How an Operating Point Moves

Why this matters

A system that is under-delivering has moved its operating point, and there are only two things that can have moved it: the machine got weaker, or the piping got more resistive. Those two have opposite repairs and identical symptoms at the register or the radiator. The reading most techs reach for to separate them, motor current, points the same direction for both, which is why a good pump gets replaced. There is a reading that does separate them, it takes thirty seconds, and there is one machine family where the whole method inverts and you have to know that before you walk in.

Before you clamp anything

• Clamp on conductors that are accessible outside the enclosure whenever the installation allows it. If the reading requires opening an energized enclosure, that is electrical work under 29 CFR 1910.333(b)(2), performed with the instrument proved on a known live source before and after per NFPA 70E-2021, 120.5, and with the protective equipment your shop's arc-flash assessment calls for at that equipment. "Just taking a reading" is not a category that exempts anything.
• A clamp around two conductors of the same circuit reads near zero. That is a measurement error, not a dead motor, and it has been reported as a fault more than once. One conductor, every time.
• Gauging the machine means standing beside a rotating shaft. Guard on, hose routed clear, nothing loose on your wrists or neck.
• Relieve and confirm zero on a gauge before opening any port, and shut the heat source and let the fluid drop below 120 F, or discharge through a hose to a drain while standing clear with a face shield and heat-rated gloves.

Two curves, so only two things can move

The operating point sits where the machine's curve crosses the system's curve. That construction has its own card; this one is about what happens when one of them shifts.

• The system curve steepens when resistance is added: a valve closes, a strainer loads, a coil fouls, a filter blinds, a duct gets crushed.
• The system curve flattens when resistance is removed: a bypass sticks open, a branch blows apart, a filter falls out of its rack, a line bursts.
• The machine curve drops when the machine gets weaker: impeller wear, an impeller trim, reduced speed, reversed rotation, a slipping belt.
• The machine curve rises when capability is added: a second machine in parallel or series, a speed increase.

Everything below is bookkeeping on those four.

The movement table

Differential means head across the machine, elevation-corrected. Current is for a radial-flow centrifugal pump or a centrifugal fan, which is the family where power tracks flow. The next section handles the families where it does not.

Two rows deserve a second look.

The differential column is the diagnostic one. Rows one, three and four all show flow down and current down. Only the differential separates them, and it separates them cleanly: added resistance drives head up, a degraded machine drives head down. If you take one extra reading on a low-flow call, take that one.

The parallel row surprises people every time. Starting a second identical pump does not double flow. It moves the crossing right and up along the fixed system curve, so total flow rises by well under double, head rises, and because each pump is now sitting further left on its own curve, each one delivers less than it did alone and each one draws less current than it did alone. A tech who clamps one pump after the second starts, sees the current fall, and concludes the first pump is failing has read a correct number and drawn the wrong conclusion.

The third axis, and the families it lies to

Motor current is genuinely useful, and it is family-dependent in a way that no other reading here is.

• Radial-flow centrifugal pumps: power rises continuously with flow. Throttle the discharge and current falls. The table above is written for these.
• Axial-flow and some mixed-flow pumps: power does not rise continuously with flow. Axial-flow propeller pumps draw their maximum power at or near shutoff, so throttling one raises current instead of lowering it. Read the published power curve for the machine rather than inferring it from the impeller type.
• Centrifugal fans: power generally falls as flow falls, so closing a damper lowers current, which is why a badly blinded filter can present with low fan amps.
• Axial fans: many designs draw peak power near stall, so closing a damper on one can raise current. Again, the published power curve settles it and the wheel type does not.
• Positive-displacement pumps of every type, gear, vane, lobe, piston, progressive cavity: flow is set by displacement and speed and barely by head at all. Adding downstream resistance raises pressure and raises current, and it keeps raising both until the relief valve opens or something fails. The entire table above inverts for these machines. Diagnose them by pressure and relief-valve behaviour, not by curve crossings.

The discipline is one sentence: name the machine family before you read amps as flow. Everything else on this page follows from a curve shape, and the curve shape is a property of the family.

The case where amps sent a tech the wrong way

A hot water loop. Design is 45 GPM at 34 ft of pump head. Complaint is one wing cold, everything else fine.

The tech clamps the circulator: 8.1 A against a nameplate of 11.0 A. The read on site was "amps are low, the pump is weak, it needs replacing." That conclusion is consistent with the reading and it is wrong, because rows one, three and four of the table all produce low current and the reading cannot tell them apart.

The second reading, one gauge moved between the two pump ports at the same elevation so no column correction applies: 18.2 psi, which is 18.2 x 2.31 = 42.0 ft. Compare that against the 34 ft design differential, which is the same quantity for the same loop.

Differential is up and current is down. That combination appears in exactly one row: the system curve steepened. A degraded pump would have driven the differential down, not up. The pump is fine and the loop is restricted.

Read flow off the published curve at 42.0 ft: about 31 GPM, against a design of 45. That is a shortfall of 14 GPM, which is 31 percent of the 45 GPM design flow.

From here the work is locating the restriction, which is a segment walk with its own card, not more readings at the pump. The value of these two readings is that they took a pump replacement off the table before anybody quoted one, and they turned a vague complaint into "flow is 31 percent short and the resistance is downstream of this gauge."

The fourth signal you already have

Temperature spread across the load corroborates a flow movement without any gauge at all, because for a given heat rate, flow and temperature difference trade off directly.

• Flow down means each pound of fluid stays in contact longer and picks up or gives up more, so the spread across the load widens.
• Flow up narrows it.

Use it as a direction check, not as a flow measurement, and do not expect the spread to widen by the full inverse of the flow ratio. Output falls as flow falls, so the heat rate is not constant either, and the observed widening always comes in smaller than the arithmetic on a fixed load would predict. A loop whose flow is 31 percent short will show a clearly wider spread than its design value, and quoting that widening as a flow number is a step further than the reading supports.

The useful application is triage before you open a toolbag: a widened spread plus a raised differential is a restriction, and a widened spread plus a fallen differential is the machine.

What a variable-speed drive does to all of this

A drive changes the machine curve continuously, which breaks the assumption every row of the table rests on. Head falls with the square of speed and power with the cube, so a pump at 80 percent speed makes 64 percent of its head and draws about 51 percent of its power at corresponding points, and all of that happens with nothing wrong at all.

So on any driven machine:

1. Read the speed first, from the drive's own display, and note it with every other reading. A differential and a current taken at unknown speed are two numbers with no comparator.
2. Compare against the curve at that speed, not against the nameplate or the full-speed curve.
3. Ask what the drive is responding to. A drive holding a differential setpoint will actively hide a system curve change by slowing down, so flow falls while the differential stays exactly where it was told to stay. On that system, the movement shows up in the speed command, and a drive that has quietly walked its speed down over a season is reporting a fouling trend that no gauge on the machine will show.
4. Log speed alongside the other three readings. A maintenance record carrying speed, differential, current and temperature spread lets the next tech see the trend rather than one afternoon's snapshot, and the trend is where fouling lives.

References

• Manufacturer published head and power curves for the installed machine at its operating speed
• Hydraulic Institute standards for centrifugal, mixed-flow and axial-flow pump application, including power characteristics by impeller type
• AMCA fan performance data for fan power characteristics by wheel type
• OSHA 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5, for taking readings inside an energized enclosure
• See related: What a Pump Curve and a System Curve Do Together; How to Find a Restriction From Upstream and Downstream Readings; Why a Bigger Pump Does Not Fix a Restriction