Author Topic: Scientific Flatwater Board Testing 101  (Read 27465 times)

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Scientific Flatwater Board Testing 101
« on: June 21, 2016, 01:28:55 AM »
Hi guys,

Here's something for you to chew over when you're on your lunchbreak. A roadmap for a bit of scientific testing for race boards.

https://www.supboardermag.com/2016/06/21/board-testing-101-how-to-find-out-what-board-is-fastest-for-yo/
« Last Edit: June 21, 2016, 01:31:27 AM by ukgm »

yugi

  • Cortez Bank Status
  • *****
  • Posts: 2228
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #1 on: June 21, 2016, 05:18:00 AM »
What are you using to count strokes?

I have a Naish Maliko and an NSP Ocean, both in 26 wide on hand. I'd be curious to see the difference between the planing hull design and the pure round hull soft edges hull design. They feel very different.

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #2 on: June 21, 2016, 07:22:27 AM »
What are you using to count strokes?

I have a Naish Maliko and an NSP Ocean, both in 26 wide on hand. I'd be curious to see the difference between the planing hull design and the pure round hull soft edges hull design. They feel very different.

I used a SUP Speedcoach 2 for this study to provide the stroke data. However, I've also used a Vaaka cadence sensor in the past successfully too.

Area 10

  • Cortez Bank Status
  • *****
  • Posts: 4057
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #3 on: June 21, 2016, 08:04:15 AM »
Very interesting. Have you performed any formal statistical testing on your data? It would be interesting to see the differences in variance between the two boards.

The main flaw in your experimental design of course is that you were not blind to which boards you were on. So this could be a placebo-type effect: you believed that the flat water board would be faster, and so this changed some subtle aspect of your technique or effort, and that is what caused the effect. In medicine, placebo- type effects can often be more substantial than actual treatment effects. It is particularly problematic in this respect that you did not also collect physiological measurements (eg. HR) and perhaps also physical ones using e.g. a strain gauge on the paddle.

But I'm sure that as a scientist you will already appreciate this. Do you have an idea of how you will deal with this criticism from the referees when you submit it to a scientific journal as you mentioned?

Of course, what we all want to know now is whether the all-round board is faster in the conditions it was designed (at least partly) for. Measuring performance differences in the sea across a variety of conditions is a substantial scientific challenge, and you may need other ways of doing the comparisons, perhaps.

It's great to see someone taking the time and effort to do this. It's extremely difficult to get scientifically convincing results from board comparisons - which is why so few people have tried.

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #4 on: June 21, 2016, 08:42:49 AM »
1) Very interesting. Have you performed any formal statistical testing on your data? It would be interesting to see the differences in variance between the two boards.

2) The main flaw in your experimental design of course is that you were not blind to which boards you were on. So this could be a placebo-type effect: you believed that the flat water board would be faster, and so this changed some subtle aspect of your technique or effort, and that is what caused the effect. In medicine, placebo- type effects can often be more substantial than actual treatment effects. It is particularly problematic in this respect that you did not also collect physiological measurements (eg. HR) and perhaps also physical ones using e.g. a strain gauge on the paddle.

But I'm sure that as a scientist you will already appreciate this. Do you have an idea of how you will deal with this criticism from the referees when you submit it to a scientific journal as you mentioned?

3) Of course, what we all want to know now is whether the all-round board is faster in the conditions it was designed (at least partly) for. Measuring performance differences in the sea across a variety of conditions is a substantial scientific challenge, and you may need other ways of doing the comparisons, perhaps.

4) It's great to see someone taking the time and effort to do this. It's extremely difficult to get scientifically convincing results from board comparisons - which is why so few people have tried.

Great reply - thank you. I've numbered your excellent points above and will attempt to address them here:

1) Yes I have. The article I wrote was for a general audience so I left out a considerable amount of statistical material that only would have confused (or frankly just bored) the hell out of most people. I did have extra stats though (such as the coefficient of variation to show how stable the tests and the boards were), t tests (to show that both boards were significantly different from each other) and a few other bits and pieces. My paper will obviously include them but this articles intent was to encourage people that a. kit choice was important and b. you can do it yourself if you make the time. Granted, a lot of this hangs on how important you believe stroke count is but HR is no good in shorter test intervals like this but if you lengthen test intervals out (and Larry Cain has done this) you might be able to use it but then you won't have a decent sample size as it will be too fatiguing. I'm hoping the margin of errors I included allay many of these concerns but for what it is worth, my coefficient of variation (defined as Standard deviation divided by the mean X 100) was no greater than 0-3% on everything (test repeatability and the boards speed and stroke data) bar the slowdown tests (which did push up towards 10 - still not bad but conditions really matter for those).

2) Yes - placebo is a huge issue generally. However, my paper will be focusing on the robustness of the tests themselves rather than worrying about the performance of boards A and B specifically. I would make the case that the statistical analysis and results are strong enough that it would be hard to be affected by placebo - especially the hydrodynamic slow down tests whereby you can't really influence the results anyway as you've just put Archimedes and Newton in the driving seat.

3) Between you, me and the internet, I suspect if you tried these tests in the more typical conditions of a flatwater race (which would have seen a little more chop than I tested in), the gaps between the two boards would come down. Plus I also suspect (and there is research in other sports like K1 to support it) that bodyweight is going to play a huge part in craft performance. It all adds to that people need to test themselves using boards and conditions that suit them. I would also recommend being aware of the test limitations. For example, the slow down tests (test 3 in the article) are fast and easy but used on their own do not consider the nuances of your stroke and if your board is slightly under volume, will perform badly in that one whereas it might do really well in others when the speed (and therefore the resulting lift) might well let it shine.

4) It is hard but not impossible. It's something I'd need to work upto and figure out the right test interval and environment to do so. The margin of errors will increase and its quite likely that as board design becomes a race of diminishing returns, the gains will be swamped by the margin of error. Bearing in mind I've only been paddling a couple of years, I was surprised how statistically repeatably I was able to perform in these trials to start with. I see this as a starting point for other ideas i have in mind.
« Last Edit: June 21, 2016, 08:55:52 AM by ukgm »

FloridaWindSUP

  • Sunset Status
  • ****
  • Posts: 335
  • Marine Biologist, Riviera Paddlesurf Ambassador
    • View Profile
    • James' Blog
Re: Scientific Flatwater Board Testing 101
« Reply #5 on: June 21, 2016, 09:55:36 AM »
Your "Test 2" is really interesting. I wonder, though, about the assumption that stroke rate is an adequate proxy for power applied. When you paddle at different speeds to generate the data for that graph, how do you ensure that you're varying only stroke frequency and not force per stroke? I know from technique drills and such that I can change speed independently of stroke rate by altering the reach and power of each stroke.
14x23 Riviera RP
14x27.25 Fanatic Falcon
11'8 Exocet WindSUP
10'4 Angulo WindSUP
...and a bunch of windsurf stuff

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #6 on: June 21, 2016, 11:46:44 AM »
Your "Test 2" is really interesting. I wonder, though, about the assumption that stroke rate is an adequate proxy for power applied. When you paddle at different speeds to generate the data for that graph, how do you ensure that you're varying only stroke frequency and not force per stroke? I know from technique drills and such that I can change speed independently of stroke rate by altering the reach and power of each stroke.

Yep, you're absolutely right - stroke rate is only part of the power equation and doesn't account for the torque you're producing. Its good practise to use at least two metrics when judging performance (with one acting as the variable and the other as a governor). Power output would be the gold standard but until that arrives its a choice between heart rate or stroke rate. I actually use both in my everyday training but as far as the decision to use it for equipment testing purely came down to stroke rate being the only one you could make statistically viable (to get the right number of runs and not require a longer interval) and secondly that there was published research already (granted in other paddle sports but not SUP) that had reliably indexed stroke rate to work intensity. However, I would openly agree that cadence has limitations when the force data is not able to be recorded. It's also possible that if you tested a board that was beyond your paddling capabilities, this could also sour the results. With all of this in mind, I'd combine tests 1 or 2 but then reinforce that with test 3 (which ignores stroke rate entirely) to likely give you a greater chance you know what is faster.

It's all a work in progress and something I can build on. Thanks for your feedback - its very useful.
« Last Edit: June 21, 2016, 11:52:48 AM by ukgm »

SUPJorge

  • Sunset Status
  • ****
  • Posts: 471
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #7 on: June 21, 2016, 01:33:13 PM »
UK- I'm not a scientist so I'll need to think a bit harder about this to really appreciate everything that is actually going on and all the underlying decisions and processes but it is very interesting stuff. One thing I do takeaway is that the robustness of the design of Test 2 and its results quiets many of my worries about the inability to hold constant all the variables involved in paddling. It'll be looking forward to see what happens once you can measure power at the blade and hold it constant for testing purposes.

An unrelated question: A large part of the game in SUP is to limit the deceleration between strokes. Have you ever had similar concerns about cycling on the road, or would you think that the issue so minor there as to be practically meaningless?
14' SIC Bullet V2 - 9'1" Naish Hokua X32 LE

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #8 on: June 21, 2016, 02:01:04 PM »
UK- I'm not a scientist so I'll need to think a bit harder about this to really appreciate everything that is actually going on and all the underlying decisions and processes but it is very interesting stuff. One thing I do takeaway is that the robustness of the design of Test 2 and its results quiets many of my worries about the inability to hold constant all the variables involved in paddling. It'll be looking forward to see what happens once you can measure power at the blade and hold it constant for testing purposes.

An unrelated question: A large part of the game in SUP is to limit the deceleration between strokes. Have you ever had similar concerns about cycling on the road, or would you think that the issue so minor there as to be practically meaningless?



Yep, this reads as three different tests you can do but actually I did have an ulterior motive when planning it.

Test 1= ascertained the repeatability of my paddling technique at race pace and the repeatability of the boards overall performance. If the results were poor, tests 2 and 3 would be pointless as my underlying scientific method would not be robust and this whole idea would be a failure.This test was a precursor to my preferred strategy of testing boards of tests 2 and 3.

Test 2= to derive a simple method of finding the typical performance of the board across a broader range of speed. The line of best fit is basically the best fit of line to the data. However, for the scientists out there, a polynomial line fitted my series dots with a R squared value of 0.99 (i.e. a near perfect fit).

Test 3 = to test a boards resistant drag by removing the variability of the paddlers technique entirely.

Test 3 won't work in chop but test 1 and 2 might. I do need to try that though and I fully expect the margin of error to increase greatly.

As for cycling, air is nearly 800 times thinner than water and a cadence nearly twice as much as paddling so pedal stroke I would suggest its less relevant. That said, I've personally seen increased resistance to my power output quality when riding on rougher road surfaces when compared to smooth ones (and I time trial at ~85-90rpm and at ~370w for 20 minutes) so its entirely possible there is a small effect. However, I would have to defer to better technical paddlers than I to advise you when it comes to paddling.
« Last Edit: June 21, 2016, 02:11:08 PM by ukgm »

PonoBill

  • Cortez Bank Status
  • *****
  • Posts: 25864
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #9 on: June 21, 2016, 03:01:35 PM »
Very interesting. A couple of observations off the top of my head. The drag test doesn't seem to specifically include optimum trim. 1. Weight position has a big effect on drag, as does total weight. You might get more conclusive results by stacking a representative weight on each board and winching it through the water with an attached strain gauge, and then moving the weights around to find the optimal trim (lowest force). You could still use coastdown for drag estimate, or just calibrate the strain gauge. You don't need much of a winch, any speed controlled motor attached to a drum reeling light line would do.

You could approximate this by doing your coastdown test in a number of positions. A relatively insensitive test for a factor with a likely narrow range of results, but it might be adequate.

2. I think you really need a power meter. You can get at the power applied by measuring acceleration and total mass, but F=ma assumes we isolate acceleration, and you don't really have a way to do that. Too many other factors acting on the board.  You might consider video as a measure of stroke rate vs. length vs. acceleration. You can certainly measure speed in a video with a calibrated float by doing stop action of the strokes. You might be able to get a relative power number from shaft flex, but it's probably going to be inaccurate as a measure.

Anyway, I'm just spitballing. Your test raises a lot of interesting possibilities. Congratulations.

Foote 10'4X34", SIC 17.5 V1 hollow and an EPS one in Hood River. Foote 9'0" x 31", L41 8'8", 18' Speedboard, etc. etc.

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #10 on: June 21, 2016, 03:31:20 PM »
Very interesting. A couple of observations off the top of my head. The drag test doesn't seem to specifically include optimum trim. 1. Weight position has a big effect on drag, as does total weight. You might get more conclusive results by stacking a representative weight on each board and winching it through the water with an attached strain gauge, and then moving the weights around to find the optimal trim (lowest force). You could still use coastdown for drag estimate, or just calibrate the strain gauge. You don't need much of a winch, any speed controlled motor attached to a drum reeling light line would do.

You could approximate this by doing your coastdown test in a number of positions. A relatively insensitive test for a factor with a likely narrow range of results, but it might be adequate.

2. I think you really need a power meter. You can get at the power applied by measuring acceleration and total mass, but F=ma assumes we isolate acceleration, and you don't really have a way to do that. Too many other factors acting on the board.  You might consider video as a measure of stroke rate vs. length vs. acceleration. You can certainly measure speed in a video with a calibrated float by doing stop action of the strokes. You might be able to get a relative power number from shaft flex, but it's probably going to be inaccurate as a measure.

Anyway, I'm just spitballing. Your test raises a lot of interesting possibilities. Congratulations.
Thanks for your useful comments.

1) Yep, optimal trim wasn't pursued as that wasn't the primary aim of the study. This experiment was primarily about establishing whether a scientific method could be achieved or not and to provide some workable examples (notice I kept the board brands and models anonymous too). I did standardise trim between runs by marking foot position on each board to ensure trim repeatability but its entirely feasible that trim could be optimised from where my feet were actually placed and this would change the results. These tests could certainly help do that in the future. Your suggestion is an interesting idea though. I've also seen published research whereby they towed craft behind a speedboat (but off beam) using a cable with strain gauges to measure the resistant force. That could achieve the same as what you're suggesting.

2) Agreed. A power meter is the holy grail (and the best solution) and I wait for the long overdue one to arrive on the market. I'll be the first one there to get my money down when it arrives. In the meantime, we've got to work with what we have available and it was important to me to suggest methods that anyone could do with relatively low-fi equipment.
« Last Edit: June 21, 2016, 03:36:38 PM by ukgm »

Eagle

  • Cortez Bank Status
  • *****
  • Posts: 2426
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #11 on: June 21, 2016, 07:55:24 PM »
I think your board test was quite reasonable - and your results seem valid.  We simply paddle around a couple of islands in a triangle course for about 5 miles - measuring 3 split times.  Works close enough for us and we can replicate the results with consistency.  We do account for wind waves and current when analyzing our data.

What we found in these simple timed tests was that - as soon as ripples turned into solid chop - our wider touring board takes over and becomes faster than our race board.  Many tests were done years ago - and were subsequently compared to more recent GPS data and HRM data as well.  The current results reflect a similar pattern to before.

So the fastest board for us was the board that we could paddle consistently using full power.  In our ocean tests - the conditions had the greatest impact - and this affected fatigue and balance the most.  In one instance on a 9 mile paddle on our race board - I had to resort to paddling on my knees - conditions deteriorated so much.  After 7 miles of punishment - just could not stay standing anymore.  My legs were toast from all the balancing.  On a subsequent day was out on our touring board paddling the same course - and was able to easily power along in even worse conditions. 

So for flat water our race board was faster - and for rough water our wider touring board was faster.  Pretty much nowadays we can tell with just a demo how easy a board is to balance on - and how efficiently it paddles.  Probably if we tested these 2 boards under your test parameters - we would arrive at similar plots and conclusions as you.
Fast is FUN!   8)
Dominator - Touring Pintail - Bullet V2 - M14 - AS23

yugi

  • Cortez Bank Status
  • *****
  • Posts: 2228
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #12 on: June 21, 2016, 09:37:55 PM »
Nice. 2 different methods come out with smooth consistent results corresponding to the outcome we’d expect (flatwater board optimised for flatwater). Seems to work correctly.

I thought something but didn’t answer yesterday, I waited and re-read the test and re-thought it. I think you are making one assumption here. That your blade is slipping. More slippage equates more effort to move craft forward.

You just happened to post this at the same time as the raging discussion (slower is faster thread) on not slipping the blade and on how, if a blade is being well set, cadence should match boat speed. Faster cadence should be on the faster craft. Ideally planted a blade would stroke the same StrokeLength*AmountOfStroks on either craft and difference should be the amount of sweat produced (effort).

You’re basically measuring how much the tires are slipping in wet grass as a definition of efficiency of the craft. Isn’t it?

Basically you are measuring how efficient a paddler is in one craft against another. I wonder if the paddle slippage assumption is the right one.

Would that Motionize sensor measure more stuff about a stroke to tell us enough more from which we could derive effort?

Would a guy like Tituan, who definitely has minimum slippage, doing this test come up with same results? You should ask him to double check it for you.

Effort would be the right thing to measure once a power meter is available.






PonoBill

  • Cortez Bank Status
  • *****
  • Posts: 25864
    • View Profile
Re: Scientific Flatwater Board Testing 101
« Reply #13 on: June 21, 2016, 10:22:20 PM »
I use a straight section of the Hood River, chosen mostly for bank access (laziness trups accuracy) but observing the river grass in this section convinces me that it's pretty straight. No wafting, no turbulence. Then again, I wanted to be convinced. I use a windsufing mast and a strain gauge from a cheap scale. I originally used a digital multimeter to read the strain guage, but decided the scale guts made more sense, the LED and the electronics are mounted on the mast with extension wire to the strain guage. I think I have $29 invested in the whole apparatus. Suprisingly accurate once I added a 3 to 1 force multiplying lever.

I thought about building a drag winch this year for lowball tank testing. We have two perfect places to do it in Hood River, the best is probably Nichols, but the pond inside the Hook would work well too.

I get that you were more interested in establishing a valid methodology than in the nuts and bolts, but I'm biased towards trying to get the tools working well so the data doesn't demand a lot of tweaking. I understand the basics of statiistical validation but I see examples in everyday life where valid numbers deliver invalid conclusions.  The 50 percent improvent that means 1.5 people had a good result instead of 1.
Foote 10'4X34", SIC 17.5 V1 hollow and an EPS one in Hood River. Foote 9'0" x 31", L41 8'8", 18' Speedboard, etc. etc.

ukgm

  • Teahupoo Status
  • ******
  • Posts: 1255
    • View Profile
    • Email
Re: Scientific Flatwater Board Testing 101
« Reply #14 on: June 22, 2016, 12:01:54 AM »
1) I think your board test was quite reasonable - and your results seem valid.  We simply paddle around a couple of islands in a triangle course for about 5 miles - measuring 3 split times.  Works close enough for us and we can replicate the results with consistency.  We do account for wind waves and current when analyzing our data.

2) What we found in these simple timed tests was that - as soon as ripples turned into solid chop - our wider touring board takes over and becomes faster than our race board.  Many tests were done years ago - and were subsequently compared to more recent GPS data and HRM data as well.  The current results reflect a similar pattern to before.

3) So the fastest board for us was the board that we could paddle consistently using full power.  In our ocean tests - the conditions had the greatest impact - and this affected fatigue and balance the most.  In one instance on a 9 mile paddle on our race board - I had to resort to paddling on my knees - conditions deteriorated so much.  After 7 miles of punishment - just could not stay standing anymore.  My legs were toast from all the balancing.  On a subsequent day was out on our touring board paddling the same course - and was able to easily power along in even worse conditions. 

Thanks for this.

1) From my point of view, I operate from the standpoint of statistical robustness as the bottom line so I personally believe that provided there are enough runs and that people are aware of their margin of error or data variability (an aspect sadly not reported or recorded in pretty much all previous attempts I've read), you could likely average out the chaoticness of a rougher water paddle if you do enough runs. I personally haven't tried it though but at the end of the day, most people won't be as concerned to obtain the same information as I would hold myself to (being an academic and someone who has to publish for a living)

2) Yep, I agree with you. I'd personally like to know what level of conditions would allow my all water board to triumph over my flatwater. That wasn't my aim here but may well form the fundamental part of what I'm going to race on for 2017.

3) I think this is an important point. Because so much of my tests hung on paddle stroke, if the board becomes to narrow to be effective, the results won't be any good (although this would likely be indicated in both the data's variability and margin of error). In that case, try test 3 as that might make any answer more suitable and helpful as your technique and balance improves.
« Last Edit: June 22, 2016, 12:37:54 AM by ukgm »

 


* Recent Posts

SimplePortal 2.3.7 © 2008-2024, SimplePortal