Anecdotal Evidence

Anecdotal Evidence

by Tony Abbey, FE Training

My Uncle started as a River Pilot on the English Humber River; a treacherous place with strong currents and constantly changing sandbanks. He then served in the Merchant Navy and was torpedoed twice. Post war he captained a survey and salvage ship for the Royal Air Force. This included marking out and monitoring offshore target ranges. 

A grim task was salvaging aircraft that crashed during bombing practice. Sometimes this occurred within sight of land. Sonar and other search methods were not so advanced as today, so to jumpstart the search he would visit witnesses onshore; either from the official list or just from casual inquiry. With a few good witnesses he could establish a cross bearing plot. It didn’t always work - but on several occasions he pinpointed the location and sped up the salvage process. Investigating anecdotal evidence is a useful method when trying to confirm the validity of an analysis. It is always an implicit part of the analyst’s job to seek out any information that can help support the simulation. 

Formal test correlation methods are an essential part of many projects. However, if unavailable, anecdotal evidence becomes important. Coming across information by chance helped me on many projects. 

In the early 1980’s I worked on an anti-tank projectile. The front section had to send a signal to the rear section on impact. Slender legs connected front to rear and had to survive long enough to allow the signal to be transmitted. Understanding leg failure mode was crucial. These were early days for high speed impact explicit analysis and nobody quite believed the new methodology.The project manager mentioned that bits and pieces were collected from the range and photographed. To our delight, several of the photos included the legs and confirmed the failure mode predictions. In a more peaceful application I reviewed a highway crash barrier test and analysis project. The barrier type (popular in the UK and Europe) used a corrugated steel strip supported at regular intervals by vertical stanchions. During vehicle impact the stanchions failed progressively and the strip formed a bow-like shape. This redirected the vehicle into the traffic lane in a, hopefully, controllable manner. The California equivalent is a massive block of concrete - which redirects, but controllability is suspect!

Key system elements were; integrity of the steel strip and rupture strength of the stanchions. Stanchions too strong meant a very hard rebound. Weak stanchions saw too much unzipping and endangerment of the opposite carriageway. The failure mode of the stanchions was uncertain. This was another explicit analysis with impact duration of the order of 5 to 10 seconds. Even today that is demanding for computational resource. The stanchion was heavily idealized using results from local detailed models. 

High-speed film of actual vehicle tests was available. However it was difficult to identify the stanchion failure mode. Talking to the project team revealed that the broken stanchions were in a pile adjacent to the test facility. It took effort to correlate specimens and tests; however we were able to confirm that the simulation failure modes matched the test evidence. This was an important step in validating the analysis method and its use in further design work with different vehicle size and impact orientation.

So if you are looking for that extra piece of evidence to confirm your analysis results, a bit of digging around (literally) may turn up something very useful. 

Until next time,