Nearing Grow Bed and DWC Completion

Status update here:

The grow bed box and DWC tanks are nearing completion.  I have affixed the 3/4" plywood to the bottoms of each box, after having pre-drilled and countersunk holes along the edges of the plywood.  I suppose it was not strictly necessary, but then the plywood was also not going to be tied to anything else and so would have been floating free.  I can always disassemble the boxes, even after the liner is installed, but I'm figuring that they will stay together until they literally fall apart.

The two boxes, upside-down

In the (unfortunately messy) image above, the two boxes are seen, the larger DWC tank sitting on top of the GB, and both upside-down.  On top of the DWC is the 2" foam board for the rafts, already cut to width but not yet to length.  I have given it 1/2" of clearance on either side of the design size of the tank.  Hopefully that will be enough.  I have enough 2" board left that I could play around with non-floating rafts.  I believe I read about these in the FAO AP publication, and they sounded interesting, albeit with their own brand of downsides (like everything else).

The beds had been sitting on the sawhorses (right side of the image) until a few nights ago.  I was able to lift the GB down to the ground myself, only by virtue of exploiting the physics of balance.  After the plywood was attached, each box gained considerable weight.  Further moves will require the assistance of at least two more people (unless I can find just one rather burly individual).

Roughing-in the stands

On the porch, it's a game of musical junk.  There are some outstanding projects that need to be completed, and for which materials remain (such as all that white junk on the floor).  Also, I have a few too many pieces of lumber, so there will be a return to the home-improvement store in the near future.  Once we relocated the DWC, however, it was then time to get the sawhorses out of the way and trial-fit the three stands.  You can see in the photo to the right the sands and the slants.

I had intended to rig up a jig to ensure clean, consistent cutting of the slants, and also purchased enough lumber to cut individual boards for both top and bottom sets.  As it turned out, a single board made two very nice pieces (thanks to the law of similar triangles), so I cut my board requirements by half.  Second, making a jig became extremely difficult due to the lengths involved.  I eventually settled on using as much of the jig as I had already assembled (which was a glorified board-holder, and not a very good one at that), and marking and free-handing the circular saw along the cut.  For the most part, the cuts went well.  I rarely strayed from the line, and when I did it was because the line was nearly invisible on some of the pieces (curse of pencil on PT lumber...I used a Sharpie for the board that followed).  I tested one of the pieces with a level directly on the porch floor, and to my exuberance the angle evaluation and cut was good!!  So now we should have nearly level grow boxes.

Next up will be purchasing the liner, the paint, and the marine sealant for the bulkhead fittings.  I am still working on the lighting, and have finished most of the reading I had started on the subject.  I'll discuss my findings in another post.  I haven't yet planned where the fittings will go on the GB, since I must account for the 6" diameter media guard.  I was thinking of locating them near the back of the unit, so as to keep the plumbing out of the way of the DWC, but it occurs to me that the DWC is wider than the GB so that point may be moot.  In any case, once the liner is installed and the GB and DWC tanks are fixed in place, I can start doing leak and siphon testing.

Rockin'....

Production Siphon Experiment 2, and (Dis)regarding the Venturi Effect

After the dismal failure of my own damn reasoning, I worked on some revisions.  The target drain pipe size for the system will be 1.25" diameter.  I performed a test with this pipe, plus some 0.75" pipe to test some more relay siphon ideas.

The basic configuration I used was 1.25" for the below-bed siphon and drain pipe, but 2" for the standpipe.  This was partly on a whim.  I also experimented with a 1.25" standpipe.  The 2" worked significantly better, especially draining into the 1.25" lower siphon pipe.  It was as though I had simply moved the "funnel" that is commonly used in other configurations down to below the grow-bed.  I suspect the funnel action here really has nothing to do with the Venturi effect.  More on that below.

During the tests, I ran the water pump and cycled water between my test tank (a 20 gallon rubber-like plastic container I picked up for cheap) and a makeshift sump tank (a sufficiently large galvanized tub - from whence I bought it, I have no idea).  This provided a great opportunity to both test the pump at the likely head-height, in the production plumbing, and to save some water.

The 1.25" standpipe produced a very undesirable result: as water spilled into it, it formed a vortex which was maintained indefinitely.  The vortex evidently allowed sufficient water to leave the tank, and prevented siphon formation.  The 2" standpipe did not appear to suffer from this problem, and worked repeatedly without fail.

Apropos: my current bell is a 6" diameter PVC pipe, covered with plastic kitchen wrap and "sealed" with tape.  The wrap provides a sort of window into the inner-workings of the siphon.  My next available bell size is 3", which is too small to work with the 2" pipe and adapters.  I am hoping that 4" pipe will offer sufficing inside diameter.  The store does not seem to carry 5".

While watching the 2" pipe, I noticed that the flood of water down to the reducer fitting caused enough turbulence and sufficient restriction of flow for a decent blockade of water to accumulate, and thus the siphon would shortly follow.  I decided to try relaying in the 3/4" pipe I had installed in the same tub.  I used a 1.5" wye fitting, normally used for waste-water collection, and (due to a lack of 3/4" fittings) drove the 1.25" plumbing into the wye at 45 degrees.  The 3/4" pipe went straight down into the fitting, and the exit was directly below.  Appropriate adapters were used to deal with the size differences.  The exit piping was unfortunately a little too short: it ran straight into the sump as I tried to lengthen it.

Unfortunately, it didn't work.  The key problem was that there was simply too much turbulence at and after the wye for a good seal to form.  Even when the 3/4" standpipe was cut so that it was very close in height to the 2" standpipe, there was insufficient suction to start the 3/4" pipe.  I suspect this issue could be averted if the 3/4" was upstream from the 2", and the 1.25" pipe had sufficient horizontal length downstream of the 2" standpipe to ensure a good blockade of water.  I plan to retest, using the other 2" bulkhead fitting I have, once I have some 4" pipe.

Regarding Venturi

I am beginning to suspect that at the pressures and water velocities we're dealing with, the Venturi Effect is a negligible effect at best when we examine the funnel-style standpipe.  I'd like to propose another hypothesis: mass of water.  Let's suppose we have a 5" length of 2" diameter standpipe.  Water begins to flow over the top circumference and down the inside faces of the pipe.  Let's assume the water flowing down is 0.1" in thickness.  Let's freeze this scene as the water reaches the bottom of the standpipe (which is also the top of the 1.25" siphon drain).  For a 2" pipe and 5" of length, that's about 3 cubic inches of water.  Were we considering the same 0.1" thickness of water inside the 1.25" pipe alone, that would equate to about 1.8 cubic inches of water.  In other words, the larger standpipe provides a deluge by virtue of its geometry alone.

Dividing the 2" pipe's low volume of water by the 1.25" pipe's cross-sectional area gives us the result that the same volume of water would occupy 2.43" linear inches of the 1.25" pipe.  This, I believe, is the reason the siphon starts so easily with this configuration: that much water flooding the pipe is significantly more than a 1.25" standpipe would have supplied, thus a water blockade can form and induce a siphon.

The picture here attempts to illustrate the example from above.  The siphon is in red (2" standpipe, 1.25" drain), water is in blue.  The drawing is to scale.  The water on either side of the standpipe represents 0.1" of thickness.   Realize that the water is actually coming down around the entire circumference of the standpipe opening, and so is much more voluminous than what I can easily illustrate.  The water shown leaving the drain pipe (which is pointing down) is the equivalent volume as what is entering the standpipe.  It, too, is to scale.  As you can see, a relatively small volume of incoming water becomes a relatively large volume of draining water, and should saturate the pipe volume sufficiently to start the siphon.

Were we to use two 2" pipes, or two 1.25" pipes (that is, the same size between the standpipe and the drain), we must then hope that the water level continues to rise sufficiently to eventually saturate the siphon.  In the meantime, there would be physically little to stop the water from simply draining through without accumulating.  Thus, I suspect the Venturi effect is of little value here, and that the geometry of the pipes provides the most potential benefit.

Progress, and a Failed Experiment

Today I was able to procure many additional materials: plywood, polystyrene panels, 2x4s and 4x4s, plus some plumbing supplies for further experiments.  I assembled the three grow-bed support frames.  I had to obtain a larger container to carry out the siphon experiments.

As I had mentioned in my last post, I had a crazy idea:

A miserable failure - do not repeat it!!

Yes, this is a two-stage inside the same bell.  IT DOES NOT WORK.  This is why I take notes on these things...it's just too bad I didn't actually re-read them before attempting this hair-brained idea.

Why does it fail?  The small red pipe feeds directly into the large main pipe.  There is no way enough water can flow into that massive blue down-pipe to form a siphon, therefore the siphon simply never forms.  What made the original 2-stage experiment successful was that the starter siphon formed a full and fast siphon. With the additional water being pumped into the larger out-piping, there was sufficient volume to start the larger siphon.

That is not the case with the above illustration.  What you see above is simply a slow flow into a very large pipe, which would happen even without that extra red pipe sticking out the side.  Without an actual siphon forming with the smaller pipe, there is insufficient water flow to trigger the main siphon.

Some other bad points about the above design: it's huge, it requires a very large diameter bell (which equates to an even larger media guard - bleh), it doesn't work (as previously mentioned), it requires many more fittings than its worth, and so on.  I will be trying some more experiments soon.  Namely, I want to see if a 3/4" starter siphon will trigger a 2" main siphon.  If it does, my next experiment will test two 2" main siphons with the 3/4 starter.  I can also modify the diameter of the out-piping, to help ensure the flow is sufficient to trip the other siphons.  The catch is that the smaller the out-pipe, the less flow it can handle, so there may cease to be a reason to run 2" siphons.

Bulkhead Fittings and Siphons

I have purchased a small selection of bulkhead fittings from US Plastics, an online supplier of those and evidently a large variety of other plastic-related items.  I am currently gearing up to procure another round of supplies, at which point I will be very close to the point of being able to fill the system with water.

Since I had to purchase fittings, it seemed like a good time to do some calculations and make sure I was sizing the plumbing appropriately.  I assume a pump rate of 800 gph (it's a 1000 gph pump, but I am attempting to account for head loss).  I can't seem to find the calculations (they must be in another spreadsheet), but for the moment I am assuming that the media will occupy about 30% of the GB space.  The GB's max operational capacity is approximately 100 gallons.  Thus, we will suppose 70 gallons will filter into the GB before the siphons start.

Given the above, the anticipated static fill time (the time it takes to fill completely before draining starts) should be about 315 seconds.  A 1 inch diameter drain pipe should pull the water out of the GB in about 200 seconds.  Figure in the constant fill rate, and the drain time jumps to 547 seconds (we'll call this the dynamic drain time), meaning it would take far longer to drain than to fill.  At 1.5 inches of drain diameter, we should expect 124 seconds of dynamic drain time.  That's better.  At 2 inches, it should drain all 70 gallons in about 59 seconds.

Now, I can only cut these holes once, and I am slightly concerned that the length of the bed will cause one side to become more stagnant than the other, should I place the drain in one corner.  The alternative is therefore to add a second drain.  Two 1.5" diameter drains equals one 2.12" diameter drain in drain area, so the dynamic drain time would be around 52 seconds.  Two 2" drains (equivalent to one 2.83" diameter drain) will reduce the drain time to 27 seconds.  These times do assume sufficient out-piping, which means I would probably need to enlarge at least a portion of the pipe running down to the DWC.

Whether or not I really need the second drain is up for debate, but by at least installing the bulkhead fitting now, I can always add the actual drain later (and otherwise just plug the hole if it need not be used.  If I do go with the larger size pipe, or two of them for that matter, I will very likely need to contend with a siphon start issue.  Thinking back to the pull-start siphon I experimented with several months ago, I decided it might be good to find a way to build that in.  Putting two bulkheads next to each other, however, did not seem like a good idea, and the loss of GB real-estate was becoming bothersome.

One potential solution, which I will be experimenting with as soon as the fittings arrive, is to put the start-pipe assembly above the bulkhead, rather than below it.  In concept, there would be only a single pipe and single bulkhead fitting.  The pipe would tee immediately above the fitting, and proceed up at full diameter and out at the start-pipe diameter (1/2" most likely).  The start pipe connection would do a bend and then run parallel to the main drain, terminating just below the top of the main.  The start pipe would therefore set the max water height in the GB.

Two siphons should not pose any special difficulty, as long as they are drained together.  If the main (pull-started) siphon is closest to the out-pipe, and the secondary siphon connects into it just before the out-pipe, the secondary siphon should get pull-started by the main siphon.  That is, once the main siphon trips, it should pull sufficient vacuum on the drain plumbing to forced the second siphon to start.  The effect will be a three-stage drain: start, primary, secondary.

It may be better to put the primary siphon upstream of the secondary, but my only concern is that the long horizontal run will cause problems for primary siphon start.

Regardless of how the siphons are started or how fast they flow, this action has an effect on the DWC tank:  given the dimensions of the DWC, a 70 gallon ebb-and-flow will equate to approximately 5.6" of rise and fall.  Consequently, the pump will need to gather water from as close to bottom as possible.  It should be noted that this height changes is slightly more than 50% of the maximum water height for the DWC.  Should it be determined that less water is required in the GB to trip the siphon, then this percentage will be reduced accordingly.  (For instance, a 50 gallon ebb-and-flow will equate to a 4" rise and fall.)

I plan to position the water pump outside the DWC, piped so that its inlet it always submerged.  I may need to find a decent filter material to keep debris from entering the pump.  One of the bulkhead fittings I acquired is intended to go through the wall of the DWC.  This is mainly to ensure that the pump can be properly primed before starting, as it is not self-priming.

Testing will hopefully commence soon!

Lighting Research - Preliminary Studies: White versus Red/Blue

I've been doing quite a bit of reading about the lighting requirements of plants.  I'll provide some details on my findings, plus references, in another post.  For now, some notes.  My goals are to build LED lighting assemblies that will provide adequate and appropriate light for my plants.  As there will be a variety of vegetables, plus fruiting plants and hopefully some root-plants, the lighting must be able to accommodate them all.

I investigated HIDs and fluorescent fixtures, and ruled both out as too costly in terms of setup and upkeep.  The former are power-hogs.  The latter's bulbs reportedly lose value after 6 months and require replacement.  LEDs to the rescue, then!  But, it's not so simple.

Firstly, manufactured LED assemblies are horribly expensive.  Second, the majority of them are red-blue fixtures.  Some also incorporate green.  I did a great deal of search about the best growth spectrum, the typical spectra for chlorophyll, and eventually found myself on a very detailed site regarding photobiology.  Here's what I have so far...

Chlorophyll-1 and -2 do indeed absorb rather specific portions of PAR light.  One takes the majority of its energy from around 450nm, the other from the 640nm area (very roughly, don't count on those numbers AT ALL).  However, their precise absorptance spectra depend on the solvent in which they are dissolved, if considered in vitro.  When one considers the effects of PAR spectra in vivo, it has been observed that a very wide range of the 400-700nm wavelengths are indeed absorbed.  These are not necessarily absorbed by the chlorophyll itself, but could be absorbed by accessory pigments and other chemical processes within the plant.  Furthermore, the presence of other proteins and chemicals either attached to the chlorophyll, or within close proximity to it, will affect (sometimes to a great degree) the spectra that the chlorophyll can and will absorb.  We cannot rely on in vitro data alone to build our lighting, as it does not accurately characterize the actual operating conditions of chlorophyll.

Some plants also make specific use of the far-red region of the spectrum.  This is evidenced by phytochromes, namely Pr and Pfr (for the red and far-red spectra).  The two convert to one another in the presence of the right light, and if I remember correctly Pfr converts without light input to Pf.  The relative quantities of these two phytochromes determine the expression of certain traits, such as the straightening of plant parts.  But the key take-away is this: we cannot alone rely on the spectra relevant to chlorophyll to determine the appropriate lighting.

Given that this will not be a monocrop, and given that the sun's spectrum in the PAR region is generally flat (as apposed to the spiked spectra for chlorophyll-1 and -2), it becomes evident that focusing on only the blue and red spectra will not be for the best of the plants.  In other words, white light with a fairly even spectral distribution will be the optimal light, with what limited information I have about the goings-on inside the plants themselves.  While it will certainly be sub-optimal from the point of view of chlorophyll - that is, I am technically "wasting" power on light that isn't immediately absorbed by the chlorophyll - it will also be the kind of light the plant has evolved to expect.  Given the number and variety of inner-plant processes that we either do not know, or do not understand, it is possible the spectra-restricted (red/blue) lighting is itself deficient, and therefore wasteful.

Many questions remain:

• How much light, in terms of PPF per day, will the plants require?
• What will happen if too much light is provided?
• What will be the best LED light source for the spectra chosen?  Will I need to augment that spectrum with additional, spectra-specific LEDs?
• That is, most of the white LEDs I've surveyed to-date (all from Cree), tend to show dips in the spectrum after blue, but before yellow/red.  This probably means emission of the green spectra is greatly reduced.  I would have to compensate with additional green LEDs, but only if that really matters in the end.  It's possible it does not.
• What should be the minimum and maximum distances for the light sources from the crops?
• On the DWC, there will be no choice
• Will the even, consistent lighting prevent fruiting?  If so, will it be necessary to implement additional lighting (in specific spectra) that can be triggered on an as-needed basis?
• If spectra-specific lighting is required for fruiting, will it cause non-fruiting plants to bolt?

The math also looks like it will be fun.  Lots of 10-to-the-very-large-number exponents.