Using Cumulus with a rotatable AWS (on a windmill)

[mcrossley]

Excellent , glad to be of help.

[uncle_bob]

I now have this working fine on a test web page, thanks for the currcond tip.

Stuart

Would love to see the web site and maybe a few pics of the mill

[beteljuice]

Ditto re a working page you are happy to share (so we can rip it apart )

Re. your site mystery machine:

Great fun !

Is it the right way up ?
Is the square 'key' relevent ?
If that is an 'in' hole - where's the 'out' ?
Why can't I see the axle shaft for the machined roller ?
Is there more than one ?

It looks a little like a chaff cutter, but there are no blades on the flywheel nor a 'feeder'.

I'm enjoying this mechanical diversion

[Stuart2007]

Hi all

The 'proof of concept' is here : http://www.oldlandwindmill.co.uk/newsite/AWS/index4.htm. Rip away! but be a bit kind I am teaching myself javascript as I go...

As you will see from the source, I simply use a webtag to pick up currcond (with the file regenerated each upload from a master file using an entry in the Files tab of Config Internet settings.

Next up, I want to add to the webpage a one hour and 4 hour average windspeed. That matters for milling.

I could just use the <#wspddate> and <#nextwindindex> webtags to heap 3600 entries into the file, but that is a big overhead when I need max 240 entries (and the internet conneciton will be a 3g one). I could pre-extract those 240 entries, and preload them into the webpage before Cumulus adds its stuff to the webtags...but that would mean the average data was always 10 mins (in my planned case) out of date. So neither ideal

Any suggestions?

At the moment, the sensors are in my garden, just working too on the mill installation.

Many thanks

Stuart

PS Phase 2 is to replace the manual entry of "Mill heading" with an arduino reading off a magnetometer and writing to the currcond file. Phase 3 is mill specific data, such as how fast the sweeps are rotating (just like a bike speedometer - magnet and read switch). Phase 4 may be a torque index sensor, perhaps measuring the deflection in the millshaft using IR....how late is the sweeptip in arriving on a radius compared to the centre)

[beteljuice]

Excuse my ignorance, but is your 3G connection charged by number of connections, duration of connections or bandwidth (or a combination).

In the last case especially you should be thinking of php and / or Ajax scripted pages where you only need to upload data (especially if you do not really need Cumulus default 'console' graphs or moon image).

You can even contruct your own version of realtime.txt (see wiki / forum on php webtags and similar)

[Stuart2007]

Excuse my ignorance, but is your 3G connection charged by number of connections, duration of connections or bandwidth (or a combination).

You can even contruct your own version of realtime.txt (see wiki / forum on php webtags and similar)

Hi Beteljuice

Our cost relates to bandwidth; we are beneath the limit but without too much headroom. It irks me to upload 3600 units of data when 240 only are needed, when I could use the bandwidth for something else.

php/Ajax are not something I have tangled with, and lower priority at present than building sensors for compass bearing and mill operating parameters...ie Phase 5!

Stuart

PS Will pass your comments/questions re mystery machine on to the windmill gang

[Stuart2007]

Quick update...thanks to your help (and Mark in particular with his gauges) I am getting well on the way to a usable website , correcting for the varying "Mill Heading". See http://www.oldlandwindmill.co.uk/newsit ... s-ssMH.htm. I still need to fix the tooltips, and the windrose.

Mark if you are reading this, I am thinking I need to rotate the data in Windrosedata uploaded from realtimegauges.txt after it arrives and before windrose.js uses it. It seems to be a set of radius measures for each of the 16 cardinal points (although your own gauge seems to eliminate the three-letter points which Fine Offset under records...I guess it will be easier to stick to the same 16 angles, not try to rotate the points to the nearest degree.

Is this a sensible approach - my gauges.js is becoming quite customised, but I would like to try to keep to your standard windrose.js

Best wishes

Stuart

[mcrossley]

Without getting into some complex maths I think you are right the simple way to 'rotate' the rose is to 'rotate' the array elements. You will only have a 16 point resolution, but that is possibly "close enough". I checked the RGraph API just in case you could specify a start angle for the data - it hasn't got anything like that

However, thinking about it, that will only work if the mill heading has been constant over the wind rose period won't it? If the mill heading changes then old directional data is invalid isn't it?

You need a mechanical linkage/chain from the mill to the wind direction sensor base to rotate it in the opposite direction to the mill so it always points true North! All your problems solved

I have dropped the number of rose points from 16 to 8 in Cumulus because of the vagaries of the FO kit, but that would make things even worse for your resolution.