Engineering guide 06 / 08

Coolant Chemistry and Filtration for Liquid Cooling

The coolant looks clear, but a chemistry reading has changed and the filter needs attention again. Here is how to work out what each observation means, choose the right test and decide whether the fluid, filter or maintenance practice needs to change.

Reliability EngineUpdated 11 min read
On this page

Check fluid chemistry and particle cleanliness separately. Start with the coolant and equipment suppliers' requirements, compare current results with the commissioning baseline, and confirm the cause before adding chemicals, replacing fluid or changing filtration.

For
Coolant suppliers, treatment specialists and filtration teams
Scope
Single-phase cold plate cooling with treated water or inhibited water-glycol fluids. Facility water and technology cooling loops require separate specifications. Immersion and two-phase fluids need their own chemistry, compatibility and handling procedures.

Key decisions

  • Set chemistry limits for the approved fluid and every material it touches.
  • Select filtration by particle removal, pressure loss and compatibility, with evidence beyond the micron label.
  • Use trends to direct investigation and suitable tests to confirm the cause.
From approved fluid to an evidence-based service decision
  1. 01Specify

    Agree the fluid, materials, methods and acceptance limits.

  2. 02Baseline

    Test the delivered fluid and the circulating loop after commissioning.

  3. 03Observe

    Trend available condition signals and filter pressure loss.

  4. 04Confirm

    Use representative samples and supplier interpretation.

  5. 05Verify

    Record the treatment and prove the loop meets its specification.

A filtration result answers a particulate question. A chemistry result answers a different question about the fluid and its treatment.

Choose the fluid for the whole loop#

A fluid that suits the coolant distribution unit (CDU) must also suit the cold plates, hoses, seals and filter it will touch. Those parts form the wetted-material list. Include temporary fill and flushing equipment too: it contacts the same fluid, even if it leaves the site after commissioning.

Name the exact coolant product, concentration and permitted makeup fluid before filling. Compare the server, CDU and fluid-supplier instructions, including their document revisions. Resolve conflicting requirements together. A limit for one formulation may assume chemistry or test conditions that do not apply to another.

Keep actual part numbers in the compatibility record. Descriptions such as rubber or stainless steel leave important details unresolved. Revisit the record when a hose, seal, filter medium or rack configuration changes. A routine replacement can introduce a new material into a previously accepted loop.

Minimum inputs to a coolant compatibility review
InputEvidence to requestDecision it supports
Fluid formulationProduct name, revision, concentration, batch certificate and supplier instructionsWhether the proposed fluid is permitted in this loop
Metals and joining materialsAlloy, plating, braze or solder details and component approvalsCompatibility across the assembled circuit
Polymers and sealsExact compound or grade and compatibility at the intended exposureWhether hoses, seals and sample lines remain suitable
Operating and storage conditionsTemperature, pressure, stagnant periods and shipping conditionsWhether approval covers the full equipment lifecycle
Service additionsMakeup fluid, cleaning agents and residual-fluid acceptanceHow maintenance preserves the approved formulation

References: Open Compute Project: Guidelines for Using Water-Based Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled Racks; Open Compute Project: Guidelines for Using Propylene Glycol-Based Heat Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled Racks

Let the measurement choose the next question#

A changed reading is a reason to investigate. Conductivity tells you about the fluid's electrical response; it cannot name the substance that changed. A pH result cannot tell you how much of every corrosion inhibitor remains. Use the table to choose evidence that answers the question you actually have.

Test the incoming batch and the circulating loop separately. A delivery certificate covers the supplied fluid. Once filled, that fluid has contacted installed parts, residual rinse water and any remaining cleaning chemicals. Keep both results with the fill record so you can see where a change first appeared.

Make comparisons under a consistent method. Conductivity depends on temperature, pH measurement must suit the formulation, and concentration instruments need product-specific calibration. Save the units, sample temperature, compensation setting and instrument identity. Otherwise a change in measurement practice can resemble a change in coolant.

A measurement plan that separates observation from diagnosis
MeasurementUseful questionWhat it cannot establish alone
ConductivityHas the measured ionic response changed under comparable conditions?Which substance changed, or whether inhibitor protection is adequate
pHHas pH moved outside the approved formulation envelope?The remaining concentration of every inhibitor or the corrosion rate
Glycol concentrationDoes the sample meet the approved mixture specification?Whether the complete additive package remains effective
Particle count or turbidityHas particulate contamination changed using a validated method?Dissolved contaminant levels or the source of every particle
Dissolved and total metalsIs there a change requiring inspection or further investigation?The damaged component or a current corrosion rate without context
Supplier-specific inhibitor assayDoes the relevant treatment component meet its approved requirement?Compatibility with an unapproved fluid or replacement component
Microbiological test where applicableIs the approved microbial-control requirement being met?That a clear-looking fluid is biologically acceptable

References: Open Compute Project: Guidelines for Using Water-Based Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled Racks; Dow: Cooling Solutions for Data Center and Cloud Performance

Choose a filter that protects flow as well as equipment#

Start with the cold plate or component you need to protect. Ask its manufacturer what particle sizes and removal performance the system requires. Then ask the filter supplier for evidence at those sizes. A micron label by itself leaves out removal efficiency and the test method; nominal and absolute labels can carry different definitions.

A finer filter also needs a hydraulic check. Compare its clean and loaded pressure loss at the intended flow, fluid viscosity and temperature. Agree when to replace it. Rising differential pressure, the pressure difference between inlet and outlet, can reflect loading, higher flow or a different fluid condition.

Put the sensor location on the drawing. A sensor after a side-stream filter sees fluid that the filter has already treated. That may suit a chemistry measurement while hiding some particles present elsewhere. Use inlet and outlet samples to test removal, and investigate dissolved chemistry separately.

  • Record removal performance at specified particle sizes and the test method.
  • Confirm housing, seals, media and extractables are compatible.
  • Provide isolation, pressure indication and a service procedure.
  • Show when filtration is bypassed or isolated.
  • Retain an abnormally loaded element for investigation.

References: Pall Corporation: Understanding Particle Removal Performance in Liquids with Cartridge and Bag Filtration

Give the laboratory a sample it can trust#

Ask the laboratory for collection instructions before opening a sample port. Bottle type, preparation, preservation and delivery time can differ between tests. Use the site's approved method for hot or pressurized fluid, with a clean collection path that keeps air, dirt and incompatible cleaners out of the circuit.

Label the sample with the loop, location, collection time, operating state and recent additions. Note whether it came before or after a filter and whether that branch was circulating. Fluid sitting in a dead leg, a branch with little or no circulation, may tell you about that branch rather than the active loop.

  • Pair samples to compare a delivered batch with circulating fluid, or filter inlet with outlet.
  • Keep laboratory instructions with the kit and train the collector.
  • Review recent flushing, refill, filter changes and equipment additions.
  • Set routine and event-triggered sampling intervals with the supplier and owner.

References: Dow: Cooling Solutions for Data Center and Cloud Performance

Check what every top-up adds to the recipe#

A refill restores the fluid level, but its chemistry depends on what went in. Record the product, concentration, quantity and reason for every addition or removal. Repeated automatic top-ups deserve review: they may conceal a recurring loss and can change the formulation if the makeup fluid is wrong.

A simple material balance can show the likely direction of that change. Confirm the final mixture with the supplier's test method before choosing a correction.

What a 40 L water addition does to concentration

For 800 L of coolant at 25% glycol by volume, adding 40 L of water without removing fluid leaves 200 L of glycol in a total volume of 840 L. The calculated concentration becomes about 23.8%, a drop of 1.2 percentage points. Additives are diluted too. After an unapproved top-up, check the service record, test the circulating mixture and ask the fluid supplier how to correct it.

Final concentration = (800 L x 0.25) / (800 L + 40 L) = 0.238 = 23.8%

This volume balance treats volumes as additive. Use the product concentration basis, temperature and mixing effects in the actual assessment, and compare the result with supplier-approved limits.

Confirm the cause before treating the loop#

Start with the instrument, sample flow and recent service history. Did someone change the calibration setting, add fluid or change the operating temperature? A conductivity rise after a planned treatment addition has a different starting explanation from a rise during stable operation. Align those events on one timeline.

Then choose a test that separates the likely causes. Particles may need identification. A chemistry excursion may need an inhibitor assay. Suspected corrosion may call for dissolved and total metal analysis plus component inspection. A bulk-fluid trend cannot locate a damaged component or prove inhibitor depletion by itself.

Close the incident against the original question. If particles fell after a filter change but chemistry remains outside its accepted range, the chemistry issue still needs a decision. Retain the before-and-after results and the reason the responsible team accepted recovery.

  • Verify the instrument and representative sampling before treating a reported excursion.
  • Compare the fluid, hydraulic and service records over the same time window.
  • Obtain the confirmatory tests that distinguish the leading explanations.
  • Agree the intervention, isolation plan and acceptance criteria with the responsible suppliers and operator.
  • Resample and document the post-service condition before closing the event.

Make the next decision someone's responsibility#

A laboratory report needs a reader with authority to act. Name who collects samples, interprets results, approves additions and keeps the history. Include a response path for a breached limit and for an unexplained trend that remains within limits. Both can require attention, but the next steps may differ.

Keep the history comparable when the program changes. A new laboratory, concentration instrument or compensation setting may change the result. Arrange an overlap comparison where appropriate. After major loop work or a formulation change, establish an accepted new baseline and retain the old one with an explanation.

Coolant service schedule

Loop and boundary: [ID]. Fluid and concentration basis: [supplier, product and revision]. Wetted materials: [register]. Limits and test methods: [schedule]. Makeup and addition authority: [define]. Samples: [locations, cadence and event triggers]. Monitoring: [signals and validity]. Filtration: [removal, flow, pressure loss and replacement]. Excursion response: [owner and confirmation tests]. Intervention approvals: [fluid supplier, equipment supplier and operator]. Closure: [samples, operating checks and records].

Coolant chemistry and filtration review

0 of 8

Common questions

What pH and conductivity should liquid cooling coolant have?

Use the approved fluid and equipment limits, including their test methods and conditions. Treated water and inhibited glycol can have different acceptable values, so one generic target is unsuitable.

Can filtration fix coolant chemistry?

Ordinary particulate filtration does not generally correct dissolved ions, glycol concentration or inhibitor depletion. Chemical treatment or specialized purification needs supplier approval and evidence that required additives remain protected.

Does increasing conductivity prove corrosion?

No. Temperature, additions, concentration and contamination can change conductivity. Validate the result, then use laboratory analysis and inspection to investigate the cause.

How often should coolant be tested?

Agree the schedule with the fluid supplier and owner. Include event-triggered tests after commissioning, unexpected makeup, equipment changes, major service and abnormal trends; adapt routine intervals using operating history.

Is clear coolant safe to use?

Clarity cannot establish compatibility, inhibitor condition or dissolved contamination. Assess it alongside chemistry tests, particle requirements and service history.

Sources and further reading

  1. Guidelines for Using Water-Based Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled RacksOpen Compute Project
  2. Guidelines for Using Propylene Glycol-Based Heat Transfer Fluids in Single-Phase Cold Plate-Based Liquid-Cooled RacksOpen Compute Project
  3. Cooling Solutions for Data Center and Cloud PerformanceDow
  4. Understanding Particle Removal Performance in Liquids with Cartridge and Bag FiltrationPall Corporation

Reliability Engine

Connect coolant condition to operating decisions

Reliability Engine brings available coolant condition and operating context into investigations. Discuss suitable measurements, interpretation limits and complementary laboratory testing for your proposed deployment.