Accident Causes vs. Likelihood of Fatalities

[rwanttaja]

Guy on another forum asked a good question...what types of accidents are more likely to result in fatalities?


BTW, "Other Mechanical" refers to mechanical failures other than the engine.

Ron Wanttaja

[gbrasch]

Great data, someone put a lot of work into this, but no real surprises....

[Frank Giger]

Fantastic work! Ron, you are a gem - and your contributions to the Experimental community are seriously under appreciated.

Knocking the dust off my statistical analyst hat, let's see if I parse this right.

At first blush, this is number of wrecks that result in fatality by cause divided by total number of wrecks, which might not give the full picture.

For instance a builder error might be fatal 80% of the time when it happens, but that it happens very rarely. Sort of like more surfers die of getting knocked on the head by their surf board and drown, but if a shark gets at them - which is incredibly rare - it's usually game over.

However, I'm more inclined to take the chart at face value and throw out the second order probabilities, as this chart rings true to certified aircraft as well. It's a "these are the most common things that will kill you" chart.

We know from long experienced there are two things that kill pilots and passengers more than anything else - flying into IFR conditions and the infamous stall on base to final (if that's what "maneuvering at low altitude" is referring to). If it means buzzing, well, Langewiesche was bemoaning this type of behavior back in 1944 and it seems we're not any wiser than our predecessors.

What I take personally striking is that in my fast taxi and first three flights in the Nieuport I can check off four* on the list that happened to me, though without injury or serious damage to the aircraft (two bent wheels and a two tail wheels). This is why I completely re-wrote my test flight program in both scale (baby steps compared to the FAA pamphlet) and time. Only until I fully understand what has happened (both good and bad) on one flight do I move onto the next....which means a visitor that comes a long way to be ground support (including flying up the previous day) may only get to witness a 10 minute flight.

Knowing the most common pitfalls of homebuilts and pilots all gets thrown in the risk assessment hopper, and I'm sure I'm not alone in bearing all of this stuff in mind.

* Here we go:

Fast taxi - brake mis-adjusted and bungees too loose; bent main wheel, tail wheel chewed and ruined at resulting 360 degree spin in place.
First flight - directional control on takeoff (minor), el cheapo replacement tail wheel fell apart on an otherwise good landing (it lasted until the turn off from the runway). I attributed it to being too slow with the throttle and dancing around the lift off speed.
Second flight - directional control on takeoff (major). This was because I was too aggressive with the throttle and didn't have time to apply left rudder. I added springs and some slack to the tail wheel cables to remedy this problem.
Third flight - directional control on takeoff (minor)/other mechanical, small loop on landing (another bent wheel), caused by a bit of foam I used in the combing around the cockpit falling off and binding intermittently with the tail wheel cable and interfering with left rudder authority. I was lucky here - it only bound the cable twice (lucky) but at the two worst times (take off and at the end of the landing).
Fourth flight - Golden. Now I can emerge from the pattern (but stay close) and do some engine and larger fight envelope testing.

I won't include my one rather bouncy landing, as I never lost directional control and it's par for the course!

[edit]

After reading that it sounds like my plane is a death trap with an incompetent fool behind the stick, but it's really not that bad, and I like to think that my skills as a pilot kept things from going much, much worse.

[rwanttaja]

Fantastic work! Ron, you are a gem - and your contributions to the Experimental community are seriously under appreciated.

Knocking the dust off my statistical analyst hat, let's see if I parse this right.

At first blush, this is number of wrecks that result in fatality by cause divided by total number of wrecks, which might not give the full picture.

For instance a builder error might be fatal 80% of the time when it happens, but that it happens very rarely. Sort of like more surfers die of getting knocked on the head by their surf board and drown, but if a shark gets at them - which is incredibly rare - it's usually game over.

Thank you for the compliment, Frank, and thanks for looking at it with a critical eye. You're right, the number of instances does make a difference. I've updated the original chart with the total number of instances of each accident type. You can scroll up to see the new version

There are indeed several causes where the actual number of accidents were so low that the fatality percentage...whether high or low...would be a bit deceptive. Thinking about it, I should probably have just left them off. Fortunately, none of them were of the particularly high fatality rate. But the Taxi Accident, Turbulence/Winds, and Engine Controller each had no more than two fatal accidents (one for the Engine Controller).

And yes, I was curious about how two fatalities stemmed from Taxi Accidents. They actually do illustrate one of the problems with my analyses, where none of the 55 cause categories I use pin down the accident, exactly. One case was an RV on landing striking another that was taxiing on the runway, and the other was the loss of control of a new design during high-speed taxi testing.

Ron Wanttaja

[Frank Giger]

Thanks for appreciating that I'm trying to understand and make the information useful rather than trying to snipe your hard work! It's just about a decade doing statistical ordering and then trend analysis from it makes me want to look under the cover and see what the engine looks like....all too often the answer is that there isn't a trend or concrete answer, just information that may be useful in explaining the here and now to some degree.

I'm very surprised at the number of midair collisions - I'd of thought it would be lower. I'd think that the odds of somebody getting seriously injured or killed in a midair would be pretty high; fortunately it's on the low risk end of my matrix - runway incursion (which could be interpreted as a "taxi accident") is one of my secret bugaboos, though.

I am curious at what meets the "inadequate pre-flight" criteria. Weight and balance problems, like putting too much baggage in a compartment, maybe, or loading past gross weight allowed?

[rwanttaja]

I'm very surprised at the number of midair collisions - I'd of thought it would be lower. I'd think that the odds of somebody getting seriously injured or killed in a midair would be pretty high; fortunately it's on the low risk end of my matrix - runway incursion (which could be interpreted as a "taxi accident") is one of my secret bugaboos, though.

From reading your saga, perhaps runway EXcursions might be more possible. :-)

I am curious at what meets the "inadequate pre-flight" criteria. Weight and balance problems, like putting too much baggage in a compartment, maybe, or loading past gross weight allowed?

Weight and balance issues are scored separately. While I track them the occurrences, in the summaries I generate, they're combined with others into a "Misc" column. Looking at the results, though, they should have been included as an individual item. 14 cases over 17 years, with ten fatalities.

That's one of the things I enjoy about getting quizzed about results and processes....it makes my product better.

Inadequate pre-flight are those cases where something was not proper on the airplane that should have been detected during pre-flight. Oil and gas caps missing, canopies not secured, tow bars/control locks still in place, etc.

Ron Wanttaja