You have a closed chilled water or process cooling loop, and a load manual says the pump's friction heat vanishes because the fluid expands as pressure drops around the piping. Do not size on that. In a closed liquid loop with no machine extracting work, the pump's shaft power ends up as heat in the fluid. Motor frame loss goes to the surrounding air unless the motor sits in the fluid stream. On a large building system the number is small next to your load-table error. On a small process chiller it can decide whether the unit keeps up.
Read the symptoms before you touch the load calc
Check the ratio of chiller capacity to pump horsepower first. That one ratio tells you whether pump heat matters on your job.
| What you see | Likely cause | First check |
|---|---|---|
| Process chiller cannot hold setpoint at high pump head or low process load | Pump heat is a large fraction of chiller capacity | Pump shaft kW vs chiller capacity in the same units |
| Insulated tank with pump running and chiller off keeps warming | Shaft power dissipating into the fluid | Log tank temperature vs time |
| Fluid heats fast on a hydraulic or positive-displacement (PD) system | Flow bypassing or relieving, so the whole input is throttled into heat | Bypass or relief valve position and flow |
| Measured temperature rise across a building chilled water pump is unmeasurable | High flow and low head make the rise tiny, not zero | Compute the rise from head and efficiency (below) |
| Coolant temperature creeps up during precision grinding or aluminum work | High-pressure multistage coolant pump adding heat over time | Coolant pump shaft power vs tank volume and chiller size |
Rule out the obvious suspect fast: if the rise is unmeasurable across the pump, that does not mean the heat is absent. The loop integrates it over time.
Trace where the pump energy goes
Split the input into three pieces:
- Motor loss. Motors run roughly 95% to 98% efficient depending on type and design (normal vs high efficiency). On a 150 hp motor that is on the order of several horsepower of raw heat at the frame, depending on whether 150 hp is the rating or the input. Read the nameplate efficiency and rating basis instead of using a round number.
- Pump internal loss. A centrifugal pump converts shaft power into head plus fluid heating. At 75% pump efficiency, 25% of shaft power enters the liquid immediately. PD pumps run at higher efficiency, so they add less internal heat.
- Hydraulic (pressure) energy. The remaining 75% becomes pressure and flow, and friction in pipe, coils, strainers and valves consumes it as the fluid circulates. Friction is irreversible, so that energy lands in the fluid as heat.
Energy is conserved, so in a closed loop where nothing extracts work, all shaft power becomes heat in the fluid at steady state. The only subtraction is a machine that takes work out of the stream, such as a hydraulic motor or turbine. Separate that fluid-power component from the pump-loss component and subtract only the work that leaves the loop.
Reject the expansion-cooling offset for liquid loops
The argument that fluid expansion around the loop exactly offsets friction heat borrows from gas behavior, such as expansion across a thermal expansion valve. Liquid water stays liquid, is nearly incompressible, and does not expand and cool by any meaningful amount as pressure falls through pipe and valves. A pressure drop through a throttle or friction path in an incompressible liquid gives a small temperature rise, not a cooling effect. Treat pump losses as heat that enters the loop immediately and travels with the fluid to wherever the chiller removes it.
A second correction on the PD side: the claim that PD pumps put more heat into water despite higher efficiency is wrong as stated. The heat comes from what the system does with the flow. A fixed-displacement pump pushing fluid across a relief valve or bypass converts its entire input to heat, which is why hydraulic systems need heat rejection sized for bypassed flow. The pump's own efficiency is not the cause.
Add the pump heat to the chiller load
-
Get shaft power. Use measured motor kW times motor efficiency, or compute it from hydraulics:
WHP = gpm x head_ft x SG / 3960, thenBHP = WHP / pump efficiency. Use the actual operating point on the pump curve, not the nameplate hp. - Decide where the motor loss goes. Frame loss heats the room or air around an open-frame or TEFC motor. It is a space load, not a fluid load. If the motor is canned, submersible or otherwise in the stream, its loss also enters the fluid.
- Convert to load units. 1 hp = 2,545 Btu/h and 1 kW = 3,412 Btu/h (standard conversions).
- Add the fluid-side heat to the chiller load. For a closed loop with no work extracted, add shaft power converted to Btu/h.
- Test the magnitude against your margin. On a large HVAC plant, compare it with load-table error and design-day variation. On a process chiller where pump hp is high relative to capacity, carry it explicitly.
Rise across the pump for water, assuming c = 1 Btu/lb-F: dT_pump(F) = head_ft x (1/eta - 1) / 778. Rise for one full pass through pump plus loop friction: dT_pass(F) = head_ft / eta / 778. At 100 ft and 75% efficiency, that is about 0.043 F across the pump and about 0.17 F per pass. Both are unmeasurable on a building loop with a big chiller and are still real heat in the loop.
Check the numbers with a temperature-rise test
Run a heat balance on your own loop before you argue the theory.
- Isolate the loop or tank, insulate it, and turn the chiller and process loads off.
- Record fluid volume, starting temperature, and pump motor kW.
- Log temperature every few minutes for at least an hour.
- Compute measured heat:
Btu/h = 8.34 x gallons x dT_per_hourfor water (assumption: c = 1 Btu/lb-F, density 8.34 lb/gal). - Compare against shaft power in Btu/h from the calculation above.
A small insulated 20 gal tank with a 1 hp stainless centrifugal pump climbed above 120 F in about an hour, and the heat balance closed against brake horsepower and pump inefficiency. As a sanity bound, 1 hp of full input into about 167 lb of water is roughly 15 F per hour, ignoring motor efficiency and the tank's own thermal mass. Starting temperature and actual motor input were not recorded, so rerun with your own values. Add the tank's mass and losses if the balance is off by more than your instrument error.
On a live loop with the chiller running, use Btu/h = 500 x gpm x dT across the chiller with a reliable flow measurement, then compare with the pump's electrical input trend before and after a pump speed or trim change.
Avoid the errors that recur on chilled water and coolant loops
- Double counting. Do not add motor frame loss to the fluid load when the motor sits outside the stream. Do not also add the pump's shaft power on top of hydraulic power.
- Ignoring small-loop scale. Building chilled water has a high ratio of refrigeration capacity to pump hp. Small process chillers with high-head centrifugal or PD pumps have a much lower ratio, so the same physics can overload the unit in some conditions.
- Forgetting throttled or bypassed flow. Hydraulic and coolant systems that bypass most flow turn nearly the full pump input into heat. Size heat rejection for that condition.
- Chasing precision on a coarse estimate. Load estimates built on a design outdoor condition (for example 90 F dry bulb, 72 F wet bulb) and table values carry large error, and summer conditions vary. Apply diversity per your design basis; a 20% system diversity allowance has been used to avoid oversizing. Pump heat is often below that noise on large systems, but it is still real heat, so calculate it and decide.
- Subtracting work that never leaves the loop. In a chilled water system with no turbine or hydraulic motor, nothing offsets the pump input.
FAQ
Why does my insulated tank keep heating with only a pump running?
Shaft power from the pump has no other path out, so it enters the fluid as heat, whether through pump inefficiency or friction and throttling of the pressure energy. Estimate the rate with Btu/h = 8.34 x gallons x dT_per_hour and compare it to shaft power at 2,545 Btu/h per hp.
Why does the temperature rise across my chilled water pump read zero?
At typical building-loop heads the rise is a few hundredths of a degree F. For example, 100 ft at 75% efficiency gives roughly 0.043 F across the pump, below normal sensor resolution, but the heat still accumulates in the loop.
Why does the chiller still miss setpoint after I add pump heat to the load?
Compare a logged heat balance (500 x gpm x dT plus the pump's measured kW) against your calculated load to find which term is wrong, and rule out bypassed flow and motor-in-stream cases first. If the measured balance still does not close after that, stop adjusting and escalate to the chiller manufacturer's official technical support with your logged temperatures, flow, pump kW and pump curve operating point. For a question about a published load-estimating method, contact the publisher of that manual through its official channels.