time the airplane was close to Butte, the left wing held 1,368 lbs. of fuel (full) and the right tank held 66 lbs. (basically empty). That massive imbalance–26 bars on the fuel gauge–surely made the airplane difficult to control, especially considering the PC-12’s long wingspan. But while an NTSB analysis determined the Pilatus was flyable, the pilot came in fast and high: just 1.8 miles from the airport he still needed to lose over 3,500 ft. of altitude.

Knowing he would not make the runway straight-in, the pilot did a smart thing and turned to enter the left downwind for runway 33. As the airplane slowed, it would have taken more and more aileron to counteract the left wing down forces. During the turn, the pilot lost control of the airplane and crashed into a cemetery just west of the runway. All 14 on board were killed.

The final NTSB report gets complicated, but comes up with a familiar refrain: pilot error. Specifically, they list three main causes: the pilot’s failure to use Prist, his failure to take remedial actions in a timely manner, and his loss of control while maneuvering. The second one may be the most important. Pilots are loath to divert or land short, feeling like they failed at their “mission.” In this case, every minute of delay lowered the odds of a successful outcome, as the imbalance grew more severe. Even descending to warmer air may have helped the problem.

Lessons

Some of the specifics of this accident, like Prist and turbine fuel systems, don’t apply to all (or even most) pilots. But there are a couple of universal lessons here that are a good reminder. If a highly experienced pilot can crash a well-maintained, high performance airplane in good weather, it should make us think.

First, and most obviously, it’s a good idea to follow the manual. It may seem overly parental at times (or written by the Legal Department), but most things are in there for a reason. In this case, adding Prist may have seemed like a minor detail, but Pilatus had very good reasons for requiring pilots to use it anytime the airplane was operated in temperatures below zero. They had done extensive testing with the fuel pumps knew that ice in the fuel system was bad news, especially in a single engine airplane.

Beyond just memorizing the POH, it’s worthwhile to understand the story behind the checklists. Why did they require Prist? Knowing the system in detail might have clued the pilot into the serious nature of the limitation.

Secondly, don’t let one mistake compound another. Most pilots are familiar with the concept of an error chain, or the “Swiss cheese model” of accidents, where a number of errors are missed by pilots and, when added together, create an unsafe situation. We learn about this in primary flight training, but our eyes may glaze over when the instructor drones on about some esoteric accident theory. Does this stuff have any use in real life?

This accident proves that it does. In this case, the pilot made a number of mistakes that were not fatal on their own. He didn’t put Prist in, but that didn’t kill him right away. He didn’t divert or descend when the fuel problem first became apparent, but that didn’t kill him. He was too high and too fast when he reached his alternate airport, which usually is not fatal in an airplane. But all of those things put together created a scenario with almost no margin for error.

Put simply, we need to admit it if we’ve screwed up. If we admit it in time, we can break the error chain, get on the ground and prevent an accident.

This article is based solely on the NTSB report covering this event.

18 Comments

  • There appear to be some holes in the NTSB theory of frozen fuel tanks that are rather obvious:

    1) If frozen fuel was the problem it should have affected both fuel tanks equally. A mechanical malfunction in the fuel system seems far more likely to be a chief cause of the imbalance.

    2) The aircraft was fueled on the ground in Vacaville, California, an airport at 117 ft MSL and therefore not “cold country”. The fuel on board should have been relatively warm, say 60-70 deg F (or 15-21 deg C) at the start of the flight … if the OAT af FL250 was say, as low as minus 5 deg C at FL250, at a typical fuel tank cooling rate of around 3-4 deg C per hour with moderately cold OAT, it would have required something like 4-7 hours for the fuel in the tanks to approach 0 deg F where it would then BEGIN to freeze. The NTSB scenario seems to assume a thoroughly cold-soaked frozen fuel load that uniformly freezes (but strangely enough, only on the left side), which is very doubtful.

    3) The fact that the aircraft was significantly over gross, loaded with CHILDREN, of all things, far in excess of the number of seats on board, seems to me to be far more telling of the negligent mindset of the PIC (no doubt, under “orders” from his boss, the aircraft owner) than anything. This was a pilot who was not only experiencing “get-there-itis” but he had legally and professionally abdicated his authority and his responsibilities as PIC from the get-go. That a pilot with that mindset would then ignore operational “details” like failing to land the aircraft with an obvious fuel system malfunction (regardless of its cause) is not surprising at all.

    This accident scenario seems less like the proverbial “accident chain” than it was an accident that was baked into the cake … beginning with the pilot’s decision to take off overloaded and then continue flight in complete disregard of the safety of flight.

    • Correction – in my item no. 2 above, I made a typo error – the fuel would not freeze until it got to 0 deg C, not 0 deg F. Again, it would have taken at least four hours for the fuel to freeze in the tanks – longer than the approx, two and a half hour flight time to Bozeman at roughly 300 cruise speed for the Pilatus PC-12. Not to mention that at least 30-40 minutes of that flight time is climb and descent time below FL250.

  • Duane, I think the temperature at 250 would be more like -25 or -30 degrees. That’s what I typically see there.

    Also, if he never used Prist, it’s possible there was some accumulated stuff in the tank.

  • John – I downloaded and read the NTSB accident report. As you wrote, it’s very complex and long (92 pages)!

    Some points that I gleaned from reading the NTSB report:

    1) You’re right, the OAT was quite a bit colder than the temp that I cited above (- 5 deg C) – the accident report said that the OAT during the accident leg averaged -32 deg C, and was -40 deg C at FL250. But I don’t think that matters all that much on a flight of the length of this one – due to the time it takes for onboard fuel temps to drop from 15+ deg C to below freezing. In fact, the NTSB report said that the fuel system began cycling (indicating a fuel pressure issue) at only one hour 13 minutes into the accident leg. NTSB was able to cite only one other PC-12 incident that allegedly involved fuel system freeze-up at similar OAT – but the fuel pump cycling in that case began more than three hours into the flight.

    2) NTSB bases its conclusions in part on post accident testing that they performed on the PC-12 fuel system, including the booster pumps and filters (which supposedly clogged with ice crystals) at -30 deg C. They (not surprisingly) found significant degradation of the fuel system performance at that temp. But they also did not witness a catastrophic fuel system failure (i.e., completely clogged fuel filter) either.

    Keep in mind that the -30 deg C temp in the NTSB post-accident testing was not OAT, that was the FUEL TEMP. There is no way that the fuel temp in this accident got anywhere near -30 deg C, probably not even below freezing.

    The NTSB also used fuel in their post-accident testing that was completely saturated with water … how likely is it that jet fuel in airport storage tanks will become completely saturated with water? That seems doubful Saturation is merely the limiting condition, not the likely condition, of the water content in the fuel.

    3) It is true that the Pilatus PC-12 AFM says that an anti-ice fuel additive must be used in fuel whenever the OAT is below zero, but that is an obviouly conservative guidance unless it’s a very long flight at very high altitudes. Ice crystals in jet fuel cannot form unless the fuel itself is below 0 deg C.

    Should the Prist additive be used when flying high? Sure. It’s always good practice to use a margin of safety. The accident pilot should have used the Prist … but it probably wouldn’t have prevented the fuel system problem he experienced.

    4) Nowhere does the NTSB report explain why only the water in the fuel in the left tank would freeze, and not also freeze in the right tank. The fuel in both tanks came from the same airport storage tanks, after all. The obvious differentiator here was between left and right fuel systems, and not in the fuel.

    5) NTSB is always searching for the “smoking gun” in their investigations, so that they can claim they “found the cause”. NTSB catches flak whenever they don’t come up with a smoking gun, so to avoid criticism, they often seize on a “cause” that isn’t necessarily well proven.

    6) It seems pretty clear that the fuel system on the accident aircraft was malfunctioning. Even if ice formed only in the left tank, and thereby clogged only the left fuel filter, the system contains a bypass relief valve that is supposed to open and allow unfiltered fuel to flow. Something else was wrong with the left side fuel system.

    7) One can argue the physical evidence forever, but we’ll never really know the specific fuel system failure.

    We can, however, conclude that the PIC failed to operate safely by continuing the flight even with an obvious and growing imbalance between right and left tanks.

    The lesson here for all of us pilots is, if the aircraft is not performing correctly, make a precautionary landing. The pilot didn’t, and 14 people died.

    • Agreed, I do not think not adding Prist had anything to do with whatever fuel system malfunction they were having. Prist in theory on lowers the freezing point of jet fuel by 6 or 8 degrees at best, particularly the newer thinner jet fuels we have. I’m just pulled a certificate of analysis from the last load we received from ConocoPhillips. The freezing point is minus 52.4C, with the minimum standard being minus 40 C. Adding Prist may get it to minus 60 at best.

      Their theory is fowler, of course it is always easier to blame the dead pilot that can’t defend himself than have the manufacture step up to the plate, which also is standing side by side in the investigation..

    • Duane,

      With all due respect,

      You’re points regarding the science behind temperature and its effect on jet fuel may be correct, but you completely miss the point. Our duty as pilots is to follow procedure, not question its validity. Seems to me you’re more interested in showing a lack of respect to the NTSB and their ugly burden of post-accident analysis. Every pilot knows the unfortunate truth – the “smoking gun” is much too often the pilot. I’m thankful that the NTSB is there once again to articulate this unfortunate reality.

      The mishap pilot made a series of mistakes that compounded and led to this tragedy. The critical link on the accident chain was his interest in questioning the validity of a known procedure, adding Prist (the temperature remarks within this limitations are a moot point to most PC12 pilots – it’s always added regardless of temperature). The mishap aircraft’s fuel system was not malfunctioning; it was compromised by nothing other then slushy freezing jet fuel. Shame on you Duane for your assumption at the expense of the NTSB. Those amazing Swiss presented the pilot with enough information to have warranted a timely and immediate diversion absent of the resulting mishap. The scientific term for that is “no-brainer”. The subsequent series of poor pilot decisions resulted in 14 fatalities.

      By the way, the cost of Prist is pennies per gallon. Six cents on six bucks irregardless of temperature. Do the math on that. This mishap pilot was clearly bottom-line orientated, another result of missing the point. Perhaps he was intimidated by his employer; although I suspect it was simply poor habit and behavior carried through from previous experience. Our “professional” duty demands influencing our bosses and peers relentlessly in the interest of safety and procedure. Waiving a procedure or ignoring a limitations is unacceptable; hence the “pilot error” annotation. Duh.

      • Craig – I completely disagree with you that fuel freezing caused this accident. The absolute proof of that is that it only happened to one side of the fuel system. If it had been fuel freezing, it would have equally affected both sides of the fuel system. The entire fuel system would have suffered degraded performance and probably forced an immediate landing by the PIC, or worse.

        Shame on me for criticizing NTSB? Gee, I didn’t realize that making a well-founded engineering-scientific and common-sense based argument that disagrees with the bureaucratic meanderings of the NTSB is the cause for any pilot or commenter to suffer any “shame”. Who appointed you the judge of shame?

        The NTDB is not an altruistic organization as you imply – they are government (i.e. taxpayer) paid employees with natural bureucratic self-preservation tendencies built in to everything they do. They are not unpaid volunteers – but highly paid bureaucrats (almost certainly earning far more in salary and benefits than the accident pilot, or most other non-major airline commercial pilots).

        I do agree with you that the pilot should have added Prist, and he should have paid attention to the ample alarms and warnings given out on his PC-12 panel system and made a precautionary landing as soon as the fuel system stopped performing correctly.

        None of my comments above were in the slightest supportive of the pilot’s actions – indeed I condemn the PIC’s actions for failing to respond to an obvious malfunction of the fuel system … a malfunction whose specific cause will never be known, other than that we KNOW it wasn’t caused by freezing that only occurred in one side of the fuel system.

  • After reviewing the NTSB report, I see that one of their primary recommendations was to have more strongly worded placards and Operating Handbooks. I’m thinking it might be better to require that turbine aircraft with an operating altitude of greater than say, 15,000 ft, have a fuel system which does not require the additive. It appears that there are a lot of variables involved with the addition of Prist which can lead to problems. Since other airplanes are built which do not require the additive, it is obviously not a technical problem.

  • Duane, you don’t by any chance happen to be involved in the litigation associated with the accident? Your point seems to be, althought the pilot recklessly disregarded the aircraft limitations and numerous indicators of a problem, there MUST have been a problem with the aircraft that caused this poor pilot to be lead to his demise.

    The families of the njured parties are looking for someone to blame and in the United States, blame means money. The attornies involved will reap a windfall while suffering no personal or emotional loss.

    This was a tragic accident brought about by the incorrect operation of the aircraft. The contributing factors may include pressure from the owner, lack or improper training, failure to use Prist, etc, but the pilot had the opportinuty to change the outcome but for some reason failed.

    • Tailhooker – to answer your question, no I have no involvement with any of the parties in this case. I am just a pilot and an engineer, and being an engineer I understand at least a little about the laws of physical science and chemistry.

      Because of that I am having a very hard time understanding why said laws of physics and chemistry (i.e., the freezing temperature of water suspended in jet fuel) behaves one way on the left side of the aircraft fuel system, but behaves entirely differently on the right side of the fuel system of the very same aircraft, at the very same time, exposed to the very same OAT, both sides of the fuel system using fuel sourced from the very same supply.

      Nobody has been yet able to explain that anomaly. Can you?

      That the fuel system obviously malfunctioned, due to whatever cause, does nothing to relieve the PIC of his responsibility to operate the aircraft safey … by making a precautionary landing as soon as he noted his inability to correct the fuel system imbalance.

      The fact that the fuel system malfunctioned – which is not debatable given the downloaded system data, and the laws of physics – does not, however, shift the primary blame for this accident to the aircraft manufacturer, nor to the maintenance department that either did or did not properly maintain this aircraft. The primary liability is on the pilot who had an opportunity to respond properly, and chose not to.

      • Duane, I appreciate your comments, and you’re right that there are other scenarios besides just the Prist one. But I think it’s going a little overboard to say that both tanks would behave identically and on the same exact schedule. One may have had more crud in it, accumulated over a few years of flying without Prist. Maybe the right pump was icing up as well, but could still deliver pressure.

        My point is, we’ll never know for sure what exactly happened. That’s an engineering debate anyway. For us pilots, the point is “what you see is what you get.” It’s our job to react and react in a timely manner.

  • John: read your blog about the utility, value and perfisnce of used / rebuilt airplanes. Good work.
    I am working on an industry effort to rebuild , restore and re – equip the GA fleet- named The Affordable Aircraft Initiative. We have a number sponsors and have 6,000 completed marketing surveys and 1,600 qualified leads so far.
    Please call me
    602-625-6814
    Tom Bliss
    Publisher
    AVweb

  • I am late to this discussion, but I ran across this article and reviewed the NTSB report and PC-12 POH in more depth than I had previously. Here is what I think happened, which is somewhat in disagreement with the NTSB, but may answer some of the L vs R questions raised above:

    1) The filter (only one after the L and R plumbing is teed together)definately iced up as shown by both boost pumps cycling. The PC12 apparnetly doesn’t have enough jet pump fuel pressure to avoid this with an iced filtereven though there is a bypass.
    2) An imbalance can begin to develop due to variations in output between the L and R pumps.
    3) The L boost pump activated to correct the imbalance. I don’t have details on the PC12, but it is typical that the activation recorded means that power to the pump was recorded, not that it was perfoming properly. The latter would require additional pressure sensors for indication.
    4) It is apparent the the output of the L pump was low, causing a rapid increase of fuel imbalance due to all fuel being fed from the R tank, plus transfer from the R to L tank through the motive flow system (motive fuel sent to L jet pumps cannot return to engine due to insufficient pressure to overcome R pump fuel pressure at the check valves, so it dumps into L tank backwards through jet pump inlets).
    5)NTSB atrributes L low output likely due to plumbing/component icing, but I think it more likely that the L pump simply failed, either partially or completely. This is because icing typically occurs at small passages such as filters (only one for both sides). Other possibilities are pump inlet screens (no mention of them I can find) check valves (never heard of them icing, but suppose it could happen) or plumbing (the plumbing icing on the BA 777 didn’t restrict flow until it broke loose and blocked the downstream fuel/oil heat exchanger).
    6) The whole problem can be made worse by failure to drain tanks, which the NTSB alludes to. A business jet was lost due to this a number of years ago.

    A few notes on icing, PT6’s and Prist:
    1) It is dissolved/entrained water in the fuel that causes the problem, not fuel freezing unless temperatures are a lot colder.
    2) PT6 powered aircraft (and others) I am more familiar with require Prist at all times, not just when cold temperatures are expected. This does not mean that it is always used.
    3) PT6 powered aircraft are required to use Prist because PWC requires the use of an airframe fuel filter, which would require the complication of airframe supplied filter deicing (such as with bleed air) for certification without Prist. Aircraft certified without Prist use only an engine fuel filter anti-iced by engine oil.
    4) Operators and fuelers don’t like Prist because of inconvenience if it is not premixed or mixed by refueling equipment, There were also toxicity concerns, particularly in Europe, with the original DIEGME, but this is of lesser concern with the current EGME. There was one business jet equipped with an onboard storage and mixing system, but it was a maintenance nightmare, so that it was removed, and airplane design changes were made for recertification without Prist.

  • Very late to this one…

    Lots of engineering/technical/judgement steps (and valid discussion) to get to the final mistake and the final mistake didn’t need to happen either. Even if it was an engine out…a single engine airplane with 3500′ of excess altitude at the airfield didn’t make the runway?

    You’ve got an emergency, you declare on Unicom/tower, take whatever @#$% runway you need pattern be damned and do a 360 off the arrival end holding enough airspeed to cover your imbalance…if you run off the end with excess rollout, better than plunging in from a stall from too little.

    Fly the airplane!

  • This was a tragic mishap. There were multiple friends and acquaintances among the victims. I think you have to work hard to explain this accident in a way other than what the NTSB report says.

    It appears the pilot got out of the habit of using Prist and simply forgot about it. Until about a month before the accident, the aircraft was based at San Bernardino International. At least some of the time, premixed Prist was used to fuel the aircraft at that airport (with or without the knowledge of the pilot). At the time of the accident, the airplane was based at Redlands where premixed Prist was not available. There’s no evidence the pilot ever added Prist to fuel that didn’t have it premixed.

    This is not the only instance of fuel pumps cycling at high altitude in low temps on a PC-12. The difference is the pilots in the other occurrences I have knowledge of landed well short of their destination and stored the aircraft in a heated hangar.

  • Even later comment based on research for purchase of a PT6 powered plane. The purpose of Prist is definitely, as already mentioned, to prevent the freezing of the water dissolved in the fuel that comes out of solution as the fuel cools. A 2014 paper in Science Direct shows that the solubility of water in JP-4 decreases linearly from 140 ppm at 30*C to 40 ppm at -10*C. Prist is hygroscopic and mops up this water so it does not freeze.

  • Duane-How could you imagine that fuel would not reach ambient air temp in a metal aircraft with 200+ knots of air exchange within minutes? They tested the fuel filters and found them working. You might think the NTSB is working with the manufacturers, but take a less consipiratorial view. If that were the case, there would be less incentive for pilots to take greater responsibility to follow guidelines. Sometimes agencies have to weigh how human beings will react. And, quite frankly, pilots are often wanna-be engineers who think they are invincible.

Leave a Reply

Your email address will not be published. Required fields are marked *