Boat Building Forum

Find advice on all aspects of building your own kayak, canoe or any lightweight boats

Re: Challenge vs Test
By:Nick Schade
Date: 8/13/1998, 5:36 pm

George's challenge is fair. He sets the rules and as long as they are clear and explicit they are fair. But it making it fair does not make it a meaningful test. In order for a test to be meaningful, it needs to be repeatable. Shooting with a 12 gauge through a kayak is not fair, but if what you are trying to get is 12-gauge-shot-gun-proof it is a good test. You can make 1000 panels and shoot each one and be sure they are being subjected to an equivalent event. Some of the panels will get a different degree of damage, but you will see a predictable trend that will give you an accurate ability to predict what will happen to the 1001st panel.

1000 paddlers paddling the same course will subject their kayaks to 1000 different series of events. If all the paddlers are different, they will have different skills and see distinctly different conditions. If the course is run 1000 times by the same paddlers, they will learn the course but they are still subject to different conditions because nature is by definition chaotic. Unfortunately, nothing can be predicted about the 1001st run.

George is trying to pass his challenge off as a test which will demonstrate in a meaningful way "strength". It is not a test, it is only a challenge. The fairness is irrelevent to whether it is meaningful.

> Sailing is different than kayaking. Here the losing boat is often
> destroyed and the winning boat is not in sale-albe condition. The
> stakes provide reasonable incentive and compensation.

> As I understand most of the posts here over the last year or
> so:

> Using a mathematical model is not acceptable, using test samples
> is not acceptable, and using real boats is not acceptable.

Mathematical models are good after validation with physical testing. Your models may well be very good, but they have yet to be validated so who knows?

> Please note what I offered Nick: no money or boats change hands,
> I pay my travel expenses, he picks the course, we have a crowd partial
> to him perhaps. Except for pride, its a no lose situation. I got the
> expected response, "its not fair".

Please quote me in context where I said: "its not fair"

> If you can come up with a better (than my challenge) test of
> kayak durability and a reward system that compensates for the risk,
> I will fund it. But lets try to keep the entry cost down. Had I won
> every foot or bicycle race I participated in I would not have paid
> my training expenses I expect that happened in sailing also.

I still intend to do some lab testing of different panel constructions. I would be happy to include some of George's panels in the test. I would not be embarrassed if they turned out stronger than the standard layup and would publish the results.

While a panel test will not fully address the real-world survivability of a boat, it would fill in some of the unknowns and can be used as a basis for validating any mathematical models.

Panels should have the following characteristics: - 11 samples of each construction method and wood sample, - Samples of each construction method should be made with wood from the same board to minimize variations in the wood. - Density of the wood in each panel determined before glassing to detect gross differences in the wood. - Samples from different boards should be included to demonstrate effects of variation in woods. - Same resin batch used to make all samples. (if possible) - Same roll of fiberglass where applicable. - Density of each panel measured before destructive testing. - All samples painted and randomly numbered to make it harder to recognize differences in layup. Numbers and characteristics of panels will be logged seperately from testing data.

Testing should be 3-point bending (strength) test or hammer drop (impact) test (or both but this requires twice as many panels). One sample of each type of panel should be held out from destructive testing for future reference. Unfiberglassed samples of each wood should also be saved.

Results would be the average (mean) of 10 tests. Standard deviation would also be interesting. Improvement on the test: more panels.

Messages In This Thread

Re: Challenge vs Test
Nick Schade -- 8/13/1998, 5:36 pm
Re: Challenge vs Test
Mark Kanzler -- 8/14/1998, 3:25 pm