Simulating complex flying scenarios: elevator failure

The crippled Cessna 172 about to touch down on Runway 35 at Boscombe Down

The crippled Cessna 172 about to touch down on Runway 35 at Boscombe Down - Credit: [None]

Eugenio Facci puts his skills to the test in this new flying simulator series. This month, he mans a Cessna 172SP with elevator failure!

A VOR/DME fix from Southampton put me on a 10 mile final to R35 at Boscombe Down

A VOR/DME fix from Southampton put me on a 10 mile final to R35 at Boscombe Down - Credit: [None]

In this new series the goal is to envision fairly complex scenarios (the kind that would be difficult to reproduce in real flight), fly them on a PC simulator, and learn some valuable lessons. The format will be that of a normal briefing, the virtual-world flight and a de-brief on the lessons learned. In each scenario the Assistant Editor will be putting his skills to the test, mentored by regular columnist and flight instructor Charlie Huke. Enjoy the flight (and get ready for some bumps).

BRIEFING

Aircraft: Cessna 172SP

Flight: Denham – Turweston

Wind: 310/12

Visibility: 6,000m

Most Read

Clouds: Broken at 2,500ft

Scenario: Elevator failure

Shortly after becoming visual with Runway 35 at Boscombe Down

Shortly after becoming visual with Runway 35 at Boscombe Down - Credit: [None]

It looks like that line of clouds is right on my path. Shall I route around it? Probably not a good idea?I have no GPS, the visibility today is not great, and there’s a fair amount of controlled airspace around me that I don’t want to infringe. I’d better fly below it: power back, pitch down, and my Cessna 172 settles in a decisive 800fpm rate of descent to stay well clear of clouds.

I’m on a leisurely cross-country flight, from Denham to Turweston, and it’s a decent flying day (aside from a fair amount of haze and clouds). But I’m in for one of those surprises that no pilot would ever want to face. When I try to stop my descent and level off at 2,300ft, nothing happens. Well, almost nothing?the nose rises a bit when I apply power, but basically keeps pointing, stubbornly, towards the ground.

A thousand thoughts race through my head: am I doing something really silly? Did I somehow leave the control lock in? By the time I realise what’s happening my field of vision is mostly filled by green fields.

I’m glad that I’m actually in my living room and not a real aircraft. This is the kind of situation that would cause serious sweating in real life, along with rushed theological considerations about the meaning of life and death. But not today: in the safe confines of my house I can explore the complexity of this scenario, and test the strategies that I have discussed with Charlie Huke, my mentor for this sim training series.

But keep in mind that the simulation is a realistic one, and X-Plane is not forgiving. It gave me this emergency at a random time (perhaps the worst possible time, it turns out) and there’s no guarantee that I will come out of this ‘alive’.

Same position, view from the cockpit

Same position, view from the cockpit - Credit: [None]

Almost deadly phugoids

In fact, for the first few minutes after my discovery of a broken elevator cable, a crash seems inevitable. First, my Cessna is descending rapidly, and by the time I take action the ground is near. To avoid an impact, I need to apply quite a lot of power and nose-up trim, which of course, sets me up for a second problem: the aircraft is now climbing steeply but decelerating rapidly! To avoid a stall (from which I would probably not recover) I need to intervene once again, quite decisively.

Here I learn?very quickly?my first important lesson: while in a conventional stall recovery you apply power, in a 172 with a failed elevator the application of power actually brings you closer to the stall. In other words, I discover that too much power pitches the nose up alarmingly, such that the aircraft decelerates toward the stall. Thus, as I cannot manage this scenario on the trimmer alone, I need to back off on the power until I am able to reduce the angle of attack with the trim.

So now, I’ve avoided a stall and learned one lesson, but I’m back to square one. To avoid the stall I pulled off all the power and trimmed fully nose down, and I’m again descending rapidly with green fields (and cows) almost entirely filling my windshield. Time for a second, fast piece of learning: I am the problem. I’m increasing the phugoid motions of the aircraft with excessive control inputs, which I am applying at the wrong time (pilot-induced oscillations, in other words).

The whole scenario is made more complex by the fact that I’m trying to control pitch by using two different inputs simultaneously, and I can’t really get a feel for either one that way (I am not learning). So, there’s another quick lesson here: unless I need to avoid an impending crash or stall, I decide to use only one control input to pitch the aircraft (trim or power), otherwise I’ll be stuck in a long and confusing series of phugoids.

In this particular case I find that it’s a bit easier to use power rather than trim, as it seems to cause smoother reactions from the aircraft (although I suspect this may vary considerably across different types). So, I start to dampen the phugoids with gradual applications of power. When the aeroplane begins to descend, I gradually and gently increase power. When the ROD starts to decrease (but I’m still descending), I progressively pull back the power lever in small increments. It turns out that this strategy works. The phugoids become smaller and smaller, until eventually I find myself in a gentle climb (which I gladly accept?I’m not particularly enjoying proximity to the ground right now).

Here I learn something else: the phugoids never completely go away (there is a fair amount of turbulence today, which again is simulated quite realistically, and continues to perturb my aeroplane), and I need to read my aircraft’s parameters by averaging its extremes. For instance, right now I realize that I’m in a 200fpm climb, but the VSI does not actually indicate that. It follows the oscillations of my aeroplane, constantly swinging from a high of plus 500fpm to a low of minus 100fpm. It seems that, in an aircraft with a failed elevator, that’s the best indication one will get, and this way of ‘reading’ the flight data will become useful when planning an appropriate ROD for the final approach.

Same moment, captured from the cockpit

Same moment, captured from the cockpit - Credit: [None]

DODAR plan

So, I avoided crashing, and now it’s time to get the aircraft back to the ground (in one piece). For situations like these I use a DODAR, one of the many structured emergency workflows that can help you work systematically through a problem. The process is: Diagnose the emergency, assess Options available, Decide what to do, Assign tasks (being single pilot this will include asking ATC for specific help), and progressively Review how my plan is unfolding.

Although the diagnosis is quite clear, I also ponder whether it’s not something else causing the problem. In addition, I evaluate what other systems in the aircraft could be impacted by this failure. Finally, with the help of a personal Quick Reference Handbook (QRH) that I use for abnormal scenarios, I assess the options available.

It’s clear that I will need a long runway, possibly without disrupting the flow of traffic over southern England (for instance, Heathrow is near but diverting there would be a pretty selfish idea). My list of long, non-busy runways in this general area includes Boscombe Down, Brize Norton, Yeovilton, and Manston (the latter not the best choice since it’s now a lorry parking facility). Because I have limited control of the aircraft, I need to choose a destination that would minimise turbulence (which today is light to moderate everywhere in southern England) and crosswind. So, with a 12kt gusty wind on the ground from 310 degrees, I decide to head for Boscombe Down’s Runway 35, about fifty miles south south-west from me.

In a real emergency I would offload navigating to ATC, letting them guide me to my destination. But that option is not available in this sim, and so now good navigation on my part becomes, suddenly, very important. First of all, I will be going to an airfield that I have never visited, and the visibility today is only six kilometres. I already have one pretty big problem, so ideally I don’t want to deal with an ‘uncertain position’ scenario as well. In addition, I will not have much control of my vertical profile, so it’s important that I show up at the start of my approach at the correct height and distance from the runway, in order to avoid large corrections on my way down. To make things a bit more spicy, I’m also simulating an inoperative GPS, and to navigate I’ll need to rely on dead reckoning, a VOR and a DME.

Of course, just like waves, problems never come alone. The extended struggle that took place when the elevator first failed means that for several minutes I did not pay attention to my navigation, and so I can’t pinpoint my position exactly. I know, however, that when the problem emerged I was about to fly over the disused Westcott airfield, that I was on a track of 310 degrees, that the struggle with controlling the aircraft lasted for about five minutes, and that my track has not changed much throughout. Based on that I should be roughly eight miles south of Turweston, an estimate confirmed by a VOR/DME fix on Daventry.

What I’ll do now, then, is navigate towards Boscombe Down, while I set up the aircraft for the approach (and in the real world I would get ATC to clear the airspace in my path as much as possible). Here my goal is to get the aircraft down to approach speed and see what parameters (primarily RPM) give me straight and level flight. This will be the configuration that I will use when I’ll show up at the beginning of my long final, before starting my descent to land.

Patient preparation

The process of slowing down the aircraft is lengthy. I need to reduce power and trim nose-up in very small increments: if I try to do things faster by applying larger power changes, then the aircraft starts to oscillate again in prolonged phugoids, which makes the scenario quite confusing.

There is also a decision to be made: use flaps or not? On the phone my mentor, Charlie told me not to use them due to the pitch effects, and I’m heading for a long runway so they will not be needed. But I want to make full use of this learning exercise, and one day in real life I may need to land in this condition on a short-ish runway. So, decision made: I’ll experiment with flaps to see what effect they have on my flight path.

I anticipate that lowering the flaps will cause a pretty big upset, and so I want to do it when the speed is well inside the white arc (upper limit 85kt in this case) but also well above stall speed (roughly 47kt). Therefore, when the aircraft reaches 76kt, I put down ten degrees of flap. As I anticipated, this action really upsets my Cessna, giving it a large pitch-up moment that brings the speed down to a scary 50kt. After seeing this reaction, then, the use of flaps becomes a no-brainer: I will not touch them again?leaving them where they are, I will landing with ten degrees of flap, which should allow me to slow down my touchdown speed in any case.

In fact, my impression is that the use of flaps in such an emergency should be generally avoided. Depending on what aircraft you are flying there may be not enough width in the white arc to avoid a stall (or losing a flap) when lowering the flaps, so in the vast majority of instance landing flapless will be the best option.

In any case, after prolonged phugoids caused by lowering them by ten degrees, the aircraft eventually settles on straight and level flight at 67kt. Normally in the 172 I would land with flap thirty and 65kt on final. In this case, since I will have flap ten and I want to keep a healthy margin on stall speed, I decide that I will increase my speed on final to 72kt. Again I apply small power increments to get the aircraft to accelerate, until eventually I’m stable at my target speed of 72kt for straight and level. Then, I write down my pitch and power datums: compass cut in half by the horizon, RPM 2,050. This is how I will be flying as I start my final approach. From there, I will power back, to a setting that I’m yet to discover, in order to fly a three-degree descent profile.

Three degrees, simple arithmetic

A three degree profile is not a random choice: it is very useful in this case for two reasons. First, it’s gentle enough so that towards the end I can attenuate the contact with the runway with small power increments. Second, it’s the standard IFR profile, and the maths are simple: roughly 300ft of descent per nautical mile, and a ROD that is ‘half’ the ground speed. In my case the headwind on approach will be about 20kt (it’s 12kt on the ground, but more higher up). Therefore, my ground speed will be roughly 50kt, and my ROD 250fpm (fifty divided in half, times ten).

Here again, then, I need to find out what parameters give me the target ROD of 250fpm. From 72kt straight and level, therefore, I slowly power back in small decrements. Eventually the aircraft settles (still oscillating a bit) on minus 250fpm. Again, I write down my pitch and power datums: compass touching the horizon (with a tantalizing dance obviously), and RPM 1800. By now I have reached and passed Boscombe Down. I have all the parameters that I wanted, and it’s time to position myself for the final approach.

I want a ten-mile final, which means that I will need to start the final at 3,400ft on the QNH (Boscombe Down is at about 400ft amsl). A bit of VOR navigation comes in handy here. The 278 radial from Southampton VOR, at 17 miles DME, puts me roughly at the beginning of my final. From there, I will turn for a track of 350 degrees, starting the timer and initiating my descent, using the parameters that I have written down (RPM 1,800 for a ROD of 250fpm being the main one). In real life I could get ATC to guide me down and along my approach path, but in this sim I can’t. So, since during the first half of the approach I won’t be able to see the runway, I had previously jotted down a descent profile that will help me keep track of whether I’m high or low. For instance, at time zero, I will be at 3,400ft. Then, with a ROD of 250fpm, I should be at 3,150ft after one minute, at 2,900ft after two minutes, and so forth.

The patient preparation to write down the correct flying parameters comes at something of a cost: with less than 45 litres remaining, the fuel level is lower than I’d like it to be when I start my long final. This also means that I don’t have time to fully practice a ‘flare’ and a go-around while approaching Boscombe, like I would have wanted to. But the reward of the long preparation is that the descent goes smoothly, and eventually I get visual with the runway from a position that does not force me to try large corrections (which I apply, of course, using only power). As it turns out?and this is often the case in aviation?time spent preparing was time well spent.

The next (and last) thing I have to do now is to smoothen out the descent into a pseudo-flare. Again here I decide to work slowly, and start to apply very small increments to my power setting when passing 600ft (200ft agl). The increments are minimal: I don’t look at the RPM, I just hear them through a barely-noticeable change of noise. This of course means that my touch down point gets pushed further down the runway – at about 2,000m, Runway 35 at Boscombe Down starts to look long but not endless at some point during my attempted flare. But the strategy works out: eventually the Cessna touches down in a semi-dignified manner, offset from the centreline but almost flat and with a gentle ROD of roughly 100fpm.

I’m ‘alive’, the aircraft is in one piece, and I really feel I have learned a lot.

Simulator details

The flight was flown using X-plane 10, a software that is almost identical to the one approved by the FAA for formal flight training. X-plane uses ‘blade element theory’, a process which the company says involves ‘breaking the aircraft down into many small elements and then finding the forces on each little element many times per second’. This allows for a realistic simulation of scenarios and effects at the edge of the flight envelope, including turbulence, stalls, spins, spiralling slipstreams, the P-factor, translational lift and, in this case, the effect of a failed elevator. Hardware used for this simulation included control column, pedals, and throttle.

My top 5 lessons

1) Use only ONE control input (either power, or trim). You may still need to use both, though, to avoid an impending crash or stall

2) In most types of aircraft to avoid an impending stall you will probably need to trim down and decrease (not increase) power

3) Be aware that you may end up overcontrolling the aircraft without noticing it

4) Make your life simple. Write down the power/pitch datums to fly (at approach speed) only two ‘modes of flying’: straight and level, and a three degree descent. Then fly only those two configurations

5) Be aware that the adrenaline will make your rush, but that time spent planning the approach will be rarely wasted

Instructor’s debrief - by Charlie Huke

The exercise was useful to learn about a few fundamental aspects of this type of emergency, like the need to plan the approach accurately and the difficulty in controlling the aircraft using two inputs simultaneously (power and trim – you should use only one).

Besides that, there are three additional points of learning that our readers might find helpful. First, I would make a different choice for my destination airfield. You excluded from your selection large commercial airports, but in this case I would have gone to Birmingham, which was straight ahead, into wind and as long as anywhere. The main point here is that, with a failed elevator, I would question the aileron integrity (they too are connected to the yoke) and I would wish to avoid long transits and multiple turns. So, don’t worry about disrupting commercial flights, diverting them won’t kill anyone, yet you’re in a genuinely life threatening situation.

Secondly, it’s also helpful to ponder how this emergency would have been different in real life. Here you had the luxury of experimenting with the flaps, but in a real emergency I would absolutely avoid their use: the risk of causing an uncontrollable upset is too great. You’re heading for a vast runway so speed isn’t an issue. You will also find that flaps will lessen the nose-up pitch with power and will decrease trim effectiveness, making the touchdown flare more difficult.

Third and finally, in real life, I would rely extensively on ATC. This situation absolutely is a Mayday on 121.5, and you could give them the job of navigating and clearing the airspace around you, preparing the destination airport for your arrival. And, when you are ready for your final approach, they could also guide you down and along your path, greatly reducing your workload. – Charlie Huke