Equations for calculating true wind with attitude?

[StuM]

Originally Posted by Juho

If you have gigabytes of data, and you have marked some segments of it as "under constant conditions", then you could automatically (or manually) adjust the parameters until your model gives you constant TWS and TWA for all points of sail,

Where in the world do you find constant TWS/TWA? Sea level atmosphere is a turbulent place.

[Juho]

Originally Posted by StuM

Where in the world do you find constant TWS/TWA? Sea level atmosphere is a turbulent place.

Yes, that is a difficult task. On the other hand all you need is three minutes of relatively constant TWS/TWA. One minute for getting averaged measurements with the first settings, then one minute for making changes, and then one minute for measuring the second settings. If you have many enough such (relatively constant) sets, you might get statistically meaningful results when trying to adjust the model. A wind tunnel with a wave machine could give us more accurate results, but also that approach has its own problems.

[mglonnro]

Originally Posted by StuM

Where in the world do you find constant TWS/TWA? Sea level atmosphere is a turbulent place.

Yes, just looking at the physical wind instrument (or windex) it's funny how turbulent it is. That's another thing I was thinking about: most instrument displays dampen/average it, but maybe there might be some interesting patterns in the non-downsampled data?

[mglonnro]

Originally Posted by Juho

Yes, that is a difficult task. On the other hand all you need is three minutes of relatively constant TWS/TWA. One minute for getting averaged measurements with the first settings, then one minute for making changes, and then one minute for measuring the second settings. If you have many enough such (relatively constant) sets, you might get statistically meaningful results when trying to adjust the model. A wind tunnel with a wave machine could give us more accurate results, but also that approach has its own problems.

If you have large enough a wind tunnel then it could provide some enjoyable sailing moments as well

[mglonnro]

Originally Posted by teppokurki

It would be absolutely wonderful to implement something like this, table based or next gen cooler than everything AI self adjusting, as part of the Signal K server. The server has a plugin system that allows simple installation and provides configuration management out of the box.

So instead of everybody doing their own bespoke systems from scratch people could collaborate on the algorithms and make more progress.

There is already a Derived Data plugin for straightforward stuff like true wind, true heading etc, but for advanced stuff like this a separate plugin would work very well.

I can help out with getting up to speed, let me know if you are interested.

https://github.com/SignalK/signalk-server-node/
https://github.com/SignalK/signalk-derived-data

Disclosure: I am one of the primary authors of Signal K spec and the server

This is cool. I checked SignalK briefly a few months ago, but will revisit!

BR,
Mikael

[wackerb]

I've mostly single handed half way around the world and have a technical background. For the last few years I've searched for the holy grail of accurate polars. I even switched all my instruments to B&G because a salesman told me that they compensated for mast head movement, not realizing that only the very expensive H5000 actualy did it.

However in the process of making manual logs to get approximate data I found out that in addition to the wind speed and direction you are trying to improve, actual performance is much more dependent on sea state, which sails are flying, reefing, and sail trim. What I'd like to see, instead of very sensitive and overly complex math is just better smoothing algorithms so it is not like looking through binoculars while bumping down a 4WD road.

We all become experts at mental averaging of jittery displays. The vendors seem to think that frequency of updates is the object, but the inertia of a 10-20 ton boat just doesn't make that an appropriate goal.

One of the most useful B&G instrument displays I have left after switching to Raymarine is the time plot of basic wind and boat speed and direction measurements. The graphs are still jittery, but you can see relationships better. More specifically, the effect on boat speed(s) as sail and wind changes settle in. On a heavy boat the useful sampling rate is on the order of minutes instead of split seconds.

That being said, the average smart phone has the electronics to compensate for mast head wind sensor movement and using that would be a good start to more stable data.

[jsanton]

I’m a retired engineer and data geek. But I try to offset that with KISS thinking and the joy of intimate mental connection with the environment. Don’t know if that’s helpful...
I too would like to build measured performance polars. I think the best approach Is to find a windy place with minimal seas. Collect the data there.

I haven’t heard anyone mention the significant delta in wind speed at mast height vs boom height.

I have a PhD statistician friend who races. He thinks that the brain can be trained to sense the environment and steer for max vmg without thought. All by reflex. Right brain vs left brain. He’s a very good racer so I think he’s onto something valuable. That provides a powerful model for tactical trimming. But we really need effective polars for route finding.

[thinwater]

Another minor point. The masthead instruments don't read the exact wind direction or speed, they read the wind coming off the sails and are materially influenced by this flow, something that becomes more and more obvious the deeper you sail. I've seen the masthead indicator off by 30 degrees. Direction and speed are also affected by heel, as the wind begins to flow more vertically.


I wouldn't get too hung up on the calculations. It's an engineering number, at best, good for comparisons.

[Alan Mighty]

Originally Posted by wackerb

One of the most useful B&G instrument displays I have left after switching to Raymarine is the time plot of basic wind and boat speed and direction measurements. The graphs are still jittery, but you can see relationships better. More specifically, the effect on boat speed(s) as sail and wind changes settle in. On a heavy boat the useful sampling rate is on the order of minutes instead of split seconds.

Ooh! Most interesting. Do you have a graphic or photo to share so I see what you mean?

[jmh2002]

Originally Posted by thinwater

Another minor point. The masthead instruments don't read the exact wind direction or speed, they read the wind coming off the sails and are materially influenced by this flow, something that becomes more and more obvious the deeper you sail. I've seen the masthead indicator off by 30 degrees. Direction and speed are also affected by heel, as the wind begins to flow more vertically.

Yeah, on race boats the wind instruments tend be mounted even higher up and forward on an elevated wand. Some use a tall custom made wand but for example purposes here is what B&G offers (yes the prices will cause a heart attack: $4250 for the old 1.8m model and not even sure if that is the carbon version) B&G WS700 https://www.bandg.com/bg/type/instru.../wind-sensors/

Originally Posted by B&G website

What gives the WS700 series its edge is the sensor design coupled with the high modulus carbon spar that elevates the sensor above any sail-plan disturbance. Clear air means clean data, free of any turbulence that might lead to biased measurements. The spar is available in 4 lengths, between 0.8 – 1.8 m. (edit: about 6 feet)

If the extra mast height isn't an issue for bridges, etc, then for cruising I would just make a taller wand to attach your normal wind instruments to using stainless or aluminium because this stuff does get damaged occasionally and no need to keep paying manufacturer prices for replacements.

Careful of the tube wall thickness though as even an errant seagull can cause issues and the old lightweight aluminium racing ones could sometimes be damaged during a broach with the spinnaker.

[wackerb]

Originally Posted by jsanton

But we really need effective polars for route finding.

That's what got me started wanting good polars. Transportation algorithms to optimize route based on winds and polars are amazing, but all that precision is based on forecasts which, in my experience, are highly inaccurate. When the actual winds are off by 180 degrees and 10 kt, an accurate polar is superfluous.

I like Predictwind routes because they give me an idea of boat position through time and multiple grib snapshots, but to actually input their optimized tracks into my GPS would be a waste of time even if I had accurate polars, could specify sail plans, and there were accurate predictions of sea state.

The weather models are much better than I could guess on my own, but they're only SWAGs based on data that is 6-12 hours old plus up to 6 hours until the next update. They simply don't support the precision of calculated route.

[wackerb]

Originally Posted by Alan Mighty

Ooh! Most interesting. Do you have a graphic or photo to share so I see what you mean?

Sorry, I'm at anchor now so all I can get is some empty graphs, but I think you can imagine what they would look like filled in. I like an hour's view. Easy to see if the 15 minutes have passed before reacting to changes.

[Montanan]

Recommended reference regarding relative accuracy of wind instruments associated with heel angle, motion, effect of sails, twist of the mast, positioning of the sensors, angle to true wind, etc.

https://junkrigassociation.org/Resou...er4pp87-96.pdf

[mglonnro]

Originally Posted by Montanan

Recommended reference regarding relative accuracy of wind instruments associated with heel angle, motion, effect of sails, twist of the mast, positioning of the sensors, angle to true wind, etc.

https://junkrigassociation.org/Resou...er4pp87-96.pdf

Cool!

[mglonnro]

Someone talked earlier about how turbulent the wind is. It's really fascinating

Here's a short video I captured a few minutes ago with live wind and some history visualisations on top of that. This is much less turbulent than what I saw yesterday. I guess there might be a correlation between wind speed and variance (?)

https://www.facebook.com/thenakedsai...dy8J&__tn__=-R

(The full page with live data is here: https://community.nakedsailor.blog/yacht/charlotte/data -- still very much a work in progress.)