How to qualify wetted materials for a liquid-cooling loop

How do you qualify wetted materials for a liquid-cooling loop?

Define the boundary before requesting evidence. This guide covers the single-phase aqueous technology cooling system (TCS), from the CDU secondary wetted surfaces through hoses, valves and cold plates. The hydraulically separate facility water system has its own material-fluid combination; immersion and two-phase refrigerants require different compatibility regimes. The direct-to-chip cooling guide explains those interfaces.

Identify the combination, review supplier reports, screen individual materials, evaluate relevant metal pairs, and validate finished assemblies. Keep three evidence packages, following OCP September 2026 §§8-9 and its cold-plate qualification paper:

A satisfactory pressure test establishes the tested pressure boundary. Keep both alongside the liquid-cooling commissioning record.

A compatibility claim must name the fluid trade name and revision, concentration basis, material grade or compound, supplier, manufacturing condition, temperature, duration, exposed area, fluid volume and acceptance source. Accept when all agreed material, fluid and assembly criteria pass; reject demonstrated failure; hold missing identity, controls, analytical capability or duty coverage. Each hold needs a named missing document or retest before purchase release.

Build a wetted-material register before ordering tests

Split each assembly into its wetted constituents. A copper cold plate also contains joining material; a hose assembly includes its liner, seals and fittings; a filter includes its binder, support and end seals. The register follows the actual fluid-contact surface, including temporary test hoses, fill carts, lubricants and service replacements.

These entries define evidence requests, not approved products. Add supplier/site, part revision, drawing/BOM identifier, lot, loop location, wetted area, temperature range, peak and cleaning-fluid exposure, report identifier and procurement disposition to every row. A controlled supplier compound code preserves confidential formulation identity while allowing changes to be traced. OCP September 2026, §§6.2-6.5; OCP Cold Plate Development and Qualification.

OCP's September 2026 guideline recommends peroxide-cured EPDM and warns that sulfur-cured EPDM can introduce sulfate into the coolant, making the cure system part of the wetted-material qualification record. Cure agents, catalysts, plasticizers and other additives distinguish compounds within one polymer family. Parker identifies a peroxide-cured EPDM tube in its 627 cooling hose; that identifies a construction feature, while the exact coolant still needs compatibility evidence. Parker cooling-system product page; manufacturer 627 product description.

Which OCP guidance and acceptance rules govern?

Use OCP's Guidelines for Using Propylene Glycol-Based Heat Transfer Fluids in Liquid Cooled Racks, Revision 2.0, Version 1.0.0, September 15, 2026. The OCP Cold Plate project index also links older contributions.

Vertiv SL-80056 Rev A, February 2025, expressly gives individual product manuals precedence over its general deployment guide. Establish the purchased equipment's limits first, then use OCP guidance to define evidence within that envelope. Conflicting fluid and equipment requirements need a written resolution accepted by affected suppliers and the design authority. Vertiv deployment guideline.

OCP's material lists are representative guidance, not product certification, a warranty or universal fluid approval. A PG25 designation identifies glycol concentration without identifying the entire inhibitor formulation. Water-based fluids follow the same evidence sequence with water-specific criteria; the PG exposure conditions below do not prescribe a water test.

The treated-water value comes from the OCP water-based-fluid guideline, Table 1; the flush-water value comes from the September 2026 PG guideline. Keep fluid and stage attached to each limit.

The verified publisher methods are ASTM D471-16a(2021) for rubber, D543-21 for plastics, D8040-18(2023) for heat-transfer-fluid glassware corrosion, G1-25 for specimen cleaning, G31-21(2025) for immersion design and G71-81(2024) for galvanic testing. Their public scopes establish applicability; obtain the selected procedures for laboratory execution.

OCP's undated cold-plate paper cites D2570 without a year; ASTM lists D2570-26 for automotive circulating-coolant testing. Its public scope permits agreed component substitutions outside the engine-coolant specification qualification it identifies, so a TCS adaptation needs its own documented scope and agreement. ASTM rubber scope; ASTM D2570-26.

What conditions belong in a coupon test?

OCP's September 2026 propylene-glycol guideline gives six weeks at 65 to 85 °C as typical polymer-compatibility exposure conditions, with a room-temperature fluid blank and a matched heated fluid blank for evaluating changes in fluid purity. Select the final exposure for the material and application, and record every departure from the method. OCP §8.7.

Fix formulation, batch, concentration, actual immersed area, liquid volume, headspace, vessel, closure, holder, temperature history and preparation. ASTM G31-21(2025) identifies solution volume, gas exposure, temperature and specimen support as influential variables in immersion tests. Match vessel and support contact in the heated blank so their contribution is not assigned to the material. ASTM G31 scope.

Three coupons sharing a jar produce one independently exposed fluid sample. Destructive tensile endpoints need their own specimens, and endpoint handling needs enough staff and equipment to meet the measurement window. Stagger exposure endings rather than leaving cooled samples waiting unrecorded.

For loaded rigid plastics, add the applicable stress-and-reagent exposure under ASTM D543-21; unloaded immersion can miss environmental stress cracking. Static jars also leave oxygen ingress, flow, pressure and thermal cycling for assembly validation. Report modified exposures as modified methods, and test production-finished surfaces separately from cleaned coupons. Temperature accelerates different reactions differently: assign field-equivalent years only through a validated model.

How much polymer change is acceptable?

OCP's September 2026 §8.12 gives general starting bands of minus 10% to plus 10% volume change and minus 30% to plus 30% tensile-strength and elongation change, subject to functional condition and component-specific acceptance. Write the purchased seal's acceptance sheet before exposure, including whether each result is a release limit or a report-only measurement. OCP polymer interpretation.

Read mass, volume, hardness, tensile behavior and visible condition together. Swelling and extraction can occur simultaneously, leaving little net mass change even while the compound changes. A seal inside a screening band still needs its actual static or dynamic sealing duty tested; ASTM D471-16a(2021) provides comparative liquid-effect evidence, not automatic service equivalence.

How do you measure coolant elastomer leachables?

Measure identified constituents in the actual coolant and compare exposed fluid with both controls. Leachables enter the intended fluid under the specified exposure; an aggressive extraction study identifies possible compounds under its own conditions. Its concentrations do not predict an operating loop. The undated OCP cold-plate paper §1.3.3.2 identifies inductively coupled plasma (ICP) metal analysis, ion chromatography, and gas/liquid chromatography (GC/LC) as complementary analytical routes.

A single “leachables ppm” result cannot settle compatibility. Glycol-rich fluid already contains organic carbon, so total organic carbon (TOC) is not a selective polymer-leachables assay; conductivity likewise does not identify the released constituent. Set formulation-specific limits with the coolant and component suppliers.

Record sample ID, exposure, collection time, container, preservation, holding time, custody, analytical method, units, dilution, matrix recovery and uncertainty. Report each independent jar and both blank results, the blank-adjusted difference, the reporting limit in the original coolant after dilution, and the decision criterion. Keep replicate spread visible rather than releasing a combination from a favorable average. A water-method reporting limit before glycol dilution does not establish coolant detection capability.

Anvilfield's analytical decision rule applies absolute fluid limits to the exposed sample and material-change criteria to its blank-adjusted result. Blank correction cannot subtract away an absolute fluid failure.

A non-detect is not zero. Preserve its reporting-limit qualifier, retain signed differences and hold any decision that dilution or blank variability prevents resolving. Repeat contaminated blanks or inadequate recovery before assigning cause.

How should leachables be normalized?

Subtract the matched heated-blank concentration from the exposed concentration, multiply by recovered fluid volume, then divide by exposed material area. With concentration in mg/L, volume in L and area in cm², the result is mg/cm²; multiply by 1,000 for µg/cm².

Anvilfield reporting calculation: q = (C_exposed − C_heated blank) × V / A.

Use this expression only for verified matching endpoint volumes. With matched starting volumes but different endpoint volumes, calculate each jar's analyte mass from its own recovered volume before comparing masses. Retain raw values and blank uncertainty; comparisons require matched fluid, time, temperature, geometry and surface-to-volume ratio.

Worked example: interpret a sulfate result

Use selected report inputs of 3.0 mg/L sulfate in exposed fluid, 0.4 mg/L in its heated blank and 0.3 mg/L in its room-temperature blank. Each jar contains 0.10 L at the endpoint; the exposed specimen area is 28 cm². All three results exceed the selected 0.1 mg/L reporting limit, and the 2.6 mg/L exposed-minus-heated difference is resolved by the analytical method.

Anvilfield estimates the net mass and area-normalized result using the reporting calculation above. The two acceptance ceilings are selected project inputs for this worked example. Meeting the absolute fluid ceiling does not override the material-change failure; investigate the compound and confirm the independent jars before release. Area normalization reports net excess remaining in the fluid, while adsorption, deposits and unidentified species require separate evidence.

Inhibitor uptake can accompany an apparently sound polymer; low dissolved metal can accompany deposited corrosion products. Resolve the lost protection or physical damage before accepting a favorable bulk-fluid result.

Area-normalized jar results do not predict loop concentration or an allowable hose count. Representative component exposure and assembly evidence must address release over time, inhibitor demand and the installed fluid-to-surface ratio.

What metal-corrosion screen should you specify?

OCP's September 2026 modified ASTM D8040 metal screen uses three separate beakers at 88 °C for 336 hours with continuous aeration and electrically insulated coupons, so it does not establish compatibility of electrically connected metal pairs. Record intended-use fluid preparation, the salt challenge and every modification. OCP §9.3.

OCP September 2026 Table 10 sets copper C11000 and brass C26000 screen ceilings of 0.2 mpy and 5 mg coupon loss, and 304L/316L stainless ceilings of 0.1 mpy and 2 mg, within its specified test configuration. Here mpy means mils per year; one mil is 0.001 inch.

Use ASTM G1-25 preparation and cleaning controls; retain raw masses, cleaning correction, area, exposure duration and density. The mass and rate criteria are tied to the source configuration, not interchangeable for arbitrary coupon geometry. Inspect each coupon before and after cleaning, retain photographs and record any localized attack.

A low average loss cannot clear unexplained pits or coating damage. OCP's nonstandard-alloy note supplies no plating-adhesion or exposed-substrate acceptance criterion.

ASTM D8040-18(2023) treats glassware corrosion results as screening evidence that cannot, by itself, establish satisfactory corrosion inhibition in service. ASTM D8040 scope and significance. ASTM D1384-24 similarly addresses engine-coolant screening. OCP notes limited copper-test discrimination around 0.1 to 0.2 mpy; use that resolution when comparing fluids, and do not calculate wall life from the screen.

How do you qualify mixed-metal and plated combinations?

ASTM G71-81(2024) addresses galvanic testing of dissimilar metals in electrical contact under low-flow electrolyte conditions. Identify the coolant, electrical connection and exposed areas, including paths through fittings or bonding. ASTM G71 scope.

OCP's undated Cold Plate Development and Qualification paper §1.2.1.3 gives a 0.15 V design recommendation for metal interfaces alongside integrated corrosion qualification. It supplies neither a universal reference-electrode procedure nor a galvanic-current pass limit. Do not convert that design number into a field voltmeter acceptance test. OCP cold-plate paper.

OCP September 2026 §6.3 favors stainless quick connects and recommends avoiding nickel/chromium-plated connectors until demonstrated satisfactory. Mating wear can expose the substrate, so include the qualified worn condition and the finished-connection leak response.

Anvilfield's pair-test record names the alloy and finish on each side, exposed areas, electrical connection, coolant batch and concentration, temperature, oxygen/flow conditions, duration, uncoupled comparators and agreed loss, pitting and functional limits. Where coupling current is recorded, retain its direction and time history along with post-test damage.

Electrolyte, oxygen, passive films, temperature, flow and crevices influence pair behavior. Include stagnant or flowing duty where relevant, explicitly adapting G71's low-flow scope when necessary.

A dielectric fitting provides galvanic isolation only when parallel conductive paths are absent, and each isolated metal still needs corrosion protection. Adding aluminum to a copper-containing loop reopens both fluid/material and pair evidence.

Vertiv SL-80056 Rev A §6.3 calls for proper grounding and bonding to be identified and followed. Preserve those protective connections when designing galvanic isolation.

What must pass on the finished assembly?

Test the purchased hose, connector, valve, cold plate and manifold construction after material screening. Finished surfaces introduce brazes, sealants, residues and geometry that a clean coupon omits. OCP's cold-plate paper and rack-manifold qualification paper treat component validation as its own evidence package.

Measure leak tightness, pressure integrity, seal movement/retention, relevant connect/disconnect duty, pressure drop and thermal performance against the purchased specification. Establish hydraulic and thermal baselines, then repeat at the same coolant, flow, temperature and mounting condition. Changed viscosity or setup can resemble fouling. Record cold-plate pressure drop and thermal resistance separately; changed thermal resistance with unchanged pressure drop also needs a mounting/contact check.

Include filled storage, stagnant operation and expected service cycles when they differ from continuous duty. Log actual fluid, pressure, temperature, flow and cycles; retain post-exposure fluid, deposits and surface observations. Exterior salt spray addresses exterior durability, not internal coolant compatibility.

Retain the hydro-test pressure-boundary record separately from material approval. The CDU commissioning guide covers operational handoff; redundancy and concurrent maintenance establishes the service and isolation duty the assemblies must support. Use their system procedures with the assembly supplier's actual pressure and cycling requirements.

Which changes require requalification?

Reopen qualification when composition, manufacturing condition, fluid or operating duty changes. A new lot of an unchanged approved product needs traceability and incoming verification; it does not automatically need another six-week exposure. Use Anvilfield's change gate below.

Treat changed inhibitors, mixed-brand top-up and added biocide as fluid-formulation changes even when nominal glycol percentage stays constant. Require affected suppliers' compatibility evidence and the corresponding tests. Changes to existing infrastructure also need the liquid-cooling retrofit scope review.

The owner/design authority accepts the project combination; component and fluid suppliers substantiate their limits. Tie the signature to BOM, coolant revision, methods, criteria, reports and resolved exceptions.

Purchase terms should require advance notice of material, site or process changes, an unchanged-configuration declaration, lot traceability and permission before substitution. Preserve superseded approvals so failures can be traced to installed lots. Apply OCP September 2026 §§6, 8-9 within the purchased component requirements.

What changes after installation?

Establish the installed-fluid commissioning baseline and transfer approved identities, limits and retained-sample locations into the water-treatment and chemistry program. Laboratory qualification selects a material-fluid combination; operating surveillance detects changes in the installed system. Salt-challenged laboratory corrosion exposure is not an installed-loop procedure.

OCP September 2026 §11.2 calls for annual qualified-laboratory testing of each TCS loop and gives the following risk-based schedules, with separate representative-sampling guidance for similar small rack loops. Record the scenario assumptions with the selected schedule. OCP sampling guidance.

“Mostly acceptable” is OCP's scenario language, not permission to install an unapproved component. Investigate changed metal, sulfate, inhibitor or particle trends against the material register and retained samples. Increase rack-sampling coverage after anomalies, and investigate inline sensor anomalies that remain unresolved after on-site checks or sensor calibration.

Collect a representative circulating-fluid sample using the approved sample port and procedure. OCP §11.2 also recommends retaining labeled as-received fluid samples from deliveries and batches for at least six months; record their locations and batch numbers at handoff.

Record flush and test-fluid contact as well. Vertiv SL-80056 Rev A §6.1 says to avoid chemical cleaners/detergents unless following the new heat-transfer-fluid supplier's guidance; OCP September 2026 §10 describes controlled cleaning and rinsing. Preserve that exception and the product-manual precedence in the approved cleaning procedure. Vertiv cleaning guidance.

How long does qualification take, and what belongs in the estimate?

Allow six weeks of exposure when the selected polymer and metal screens run in parallel, or eight weeks when they run sequentially. These Anvilfield schedule estimates use OCP's September 2026 starting conditions; preparation, analytics, pair testing, assembly validation and review need additional scheduled time.

For an order covering three polymer formulations, multiply formulations by temperatures by independent jars, then add matching fluid controls. The selected scope uses one inhibited PG25 formulation and batch, identified peroxide-cured EPDM, FKM and HDPE formulations, and exposures at 65 °C and 85 °C. Order three independent jars per material/temperature combination; the isolated metals are C11000 copper, C26000 brass, 304L stainless and 316L stainless. Share controls only when fluid batch, volume, container and thermal run match.

Anvilfield estimates these quantities from OCP's September 2026 exposure conditions and the selected scope above. Five exposed tensile specimens per jar is the order's choice; the laboratory confirms destructive endpoints, conditioning and material-specific geometry.

Use the table's rubber geometry for the rubber rectangles; OCP describes plastic tensile preparation separately through its ASTM D638 reference. Five unexposed controls per formulation are shared only with identical baseline conditioning and material lot.

Adding the rectangle's cut edges increases face-only area by 12%; omitting them understates total area by about 11%. In a selected 100 mL jar, that rectangle contributes 0.28 cm²/mL; add all tensile-specimen areas and subtract holes or covered surfaces to obtain the actual ratio.

For the salt batch, multiply each published salt mass per liter by 0.5 L final solution. Use qualified stock preparation when direct weighing is outside suitable balance capability. Set beaker fill volumes and ready-to-use-fluid preparation through their respective method branches.

Request separate laboratory quotation lines for these quantities, endpoint methods, detection capability, handling capacity, pair conditions and assembly duty. Name who supplies specimens, owns reports and funds agreed retests. Obtain the selected ASTM procedures and agree every TCS modification before testing.

A public cooling bid needs clear responsibility for test specimens, laboratory reports and retests; estimators use PorchRocket's Bid Control to find the HVAC and refrigeration contracts where they price that scope.

Field checklist: release the material combination

• Mark the TCS boundary and temporary fluid contacts; hold incomplete scope.
• Record part, revision, compound/alloy, cure, coating, lot and process; hold unnamed formulations.
• Name fluid product, revision, batch, concentration and thermal envelope; resolve conflicts.
• Match supplier reports to the actual combination; document equivalence or order tests.
• Approve geometry, area, volume, controls, independent jars and analytical limits.
• Agree mechanical units and limits; allocate destructive specimens and endpoint handling.
• Retain raw results and analytical QA; hold unusable blanks or inadequate detection.
• Inspect pits, dezincification, coating damage and deposits; resolve localized attack.
• Map electrical paths and area ratios; test uncovered pairs.
• Complete finished-component duty; hold functional failures despite coupon passes.
• Sign the exact BOM/fluid release and change triggers; block unreviewed substitutions.
• Transfer baseline, sampling schedule and retained samples; assign trend-review responsibility.

Release record and citation

Keep one signed cover record for the approved combination: register revision; purchased part/BOM revisions and lots; coolant product, revision, batch and concentration; allowable exposure envelope; report IDs and specimen identities; accepted criteria; disposition of deviations; approving owner/design authority; supplier change-notification terms; and the operating baseline. Link it to raw measurements, retained samples, photographs and assembly results so a service replacement can be checked against the same approval.

Anvilfield. Sources checked and reviewed: 2026-11-05. Update log, 2026-11-05: clarified alloy-screen scope and acceptance records; added a worked leachables decision; revised the quotation scope and operating handoff.

How to cite this page: Anvilfield, How to qualify wetted materials for a liquid-cooling loop, 2026, https://anvilfield.com/field-guides/datacenter/coolant-material-leachables-qualification/

Related tools

FAQ

People also ask

• What coolant is used in data center liquid cooling?
• What is direct-to-chip liquid cooling?
• What pressure do you test chilled-water piping to?
• What is a CDU in a data center?
• What is direct-to-chip liquid cooling?

Sources checked

• S1. Open Compute Project, Guidelines for Using Propylene Glycol-Based Heat Transfer Fluids in Liquid Cooled Racks, Revision 2.0, Version 1.0.0, September 15, 2026. Primary PG exposure, acceptance and sampling guidance, §§6, 8-11. Sources checked: 2026-11-05.
• S2. Vertiv, Guidelines for Deployment of Single Phase Technology Cooling System, SL-80056_REVA_02-25, ©2025. Product-manual precedence, acceptable materials, sealants and conditional cleaning instructions. Sources checked: 2026-11-05.
• S3. Open Compute Project, White Paper: Cold Plate Development and Qualification, undated public edition at the opened URL. Undated public edition; integrated corrosion, analytical methods and assembly qualification. Sources checked: 2026-11-05.
• S4. Open Compute Project, Guidelines for Using Water-Based Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled Racks, undated final public PDF. Undated final public PDF; water-specific chemistry and scope. Sources checked: 2026-11-05.
• S5. ASTM D471-16a(2021), Standard Test Method for Rubber Property: Effect of Liquids. Public scope and active rubber-method edition; OCP bibliography has a differing designation. Sources checked: 2026-11-05.
• S6. ASTM D543-21, Standard Practices for Evaluating the Resistance of Plastics to Chemical Reagents. Public scope and stressed-plastic exposure. Sources checked: 2026-11-05.
• S7. ASTM D8040-18(2023), Standard Test Method for Corrosion Test for Heat Transfer Fluids in Glassware. Public scope and screening limitations. Sources checked: 2026-11-05.
• S8. ASTM G1-25, Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens. Public scope and specimen preparation, cleaning and damage evaluation. Sources checked: 2026-11-05.
• S9. ASTM G31-21(2025), Standard Guide for Laboratory Immersion Corrosion Testing of Metals. Public scope, immersion variables and localized-attack limitations. Sources checked: 2026-11-05.
• S10. ASTM G71-81(2024), Standard Guide for Conducting and Evaluating Galvanic Corrosion Tests in Electrolytes. Public scope, electrical contact and low-flow galvanic testing. Sources checked: 2026-11-05.
• S11. ASTM D2570-26, Standard Test Method for Simulated Service Corrosion Testing of Engine Coolants. Public scope, automotive circulating-coolant application and substitutions. Sources checked: 2026-11-05.
• S12. Parker, Data Center Cooling Thermal Management System, undated manufacturer product/application page. Undated manufacturer cooling-system page; linked HOSE627.js product description confirms peroxide-cured EPDM tube construction. Sources checked: 2026-11-05.
• S13. Open Compute Project, Cold Plate subproject, live index checked 2026-11-05. Live index; document discovery and cold-plate workstream. Sources checked: 2026-11-05.
• S14. ASTM D1384-24, Standard Test Method for Corrosion Test for Engine Coolants in Glassware. Public scope and engine-coolant glassware-screen limitations. Sources checked: 2026-11-05.
• S15. Open Compute Project, White Paper: Guidelines to Rack Manifold Requirements and Qualification, undated public PDF at filename v3. Undated public PDF with v3 filename; manifold component qualification. Sources checked: 2026-11-05.