Dafydd Llewellyn

The after-effects of prostate cancer require this, also; so the same answer will work for both halves of our team. I carry a travel mug in the car, and sip to prevent dehydration (which is not all that much different to flight, in a car with the aircon on, as it needs to be most of the year in QLD). An insulated cyclist's bottle with Staminade and a length of TYGON tube (don't use PVC) with an end-plug will work in the air, to prevent dehydration; the trick is little & often. I can manage 3 hours in a car, so I expect 2 - hour legs in an aircraft will be OK.

Yes - well, for the models that were certificated, this work was handled by either Alan Kerr and myself, or Alan Kerr and Keith Engelsman; we used our own instruments, and we gave the factory advice, not vice-versa. Look up MS 28034-1 on the internet; that's pretty much the GA industry standard oil temperature probe, and that's what I used. It reaches much deeper into the sump than the normal factory probe.

Just an observation, here: You need to be aware of which oil temperature you are measuring - the oil INLET or the oil OUTLET. Aircraft engine oil temperature limits are normally specified as oil INLET temperature. Rod has stayed with the oil temp. sensor in the sump, because for a long time he was trying to cool the oil by fins on the sump. In that situation, the sump temperature near the suction line inlet is the oil inlet temperature. But when you install an oil cooler that picks up from an adapter under the oil filter, the oil going into the main gallery is no longer at the sump temperature; it's cooler. Just how much cooler I don't yet know. The oil temperature in the sump is in reality the oil OUTLET temperature, for any installation that uses an external oil cooler. The real oil INLET temperature should be measured in the oil cooler adapter, where the flow is returning from the oil cooler. This is one of the things I will be measuring in the test cell, and later in flight. Unfortunately, the "oil inlet temperature" limits given in the TCDS - and hence in the POH - are still based on a sensor in the sump. I'd like to change that, for the CAMit engine, but the basis for approval may prevent that, at least until Ian can manage a supplemental type certificate. Until then, we're still stuck with the existing TCDS basis. So your measurement down the dipstick tube would indeed show a high oil temperature. It may well differ from the value shown on the Jabiru panel instrument, because the short sender used by that instrument will be somewhat affected by the boundary layer of somewhat cooler oil due to the air blast on the underside of the sump. There are all sorts of tricks for inexperienced players, in certification instrumentation; it's been, as I have said, the biggest single problem in the Type certification work in which I have been involved.

That's why one has to do proper tests. More than half the problem is eliminating the irrelevant possibilities. Guessing won't get you there, tho it can produce a starting point. The CAMit heads are made from a more thermally-stable alloy; that will help retain the valve seats - but the temperature range is always a factor.

Well, side-stepped, anyway. Checking the valve clearances is a useful guide to the health of the cylinder head, in any case.

Merv, I'm not arguing with you. In due course I'll get a 3300 into the test cell (may have to put a larger cooling fan in, to cool it) and then I may get some numbers that could help. None of the 3300 - engined aircraft is Type Certificated, and I have my doubts about its cooling system - but no hard data as yet. We'll get to it, but Ian's four-cylinder comes first.

Is your aircraft controlled by weight shift? If not, you need to do the inspection. I'm not sure, in regard to weight-shift types.

Andy, I agree wholeheartedly; I want full CHT / EGT data for all cylinders of any engine in my aircraft. I recommend that people fit a recording EMS system; it can be done as an add-on to the basic required instruments. In anything up to and including a J 160C, that would be a trivial approval exercise under CASR 21M. If you have one of the LSA models, I doubt Rod would object to it as an add-on, though he may well object to it as an instead-of. If you have one under -19, just do it. I don't want to get into a debate about "I shouldn't have to do that on a 'certified' aircraft", etc - I would just want such an instrument, by whatever means. I'd go find the means. The accident that started this thread could have a cause as remote as a distorted outlet on the air selector box, causing swirl in the induction duct and thus a mixture mal-distribution. Only a full set of temperature instruments will allow one to see something like that coming, and it may take years to develop.

Mind you, the oil flow is just one of a number of factors. Provided the guides were reamed correctly, so the valve did not go into its seat off-centre as initially assembled, then the excessive clearance is very likely due to the guide getting just a bit too hot. The guide will inevitably get a lot of heat from the valve, and it has to get rid of that heat in various ways - oil flow is one, thermal conduction into the cylinder head is another - if the engine had run at almost any time with too high a CHT, for whatever reason, that makes it harder for the valve guide to get rid of its heat, and that could start the process. Rocker geometry has a potential part to play, too. If one draws a "causation tree" it becomes obvious that what may seem simple is in fact very complex. One might learn something from a detailed examination of the other three heads.

Ha! Yes, that fits the pattern as Ian Bent explained it. Thanks. It's not necessarily a result of poor installation practice; excessive guide clearance can also result from insufficient oil flow in the spring pocket area inside the rocker box (which cools the exhaust valve guide).

It's been a requirement for damn nearly a century. People normally help one another by doing a dual inspection, without charging anything for it; anybody with a pilot licence can do one in GA. I've done hundreds of them. It's cheap insurance in my book; and likely culpable negligence if you don't.

OK, thanks - well, as anybody can see, that changes the relative occurrence rates, from 11.4 to (40/7) x (1407/1078) = 7.5 - i.e. over the period covered by the data, the Jabiru engines had 7.5 times the occurrence rate for the Rotax 912 series. I emphasize again, that the lack of accurate data and the unknowns such as what percentage of occurrences are reported, makes all these figures decidedly "rubbery"; however, they do, I think, give a tolerable relative notion of the rates as they existed over that period (2007 ~ 2012).

I suggest you look at the thread on Recreational Airfield Safety Operations, in General Discussion, especially page 3 and on. This subject was discussed at some length there

Here's my tabulation of Turb's data for the Rotax 912 series. It adds very little to his spreadsheet - but it's in Word, which may be more convenient for some readers. The info is just as wanting as for the Jab tabulation, and subject to the same limitations. As you can see, there are no obvious systemic fault causes; the faults are essentially random, and as Turbs pointed out, mainly indicative of faulty maintenance. The total number of occurrences was seven; the jab. occurrences for the same time interval numbered forty. Given an assumed percentage of the fleet double that of the Jabirus, and assuming a similar utilisation per aircraft (which may not be correct, but I have no better information), that would suggest that the Jabiru occurrence rate over this period was approximately (40/7) x 2 = 11.4 times that for the Rotax 912 series. The MTBF, for what that number may be worth, was not better for the Rotax than for the Jabiru. So there is indeed some point in what Ian Bent is doing. However, the failure probability for the Jab. engines is (as far as I can estimate, given the lack of data) still less than 1 in 1000 per flying hour; and three-quarters of the Jab rate is due to two identified causes, which are (in the CAMit mods, and for the through bolts, by the Jabiru Service Bulletins) arguably fixed (tho time will tell, of course) - so the real issue here is, I think, what the statistics do not show, which is the rate at which engines are being pulled out of service for major overhaul - i.e. the mean time between overhauls. The CAMit mods will, I think, make a considerable improvement in the TBO - but again, that will only be proven in the light of experience. So yes, the Rotax shows up as having a significant advantage over that time interval. Keeping the temperatures down in your Jab will pay dividends; and do pay attention to the manufacturer's Service Bulletins. 912 failure statistics.doc 912 failure statistics.doc 912 failure statistics.doc

See Post # 265 on this thread.

Andy, a lot of the non-reporting must happen because the operator concerned has a guilty conscience about it; whereas somebody who feels a genuine grievance is far more likely to report it. So to some degree the non-reporting may be offset by what amounts to an operator acknowledgement that it really wasn't an unprovoked occurrence. Not everybody is as honest about it as yourself. One can't put numbers on this, of course; but in an arm-waving kind of way, one can make at least as reasonable a stab at the dimensions of the problem, as the "Jab Bashers" have pretended to do. Try this for size: 1. The data Turbs supplied covers the period March 2007 to March 2012, i.e. five years. 2. What's a reasonable guesstimate for the average annual utilisation? Let's say, for the purposes of the discussion, 40 hours per year for a private owner, 250 hours per year for a training organisation. If ten percent of the active fleet are used for training, that works out at 61 hours per year, on average. 3. What percentage of the fleet is grounded for want of a serviceable engine, or whatever? Let's say half. That brings the average utilisation down to around 30 hours per year. 4. How many Jabirus in the fleet? For 2012 it was about 930; I don't know the annual growth rate , but would 800 be a fair guess for the average over that five year interval? If we accept all the above guesses, then the total utilisation would have been 30 x 800 x 5 = 120,000 hours. We have 40 reported "engine problem" occurrences; let's suppose there are an equal number of unreported occurrences of similar severity. That's 80 occurrences. The rate from those numbers is therefore 80 / 120,000 = 1 / 1500 - i.e. one per 1500 hours flying, or 1 in 1500 per flying hour. That's STILL less than the benchmark probability of one in a thousand per flying hour, for an aircraft approaching the end of its "safe life" for structural fatigue. Anybody can put their own numbers into this sort of estimate; but unless I'm way off the mark, that probability is nothing out of the ordinary. On these sort of numbers, I'd be surprised if CASA got excited about it. They may use the RAA bleat as an excuse to push their own agenda, of course, assuming they have one.

Well, that's a direct result of the CAO 95.55 exemption to CAR Part 4B

Ah, OK, found it - I'll do the same with it. However, before anybody gets too excited over this wealth of information, consider what we DON'T know from it: We don't know anything at all about the engines that did NOT fail. We do not know the total fleet hours or the fleet hours by engine type. We do not know anything about the TIS at which engines are being pulled for major maintenance, or what causes them to be pulled. Without this information, one can only make guesses at the failure rate per flying hour. Getting more specific, looking at the 17 through-bolt-related failures for Jabirus: We don't know how many of these would have been prevented by timely application of the Jabiru Service Bulletins on the subject. We don't know how many of them occurred in spite of timely application of the Jabiru Service Bulletins; or how many of those were caused by inept application of the Service Bulletins (which would show as failure shortly after the SB was implemented, presumably). None of this is really getting into the difficult forensic stuff, so it's all within the capability of the RAA, if they wanted to. There's nothing at all difficult about what Turbs has done, other than the tedium of picking it out of the magazines and putting it into the spreadsheet. Turbs has very effectively highlighted how easily RAA could at least get this far with even the inadequate data it has collected. GFA has an electronic means of airworthiness incident reporting in place; why not RAA? Once you have this, it's not rocket science to at least sort it into a data list, surely? And why not put that in the public domain?

Many thanks, Turbs. Here's my tabulation from the spreadsheet; it gives a MTBF (not counting engine that did not fail but were removed from service due to problems, or are still in service) of 588 hours. As for the previous version of the data, the two major faults were through bolts (42.5%) and valve/exhaust valve failure (30%). The remaining 27.5% of occurrences have insufficient data to assign any likely causes, except perhaps occurrence # 39 which looks like faulty maintenance. So this more complete data set does not change the overall picture at all, really. Fixing the through bolt issue and the exhaust valve/valve issue (both of which have complex causes) would reduce the failure rate as shown by this data set, to about one quarter. It's not clear whether the various Jabiru service bulletins have accomplished this fully for the through bolts; the data are insufficient to allow that. Issues such as loss of compression, causing engines to be withdrawn from service, are not covered by this analysis, as they do not appear in the data set. I don't think we can get any more data than this, from the available RAA statistics; it would be useful to compare the total engine failure rate to the total hours flown, but I do not have the latter information. Further analysis.doc Further analysis.doc Further analysis.doc

Insufficient data, I suspect. From what little I've seen so far, Ian Bent has already covered all those points, and he's into aspects to increase the TBO now, rather than the MTBF. Temperatures are the principal governing factor in that, especially insofar as they affect ring life. Ian's well-researched answers to the exhaust guide and the exhaust valve stem temperature are quite different to Bex's suggestion, by the way. I rather suggest that our speculation would prove to be way behind Ian's thinking. I have no idea whatsoever of what Rod Stiff may have up his sleeve; but Rod has been beset by "Wydonchas" from day one, which hasn't helped his receptivity to hypothetical input. I can't really blame him; everybody that came along had some idea as to how it could be done better; I'm amazed he managed not to scream. The through-bolt and exhaust valve issues are very reminiscent of similar problems that Lycoming had, with the early 0-320s and o-360s, about 30 years ago. If you remove them from the picture, as Ian Bent's mods seem very likely to do, and reduce the piston and cylinder wall temperatures, so the ring life is what it should be, and use a more thermally-stable alloy for the cylinder head, the statistics will almost certainly jump into 912 territory. The sooner we can formally test one of Ian's engines, the better.

It's simply the total hours to failure divided by the number of samples, for that 35 specimen example that Turbs provided. So it's definitely NOT an average life expectancy for Jabiru engine in general. I agree this is of limited value, but that's all I can get from that sample of data.

OK, here's my tabulation of it: From the total no. of hours, the average TIS was 602 hours. As far as I can determine, 14 of the occurrences were through-bolt related. What the data do not say, is how many of those occurred because the operator did not comply with the relevant Jabiru Service bulletins (there were about three of them) within the specified time; and how many represent failures that have occurred after all service bulletins were incorporated. Apart from the through bolts issue, the next principal fault was exhaust valve failure (12 cases). I haven't yet looked up the Jabiru Service Bulletins on that one. I suspect that some of these may have been due to a dud batch of valves from the supplier. If so, that's a quality assurance issue, not a design issue. There are two instances of cracked cylinder barrels; that's part of the through bolt problem; it occurs due to flexure of the cylinder base flange. That's a design issue, in my opinion. The three instances of seizure have me scratching my head; were these due to insufficient oil, or what? What I tentatively get from this, in conjunction with the previous "guesstimate" analysis (which I'll try to revise when more data comes from Turbs), is that the panic over the probability of an engine failure is over-played; the probability is likely to be about one in a thousand for a four-hour flight. That's one quarter of the probability of a structural fatigue failure for an aircraft that is nearing the end of its permissible fatigue life; the "safe life" of an aircraft is set by the probability of a failure reaching one in one thousand per flying hour. The real issue is the short average service life; and that's what Ian Bent has set out to address. Jab engine failures.doc Jab engine failures.doc Jab engine failures.doc

Ta - yes, even that much helps to narrow it down somewhat. I wouldn't expect more from RAA data. Aye, belike.

Ta. Firstly, as I have repeatedly stated, I'm NOT a member of RAA - I was so disgusted by what I learned as an expert witness for Carol Smith, that I let my membership lapse, and went back to the GFA. Secondly, I do not have the time to spare to wade through years of RAA reports; so I will be most appreciative to see your results.

If only it were that simple. I've spent a lifetime sorting out other people's problems; that's what professional engineers are for, mostly. It's meant doing things like spin testing, flutter testing and so on; and it's also involved a lot of plain slog. Haven't seen much glory in it, frankly. More like 99% perspiration plus a bit of sheer terror. It hasn't earned me the money to get an engine manufactured, though I have designed one. Being a CASA design signatory means you get to carry the can, mostly.