software to analyse boat roll data for a flopper stopper

Chriscraft

New member
Joined
18 Apr 2021
Messages
5
Visit site
Hi

My vintage chris craft rolls badly at anchor, so I am testing some flopper stoppers to see if they help.

Ive used data from a smartphone app to record rolling etc with and without the flopper stopper (see attached) but have no idea how to analyse it.

Are there any free or cheap ways to analyse the data and see what difference the flopper stoppers make? A sample of the data is attached.

Any tips much appreciated

thanks

david m
 

Attachments

  • flopper data.txt
    65.8 KB · Views: 29

Birdseye

Well-known member
Joined
9 Mar 2003
Messages
28,092
Location
s e wales
Visit site
Looks to me as if it is recording movement in 3 dimensions viz x,y,z and likelt the last column is some sort of G force ie acceleration. In terms of what you feel, the accelaeration is significant. But if you are worried about having to keep hold os a G&T on the table ( or a bottle of milk as I experienced) then it would be x or y assuming z is vertical.

Suggest you records with and without, one after the other and then average the x,y, and G force data for each session ( ie with and without) and then compare.

That said it is a classic case of playing with numbers when what matters is whether you feel a noticeable improvement. Still, playing with numbers is fun.

P.S. I wonder if I can do an analysis of how often I hit thw wrong key, and then try to correlate it with age. :cry:
 

Zing

Well-known member
Joined
7 Feb 2014
Messages
7,794
Visit site
There are times when an elaborate mathematical analysis is the wrong thing to do. If the difference is not bloody obvious, then you’ve failed.

You are not a university professor are you? ?
 

Neeves

Well-known member
Joined
20 Nov 2011
Messages
12,074
Location
Sydney, Australia.
Visit site
Down load SCraMP - its free and developed to measure the 3 dimensional movement of vessels. It might only be available for Apple devices, don't know.

The data is graphical, which is going to be easier to compare visually.

I cannot tell you how to optimise the data collection speed - as I have not used it, in anger, yet.

There are a couple of similar Apps, Accelerometer and VibSenson that do the same sort of thing - but I'm sticking with SCraMP - and I'd be interested in an analysis you might make.

I might agree that if you cannot tell the difference between with and without the flopper stopper than you are wasting your time - but assuming you are sensitive (or you wife is) then a later development might find SCraMP useful to fine tune the flopper stoppers, location size, perforated or not etc.


In terms of correlation of miss keying and age

I often read what I have scripted and autocorrect can turn something quite intelligent into gibberish - so I don't necessarily think age and miskeying will necessarily give an accurate answer :)

Jonathan
 

Chriscraft

New member
Joined
18 Apr 2021
Messages
5
Visit site
Thanks Jonathan and birdseye

indeed I want to be able to justify the clutter of the flippers to key parties, and also confirm the best arrangement, depth etc for using them. My first test run (where I also collected the data) was positive, most notably at the end when I certainly noticed the change when I took them out of the water! I’ll take a look at a scramp and keep looking for a way to analyse the data. The problem is that all roll measurements eventually average out to the same number no matter how strong or weak

David
 

Neeves

Well-known member
Joined
20 Nov 2011
Messages
12,074
Location
Sydney, Australia.
Visit site
David

I recall,

SCraMP was developed, I think by a female (its not relevant but might indicate I might have remembered correctly, or not :( ) and she has a description of why she developed the App and I think how it has been used. I must have googled the app and this led me to the script. It might be worth your time to investigate - some of the background.

I cannot see why, if you noticed a difference, the roll data would average to the same result - unless you are looking at pitching, rather than rolling?

I do understand, or appreciate, why you might want to do the work - its an interesting project - which can only been done now - because of a 'tool' within current hardware (that we have all paid for) - that so far has not seen much use.

Good luck,

Jonathan
 

Mistroma

Well-known member
Joined
22 Feb 2009
Messages
4,882
Location
Greece briefly then Scotland for rest of summer
www.mistroma.com
It will be interesting to see what the analysis shows. A couple of minutes with Excel looks promising, assuming this was a run where the gear was deployed part way through recording. Significant reduction in roll on the X-axis, smaller reduction in Z-axis. It looks as if the period is greatly increased on the Y-axis but recording time is not long enough to see what's happening.

I'd say that it is worth continuing to record data and analyse it as the results look sensible. Of course it could be complete rubbish if the example data did not show you deploying the gear around time 1185.:D I'm assuming that the times aren't seconds.
 

Chriscraft

New member
Joined
18 Apr 2021
Messages
5
Visit site
Jonathan/Mistroma

- re- averaging to zero, I had thought it would be like a pendulum, all the swings each way would (in the data) even themselves out. I've now been in touch with the SCraMP developer (your recollection was correct) who gave some good tips.

- re the test data, unfortunately that came from another app (sensor suite) which pumped at measurements at 400 per second, and measured 'g-force' rather than roll, and the segment i posted is in fact just 5 seconds worth. And unfortunately I'm not a university professor so the data is doing my head in

I'll be out again on the weekend, with looking at particularly how to extend the floppers out a metre or so, which seems to be the key. my initial attempts were bodgy to say the least, flopper.JPG
 

Neeves

Well-known member
Joined
20 Nov 2011
Messages
12,074
Location
Sydney, Australia.
Visit site
David

Don't, or never, apologise for bodgyness. You have to start somewhere and develop as you learn. If you knew all the answers you would not be asking the questions here. If you read, or at least look at the pictures, in the link below you will see how many prototypes I made for a very vaguely similar problem.

I had noted that many of the apps (and there are quite a few) developed so much data as to be useless - because you could never process it.

But you actually have 2 problems, best way to use the App to achieve the answers you need

and

developing the flopper stoppers

Keep posting - I'm interested from the same, or similar, point of view (anchoring). I may have some answers to questions that might cross your mind.

I posted this:

A Snubber & Hook for all Occasions | Practical Sailor

a week or so ago and updating myself on this thread of yours made me wonder if a bit of elasticity in your suspension ropes might smooth things out for you (though how you get useful elasticity into your cordage - I'm not sure (those rubber dog bone things?). I also wondered if you have seesawing, pitching, as an issue as well as roll.

Take care, stay safe

Jonathan
 

Mistroma

Well-known member
Joined
22 Feb 2009
Messages
4,882
Location
Greece briefly then Scotland for rest of summer
www.mistroma.com
The changes in period and amplitude looked promising if you were only recording a few readings a second. At 400 readings per second it was just vibration, rather than rolling at anchor. It looks as if you were recording someone using a hammer drill if it was a 5 second run. :D :D

Perhaps you will see much larger changes over a much longer run with fewer readings per second.
 

Neeves

Well-known member
Joined
20 Nov 2011
Messages
12,074
Location
Sydney, Australia.
Visit site
I confess that I have not tried to analyse your data - it was simply too daunting!

If the data represents a hammer drill - or something like - were you running you engine? I cannot think of anything else to cause significant vibration.

You said you contacted Leigh - did she mention how to reduce the amount of data collected (to something that is manageable) by reducing the readings to 1 per second, or something those of us not employed by a University might get out heads round :)

Jonathan
 

GHA

Well-known member
Joined
26 Jun 2013
Messages
12,104
Location
Hopefully somewhere warm
Visit site
4 lines of python is what you want :) I'll have play later with a few more lines to make a better plot.
If you really want to go to town then a rapsberry pi & 9DOF cheap sensor gives you accel, compass & gyro data.

1619197002152.png
And a bigger one..
Y6aLgjG.png


((Should this be in a racing forum?)
 
Last edited:

Mistroma

Well-known member
Joined
22 Feb 2009
Messages
4,882
Location
Greece briefly then Scotland for rest of summer
www.mistroma.com
4 lines of python is what you want :) I'll have play later with a few more lines to make a better plot.
If you really want to go to town then a rapsberry pi & 9DOF cheap sensor gives you accel, compass & gyro data.

View attachment 114036
And a bigger one..
Y6aLgjG.png


((Should this be in a racing forum?)
I plotted the same thing but expanded X & Y which made a reduction in amplitude really stand out. Of course it turned out to be meaningless as OP was sampling at 400 Hz and it was only a 5 second snapshot rather than the several minutes I was expecting.

My approach was fairly basic to say the least (I renamed the TXT file as CSV and opened it in Excel :D:D). I like the idea of a cheap sensor and some python.
 
  • Like
Reactions: GHA

GHA

Well-known member
Joined
26 Jun 2013
Messages
12,104
Location
Hopefully somewhere warm
Visit site
And with a moving average to get rid of a bit of the noise.., the bokeh library for python can do some really funky plotting quite easily :cool: , this is just a few lines of code>

It3CaR6.png


Runs in Thonny - Thonny, Python IDE for beginners
I swapped the tabs for commas in sublime as it was quicker that doing a search for the pandas command.

Python:
import pandas as pd
import matplotlib.pyplot as plt
filename ='flopper_data.txt'
df = pd.read_csv (filename)
df['gFx-sma'] = df.iloc[:,1].rolling(window=100).mean()
df['gFy-sma'] = df.iloc[:,2].rolling(window=100).mean()
df['gFz-sma'] = df.iloc[:,3].rolling(window=100).mean()
df['TgF-sma'] = df.iloc[:,4].rolling(window=100).mean()
# df.plot(x="time", y=["gFx", "gFy",  "TgF"])
df.plot(x="time", y=["gFx-sma", "gFy-sma",  "gFz-sma", "TgF-sma"])
plt.show()
 

GHA

Well-known member
Joined
26 Jun 2013
Messages
12,104
Location
Hopefully somewhere warm
Visit site
. I like the idea of a cheap sensor and some python.
Going the Pi route you don't even need any python, it's all done already. Sensor is one of these though bit of research would be wise, lots of duds around apparently. IIC/I2C MPU-9255 MPU9255 Three-Axis Gyroscope Accelerator Magnetometer Sensor Module GY-9255 3-5V Power Board: Amazon.com: Industrial & Scientific
All the calibration routines exist already, it's recommended for an open source autopilot
start [pypilot]
Calibrates continually . Bit overkill for a little bit of playing around but all this stuff costs little and the huge amount of easily accessible accurate data for little cost or power really is something else :cool:

PS - Chriscraft - welcome to the forums! And ignore the "if you can't notice the difference..." comments, humans are useless at analyzing sense data, completely rubbish, riddled with bias - measure it and graph it, then the real world will open up to you :)
 

Mistroma

Well-known member
Joined
22 Feb 2009
Messages
4,882
Location
Greece briefly then Scotland for rest of summer
www.mistroma.com
Going the Pi route you don't even need any python, it's all done already. Sensor is one of these though bit of research would be wise, lots of duds around apparently. IIC/I2C MPU-9255 MPU9255 Three-Axis Gyroscope Accelerator Magnetometer Sensor Module GY-9255 3-5V Power Board: Amazon.com: Industrial & Scientific
All the calibration routines exist already, it's recommended for an open source autopilot
start [pypilot]
Calibrates continually . Bit overkill for a little bit of playing around but all this stuff costs little and the huge amount of easily accessible accurate data for little cost or power really is something else :cool:

PS - Chriscraft - welcome to the forums! And ignore the "if you can't notice the difference..." comments, humans are useless at analyzing sense data, completely rubbish, riddled with bias - measure it and graph it, then the real world will open up to you :)
Thanks for the link, I also found one on eBay. I'm considering a re-write of the multi-display I built to replace my dead Scheiber unit and could add one to the 2 ADS1115s on the same I2C bus. I found a better serial LCD and can simplify my original setup (2 Arduino UNO units, one for display and 2nd I/O). A serial LCD with clearer display and single ESP32 with Bluetooth/Wifi seems worth considering. I could link it to my ESP32 based chain counter as a secondary display.

I can't actually think of a use for it just now but will keep a note in case I need these features for some future expansion.
 

Chriscraft

New member
Joined
18 Apr 2021
Messages
5
Visit site
Hi All

I went out on the weekend to two spots in Sydney Harbour. One is Snails Bay , a relatively quiet and pleasant spot, the other Greenwich Point, is also nice but right at the choke point of the Harbour, lots of boat traffic, especially ferries, going by so a lot of roll.

I tried the flopper stoppers as follows:
- Snails Bay: no FS, FS hung over the side on the cleat ('inboard'), on one side (stbd)
- Greenwich Point: no FS, FS both outboard. Unfortunately there is a large gap between the no FS and FS measurements, as getting both out in the bumpy conditions was a bit challenging!

I have linked to the data, SCraMP proved a bit trickier to save/up load, so some is from another program , Physics Toolbox Suite and the key file are:
- 2021-04-2511.21.35 Snails without and with floppers
- 2021-04-2513.56.12 greenwich no floppers
- SCraMP1619326840 both out

The SCraMP data is about 1/50th of a second, the Toolbox about every 10th

You will see a few random bumps where I knocked the phone....

Also, the SCraMP data is in UNIX time which I had never heard of until now, but suspect the python experts here know all about.

In terms of my gut feel, the floppers seemed to help somewhat, but I think were taking too long to 'open', I suspect because the flaps are too large.

Really looking forward to anyone's analysis!


cheers

david m

Flopper Data 25-7-2021 - Copy
 
Top