Pelican's Perch #57:
The Whyalla Report — Junk Science?
On May 31, 2000, a Piper PA31-350 Chieftain operated by Whyalla Airlines crashed in South Australia after suffering catastrophic in-flight failure of both engines, killing the pilot and seven passengers. A year and a half later, the Australian Transport Safety Bureau (ATSB) issued their 150-page investigation report and blamed the accident on buildup of
In December, 2001, the ATSB (Australian Transport Safety Bureau, similar to the U.S.'s NTSB) published one of the worst accident reports I can remember reading.
In my opinion, the ATSB has taken junk science, pure speculation and profound ignorance to levels seldom before seen. They have not only perpetuated "Old Wives' Tales," but they have invented a few new ones. I'm afraid we'll be hearing about "lead oxybromide deposits" for all eternity. As best I can tell, that term seems to have been INVENTED in this accident report. (Try a search on the Internet for "lead oxybromide"!)
My main complaints about the ATSB report are:
- The ATSB makes "lead oxybromide" deposits the central focus of
the accident, when it is very unlikely that ANY "deposits"
played ANY role in the accident;
- The ATSB seems to leave the impression that the rich side of peak EGT is
the only safe mixture setting to use; and
- The "probable cause" is nowhere to be found in the voluminous 150-page report.
I had hoped that dreadful report would not be widely read before it might be corrected. Alas, FLYING magazine has now picked it up, it's all over the Internet, and those of us involved in dispelling some of the longstanding myths in the engine world are starting to get increasing numbers of questions about some of the utter drivel in the ATSB report.
So far, the only intelligent comments I've seen in the printed media are in The Aviation Consumer (June 2002, page 22), but space did not permit them to say very much.
FLYING (July 2002, page 72) did rather poorly in their main article, with several errors, but in relying utterly upon the information in the ATSB report, they were badly led astray. The sidebar to that article is even worse. More on this later, below.
The ATSB didn't exactly condemn lean-of-peak EGT (LOP) operations, but their investigation report sure implied that the leaner an engine is run, the more likely damage will be. Then some time later, in a letter, they admitted that LOP operations had nothing to do with the failure of the right engine with the hole in the piston from detonation and/or pre-ignition. Every single person who has asked me about the Whyalla report had the distinct impression that the ATSB was highly critical of LOP operations. It seems the report's authors bought into Lycoming's recommendations completely (even when they contradicted themselves), and proceeded down the primrose path from there. They speak of "aggressive leaning" in a very negative manner, leaving the impression that if leaning to 50°F rich of peak EGT (ROP) is "too aggressive" (and it IS!), then leaning further must be even "MORE aggressive" (which it isn't).
This is a fable shared by most of the flying world, and by both TCM and Lycoming (in spite of their own engineering data from years ago). It is directly at odds with ALL of the old engineering data, with current test stand data, and with common sense.
Perhaps we need to add another curve to the power charts, and call it "Aggressiveness." Like EGT, CHT, and HP, it would "peak" at about the same point on the chart as CHT. That peak would then be "maximum aggressiveness," with "less aggressive" falling on BOTH sides of the peak.
Let me hammer this to death, and put it several different ways, for it is central to the issue, and central to the accident.
The problem lies in the widespread belief that "too lean," always refers to LOP operations. That's absolutely not true! In one recent gathering of pilots, the question was asked, "At what mixture setting (reference to peak EGT) will the engine (CHT) be the hottest?" The multiple-choice answers offered were 100 ROP, 50 ROP, Peak, 50 LOP, and 100 LOP. The result: 49 of the 50 chose "50 LOP."
The correct answer is 50 ROP. TCM, Lycoming, the old books, engineering principles, and actual data all confirm this. There is NO argument about this. So why do 98% of all pilots believe otherwise?
You can have a VERY rich mixture setting, and be "too lean." This might be the case if your engine requires 30 GPH for 38" MP during climb, and you set anything less (27 GPH, as Whyalla did).
You can have a very lean mixture setting, and ALSO be "too lean." This might be the case where you have leaned so much that "too much" power is lost, or where the engine runs rough, or quits entirely.
You can also be "too rich" on both sides of the mixture curve. Engine manufacturers specify a fuel flow for full power, and frankly, most engines need every bit of it, some just a touch more. But if you supply too much fuel (with a boost pump, for example), you may cause it to produce less power, and in extreme cases, it will "flood" and quit. That's definitely "too rich!" If you are operating on the lean side of peak EGT and your CHTs get too high (380°F to 400°F or more), you are "too rich" and need to lean it a bit.
Without qualifying language, terms like "lean," "too lean," "rich," and "too rich" are meaningless! We all need to be MUCH more specific when using these terms, now that our "flat six" and "flat four" engines are capable of running on the lean side of peak EGT.
Follow me through here, this is CRITICAL for understanding:
When you lean from FULL RICH on a properly set up engine, EGT, CHT, power output, horsepower per pound of fuel, and internal combustion temperatures and pressures ALL rise at first. At some point during that leaning process, FIRST the Horsepower peaks, THEN the CHT and internal pressures peak, THEN the EGT peaks, THEN the HP per gallon peaks, all at slightly different fuel flows.
Take a look at this chart, courtesy of "Advanced Pilot Seminars Inc.," a new company being formed for the express purpose of giving seminars on engine management.
This is a "concept" chart, intended to represent all internal combustion, spark-fired, fixed-timing, gasoline-powered engines. It is taken from the TCM chart in the back of the Installation and Operations Manual, generalized, and the ICP (Internal combustion pressure) curve has been added from actual test stand data. The Brake Specific Fuel Consumption (BSFC) curve has been inverted, so instead of the customary "pounds per horsepower," it shows "horsepower per pound." This way, the BSFC curve also peaks.
Imagine a ruler held vertically, and moved from left (rich) to right (lean), representing the leaning process (a "mixture sweep"). Each of the curves will touch the ruler and show what that parameter is doing at that point on the mixture curve.
Once each parameter peaks, then with further leaning, it will start to drop again. If this were not so, we could not talk about "peak" at all!
You simply cannot say "too rich" or "too lean" without some qualification. You need to state what you want to do, what the parameters are, or where you start, and where you end.
For example, Lycoming is fond of saying something like, "Mixtures that are too lean are harmful to your engine."
The more accurate statement is, "Mixtures that are not rich enough, or mixtures that are not lean enough may be harmful to your engine." This is NOT a contradiction! The "harmful point" is a CENTER POINT. Again, look back at the graph.
Why doesn't Lycoming (or TCM) talk much about the lean side? Both have admitted that their engines CAN be run LOP, but they seem to feel that pilots are morons, and cannot be trusted to understand the issues. They seem to think operating LOP is more difficult (it's actually easier), and more dangerous (it's actually safer). I can't help but wonder if anyone at Lycoming has ever seen LOP operations demonstrated in a properly-instrumented engine, in-flight. I find it hard to believe they have, or I don't think they could maintain the views they seem to hold.
Take an engine set up at 50 ROP (rich of peak). Is it "too rich" or "too lean"? We cannot say, without knowing what else is going on. At any power setting above about 60 or 65%, 50 ROP is BOTH "too rich," and "too lean," in that it is the worst possible mixture setting for the engine, and moving the mixture EITHER way will improve things. On the other hand, at about 9,000 feet and above, with no turbo, 50 to 80 ROP is an EXCELLENT mixture setting, for it will produce just about the maximum possible power, and it won't overheat anything at and above that altitude. You NEED 50 to 80 ROP up there! But 50 to 80 ROP and full power at sea level will destroy your engine.
Lycoming, TCM, virtually all the media, and sadly, most pilots and mechanics are much too careless with these terms.
What Really Caused the Whyalla Crash?
The Underlying Cause: Inadequate Instrumentation?
In my opinion, the primary cause of this accident was CASA's failure to require, and the operator's failure to install and use, proper engine instruments that monitor the EGT and CHT of ALL cylinders, as the JPI EDM-700 and 800 do.
(Standard disclaimer here, no, I don't work with or for JPI, and I receive nothing of value from them. I bought mine at retail. They did (once) do a free minor upgrade for me once, without comment. When I called to say thanks, they just said, "Aw, it's nothing, we appreciate the nice things you've said." They have nothing I want, and I would not accept anything from them, anyway. The simple fact is that JPI makes the flat-out, hands-down, best device on the market for monitoring an internal combustion engine. I think it's negligent to operate without one, and it's criminal that CASA and the FAA doesn't require them in public-use recips.)
This simple device would have allowed Whyalla (and others) to operate this very high-performance engine much more intelligently, it would have paid for itself in fuel savings alone, and it would have warned them of their folly the instant they started using their dreadfully uninformed leaning procedures for climb in the Navajo Chieftain. The recording feature would have acted as an excellent flight recorder, for the memory chip would have preserved the final 70-odd hours of engine operation on both engines. That could have told us exactly what happened to the engines on the Whyalla crash aircraft, AND helped prevent such accidents in the future.
Instead, the FAA (and probably CASA) throw roadblocks in the way! You want to pull a useless factory single-point EGT gauge out and put a JPI in? "Oh, horrors, no, if the airplane came with that worthless EGT, you have to keep it, forever!" Suggest tossing the equally useless CHT gauge most factories install, and see how far you get with your local FSDO.
The Lycoming TIO-540J2 series of engines are good engines. They have some "different" design features, which have worked out pretty well in years of field experience. However, they are very high-powered engines that take the piston engine art and technology to just about the same performance levels found in the old high-powered big radial engines of yesteryear (some of which enjoyed 3,600-hour TBOs with decent instrumentation and techniques). The TIO-540J2x-series engine is universally considered the most difficult engine in the general aviation fleet with respect to detonation margins.
All this is somewhat reminiscent of the old Wright R-3350 engine. It had some early problems, and Wright did a number of changes mechanically, and together with the airlines and the military, they developed operating techniques that eventually made it perhaps the finest big radial of all. But they absolutely had to have BMEP (Brake Mean Effective Pressure) and fuel flow gauges to do it. Today's JPI is the "new and improved" equivalent of the old instrumentation.
Closely related to the failure to install a multi-probe engine monitor is the operator's ill-conceived procedure of leaning to an EGT of 1,500º F on the rich side of peak EGT during climb at high (36-38" MP) power (note those qualifiers, they're critical!). NO ONE, to my knowledge, even hints that this might be "good." It is an absolute fact that it is very, very bad for any engine. It is a tribute to the basic engine that they suffer this insult for more than a few flights.
Please note it's not the EGT itself that is "bad for the engine." An EGT of 1,500º F CAN be a very nice, very good setting to use, IF IT'S ON THE LEAN SIDE OF PEAK. It CAN be okay, if the engine is operated at lower power settings (60% or so), if you can get the EGT that high. 1,500º F isn't going to melt a thing, rich or lean. However, on the rich side, at high power, we know from experience (and test stand data) that an EGT of 1500ºF or more at 38" MP while ROP means other BAD things are going on in THIS engine. CHTs will be unacceptably high and going higher, peak pressures and temperatures within the combustion chamber will be soaring out of limits, and detonation will not be far away. At this point, the margins become so thin that if anything at all goes wrong, major engine damage or failure will result. This could well cause a pre-existing defect (as in a crankshaft) to suffer fatigue and failure much sooner, and it could cause long-term damage from detonation. This procedure could have accelerated the damage on both engines in the Whyalla crash, over time.
The Proximate Cause: A Flawed Crankshaft?
Beyond that, the cause of the crash was probably rather simple. BOTH engines were probably damaged to some degree before the fatal flight departed.
In my opinion, the left engine probably had a long-term defect in the crankshaft just waiting for some final stress to break it, a true time bomb. That defect may well have come with the crankshaft from the factory some 262 flight hours earlier. It strikes me as unlikely the damage was caused by bearing failure, since there was very little bearing debris found in the screens, AND the right engine was operated the same way, for much longer (presumably without a factory defect). If there had been a progressive bearing failure bad enough to damage the crankshaft, there would have been a lot more evidence in the oil screens. I cannot find any mention of oil analysis in the report. If this wasn't done, why not? If it was done, why wasn't it mentioned? Did the operator use oil analysis?
The right engine may have had some detonation damage from the operator's failure to maintain an adequately rich mixture during the climb phase of each flight. It probably had something else, perhaps a damaged helicoil, the part into which a spark plug is inserted, or perhaps a partially plugged fuel injector, or even just a single fuel injector that was within factory tolerances, but at the extreme low end of the +/- 4% flow tolerance for Lycoming nozzles. The report does mention that the injectors were tested, and found within limits. But what no one, including Lycoming, seems to understand is that if you have an engine with injectors at both ends of the 4% limit, you have an engine doomed to problems, or failure. Suppose five of the injectors are flowing 4% high, and one is flowing 4% low? That is enough to show a HUGE difference in the parameters, and may easily place one cylinder in detonation without any sign of it — until it fails. The single-point EGT won't show this, but the JPI Engine Monitor will, AND will issue a warning, as well!
I find it odd that a later Lycoming Mandatory Service Bulletin (SB-550) called for a massive recall of crankshafts built around the same time as the accident aircraft's left crankshaft, but I don't see any evidence that anyone pursued an examination of the failed crankshaft to see if it is — or should have been — part of the recall.
Consider the time line:
- The Whyalla left engine was overhauled by Lycoming on or about January
6, 2000. It was installed in the airframe on February 13, 2000.
- Reports from the field in Australia claim that it was acknowledged at
the time of the crash in June of 2000 that the crankshaft was new when it
was installed during the overhaul on January 6, 2000 by Lycoming.
- Then, on February 1, 2002, just a little over a month after the ATSB report is issued, Lycoming issues SB-550. It has a long list of engine serial numbers, including a couple of dozen TIO-540J2B engines identical to the Whyalla engine that were grounded by the Service Bulletin.
At the end of SB-550, there is the following statement:
NOTE: If the crankshaft in your engine was replaced with a factory new part in the time period from 9/30/99 to the present, and your engine is a Lycoming turbocharged (L) TIO-540 engine rated at 300 horsepower or higher, please contact the factory to determine whether your engine is affected.
With this time line firmly in mind, it seems to me that there are only two possibilities:
- The ATSB didn't know about the upcoming Lycoming crankshaft recall
(surely Lycoming knew of the crash, and the failed crankshaft); or
- The ATSB did know about the upcoming crankshaft recall — but then why wasn't there any mention in the report that there was a problem series of crankshafts made at the same time as the crankshaft in the Whyalla aircraft, nor any indication that a metallurgical investigation of that specific crankshaft was done?
I find either scenario troubling.
... and/or Bad Bearings?
What Lycoming doesn't mention in SB-550 (nor does the accident report) is that there is also an issue with connecting rod and main bearings. Apparently sometime in the last three years, there were some changes in the composition of the bearing alloy comprising the middle layer of this three-part bearing (three layers, bonded together). There have been a number of problems with bearing delaminations in Australia during the last three years, and this is apparently under investigation. This three-year period covers the same time frame in which the left engine on the Whyalla Navajo was installed. Sometime while these bearing failures were surfacing on the radar screen, Lycoming began supplying "Dana Clevite" bearings (a different brand) for replacement and field overhaul use in the Australian engines (One wonders if engines have passports to identify their nationality, or if the air is somehow "different" in Australia?).
Apparently the Australian CASA ("Civil Aviation Safety Authority," the Australian counterpart of the U.S. FAA) asked Lycoming, "why the change, and why only in Australia?" In reply, CASA was reportedly told that Aussies are the only ones with a bearing problem, and they need to check their gasoline.
Say what? Unless the gasoline was substantially out of octane specification (it wasn't), what could gasoline have to do with bearing failures? I suspect CASA was as speechless as I am over that VERY strange comment. Lycoming might as well have told them to check their beer! In fact, considering the taste of Aussie beer... well, never mind, I've got too many Aussie friends who actually like their beer, and I'm too old for bar fights.
When CASA regained the power of speech from that absurdity, they asked Lycoming why they were changing out ALL the bearings in ALL the engines in Piper Malibu Mirages (SB-547).
Lycoming's response was reportedly, "Oh, that was for 'product improvement' and there was no connection to the problem in Australia." Riiiight. Boy, do I have a crankshaft for you!
That the left crankshaft failed is undeniable. But crankshafts are MASSIVELY overbuilt because of their inherent critical single-point-failure nature. They simply do not fail in the absence of a metallurgical fault, or some very abnormal cause — such as a failed bearing.
Whether the crankshaft failed from a defect in the crankshaft itself, or from a failed bearing isn't germane to the accident — but it's interesting, to say the least. With a defective crankshaft and long-term fatigue (over 262 hours, some of it leaned for climb with attendant very high combustion chamber pressures), the failure would have been instant and massive, with no warning at all. The accessories would have stopped, depriving the engine of oil pressure. Without oil pressure, the left prop would have instantly feathered all by itself (by design on light twins), with no action from the pilot. "Autofeather" if you will.
With a bearing failure, it's even more interesting. When bearings fail, they get hot. WHITE hot, very quickly, and oil temperature will rise dramatically. This is especially true on the main crankshaft journal bearing on the big radial engines, for there is only one, and lots of oil passes through it. On the Whyalla "flat six," the rise might have been less evident. It's odd that the accident report mentions a series of flights a few months before the accident where the left engine was showing an oil temperature of 250 degrees. That's probably not related, because I don't think the engine would have operated for that long with a badly damaged con-rod bearing, but I wonder if it's possible ... .
View from the Left Seat
The Obvious Scenario ...
The obvious scenario is that the left engine crankshaft failed abruptly from fatigue at a long-term defect, and the engine automatically feathered due to loss of oil pressure. It occurred on a cloudy, moonless night over dark terrain. It would have been a nasty surprise, and a difficult situation, one I'd not like to face.
The very young pilot (22), perhaps a bit rattled (as anyone would have been), might have advanced the power on the right engine, while forgetting to increase the mixture. This is a VERY common error in training and on check rides, even with very high-time captains. We have ALL done that a time or three. When you do that, the cylinder head temperatures rise rapidly. After some short period of time (it could be as short as a few minutes) at that power setting, it is likely detonation would have begun. Another minute or two of light detonation, progressing to moderate and worse, and it would have turned into pre-ignition (a spur from a bad helicoil can easily get hot enough to remain white hot through the exhaust and intake stroke, lighting off the fuel/air mixture sometime before the spark plug's normal timing). Pre-ignition is the worst possible malfunction in the combustion chamber. This whole problem of failing to increase the mixture when the pilot increases power on the remaining "good" engine would be substantially more dangerous if there was a partially plugged injector, or if one injector was at the low end of Lycoming's 4% tolerance band, with the rest somewhere above that. This is fairly common and cannot be readily detected without a multi-probe engine monitor.
(Note that a JPI EDM-700/800 engine monitor not only would show that to the pilot instantly, but to the maintenance staff the first time they looked at the data.)
Once pre-ignition began on the right engine, it would have been catastrophic within seconds, causing that one cylinder to fail, with major vibration that could easily lead the pilot to believe the engine had failed completely.
However, this scenario does not explain the sudden drop in groundspeed at time 18:47:15. (See plots on page 3 and 4 of the accident report). To make the speed they did after that failure, according to the report, it would have been necessary to use a total of about 375 HP. (This calculation is not in the report, but the 375 HP number is.) That would mean that with one feathered, the pilot would have had to have set takeoff MP (or more), takeoff RPM, AND leaned it for "best power," substantially more power than the engine was rated or designed for. While he might have done this (briefly) to keep the airplane in the air, I cannot imagine a pilot dumb enough to press on to a distant destination, at night, over water, with that condition, especially with paying passengers aboard.
Sadly, the FAA, the media, and the public tend to jump to the conclusion that all pilots are stupid, and somehow cause the problems that lead to accidents. Pilot error IS usually involved in accidents, but we pilots usually don't make those mistakes without something leading us there.
... and an Alternative Theory
There is yet another theory that also makes a lot of sense, and does cover all the known facts.
Suppose that during or after yet another grossly insulting mixture setting on climbout the right engine finally holed the number six piston? This could have been from the 1500°F TIT ("too lean" or "not rich enough," however you want to look at it) climb, or from the climb and the combination of a partially plugged nozzle or from a bad helicoil, or some combination thereof. At first, the vibration might not have been noticeable, but as the hole burned through the piston, no pilot could ignore it. The natural response would have been to reduce power (and RPM) on that (right) engine. The vibration would have gone away at some point, and with an initially small hole, the engine, throttled back, might well have produced a 100 or 150 HP, and would have seemed pretty normal. I've seen cases where an engine like that operated pretty normally for many hours at low cruise power settings. On each compression stroke, the fuel/air mixture will blow through the hole in the piston, pressurizing the case with fuel and air, and blowing out the breather, taking some small amount of oil vapor and oil spray with it. The spark plugs firing would probably light the fuel/air mix, adding heat, and eating at the hole. Eventually, that might have led to oil starvation, but it would have taken some time.
There are rumors the young pilot had been previously criticized for landing an airplane away from home with a problem, rather than bringing it home where maintenance could have been performed. I'd guess the young pilot felt his job was on thin ice.
Okay, so now (continuing with this theory) he's cruising along at reduced speed, with the right engine at reduced power (but "running fine"), left engine normal (or perhaps a little higher power setting than normal), everything is just fine. With the right engine making 125 HP, and the left engine making 250 HP (about 70%), the aircraft would have the required 375 HP to maintain the speed observed by reference to the radar track and known winds aloft. Under these circumstances, any of us might have been tempted to take it home, for all the usual reasons. Schedule, maintenance and personal needs might have driven that decision. Pilots usually prefer to go home, the passengers probably wanted to press on, etc. It might not have been the best decision, but in my opinion, it wasn't all that unreasonable, either. MANY of us have done similar things and survived, perhaps because of better luck. That's certainly not as dumb as pressing on home for a few miles, with one engine out completely, after a massive failure, as in the first scenario.
In hindsight, of course, it would have been a dreadful decision, but that's "Monday morning quarterbacking."
Somewhere out over the Spencer Gulf (page 3 of the report), Murphy stepped in, as he so often does. The 262 hours of fatigue accumulation in the left engine crankshaft (whether from failing bearings or original metallurgical origin, it doesn't make any difference) finally became critical, and the crankshaft failed. The prop probably feathered before the pilot even knew it, and the engine was useless.
Any attempt to pull more power out of the right engine (with the hole in the piston) would have resulted in heavy vibration, and would have quickly began eating that hole larger and larger. There wouldn't have been any more detonation or preignition, as there wasn't enough pressure and temperature to get there, but the hot gasses blasting through that hole would have acted like a cutting torch. The oil would disappear out the breather in minutes at high power. With one engine feathered, and the other at very limited power, it wouldn't have taken long to the splash. Depending on the RPM of the right engine, there might not have been enough vacuum to drive both horizons and both directional gyros, leaving only an electric turn and bank (which the ATSB called a "Turn and Balance" indicator — is that common Australian usage, perhaps?)
Could It Have Been Prevented?
A couple of scattered thoughts.
First, in either scenario, let's look at the possibility that the accident pilot added power to either engine, without enriching the mixture. I believe that mistakes like this come from engine failure procedures that are much too complex, with too many memory items, as taught in the old days by the military to fledgling pilots, and espoused in the FAA's "Flight Training Manual" for student pilots.
The best procedure is to identify the engine, feather it, maybe cut the mixture (to keep the engine from running in full feather), then take a deep breath, forget the failed engine, and do what is needed to FLY THE AIRPLANE NORMALLY on the remaining engine. When the situation is stable, then run any appropriate checklists for the engine failure/shutdown. In the accident aircraft, under the first scenario above (left engine failed first), the pilot probably had plenty of power already set on the right engine to make the destination, as he had already started a normal descent. But most multiengine instructors insist on "max power, right now" on the good engine, or "at least the next higher power setting," plus all kinds of other memory items. (I've seen up to 13 memory items on an emergency engine-failure checklist!) Many check pilots will insist on that on all check rides, whether the power is needed or not. Doing "as trained and checked" may have contributed to this fatal crash.
We DESPERATELY need BETTER multiengine training in general aviation! (I've written about this subject before, in "Pelican's Perch #4: Engine Failure!")
Second scattered thought: The Navajo and Chieftain engines (Lycoming TIO-540-J2B) are, as noted above, good engines, but notoriously among the most difficult-to-manage and easy-to-abuse engines in the general aviation fleet. Operated with proper intelligence and instrumentation (JPI or similar) and flown often, it is a fine engine, fully capable of going to TBO and beyond. It has a fine control system for controlling mixture in close coordination with the throttle, prop, and turbo system. In accordance with the very best engineering principles, it gives a very, very rich mixture at takeoff power, and then runs it somewhat leaner (but still very rich on the rich side) at the higher power settings just below takeoff power.
But even when operated in accordance with the Lycoming recommendations it is possible for the engine to operate so close to the detonation threshhold that any anomaly in the fuel setup, the spark timing, or a slightly damaged helicoil can lead to rapid destruction of the engine. Those recommendations may make for good marketing in terms of fuel flow and performance, but it's lousy engineering. Of course, by the time failures do occur, the warranty will have expired.
Given the critical margins, it's amazing the engines did as well as they did with the ill-conceived leaning procedures used during climbs at Whyalla Airlines.
The same engine CAN be operated very successfully at very high cruise power settings — if operated lean of peak TIT. Most will probably go far beyond TBO that way. To do that, you must have a good all-cylinder engine monitor, and you must have all cylinders at the same fuel/air ratios. There is good data on these engines that shows LOP works very well, with nicely lowered cylinder temperatures and reduced peak cylinder pressures for any given power. Apparently, from comments Lycoming makes to FLYING and in its official Lycoming publications (see "'Experts' Are Everywhere To Help You" on Lycoming's web site), someone at Lycoming does in fact understand this, but holds the opinion that pilots are too dumb to do it right, and therefore has shown a total lack of interest in seeing or acknowledging the existence of ANY data on this subject.
Lack of Communication
In fact, let me take a major slam at both Lycoming and TCM for their utter lack of interest in any form of communication with the outside world. From what I understand, letters generally go unanswered unless there is an immediate threat of litigation, and then the lawyers handle it. There seems to be NO way to communicate with these companies by email, a major blunder in my opinion. Both companies should have INTERACTIVE Web sites, with competent engineers responding directly to questions and correcting "those experts on the Internet." We could all learn something, and we'd have better engines, too. In fact, why hasn't a Lycoming person EVER made a single comment on my columns in an effort to correct them?
Let me take a slam at FLYING, while I'm at it. In the July issue, FLYING's writers print leaning recommendations obtained from both Lycoming and TCM. FLYING correctly notes that TCM agrees with GAMI's recommendations on leaning and its effects in the critical area from peak out to 50°F rich of peak (there are some missing details here, but that's good enough for our purposes today). But then FLYING does something that I consider indefensible for a magazine article dealing with the subject of leaning: FLYING simply ignores the fact that the Lycoming leaning recommendation is directly and irreconcilably in conflict with the TCM recommendation.
There is NO significant difference in the combustion process or the metallurgy between TCM and Lycoming engines. If one can be safely operated lean of peak, then both can.
For the record, the comments attributed by FLYING to TCM, expressing agreement with the statements quoted from GAMI, is the side of the discussion which is correct from an engineering point of view. In my opinion, FLYING should not have given Lycoming a pass on that glaring conflict in leaning recommendations. It seems to me that they did a profound disservice to their readers.
(Say, do you think there is any chance that FLYING will ever ask me to write for them?)
|More on "Lead Oxybromide"
The ATSB seems fixated on the term "lead oxybromide," and I could find NO reference to it, anywhere. A friend, knowledgeable in this area, did manage to dig up the following data, which MAY be what the ATSB is talking about, under another name, perhaps a colloquial name.
A search of Chemical Abstracts, 1967-date, and of Gmelin, Inorganic Chemistry Handbook shows (for the scientists out there):
Lead bromide oxide (Pb3Br2O2) (8CI, 9CI)
Other names: Lead bromide (Pb3Br2O2) (6CI)
Other names: Lead bromine oxide (Pb3Br2O2)
GMELIN Reg. No. (GRN): 419194 GMELIN
Melting Point = 709 degrees Celcius (sic)
If we convert 709°C, we get 1,308°F.
(By the way, I so wish we Americans could/would switch to the far simpler and much more logical metric system! But all our data is in °F, so I go along.)
That 1,308°F melting point destroys the ATSB's whole thesis, if this is the "lead oxybromide" to which they refer.
Temperatures inside the combustion chamber vary widely. The actual fuel/air mixture can burn within that space at well over 3,000°F. The tips of a spark plug, protruding slightly into that space normally run temperatures around 1,200° to 1,300°F, and in very high power applications, up to 1,500°F. Somewhere above that, the spark plug tip is known to cause pre-ignition. (These temperatures can only be measured by special spark plugs, and rather exotic equipment, but they are pretty accurate.)
We also know that pistons, being aluminum, lose half their strength at 400°F, and melt around 700°F. They actually track CHT fairly well, so figure they run a face temperature of about 400 to 450°F. Close enough for our purposes, anyway.
But wait! If the combustion process runs at 3,000 degrees, how do pistons, or even the steel cylinder walls survive? First, there is a very thin thermal barrier, a "transition zone," with the cooler metal on one side, and the hot, intermittent combustion event on the other.
That combustion event is just a quick flash of that very high heat, during 20 degrees or so of crankshaft rotation. The rest of the 340 degrees of crank rotation, the temperatures are much cooler. During the intake stroke, the combustion chamber may even see a blast of sub-freezing air! It's like passing your hand through a candle flame. Do it quickly and you feel nothing, keep it there longer, and you'll find the melting point of skin.
Now, the key to all this: If a spark plug tip at 1,500°F will not cause pre-ignition, then how can a substance that MELTS at 1,300°F hang around in the combustion chamber long enough to rise to more than 1,500°F, and trigger pre-ignition? That's right, it can't.
Conversely, if that same substance somehow attaches itself to the face of the piston, and acts as a heat sink, how can it possibly rise to a temperature sufficient to cause pre-ignition, without melting and blowing out the exhaust? If the surface gets hot enough to melt (1,300°F), it simply erodes away, blows out the exhaust, and each combustion event will melt a little more. Pretty soon, no more "stuff."
Oh, by the way, you may be wondering, just what are those deposits?
Carbon, the raw material from which diamonds are made. Carbon, which has a very high melting point. Carbon, which IS capable of causing pre-ignition, once the carbon is hot enough - from detonation. Carbon deposits are found in engines run rich of peak for many hours. ONLY in engines run rich of peak, because if you run lean of peak, carbon deposits are eroded, converted to CO2, and blown out the exhaust!
Lead oxybromide? Bah, humbug!
— John Deakin
Dissecting the ATSB Report
Okay! Off my soapbox for a moment. Let's take a peek at some of the stuff in the report. I have extracted bits and pieces for comment, doing my best to maintain context. The report is in italics, my comments are in normal typeface.
On the evening of 31 May 2000, Piper Chieftain, VH-MZK, was being operated by Whyalla Airlines as Flight WW904 on a regular public transport service from Adelaide to Whyalla, South Australia...(text omitted)…The entire flight was conducted in darkness.
They might have added here, "… darkness, on a moonless, cloudy night, over barren terrain with few lights." I think that's very pertinent, it's a nightmare for any pilot.
The aircraft reached 6,000 ft and proceeded apparently normally at that altitude on the direct track to Whyalla. At 1856 CST, the pilot reported to Adelaide Flight Information Service (FIS) that the aircraft was 35 NM south-south-east of Whyalla, commencing descent from 6,000 ft. Five minutes later the pilot transmitted a MAYDAY report to FIS. He indicated that both engines of the aircraft had failed, that there were eight persons on board and that he was going to have to ditch the aircraft, but was trying to reach Whyalla.
Note that within FIVE MINUTES, he apparently went from beginning a normal descent with no mention of problems, to a double engine failure. He was apparently very calm, cool and collected on the radio.
On 9 June 2000, the wreckage of the aircraft was recovered for examination. Aside from the engines, no fault was found in the aircraft that might have contributed to the accident. Both engines had malfunctioned due to the failure of components of the engines.
Left Engine Failure
I get the impression that the broken crankshaft was found (but not by the ATSB), and the "investigation" focused on that almost to the exclusion of the hole in the piston on cylinder #6 on the right engine. One question that occurs immediately is about the magneto timing. That should have been checked on both engines, and reported. It might have had a major bearing on the accident, as a mistimed engine could have greatly accelerated detonation, pre-ignition, and contributed to the fatigue in the crankshaft. (Later information indicates the timing was checked and found normal - but not reported.)
The crankshaft of the left engine fractured at the Number 6 connecting rod journal.
Fatigue cracking was initiated by the presence of a planar discontinuity in the journal surface. It was evident that the discontinuity had been caused by localised thermal expansion of the nitrided journal surface following contact with the edge of the Number 6 connecting rod big end bearing insert. The crankshaft failed approximately 50 flights after fatigue crack initiation.
Please, will someone tell me how they came to that "50 flights" conclusion? The stress riser that was the point of origin for the fatigue crack was mostly likely there when the crankshaft was installed, or it was created later by the bearing delamination problem, nowhere discussed in the ATSB report. Cranks are massively overbuilt, and fatigue cracks don't appear by magic.
The Number 6 bearing insert was damaged during engine operation through the combined effect of:
- high bearing loads created by lead oxybromide deposit induced preignition, and
Oh, RUBBISH. This accident report is the first mention of "lead oxybromide" I've ever seen! A fairly serious search of the Internet with several different search engines produced NO "hits" on that new term, EXCEPT ON THIS ACCIDENT REPORT! The deposits pictured are NORMAL deposits in any engine normally run rich of peak (ROP), and will disappear if temperatures rise above normal. In fact, a little detonation is a good way to clean those deposits off entirely! The richer the engine is run, the greater are those normal deposits. It's more common to see normal engines with them than without. There is no evidence that the ATSB bothered to check any normal engines, and now people will be pulling engines for no good reason.
(In fact, one well-meaning person has already tried to find a lab to "check his spark plugs for lead oxybromide deposits every 100 hours! No, I'm not joking! Hmm, now there's an opportunity for a new business, perhaps? Call it "Snake Oil Detection Services," or something.)
The ATSB used a picture from John Schwaner's "Sacramento Sky Ranch" website, and made it one of the central issues of the whole report! The picture can be found at http://www.sacskyranch.com/photopis.htm. In the report, the ATSB claims that the swirl pattern on the piston in the picture is a swirl pattern that is characteristic of detonation. This statement by the ATSB is one of those monumental blunders in report writing. It is the kind of blunder one makes when trying desperately to find evidence to support an erroneous pre-conceived notion of how some series of events took place.
The author of the ATSB report seems to have failed entirely to read what John Schwaner (correctly) said in the caption right under the picture! Here it is:
Swirl markings on top of piston are normal combustion pattern markings. They show how the hemispherical (dome shaped) cylinder head and the induction system swirls the mixture for better mixing and burning. [Emphasis by Deakin.]
Bullet item two in the ATSB report on why the bearing failed:
- lowered bearing insert retention forces associated with the inclusion of an anti-galling compound between the bearing inserts and the housings.
The use of anti-galling compound is both common and recommended in the overhaul manuals. This one is a complete red herring.
Fatigue cracking in the Number 6 connecting rod big end housing had developed following the gradual destruction of the bearing insert.
The fatigue crack grew over time, perhaps even taking the full 262 hours to develop to the point of failure. I think the failure occurred without any warning to the pilot whatsoever, was quite abrupt, with the jagged ends rubbing against each other long enough to create the heat from friction that caused the rest of the damage, plus immediate feathering from lack of oil pressure.
(Most light twins are designed to be feathered by a large spring, and are kept out of feathering by engine oil pressure and governor oil pressure, both of which regulate RPM. To prevent the engines from feathering on shutdown, there are small centrifugal latches that prevent the blades from going into feather below about 400 RPM or so. Most single-engine airplanes are designed with the opposite setup, with the prop going flat with no oil pressure. Radials have a different system entirely.)
Right Engine Failure
The physical damage sustained by the right engine was restricted to the localised melting of the Number 6 cylinder head and piston. The piston damage had allowed combustion gases to bypass the piston rings.
I think it might be significant that only the #6 cylinder suffered this destruction. The most likely things that can cause this damage in a single cylinder are:
- A "tang" from a damaged helicoil (the insert into which the
spark plug is screwed). This tiny projection would probably have been
harmless in normal operation, and might have existed for many, many hours,
or even until TBO. However, once detonation began due to the application
of high power with a mixture setting rich of peak, but not rich enough,
that small helicoil projection could have become white hot during each
combustion event, getting hotter and hotter, finally getting hot enough to
keep enough heat through the subsequent exhaust and intake strokes to
"pre-ignite" the fuel/air charge before the spark plug did so.
It takes only a few degrees of this "pre-ignition" in advance of
the normal timing to almost instantly run the temperature and pressure
right out of sight, causing immediate destruction. The rest of the
cylinders may have also experienced some detonation, but without a damaged
helicoil, they didn't suffer the pre-ignition event.
- If a single fuel injector was partially plugged, or even just out of
calibration on the low side, that defect, in combination with the failure
to use adequate fuel flow for the whole engine during the climb, or later
when the pilot may have increased the power on the right engine, could
have been enough to start the detonation that led to pre-ignition and the
hole in the number six piston. Without a multi-probe engine monitor, the
pilot would never have any idea that the fuel injector was partially
- It could have been a combination of (1) and (2).
The damaged piston would have caused a loss of engine oil and erratic engine operation, particularly at higher power settings. Engine lubrication was still effective at impact, indicating that oil loss was incomplete and that the piston holing occurred at a late stage of the flight.
This is not completely correct. The progressively enlarging hole in the #6 piston would have ultimately prevented compression, and blown the fuel and air charge through the hole, pressurizing the crankcase. That would gradually, over several minutes, blow some of the oil in the sump overboard, and eventually that might have unported the oil pickup and starved the engine of essential lubrication and cooling. At reduced power settings, the engine should have been capable of pulling some power long enough to make the Whyalla airport (absent the failure of the left engine), but would have been running very, very rough with only five cylinders working. (Large radials with 18 cylinders have been known to run for hours with an inoperative (or even a detached) cylinder, with barely noticeable vibration.)
Speed variations, accompanied by track irregularities, then became more pronounced.
I wouldn't place too much credence on this data after the top of descent. We'd all wander around a bit, with a catastrophic engine failure at night, on instruments (Yes, the weather was good, but without surface lights, it would have been the same as instrument flight). Add in the possibility there was insufficient suction to drive the four primary gyros, two on the pilot's panel, two on the co-pilots (yes, ALL FOUR were vacuum-driven), and there's enough to cause some wandering, if not outright loss of control. Note too, there was one stall warning sound on the ATC tapes, during transmission from the pilot.
Executive Summary Conclusion
The power output from the left engine deteriorated during the first third of the cruise segment of the flight after the Number 6 connecting rod big end housing had fractured. The engine ceased operating completely 8-10 minutes later.
Okay. Yes, it's okay. Go ahead. Laugh out loud! There, now didn't that feel good? With that kind of damage, the failure of the crankshaft and engine would have been instant, and catastrophic, probably causing a hole in the case (which was apparently found a few weeks after the ATSB completed their initial examination of the engine without even seeing the hole). I don't believe this was the failure mode — the crank probably failed first, destroying the end of the con rod — but that's not all that important.
In response to the failure of the left engine, the pilot increased the power setting of the right engine.
A good bet, although it might not have been needed in the first scenario I suggested. A gentle "drift-down" might have been better, but that's "Monday-morning quarterbacking" and I won't criticize the pilot for that kind of decision. He probably did just what he'd been trained to do, but probably forgot to increase the mixture when he added manifold pressure.
Increased combustion chamber component temperatures via detonation within the right engine led to the Number 6 piston being holed. That resulted in the erratic operation of the right engine with reduced power and controllability and left the pilot with little alternative but to ditch the aircraft.
Maybe, maybe not. I believe there is a significant possibility in the first scenario above that the pilot might have been able to extract enough power from that engine to provide a slow "drift-down" to the runway at Whyalla, or to a closer airport, using cruise power or less.
- The double engine failure was a dependent failure.
I am uncertain as to what this means. The one failure had little or nothing to do with the other. If the bearing in the left engine failed, and caused the crankshaft failure, THAT'S a "dependent" failure to me. The ATSB must have a different definition.
Examination of eight failures of Textron Lycoming engines from a number of operators that had occurred over the period January 2000 to November 2001 revealed that deposits of lead oxybromide on combustion chamber surfaces were not restricted to the engines from MZK; seven other engines had such [NORMAL] deposits.
I don't believe any were identified by other parties as "lead oxybromide," a term seemingly invented by the ATSB for this accident report. But in any event, they'd find those same normal deposits on engines that went to TBO ... if they looked. The absence or presence of those deposits will depend on how the engine is operated. Mine are utterly clean, because I operate at very high cruise power, and LOP, leaving no deposits.
Anecdotal reports indicated that there were fewer engine problems (including component failures) in engines that were operated full rich during climb, and 'best power' during cruise, compared with those where the mixture was leaned during climb and 'best economy' cruise power was used.
This is disingenuous at best. They compare "good" (full rich at high power) with "bad" (best power at cruise) as if they go together, perhaps to prove the point they invented. The two power settings have nothing to do with each other. Very little leaning at full climb power (and higher) on the rich side of peak EGT is tolerable by any of the "high performance" engines, and the practice used by the Whyalla operator gives me cold chills.
Setting "best power" (approximately 80°F ROP) at any power setting much above 60% is nearly the worst possible place to run any of these engines, unless you are a marketing type interested only in speed, and NOT interested in engine longevity! (The worst is at max CHT, at about 40°F ROP.)
No, I'm Not Done
Next month, we'll turn to the accident report itself, and pinpoint the numerous errors that led to an utterly silly conclusion. It may take TWO more columns to cover all the points.
While I have presented two alternative scenarios above, we absolutely must leave open the possibility there are others. Accident investigation is not an exact science, and following the shreds of evidence is challenging indeed. Sadly, the ATSB blew it, badly.
Lead oxybromide, indeed! Good grief!
Be careful up there!