Injection issues

[varadhammo]

Ok, moving to new post. Old post here: https://www.hearth.com/talk/threads/question-circ-in-closed-vs-open-system.138869

So I'm having some issues with my tekmar 356 injection. The pump is a UPS 15-58 (although right now I have it swapped with a 008 for comparison). It's a "minitube" type injection setup, the injection bridge is the underground piping between close tees off the GARN primary loop into close tees in the system loop in the building, 1.25" PEX, changing to 1" copper inside. I've got a ball valve that I've been using for balancing the injection flow although I think my next step is to put in a proper balancing valve (caleffi quicksetter). The system pump is an Alpha set on constant pressure mode.

I realize that doing injection from storage with variable tank temps (say, 190 down to 120) and variable system flows (9gpm with all zones wide open down to 2 gpm with most zones closed) and variable mix temps (~130 at design down to 80 in mild weather) is asking a lot of range out of the injection system, and I would expect some loss of temperature stability at the low end of these flows. But what seems to be happening is that the injection circ stalls--stops moving any water--when the controller ranges down to about 30%. Then the controller ranges up again, but the pump doesn't un-stall until about 50% output at which point it sends a slug of hot water and overshoots badly, then the controller ramps down, the pump stalls, it undershoots badly, and cycles on like this. I've watched it try to hit 110* and it'll swing down to 100* and up to 120* with the pump stalling and starting.

I can play with the balancing valve and get the pump working in the upper range on the tekmar (60-80%) for a given flow situation, say with all zones open and storage at 140*, but then in a different situation, say storage at 180*, and now the circ is back into this stalling and starting range. Or I can balance it for good operation with storage hot and some zones closed, and it will stabilize, but then it can't provide enough injection flow when storage is cooler/more zones are open. I'm just not getting enough range out of the 356.

I swapped a 008 in and it does the same thing, except I can actually hear a brief rubbing sound when the pump stalls/unstalls. With the 15-58 I can hear the tone of the pump change from a purer tone to a more PWM modulated sound with a wider frequency spread when it stalls.

Someone on another forum suggested that the tekmar injection control is designed to operate a short, small diameter tubing injection bridge, and it just won't work as I have it configured. He may be right in the end, but I've read about "minitube" injection systems and those are working with a long pipe run to the terminal manifold, so the length alone shouldn't be a factor right? When I calculated out the pipe resistances, I found that 200 feet of 1.25" PEX has an equivalent resistance to 10 feet of 1/2" copper or 53 feet of 3/4" copper. The 15-58 circ on speed 3 is more than capable of my worst-case injection flow. So with the proper balancing valve setting this should be well within range, methinks.

Am I just expecting more than this setup can deliver, or is there something wrong here? Hopefully you guys have some insight. Thanks in advance.

 

[maple1]

This is all kind of over my head since I don't do injection or radiant - but you don't have a HX between the Garn & the system?

 

[varadhammo]

No, in the end I decided to go "open". According to the Garn guys, this is OK, as the water treatment program takes care of all the ferrous components in the system, along with the Garn itself.

 

[maple1]

What is the elevation difference between the Garn & the injection pump?

 

[Clarkbug]

I dont think that the problem is necessarily with your piping, but I think that perhaps you are trying to lower the speed of the pumps outside of a range where it wants to operate. Need to do a little more reading, but it could be as simple as the tekmar and the two pumps you have dont play nicely with each other.

EDIT: According to this essay from Tekmar, those pumps arent on the "approved" list... http://tekmarcontrols.com/images/_literature/e0021_06.pdf

 

[varadhammo]

Sorry, forgot to indicate the elevation. The 0-15psi gauge I have reads 10.5 psi in the basement, so if you trust that, it's 24' below Garn water level. Seems about right (the boiler shed is up a hill from the building).

 

[varadhammo]

I also forgot some detail about when the pump stalls. I put one hand on the pipe just before the injection tees, and one right after the tees. After the pump stalls, the 356 is ramping up in small increments (I verified this both aurally and with a voltmeter across the triac output) but there is zero heat input at the tees until the threshold point where the pump "un-stalls" and gives a big slug of hot water, starting the overswing.

 

[Clarkbug]

Looked at your post again, and your photos. A couple of questions.

1. Are you heating your in-slab loads as well as other loads with the same temp water? Looks like your system circ feeds both equally, and I didnt see a mixing valve on there to mix down temps for the in-floor.
2. Your headloss is probably somewhere between 4-6 feet of head per 100' of piping at 9 GPM. So depending on the balancing valve, at max flow you may indeed be hitting the circulator limits.

I think you need to drop the 30% if you can, leave 50% as the minimum and see if that works better for you.

 

[varadhammo]

1. Are you heating your in-slab loads as well as other loads with the same temp water? Looks like your system circ feeds both equally, and I didnt see a mixing valve on there to mix down temps for the in-floor.

Yes, the slab zones and tube-and-plate zones are the same water temp. According to my load calcs accounting for [future] floor coverings as well, the required water temps worked out to within 10% of each other, so I went with the single mix temp for now. For the basement slab I'm just going to rely on a t-stat (Uponor PWM radiant stat) to keep it from getting to 90* down there It seems to be working out fine so far, although the finish floors aren't in yet, so eventually I'll have to bump the reset curve up a bit more. Recent cold snap we got down to 3*F overnight, Tekmar was mixing 120* to the floors. Upstairs air temp was 70* running constant circulation, basement was 65*, spot on the t-stat setting.

2. Your headloss is probably somewhere between 4-6 feet of head per 100' of piping at 9 GPM. So depending on the balancing valve, at max flow you may indeed be hitting the circulator limits.

Yes, 1.25" pex at 9 gpm comes to 3.7 ft per 100' so speed 3 on the 15-58 will match that just about perfectly. Granted, the injection pump would only ever need to run that fast if storage was all the way down to the system mix temp.

I think you need to drop the 30% if you can, leave 50% as the minimum and see if that works better for you.

Yeah, the problem right now is that if I can only use 50%+ on the 356, I can't seem to find a balancing valve setting that gives me anything close to a full range. Even with all the zones open, I can't seem to find a balancing setting that will give stable mix temp when the garn is 180 and also when it's 130. Seems like I should be able to do better than that. But maybe not. It still make building warm Just hear "tick tick tick tick" temp goes up, temp goes down
.

 

[Clarkbug]

I understand why you want a full range. Makes perfect sense.

I just wonder if the cold water coming back from the slab makes things tricky with trying to control temps? It looks like its a single zone valve to all your radiant loops? My thought is that when that opens its a big slug of cold water, the injection pump has to try to hit it, and things get all wonky from there.

For the stalling problem any possibility of getting one of the approved pumps on the list and seeing if that will go to 30%? That stall sounds like a control signal issue, not a hydronic issue.

 

[varadhammo]

It looks like its a single zone valve to all your radiant loops?

Sorry, should have specified what my zones are. The first supply/return goes a manifold, 3 zones tube-and-plate under subfloor, zoned with actuators. The second is the basement slab which you can see with the Zone Sentry valve. The third is a single zone great room (7 loops 1/2" pex), also with the Zone Sentry. The fourth is three rooms, slab on grade, zoned by actuators.

I think you have a point with the slab return temps, when the basement zone turns on it definitely sends back a slug of cold water. But this actually seems to help the injection control since it makes it have to run a higher flow. When transient conditions like that happen, I can certainly understand some temperature hunting.

For the stalling problem any possibility of getting one of the approved pumps on the list and seeing if that will go to 30%? That stall sounds like a control signal issue, not a hydronic issue.

Yeah, I wonder that too. I talked to Tekmar on the phone and they said they know of a lot of setups using 15-58's and don't know of any problems. They even said that the 15-58 was one of their tested pumps, although like you pointed out it's not on the list in their documentation. Maybe I will get a 007 just to see, can always keep it as a spare...

I have a Caleffi Quicksetter balancing valve on the way, we'll see if that helps me dial it in.

 

[Clarkbug]

Silly question. Is the tekmar trying to prevent cold water from getting to your boiler? I see on their webpage it mentions that return water protection is part of the package.... Can boiler min be set lower?

 

[varadhammo]

I checked that too, I didn't even hook up the boiler sensor at first, and the boiler relay (which I'm using to trigger the Garn primary circ) turns on whenever there's a mixing demand (wired from the Taco zone valve box). Just to make sure, I hooked up the boiler sensor, set it to return, and set the minimum to OFF. No difference.

 

[Karl_northwind]

look for sediment in the pump. and potentially flush any sediment out of the underground lines. you can disassemble that circ in place with the isolation valves. all you need is a 5/32" allen wrench (long tee handle style for preference) pull out the cartridge and rinse the whole mess out. my 15-42 was doing that with my 356 last fall. cleaning and flushing did the trick. mine is on domestic hot water, so totally open, not like your semi-closed system. there was a fair bit of rust in there (it is a SS circ)

edit: i just saw that you changed out to a 008, with similar results. so I got nothin. but it might help someone else, so I'll leave it here. is the 008 new?

as a thought, how about connecting the two boiler drains on your manifolds together with a washing machine hose, opening them, leaving the alpha system circulator running( turn on thermostat? ) which will not circulate to your floor, but will allow system circulation, and see if that changes anything as far as the 356 and 008/1558 combination. if it hums along and maintains system temp, you know there is somewhere else to look. I'll look at this thread in the am as I'm leaving work, and don't have internet at the homestead.

how long has the system been operational in the current configuration?

karl

 

[varadhammo]

@Karl_northwind,

The system has just been up for a week or so. I'll try the washing machine hose bypass in the AM, although I suspect it will maintain the temp fine that way, since when I have my great room zone open and the system flow goes up to 7 gpm or so, the 356 is able to stabilize ok. It's really the low system flows that it can't deal with. Maybe that's just normal. I was just watching it, turned all zones off except one manifold with 4 loops, system flow was around 2 gpm and mix target was 106*. The thermometer on the supply manifold for the floor loops was swinging pretty smoothly between 102* and 115*. So it actually seems to be hitting the right temperature, on average. Then I open the great room zone, the 008 ramps up a bit to pick up the slack, and then it coasts right at 106*.

The thing that concerns me is that if the pump is really stalling and restarting like that, there could be some undue wear and tear. But I'm just guessing on that one.

Oh, the 008 is not new. It's actually my boiler side DHW pump, pulled it temporarily to experiment...
Thanks...

 

[__dan]

Just guessing but I'll throw something else out for you.

Those injection controller are PID, proportional, integral, derivative, and learn the system response conditions. The control monitors and thinks, 'how long at this input did it take to get this result'. AFAIK the internal PID parameters are fixed and not field adjustable (just guessing), where they would be adjustable parameters in a complex building automation system for example.

Here is where this info may come in the picture. My read of your picture shows a (guessing) long distance injection loop circuit with the source closely spaced tees in the Garn barn and the load set closely spaced tees at the house. I suspect if the supplied water temp at the injection pump was very stable and varied slowly over time (with tank temp for example) the controller would have no trouble dialing in injection pump speed to make the reset setpoint water temp at the controlled sensor location.

What may be happening is the long injector piping circuit is introducing a time delay that the controller may not be optimized for in the PID parameter setting (which may not be field programmable in that controller). The controller is seeing, 'it's too cold, I need to ramp to make setpoint', "ramp is taking too long I need more pump speed', about this time the supply water temp at the injector pump is varying, going from cooler to hotter, then the controller sees, 'I'm too hot at the controlled sensor, need to ramp down'. However, the time delay in the long piping loop causes hunting, under and over shooting, beginning the cycle allover again, 'I'm too cold, need more pump speed'.

Seen this happen with large systems, 600 tons of condenser water cycling out to the towers from the chillers. .PID controls with the wrong PID parameters can not dial into the time delay of the water temp over the loop and will hunt continually or go unstable.

To confirm this view, monitor the supply water temp into the injector pump to see if the water temp is stable and varies slowly or not at all over 5 to 10 minutes or if the feed water to the injector pump is also hunting, going up and down.

If the supply water into the injector pump is hunting in temp, the controller may not be able to dial into a correct speed for making reset mixed water temp at the sensor location. The varying water temp supplied to the pump causes varying pump speed (too much, too little) and this makes its way back to the pump with a time delay from the long loop, from the supply to the return with mixing at the closely spaced tees.

Note that I ignored your suggestion that the pump may be stalling at low speed. If that is happening it will need to be fixed. This is a separate area of inquiry.

I would advise confirming by monitoring water temp supplied to the injector pump. If it is stable and varies very little, the controller should be able to dial into a stable speed to meet demand. If water temp supplied to the pump varies, the PID parameter setting of the controller may not be optimized for the long time delay of the loop and short term varying supply water temp.

If you confirm this is happening, my next step would be to copy this email and send it to Tekmar for their confirmation of a possible diagnoses and consult with them about what remedy may be applied.

 

[varadhammo]

Dan,

Thanks very much, I will monitor the injection pump supply temp and report back.

 

[varadhammo]

To confirm this view, monitor the supply water temp into the injector pump to see if the water temp is stable and varies slowly or not at all over 5 to 10 minutes or if the feed water to the injector pump is also hunting, going up and down.

OK, some good clues here. The feed water to the injection pump is indeed hunting up and down. It swings in the same cycle as the mix temp. Just now, I measured it in three places: right before the injection pump isolation flange; on the the brass webstone main shutoff valve; and on the brass pex x MPT adaptor where the pex line transitions to the SS flex hose. I see the biggest swing right before the injection pump flange: up to 160, down to to 120. Smallest at the PEX connection, 160 down to 147. At the brass valve, somewhere in between those two. So the swinging seems to be coming from the flow stopping and the water in that section of piping losing heat to the air, right?

If the supply water into the injector pump is hunting in temp, the controller may not be able to dial into a correct speed for making reset mixed water temp at the sensor location. The varying water temp supplied to the pump causes varying pump speed (too much, too little) and this makes its way back to the pump with a time delay from the long loop, from the supply to the return with mixing at the closely spaced tees.

Understood, but since the temp swing at the injection pump inlet is directly corresponding with the controller ramping the pump up and down, I pretty sure we're not seeing a time-delayed effect. I ruled this out entirely by closing the ball valve between the close tees on the Garn primary loop and turning off the primary pump, so that the injection pump is drawing straight off the Garn, and returning straight to the Garn, not the primary. So there is no return->supply mixing going on, but same swing happening. I do definitely have some underground line-to-line temp loss, since with low flows I never see the full Garn temp in the building. The supply line is losing heat to the return line. The lines are well insulated from the ground, but not so well from each other, unfortunately.

Note too, that at higher injection flows, the control is able to dial in a precise pump speed to maintain its target.

Note that I ignored your suggestion that the pump may be stalling at low speed. If that is happening it will need to be fixed. This is a separate area of inquiry.

It's when the pump makes this momentary rubbing noise after being ramped down to a certain point, that the injection flow stops, as far as I can tell by monitoring temperatures. Then the temp in the system loop starts to fall, as well as the temp at the injection pump intake. The controller starts to ramp up but there is again no flow until I hear the pump make this rubbing noise again, at which point I all of a sudden feel a surge of heat being injected. Then the control quickly ramps down until the pump "stalls" again. I am almost certain that there is a "deadband" where the pump is not responding to modulation from the controller. Whether this is literally stalling, or what, I really don't know.

In other words, the temp swing at the injection pump inlet seems to be an effect of the pump "stalling/starting". The pump "stalls", flow stops and the heat sitting there in the copper pipe ahead of the injection pump dissipates to the air. The mix temp drops, the controller ramps up. It reaches a threshold point and the pump "unstalls" and kicks back in, delivers that slug of much cooler water, which makes the controller ramp up quickly. Then once there is flow again, the supply to the injection pump recovers and hits the mix with a bunch of heat that the controller wasn't expecting. Et cetera...

Oops, I think I just said the same thing twice Anyway hope it's clear enough. Really appreciate the help.

 

[__dan]

I cannot see or agree with all of what you are saying above. I will try to stay with what is supposed to be happening and what may be happening. I would say the piping loss of heat to the air over the time span of a few minutes is minimal and I don't know about the low speed performance of the pump, if it works well at minimal speed. Certainly, the temperature hunting of the supply water on the suction side of the injection pump is confusing the Tekmar reset controller and preventing it from dialing in to a correct stable speed to meet the mixed reset demand setpoint.

Your system basically has three flow rates, the primary loop around the Garn, the bridge or injection pump loop between the two sets of closely spaced tees, and the secondary or load loop flow.

The closely spaced tees create hydraulic separation between the different loops and prevent one loop from pumping the other. Pump force from one loop to the other is zero and instead of pumping, mixing of the different loop waters occurs in the closely spaced tees.

In the very short term, the Garn water temp is not changing and the load demand is not changing. The load maybe drops 20 deg around the load loop, fixed regardless of flow rate in the short term of minutes. What is changing is the quantity of mixing happening in the closely spaced tees because of the varying injection pump speed. The water at the suction side of the injection pump is basically the exact same water that was previously at the same location, just separated by the time delay of travelling around the loop from the discharge of the pump, through two sets of closely spaced tees, and back to the suction side.

The longer injection pump loop piping is creating a probably non standard time delay and the wrong pump speed creates a feedback signal that is also wrong. The controller expects to dial into a stable speed based on a stable water temp supplied to the injection pump. The water that is hunting and too hot or too cold is basically the same water that left the pump minutes earlier and was too hot or too cold, it has just travelled the loop and is back at the pump.

This could probably be fixed just by adjusting the PID parameters to something that responds a little slower but the controller is probably optimized for the most standard applications and the PID parameter of that controller is probably not field programmable (check with Tekmar).

I see one possible, easy something to try that may work.

You may be able to move the primary side set of closely spaced tees from the Garn outbuilding to inside the house pretty close to the injection pump. I believe you tried something like this by opening and closing valves at the Garn side tees, but I don't know.

Moving the supply side closely spaced tees close to the injection pump will eliminate the time delay of the water travelling around that loop and is probably the standard way that the Tekmar's PID parameter are optimized for. In the short term that set of tees will have a stable water temp and as the Tekamr controller does its startup trial hunting of pump speed, it will see the effect instantly as it stabilizes. It will not be minutes later that the too hot water is, oops, back at the injection pump (ramp down), then minutes later the ,opps, too cool water has travelled the bridge loop and is now back at the injection pump (ramp up).

I would definitely say the hunting of the water temp at the suction side of the injection pump is caused by the time delay of the same water travelling around the injection pump loop.

You may consult Tekmar for their view before making changes, there may be another way to change either the response time of the system or something with the flow rates may work.

 

[varadhammo]

Today I swapped the UPS 15-58 back in, this time with the internal flow check installed. The "stalling" and temperature swinging in the system loop at low flows didn't change, but the swinging at the suction side of the injection pump did pretty much disappear. Before, with the 008 (no flow check) installed, I would see the temp there swing by almost 40 degrees in a matter of a minute. I agree with you that that is way too much to be simply heat dissipating out of the pipe. Now, with the 15-58 with IFC installed, the temp drops by only 2-3 degrees when the pump "stalls". My only theory is that with no check valve, when the pump stalled, the flow in the system loop passing by the injection tees was actually inducing a bit of flow from the injection return to the injection supply, backward through the pump, pushing cooler return water up the supply. I know this isn't supposed to happen with closely spaced tees in theory, but my guess is since my close tees are 1" into 1" pipe (Webstone P/S valve), it seems a bit more likely that this could happen. Anyway, that's my best theory.

Anyway, if what you are suggesting is really happening and pulses of hot and cool water are travelling around the injection loop causing a time-delay problem with the PID control, then I would have to expect that the temperature of the return injection leg would have to be pulsing hot and cold too. I checked it, and it isn't, maybe 1-2 degrees max when the injection pump "surges". And the return injection leg temperature is equal to the system loop return temperature, always, give or take 1-2 degrees.

The water from the injection supply tee doesn't cycle back to the injection return tee unless the system flow is less than the injection flow. The hot injection flow gets swept downstream by the system loop flow, and the return injection flow is straight from the system loop return.

And when the injection return dumps back into the Garn primary loop, the Garn primary loop flow, which is much higher than the injection loop flow, sweeps all that cool water right back into the Garn. So the temperature going into the injection supply leg from the Garn loop is always Garn temperature, and it shouldn't see any effect from the return leg temperature whatsover. It would only mix back on itself if the Garn primary loop flow were less than the injection loop flow.

Anyway, with the IFC installed in the pump, it is working a bit better, since I'm not getting that temperature swing at the injection pump inlet. I've got a Caleffi quicksetter balancing valve on the way, once I get that in there it should help me balance the injection a bit better.

Thanks for your help, really appreciate it.

 

[ewdudley]

And when the injection return dumps back into the Garn primary loop, the Garn primary loop flow, which is much higher than the injection loop flow, sweeps all that cool water right back into the Garn.

I'm curious to know why the Garn loop is needed, as opposed to drawing directly from the boiler tank with the injection loop.

 

[varadhammo]

@ewdudley,

You're right, it's not strictly needed. The way I have it configured I actually can close those valves between the close tees and turn off the primary circ and draw straight off the tank. I plumbed the primary loop because of some advice I got that with very low flows (all radiant floor loads) the stratification can become too extreme and you can end up with a fair amount of hot water stranded above the supply tapping. And I've also heard some say that the Garn doesn't fully mix itself during charging, so it's better to pump it P/S. Right now I've been playing with putting it in P/S mode while charging (pump on, valves between the tees open) and then putting it in "direct" mode after charge is complete. So far I've noticed that with the alpha primary circ pumping 13 gpm from supply to return it really does almost kill the stratification. I remember someone saying that it would take 20-30gpm to really mix the Garn tank but in my experience so far, 12-13gpm does it pretty thorougly. So.... this is really another thread, I know it's been discussed a lot on here...

I think the ideal thing would be to draw directly off the tank, and have a separate tank mixing loop/circ with it's own tappings that is wired to the draft inducer relay...

 

[ewdudley]

Thanks for the explanation, those are some interesting subtleties to take into consideration.

 

[Clarkbug]

I still think you are dealing with two separate issues here. The first is that the circle pump for whatever reason, doesnt like the 30% setting of the Tekmar. Thats issue #1 to address, and I think the other stuff can be sorted after that.

So either the pump should be swapped, or a change made to the system so the controller won't drop the demand below 50%.

What about closing a valve on the discharge of the injection pump to see if that helps?

 

[varadhammo]

The first is that the circle pump for whatever reason, doesnt like the 30% setting of the Tekmar.

Yeah, that's my biggest question. The weird thing is it was doing more or less the same thing when I swapped the 008 in for the 15-58. Except it actually made a scraping/rubbing noise when the pump "stalled" and "unstalled". So that makes me think it's something to do with the hydraulics of the loop in combination with the controller. Or find a proper balancing/throttling setting for the injection pump, like you said. I might try one more pump, put a plain old 007 in there.

Today before I put the 15-58 back in, I removed the cartridge and found a bunch of crap jammed in the impeller (actually bits of TWIGS, how did that get in there, maybe some mouse living in a piece of 1-1/2" black pipe before I assembled it or something!?) . I thought, ahah! that HAS to have been the problem. But then I realized that the 008 did the same thing, so I took that one apart. Nothing in there, clear. Put the 15-58 back in after cleaning it out. Same stalling problem as before. So, false alarm. Although a "mouse's nest" in my pump certainly couldn't have been helping anything.

Anyway, it's really not that bad, it only happens at the lowest end of the injection flow, and watching the return temps on the floor manifolds, all this temperature swinging is evened out by the time it gets through the floor loops anyway. Even with the jumpy pump control, on average the 356 seems to be hitting the right temperature. So I'll just keep tinkering, if anyone has any more ideas, I'm all ears...

By the way, why is Grundfos standard flange = Taco rotated flange? Argh....