GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

How do I get and keep air out of my geothermal loop?

[email protected] | Posted in Mechanicals on

We have a “closed loop” geothermal heating and cooling system. 97 wells 300 ft. deep. 4 story building. 36 water source heat pumps. 72 electric water pumps. 219 ball valves.
There is no way at the high point (attic over 4 story) to release air and no easy way to add water to displace the air. The contractor keeps trying with their big pump and tank trailers to purge the system by driving the air bubbles out through the basement. No surprise to me, having observed the bubbles in my bathtub travelling upward, this approach keeps failing.
Our piping is very convoluted and there are many branches. I don’t see where anything with 219 brass ball valves and 76 electric water pumps could be considered a “closed loop” that won’t regularly need water in and air out. I suspect that they have designed it that way to cause dependency on their big tank and pump trailers to do any service work.
Seems to me, a pipe delivering city water through a couple check valves in the basement, and a high point air release valve (that lets a bit go whenever its paddle isn’t floating) would do the trick. Then changing out a pump, valve or heat exchanger would no longer require calling the guys with their giant operation to come from the big city to put a couple gallons back in. That and it would end the repeated failures which are most inconvenient. Is there a flaw in my logic? Has anyone had a similar problem? What is the solution here? Is chasing the air through 72 branches of piping, through little intricate heat exchanger coils, from the attic over the fourth floor and all the way down and out through the basement really the best way to eliminate trapped air? There are a lot of deareators on the market, is there one in particular I should be considering? Thank you all.

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. GBA Editor
    Martin Holladay | | #1

    This question is beyond my expertise. But I'll share a few reactions:

    1. Instinctively, I agree with you. The way to get air out of a closed loop is to introduce fluid into the loop while opening a valve at the highest point of the loop to release any air in the system.

    2. I know that residential ground-source heat pump systems often have commissioning and maintenance problems, due to the complexity of these systems and the fact that they are cobbled together on site as one-off systems. Your example makes me realize that these problems don't go away when these systems are scaled up and installed in a large commercial building.

    3. I'm wondering how much this system cost to install. If I were a well driller, I would have liked to have the contract to drill 97 wells, each 300 feet deep. That contract wasn't cheap...

  2. charlie_sullivan | | #2

    I'm not familiar with how large commercial geothermal systems are done, but residential systems typically have a "flow center" that serve to release air, while also providing a small tank to provide a limited amount of make-up water, and to provide expansion space to allow for thermal expansion and contraction of the water. An example of such a unit, randomly chosen--I don't know which are best.

    They aren't connected directly to city water for make-up for several reasons:
    1) The typical design is not pressurized.
    2) Most run a low-concentration glycol mix in the ground loop to ensure it won't freeze in a worst case scenario, even if the design temperature is above freezing (as it should be).

    Perhaps also because if you have a leak you want to know about it.

    I'm not sure whether a large system like yours should have a huge flow center equivalent with a big tank, or should have many little ones, but it should have something to provide a way for air to get out, provide expansion space, and allow for putting in a little make-up water. And that should be located at or near the high point in the system.

    Getting the air out in the first place is still a challenge with such a complex system, but it should be set up to take care of itself after that.

  3. [email protected] | | #3

    Thanks Martin Holladay for the input. I don't know off hand what the system cost. It wasn't cheap. What blew my mind was that they drilled 97 wells, dropped the Poly Pipe loops in them and poured the well caps all in 3 days. They made a little money. The building is Historical so there was a lot of labor involved in hiding 36 Heat Pumps and all that piping in such a manner as to not detract from the historical fabric of the building built in 1911. I'm sure that cut into their profit margin a little.

  4. [email protected] | | #4

    Charlie Sullivan, Thank you. Our system clearly differs substantially from what you are describing. For instance, what you're describing sounds like it would work. Our system is pressurized and all the pumps are inline pumps with no tank to allow separation of air. There is only one itty bitty little expansion tank (maybe 3 gal.) in the system.There is no glycol mix in the loop at all. The system is clearly not set up to take care of its self after the initial purge. There is currently no way for air to get out without the trailer full of equipment and no make up water or reservoir at all. If I ever have to change a pump or a valve or a condenser coil I would have to call them back with their two huge trailers full of purge equipment and 3 men for an entire day. I do really like the idea of the makeup water being stored in the system in a tank. Even pressurized I could do that. Just put an in line vessel in the system piped out near the bottom. It would start out full and then eventually have a head of air in it. I could inject a shot of city water in through a garden hose if need should arise. The top of that tank would be a good place to put a bleeder valve to get rid of air too. If I place the tank in the tower (the high point) then it would double as the high point bleeder for the whole system. I'll have to refine this idea a bit but that could work. Thank you.

  5. user-2890856 | | #5

    Todd ,

    Is it possible for you to post a few pictures of the areas around a couple of these Heat pumps and include the piping and pumps ? Air elimination is the most neglected part of many systems from design right through install and elimination at other key points gets totally overlooked . Valves pumps and fittings and even air elimination devices allow alot of air into the system . Initial purging is a key factor and many don't do it well , pump sizing also comes into play , pumps must be sized properly for the worst case head loss of the system and fluid velocities in a building such as yours must be able to move entrained air around the system . Position of air elimination devices is also an issue since your air will never be entirely removed until you reach design temps . In other words , the location will vary system to system and placing all your air elimination right near the source may not be correct in every instance . When designers understand these factors and spend a bit of time during design as if they were water and air it becomes easier . However , as I have found , when you are sitting there being air and water and moving through a system people tend to look at you funny or ask if this is what they are paying you for . Being able to look them square in the face and tell them , " Yes , this is what you're paying me for " is a forgotten skill .
    Some pictures may help greatly . Why are there so many HPs also , is this a multi unit application where utility bills are the driving factor ?

  6. [email protected] | | #6

    Richard, I'll get pictures next time I'm opening it up for service for you. The reason for so many heat pumps is because the goal is "invisible". This is a 104 year old Historic Landmark and while we want modern conveniences we don't want anything that looks modern. Water Pipes are smaller than air ducts and hiding 36 little units in closets, chases and false ceilings left less visual footprint than other choices. Long term cost played a roll. We are Texans, I don't know if being renewable was any kind of selling point or not. I wasn't here for the decision making. From a maintenance man's stand point this system will be an endless challenge. Everything is jam packed into as little space as possible in a building not designed to accommodate it all. Then there's the factor that I have 14 different sizes of air filters in here ( not even sizes either, 23-7/8 X 19-3/8 X 1-3/4) and most are extremely hard to reach. Then the fancy, fragile drop ceiling tiles that cost over $22.00 each for a 2 X 2ft. and have to be custom cut with a rabbet if I break one. Every unit has a wonderful strainer with a valve to clean it out installed on it but if I open one I've let in air and now I need the huge purge trailers again. Job security.

  7. Guilbault | | #7

    You absolutely need an automatic bleeder valve installed on the highest point of the loop. It should have been required by code (and by common sense). You're never going to be able to refill and operate the system successfully without one.

    But this all sounds very strange... Is everything a single loop? It sounds that way. Or is the condenser water for the heat pumps on a separate loop with the ground coupled side on another loop across a heat exchanger?

    Also, you have 72 pumps... Is that for 72 individual secondary loops (I'd like to see that primary manifold!), or 72 completely separate loops? Are they for the wells (but you have 97 wells...)?

    I can't believe someone would engineer a multi-unit building that requires 72 pumps. That engineer should be drawn and quartered. The initial cost + the maintenance cost + the utility cost... It boggles my mind. There's no sensible explanation for it.

  8. [email protected] | | #8

    Jonathan Gilbault. Basically one loop. The wells are headered together outside and one main trunk line loop runs throughout the building with no water pump directly on it. The 72 water pumps are serving 2 to a heat pump. one in and one out, extracting water from one side of the loop and putting it back in to the other. The 2 pumps run when the heat pump does. Nothing else circulates the water in the loop but the water pumps on the individual heat pumps. I don't understand why water should need to be pumped twice to flow through one heat exchanger coil. Seems to me that one pump at each unit would have worked just fine. In fact, in the attic, the three biggest units each have only one pump which is larger than the rest. Which statement begs a correction on my part. There are really only 69 water pumps (not 72) as three of the 36 HPs only have one water pump.

  9. charlie_sullivan | | #9

    The multi-pump configuration was probably chosen to avoid the power draw of a single large pump running when only a few heat pumps are active. Having one pump per heat pump would mean you'd automatically get the right amount of pumping for each scenario. I imagine one big variable-speed pump with the right controls would be more efficient, but the availability of the right system for that might have been a problem. Still, for the cost of 69 pumps, you could get something custom configured.

    The only reason I can think of for two pumps per heat pump would be if they figured out they needed twice the head that one pump can deliver. That could be a reasonable design, depending on availability of the right size high efficiency pumps at a reasonable cost. It's hard to tell how much analysis was done on that design.

    But the lack of a way to remove air seems inexcusable.

  10. Guilbault | | #10

    Hey Todd, it sounds like the little "secondary" unit pumps also pump the water through the wells by moving it from a supply manifold to a return manifold that all the wells are run off of (otherwise, how do we get flow through our ground loops?). I use secondary and quotes because it sounds like you don't have a primary-secondary hydronic loop design.

    If those are the only pumps you got, that's the only way to design the system. This is a problem (but doesn't really matter for your air).

    On the air side, you just need to get an automatic bleeder valve attached on the highest point of the loop. If that point is not the return manifold, you're also going to want another on the top of the return manifold.

    Regarding the plumbing and the pumps, unfortunately, from what you're telling me it's not a proper primary secondary loop layout. This type of plumbing layout, where you put the load between the supply and return manifolds isn't strictly banned in the code in many places, but no engineer should ever design one. They don't work reliably. Even in an ideal constant flow scenario they are annoying to balance and work on. They are prone to a variety of problems which are hard to diagnose and correct.

    You don't have a constant flow system since the pumps only run when the unit thermostat calls. This makes this loop design much more problematic.

    Your system cannot be balanced, and condenser water flow can't be controlled. As long as the heat pumps are tolerant enough they may work fine. The problem you have is that in some load situations it's entirely possible that instead of pushing/pulling water through the wells, you'll just pull water backwards through the lines of the units that aren't running. Because it's not a primary secondary loop, every line is going to "feel" a push or suction when other lines run. When unit owner 'a' calls for cooling, that cooling call is going to change the pressures at the supply and return manifold, changing the flow at every other heat pump and possibly even reversing it depending on the relative resistances of the ground loops (likely very high) and the heat pump "secondary" loops (also pretty high because you have to spin the pump rotor).

    You can get flow through pumps that are off (you may see rotors turning slowly without power)... You can get reverse flow through pumps that are off, bypassing your ground loops... You can get all sorts of weird problems that primary secondary loop plumbing would have prevented. Hopefully you get lucky and it just works, but a system like this works through luck --not design-- especially with variable flow as pumps turn on and off.

    I hope you don't have too many other problems once the air issue gets sorted. The types of flow issues this type of plumbing design is subject too can be super annoying. If you experience issues that could be related to flow (heat pumps going into alarm on high condenser pressure for example), you may have to convert the plumbing to a proper primary secondary loop.

  11. Guilbault | | #11

    Hey Charlie, it sounds like from your post you understand the issue. Fewer pumps with drives (or even ECM pumps) are (1) more reliable, (2) higher performing, and cheaper to (3) install, (4) run, (5) maintain, and (6) replace.

    There's no way they saved money by adding 69 pumps. Just think: 2 ball valves per pump so it can be replaced and maintained. You can lose a modulating valve with this design I suppose (and probably did), and replace them with balancing valves which would be slightly cheaper when you factor in the controls (although I'd be willing to bet there are no balancing valves at this installation), plus the labour to install 69 pumps instead of 1 pump to make this design work, or 3 pumps to make a proper primary-secondary design work.

    You can't properly commission this system because you've given up control of flow and pressure in the design phase, so I guess(?!?) you save money there. You would want to commission your pumps if they had drives on them.

    We've had VFDs for decades, and ECM pumps for years even in North America. People shouldn't be designing plumbing loops this way. The only people who benefit in this situation are the installers. And frankly, even they probably hated it.

    I'm sure the engineer isn't a bad guy. But this shouldn't have been designed this way. It strikes me as value engineering gone wrong, where people say "hey we can get a deal on a ton of these little pumps; they'll be cheaper than that big fancy pump with the controls we were going to buy". The plumbers are super stoked because they get to bill an insane amount of labour for a kazillion little valves and pumps.

    Even from a number of unions count it's terrible. You are now well on your way to 500 extra connections, all of which are a potential leak... Bleh...

  12. Guilbault | | #12

    Todd, and you're right that 1 pump on each line would have been fine. Water is incompressible, the suction and push design you see is unnecessary unless they improperly sized the pumps and had to add extras later.

    With air, which is compressible, sometimes it makes sense to have two smaller fans inline, rather than one big fan. But generally even with air where there's a theoretical performance benefit in some edge cases it's best avoided for maintenance reasons.

    Sounds like you're going to have your hands full at this building.

  13. Guilbault | | #13

    Ironically, the 2 pumps in line might help reduce the potential for reverse flow on the unit lines by increasing the resistance in lines that are off, making the system more reliable.

  14. wjrobinson | | #14


    Mind blowing

    Oh how we all could use subs that can complete a years work in 3 days.


    Mind blowing details

    All of them

    And every post

    It is oft said to live "in the present."

    Now you may have insight to how some of us feel about hysterical preservation.

  15. [email protected] | | #15

    I left out three details. (1) There is a one way check valve installed in each of the individual loops servicing the individual heat pump units. This should help a great deal in assuring that each unit is pulling off the supply side and dumping into the return side. (2) At their ends in the attic, the supply and return pipes are connected through a huge plastic ball valve which could be opened for flushing and closed for operation, further reducing mixing or back flow. (3) They also have 2 or three of the pump pairs in strategic locations constantly running instead of being powered through the typical demand actuated contactor. I guess that would help assure you don't have water sitting in the pipes stagnant and eventually equalizing at ambient temperature.

  16. [email protected] | | #16

    AJ Builder, I've made enough of my living off of Hysterical Preservation to feel it prudent to plead the fifth. I love and appreciate History but I will admit these projects can get ridiculous at times. I love it when the Historical Society, The ADA (Disability Advocates), The Fire Marshal and the owner of the building all have differing views on one door or what have you. All a contractor can do is wait for contest to end and build whatever the winner demanded. Here we are going to have to lower the vision lites (glass) 1-1/4 inch in every door that is merely a replica of the originals to the ADA standard of 43 inches but we have to leave the glass as original in all the doors we were able to salvage and repair. The genuine historic value of a 104 yr. old door warrants an ADA variance but the other doors next to them in the same hallway are undeserving of the variance, being mere replicas. Now we have to have a hallway checkered with mismatched doors. And all this after completion.

  17. [email protected] | | #17

    Still struggling with this issue. I'm not an engineer so what seems to me an obvious, common sense solution just has to wait until those more qualified weigh in. Very frustrating. Maybe I'm just overlooking something?

  18. Home_GSHP | | #18

    I would appreciate an update on the air purging issue if there was one- I agree there ought to be an automatic purge valve.
    I suspect the 72 duplex water pumps are designed in to save power cost and to prevent hydraulic bypass through the other heat pumps. By pumping into and out of the ground loops, the system ensures that the flow travels outside and then back inside, whereas with just one water pump per heat pump the water would travel through the parallel paths through the other heat pumps. If one giant pump was used for the whole system, it would have to be sized to pump sufficient water to support all heat pumps being on simultaneously- way oversized if only one of the 36 were to come on.

  19. tommay | | #19

    Sounds like all those individual pumps as well as fittings and other restrictions are causing cavitation likely leading to air bubbles forming in the loops. Like on any heating loop, air vents such as hy vents should have been installed. On most systems, an air scoop to separate the air from the water is normally installed along with hy vent. There are many types of vents available that could be considered, especially on the higher levels, such as duo vents for air removal and air intake for draining situations.
    I like how you consider them the hysterical society.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |