[conspire] Watering automation: an edge case cautionary tale

[Paul Zander]

Since OpenSprinkler is open source, you could, in theory modify the software so the green house gets watered on a fixed schedule independent of rain.
In my experience, projects like that get put on the round-to-it-list.  On the other hand, if I have such a problem with a closed source project, I tend to get annoyed until I find a F.O.S.S. that I can use.  Then my blood pressure relaxes and I add another item on the list of projects I might get to someday...

      From: Rick Moen <rick at linuxmafia.com>
 To: conspire at linuxmafia.com 
 Sent: Monday, March 6, 2017 6:53 PM
 Subject: [conspire] Watering automation: an edge case cautionary tale
   
I think it was around 2014 that the state started talking about
requiring residents to hit 'water budget' targets, and hard enforcement
started in June 2015.  Each household was required to reduce water each
month going forward by 32% compared to amounts they used in calendar
year 2013 (the 'base year') subject to a floor allocation of at least 6
CCF (centum cubic feet; 1 CCF = 748 gal).  Starting Aug. 2016 with the 
easing of the drought, these were changed to 'conservation targets',
though budgeting will return if trends reverse.

I was pretty worried when this happened, because I have 1/3 acre with
many fruit trees and plants.  Worse, over the 9 years since we moved
here, I'd not yet set up a real watering system.  Instead, there was a
crazy-quilt of soaker hoses and no automation.  Soaker hoses are better
than nothing, but just barely:  They apply water in the same Zip Code as
the trees and plants.  And, of course, if you forget to turn off the
faucet and let it run overnight, oops.

June, July, and August 2015, we exceeded assigned water budgets -- 
but I was frantically yanking soaker hoses (anyone need some?) and
installing a drip irrigation system.  And then, I'm guessing it must
have been right at the beginning of 2016, I also completed the primary
system by installing solenoid-driven water switches so that the watering
could be run by an automated timer.  Even though I had a perfectly
adequate 'dumb' Rainbird controller, I replaced it with OpenSprinkler
(https://opensprinkler.com/), an Arduino-based replacement.  (They also
have a very cool Raspberry Pi version.)  OpenSprinkler keeps track of
the weather and changes watering times accordingly, and can do other
nice tricks, and can be programmed & monitored from its built-in Web
server.  From Jan. 2016 on, water usage has gone _way_ down, especially
during our long Mediterranean summer.[1]

This past summer, I also connected in a water sensor, a gadget you mount
on the side of the roof to send the controller a signal if it's rained
recently (so it can decide not to water for a few days).

Result:  This winter, our water usage has fallen down to just above zero.
The February bill reflected a meter read of 1 CCF, and the March bill
that just posted showed a meter read of _0 CCF_.  We continue to be way
under usage targets, but the last two months are our lowest usage ever.

Yay, automation.  Yay, water sensors.


Automation's your friend, right?  In computing, it's practically a
mantra:  Script anything you can.  Automate everything you can.

And water automation has been a lifesaver around here, let me tell you
-- but let me tell you about a weird edge case:


One of the most recent places I extended drip irrigation to is to plants
overwintered inside my about 12' x 8' kit-built greenhouse.  A couple 
of large lemongrass plants (native to the tropics) spend winters there, 
and this winter some hot pepper plants are surviving there, too.  All
roof and side vents and the greenhouse door are closed up tight, to
protect the plants against frosty nights.

But there's a hidden problem -- that arrived when I added the water
sensor.  Can you spot it?


Before I added the water sensor, my OpenSprinkler controller would open
the solenoids to feed water to the greenhouse (among other places) twice
a day like clockwork.  Perfect.  Exactly what the frost-sensitive plants
in there need.

But imagine during a wet winter like the present one, a two or three
week period of rain almost every day.  Then, the sensor would suppress
all watering -- which is fine for plants outside the greenhouse that
would get their share of the rain, but would be nicely automated death 
for plant inside the tightly sealed greenhouse.


Automation's great -- except where it does clockwork automated failure.


(I'm about to go outside and water the plants in the greenhouse,
manually.)



[1] Historical household usage in CCF billed:
    2014  2015  2016
Apr    3    10    2    
May    6    7    2
Jun    16    15    5
Jul    18    15    6
Aug    24    13    7
Sep    9    5    7
Oct    13    7    6


_______________________________________________
conspire mailing list
conspire at linuxmafia.com
http://linuxmafia.com/mailman/listinfo/conspire


   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://linuxmafia.com/pipermail/conspire/attachments/20170307/ee307c3a/attachment.html>