Fire Pump Churn, Thermowells & Flow Indicators

Annual fire pump testing puts three small parts under pressure: the pump casing at churn, a thermowell in the test header, and a flow indicator downstream. Each fails for a reason you can calculate. Here is the heat balance at shutoff, a vortex check for a thermowell, and what weld debris does to an indicator, plus how a cold-assembled test header with a tapped grooved cap fits in.

• A 20 kW churn load heats 30 to 60 L of casing water by about 5 to 10°C per minute
• A short stainless thermowell in a 4.5 m/s header runs far from resonance, while a long, slender one can approach it
• A 2 mm slag particle drops out of water below about 0.23 m/s, so it travels with a flush and stops at the first trap

How Fast Does a Fire Pump Heat Up at Churn?

At churn, or shutoff, the pump runs against a closed discharge. No water leaves, and the shaft power still has to go somewhere. Nearly all of it ends up as heat in the water inside the casing.

Take a pump rated at 63 L/s (1,000 gpm) and 0.69 MPa. Its hydraulic power is 0.69 × 10⁶ × 0.063 ≈ 43 kW. At 75% efficiency the shaft draws about 58 kW at rated flow. A centrifugal pump at shutoff commonly draws roughly a third of that, so assume 20 kW. This is an estimate, and the pump curve gives the real figure.

The heating rate is P ÷ (m × c_p). With 30 kg of water in the casing, 20,000 W ÷ (30 × 4,186 J/kg·K) = 0.16 K/s, or about 9.6 K per minute. With 60 kg it is about 4.8 K per minute. Rising from 20°C to 49°C takes roughly 3 to 6 minutes. That matches what site crews see: temperature climbs quickly once the valve is shut.

One correction to a common claim. Water at 55°C does not cavitate or boil in a pressurized casing. At churn the casing pressure is near 0.7 MPa or more, and water at that pressure boils well above 150°C. The vapor pressure of water at 55°C is only about 16 kPa absolute. Danger comes from other places: seals, packing and wear rings have temperature limits, and the suction eye sits at much lower pressure. If the supply is near atmospheric pressure, water close to 100°C can flash there. The temperature limit for your pump comes from its datasheet.

The standard remedy is a small continuous flow that carries the heat away. The required flow is Q = P ÷ (ρ × c_p × ΔT). To hold a 10 K rise at 20 kW, Q = 20,000 ÷ (1,000 × 4,186 × 10) ≈ 0.48 L/s, about 7.6 gpm, or under 1% of rated flow. A circulation relief valve does this job. Check your edition of NFPA 20 and NFPA 25 for the exact requirements and run times, because they differ by driver type.

A temperature port helps you see this happen. A tapped grooved cap or saddle tee on the test header gives a gauge or sensor connection without hot work. Put the sensor near the pump discharge, not at a dead end where the water is stagnant.

Can a Thermowell Resonate in a Fire Pump Test Header?

A thermowell is a closed tube that protects a sensor from the flow. Water passing a cylinder sheds vortices at a frequency f_s = St × V ÷ d, with St near 0.22. If that frequency approaches the well's own natural frequency, it can vibrate and fatigue. ASME PTC 19.3 TW sets the full calculation, and a common screening limit is a frequency ratio f_s ÷ f_n of 0.8 or less.

Run the numbers for a plain 316 stainless well with a 19 mm diameter in 4.5 m/s water. Vortex shedding comes to 0.22 × 4.5 ÷ 0.019 ≈ 52 Hz. Treat the well as a solid cantilever, with E = 193 GPa and a density of 8,000 kg/m³. The first natural frequency is f_n ≈ 0.56 × (d ÷ L²) × √(E ÷ 16ρ).

• 76 mm insertion: f_n about 2,270 Hz, ratio about 0.02
• 100 mm insertion: f_n about 1,310 Hz, ratio about 0.04
• 250 mm insertion: f_n about 210 Hz, ratio about 0.25

A short well is nowhere near resonance. Even the 250 mm well stays below 0.8 until the water reaches about 14.5 m/s. This is a rough screening result. It ignores the bore, the added fluid mass and the stress limits, so a real thermowell still needs the standard's full calculation from its maker. We don't certify a thermowell. We provide the tapped port, and the sensor supplier provides the wake-frequency report.

Also keep the flow regime in mind. During a churn test the header carries no flow, so there is no vortex shedding at all. The risk belongs to the flow test, when velocity peaks.

A word on pipe wall strength. A plain round hole in a pressurized cylinder raises the peak stress at its edge to about 2.5 times the nominal hoop stress. A branch needs reinforcement, by code, to make that up. A cap or a saddle tee does not cut the pipe wall for the tap, because the tapped port sits in a separate part. We therefore don't quote a single concentration factor for a welded tap. It depends on the weld, the pad and the procedure.

Test Header Tap Comparison by Method

Property	LEDE Cold-Assembled Header With Tapped Cap or Saddle Tee	Field-Welded Half-Coupling Tap	Forged Flange With Drilled Port	Unspecified Tapped Grooved Cap
Hot work	None	Permit, fire watch and clean-up	Welding if the flange is not bolted on	None
Weld slag inside the line	None from the tap, flush for cut chips	Possible, depends on the crew	Possible	None from the tap, check casting sand
Cut of the pipe wall	None for the cap, a reinforced hole for a saddle tee	Hole in the pipe wall, reinforcement needed	Wall intact	None
Thread quality	Factory-cut thread, concentricity target ≤ 0.15 mm	Depends on the fitting	Machined, depends on the shop	Not stated, ask for a gauge check
Thermowell mounting	Tapped port, sensor supplier does the wake-frequency calculation	Same, plus weld quality	Same	Same, if the thread suits
Pressure rating	2.5 MPa working target, burst ≥ 12.5 MPa target	Set by pipe and weld procedure	Set by flange class	Often not stated
Planning time per tap	Under 5 minutes	35 to 55 minutes	15 to 30 minutes	Under 5 minutes
What to ask for	Test report and thread gauge record	Weld procedure and inspection record	Drilling drawing	Rating and material certificate

"People treat instrument ports as an afterthought, and then wonder why the numbers drift. A port is part of the pressure boundary. It needs the same questions as any other joint: what does it do to the wall, what does it leave inside the pipe, and who checked the sensor that screws into it? I'd ask for those answers before the test, not after."

— Guo Wei, Chief Metallurgy & Critical Infrastructure Piping Systems Specialist at LEDE

What Does Weld Slag Do to a Flow Indicator?

A paddle-type flow indicator has a vane that swings with flow and trips a switch. It has small clearances at the pivot, and it sits in the path of anything the water carries.

Weld slag is hard, rough and heavy. Take a 2 mm particle with a density of 3,000 kg/m³. Its settling velocity in water is about √(4 ÷ 3 × g × d × (ρs − ρ) ÷ ρ ÷ Cd). With Cd near 1, that is √(4/3 × 9.81 × 0.002 × 2) ≈ 0.23 m/s. Flowing water at 1 m/s lifts and moves it easily, and a flush at 3 m/s carries it along. It then stops where the flow slows or the path narrows. A flow indicator is often that place.

The usual result is a stuck or slow vane. We have no failure rate to quote, and we don't put a percentage on it. We do know that a jammed vane gives no alarm, and a project spec will name a retard setting and a response time to test against.

Cold-cut grooved joints help in a defined way. Rolling a groove and fitting a coupling add no weld metal or slag to the line. A cut pipe still leaves burrs and chips, so we ream the ends and flush before any device goes in. Flushing sequence matters too:

• Flush through a temporary spool, with the indicator removed, at a velocity that moves debris
• Fit the flow indicator after the flush, using a saddle tee or a grooved tee
• Run a flow test at the end and confirm the signal reaches the panel within the time in the specification

As a fire grooved fittings supplier, we say plainly that no fitting removes the need for a flush. We can remove one source of debris. The crew still has to clear the rest.

Three-step acceptance checklist for pump test headers:

1. Check each tapped cap or tee: gauge the thread, confirm the pressure test record and the rating on the drawing
2. Ask the sensor supplier for the wake-frequency calculation for each thermowell at your test velocity, and record the insertion length
3. Flush before fitting the flow indicator, then run a flow test and confirm the alarm signal arrives within the time in the specification

LEDE (ledefittings.com) supplies grooved fittings, tapped caps and saddle tees for fire pump test headers, with thread, pressure-test and dimensional documentation available on request.