Erosion and Corrosion in Liquid Cooling Systems: Investigating Velocity-Dependent Degradation Mechanisms (OCP EMEA Summit 2025)

Overview

This presentation by Christy Gonis and Reza Najjari, given in the Liquid Cooling track of the OCP EMEA Summit 2025 in Dublin, describes early results of a programme run by CoolIT. The programme aims to find safe coolant velocity limits in single-phase direct liquid cooling (DLC) loops.

The deck first reviews published velocity limits for copper: the OCP OAI liquid-cooling guidance gives 1-2 L/min per cold plate; the Copper Development Association gives 1.5-2.4 m/s for cold water and 0.6-0.9 m/s for water above 60 C; another source gives 2 m/s cold and 0.5 m/s hot. That these values differ from one another shows that a common empirical basis is lacking in the industry.

The mechanisms section explains erosion, corrosion and their combined acceleration: damage accelerates sharply above a system-specific critical flow velocity or critical wall shear stress when erosion combines with corrosion. The test method uses impinging-jet flow loops on copper coupons with CFD to derive wall shear stress, and 384 data points were collected in the first round. Indicators tracked include azole inhibitor depletion, copper particles, erosion depth, weight loss and roughness.

Preliminary results at 5 m/s with passivated copper, in heated and room-temperature loops, showed no measurable weight change after more than 10 days, but visual differences appeared after about 3 days and surface changes after 146-318 hours. The talk stresses the need for empirical industry data to set DLC velocity guidance. The results are preliminary and should not be read as a final design limit.

Key points

1. Published limits cited: OCP OAI 1-2 L/min per cold plate; Copper Development Association 1.5-2.4 m/s cold water and 0.6-0.9 m/s above 60 C; another source 2 m/s cold and 0.5 m/s hot.
2. Damage accelerates sharply above a system-specific critical flow velocity or wall shear stress when erosion combines with corrosion.
3. Method: impinging-jet flow loops on copper coupons with CFD-derived wall shear stress; 384 data points in the first round.
4. Heated and room-temperature loops at 5 m/s with passivated copper showed no measurable weight change after 10+ days.
5. Visual differences appeared after about 3 days and surface changes after 146-318 hours.
6. Indicators tracked include azole inhibitor depletion, copper particles, erosion depth, weight loss and roughness.
7. The talk calls for empirical industry data to set DLC velocity guidance.

Using it in design

• In pipe and manifold sizing rules, check coolant velocity against a configurable copper limit, warning above about 1.5 m/s for warm-water loops and at a lower threshold above 60 C.
• Record the inhibitor type for wetted copper surfaces.
• State the source of the velocity limit used (OCP, Copper Development Association and so on) in the design report.
• Use the preliminary results as justification for a cautious approach, not as a firm limit.

Who should read it

Mechanical engineers sizing direct liquid cooling loops and cold plate or CDU designers.

Related standards

• OCP cold plate loop requirements

Publication details

Authors

Christy Gonis; Reza Najjari

Publisher

Open Compute Project Foundation

Year

2025

Document number

OCP EMEA Summit 2025, Liquid Cooling track

Event

OCP EMEA Summit 2025, Dublin

Document language

English

Source type

Industry body

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation