Persistent flow anomaly: Grundfos UPS15-58FC requires manual ball-valve throttling to establish flow

Evan456

Hey everyone, I’m chasing down a persistent hydraulic issue on a residential radiant setup and could use some expert eyes.

System Configuration:

 Heat Source: Navien combi boiler configured with primary/secondary piping.

 Circulators: Dual Grundfos UPS15-58FC wet-rotor pumps (multi-speed).

 Distribution: Uponor radiant manifolds with built-in flow meters.

 Controls: Tekmar 519 thermostats managing zone calls.

The Symptom:

The system exhibits a bizarre hydraulic "run-out-on-the-curve" behavior on the secondary loop. When the manifold balancing/ball valves are fully open, the Grundfos circulator appears to stall or lose its grip entirely—flow drops to near zero across the manifold flow meters, regardless of whether the pump is set to Speed 1, 2, or 3.

However, if I partially throttle a downstream ball valve to artificially introduce head pressure, the pump instantly catches, flow stabilizes on the meters, and the system circulates properly.

What's Been Tried:

My plumber has already swapped in three brand-new pumps trying to solve this, and the exact same behavior happens on all of them. Changing pump speeds (1, 2, 3) has zero effect on the stall behavior when the valves are wide open; the pump acts identical across all speeds.

Here is a photo of the mechanical board layout:

(Attach the photo here)

Has anyone run into a similar scenario where a Grundfos circulator on a low-resistance loop requires added downstream head resistance just to maintain its vector flow? Any insights on whether this points to an air-binding quirk, a primary-secondary hydraulic decoupling issue, or differential pressure across the circuit would be greatly appreciated.

IMG_7198.jpeg

EdTheHeaterMan

Does this happen on both zone pumps? Yellow or orange of both?

EdTheHeaterMan

There is a reference to a pump failing to start under some conditions. it refers to the bearings not having sufficient water pressure to lubricate the water lubricated bearings.

Perhaps this is the reason for the slow operation without the artificial head or pressure resistance you cause with the partially closed ball valve. Seems weird that you are having the problem you describe, but I can't think of any other explanation.

Evan456

Hello Ed, thanks for the response.

It only happens to the pump you circled in orange. The yellow one works normally. I’ve had three local plumbers, HVAC, and hydronic people over, and none of them have an explanation—including the guy who installed the system this spring and Ferguson (who engineered/designed it).

Regarding the 'failure to operate' scenario: I can leave that pump running for 12 hours straight without an issue if the valve is throttled. But the second I fully open the ball valve, it immediately stops moving water. Because it moves liquid just fine under restriction, I have to assume the impeller and motor are completely fine. It's strictly acting up when the system resistance drops away.

hot_rod

checks working in both pumps?

Flowing the correct direction?

System purged adequately?

Lower pump should be rotated so motor is horizontal

With a 3-31/2 gpm flow requirement maybe a single pump and two zone valves with a delta p circ is a better option. Even on speed 1 you are probably over pumping.
But it should circulate on any speed setting

Steamhead

Maybe it's cavitating?

Evan456

Thanks for taking the time to respond and help Hot Rod! I appreciate it very much! To give a bit more background: I actually paid to have this system professionally installed about 6 months ago. It ran for a few days before it froze up and stopped working completely. I called the installer back, and he finally came out to fix it just last week—and it still isn't working 100% the way it should. He can't explain why this is happening, so I'm having to take the initiative to figure out what's going on and how to fix it myself.

While reading through all the instruction booklets that came with the equipment, I agree that the garage pump orientation is wrong, and I'm going to make him fix that. However, that garage loop actually works fine.

Before I talk to him again, I am trying to get a solid understanding of what is causing the house loop to only work when the valve is partially cracked. To test this, I went into the Navien boiler settings, turned on test mode to run the 4-loop zone pump, and shut off the valves to the garage and the boiler loop. Even isolated like that, it still has to be throttled down to move water.

For more context on the layout that I probably should have mentioned initially: the 4 loops add up to just shy of 1,200 feet total, the boiler is down in the basement, and the radiant tubing is run up in the wood floor joists above the system.

Evan456

As for the checks, my plumber initially thought a stuck check valve might be the culprit, so he pulled the original pump and found it didn't even have one installed. He temporarily swapped in a smaller-head spare (I believe a 15-55 model) that had a check valve, but experienced the exact same issue. He then warranty-replaced the original pump, and I personally saw the check valve built into that new replacement.

Jamie Hall

Which way is the flow through that pump? If the flow is downward, it may be cavitating, as @Steamhead suggested. If it is, it won't pump much (it will pump a little) and closing the outlet valve partly will allow the impellor to fill and work.

hot_rod

turn the isolation valves off on the small loop and try it

Also bleed the upper manifold at the bleeder screw as it runs, and maintain 12-15 psi as you do

Evan456

Hi Jami and Hot rod, I do agree that the pump is too powerful for the amount of resistance it sees in my pipes. I am going to contact the installer about getting the pump switched to one with a variable ECM so it can go low enough to have no cavitation. Thanks much for the help, especially the graph—it appears my current pump pushes 8–12 GPM, and my loops only need 3–4 GPM total.

Evan456

What Happens in Your House Loops

1. Designed for Resistance: Pumps are engineered to work against a certain amount of natural friction and restriction (head pressure) from piping, valves, and fittings. That resistance acts like a brake, keeping the pump operating safely in the middle of its curve where it maintains a steady, controlled flow.

2. The "Wide Open" Trap: Because your house loops have low natural resistance and wide-open parallel paths, there isn't enough friction to act as that brake.

3. Sliding to the Bottom Right: With no resistance holding it back, the pump's operating point slides all the way down and out to the extreme bottom-right edge of its curve (hence, "running out on the curve").

When it gets pushed out there, two bad things happen simultaneously:

 Maxed-Out Volume: It tries to shove way too much water (~8–10 GPM) through pipes that can only handle a fraction of that.

 Zero Backpressure & Cavitation: The pressure at the impeller drops so low that the water literally vaporizes into microscopic vacuum bubbles, causing that "gravel in a blender" noise and stalling the flow.

When you partially close your ball valve, you are manually forcing artificial resistance back into the system. That "brake" pushes the pump's operating point back up and to the left, right into its happy, quiet zone. A variable-speed pump does that automatically by electronic magic instead of needing a closed valve.

HomerJSmith

Does it only happen when the pump circled in yellow operate at the same time as the other pump is running?

There is no hydrolic separation between these two pumps??

Why is one pump mounted vertically?

Evan456

I can shut the valves off to the lower pump to make the system like a “1 zone system” and it still occurs. There is nothing wrong with a pump being positioned vertically?

hot_rod

The arrow on the pump body is pointing towards the manifold?

All the manifold valves open?

Evan456

Yes they are flowing proper way. And yes the one pump is wrong, going to get it fixed.