Fill, flush and commissioning
A loop is cleanest the day it is assembled, dirtiest the day it is filled. Three rules follow.
Ask for the specification sheet
Flushing afterwards drives debris into channels too narrow to clear.
Contaminant migrates to its heat exchanger.
Differential pressure is read at the loaded state; a 5 µm element loads fast without prefiltration.
Each gate releases the next
| Step | Gate that releases it |
|---|---|
| Flush before IT install | Velocity at or above maximum operating value |
| Verify cleanliness | C13/D9/E-N00; zero particles at 50 µm |
| Drain and dry | Dry if fill is over 24 hours away |
| Fill | Inline filter in the fill line |
| Remove air | 25–30 % for 30 min, then +10 % every 15 min to 75 % |
| Pressure test | 1.5× rated maximum working pressure |
| Baseline and alarms | Difference, flow, pressure and conductivity logged |
Flush as fluid mechanics
| Figure | Category | What it governs |
|---|---|---|
| Commissioning staging | A procedure for a new loop, not a product specification | |
| Protection rating | An absolute grade against the finest channel | |
| Protection rating | The multiplier by where the element sits | |
| Strainer rating | A guard at each unit inlet; optional secondary strainer within 3 m |
Flush velocity meets or exceeds the maximum operating velocity.
Replace flush-until-clean with a class
| Class | 15–25 µm | 25–50 µm | 50 µm up |
|---|---|---|---|
| C13/D9/E-N00 | 5,000 max | 350 max | zero |
| A(C16/D12-N00) | 50,000 max | 3,500 max | zero |
Proving it takes internal rinsing extraction, a 2 µm analysis filter and light-optical analysis. Weld joints are ground to 0.5 microinch and passivated per ASTM A380 or A967; surface condition decides whether the class is reachable.
Air is a defect with a delayed symptom
| Step | Value |
|---|---|
| Pump, first stage | 25–30 % for 30 min minimum |
| Ramp | +10 % every 15 min, to 75 % |
| Minimum speed | 15 % |
| Vent at | Every high point |
| Separator position | Warm low-pressure return, at the unit inlet |
| Avoid | Traps, vertical dead legs, inverted U-bends, unvented hose |
Degassing works best at elevated temperature, which a loop reaches only under heavy load, so air symptoms appear weeks after startup.
The acceptance test and its reference
Relief valves set below the rated maximum.
Pressure drop not over 0.02 MPa.
100 psig nominal.
Isolate equipment that cannot take test pressure, and record the gauge used.
One curve, and the viscosity beside it
| Concentration | 20 % | 25 % | 30 % | 40 % | 50 % | 55 % |
|---|---|---|---|---|---|---|
| Freeze point, °C | −7.2 | −10.1 | −13.3 | −22.2 | −35.0 | −43.3 |
| Viscosity, mPa·s | 1.68 | 2.48 | — | — | — | 8.20 |
The 25 % point is corroborated at −11 °C by a fluid data sheet, and a refractometer resolves 0.2 % (V/V) at ±0.4 % accuracy. Viscosity climbs from 1.68 to 8.20 mPa·s between 15 % and 55 %, which shifts the low-flow cut-off of every meter on the loop.
Recorded before handover
| Reading | Value | Why it matters |
|---|---|---|
| often only 5–10 °F | Two sensors at ±1 °F inject almost 50 % uncertainty | |
| one part number, matched pair | The measurement chain is what fails | |
| electromagnetic needs 5 µS/cm or more | The de-ionised band runs 0.2–20 µS/cm, so that loop takes another principle | |
| 5× DN upstream, 3× DN downstream | Omit them and error rises 1.5 % | |
| make-up alarm after 5 s; elements at 35 psid | Signed before handover |
The hardware each step uses
Flushing skid EMS-FF-FL75. Filtration skid EMS-FF-FS20 at 5–15 % of loop volume. Fill cart EMS-FF-C200 with dual 25 µm filters. Coolant EMS-CL-PG25. Telemetry EMS-TL-K300 with EMS-TL-SDP and EMS-TL-G50.