So, I’ve finished my replumbing project complete with DIY on demand recirculation loop. I thought I’d leave a few notes here for anybody wanting to try this.
The good news: It works! Nearly instantaneous hot water at ever fixture, as often as I’ve tried them in the last day or two since finishing the project anyway.
The bad news: Was it cheap? Mmmm maybe. It depends on which demand based recirculation system you’re looking at.
There are 2 dedicate recirculation options sold by my water heater manufacturer, Rheem, which go with my tankless RTGH-84xln-1. The first one is their timer based option and it can be had for about $250. The second is their “on demand” kit which goes for upwards of $500. Most other “demand” systems I looked at were around $500 and up too. All in here I’ve got $199 Watts pump kit from Amazon , and another maybe $50 into the fibaro sensor and DS18B20. Of course the benefit of being fully ST integrated is not something you can buy at any price.
The Watts pump is a “timer based” pump though I’m using it as a “demand” pump. It’s unclear to me if there is a difference in flow velocity between “demand” pumps and “timer” pumps. I’d expect that there is, but it’s really not clear from the literature of recirc system manufacturers. The specs on the seemingly identical Grundfos version of the Watts pump claim that it’s a 1/25 HP pump which can deliver “0-7 GPM” depending on the “head”. I’m still unclear on what pipe head is so… uh, yeah. These specs do not seem to be out of line with what a “demand” pump claims be able to do however, so again, is there a difference? I don’t know enough about pumps and plumbing engineering to be able to tell.
The Watts pump seems to do the trick, though I’m still experimenting with it. It’s delivering enough flow back through my 1/2" recirc line to turn on my tankless water heater anyway. The Rheem tankless requires a minimum flow rate of .26 gpm (best in the industry! says rheem), so the flow rate the pump is delivering is somewhere between .26 GPM and 7? Yeah, not very helpful, I know.
It gets hot water to my furthest fixture in under a minute, which is the runtime I have my WebCoRE piston set at right now. Would 30-45 seconds work, which is about the time that dedicated on demand systems claim they can deliver hot water? Still experimenting.
The other issue here is that there’s significant delay in the feedback loop from the fibaro/DS18B20 sensor. Since the fibaro is a battery powered device, you can’t really set it up with real time temperature monitoring such that when you sense your desired temp you shut off the pump. Mine is sampling every 2 minutes, so the best I can do with it is to use it to decide if the water in the pipe is already hot enough to not have to run the pump for a motion event.
The last issue that I’ve noticed is that when waking up in the morning, it takes the fibaro/DS18B20 setup a while to “warm up”. I have the sensor glued with arctic silver thermal adhesive to a 6" piece of 3/4" type L copper, hooked up to pex on either side with 2 sharkbite fittings, all wrapped in foam pipe insulation. When the first motion event of the morning triggers a pump run, I get hot water at my fixtures as expected, but the sensor is still reporting much lower temps than after a pump run later in the day.
It appears that it takes a bit of hot water running through it to warm up my 6" piece of copper. Perhaps I should have used the thinner walled M type pipe for this application, or maybe sweated on the barb ends to connect to the pex, rather than using the more thermally massive Sharkbites? I’m unlikely to change it now, but for future builders this, or locatring your sensor on a 1/2" line might be something to consider.
So bottom line, the system is working but if there’s any room for improvement it would be with 2 things, a powered SmartThings enabled in-line aquastat, and possibly a faster pump. That would take nearly all of the play out of the system.