While simulator training provides an invaluable resource for training rare, abnormal, and emergency events, it can unintentionally foster mindsets and biases that interfere with our successfully handling of real-world situations. Following are two events from the NASA ASRS database that demonstrate this mismatch. The first (ASRS event #1691666) is from a crew that misapplied their windshear response procedures because of the training scenarios used in simulator training.
Pilot Flying narrative: …The windy conditions had been a subject of discussion and mentioned during the briefing of the arrival and approach. During the descent we had some light chop and the surface reports were easing up as we made the descent on the RNAV Approach. We were fully configured, on speed, Vref +15, [with] the runway in sight [and] the autopilot off.
At about 1,500 ft. AGL on final, we received a “windshear ahead” warning. Since it was a lot smoother ride than anticipated and I didn’t have any other cues about wind shear with either the airspeed or vertical speed, it took me a couple of seconds to interpret the warning [during] which time it went away. Since we remained on speed and profile after the warning went away, I continued the approach and landed.
In debriefing the approach, we both agreed that a go around should have been initiated, but that a lack of other cues was a factor in a few moments of hesitation about whether initiating a go around was necessary. I felt like we otherwise had good monitoring, cross checking, and communication about the aircraft state throughout the arrival and approach. So often, wind shear training in the simulator is accompanied by the cues that help us identify the onset of wind shear, wind shift, airspeed vertical speed etc. I need to brief or at least tell myself to be ready to go around when directed by the predictive computer.
The reporter brings up important points. We use simulator windshear training profiles to teach the full range of indications and procedures. To get the most from our limited simulator time, airlines pack a wide range of experiences and stimuli into a single training scenario. We even instruct pilots to verbally announce when they would initiate their go around from increasing turbulence, but to delay their actual recovery maneuver until they receive the automated windshear warning (when the windshear conditions have progressed to the most challenging state). While this trains us to handle the worst possible windshear events, it unintentionally instills a mindset that windshear events start with turbulence, which intensify to flightpath deviations, and finally progress to an automated warning as we enter severe windshear conditions. In this case, the pilots got the automated warning first while experiencing “a lot smoother ride than anticipated”. While they were mentally processing this mismatch from their training, the audible warning ceased. At this point, it felt natural for the pilots to classify the warning as an anomaly or a transient condition, so they continued to land. Procedurally, a windshear warning on final approach requires an automatic, non-discretionary go around.
Another factor is the strong incentive to land from each real-world flight versus our experience in simulator training where going around feels much more appropriate. The vast majority of daily flights end with landing, while a large percentage of simulator flights end with a go around or missed approach. Going around from a real-world flight tends to feel a bit like failure – and pilots hate failing to get their passengers to their destinations. This creates a strong incentive to rationalize landing.
In this second situation (ASRS event #1876901), a crew discovers a fuel imbalance which they quickly diagnose as a fuel leak.
Captain’s narrative: I operated Aircraft X ZZZ-ZZZ1 on DATE with a suspected fuel leak that resulted in an engine shutdown and diversion to ZZZ2. I was the Pilot Monitoring and the First Officer was the Pilot Flying on this flight. Around the ZZZ3 area the left center tank low pressure lights flickered and I turned the left center pump off with about 400 pounds remaining in the center tank. I didn’t see a fuel imbalance between the left and right tanks at this time. Several minutes later the center fuel tank was empty and I turned the right center pump switch off. In addition at this time I didn’t notice anything abnormal about the cross feed valve selector or light. At approximately 70 miles SE of ZZZ2, I heard the First Officer say we had a fuel imbalance and looked over to see the fuel IMBAL light illuminated on the right fuel tank. At this point I noted a 1,000 pound fuel imbalance and that in my experience the fuel in the right tank was decreasing at an abnormally fast rate. It had only been around 10 minutes since I turned off the center fuel tanks. At this point the First Officer and myself thought a fuel leak was plausible due to a 1,000 pound fuel imbalance occurring in around 10 minutes and observing an abnormally high rate of fuel decreasing from the right tank. I proceeded to run the fuel leak-engine QRH and contacted the flight attendants to request one of them check for a fuel leak/mist coming from the back of the right (#2) engine. While running the QRH, I felt time pressures to stop the fuel imbalance before it led to adverse control issues. At step 5, I recorded the total fuel and time (I don’t remember what I recorded) and proceeded to the condition statement in step 6. After reading the condition statements and based on the abnormally high rate of fuel decrease in the right tank I proceeded to step 7. At this point I thought I confirmed an engine fuel leak because we were now at 1,200 pounds imbalance in 10-15 minutes. Far greater than the 500 pounds in less than 30 minutes that the QRH states. By this time the FA reported not seeing any fuel leaking from behind the right (#2) engine. Knowing that we were going to shut down the #2 engine I requested priority handling and requested a lower altitude. In addition we requested vectors to ZZZ2. By the time we shut down the #2 engine we had a 1,400 pound fuel imbalance. I’d like to add that after shutting down the #2 engine the QRH calls for the cross feed selector to be opened. The cross feed valve opened normally with no abnormal indications and closed normally with no abnormal indications several minutes later when we decided to even out the imbalance by burning fuel from the left tank. .... While rereading the Fuel leak-engine QRH step 6 after the event, I realize that my decision to proceed to step 7 was based on what I read the step 6 condition statement to say of “the fuel quantity is decreasing at an abnormal rate out of the right tank”. Rather than basing it on what the condition statement actually said “or the total fuel quantity is decreasing at an abnormal rate”. This was a mistake. I can say that the fuel imbalance QRH checklist would probably have been more appropriate to call first. However this wasn’t an imbalance that took time to develop. We experienced the fuel IMBAL light and an abnormally high rate of fuel decrease from the right tank around 10 minutes after turning the center pumps off. I didn’t notice any fuel imbalance prior to turning the center tanks off. I don’t think that it was unreasonable to run the fuel leak-engine QRH in these circumstances.
First Officer’s narrative: … While looking at the fuel quantity, I noticed that quantity in the right tank was decreasing at a higher-than-normal rate, and the fuel imbalance was approximately 1,000 pounds. The Captain pulled out the QRH. I also may have verbalized “do we have a fuel leak,” or “is this a fuel leak.” This was probably confirmation bias on my part. I recently completed my annual simulator training, about 2.5 months ago. The scenario that I had on day 3 was depart ZZZ4 for ZZZ5. During that sim session we had a fuel imbalance shortly after takeoff, which was actually a fuel leak. So, the scenario we were experiencing in the airplane seemed similar to a recent training event. Since the Captain and I both thought a fuel leak was possible, he started to run the Fuel Leak checklist in the QRH. The Captain also contacted the flight attendants and asked them to check the right wing and engine for any visible fuel spray. The flight attendants reported back to us that they didn’t see anything. However, based on the fuel imbalance rapidly getting worse, exceeding the 500 pounds within 30 minutes the Captain and I confirmed a fuel leak. We requested priority handling, lower altitude requested, as well as vectors for ZZZ2. Once on a localizer intercept vector, the Captain took controls and landed the airplane. We taxied clear of the runway, and emergency personnel (crash fire and rescue) visual inspected the airplane to make sure fuel was not leaking from the aircraft. We then taxied towards the gate. Since I was the Pilot Flying, and the first person to notice the imbalance I should have asked for imbalance checklist first.
While neither pilot reveals the maintenance finding in their report, their tone implies that the cause of the fuel imbalance may have been a malfunctioning crossfeed valve that allowed fuel to transfer from the right wing fuel tank to the left resulting in the imbalance or perhaps a fuel quantity sensing probe issue. The First Officer reports that their recent simulator training event included a fuel leak scenario which biased them to interpret their imbalance as a leak. The fuel leak checklist directs an engine shutdown, while the fuel imbalance checklist does not. The FO admits that his bias may have influenced the Captain to follow the leak scenario. The crew elected to divert and land which was the safest call when single engine, but shutting down a engine may have been unnecessary.
Another interesting human factors bias is crew behavior created by the first conclusion expressed. Once the FO mentioned “fuel leak” and the crew established a fuel leak mindset, they appeared to abandon further analysis and inquiry. Granted, fuel leaks generate a high level of urgency which encourages prompt action. The Captain even mentions “time pressure” to mitigate the leak. They did take the additional analytical step of asking the cabin crew to check for fuel misting from the suspected engine. Even after the cabin crew reported no misting, they didn’t investigate further. Ideally, the crew could have compared expected fuel burn rates from their flight plan against the actual totals to see if the total fuel quantity was dropping significantly. This might have raised their curiosity to expand their inquiry and discover that the imbalance was caused by unintended fuel transfer instead of a fuel leak. Fuel imbalance procedures direct a series of steps to cure the imbalance.
In both of these events, crews followed mindsets learned from simulator training. As an industry, we should take steps to encourage pilots to expand their mindset beyond the lessons learned in the simulator. Simulators teach procedures, but they need to also teach abnormal event processing strategies. Especially with rare events, we want crews to take enough time to examine the range of possibilities to not only select the first cause they think of. They need to take the extra step of excluding other possible causes. Had the fuel imbalance crew expanded their mindset to weigh fuel leak versus imbalance, they would have taken a closer look at their total fuel burn. Since they landed short of destination, we can assume that they had adequate fuel to take an extra minute to conduct this analysis.
Of course, we are analyzing these events from our perfect hindsight perspective. The bottom line is that both crews landed their aircraft safely following stressful and rarely encountered conditions. I submit these events for our collective analysis to better understand the latent problems promoted by simulator versus real-world mindsets. In my book, Master Airline Pilot, I share a range of strategies for countering these biases. I encourage you to post your comments and suggestions to this forum thread.
