On-Demand Hot Water Circulation

[Kabra]

Wondering if I can get some input on water velocity concerns from a pro on a redesign I did of my on-demand hot water recirculation system.

Originally I couldn't find anyone who could do what I wanted so I wound up building the system myself about four years ago. It's worked great but I always wished I could get faster hot water delivery which used to take 30 to 50 seconds depending on length of the branch. I've now cut that time by almost 1/2 with a redesign that includes putting two pumps in series.

I have momentary switches near faucets that initiate a timer and in turn starts a pump. It's a single story ranch with a big footprint and also has a basement. The hot water leaves the water heater (tank type) and splits off in four quadrants. The initial design had four Taco 006E3 pumps, one for each of the quadrants. The recirculation lines in turn come back and then join a common return to the cold water line on the water heater.

My redesign (see photo) simply uses two of the Taco pumps in series and are common to all of the zones. Each zone has a solenoid valve that's activated along with the pumps. The pumps are set to max speed which I'm aware would likely be problematic if they ran all the time, but I have the timers set for 20 - 25 seconds so obviously they don't spend a lot of time moving water. I do hear a hum when both pumps are running but not when only one is running. The noise doesn't seem to be from the pumps but from the pipes so I assume that's some sort of resonance from the water velocity in the pipes. So I'm wondering if that's a concern at all.

If there was some way I could measure the actual head of each zone I could probably do the calculations to determine optimal pump speed but I have no idea how to do that. But overall with the two pumps in series the system works as I hoped it would but I'm just concerned whether the MAX pump settings might be problematic in some way.

I appreciate any input on this.

 

[hj]

I am sure your flow velocities are off the chart. You will probably encounter a lot of erosion, and therefore leaks, at all turns where turbulence is highest. The PEX insert fittings may also be a problem at high velocities.

 

[Sponsor]

 

[Kabra]

I am sure your flow velocities are off the chart. You will probably encounter a lot of erosion, and therefore leaks, at all turns where turbulence is highest. The PEX insert fittings may also be a problem at high velocities.

Thanks HJ. I also assume they're off the chart but doesn't the fact that the pumps run only a few minutes a day mitigate the concern about erosion? I forgot to mention that each pump has 13' of head at the max setting per the data sheet:

https://www.tacocomfort.com/documents/FileLibrary/006e3_Catalog_100-143.pdf

 

[John Gayewski]

Pretty sure "pumps in series" isn't a thing and does not do what you think it does. When I say pretty sure I mean I know it's not doing what you think.

Your only using one pump,and they are both being ruined.

 

[Kabra]

Pretty sure "pumps in series" isn't a thing and does not do what you think it does. When I say pretty sure I mean I know it's not doing what you think.

Your only using one pump,and they are both being ruined.

John, pumps in series and in parallel are very much a thing. In series the head is additive and in parallel the flow is additive as I understand it. I'm sure there's more to it but it's definitely something that is engineered for various reasons but from what I've seen typically in more industrial applications.

I see from the shapes of the curves in the product literature that there's potential to turn the pumps down without losing much performance. But without knowing how to measure something (like velocity?) I have no idea where each branch sits on the curve. If I ruin the pumps I still have two more. It's more the pex lines and fittings I'd rather not destroy. But even then almost all of it is exposed in the basement. But this is why I'm asking. I'm just hoping to understand this more definitively because I really like how quickly hot water comes out of the tap.

 

[Bgard]

Generally Pumps in series doubles the head, pumps in parallel doubles the flow.

 

[John Gayewski]

John, pumps in series and in parallel are very much a thing. In series the head is additive and in parallel the flow is additive as I understand it. I'm sure there's more to it but it's definitely something that is engineered for various reasons but from what I've seen typically in more industrial applications.

I see from the shapes of the curves in the product literature that there's potential to turn the pumps down without losing much performance. But without knowing how to measure something (like velocity?) I have no idea where each branch sits on the curve. If I ruin the pumps I still have two more. It's more the pex lines and fittings I'd rather not destroy. But even then almost all of it is exposed in the basement. But this is why I'm asking. I'm just hoping to understand this more definitively because I really like how quickly hot water comes out of the tap.

Pumps in parallel are a thing. Series not a thing. That's only for lifting water one way. Definitely not for what your doing.

Think of it like this. Water is not compressable so let's use marbles. The pump moves one marble in and one marble out at a set rate. Adding another pump directly next to it doesn't change that. You have essentially a closed loop when one marble moves they all move. All the other pump is doing is moving one marble in and out. It doesn't affect anything other than making turbulence for the other pump.

The number that professionals use to size systems depends on the type of piping. For pex you can use 0.03

Take your longest loop (you only need the longest loop since they are in parallel) in feet and multiply. That's your effective head.

 

[John Gayewski]

How long is your longest loop? What size pipe?

 

[Kabra]

Pumps in parallel are a thing. Series not a thing. That's only for lifting water one way. Definitely not for what your doing.

Think of it like this. Water is not compressable so let's use marbles. The pump moves one marble in and one marble out at a set rate. Adding another pump directly next to it doesn't change that. You have essentially a closed loop when one marble moves they all move. All the other pump is doing is moving one marble in and out. It doesn't affect anything other than making turbulence for the other pump.

The number that professionals use to size systems depends on the type of piping. For pex you can use 0.03

Take your longest loop (you only need the longest loop since they are in parallel) in feet and multiply. That's your effective head.

Your marbles being moved at a set rate example brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move. The huge performance improvement from the second pump

Pumps in parallel are a thing. Series not a thing. That's only for lifting water one way. Definitely not for what your doing.

Think of it like this. Water is not compressable so let's use marbles. The pump moves one marble in and one marble out at a set rate. Adding another pump directly next to it doesn't change that. You have essentially a closed loop when one marble moves they all move. All the other pump is doing is moving one marble in and out. It doesn't affect anything other than making turbulence for the other pump.

The number that professionals use to size systems depends on the type of piping. For pex you can use 0.03

Take your longest loop (you only need the longest loop since they are in parallel) in feet and multiply. That's your effective head.

Your example of marbles being moved at a set rate brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move through a given medium. The huge performance improvement from the second pump indicates there was a lot of room to move those marbles faster, in fact nearly twice as fast. At what cost, I'm not

Regarding the effective head using .03 I would say the longest loop is 150' so that's a meager 4.5' of head and the pumps add up 26' at the max setting. Thanks for that factor, this at least gives me a starting reference point where before this I had no concept. If that calculation is anywhere in the ballpark that's a little less than six times actual head.

I'm sure that is massive as you suggest, but I wonder what would be a safe multiple given that the pumps run so infrequently and for extremely short periods of time? My assumption of course (right or wrong) is that some level of turbulence isn't of too much practical concern given the cumulative operating time of the pumps. I'm thinking the kitchen would be the most frequently used and I'm guessing that might be an average of six times per day x 20 seconds or a total of 2 minutes.

I really appreciate your input and it's helping me understand more about this.

 

[Kabra]

How long is your longest loop? What size pipe?

As I mentioned in the prior post the longest loop is roughly 150'.

You also mentioned the longest loop is the one that matters because they're in parallel with the other loops. But if I understand correctly I'm not sure they are in parallel and in fact are completely isolated due to the solenoid valves. I suppose if two zones were triggered simultaneously (opening two valves) they would be in parallel but the chances of that happening are totally miniscule. As you can see from the photo I sent in the original post all the zones join a common return to the water heater after the solenoid valve. So nothing flows through any loop where the valve is not open.

I used 3/4 pex for as much of the return system as possible. The logic being to reduce head and the fact that the volume of water on the return side (I think!) is totally inconsequential because its movement is dictated by how much the supply side can pass through to the return side. Three of the four return loops have tee off into two fairly short 1/2" lines to faucets. The fourth zone is the kitchen which only has a single 1/2" return for the sink.

 

[Kabra]

Generally Pumps in series doubles the head, pumps in parallel doubles the flow.

That is my understanding as well. Having decided to use two pumps with the desire to speed up hot water delivery I decided to put them in series rather than parallel. But honestly that was a guess so I tested it before being committed and like I said earlier, it significantly improved delivery speed. I still don't really have an understanding if this problem is best solved through increased head or higher flow. It seemed like pressure to me. I thought about testing a parallel setup but I wasn't sure how to setup the pumps for a meaningful experiment. Just using tees going from one 3/4 line to two pumps and back to one 3/4 line before and after the two pumps seemed kind of pointless so I didn't bother. And like I said the results of the pumps in series met my performance objective completely so I left it at that. Of course was somewhat ignoring the turbulence concern that was created with the pumps in series.

 

[John Gayewski]

Your marbles being moved at a set rate example brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move. The huge performance improvement from the second pump

Your example of marbles being moved at a set rate brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move through a given medium. The huge performance improvement from the second pump indicates there was a lot of room to move those marbles faster, in fact nearly twice as fast. At what cost, I'm not

Regarding the effective head using .03 I would say the longest loop is 150' so that's a meager 4.5' of head and the pumps add up 26' at the max setting. Thanks for that factor, this at least gives me a starting reference point where before this I had no concept. If that calculation is anywhere in the ballpark that's a little less than six times actual head.

I'm sure that is massive as you suggest, but I wonder what would be a safe multiple given that the pumps run so infrequently and for extremely short periods of time? My assumption of course (right or wrong) is that some level of turbulence isn't of too much practical concern given the cumulative operating time of the pumps. I'm thinking the kitchen would be the most frequently used and I'm guessing that might be an average of six times per day x 20 seconds or a total of 2 minutes.

I really appreciate your input and it's helping me understand more about this.

I think your missing the point, but your velocity with one taco 006 is around 4.4 ft per sec. Two pumps isn't getting you anything but a chuckle from me and two bad pumps in a much shorter time than would be expected with a properly configured system.

You want more velocity you need a bigger pump, two in a row gets you nothing as your only moving water in a circle.

 

[John Gayewski]

Your max is gonna be around 8gpm which will get you around 7ft per sec. You'll need to find a pump with a curve that does 8gpm with around 5ft of head.

 

[John Gayewski]

Your marbles being moved at a set rate example brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move. The huge performance improvement from the second pump

Your example of marbles being moved at a set rate brings up a good point. I think what you're inferring is that there's a limit on how fast those marbles can move through a given medium. The huge performance improvement from the second pump indicates there was a lot of room to move those marbles faster, in fact nearly twice as fast. At what cost, I'm not

Regarding the effective head using .03 I would say the longest loop is 150' so that's a meager 4.5' of head and the pumps add up 26' at the max setting. Thanks for that factor, this at least gives me a starting reference point where before this I had no concept. If that calculation is anywhere in the ballpark that's a little less than six times actual head.

I'm sure that is massive as you suggest, but I wonder what would be a safe multiple given that the pumps run so infrequently and for extremely short periods of time? My assumption of course (right or wrong) is that some level of turbulence isn't of too much practical concern given the cumulative operating time of the pumps. I'm thinking the kitchen would be the most frequently used and I'm guessing that might be an average of six times per day x 20 seconds or a total of 2 minutes.

I really appreciate your input and it's helping me understand more about this.

No the first pump is only getting that the second pump gives it and vice versa. There is no speed increase. One marble in one marble out, next marble in, next marble out. The is no point where one pump moves one marble then the next pump moves two marbles. Your only stirring the marbles up and trying to jamb two in at a time, but only one will fit (cavitation).

The reason you only need to factor head loss on the longest loop is because the pump is only overcoming that amount of head. The other loops being shorter doesn't effect what the pump can drink. So if your pump can drink one marble at 150' it can still drink one marble at 100'. It can drink 100' a little faster, but it's still one marble, and the speed doesn't change that much, not enough worry about. Even when they are in parallel with multiple loops running the pump is moving one marble, how they are getting there doesn't matter if the longest run is only 150'.

 

[Kabra]

I think your missing the point, but your velocity with one taco 006 is around 4.4 ft per sec. Two pumps isn't getting you anything but a chuckle from me and two bad pumps in a much shorter time than would be expected with a properly configured system.

You want more velocity you need a bigger pump, two in a row gets you nothing as your only moving water in a circle.

No the first pump is only getting that the second pump gives it and vice versa. There is no speed increase. One marble in one marble out, next marble in, next marble out. The is no point where one pump moves one marble then the next pump moves two marbles. Your only stirring the marbles up and trying to jamb two in at a time, but only one will fit (cavitation).

The reason you only need to factor head loss on the longest loop is because the pump is only overcoming that amount of head. The other loops being shorter doesn't effect what the pump can drink. So if your pump can drink one marble at 150' it can still drink one marble at 100'. It can drink 100' a little faster, but it's still one marble, and the speed doesn't change that much, not enough worry about. Even when they are in parallel with multiple loops running the pump is moving one marble, how they are getting there doesn't matter if the longest run is only 150'.

Ok I've got to read this a few more times and really digest it. I'm not qualified to debate any of this with you but what I'm seeing does seem to contradict what I think you're saying that at some point water hits an absolute speed limit and I'm well past that point. But three things I see indicate otherwise:

1. As I've pointed out there's a very substantial and obvious speed increase in adding that second pump.

2. Any pump performance chart shows that flow increases as pump pressure increases but of course the curve flattens as pressure increases.

3. I've turned the speed of the pumps down and performance goes down. I unplugged the lower pump (although didn't remove it) and performance also goes way down.

So I'm not sure how to reconcile what I'm seeing with this absolute speed limit notion.

 

[Kabra]

Your max is gonna be around 8gpm which will get you around 7ft per sec. You'll need to find a pump with a curve that does 8gpm with around 5ft of head.

The pumps I have can be set to work near those parameters. The Max setting per the chart deliver 9gpm at 5' of head. And that is exactly what I had prior to adding the second pump. But as I said that adding the second pump made a massive improvement for whatever reason.

 

[John Gayewski]

Your pumping through a pump. That is adding resistence. Not sure how much as no one pumps through a pump and I don't think pumps have cv ratings. Pumps don't have cv ratings because people don't pump through pumps.

The speed isn't absolute it's dependant on head resistant. Hence the curve. Head goes up (by putting another pump in the way) flow goes down, speed goes down.

We don't use pressure for much in this type of sizing and design. Differential pressure is the only thing we care about.

 

[Kabra]

John, I went away for a few days to take your very last comment to heart and try and understand the dynamics of what I have a little better. I installed pressure gauges before and after the pumps to see what the real differential pressure is and it turns out to be consistently about 8 psi. If I understand correctly that indicates (8 x 2.31) 18.5' of head and using pressure drop tables for 150' (the longest of the loops) of pex is around 5gpm. And since I only need to displace the water through the supply side (less than 1/2 the loop) that comes pretty close to the 15 - 25 seconds I'm experiencing depending on the length of the particular loop.

I'm not fully confidant that's correct logic but I have a pretty good sense that you'll point out how it's wrong. I realize even 5gpm is twice what you like to see, but in a practical sense it's hard to imagine it will cause any meaningful issues running for just a few minutes a day the way I use the system.

 

[John Gayewski]

I think i distracted you by mentioning pressure differential. All that means in this case is the your pump(s) are creating 8 psi and adding it to your system's pressure. Doesn't mean much to me as those pumps are meant to move water not create pressure.

I'm not sure there's much left to say as I don't know what your question is. Any time you feel like it you can get rid of one of those pumps and have the same (or better) performance and lengthen the life of the pump.

I think i mentioned before those pumps should run 20 or more years depending on your water chemistry and how often they are turned on and off. You'd be better off running them all the time and having one pump. You'd get better faster hot water and your pumps will last twice as long.

 

[Kabra]

John, the question is so simple and I've asked it numerous times in this thread.

I simply would like to know if the setup I have is causing a significant amount of erosion considering that it only runs on demand and the pumps run just a few minutes per day?

I've been using a slightly different single pump setup for four years and the addition of the second pump nearly doubles the speed of delivery to the point I couldn't be happier. You've made it clear that you think this is impossible but it works and it works very well! I would just like to have a better understanding of possible erosion issues and it would be really great if you or someone else could provide some numbers based on the info I've provided.

I assume the important estimate is flow rate. I estimated about 5GPM if I calculated correctly based on the 8 PSI differential pressure and pressure drop of 150' of 3/4" PEX.

Here is a summary of the info:

The 3/4" PEX main hot water supply branches into four quadrants of a single story home with a basement. The faucets are supplied with 1/2" PEX.

Each quadrant has a separate return that tees into a common return to the water heater. Each return is 3/4" PEX except two 1/2" PEX lines that tee close to faucets.

I use two Taco 006e3 pumps in series and are set to maximum. Pump specs and curves are here: https://www.tacocomfort.com/documents/FileLibrary/006e3_Catalog_100-143.pdf

Differential Pressure is about 8 psi.

Following are pictures of the current setup with two pumps in series and the old setup with separate single pumps for each quadrant. Pressing a momentary switch near a faucet activates a timer relay for that quadrant which opens the quadrant's normally closed solenoid valve and turns on the two pumps for 20 - 25 seconds depending on which quadrant was activated.

Here is an image of the new setup using two pumps in series that are common to all four quadrants.

The old setup has separate pumps for each quadrant with a common return to the the water heater.