Dafydd Llewellyn

Do you normally fly without looking at your speed? If so, I suggest you go and do some gliding; because in a glider it becomes very obvious that your only control of speed is via the pitch attitude (that's really true in any aeroplane; power just gives you a greater choice of pitch attitudes). You'll soon learn to maintain speed by pitch attitude; it becomes automatic. If you have trouble watching the skid ball, try putting a yaw string on the windscreen. Next, the speed you need to fly at whilst you're turning is 1.5 times the flaps-up stall speed, NOT the speed for best glide. AFTER you have finished turning, you can reduce the speed to the best glide speed. Thirdly, the trim setting will vary according to the centre of gravity. There's no requirement that full back trim should not stall the aircraft; it just happens that way at the forward CG limit. So you should trim for 1.5 Vs1, AND keep an eye on the ASI. All this is just normal flying; the whole of the turn-back manoeuvre is simply a pre-planned piece of normal flying. Keep your adrenalin under control.

If you want to be precise about it, the load factor is 2 in a balanced (i.e. no slip or skid) 60 degree turn, whether it's climbing, level, or descending. However, I agree that one can turn the aircraft via a variety of other manoeuvres - even a stall turn if you have sufficient speed in hand - that involve a lower load factor whilst you are actually turning; but they all involve a recovery that DOES require an increased load factor; in a sense, they simply postpone the increased load factor. They're all aerobatic manoeuvres, however, and can easily exceed the flight envelope of a recreational aircraft. Since it turns out that the optimum bank angle is less than 60 degrees, the turn-back manoeuvre I described is well within the flight envelope, and also within the normal pilot capability of an RAA pilot. The cable-break situation in gliding is relevant to this discussion, I believe, because it does NOT require any extreme flying whatsoever; once you have the nose down and the speed back to what it should be, the rest of it is ordinary flying requiring no special skills. However, if you make the mistake of racking the glider into a turn as soon as the nose is down but before the speed is back up, all seems to go well until you come to try to roll out of the turn - at which point you discover that the glider is actually in a fully developed spin; the result is almost inevitably fatal. So hasty or fancy flying is the LAST thing anybody should be teaching. The KISS principle applies, with a vengeance.

Either way would do it. However, if people wanted a simple gadget to make the pilot conscious of the turn-back height-loss issue, then one could put two marks on the perspex disc, spaced apart by a suitable margin to cover a typical turn-back manoeuvre plus, say, 100 feet - and just turn the disc to align the lower mark with the altimeter large needle as part of the pre-take off check. One could add "not above" and the density altitude at which the zero-flap climb gradient equalled the best glide gradient. The chinagraph pencil way does have the advantage of requiring the pilot to think about it, I suppose. This thread asked for constructive suggestions as to how to improve the turn-back accident rate; making the pilot conscious of the physics of the situation and giving him a simple go/noGo criterion would seem to me more constructive than arguing about the definition of a steep turn. There seems to be an underlying misapprehension that the steeper the turn one can make in a turn-back manoeuvre, the better. This is incorrect; there's a trade-off between how long it takes to fly the turn, and the rate of height loss due to the increased induced drag. RAA aircraft are certificated for "normal category" flight manoeuvres, which include turns up to 60 degrees angle of bank. In a 60 degree turn, the load factor is 2.0, so the speed needs to be not less than 1.5 times the stall speed. Now, the minimum drag speed (at which the parasite and induced components of drag are equal) is usually less than 1.5 times stall speed, so the induced drag may contribute perhaps 40% of the total drag, in level flight. However, the induced drag increases in proportion to the square of the load factor, so in this case, the rate of sink would increase by (0.6 + 0.4n^2), where n = load factor, compared to level flight. At 60 degrees bank, that gives 2.2 times the level-flight rate of sink. Therefore, there is an optimum angle of bank in order to turn back with the minimum loss of height. It works out at 44 degrees for the case I analysed, but as it depends solely on the bank angle and how much above the speed for minimum drag one has to fly, that's not going to vary much from one aircraft type to another, and so the optimum bank angle isn't going to vary much from one aircraft to another. This is hardly rocket science; all you need to know is what the minimum drag speed is in a level glide, and even the average flight instructor should then be able to work out what the optimum bank angle is for that aircraft.

Have you not met the glycerine trick? If you need to put some sort of adjustable index onto an instrument that does not have a turnable bezel ring, a disc of thin Perspex will stay on the face for years if attached by a bit of glycerine; the glycerine fills the space between the two and gives perfect vision through the combination. Used to be a common trick for putting a speed-to-fly ring onto a variometer. You can turn it with a finger, and it will stay where you leave it.

Well, I thought I was writing in English, but judging by the responses on this thread, maybe not. The aircraft I'll be flying is a motorglider of fairly modest performance. My preparation for an EFATO will be, to have determined the height loss in the turn-back manoeuvre that I have described, on a hot day at altitude. I'll put a disc of thin perspex on the face of the altimeter, held on with a dab of glycerine, with a reference marker on it. Part of the pre-takeoff check will be to set the reference marker to a height equal to the runway altitude, plus the height loss in the EFATO turn, plus the estimated height of any trees , power lines etc that would have to be cleared when gliding back in, plus 50 feet. If the altimeter does not reach the reference mark as I climb out over the obstacles, I will know that I cannot make a turn-back until after the crosswind turn. The rest of you can do as you damn well please, but that's my way of tackling the issue.

What you describe is what I was taught, too - but generally from about 100 to 150 feet (and with risk of scaring somebody's chooks). What I am describing is quite different; for a start, you DON'T do it at low level - you could do it from 500 feet, in a designated low flying area, and break off at 100 feet, saying, "There, you see it takes over 400 feet" - or whatever. It's a demonstration of how much height is actually required, i.e. more than the student expected; and that is NOT in the syllabus, to my understanding. This is not something to bring in at early solo stage; but I suggest pilot training isn't complete without something of this sort. For your interest, here's what my spreadsheet showed on height loss Vs bank angle. The bottom curve is at sea level; the top one at 8000 ft density altitude. Note that the turn is not the dominating factor in the height loss; the loss on the initial roll-in and the turn reversal is about equal to the loss in the actual turn itself. This is for a glider having about 23:1 glide ratio at 60 KCAS; but the relativities will be similar for most aircraft.

I was looking at what might form a reasonable basis for a criterion that an experienced pilot might use. Obviously, an inexperienced pilot won't be able to fly sufficiently accurately to achieve either an "optimum turn" or a "zero-G manoeuvre" . I am aware of the concept of getting the aeroplane pointed the opposite way whilst at reduced G so it doesn't stall in the process (but you better have the necessary speed when you try to recover, or it will go into a fully-developed spin faster than you can blink). What I'm seeing is that it won't work unless the strip is long enough that you are well above your "turn-back manoeuvre" height loss figure - however you choose to try that - by the time you reach the upwind end of the runway. Short wings and windmilling propellers and draggy undercarriages all reinforce that conclusion. An instructor could certainly close the throttle and demonstrate the height loss from a balanced 45 degree banked turn of the sort I described, at a safe height, and thus make the point that the terror will most definitely become firmer unless the aircraft is comfortably above that height, as part of the normal forced-landing training. Whether or not that's psychologically a sound notion, I have no idea - but if I were shown that trying to turn back will definitely fail if you have less than X feet in hand as you cross the trees on takeoff, that would certainly deter me from trying it - and X is likely to be at least 250 feet for a motor-glider, and likely around 450 feet for a reasonably clean recreational aircraft, so all in all, it says that it's not something to try on the upwind leg of a normal circuit unless you're operating out of an airfield large enough that you could just about turn crosswind before you reach the boundary fence. Is that worth adding to the training syllabus, or not?

Did a bit of analysis of a possible turn-back manoeuvre; and what it shows is perhaps a little different to what one might expect: Firstly, if the pilot takes two seconds to decide to turn back, and three seconds to roll-in to the optimum bank angle (which came out to be 44 degrees, for the aircraft I was studying), and the aircraft is travelling at around 70 kts, it travels some 500 feet straight ahead before it develops the full rate of turn. The turn radius came out at around 180 feet; and allowing five seconds to reverse the turn, one needs to commence the reversal after about 210 degrees of turn. The reversal carries the aircraft back towards the runway centreline, and only 30 degrees of turn at perhaps 20 degrees of bank is needed to straighten it out. I've allowed a bit of benefit from drift from the crosswind component. The overall shape of the manoeuvre is shown in the attached document. What the manoeuvre achieves, pretty much, is to bring the aircraft back onto a reciprocal heading to the runway, at about the place where the engine failure occurred, but several hundred feet lower (about 250 feet lower, for the motor-glider I was studying). It does NOT bring it out any closer to the threshold than the point of engine failure. This says two things, to me: Firstly, it confirms what I previously said - you need to clear any obstacles at the threshold by at least the height loss in the turn-back manoeuvre plus 50 feet; and your glide gradient in a straight glide needs to be no steeper than the climb gradient before the engine started to fail. Secondly, this sort of manoeuvre will only work if you are departing from a runway that is sufficiently long that you will have this height in hand at the upwind end of the runway - which probably means about a 1000 metre strip, for a typical lightie. On a short strip, forget it; it also suggests that climbing out to 500 feet on the runway heading is a pretty dumb thing to do, if you're operating on a short bush strip where that is not mandatory; you'd likely be better off starting a gentle turn into any crosswind, from about the point at which landing on the remaining runway is impossible, assuming you're airborne and clear of obstacles. turnback.docx turnback.docx turnback.docx

This is a good point, and it brings up something I've been thinking about for a while. One of the requirements for certification of a factory-built aircraft, is the supply of performance data for its flight manual. Now, I know only too well, that about 95% of pilots never look at the performance data in the FM, but it's a resource you might be able to use: By way of illustration, the data required usually includes climb gradient versus weight, pressure altitude and temperature. Assuming you are taking off into wind, the gradient made good over the ground will always exceed the "still air" gradient. Climb gradients are normally expressed as "% gradient" - so a 10% gradient means the aircraft will gain 10 metres in altitude for every 100 metres it travels horizontally. You will also normally find the best glide ratio. A glide ratio of 12:1 means the aircraft descends 1 metre for every 12 metres it travels horizontally, which translates to a gradient of 1/12 = 0.0833 or 8.33%. So long as the climb gradient exceeds the glide gradient, you will be going up in the upwind leg climb, faster than you would descend in a straight downwind glide. The climb gradient reduces with increasing weight, and increasing density altitude, but the glide gradient does not change. Now, go out and measure the height you lose in the "turn back" manoeuvre. (If you assume some drift on the upwind leg, that would comprise the height lost in stabilising the airspeed at 1.5 times the stall speed - remember, you would likely have been climbing at 1.3 times the stall speed, plus the height lost in turning about 210 degrees, followed by a reversed turn of about 30 degrees.) I'm NOT talking about any form of extreme manoeuvre that would put you in danger of spinning; the turn-back manoeuvre should be within the range of normal flying practices, i.e. not more than 60 degrees of bank; and to do that sort of turn, your airspeed MUST be at LEAST 1.5 times the stall speed. Let's assume, for the sake of this discussion, that turns out to be about 250 feet. Add 50% to that, i.e. 375 feet. So, provided your climb gradient exceeds the glide gradient, and you clear any obstructions at the upwind end of the strip at not less than 375 feet, a turn back will be possible. Unless those two criteria are met, a turn back will not be possible on the upwind leg of the circuit. At some density altitude, the climb gradient will be less than the glide gradient, and a turn back on upwind leg should not be attempted. If you know the length of the strip, you should be able to work out how high you would be at the upwind end of it, from the takeoff data and the climb data. Using these principles, it's not at all difficult to work out a safe turn-back criterion. I'd do it for standard sea-level conditions, and for a 30 degree day at, say, 2500 feet, and work out a rule of thumb for extrapolating. Not too difficult to set up on a spreadsheet. There will be a point on the airstrip beyond which it is impossible to stop the aircraft without overrunning the strip end. From that point, to your manoeuvre height - 375 feet in the above example - is a "non-manoeuvring area" - if it quits in that area, pick two trees and aim between them. So, what is so difficult about it?

I can only speak of my own experience; I've had aileron flutter, a disconnected elevator, a forced landing, and a desperately marginal situation during an attempted landing for an aerotow retrieve, plus a few others as well as the usual few in driving. None of them triggered either a "freeze" or dirty underpants; so I can only assume that the training I was fortunate enough to receive, and my subsequent experience, led me to keep control of the situation. I've had a student who screamed and grabbed the canopy release & canopy jettison, when the aircraft encountered a mild bump; and one who froze on the wrong rudder pedal in a spin. I know the difference, so please do not tell me it's all the same. I am not denying that looking at the problem through fresh eyes may be valuable; however, if you start with the precept that one should NEVER turn back, and ask the Psych why people do, you will only get half the answer. The other half, is about how to teach people to make a rational choice. Winch cable break training emphasizes the necessity to keep control of the aircraft FIRST; and then assess the situation calmly; and only then act decisively.

In my personal experience, if you have rehearsed the situation either in your mind or physically, your WTF moment will be greatly reduced. I've had a few such moments, and for me, time slowed down - one seems to go into a "superconductive" mode, or something of the sort, for a few seconds, and things happen in slow motion. I also had this described to me, by a pilot who had to bail out of a glider in a hurry. I suspect this is a consequence of experience and training. I have no idea whether this happens to everybody. I've also observed that students whose instructors have not exposed them to some unexpected events, often "freeze" when they are overloaded. I can't, offhand, think of anything much worse than "freezing" when you're climbing at about 45 degrees nose-up, at around 450 ft AGL, and the winch cable breaks; you HAVE to get the nose down rapidly. So understanding the physics of the situation and reacting fast and appropriately are both training requirements. Recovery from unusual attitudes is potentially useful for this, I would expect. This being the case, I wonder at what will come out of a psychological study, unless the psychologists concerned are themselves competent pilots. Please excuse my skepticicsm.

Go to a gliding club that uses winch launch, and do the GFA required training for winch-launch cable break. Then relate that to the actual L/D ratio for the aircraft you fly. A typical two-seat training glider can make a "modified circuit" - usually a 135 degree turn, fly out to a close downwind leg spacing to allow a 135 degree turn to a "base" leg, followed by a final turn, from a cable break at about 450 feet AGL; the critical thing is to stabilise the speed BEFORE commencing the first turn, which means the decision height after stabilising the speed, is at about 400 feet. That normally allows the subsequent manoeuvring to be done with a comfortable margin to clear trees etc. If you have less than 400 feet, there will normally be sufficient strip to land straight ahead (remember, gliders have dive brakes). That's with a glider whose L/D ratio is around 26:1 ~ 30:1. GFA sets strip requirements so there is no "non-manoeuvring area"; and it's worth studying the basis on which this is calculated. If your aircraft can only manage, say 12:1, that height would need to be multiplied by 26/12 = 2.17, so the equivalent "decision height" would be around 870 feet. However, the length of a gliding strip used for winch-launching is such that the glider is only about 1/4 of the way along the strip when it reaches 450 feet; whereas a powered aircraft may be past the upwind end of the strip - so the shape of the necessary manoeuvre will be quite different. In effect, a single-engine light aeroplane has a substantial "non-manoeuvring area" . It's not all that difficult to work out what that area is likely to be, for your home strip. The result is likely to be sobering.

I find this whole subject bewildering; may I ask some questions? My personal experience is that there is no satisfactory substitute for a real understanding of WHY things work; when I started my flying, I did so by completing the full DCA PPL theory course before I got into an aeroplane as a student. Back then, the "short answer" style of theory course had not been invented; the exams required essay-type responses and the examiner was looking to see that the student really had an underlying comprehension of the principles involved. I saved all my course books; but when my wife tried to use them for her training, about five years later, it had all changed. Also, by then I had an aeronautical engineering degree; but when I tried to explain the basis behind the theory she was being expected to regurgitate, it emerged that the course was not at all interested in such basics; it taught glib, over-simplified "facts" and the trick was to understand what words the setter of the course expected to see, NOT what the underlying physics actually was. This seems to have become progressively more the case as each generation of pilots set revised "newer" training courses; each generation seems to further distil the "knowledge" into yet a higher level of "pilot fiction" which is even more remote from the fundamental facts - it's very much folk-lore being progressively evolved. The RAA training material of recent times - from what I have seen of it - has made a notable effort to reverse that trend; but there are limits to what it can achieve. People who look for a simplified, condensed "knowledge pill" that they swallow in order to be allowed to commit aviation, will never have that real underlying understanding. One cannot, in my experience, simply sit there and allow somebody to shovel knowledge in. You get what you work for; others can teach, but you have to do the learning. That takes real effort. And a lot of reading. And actually doing it. So, just exactly how are these "training providers" going to work?

You'll likely find the Airpath or whatever just as badly affected. The problem is the magnetic properties of the canopy bow, not the compass - it's just doing its job. What's the aircraft? I assume it has a 4130 steel canopy bow; excellent for strength, but 4130 is notorious for becoming magnetised. This problem is common in welded-steel tube fuselages. The logical answer, one might imagine, would be to install a remote flux-gate compass, but CAO 20.18 Appendix 1 requires a direct-reading compass as well. There are methods for de-magnetizing such a component - but unless you remove whatever the cause of its becoming magnetized in the first place, it will simply keep doing it. You either have to re-locate the compass or use a non-magnetic material for the canopy bow.

If there's anything worse than the British Thomson-Houston magnetos on the Gipsy, I don't want to know about it. Put the Slicks on.

Yes - that's why he pinches my reference books. Insanity is hereditary - you get it from your kids . . .

I'd like to make a point, here: Catalytic converters were introduced in order to reduce the oxides of nitrogen that produce "brown smog". Unleaded fuel was introduced because leaded fuel "poisons" the catalyst - with the result that the vehicle then emits hydrogen sulphide, which is not only smelly but VERY poisonous (up there with hydrogen cyanide). However, the reason given by governments at the time, was that the lead was affecting the brains of infants. This in Britain, where they still have a proportion of lead water-pipes! Brown smog is a product of big cities; Sydney used to be very bad for it in the '60s, when I was attending Sydney University. The public transport system was inadequate, and so were the roads - so it typically took me over an hour in the morning and evening rush ours, to get to & from the uni. There were tens of thousands of cars sitting bumper to bumper for at least two hours every day, with their engines idling. Too many people in one place, in short. The "fix" was - better road systems, better public transport, staggered working hours, decentralisation, car sharing, and yes, the introduction of catalytic converters. I suspect that catalytic converters probably did not play all that great a part in it; higher fuel prices meant less people could afford to sit in traffic jams by the hour. We have a great tendency to fix the wrong problem, for the wrong reasons, and to tell political lies about the reasons behind it. The CO2 / global warming thing is another such, in my view. So was the Y2K business. It keeps happening.

Try your local lending library

No, that's OK; but it's not my major field of expertise. I've told you what I am sure of. I am in an outlying area and have only limited broadband capacity, so I cannot indulge in much in the way of a discussion. Nor do I really have the time. I try to alert people to potential hazards; and propeller blade failure is one of them when you start playing about with the engine's ignition system. You'll find a reference to this on the U.K. PFA site.

I am not going to try to write a textbook on this subject, in this thread. The combustion process is affected by a lot of factors, many of which are incompletely understood. But the fuel-air mixture burns, it does NOT explode (at least, if the engine is to stay together more that a few seconds). Losing a mag. does NOT necessarily mean all the fuel isn't burned; it means it isn't burned at the right time. If you can find a copy, look up "The High-Speed Internal-Combustion engine" by Ricardo; that's the classic textbook on this subject. There's a vast amount of later information on this, on the internet; but it's difficult to put it all together unless you have read Ricardo's explanation first. However, the characteristic frequency of detonation is related to the cylinder bore diameter, which tends to confirm that it's a result of the rate of propagation of the flame front. (I managed to confirm that during detonation testing of the Jabiru 2200; you'll find the formula in the literature if you look for it.) I agree that going to full rich is not necessarily the best way to fix an engine running problem - but it's normally a least-worst initial response.

There's no such thing as unleaded Avgas. That may change in the near future, but it hasn't yet.

The two-magneto dual ignition system on Lycomings and Continentals (and all large radials, except the Guiberson) has a fixed advance, normally around 25 degrees BTDC (tho that varies a bit from one model to another). The impulse coupling provides 25 degrees retard, so it make it possible to hand-swing the propeller. Or for a fairly small starter motor to start an engine that has around two litres capacity per cylinder. The fixed advance means the two magnetos fire simultaneously, which is necessary to get full power from a large cylinder - and also to increase the detonation margin at lean mixture. Detonation is caused by the pressure rise as the flame-front travels across the combustion chamber; the further the flame-front has to travel, the greater the tendency to detonation. In a big-bore cylinder, the combustion chamber shape has been developed to take advantage of the shorter flame-front travel provided by dual ignition. If you lose one magneto, it's advisable to keep the mixture full rich - tho that is not commonly taught, except that if the engine runs rough, or the EGT rises noticeably, most pilots go to full rich automatically. The concept of having one of the magnetos replaced by a CDI system seems to be, if I understand it correctly, that at reduced-power cruise, the fuel economy might benefit from a bit more ignition advance. However that loses the advantage of the shorter flame-front travel when both plugs fire simultaneously; and as lean mixture settings such as one uses anyway for cruise economy take the engine closer to detonation, this longer flame travel is liable to result in the combustion in the part of the cylinder furthest from the plug fired by the CDI system, getting much closer to detonation than with the correct dual ignition setup. Even if detonation is not sufficient to show up as increased EGT, this can result in a sharp pressure rise late in the combustion process, which puts a sharp edge on the torque pulse reaching the propeller. The second plug firing late, has no effect if the flame front from the first plug has already passed it. NO Jabiru engine has a magneto, in the sense normally understood. It does have distributors, rather than being a "wasted-spark" system. The jab ignition system is self-powering, so it does NOT rely on the battery; it's all out in the open, so you can see how the ignition is excited by magnets on the ring-gear disc. However, the contact-breaker points are replaced by an electronic gizmo that detects the rate of rise of the current in the primary winding. Because the rise-point occurs a bit earlier at higher RPM, this provides sufficient ignition advance. It's a clever system, but it won't produce sufficient power to produce a spark unless the engine is turning fairly fast on the starter. I'm not familiar with how the slow-start boost setup works.

Self-sustaining or not, it may still break the propeller.

I gather you were trying to discover the wind speed and direction at your height? A flight data computer (which needs GPS or inertial input, as well as flux-gate compass data, and pitot and static inputs) will give you a windspeed. Problem is, they usually cost almost as much as a recreational aircraft. (Maybe there's an affordable one by now, but it will still need a flux-gate compass input.) Some gliding computers get the wind by averaging the GPS ground speed over a period of thermalling, assuming the rate of turn is fairly constant. Flying VFR cross-country in the days before GPS, one could eliminate the need to guess the wind, by using a time-scale, which has an elastic scale that you can stretch so it reads the distance across the map in minutes, rather than in hours. How much you stretch the scale is indicated in groundspeed, i.e. the groundspeed you are actually achieving - which is what you want to know, for fuel usage calculations.

There has been a practice of replacing one of the magnetos on a Lycoming or Continental, with a CDI system, which is claimed to improve the cruise fuel economy by advancing the ignition. There are two things against this - firstly, it alters the status of the engine from being type-certificated to being experimental; this may have (will have, in Australia) the effect of prohibiting flight in controlled airspace, as I understand the current rules. Secondly, it alters the engine's vibratory inputs into the propeller. If you have a certificated metal or composite propeller, or a wood propeller that is ground-adjustable, it must be vibration-cleared for the engine; you can find which propellers are cleared for which engines, by looking up the propeller Type Certificate Data Sheet. However, that clearance is only valid if the engine fully conforms to its certificated condition. There have been a number of propeller blade failures, including one fatal one involving an almost new Hartzell on an RV-6 in Australia, from this cause. If you're running a fixed-pitch wood prop, this is unlikely to shake a blade off, but it may make it more likely for the propeller hub face to char, which loosens the propeller, so both blades come off.