How to Size and Pipe Buffer Tanks

HeatingHelp

How to Size and Pipe Buffer Tanks

In this episode, Bob "Hot Rod" Rohr shares his knowledge about buffer tanks. He talks about when and why you should consider a hydronic buffer tank, how to size buffer tanks, and how to pipe buffer tanks.

Read the full story here

Electricman15

@hot_rod how should I pipe an electric water heater to use as a buffer tank. Thanks

hot_rod

How big of a load, either BTU/hr or gpm?

I've used tanks as small as this six gallon, up to 500 gallon!
A 3/4 mip X 1-1/4 copper adapter shown below works on tanks with 3/4" ports.
Larger electric tanks with 1" element holes, use a 1X 1-1/4 copper to male adapt.

This connection size, 1-1/4 is good up to 100,000 maybe 120,000 BTU/ hr loads

I like 2 or 3 pipe buffers, depending on what connections you have to work with. The math for calculating tank size.

aaronhjelt

aaronhjelt

Hello, my questions are regarding the "differential pressure valve set for 1-1.5 psi" in the 2-pipe diagram. I have repurposed a 50 gallon electric tank using 1-1/4 x 1 x 3/4" tees feeding a 1" manifold. I've more-or-less copied the "extended manifold" design presented by John Siegnethaler here. I am using an 80k Burnham Alta modcon and have an Alpha 15-58f as the system circulator.

1. Is the valve necessary if I haven't had issues with excess noise or hammering during operation?
2. Does the valve support the efficiency of the circulator?
3. I can't find a 3/4" valve that operates lower than 1.5 psi. Any recommendations?

Thanks,

Aaron Hjelt

Screen Shot 2024-01-10 at 11.38.59 AM.png

Screen Shot 2024-01-14 at 9.46.38 PM.png

Ironman

Scrap the bypass valve and use a delta P pump like a Grundfos Alpha.

DCContrarian

Thanks for this, Bob. It was interesting and entertaining, I particularly liked seeing some of your experimental devices.

Heat pumps present special challenges. First, you always want to do 3-pipe plumbing, with heat pumps if you're mixing you're doing something wrong. The latest generation of heat pumps have variable speed compressors to modulate the output to match the load, as well as variable speed circulators to modulate the flow to match the output. Since they can modulate rather low the buffer tank can be fairly small.

Typically they're installed with a separate circulator for the emitters. The flow through the emitters is going to be determined by how many zones are open, the flow through the heat pump is determined by the internal controller, and the two flows aren't going to match, and sometimes one is going to be bigger and sometimes the other. When the heat pump flow is bigger, the excess goes into the buffer at the top and out the bottom, and when the emitter flow is bigger the excess flows in the opposite direction. The heat pump control logic is trying to get it to operate continuously at an output that is matched to the load, and when it does that the buffer tank will rather quickly fill up with constant temperature water. If the heat pump flow is bigger the buffer will stabilize at the output temperature, and if the emitter flow is bigger the buffer will stabilize at the return temperature.

So stratification isn't going to be an issue. Which is good, because typically these systems are used both for heating and cooling, and you can't set up to stratify for both.

The way that the heat pump measures the load as it attempts to match output is to look at the send and return temperatures and the flow through the heat pump. So it's important that the actual return temperature is measured. I've seen some recommendations that the return temperature sensor be put into the buffer tank. This will trick the heat pump and cause it to cycle unnecessarily.

DCContrarian

One other comment: I agree with the caller (didn't catch his name) who said that buffer tanks should be sized for low loads, at high loads you don't need buffering. This is also true with heat pumps.

I've come around to viewing under-floor heat this way as well. I think most people view it more like the towel warmer in your example than as a principle source of heat, something to give a warm comfortable sensation rather than heat the house. So you want to size it small enough that it can run full out whenever temperatures are below around 55F. Then have other zones to maintain temperature at the design point.

hot_rod

the heat pump buffer tank can also be the heat source for the defrost cycle.

You want to use warm water generated at that 2 or higher COP instead of kicking on a resistance element at 1 COP

leveraging COP is why you bought a heat pump

So a small 20 gallon buffer is still a good option

Kaos

I went through a fair bit of back and forth on this and some basic modeling in Excel when I was doing the air to water heat pump conversion.

At first glance, I would have needed a buffer tank to keep the unit from short cycling as the min capacity on the outdoor units 20k BTU/h which is well above my heat load in the shoulder season.

In the end I didn't put any in. I do have a long primary loop that does hold about 3 gallons of water which helps.

What I have noticed is the unit does cycle in the shoulder season as expected but the runtime are still pretty long. What the heat pump controls seem to do is run for a minimum time no matter the load so when it is warmer the peak tempeartuer the loop will increase a fair bit. On a mild day I can see peak of about 45C with a slow decay as the heat is used up down to 28C when the heat pump kicks in again. Since the outdoor temps are mild making this extra hot water doesn't cost all that much extra power.

When it gets colder, the unit runs pretty much at steady sate. If a number of zones are not calling for heat, the loop temp will slowly drift up again (much slower as there are always some zones running) until the zones need heat and it will drift down again.

Defrost happens most often near freezing conditions. At that point there is always a zone running, so if the heat pump goes into defrost, it will take the heat from that zone plus the loop. I have not seen much more than a 8C drop (ie 33C down to 25C) in return water temp during a defrost event. Compared to minisplits I'm used defrost seems to be surprisingly quick. I guess it could happen that no zones are calling for heat during a defrost event, but it would be highly unlikely.

So far, I think I'm content to keep the setup as is and continue without a buffer tank. Not sure if the cost of the buffer tank will save enough electricity to bother installing it.

DCContrarian

Interest, Kaos. If you have zone valves, how do you keep it from dead-heading when all of the zones are off?

I've written about this before, but I've found that my heat pump seems to work better in low load conditions when I bypass the buffer tank using valves. By "work better" what I mean is that it turns the modulation down and runs continuously at a very low output, rather than cycling on and off and running at a higher level when it's on.

"Not sure if the cost of the buffer tank will save enough electricity to bother installing it."

It's only at very low output levels that a buffer tank even makes a difference, once you get to the point where the heat pump can run continuously the buffer tank doesn't really matter. At those low load levels, COP is very high and the total amount of energy consumed is low, so there's probably not much to be gained by optimization.

hot_rod

https://forum.heatinghelp.com/discussion/comment/1823550#Comment_1823550

It's not so much the cost of operatiop that I am concerned about, but the short cycling of equipment. It your system has logic built in to prevent that, all is well. Does the HP industry have a rule of thumb for cycle length?

In the boiler world, 20 minute run time is what most agree on. Especially with high mass non condensing boilers that need to warm up before going to sleep wet.

With modulating boilers it is a bit different story, I would look at the lowest turndown against the smallest load to determine if a buffer is worth the addition.

Seems the biggest fear for mod con shoppers is repair costs. When circuit boards, gas valves and fans go out, repairs get expensive, and sometimes a long wait for parts. Limit the on/off cycles to extend their life.

Although tankless water heaters sure do cycle on off a lot. I wonder if any have cycle counters like boilers do?

That was one of the first things Max checked as a Lochinvar rep on a troubleshoot call. How many on/off cycles. He saw one boiler that ran out the counter in just a few years time!

Kaos

With proper sizing and design on a fuel burner it is possible to design out the buffer tank. It is also very easy to build one that will cycle the unit to death.

Tankless units are not immune to cycling issues. In actual DHW use, they don't cycle all that much since they are off most of the day and night. Same as a modcon, you can still break them if used with continuous recirc or in combi heat with too small of a load.

I don't think modulating heat air to water suffer much COP from short cycling. The number seems to be about 5min. I have seen graphs for a unit showing real time COP and there were no major spikes when the unit was turning on/off.

If you want to get into the nitty gritty, the issue with cycling is the mean temperature supplied by the heat pump is higher than the mean temperature supplied to the emitters, so any time you cycle no matter the buffer size, you are effectively running at lower COP. You definitely don't want cycling when you need a lot of heat or hoter water, so oversizing can still be an issue.

@DCContrarian My setup is primary/secondary loop, so the heat pump loop is never dead headed. The only things on it are the plate HX for DWH (didn't see a reason to install a 3 way) and the Ts for space heat.

DCContrarian

"I don't think modulating heat air to water suffer much COP from short cycling. The number seems to be about 5min. "

As a data point, Chiltrix recommends a 17 gallon buffer tank for their two-ton heat pump. The swing from cycle on to cycle off is 4C or 7.2F. A 7.2F swing on 17 gallons is just about 1000 BTU. Their minimum modulation is about 25%, or 6000 BTU/hr or 100 BTU/min. So that's ten minutes at minimum modulation.

That's assuming zero load, if there is a load the cycle time is higher.

"If you want to get into the nitty gritty, the issue with cycling is the mean temperature supplied by the heat pump is higher than the mean temperature supplied to the emitters, so any time you cycle no matter the buffer size, you are effectively running at lower COP. You definitely don't want cycling when you need a lot of heat or hotter water, so oversizing can still be an issue."

Agree with both of these statements. I'd add that a heat pump is trying to measure the load and modulate to meet it, and the buffer tank makes the load look bigger than it actually is when heat is flowing into it, which makes the heat pump tend toward cycling.

DCContrarian

"My setup is primary/secondary loop, so the heat pump loop is never dead headed. The only things on it are the plate HX for DWH (didn't see a reason to install a 3 way) and the Ts for space heat."

In other words, the primary loop is the buffer tank!

DCContrarian

https://forum.heatinghelp.com/discussion/comment/1823651#Comment_1823651

@hot_rod and I watched a training from Ener-tech a couple of weeks ago where they claimed with their new heat pump they can serve loads as low as a single towel warmer with no buffer tank. They didn't explain how they did it. I kept looking for a hidden tank somewhere but couldn't find it.

hot_rod

the buffer tank for HPs add cost and piping labor. at leads 3 manufacturers include the tank in the package. Several brands have reverse indirects in their offering, do dhw and buffer comes together. So the HP designer must know they add not subtract from the performance, efficiency and longevity?

It seems Entertech is the only brand suggesting you do not need a buffer.

Just as mod cons evolved so will the heat pumps as far as modulation, and the option to go buffer-less to the load

DCContrarian

My thinking has evolved to the point where I view buffer tanks in a heat pump as a necessary evil, only necessary under low-load or no-load conditions to protect the compressor, and evil the rest of the time. Now @Kaos and Ener-tech have me thinking they may not even be necessary, that if you use primary-secondary piping to provide a flow path when all the zone valves are closed you only really need a few gallons in the piping to protect the compressor. I need to stew on that for a while.

hot_rod

i would have to see the piping also. Often in closely spaced tees like a hydr sep, you are blending temperature . So the hp need to run a higher temperature to deliver the necessary SWT

Or you could be mixing up the rwt to the hp

And with variable speed circs on one or both circuits it is even harder to predict those mixed temperatures.
That is where the direct to load buffer makes sense, whatever comes from the hp goes to the load

Screenshot 2024-12-01 at 6.22.42 PM.png

Kaos

The Ts for the space heat are staggered, supplies first for all zones and returns after. Not quite closed spaced Ts but I'm running IFCs on all the pumps so perfect hydraulic separation is not an issue. This avoids the problems with mixing. I have also set the overall zone flow rates to be a bit bellow the heat pump max so water is never circulating backwards in the primary loop which could cause mixing. Hydraulic separators do seem like not the best idea here.

I do agree direct to load with a differential bypass is the way to go with heat pumps. Your reference zone can serve as the buffer and usually most of the water supplied by the unit will be used by the emitters. The problem is in north America TRV just cost way too much, so I doubt they would make financial sense (never mind the pain to retrofit them).

DCContrarian

https://forum.heatinghelp.com/discussion/comment/1823783#Comment_1823783

See my comment above (I don't know how to link to a comment but it's the sixth one on the thread.) Basically, with three-pipe plumbing you don't get much mixing in the buffer tank, it stabilizes at either the leaving water temperature or the return temperature, depending on whether the heat pump flow or the emitter flow is greater. With primary-secondary piping you'd get the same thing, it would just stabilize more quickly. I think that's a good thing because once the temperature is stabilized the heat pump is seeing the actual load and can modulate accordingly.

It's generally better to have the heat pump circulator flow be higher than the emitter circulator flow, that way the emitters see water coming straight off the heat pump, unmixed. But I don't know how you guarantee that when the heat pump is modulating its flow to match output.

hot_rod

The heat pump, mine at least, seems to work well as the compressor and variable speed pump in the IDU communicate well. I hear that pump ramping up and down all day long

My distribution is simple, 1 radiant zone constant circulation. I have not set ODR, 115

I can see if you require multiple temperatures and have multiple zoned, including micro zones , it will be more complicated maximizing the system for peak efficiency.

For me the definition of primary secondary is closely spaced tees. Without properly installed closely tees you don’t have primary secondary piping. Separators are a version of PS in a convenient package

We get pictures and drawings constantly of systems to trouble shoot, that are believed to be P/S, and are not even close

That is why we put together idronics 19, the most common piping mistakes are in there , with suggested fixes

https://www.caleffi.com/sites/default/files/media/external-file/Idronics_19_NA_Proven%20hydronic%20distribution%20systems.pdf

Kaos

You have to be careful with heat pumps. The biggest mistake there is mixing water and lot of the standard topologies are prone to it. I would like to add the Kaos special as a reasonable compromise. It doesn't have proper hydraulic separation. The spring checks seem to keep it under control but even if there is a little bit of a bleed into zones that are off is not much of an issue when most zones are on a good part of the day.

The check at the end of loop is not really needed if zone flow is set to be always bellow loop flow.

You can also replace the zone pumps with TRVs and the loop spring check with a differential bypass for a direct to load setup.

Open to other ways of piping these up.

hot_rod

P-1 is the circ supplied with the indoor unit? A variable speed pump that modulates with the compressor modulation?

If so I think it runs all the time, modulates from maybe 3 gpm to max output, 8 gpm or more?

As I see it without hydraulic separation P1 is in series with your circuit 1-3 circs? So the flow rates to those circuits is fairly unpredictable? So the temperature of the purple return line is always changing also, depending on which zones and what their load is.

Your loads are in series, so temperature drops across the three. That temperature drop is dependent on which zones are calling. Supply temperature to #3 depends on if 1 and 2 are operating. If only #1 is running, #3 sees a different SWT. There is no way all 3 could ever get the same SWT. As the color of the SWT line indicates :) \

There is a complicated piping to equal out the SWT to all 3 circuits with crossovers and balance valves.

Even if you have a high, medium and low temperature requirement in those 3 circuits, SWT depends on which circs are running.

If you want or need a fixed gpm, you could add a dynamic balance valve at each of those circs, a PIC valve. With a fixed flow there you could better predict the mixed temperature. But temperatures are still a moving target.

What does the pressure gauge show as various pumps turn on and off.

It seems the PONPC for the circuit pumps is through a high pressure drop HP.

Or maybe the circuit pumps are pumping at the PONPC? When that happens you can pull a negative, sub atmospheric condition somewhere out in the piping. I guess if P1 always runs you are always add some ∆P. If P1 is a variable speed circ the ∆P is changing. A large dial 30 psi gauge is good for observing this.

More on pumping at the PONPC in the attachment.

It would be interesting to add a bunch of Petes plugs to see what pressure is doing throughout the loop, including out at the heat emitter, that is where I expect to see a negative pressure. Maybe?

I have to sleep on this one.

If it works to your desire…

Screenshot 2022-12-13 at 8.43.11 PM.png

Screenshot 2023-10-21 at 1.23.32 PM.png

Screenshot 2024-12-02 at 8.45.12 PM.png

DCContrarian

https://forum.heatinghelp.com/discussion/comment/1824165#Comment_1824165

Well, that answers the question I posed above:

"It's generally better to have the heat pump circulator flow be higher than the emitter circulator flow, that way the emitters see water coming straight off the heat pump, unmixed. But I don't know how you guarantee that when the heat pump is modulating its flow to match output."

With the check valve, emitter flow can never be greater than the heat pump flow, it it is the check valve closes and the two flows are forced to be the same. I'm not sure which way the forcing goes. If heat pump flow is greater the check valve opens and the excess bypasses the emitters so the return water is mixed and the heat pump doesn't see as much delta as it would if the flows were matched. But the important thing is the heat pump sees the actual load and can modulate accordingly.

Kaos

Not running a loop check and have not had issues with flow reversal.

@hot_rod P1 is the modulating pump inside the monoblock. The heat pump runs it based on delta T (min target around 3C, 5C at full load) which is much smaller than I expected. This does mean that there is a fair bit of flow from P1 so with reasonable emitter temp drops the zone flow is always bellow that.

I initially had the check valve in as per my diagram but the one I had was too restrictive and was getting low flow errors. After removing it, the setup still runs just as well, so I would put it as optional.

Because the loop flow is always above zone, the supply to each zone is at the same temp as the heat pump output, each zone can run without issues.

In my case the zone pumps feed manifolds so it is a much higher resistance flow path than the main loop. Without the zone pumps running I don't see any flow on the manifold gauges so ghost flow is low enough not to matter.

There is no cross talk that I can see on each zone. They all run at their designed flow rate regardless of what the heat pump or the other zones are doing. I know it looks like series but works pretty close to how primary/secondary should run. Or at least close enough.

@DCContrarian I'm still trying to figure out if return water temp matters much. My gut feel says yes but would need a proper BTU meter setup to check. Not in the cards any time soon. With the current outdoor reset config, there are pretty long runtimes and overlap on all the zones, so the mixing is minimal. The RWT is pretty close to what comes out of each return manifold.

hot_rod

https://forum.heatinghelp.com/discussion/comment/1824204#Comment_1824204

does your display on the IDU show flow rate? On my diagnostic screen it shows gpm and also compressor operating condition. Without some gauges it is a guess as to flow rates and temperature mixing along your loop

Your manual must show max gpm they want though the ODU?

With some calculating you could define the system curve, Lay it over the pump curve and see where the operating point falls. Same with the HP loop

With that info you could define temperature conditions idronic 26 shows an example how to develop system curves

My 2 ton has about 6’ head at. 8 gpm through the outdoor unit HX, so it is small Wilo ECM.

Do you have a meter to monitor power consumption? A GEO contractor I know puts an EKM meter on all the systems to gauge power consumption against loads. Unless your control tells you this info?

IMG_9084.jpeg