How to tell if a candidate has actually run a liquid cooling loop

1. What coolant runs in your TCS, and how do you check it?

   Strong: PG25. Glycol checked with a refractometer and kept within 24.5 to 27.5%. pH within 8.0 to 10.5 (7.0 to 10.5 for OAT fluids). Samples sent out for lab analysis. Every inhibitor addition logged.

   Red flag: "We top it off with water." Plain water dilutes the glycol and the inhibitors.

2. How fine are your CDU filters, and how do you know it's time to change them?

   Strong: 25 or 50 micron. They watch differential pressure across the filter. Duplex filters let them swap an element without shutting down. Bonus if they know filtering too fine can strip the antifoam out of the coolant.

3. A turbidity sensor alarms. What do you check first?

   Strong: whether it's air or debris. Turbidity sensors read air bubbles as particles. Air can mean a leak, or a hardware swap that wasn't purged. Then they check the buffer tank level.

4. Where do leaks actually happen on your loop?

   Strong: at couplings, quick disconnects and joints, rarely the pipe itself. They can describe the rope sensors along the piping, spot sensors at the low point of drip pans, and a drop in tank level as an early sign. They check for mineral traces on visual rounds.

5. What happens to a loop that sits unused?

   Strong: stagnant sections grow bacteria, which shifts pH and speeds up corrosion. They circulate and filter backup loops and unused feeders instead of leaving them idle.

6. Why not drop the chilled water setpoint to cool the chips harder?

   Strong: condensation. When one plant serves both air and liquid loads, the water has to stay above dew point, which limits how low the setpoint can go.

7. Walk me through a fill and flush.

   Strong: flush water clean enough to meet a conductivity limit (under 25 µS/cm in the OCP guideline), no biocides or soap in the final flush, then fill with PG25 within 24 hours. Uninhibited water never sits in the loop past a day.

8. A cold plate leaks inside a server. Who fixes it?

   Strong: they know their site's responsibility split between the installer, the cooling vendor and the IT vendor, and when warranty terms mean they isolate and call instead of opening the server.

• ●Can't name the coolant or its glycol percentage.
• ●Every answer is about CRAHs and air.
• ●Has never seen a fluid analysis report.
• ●Describes immersion tanks when you asked about cold plates. Fine for an immersion site, a gap for a direct-to-chip one.

Schneider Electric's research team puts it plainly: facility teams know chilled water plants well, and distributing coolant to racks is still very new to most of them. What transfers straight across: chillers, pumps, hydronic balancing, BAS, and water treatment from cooling towers. What you'll teach: PG25 limits, quick disconnect handling, purging after hardware swaps, and where the vendor's responsibility starts.