seeing as the bottoms of those beds are even with the top of the fish tank, I would simply plumb them to drain back to the fish tank. Just plumb the drain through the side of the fish tank. Since the water level in the fish tank will rarely be all the way up to the rim it will be fine. That will eliminate the need for a second pump and separate sump tank. (pumps on float switches will fail to activate on occasion, that is just the way it is. Something could tangle up the switch or wedge it so it either stays on or off and that causes an issue.)
I would only go with a second pump if there is no way possible you can set up to do a CHOP CHOP 2 or CHIFT PIST system.
I like your planned use of the 3 way valve. I advise installing it over the tank or over a bed because I have had trouble with at least one brand of the valve starting to leak at the handle after a couple weeks in service and changing out the valve only fixed it for a few weeks before the new one started leaking. If it does this over the tank, not a big deal but if it leaks out of the system you are loosing too much water.
as usual, your knowledge and experience show in your words tclynx.
the figure is a simplified representation of a larger process control system. i have not indicated the sensors nor the feedback loops.
a lot of aquaponics technology is tied to the garden design, it is hard to change the functionality without changing something within the garden components. with process control, the functionality is tied to how the garden responds and is not determined by the physical garden design. eg. because of feedback i can flood my gardens at 75% for 5.5 minutes if i choose to as compared to flood cycles being decided by pump flow rates, standpipes, loops or float switches. to make it behave like a continuous flow dwc system is as easy as clicking. i can combine the two to create new functionality by adding a new shell script and updating initGardenUtil - yes the controllers run linux.
since the garden is controlled using logic through a single point - the valve actuator, a million things could go wrong. so feedback is necessary, meaning the process outputs at each point have to be sampled using sensors and used to modify the input control signals. this is further modified using pid control.
in this particular case, the garden is at least 2 hours away from me and i assist manage this remotely. the sump tank is actually a 300gal fish tank with gold fish and koi (and looks like the main tank) as it is inside the greenhouse with the growbeds. the main tank is outside some ways away under a tree with bluegill. the flow switch sensors in the sump return monitor this to mitigate the potential problems you have indicated below - the pumps will stop and a notification alert sent should the float switch on the sump fail or tank overflow.
i have found that everything seems to work ok when i look with my eyes, as aquaponics gardens respond slowly. in this case you have to be there for a full hour to ensure everything is working just for that hour. when i have added sensors to gardens then i have been amazed at how often they don't function as expected. siphons stick - in the bottom, middle or top if they are not monitored often. float switches stop to work if you don't have enough water.
they weakest point of this design unfortunately is the indexing valves they are finicky. you could never tell with your eyes how often they fail to index. in a commercial environment, i could never touch them without sensors to monitor them and a smart controller to respond to problems that arise.
then again, my friends say i have supervisory control and data acquisition (scada) issues.
the following images show some random shots showing indexing issues for 3 growbeds. there is a sensor embedded in the gravel of each growbed. the graphs are for water levels in the growbeds - 100% is full with water flooding through the overflow, 0% is for empty ( 1" or water left in gb). the sensors need calibrating so they come out of the negative. the red negative spikes are noise that i will suppress shortly using code. the graphs alternate when everything is working correctly.
we are using pressure relieve valves to reduce these issues.
It is true that anything mechanical, electrical or sensor related in aquaponics is going to require some tweeking/attention. And no matter how perfect you get it there will occasionally be hiccups.
The trick is making sure those hiccups don't actually cause catastrophe. You can pick and choose what methods you like best since you are the one doing it and part of the fun can be tinkering to get a method or process working to your liking. I've personally had problems with sensors and float switches and the electronics failing in some cases so you have to choose for yourself.
I like the idea of being able to log the data and am interested in methods to do that but I need them to be affordable or I have to wait to save up to be able to do them since starting up a new farm is costly and the budget is getting tight now.
it takes a lot to put together a decent controller tclynx. reliable high quality electronics, sensors and actuators are pricey, good engineering and software development is expensive.
i will open source many parts of this and have already started a series on howto build a smart aquaponics garden in my blog
i have much respects for you tclynx ... please give me your specs, i'll build and give you a smart aquaponics controller prototype for your evaluation.
Hum, well the homemade controllers I've been slapping together lately are really just to operate the automated pool valves so there isn't much too them and so far they have done fine if I install them in a weather proof box. (very simple, pic axe chip timer really set to a 10 minute one way 10 minute the other sort of schedule operating a spdt relay that feeds the power to one phase of the actuator or the other.)
I might like to have something that I could sell and people would have an easier way to change the timing (besides pulling out the chip and hooking it up on a programing board to their computer but the truth is that isn't so hard and most of the re-programing methods I've seen for such a simple set up are too cumbersome or they escalate the cost to the point where people might prefer to replace pump impellers a couple times a year because they are turning the pump on/off instead.)
So anyway, I'm interested in the sensors and data logging more than the actual control. I would really like to have a temperature data logger that can track more than two or three temperatures and can log it to a computer so I don't have to remember to do write the min/max down every day or every 4 days. I've also has a terrible time with thermometers failing. I say I'd like this but at this point in time, I don't get to spend much money on anything for a while. Setting up the farm has used up much of the budget and it will be a while for me to recoup.
I might also be interested in data logging the levels in grow beds just to make indexing valve testing a bit easier (less time consuming) but at this point in time it isn't important enough for me to spend money on.
do you mean a digital cyclic timer with lcd/led display, buttons for control, switched power (12v or 110v)?. or would something like bluetooth work better for configuration. further upgrades could include sensors to monitor flow in pipes or gb levels to ensure that it is indexing correctly. then there is data logging - local or remotely, and visualization of data.
the v1.1 smart controller will do multiple temperature readings as it uses 1-wire water proof sensors probes. the v1.1 controller consists of two main separate parts, the switched power strip with build in relays and the microcontrollers. this is shown below - i think you are talking about the sensor and micro end of it. the micro is also two computers, the arduino with the smart aquaponics shield for interfacing sensors (and relays) and a linux single board computer to handle garden high level logic and communications. it is possible further decouple the sbc from the aduino/shield making it completely local.
Well currently the only power switching I'm doing on my systems is 24V AC for the diversion valve actuators. At the moment I don't have any repeat cycle timers installed at the new farm.
And I haven't had time to really consider integrating or adding sensors. So far in my experience with the indexing valve media bed systems, if the valve occasionally gets stuck in one zone for an extra cycle, the plants and fish haven't been affected. It is only a worry to me if it is getting stuck all day in one position and I can usually tell if that has happened by walking through the system and listening to what beds are draining (if only one bed is draining and all the others not even trickling anymore, then I know the valve has been stuck on the one bed too long.) Media beds are pretty forgiving so far in my experience.
Now if I were using an indexing valve to feed towers, that might be a different story since if they go without water for an extra hour or more the plants could really start suffering on a hot day and going without water for a whole day could kill plants and dry out the filter media. So I figure I'm better off simply alternating the flow between banks of towers using the diversion valve. In my climate I doubt I would want to leave a tower without irrigation for more than perhaps a half hour during the summer heat so I just set all the diversion valves to alternate on a 10 minute by 10 minute schedule (which also works for the media beds as two banks of 6 zone indexing valves of a diversion valve will water each bed once every two hours providing nothing is getting stuck.)
Now please keep posting about the process controls you have and how they work because some day I might be able to afford them and I'm sure there are others out there who do want all that stuff now.