Sunday, April 20, 2014

Ode to JungleBus

Well, Basic Indoc class is over, and I'm officially a [nameless major airline] pilot, wings and all. It was a really good experience, extremely welcoming regardless of flow status and quite a bit more chilled out than previous indocs. But I still have to get through Mad Dog school, which starts in a week, with computerized home study until then. It will not be chilled out, it's notoriously fast and furious, so I'll be hitting the books hard this week. My life at NewCo already seems impossibly long ago, and the "brain dump" on JungleBus systems and procedures is already well underway. You have to forget your previous airplanes as an airline pilot, or that knowledge will interfere with the one you're flying now. But I can't let all things JungleBus slip from my brain without some acknowledgement. This is the airplane I've flown far more than any others in my career, a plane I came to know rather well. I left NewCo with 4801.6 hours in the JungleBus, which is slightly over half of my total time. In all that time I never had a serious systems-related emergency. I had perhaps a dozen maintenance cancellations in over six years. JB seldom let me down; she deserves a tribute before I become intimate with another mistress.

The first time I saw JungleBus, I was a Q400 FO at Horizon, who had recently lost their Frontier JetExpress flying only three years into a supposed 12-year contract. As our CRJ-700s came back from Frontier they were being put on inefficient routes like Portland-Seattle, and so the factors that would eventually lead me to leave Horizon were already evident. In Spokane, I saw a Republic JB in Frontier colors taxiing into the gate, and my initial reaction was profound dislike. It looked like a child's pudgy plastic toy airplane to me, and the fuzzy baby animal on the tail added to the impression. My critique of the airplane's aesthetics never entirely changed even after I was flying it. But that first impression was likely tinged by jealousy and frustration over my stalled career at Horizon. Six months later I was in class at NewCo, learning to fly that pudgy toy plane.

The training began with a home self-study CD-ROM, which seemed awfully slim on in-depth systems knowledge. I figured the classroom systems lectures would go into greater detail, but when I got to Montreal the ground instructors seemed to know even less about the plane than we did. Alarmingly, the same was true of our simulator instructors. The checkride examiners asked plenty about each control and button's function and system logic, but very little about the underlying systems themselves. I entered OE feeling like I knew less about this airplane than any I'd ever flown before - and despite that, got through in only 25 hours feeling pretty comfortable with line operations. And that wasn't me being superpilot; everyone was getting through with 25 hours of OE. I realized then that the lack of systems instruction wasn't a byproduct of a new airline or a relatively new type, it was an integral feature of a design meant to remove pilots and their pesky human failings from its operation to the greatest degree possible.

The JungleBus is in fact a fairly complex airplane, but from the flight deck it is simplicity itself. The overhead panel features a few rows of large, rarely-touched "dusty buttons" which select the systems to "Auto" or "Off." When everything is selected to its normal (auto) position, the switchlights are dark. The ram's horn yokes are beautifully ergonomic and nicely balanced, which they well should be since they're not connected to much of anything - it's all computer-augmented fly by wire with the exception of the ailerons, which actually have control cables running to their hydraulic actuators. The FADEC-controlling thrust levers glide smoothly from idle to the rating detent, which you seldom feel since the autothrottles are generally engaged before you ever get close to it. The secondary controls - Flaps, Spoilers - are big and unmistakable, and protect themselves by refusing to deploy when they shouldn't. The real heart of the cockpit, the Flight Guidance Panel, has quite a few easy to understand and use modes for lateral and vertical guidance - the most commonly used of which are the FMS-derived LNAV and VNAV. When everything is working, it's a rather easy airplane to fly. 

But what about when things break, you ask - what then? Again, the airplane makes it as simple for the pilot as possible. Most major systems go through extensive self-tests, transparent to the pilot, on aircraft power-up. If they fail their self-test, they display an error message and you call maintenance. The only systems the pilot tests himself are the fire protection, annunciator lights, and trim switches. After startup and proceeding through the flight, the EICAS system inhibits many faults, discreetly sending minor ones directly to the airline and displaying others after landing or shutdown. Only outright failures are shown on the EICAS display in real-time or close to it (most are inhibited between 80 knots and V1), and then the system prioritizes the order of their display so that malfunctions most likely to be the root cause of secondary faults are addressed first. You just pick the highest priority message and run the associated QRH checklist. Nine times out of ten, the checklist tells you to select the system off, wait sixty seconds, and turn it back to auto, and nine times out of ten this resolves the fault – the aeronautical equivalent of Ctrl-Alt-Del. If the QRH leaves any room for confusion, you can bring up the appropriate system schematic on the Multi-Function Display (MFD), and the status and relationship of every component in the system is displayed. If our instructors never placed much emphasis on knowing exactly what is powered by each hydraulic system, for example, their laxity makes a little more sense when you realize that the hydraulic synoptic page shows exactly what is being powered at any given time, no rote memorization required. My colleagues and I have often theorized that this airplane was designed to be flown in challenging third-world conditions by inexperienced, occasionally fatigued crews, and most of the time it will keep even this “lowest common denominator” crew out of serious trouble.

Of course, JungleBus’ front-end simplicity is achieved through a great deal of back-end complexity of the electronic variety, and this means it is still prone to human error – but these errors can lay hidden within layers of code or imbedded on circuit boards for months or even years before surfacing in unexpected ways. In six years I’ve seen this airplane do some truly mind-boggling things, like spontaneously change a pilot-entered VNAV descent angle from a reasonable 3.5 degrees to an untenable 6 degrees, or abruptly turning 20 degrees off course upon crossing an intermediate fix where the airway didn’t bend. A good friend had a baffling situation where the pneumatic synoptic page showed that both pack valves had closed themselves and the cabin altitude was rapidly increasing. He quite reasonably made an emergency descent and subsequently diverted to our maintenance base in Louisville, where subsequent investigation showed that it was an indication error – both packs were working correctly and cabin altitude had actually remained steady at 8000 feet! I personally experienced a current surge event in one of the Secondary Power Distribution Assemblies (SPDA) that fried multiple circuit boards in one fell swoop, creating a puzzling array of seemingly unrelated faults; this fortunately happened on the ground, during the Power-On Self Test.

These type of events are not a slight against the JungleBus’ basic design; they happen in every modern airplane, and I dare say more often in many than in the JungleBus today (early on it had many more bugs to work through). I would say they are notable in the JungleBus primarily because most of the time the airplane does work so well that it lulls crew into less active monitoring than they might practice in a less reliable design. This complacency is the easiest way to get into trouble in a JungleBus, and usually not because of some insidious bug hidden deep in the software. Most of the time JungleBus crews screw up, they do it to themselves with bad input, and subsequently fail to catch the error through lax monitoring. Too often when the error is finally discovered, they attempt to fix the automation in a hurried, panicked manner rather than simply flying the airplane and having the other pilot fix the automation when time allows it to be done methodically. And this problem, I would argue, is a direct result of JungleBus’ design philosophy and the resulting way we have trained her crews. If you take so much of an aircraft’s operation out of a pilot’s hands, encourage him to continually use full automation, and keep him in the dark as to much of what his aircraft is doing, you cannot reasonably expect him to remain alert and engaged in the aircraft’s operation over countless uneventful hours, and you cannot expect him to become instantly attuned when something does go amiss. It goes against human nature. The FAA and the airlines have become quite aware of this and the pendulum is finally swinging the other way across the airlines and across fleets, toward a greater emphasis on manual flight skills and automation shedding when appropriate.

Fortunately, many of us who’ve been flying JungleBus for a while recognized her potential for creating automation dependency early on, and took our own steps to ward it off. We hand flew whenever appropriate, and occasionally turned off the autothrottles and flight director too. We backed up the VNAV with good old 3-to-1 ratio mental math (which saved my bacon on that spontaneous path angle change incident). We talked amongst ourselves and learned some of the more insidious “bad input, bad output” scenarios, like accidentally hitting PREV twice and setting ourselves up for an offside-localizer capture and subsequent reversion to ROLL HOLD mode. We talked to our mechanics and learned the nuts and bolts behind the Auto mode buttons. We opened the QRH in cruise with synoptic pages displayed on the MFD, unraveling the reasoning behind the procedures. We followed our progress across the country and played the “if everything goes blank now, where do I go?” game.

Fortunately, none of us ever had those “everything goes blank” scenarios actually happen. Everything generally worked very well. When JungleBus did throw us a curve, we usually caught it early on. When I made dumb inputs, my FOs almost always caught them before agreeing to “Activate.” Once in a while the autopilot and/or autothrottles would do something funky and I’d turn them off, and I’d appreciate all over what a wonderfully honest, smooth-flying airplane it is when you just fly the damn thing – a big Cherokee, really. JungleBus never did anything to put me or my ticket in danger, as long as I just paid attention. I finished my stint at the regionals with a clean record and about 7000 accident-free, violation-free, and mostly uneventful airline hours. I have useful training & prior experience, great coworkers, and two very different but ultimately very good airplanes to thank for that. Goodbye, JungleBus. I’ll miss you.