Large Drainback System for 16 Unit Apartment


This is a large drainback system to provide domestic hot water for a 16 unit apartment building in the Pennsylvania.


The system was designed and built by Alan Rushforth of RushforthSolar LLC -- here is his description of the system:

"Attached are some photos (before and after) of a solar DHW system I recently finished.  It is on a roof of a 16 unit apartment building.  It is a drain back system. 


Prior to the installation, the building was using $4,600 annually in natural gas solely for domestic hot water.  A new 95% efficient Polaris gas water heater was installed along with this solar DHW system.  Not counting my labor, or the gas heater installation, the project cost $13,645.    Between the new heater and the SDHW system, gas bills will hopefully be cut by 2/3.  I believe a lot of commercial SDHW applications can make more financial sense than residential applications with the increased economies of scale.  I have adopted commercial solar DHW as a hobby, and would be happy to help anyone with questions. "   


Really a nice job -- thanks very much to Alan for providing all the pictures and detailed captions!


The pictures show most of the construction detail.


If you have questions, Alan can be emailed at:   alanrushforth AT   (change the AT to @)



There are 3 rows of 3 Sunflower evacuated tube collectors from Fitch Consulting

(6 x 30 tube collectors and 3 x 20 tube collectors)



Collector Installation and Supports:

Here is the framework in process. The steel pyramid bases are 1/8" steel plate. Some cuts and the bends were done a Fazzio's (a great metal yard in South Jersey). I did the angle cuts and welding. They were screwed down into wood roof joists and then roofed over with torch down roofing to make them pretty much zero maintenance. The rest of the frame is all aluminum with stainless bolts. The horizontal plane of the entire frame is set on a slight diagonal pitch so when the collectors are mounted, everything will tilt back to the tank. If the roof were flatter/all one pitch, it might have been easier and cheaper to mount frames right on the roof, however the raised framework here allows for easy roof maintenance, and also creates the nice drainback pitch. The main beams are 3x3x1/8"x 24' long aluminum square tube. Longitudinal cross members were 2x2x20'x1/8 aluminum square tube. Diagonal bracing is 1 1/2 x 1 1/2 x 1/8 aluminum angle. The frame came out about the way I wanted it. Not bouncy. Not overkilled. Galvanized steel would have worked too and been cheaper, but aluminum should last longer and is easier to work with.



Above is a close up one of the 6 2'x2' x 1/8" steel pyramid bases. The 7/8" threaded rods made adjusting the pitch easy, although cutting and welding up the bases was a job. Somewhat lighter gauge steel (like 14 gauge) would have probably been adequate.


Here we have evacuated tube manifolds installed, and are teeing the arrangement together with copper piping. Short sections of high temp silicone rubber hose (from Fitch Consulting) worked well for hose clamping the manifolds together. 1" thick armaflex pipe insulation was installed over the piping. Installing the tubes is the last thing to get done. You don't want to inadvertently create a stagnation situation with tubes installed before the pumps and controller are ready to turn on.


To protect the Armaflex from sun and UV deterioration (this system should last for decades) I wrapped it in foil faced foam duct insulation from Home Depot. It was not as sturdy and sticky as I wanted, so I wrapped it again (later photo) with 12' width Peal and Seal roofing (from ) which is a more durable foil faced self adhesive roof product.



Above is not wrapped with Peel and Seal yet. The copper pipe runs that span between the banks of collectors need to be supported so that they do not sag and create a draining problem. I attached a 2 x2 x 20' aluminum square tube along the top of each run, to prevent sag. Then covered it all with Peel and Seal.


Finished product all wrapped up and ready for many years of abuse in the elements.


Plumbing and Storage Tank Installation:

Here is the roof before we started. The chimney in the middle was no longer used since separate heat pumps were installed in each apartment and the hot water heater was replaced with a wall vented 95% Polaris heater. So we were able to take down the top few feet and convert it to a pipe chase.


Here are the 1" pex/al/pex lines covered in 1" armaflex running down the pipe chase (old chimney). Plywood, can foam, and roofing were sealed up the opening.


Here is the basement before we started. A bit of a mess.


Here is a 822 gallon STSS tank ( ). You can see the folded EPDM liner. Behind it is 2" foam insulation board. The aluminum skin holds it all together. The tank comes flattened into an oval shape, in a crate about 4 1/2' high, 11' long, and 18" wide. We uncrated it, removed the lid on the inside, wrapped the flatted tank in some cardboard and rope to protect the edges, and dragged it down the cramped basement stairs. Once down, we tilted it up, opened it into a circle, lifted an edge, and slid the insulation board bottom circle under, and we had a tank! The orange pipe is 1" pex/al/pex (which is pretty impressive stuff, probably better than copper in many ways). We ended up converting the pex/al/pex over to copper before entering the tank. Special brass compression fittings from  did the trick. The return line dumps into the top above the water line. We converted it to copper just outside the tank to put on a ball valve I thought would be needed to modulate flow. Turns out that valve was not needed. The other line was the suction line for the pumps that enters the tank above the water line but dips down to draw water from the bottom of the tank.


Here is the tank with the insulated lid on and the plumbing complete. Suspended inside the tank is a 1 1/2" x 120' copper coil. I used the oversize diameter coil for a bit of tank effect within the coil - it holds 10 gallons of hot water. When a tenant opens a hot water faucet, pressurized cold water from the street flows through this submersed coil on its way to the heater and warms on its way. The water entering the heater is partially or fully heated by the time it reaches the heater, and the heater come on much less. On a repeat job, rather than use one large diameter coil, I might use several smaller coils in parallel that might work equally well and cost less. The upper black line is the return from the solar collectors. The black line that runs down to the two pumps on the floor is the copper suction line that goes from the bottom of the tank up to the collectors. There is a thing called pump cavitation. As I understand it, under suction within the pump, water (especially hot water) can vaporize (essentially boil) inside the pump making noise and shortening pump life. Having adequate pressure in the inlet side of the pump can eliminate cavitation, as can restricting flow on the exit side of the pumps. To maximize the inlet pressure in to the pumps, I kept them as low to the ground as possible. Cavitation was not a problem.



Here are 2 brass Taco 009 pumps. Brass pumps cost more than iron, but because the tank is vented, and exposed to oxygen, brass is best because it does not rust.

In this photo you can see the right hand ball valve is throttled half way off to get the flow from about 10 gpm down where we want it - about 4.5 gpm (.5 gpm per collector). The Goldline GL30 controller needs return water to be at least 4 degrees hotter than the feed water or it will shut down the pumps. If one left the flow at 10 gpm, the controller would regularly cycle the pumps on and off. As a subsequent refinement, rather than throttle the flow down, I hooked up a delay on timer (for about $30 from that shuts off one pump after a couple minutes (once return flow with siphon effect is established). With valves wide open, flow then drops from 10 gpm to 4.5 gpm, just where we want it.


Here you can see the Goldline GL30 on the wall - pretty simple to wire up. The rectangular metal/glass thing on the right is a flow meter. There are several makers. The key is get one rated for higher temps, like 180F, with an appropriate flow scale. I originally had it mounted lower, but then moved it up centered at the level of the water line in the tank. That way, when the pumps shut off and the water finishes draining back, it shows where the tank water line is, which is nice.




RACKS: Racks can be a major labor and material expense. I was looking for zero maintenance racks, that would not prevent roof repairs or maintenance. The raised rack design I used achieves that but was a LOT of work. An added benefit of this rack design was that I was able to tilt the plane of the racks to facilitate water draining back in to the tank when the controller turned the pump off. Racks mounted flat on the roof would save money but would have been less convenient for roofing, and would have posed separate challenges for how to achieve that critical drain-back tilt.

They were from Fitch Consulting . Bill Fitch over there is very knowledgeable and was very helpful with technical issues. Time will tell how much the gas bill is reduced, but they seem to be working well. On the first full day of operation, the 822 gallon tank went from 86F to 119F. Since that day was full sun and since the coil was not valved on yet, it accidentally provided a good test. 6,855 lbs of water was raised 33 degrees equating to about 226,000 btu net deposited into the tank in one day or just over 28,000 btu/30 tube collector (ignoring piping and tank losses). A word or caution - evac tube collectors contain a million (seems like it) little parts to assemble. Allow extra labor. Also the boxing for the glass tubes is pretty skimpy. Handle carefully and do not stand on end - thin foam end packing does not cushion the nibs adequately on the end of the glass tubes. 3 tubes out of 240 were cracked at the end, by the time I was installing - but it was not a big problem. Bill supplied a spare tube with every manifold.

If one needs to pump water 50' up using circulating pumps that only have a 33' head, theoretically 2 in series should do the trick. The theory proved correct - together they do the job. To pump the coolest water from the bottom of the tank one could cut a fitting into the bottom, or theoretically it could also be drawn up above the water line from a dip tube. That theory worked also.

Don't oversize the pipes. For cost purposes, and heat loss purposes, bigger is not necessarily better. On this size system, if the head were lower, and only one pump were used, or if one were going to leave both pumps on, 3/4" pipe would be adequate (instead of the 1" I used). Using the smallest pipe that will do the job, will help hold down pipe and pipe insulation costs as well as cut surface area heat loss a little.

The common wisdom is you must use copper for lines back and forth from the collectors to handle high stagnation temps. I felt on a drain back system, where during stagnation, there is no liquid in the collectors, Pex/al/pex (premium pex with an aluminum layer inside) should work for all the run except the last couple feet to the collector, where you might pick up more of that stagnation heat. It turns out Pex/al/pex is very impressive pipe. The aluminum embedded in the pex seems to be about 1/32" thick and the pipe is rated for 80 psi at 200F. So far it is working fine, and I have confidence it will continue to do so. Note: pex/al/pex has a slightly bigger diameter than regular pex and uses its own special compression fittings. You do not need to buy a pex crimper to use it - just the right compression fittings. (I learned that the hard way).

1" armaflex (extra thick -special ordered from a HVAC supply) worked well. Peel and Seal roofing worked well to wrap it and protect it from the elements. To boost tank insulation, (especially desirable on more oversized tanks) one might pick up 2) 4x8x 2" foam sheets and lay them down first under the tank before installation, to double the bottom insulation. One can always wrap the sides and top later with extra insulation if wanted.

I believe in this type system where the tank is a simple non-pressurized, relatively low cost item, comprising a small percentage of total system cost (less than 15%) one can afford to spring for a larger tank that contains more gallons per person than normally recommended. I used 822 gallons with 51 gallons per apartment, (typically one person in each of these apartments). Although this is more storage than the 20 to 30 gallons per person often recommended, it still did not provide as much 'flywheel effect' as I had hoped. From spot checks, November tank temps bounced from the mid 70's to the upper 120's. If I were to do it again, and if the tank would fit, I would increase the tank size to provide 80 gallons or more per unit which I believe would better absorb heat from sunny spells, carry it into cloudy spells, improving overall performance.

When ordering an STSS tank, one specifies how many tank inlet sleeves one wants installed. In this case it was 4 total, 1 for the collector feed, 1 for the collector return, 1 for street pressure feed to the coil, and one for the exit from the coil to the gas heater. Unless you specify otherwise, all 4 will be mounted all in a row spaced about 6" apart. For optimal piping runs, I needed one on the opposite side of the tank, so I removed one, moved it, and patched the hole (all above the water line). If I were to do it again, it would be better to plan out and specify tank pipe entry locations, or at least order an extra sleeve fitting or 2 in case field modification are desired.


Alan Rushforth

Rushforth Solar LLC

Bryn Mawr, PA



Gary 1/2/0