Coolant chemistry in Liquid Cooling: the design-stage variable that defines your data center service life

In liquid cooling —fluid-based cooling of computing infrastructure—, conversations about service life usually focus on equipment: the chiller, the CDU (Coolant Distribution Unit — the unit that isolates the IT cooling loop from the facility water loop), the cold plates, the pumps. The international industry documents something different: the variable that ultimately defines availability and system service life is not any of those pieces of equipment in isolation, but the chemistry of the coolant that flows through them. A 0.2-point pH deviation can initiate galvanic corrosion in a copper-aluminum heat exchanger. An inadequate propylene glycol concentration lowers heat transfer and raises viscosity. A depleted corrosion inhibitor leaves unprotected the metals the design assumed protected. Coolant chemistry is the system X-ray; the decision about that chemistry is a project decision, not an operational one.

This article makes the case for why chemistry is decided in project engineering and not in the later operations phase. We walk through the four master variables of the coolant with their internationally documented valid ranges, the circuit materials as a simultaneous constraint on that decision, the verification frequencies the industry recognizes, the design errors that are paid for later, and the closing documentation that sustains the decision through fifteen years of operation.

Coolant chemistry is a design decision, not an operations decision

Two planes that industrial conversation tends to conflate are worth separating.

The first plane is the chemistry decision: which base fluid is specified for the circuit, which corrosion inhibitors it carries, which target glycol concentration, which valid pH and conductivity ranges, which biocide treatment in circuits with Legionella risk. This decision is made during project engineering, jointly with equipment selection and hydraulic topology. Whoever decides which pump, which cold plate and which CDU are installed must simultaneously decide which fluid flows through them, because the chemical compatibility between fluid and materials determines the service life of the assembly.

The second plane is chemistry verification: periodically measuring the fluid variables, adjusting concentrations, replenishing depleted inhibitors, draining and replacing when the analysis calls for it. This verification is the responsibility of subsequent operations and is executed by the client with its in-house team or with a specialized analytical service provider.

Conflating both planes produces two common failure modes. The first occurs when chemistry is not specified in the project engineering and is left to the discretion of the later operator, who does not necessarily know the circuit materials or the ranges the design assumed. The second occurs when later verification is executed against a poorly designed chemistry: the analyses come back "in range" but the system degrades anyway because the ranges were wrong for the installed materials.

ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers), through its technical committee TC 9.9 (Mission Critical Facilities), produces the Thermal Guidelines for Data Processing Environments and the Liquid Cooling: Resiliency Guidance for Cold Plate Deployments published in 2024. Both documents treat coolant chemistry as a design parameter in their H1, H2 and H3 environmental classes for liquid-cooled systems.

The four master variables of the coolant

The international industry documents four variables that define the chemical state of a coolant. Each one has valid ranges associated with system service life, and each one has operational consequences when it leaves its range.

pH.
The coolant must run slightly alkaline, in the 8.0 to 9.0 range, to minimize corrosion of the circuit metals. A deviation of just 0.2 points below the lower bound can initiate galvanic corrosion in heat exchangers that combine copper and aluminum, a common geometry in commercial CDUs. If pH drops below 7, two hypotheses are likely: corrosion inhibitors are depleted, or dissolved metal products are shifting the acid-base equilibrium. ASHRAE Guideline 12-2023, Managing the Risk of Legionellosis Associated with Building Water Systems, treats pH as a control variable in building water systems.

Electrical conductivity.
Conductivity is the most sensitive indicator of fluid contamination. In deionized makeup water for closed-loop circuits, the industry recommends keeping it below 10 µS/cm. In circuits with inhibitor-treated water, the acceptable range rises to 500 µS/cm. When conductivity rises with no explainable cause, the usual hypotheses are three: cross-contamination with a different fluid, metal dissolution from active corrosion, or additive degradation. Monthly verification of this variable catches the problem while it is still small.

Corrosion inhibitors.
Coolant fluids carry inhibitor packages designed for the specific metals in the system. Azoles (tolyltriazole, benzotriazole) protect copper and brass. Nitrites and molybdates protect steels and ferrous alloys. Silicates protect aluminum. The industry evaluates the effectiveness of these inhibitors with two standardized tests from ASTM (ASTM International, U.S. standards organization): D1384-05(2019), Standard Test Method for Corrosion Test for Engine Coolants in Glassware, a fast laboratory procedure to screen out clearly deficient fluids; and D2570-16, Standard Test Method for Simulated Service Corrosion Testing of Engine Coolants, which subjects the fluid to 1,064 hours at 88°C in a loop that simulates real operating conditions. These tests are also applied, with adaptations, to industrial cooling fluids.

Biocides.
In circuits with cooling tower, air-open loops, or any point with aerosolization potential, biocide treatment is not optional. ANSI/ASHRAE Standard 188-2021, Legionellosis: Risk Management for Building Water Systems, requires a documented Legionella risk management plan for the building water systems within its scope. The plan includes periodic monitoring, chemical treatment, and documented response to deviations. The project chemistry specification must address which biocide is compatible with the selected corrosion inhibitors; some combinations are antagonistic and cancel each other out.

The four variables are measured at sampling points in the circuit that engineering must plan from the start. Without accessible sampling points documented on the as-built drawings, later verification becomes operationally expensive and is often skipped.

Circuit materials as a chemistry constraint

Coolant chemistry is not chosen in the abstract. It is chosen against the circuit materials already specified in engineering. Each material has a chemical window in which it operates with maximum service life; leaving that window accelerates degradation.

Copper and its alloys (brass, bronze) are common in heat exchangers and connections. They operate well with alkaline pH and require azole-type inhibitors. They are sensitive to ammonia and chlorides.

Aluminum is common in cold plates and CDU sections because of its thermal conductivity. It requires pH controlled in the 8.0-9.0 range and is affected by sharp deviations toward both extremes. Its galvanic coupling with copper, without an adequate inhibitor, is one of the most documented failure modes in the industry.

Stainless steel 304 and 316 is common in header piping and tanks. It tolerates wider chemical windows but is susceptible to pitting corrosion in the presence of chlorides and to stress corrosion under specific conditions.

Polymers (PVDF, PEX-Al-PEX, EPDM in seals and hoses) have their own chemical compatibility and are sensitive to temperature, fluid plasticizers and oxidizing biocides.

The project chemistry specification must explicitly declare which materials are present in the circuit and which fluid is compatible with that mix. When the client or a later integrator introduces a different fluid than the specified one, because it seemed equivalent or because their supplier recommended it without knowing the installation, that decision breaks the compatibility chain. The Open Compute Project Foundation (OCP), an international hyperscale industry consortium launched by Meta in 2011, in its OCP OAI System Liquid Cooling Guidelines and its ACS Cold Plate Requirements, treats chemical compatibility with materials as a non-negotiable design requirement.

Verification frequencies the industry documents

Once chemistry is decided, later verification runs on three frequency tiers the industry documents consistently.

The first tier is continuous and runs through the BMS (Building Management System — centralized building operations monitoring and control platform) or DCIM (Data Center Infrastructure Management — the equivalent platform specialized in computing infrastructure): in-line sensors measure temperature, pressure and flow in real time. Deviations raise alarms the operator can correlate with chemical changes before they escalate.

The second tier is basic chemical analytics, monthly or quarterly depending on the risk of the circuit. It covers pH, conductivity, glycol concentration and turbidity. It is executed with field instrumentation or with samples sent to an external laboratory. ASTM D3306-21, Standard Specification for Glycol Base Engine Coolant for Automobile and Light-Duty Service, defines quality parameters and verification methods for glycol-base coolants, and is the most widely used international reference for chemical specification of glycol fluids. Uptime Institute, an independent organization founded in 1993 and global reference for data center availability certification, documents that the most widely deployed operating mixture is PG25 (roughly 75% deionized water + 25% propylene glycol), a chemistry with well-characterized material compatibility.

The third tier is extended chemical analytics, quarterly or semi-annually. It covers dissolved metals, microbiological counts, residuals of inhibitors and the additive package composition. This analysis is only reliable when executed by a laboratory accredited under ISO/IEC 17025:2017 (a joint standard from the International Organization for Standardization and the International Electrotechnical Commission), General requirements for the competence of testing and calibration laboratories, the international standard that certifies traceability and technical competence of the laboratory.

For systems within the scope of ANSI/ASHRAE Standard 188-2021, the Legionella management plan also requires periodic microbiological verification with methodology and frequency documented in the building plan.

Project engineering must declare which frequencies apply to the specific circuit, not as abstract recommendation, but as a condition of the technical warranty of the delivery.

Five chemistry design errors paid for later

The industry consistently documents five chemistry design errors that later operations cannot correct.

1. Not declaring target chemistry in the project specification.
The operator inherits a circuit without knowing which fluid is in it, which inhibitors should be present, which ranges are valid. Any decision they make is by default incorrect because the design gave them no criterion.

2. Mixing fluids from different manufacturers in the same circuit.
Inhibitor packages are not interchangeable; combining them can nullify the protection of both. When a top-up is done with a different fluid than the original, chemical traceability is broken and later verification loses meaning. The project must specify brand and model of valid fluid and plan how losses are replenished with the same product.

3. Not planning accessible sampling points in the circuit engineering.
Without a sampling point in each significant hydraulic loop, taking samples requires dismantling connections or introducing contamination into the sample. Quarterly verification becomes operationally expensive and, over time, is dropped. Sampling points are as important as venting points or drain points, and they are designed at the same time.

4. Sizing the expansion tank without considering safe replenishment volume.
Each top-up with untreated water dilutes inhibitors and lowers glycol concentration. If the tank does not have enough capacity for scheduled top-ups with treated fluid, the operator fills with mains water and chemistry silently degrades. Tank sizing must consider evaporative losses, equilibrium leaks in dynamic seals and a maintenance margin.

5. Not documenting valid ranges in the closing folder..
If the operator does not receive in writing the target values for pH, conductivity, concentration and residuals, together with the alarm and critical thresholds, they cannot verify anything because they have nothing to compare against. The closing folder is the last link of the original chemistry decision; without it, the decision is lost with the first personnel change.

What is documented in the Reaclima delivery

In a Reaclima delivery, the project chemistry decision is documented in the closing folder as a standard part of the technical package, not as an optional annex.

The as-built drawings of the circuit explicitly identify the sampling points planned in engineering, with their location and connectors. The technical datasheet of the specified fluid includes brand, model, manufacturer-declared composition, inhibitor and biocide package, and traceability of the supplied lot. The valid operating ranges —target pH and alarm thresholds, maximum conductivity, minimum and maximum glycol concentration, inhibitor residuals— are recorded in the commissioning log, signed at project closing.

The documented BMS control logic includes the planned chemical alarms: out-of-range pH, conductivity above threshold, sustained differential pressure drop in cold plates consistent with internal deposit. Operator training, part of the standard delivery, includes reading and interpreting these alarms and the initial response procedure.

What Reaclima does not include in the delivery is the later recurring analytical service. The accredited laboratory that executes the quarterly chemical verification is contracted by the client directly, at the frequency and scope defined in the closing documentation. This separation keeps the project chemistry decision at its correct level —an engineering decision— and the later verification at its own —operations with independent analytical backing.

The decision that lasts fifteen years

The arithmetic of the chemistry decision is the same as in the previous installment of this cycle on the cost of deferring preventive maintenance. A good engineering decision at project inception sustains fifteen to twenty years of stable operation. A bad initial decision —chemistry not specified, fluid incompatible with materials, sampling points not planned, ranges not documented— is chemical debt with compound interest, paid in accelerated corrosion, premature replacement of heat exchangers, degradation of documented availability.

The international industry has codified this logic in a sequence of standards: ASHRAE 188 and Guideline 12 for Legionellosis, ASTM D1384 and D2570 for corrosion, ASTM D3306 for glycol-base fluids, ISO/IEC 17025 for analytical verification, OCP ACS for material compatibility in liquid cooling. Each one is verifiable in a single search against its original source, with direct links available at the end of this article. The point of the documentation is that the chemistry decision is not left to the individual judgment of the later operator, but rests on traceable technical backing.

Does your next Liquid Cooling project need its coolant chemistry defined from engineering? Let us talk.