Welcome! Here are the website rules, as well as some tips for using this forum.
Need to contact us? Visit https://heatinghelp.com/contact-us/.
Click here to Find a Contractor in your area.

Circulator Pump Selection for Fan Coil Units - Please Help

E_Guy
E_Guy Member Posts: 8

I'm a newbie here. I'm need to select a pump, presumably a circulator pump, for my house, for 3-4 fan coil units(FCUs), to be used for both cooling and low temp heating. (My system is a low temp(~120deg) AWHP & buffer tank, supplying a radiant floor, and I'm now adding the FCU's.). For FCU's, I estimate I will need ~6gpm and ~20ft of head, and system will only be used occasionally, say a few weeks per year for each heating and cooling. What is an appropriate pump for this? I see Grundfos Alpha1 15-55, or Alpha1 26-99, which is what the Grundfos site (quick selection feature) gives me, but I don't really know if these are appropriate. I assume variable speed pump would be appropriate, but again I"m not sure. Please comment.

Comments

  • Alan (California Radiant) Forbes
    Alan (California Radiant) Forbes Member Posts: 4,217
    edited May 27

    That's a pretty high pressure drop. Are you adding all three or four together?

    8.33 lbs./gal. x 60 min./hr. x 20°ΔT = 10,000 BTU's/hour

    Two btu per sq ft for degree difference for a slab
  • EBEBRATT-Ed
    EBEBRATT-Ed Member Posts: 16,507

    What @Alan (California Radiant) Forbes means is what is the head loss for each air handler? The total head the circulator needs to overcome is the air handler with the highest head loss not all of them added together.

  • hot_rod
    hot_rod Member Posts: 23,416

    look up the spec sheet on the FCU, here is an example.

    That will give you all the info, flow, head, and required SWT for the desired heat output.

    Then the pressure drop for the piping, valve, boiler, etc.

    This Idronics issue takes you through the steps.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • EdTheHeaterMan
    EdTheHeaterMan Member Posts: 9,407
    edited May 27

    If a pump can handle the highest head zone, then it will handle all the other zones that have a lower head.

    The part of the pump performance curve that you Add Together is the total of all the gallons per minute of each zone.

    If the longest of most restrictive zone or loop has a head requirement of 6 ft, then you need a pump that can overcome 6 ft of head. That's it.

    But if each zone required say 4.5 gallons per minute and you have 4 zones, then you need a pump that will move 18 GPM at 6 ft of head.

    A variable speed pump is best for your situation since all the zones may not be calling at one time. so you need a pump to handle say 4.5 GPM at say 5 ft of head for a FCU that is right next to the boiler, and you will need say 4.5 GPM with 6 ft head another time, when the FCU that is the farthest away is calling. That variable speed pump can be set at a ∆T or a ∆P based on your circumstances to produce the proper results whenever all 4 zones are not calling. But it must be able to handle the 18 GPM at 6 ft head when all four zones are calling.

    Does that make sense?

    Edward Young Retired

    After you make that expensive repair and you still have the same problem, What will you check next?

  • E_Guy
    E_Guy Member Posts: 8

    Ed, Thanks - that's very helpful. I understand your gpm and ft head examples (see footnote).

    Apart from the gpm and ft head, my real question is pump selection. You agree variable speed pump is appropriate. Assuming my 6gpm & 20ft head is correct, I mentioned I identified Grundfos Alpha1 15-55, or Alpha1 26-99. Are these reasonable? Price range is roughly several hundred dollars, perhaps $300-500 range. Is that expected ballpark? I have heard of Compass H models too - seems similar price/gpm/fthead. Are these reasonable options too? Any others brands of models suggested, or less expensive options?

    This part of my hydronic system is closed loop, so I assume cast iron is ok? Do I need to worry about 'wet rotor' or other details? Thanks for mentioning the ∆T option - I didn't know about that - just googled it, now I know. (I infer the Grundfos Auto-Adapt feature may be their fancier version of ∆T mode.)

    Footnote: Concerning your gpm & head example, apart from suspected typo, I think you were saying for further located FCU case, one would have (for example) 5ft hd (from FCU) + say 6 ft hd (from line losses) giving net of 11ft hd. Is that what you meant? My case of 6gpm and 20ft head value I believe is accurate, with the slightly high head due to assuming 3/4" PEX (and 100ft round trip & some elbows)- although I may be able to go to 1"PEX.

  • EBEBRATT-Ed
    EBEBRATT-Ed Member Posts: 16,507
    edited May 28

    @E_Guy

    What you need is an accurate calculation of the head loss and GPM requirements. Start by determining the AHU with the coil with the highest head loss Calculate the pipe and fittings needed to connect that one loop to the boiler or to the common header.

    Then repeat for the other AHUs (chances are the AHU with the higher loop will be the loop with the most resistance) but not always.

    Once you have this done calculate the boiler and any common piping with the total flow of all the zones. Take this resistance of the common piping and add the head of the highest AHU loop to that. That is you total head loss

    Now take the total head loss calculated and the total GPM required by all the AHUs. This is what your new pump should be selected from.

    I would suggest downloading the B & G system sizer.

  • hot_rod
    hot_rod Member Posts: 23,416

    6 gpm for 4 FCU? How did you come up with that number? That would be 1.5 gpm per unit. If they are all the same size.

    Start with the spec on the FCU and work backward.

    Look at the Alpha 15-58, it is replacing the 15-55 and has at bit more performance and features.

    26-99 is quite a bit more pump, that you may not need.

    Add a magnetic separator if you use an ECM type circulator.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • E_Guy
    E_Guy Member Posts: 8

    Folks

    Thanks for the feedback. Hot_Rod, good - the Alpha 15-58 seems ideal for me. (I noticed this model after I posted above comments.). Currently this is my top choice of pumps.

    As there was so much feedback on my gpm+head estimates, I tuned my head estimates, and now have 7.3gpm, and 12.9ft head on the worst zone (this all for 1" PEX and four FCUs). (My earlier numbers were assuming 3/4" PEX, but I will plan to switch to 1".)

    My full table for ft head calc (and gpm) is here:

    and here is spreadsheet with full table plus all referenced data from FCU's, PEX, etc:

    Real head will actually be 5-10% higher due to 10% glycol and a few more unions, not included in above table. Please let me know if there is still concern with my gpm & head estimates.

    EdTheHeaterMan
  • Zman
    Zman Member Posts: 7,611

    You will probably have some balancing issues with the range of head loss of the different FC's. Adding a balancing valve like the Caleffi quick setter will help if the flow rates are critical. https://www.supplyhouse.com/Caleffi-132559AFC-3-4-Sweat-QuickSetter-Balancing-Valve-w-Flow-Meter-2-0-to-7-0-GPM-Lead-Free

    "If you can't explain it simply, you don't understand it well enough"
    Albert Einstein
    Alan (California Radiant) Forbes
  • Alan (California Radiant) Forbes
    Alan (California Radiant) Forbes Member Posts: 4,217
    edited May 29

    The Alpha I 15-58 is the ideal choice. Even though the performance curve shows missing your calculated target, I don't think you will notice a difference. Also, the actual performance of pumps always seems to exceed the published data.

    8.33 lbs./gal. x 60 min./hr. x 20°ΔT = 10,000 BTU's/hour

    Two btu per sq ft for degree difference for a slab
  • EdTheHeaterMan
    EdTheHeaterMan Member Posts: 9,407

    I'm happy with your tuned calculation. The FCU pressure drop plus the longest zone pressure drop total plus a little extra, sounds good for your Pump Head.

    Since some folks come to the supply house asking for a pump that can move 6 GPM, with 50 ft of head because it is going into a 5 story building, when we know that all he needed was a pump that needs to move 6 GPM at maybe 9 ft head. Not all folks understand what Pump Head measurement is. We just wanted to be sure you understood that if you have a zone with a 4 head requirement plus a 6 head requirement plus a 5.5 head requirement and a 4.5 head requirement that you don't add up all the zones to get a total of 20 head.

    Just makin' sure you got it , and it appears that you do. 👍️👍️👍️

    Edward Young Retired

    After you make that expensive repair and you still have the same problem, What will you check next?

    Zman
  • E_Guy
    E_Guy Member Posts: 8

    ZMan - ok, yeah I was worried about that. I was considering using Caleffi twistflow manifold, which if I understand has flow regulators. Would use of that make sense for balancing? Any problem with the Caleffi manifolds?

  • Zman
    Zman Member Posts: 7,611

    It looks like the twistflows max out at 2 GPM. If you have the ability to home run the zones, there may be a manifold option out there. @hot_rod can walk you through the options and associated restrictions.

    "If you can't explain it simply, you don't understand it well enough"
    Albert Einstein
  • hot_rod
    hot_rod Member Posts: 23,416

    If you want super accurate control for flow, I would consider a dynamic ( automatic, or pressure independent PIC) type balance valve. A Caleffi 127. This valve adjust automatically as the zones open and close, unlike a manual balance valve.

    Here is a graphic example of both under different zone valve operation conditions.

    Build a copper header or manifold, add a zone valve and 127. Or the zone valve and 127 could be at the FCU. That is how some of the factory FCU unit ship to large job. The ZV and flow control is built onto the FCU, sized and labeled to the engineered spec.

    A dynamic balance valve is an excellent match with a delta P pump logic. As the valve modulate, zone come on and off, the circ reponds by varying its flow rate. So you never over or under pump, and use the least amount of pump power.

    Usually this degree of balance accuracy is seen on engineered designs where maybe dozens, hundreds, or thousands of FCU need exact control. It can make a big difference in performance and comfort, as well as efficiency with many FCUs.

    Certainly no harm in adding auto balance valve on your project. Include a Y strainer or DirtMag to keep them clean.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • DCContrarian
    DCContrarian Member Posts: 686

    Looking at your spreadsheet, it looks like you're using a head drop based on 7 GPM but the FCU's are only rated for 1.6 (and the Chiltrix website says 1.3 GPM for the CXI65) https://www.chiltrix.com/chiller-fan-coil.html

    Using the calculator here ( https://www.plasticpipecalculator.com/PressureDropHeadLoss.aspx ) with a developed length of 116 feet with 3/4" PEX I get a drop of 1.56 feet of head at 1.3 GPM.

    I assume since you're using Chiltrix FCU's you're installing a Chiltrix heat pump as well? Are you going with a CX34/35 or a CX50? The smaller one is only two tons and probably won't be able to serve the four FCU's you have listed. The CX50 will.

    There is a Chiltrix user group on Facebook, https://www.facebook.com/groups/chiltrix.owners You should join that, there's a lot of firsthand knowledge there.

  • E_Guy
    E_Guy Member Posts: 8

    Thanks all.

    ZMan, Thanks for pointing out the 2gpm max scale an Caleffi - so then question is, as that is only the meter limit, is it the case that one cannot achieve >2gpm on any zone, or rather that the balancer may allow higher flow but the meter simply cannot measure that higher flow level rate? I suspect it is the later, yes?

    Anyone - I've heard some talk that Caleffi twistflow or similar manifolds are not worth it, or not ideal, I think bcs the meters may easily cloud up (or fail), and the valves may be noisy, or maybe there are other issues? Any comments on this?

    I don't need "accurate" flow control, but I want some rough idea what the flow is thru my different zones. One of my FCUs has 2.5x higher flow impedance (more pressure drop), and needs almost 2x higher gpm than all the other FCUs. Balancing valves for sure will help with that, but without a meter, I really have no idea whats going on (other than from the room's thermometers/thermostats). So any arrangement with flow meters would be nice. This Quicksetter meter seems nice(https://www.supplyhouse.com/Caleffi-132559AFC-3-4-Sweat-QuickSetter-Balancing-Valve-w-Flow-Meter-2-0-to-7-0-GPM-Lead-Free

    ), it can measure the higher flow (2-7gpm range), but a little pricey. The balancing valve here seems to be a ball valve - I've heard ball valves for balancing will be noisy, and somehow problematic. What is your experience? It was suggested to me use a high quality gate valve instead. Any comments?

    DCContrarian, Thanks for the Chiltrix facebook link - I had no idea. Will check it. I have a CX50, and I have a fair amount of data on COP, COP vs compressor speed, and COP vs fan speed, COP vs Temp, and Noise vs Fan speed. I need to plot it up, and I will share it there. Its low on my priority list, but I will do it within a few months.

  • jumper
    jumper Member Posts: 2,385

    I've put pumps on each fan coil. For cooling you want less than recommended maximum flow through chiller when all coil pumps are on. And hopefully more than recommended minimum flow through chiller when only one coil pump is on. If zone control only turns off fans then return water to chiller can be chilly which I don't like. If you decide that you need a system pump as well then you need a shunt (bypass) somewhere far away from chiller and a low head/high flow pump somewhere.

  • DCContrarian
    DCContrarian Member Posts: 686

    Here's the problem you're going to run into. Zone 3 requires 2.5 GPM, the FCU drops 10.7 feet of head at that flow. The 76 feet of 3/4" PEX is another 3.18 feet, so 13.9 feet total.

    If all four zones are on the same circulator, they all have to be on the same manifold so they all have to see that 13.9 feet of head.

    The CXI65's on the other three zones drop 4.3 feet of head at 1.3 GPM. Using the flow calculator I linked to earlier, I see that's equivalent to 320 feet of 3/4" PEX. Taking the longest zone with 116 feet of pex, that's a total of 436 feet of PEX. At 13.9 feet of head I get a flow of 2.1 GPM through 436 feet of PEX. So if you apply enough pressure to get 2.5 GPM in zone 3, you're going to get at least 2.1 GPM in the other zones. So you need 8.8 GPM total at 13.9' of head.

    Because zone 3 needs so much more head, it increased the flow through the other zones. You could try choking the other zones down with balancing valves, or even just piping them with 1/2" PEX.

    If the zone 3 is on its own circulator, it needs 2.5 GPM at 13.9', and the other three zones need 3.9 GPM at 5.86'. That might be more attainable.

  • E_Guy
    E_Guy Member Posts: 8

    Hi DCContrarian,

    Yes the Zone 3 pressure drop (PD) is much higher than the other zones. Right, thats my issue. So yes I was planning to choke down the other zones to get reduced flow into those zones. I only expect to be using these FCU's a few weeks per year at most(due to our mild climate here), and maybe not running all zones, so my thinking is that this is a reasonable approach. If I understand the Alpha 15-58 wattage spec (in my excel file) then the pump would be running at ~37watts max, more likely 10-30watts most of the time (fewer zones on, at lower flow), so the kWh penalty of choking down with balancing valve is really small/negligable. With a Caleffi twistflow or equivalent manifold, should be easy to choke down the flow, right?

  • hot_rod
    hot_rod Member Posts: 23,416

    I think the Alpha 15-58 is the correct choice. With all zones open if the spec is 8.8 gpm, 13' head?

    The autoadapt may regulate the pump accuratly enough without balance valves. If not, three 2.5 gpm PIC type balance valves would give you the exact flow at any point in time.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • DCContrarian
    DCContrarian Member Posts: 686

    Go to the head drop calculator I linked to earlier, and play around with some numbers. What you'll see is that small changes in GPM lead to large changes in head drop, it's highly non-linear. So there's not going to a single balancing valve setting that leads to the desired flow in all conditions.

    This is the problem that constant pressure pumps were designed to solve, the pump itself keeps the manifold pressure constant under all flow conditions. This will give some zones more flow than you need, you can choke them down a bit to reduce the flow if you want.

  • hot_rod
    hot_rod Member Posts: 23,416

    The pressure independant type of balance valves will have a PS

    ID range stamped on them, or assigned to the valve. So in the case of this Caleffi 127, as long as the circ can develop between 2 and 32 PSID the valve will accurately regulate the flow. This will be the case with fixed speed circulators or with delta P mode on ECMs.

    The 2-32 PSID should cover a wide range of circs.

    A PICV is a pressure independent control valve that can also act as the zone valve. It can be have a typical on/off actuator, or a modulating type actuator driven by a BAS, for example.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • DCContrarian
    DCContrarian Member Posts: 686

    Note that the CX-50 comes equipped with a Grundfos UPMXL 25-124 which is capable of 10 GPM at 24' of head.

    There is a pump curve on page 12 of this document:

    https://www.chiltrix.com/documents/CX50-2-IOM.pdf

    With FCU's, you should be able to let the water circulate continuously and modulate their output by modulating the fan speed. So no zone valves, no external circulator, no buffer tanks. I'd verify that with Chiltrix support and maybe bounce it off the Facebook group.

  • hot_rod
    hot_rod Member Posts: 23,416

    I would use a buffer tank as they recommend. So the variable speed circulator they include just handles the heat pump. The heat pump sees the buffer tank as it,s load, not the system, per say.

    The secondary circulator gets sized just to the distribution loads, as you have already calculated, and will be much smaller.
    Although their schematic doesn’t seem to show a circ for the FCUs?. But one on the radiant?

    And I would not have the back up heat piped that way, no need to heat the heat pump with a backup?

    I’m not sure what they mean by ‘wild” coil loops. If that is a 4 or 6 pipe buffer, I don’t see any flow getting to the FCUs. The buffer is a hydraulic sep.

    I’d use a dirt sep not a y strainer as strainers can plug and reduce flow if not serviced properly.

    I would send one of the Siegenthaler piping designs from Idronics to them for approval.

    One last thought, you should not use pvc or CPVC with glycol. Glycerin is used in CPVC as a freeze protection fluid, at least in fire protection systems.


    other than that🤔

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • DCContrarian
    DCContrarian Member Posts: 686

    "Wild" is a term that Chiltrix uses, I think it's unique to them, to describe a FCU that has no zone valve and uses no circulator other than the one built into the heat pump. It's the simplest possible configuration and the one I think is common in Europe, where their units are designed for.

    The idea is you have all of the FCU's attached to a manifold and then they're in parallel. Water flows continuously through all of the FCU's. The thermostat on each FCU determines whether the fan runs, which determines how much heating or cooling the FCU produces.

    The Chiltrix heat pump has no sensors inside the house. It circulates water continuously and monitors the return temperature of the water to sense the demand for heating or cooling. Both the compressor and the circulator inside the heat pump are variable speed, the compressor speed is modulated to match the demand and the circulator speed is modulated to match the compressor output while maintaining a 10F delta between the output water and the return water.

    The compressor is able to modulate to a minimum of about 25% of rated output. If the demand is less than that, it will cycle on and off. In heating mode, it will sense that it needs to cycle by the water temperature rising even thought the compressor is at minimum. If the temperature rises 2C (3.6F) above the set point, it cycles off, and stays off until the water is 3.6F below the set point. Cooling is just the reverse.

    In order to avoid short cycling, in "wild" mode you need to have a certain minimum amount of water in the system, for the CX50 Chiltrix says 15 gallons. With a 7.2F swing that's 900 BTU's, which is about 4 minutes of output for the CX50 at 25%. Note that the smaller CX34 also says 15 gallons, for that unit it's about 6 minutes of output. I would think the CX50 would do better with maybe 20 gallons of water.

    Your system looks to be only a few gallons so you'd have to include a tank for extra capacity.

    "Wild" mode can't really be mixed with radiant loops or any heat emitter that relies on zone valves for control.

  • hot_rod
    hot_rod Member Posts: 23,416

    The term wild loop is not unique or new to me. It appeared in the IBR Hydronic Institute trainings back in the 1980's, when I took the course. Theyn showed fin tube wild loops.

    It is as you state a loop that has no on/off controls, it flows whenever a circulator is running.

    I'm having a hard time with their explanation of the buffer/ hydraulic separator VCT 37 tank. It reads like something the marketing department put together :)

    They explain it as a hydraulic separator and that the radiant loops need a dedicated circulator as shown. I agree.

    But ignore that concept for the FCUs?

    So how does the fan coil circuit get any circulation as the same hydraulic function would apply. How do they size their built in radiant pump? Without any loop design data? It could be a 1 gpm load or a 40 gpm load.

    The pressure drop from the upper to lower ports on the left, HP side will be near zero. So flow from HP would just return to the HP.

    There must be some head involved with the FC units as the math you worked through above shows 10' head at 2.5 gpm for the zone 3 example.

    Here is an example of how the math works out on this 2" Caleffi HydroSep. The diameter of the tank in relationship to the diameter of the in out ports is how separation is accomplished. The "barrel want to be at least 3 times the diameter of the inlet/ outlet piping.

    With a 40 gpm flow in and out of the tank, there is no flow or movement to the distribution side.

    In fact air rises up through the tank with a 40 gpm flow going through it, due to the diameter 9A area ratio. Dirt drops for the same reason.

    It looks like their tank far exceeds that ratio?

    I've built a number of clear hydraulic separators to show how this actually works.

    I've also tested brands that don't do as they advertize due to the wrong barrel to connection ration. I have videos to show how secondary flow happens when you exceed the 3-1 ratio, by injecting food coloring.

    Unless there is a jumper pipe inside, I don't see how any flow happens in a FCU that has a head associated to the flow rate.

    Here is their parable of buffer/separation functions.

    Bob "hot rod" Rohr
    trainer for Caleffi NA
    Living the hydronic dream
  • DCContrarian
    DCContrarian Member Posts: 686

    So after all this discussion, here's the $64 question: Why not just use minisplits instead?

    In "wild" mode, a chiller plus a FCU is functionally equivalent to a minisplit. What you've laid out is four heads, three half-ton and one one-ton. A minisplit version is something that any HVAC tech in America can install and maintain. If low maintenance is important to you this is a big advantage.

    Bill of materials, from the Chiltrix website:

    CX-50 heat pump: $7989

    CXI-65 FCU, 3 @ $939= $2817

    4HBC3: $897

    Total: $11,703

    Plus probably at least another $2000 for hydronic fittings.

    You can get a comparable Mitsubishi system for less than half that:

    https://hvacdirect.com/mitsubishi-36-000-btu-19-seer-quad-zone-heat-pump-system-6-6-6-12-id8332.html

    You can get a really nice Mitsubishi Hyper-Heat, 19 SEER system for $7725

    https://hvacdirect.com/mitsubishi-42-000-btu-19-seer-quad-zone-heat-pump-system-6-6-6-12-id8427.html

  • DCContrarian
    DCContrarian Member Posts: 686

    Forgot about the 19 gallon buffer tank, another $1149.

  • E_Guy
    E_Guy Member Posts: 8

    A few comments. I have a buffer tank, the 19Gallon Chiltrix one. A circ pump is absolutely needed on the load side of buf tank, for both radiant or FCU. Chiltrix doc say (at various points) that the diagrams do not show all needed components. Without a pump on load side, there would be no circulation. My wye strainer has indeed been blocked ; I agree the dirt remover would be preferable, good idea - the local rep suggested that to me too. There should be no problem with the right glycol in the closed loop system - it is recommended by Chiltrix. About mini-splits, there are several reasons I don't go that path. One, I would need to find space for another HP - I don't have that space. Two, I have a perfectly good performing AWHP (CX-50) with buf tank already (supplying my radiant system) , so I'm just tapping into that. I will start with 1 or 2 FCUs, and easy to add on down the road if it all works out well. And with monobloc there is less risk of refrigerant leaks, a climate concern of mine; also I know how to do water lines, I don't want to mess with refrigerant lines. (Also, I don't like the optics of the minisplit heads, and I plan to 'conceal' my FCUs.). About my floor radiant, while I can use it for cooling, it is circa 1959, buried below serval inches of concrete slab; it is inefficient for quick cooling, and quick cooling (for a few hours a day) is my dominant use case. It's almost as inefficient for winter heating, but the radiant for heating is nice, and it works, and we want to keep it.

  • DCContrarian
    DCContrarian Member Posts: 686

    OK, didn't realize you already had a radiant system. Strike everything I said in my previous three posts, "wild" mode only works if the FCU's are the only emitters and all of that only applies to wild mode.

    If you're using radiant tubing you have to have a zone valve and if you have a zone valve you have to have at least one additional circulator and a buffer tank.

    One thing to look into is 3-pipe plumbing at the buffer tank. The supply from the heat pump goes into a tee, one leg of the tee goes directly into the circulator and the other into the buffer tank. When the heat pump is running the circulator pulls water directly off of the heat pump, which in heating mode is going to be about 10F warmer than the water in the buffer tank. Since heat transfer is all about temperature delta, and the temperatures you run are relatively modest, this makes a significant difference in your output.

    In cooling mode the water will be cooler. This means the FCU's will do a better job of removing humidity, which in my climate is a crucial consideration.

    I believe Space-Pak has a diagram for 3-pipe in their manual.

  • jumper
    jumper Member Posts: 2,385

    I believe that there's circulators with built in check valve so that when pump is off no water flows. My solution of dedicated coil pumps is simplest; on/off is not perfect but it is adequate A/C control. Shutting off fan only aggravates problem of cold water return.