ANALYSIS OF WING SEPARATION AND MID-AIR BREAKUP IN LIGHT TRAINING AIRCRAFT
Matthew HIRABAYASHI
Department of Ophthalmology, University of Missouri Columbia, Columbia, USA
Received 9 November 2020; accepted 5 April 2021
Abstract.
Despite increasing discussions concerning the recently published wing spar airworthiness directive (AD) that af- fects many training aircraft and several current ADs for wing struts, there remains limited objective literature on incidents of wing separation or mid-air breakup. This paper attempts to report and analyse instances of wing separation and mid-air breakup of light training aircraft. A careful review of the United States National Transportation Safety Board (NTSB) aircraft accident database revealed that wing separations were more likely occur as mid-air breakup in PA28s than 172s/182s (OR: 3.06, 95 % CI: 1.3682 to 6.8536, p = .008). Additionally, wing separations were less likely to occur as mid-air breakups in the strutted 172s/182s than 177s/210s that don’t have a wing strut (OR: 0.11, 95 % CI: 0.04 to 0.29, p = <.001). This implies that non-strutted wing designs may be more susceptability to mid-air breakup than the strutted design of similar aircraft.
Snoopy wrote:
Department of Ophthalmology,
Serioulsy ??
It does read like a medical paper. I’m not sure about the generalisability of the conclusion, given that he looks at so few types, but it seems possible.
The component loads in strutted wing structures are so low they are often over strength in relation to flight loads, just to make them able to withstand ground handling and potential hangar rash.
A friend of mine sized his Wittman Tailwind wing struts for flight loads, knowing that nobody but him would be pushing it around on tbe ground, and looking out the window while flying along at 160 kts they look very, very skinny.
Serioulsy ??
Why not. He might have had good eyes going through the NTSB database.
Nice paper but I don’t think the conclusion is somewhat more value add to what we already know. I find it hard to draw conclusions from 2 “insidious” wing seperation cases in a PA28 since the day the first PA28 rolled out of the factory. Has anyone seen anywhere Piper or any authority publish the results of all eddy current inspections done so far? That’d be quite interesting to see and could paint a clearer picture.
I was expecting a conclusion on C172 vs PA28 to involve mechanics & physics (structural analysis, sheer & load tensors, crack propagation, elasticity-plasticity, solid-fluid harmonics…)
Not rare events stats with 2 data points where one go to look into NTSB database to extrapolate Gaussian variables beyond the 0.01% quantiles !
By9468840 wrote:
Has anyone seen anywhere Piper or any authority publish the results of all eddy current inspections done so far?
This is based on hearsay, but it is actually not low single % figures, but somewhere between 5% and 10%, in some cases the NDT inspection failed on relatively low time examples not covered by the AD. It would be helpful if Piper did publish the actual failure rate, so that you are not relying on local engineers word of mouth/gossip.
As a consequence pretty well all PA28 have, or will have the NDT inspection to maintain their marketability.
Ibra wrote:
I was expecting a conclusion on C172 vs PA28 to involve mechanics & physics (structural analysis, sheer & load tensors, crack propagation, elasticity-plasticity, solid-fluid harmonics…)
I found it interesting to contemplate the possible value of a paper that didn’t.
In engineering, we work on the assumption that everything is more-or-less understandable. In biosciences, it’s much harder to work anything out from first principles. Even when you can, you don’t trust it until it’s been validated experimentally – and it’s much harder to do that rigorously because there are rightly lots of rules on how much you can experiment on people, and because people are far more diverse than rivets so you need larger sample sizes to reach firm conclusions. But what if engineering is actually more like the biosciences than we like to admit? We all know there are some unknown unknowns, but perhaps they are more numerous than we realise.
My understanding of aircraft design is that you have structures that are designed analytically, but padded with safety margins and design decisions that are perhaps based more on rules of thumb, and hopefully take into account the fact that people don’t always treat the structures in the manner for which they were designed (see Silvaire’s excellent point about some structures being sized in practice to resist hangar rash rather than structural loads). Wing roots with all their holes and bolts strike me as something that will only partially yield to analysis (particularly back when the Cessnas and Pipers in question were designed), and the safety margins are likely to be drawn as much from experience as theory.
No aircraft should fall apart in reasonable use, so if cantilevered wings are significantly more likely to do so then perhaps our rules of thumb are not as conservative as they should be. On the other hand, if we’d found that strutted wings were more likely to fail due to the end fasteners on the wing struts being prone to failure due to stress concentrations that weren’t fully accounted for, or moisture flowing to the bottom of the struts and causing corrosion – then post-hoc we would have said ‘sure, that makes sense’. Perhaps sometimes it makes sense to throw the theory aside and ask, ‘what is?’
My question would be how generalisable the findings are. If all the companies building planes are using similar rules of thumb to design their aircraft, then we might find that Eurostars and low wing Zenairs are more prone to wings breaking than C42s or high wing Zenairs. However, I would not jump to this conclusion because I suspect that there will be cultural and temporal differences in how wings are designed (i.e. Zenair might use different rules of thumb from the legacy Cessnas, partly because it’s a different company and partly because the designs are more modern, and materials and design philosophies have changed). Personally I’m happy to fly any individual design with a decent track record.
Yes I agree, in certified aircraft, the design is done based on high level rules and principals and likely to lead to similar guarantees across all CS23 airframes, anything else is ‘just noise’ but some, including FAA, would even go to attribute even the tiniest of noise to specific type based one single rare events !
On the modelling side, things like robustness to corrosion and abuse are not well understood in Mechanical Engineering even with ‘rule of thumbs’ and plenty of margins on loads…if something ‘wrong in the data’ of Piper wing vs Cessna wing, it will have to do with water? rough surfaces? history of flight? or type of pilots?
PS: while ago when working on jet engines, ‘technically’ one could predict 3D crack propagation in titanium alloys fans for given jet engine based on it’s unique recorded usage and maintenance history, obviously, you can’t use that (non-approved) model to certify an engine with 300 pax on it, even if the maths add up, you have to stick ‘rules of thumb’ with load of margins and throw in ‘regular inspections’ 
