Adding Glycol Means 4 Percent More Flow: Coolant Properties and Compatibility

A 30 kW cabinet with a temperature rise of 8 K requires 0.897 L/s of pure water at 20°C to maintain thermal balance. When 25% glycol is added, the flow rate increases to 0.94 L/s, a 4.8% increase. This subtle but critical adjustment can be easily overlooked, leading to thermal issues and potential equipment failure.

Understanding the properties and compatibility of glycol-based coolants with stainless steel and other materials is essential for effective cooling system design. This article will explore the specific changes required in flow rates, material selection, and joint types when using glycol as a coolant.

Heat Capacity and Thermal Conductivity with Glycol Concentration

The specific heat capacity (cp) and thermal conductivity (λ) of the coolant are crucial parameters that influence the overall cooling efficiency. For pure water at 20°C, the cp is 4187 J/(kg K) and λ is 0.598 W/(m K). In contrast, a 25% glycol solution has a cp of 3850 J/(kg K) and λ of approximately 0.51 W/(m K).

These differences mean that the coolant's ability to absorb and transfer heat is reduced, necessitating an increase in flow rate to compensate. The formula for calculating flow rate is:

Where V is the flow rate, Q is the heat load, ρ is the density, cp is the specific heat capacity, and dT is the temperature difference. Using this formula, we can see why the flow rate must be adjusted when glycol is added.

Worked Example: Flow Rate Calculation

Let's calculate the flow rate for a 30 kW cabinet with a temperature rise of 8 K using pure water and a 25% glycol solution at 20°C. For pure water:

• Q = 30,000 W (30 kW)
• ρ = 998 kg/m³
• cp = 4187 J/(kg K)
• dT = 8 K

Substituting these values into the formula:

For a 25% glycol solution:

• Q = 30,000 W (30 kW)
• ρ = 1036 kg/m³
• cp = 3850 J/(kg K)
• dT = 8 K

Substituting these values into the formula:

This example clearly shows the 4.8% increase in flow rate required when using a 25% glycol solution.

Comparison Across Different Values

Let's compare the flow rates for different glycol concentrations at 20°C:

• 0% glycol: 0.897 L/s
• 25% glycol: 0.94 L/s
• 40% glycol: 0.960 L/s (cp = 3700 J/(kg K), ρ = 1056 kg/m³)
• 50% glycol: 0.974 L/s (cp = 3600 J/(kg K), ρ = 1070 kg/m³)

As the glycol concentration increases, the flow rate also increases to maintain the same cooling efficiency. This trend highlights the importance of adjusting the flow rate based on the coolant composition.

Field Consequences of Incorrect Flow Rates

Using an incorrect flow rate can lead to several field consequences. For instance, if the flow rate is too low, the coolant may not effectively remove heat from the cabinet, causing overheating and potential equipment failure. Conversely, if the flow rate is too high, it can cause excessive pressure drop and increased energy consumption. A 4.8% error in flow rate, as seen with the 25% glycol solution, can result in a 4.8% reduction in cooling efficiency, which can be significant in high-power applications.

Flow Rate Adjustment Example

For a 30 kW cabinet with a temperature rise of 8 K, the required flow rate for pure water at 20°C is 0.897 L/s. When 25% glycol is added, the flow rate increases to 0.94 L/s, a 4.8% increase. This calculation is based on the following values:

• Pure water at 20°C: ρ = 998 kg/m³, cp = 4187 J/(kg K)
• 25% glycol at 20°C: ρ = 1036 kg/m³, cp = 3850 J/(kg K)

To ensure consistency, all calculations should use the same temperature baseline. At 40°C, the values are:

• Pure water at 40°C: ρ = 992.2 kg/m³, cp = 4179 J/(kg K) -> 0.904 L/s
• 25% glycol at 40°C: ρ = 1020 kg/m³, cp = 3920 J/(kg K) -> 0.938 L/s, a 3.7% increase — smaller than the 4.8% obtained on the 20°C basis, which is precisely why the two bases must never be mixed

Mixing different temperature baselines can lead to incorrect flow rate calculations, which can result in inadequate cooling.

Field Consequences of Mixing Temperature Baselines

Using different temperature baselines for water and glycol can lead to significant errors in flow rate calculations. For example, if the flow rate for pure water is calculated at 20°C (0.897 L/s) and for 25% glycol at 40°C (0.938 L/s), the difference in flow rate is 3.6%. If the flow rate is set based on the 20°C value, the system may be undercooled by 3.6%, leading to thermal issues and potential equipment failure. Consistency in temperature baselines is crucial for accurate flow rate calculations.

What Actually Sets the Corrosion Rate: Coolant Conductivity

The corrosion rate is not set by the potential difference alone. The galvanic current I equals the potential difference divided by the loop resistance, and that resistance is set by the coolant resistivity and the geometry of the current path: R = ρ_e · L / A, where ρ_e is resistivity (the reciprocal of conductivity κ), L the characteristic path length and A the path area. Taking ΔE = 0.75 V, L = 0.05 m and A = 0.01 m² (these geometric terms must be measured on the actual joint; they are stated here as worked-example assumptions), the geometric factor L/A is 5 m⁻¹, so:

The electrochemical equivalent of aluminium is k = M / (nF) = 26.98 / (3 × 96485) = 9.321×10⁻⁵ g/C, and the annual loss is the current multiplied by the seconds in a year and by k. Because the current is directly proportional to conductivity, raising conductivity from 50 to 500 µS/cm (10×) multiplies the aluminium loss rate by ten, from 2.21 g/year to 22.05 g/year. This is the quantitative reason liquid-cooling loops hold coolant conductivity below 50 µS/cm: conductivity is not harmful in itself, but it scales the galvanic current linearly.

Freezing Point and Glycol Concentration

The freezing point of the coolant is another critical factor. A 25% glycol solution has a freezing point of approximately -12°C, while a 40% solution freezes at about -24°C. For extremely cold environments, a 50% glycol solution can be used, with a freezing point of around -37°C.

The table below summarizes the relationship between glycol concentration and freezing point:

Selecting the appropriate glycol concentration ensures that the coolant remains liquid even in very cold conditions, preventing damage to the cooling system.

Field Consequences of Incorrect Glycol Concentration

Using the wrong glycol concentration can have severe consequences. For example, if a 25% glycol solution is used in an environment where temperatures drop below -12°C, the coolant may freeze, causing blockages and potential system failure. Similarly, using a 50% glycol solution in a milder environment can lead to unnecessary costs and reduced cooling efficiency due to the higher viscosity and lower heat capacity of the solution. Proper selection of glycol concentration based on the environmental conditions is essential for reliable operation.

Chloride Ion Limits for Stainless Steel

The chloride ion content in the coolant can cause corrosion in stainless steel. For 304 stainless steel, the recommended long-term limit is 50 ppm, while for 316L, it is 250 ppm. These limits are only meaningful with a temperature attached. The 50 ppm figure corresponds to the chloride stress corrosion cracking (Cl-SCC) threshold near 60°C; it is not a temperature-independent constant. Coolant in a liquid-cooling CDU normally runs at 35–45°C, below 60°C, where Chinese industry practice allows roughly 200 ppm for 304 in routine service. Once local temperature exceeds 60°C the limit halves, and above 100 ppm the Cl-SCC risk becomes real. Two further code anchors: hydrostatic test water for stainless pipe must not exceed 25 mg/L chloride (GB 50235), and recirculating cooling water may be relaxed to 1000 mg/L with an effective inhibitor programme (GB 50050-2017).

The table below outlines the chloride ion limits for different types of stainless steel:

Regular monitoring and maintenance are essential to ensure that the chloride ion levels remain within safe limits.

Field Consequences of Exceeding Chloride Ion Limits

Exceeding the chloride ion limits can lead to significant corrosion issues. For example, if the chloride ion level in the coolant exceeds 50 ppm for 304 stainless steel, pitting and stress corrosion cracking can occur. This can result in leaks, reduced system efficiency, and costly repairs. Regular monitoring and maintenance, including periodic testing of the coolant, are necessary to prevent such issues. For 316L stainless steel, the tolerance is higher, but it is still important to keep the chloride ion levels below 250 ppm for long-term reliability.

Crevice concentration is the failure path that gets missed. In narrow gaps under flange faces, threads and gaskets the fluid stagnates and chloride can concentrate by a factor of 10 to 100. At a bulk level of only 25 ppm the crevice already sits at 250–2500 ppm, which breaks the 50 ppm limit for 304 outright and reaches or exceeds the 250 ppm limit for 316L. That is why the field pattern is so often "bulk sample passes, flange leaks first": the criterion has to be written against the crevice concentration factor, not against the bulk reading alone.

Galvanic Corrosion and Protective Measures

When aluminum and stainless steel are in contact, galvanic corrosion can occur. The potential difference between an aluminum cold plate and 316 stainless steel is approximately 0.75 V, making aluminum the anode and more susceptible to corrosion. To prevent this, several protective measures can be taken:

• Insulating joints to break the electrical path
• Using sacrificial anodes (zinc or magnesium) that dissolve in place of the aluminium, which is the anode of the couple
• Adding corrosion inhibitors to the coolant

Calculation basis. All primary examples use 20°C properties (water: cp 4187 J/(kg·K), λ 0.598 W/(m·K), ρ 998 kg/m³; 25% glycol: cp 3850, λ 0.51, ρ 1036) for a 30 kW cabinet with an 8 K rise. On that basis water needs 0.897 L/s and 25% glycol needs 0.940 L/s — a 4.8% increase (the "4 percent" in the headline is the rounded-down conservative statement; allow 5% when sizing). The 40% glycol properties are linearly interpolated from the 25% and 50% points (cp 3700 J/(kg·K), ρ 1056 kg/m³) and agree to within 0.5% with independently published values (cp 3680, ρ 1056 → 0.965 L/s). On a 40°C operating basis the same case gives 0.904 L/s and 0.938 L/s, a 3.7% increase — the two bases must not be mixed: taking water at 20°C and glycol at 40°C produces a spurious 3.65% discrepancy. Included: sensible-heat flow correction, chloride concentration and temperature, galvanic loop resistance, straight-pipe pressure drop. Not included: boiling margin, inhibitor film kinetics, biofouling, and local electrochemistry at sensitised weld zones.

Field Consequences of Ignoring Galvanic Corrosion

Ignoring galvanic corrosion can lead to rapid degradation of the aluminum components. For example, if the potential difference between an aluminum cold plate and 316 stainless steel is not managed, the aluminum can corrode, leading to leaks and system failure. This can result in significant downtime and repair costs. Implementing protective measures, such as insulating joints and using sacrificial anodes, is essential to prevent galvanic corrosion and ensure the longevity of the cooling system.

Why the Velocity Ceiling Is 5 m/s: Pressure Drop, Not Erosion

For DN25 stainless tube, 10 m long, carrying 25% glycol at 20°C (dynamic viscosity 2.23 mPa·s, density 1036 kg/m³, kinematic viscosity 2.153×10⁻⁶ m²/s), the Blasius smooth-pipe friction factor is f = 0.316 / Re^0.25, and the straight-pipe pressure drop is Δp = f · (L/d) · ρv²/2:

Raising velocity from 3 to 5 m/s is only a 1.67× change, yet pressure drop grows 2.45× and shaft power 4.07× — pressure drop scales at roughly v^1.9, and power adds one more factor of flow. The economic meaning of the 5 m/s ceiling is therefore a pump-power wall rather than a pure erosion limit. The 3 m/s floor exists to keep the flow turbulent (Re ≈ 35,000 here, far above the 4000 transition) so that the heat-transfer coefficient holds up and stagnant pockets cannot form.

Common Mistakes and Their Costs

1. Incorrect Flow Rate Calculation: Using different temperature baselines for water and glycol. This can lead to a 3.65% error in flow rate, causing insufficient cooling. The correct approach is to use a consistent temperature baseline, such as 40°C, for all calculations.
2. Exceeding Chloride Ion Limits: Allowing chloride ion levels to exceed 50 ppm for 304 stainless steel or 250 ppm for 316L. This can result in pitting and stress corrosion, costing thousands in repairs and downtime. Regular monitoring and maintenance are necessary to keep chloride ion levels within safe limits.
3. Ignoring Galvanic Corrosion: Failing to insulate joints or use sacrificial anodes when aluminum and stainless steel are in contact. This can lead to rapid corrosion of the aluminum, resulting in equipment failure. Proper design and installation of protective measures are essential.
4. Excessive Flow Velocity: Operating stainless steel components at flow velocities above 5 m/s. This can cause erosion and mechanical wear, reducing the lifespan of the cooling system. The recommended flow velocity for stainless steel is 3-5 m/s.
5. Inadequate Maintenance: Neglecting regular testing and maintenance of the coolant. This can lead to degraded coolant quality, increased corrosion, and system inefficiency. A comprehensive maintenance plan, including regular testing and coolant replacement, is necessary to ensure optimal performance.

Adjusting the flow rate by 4.8% when adding 25% glycol, ensuring chloride ion limits, and implementing protective measures against galvanic corrosion are critical steps in maintaining an efficient and reliable cooling system.

Cross-Check Against Published Property Tables

Whether the properties used here match published data decides whether the flow conclusion holds. Against a published property table at 20°C on a mass-fraction basis: at 20% glycol cp = 3880 J/(kg·K), λ = 0.52 W/(m·K), ρ = 1027 kg/m³; at 30% glycol cp = 3690, λ = 0.48, ρ = 1041. Interpolating to 25% gives cp ≈ 3785, λ ≈ 0.50, ρ ≈ 1034.

All three sit within 2%, so the values used here fall inside the mainstream engineering handbook family and the flow conclusion (a 4.8% increase) is robust. One caveat worth stating: glycol property tables exist on both a mass-fraction and a volume-fraction basis, and at the same nominal concentration the two bases can differ by more than 1% in density and noticeably more in freezing point (40% by volume corresponds to about -23°C). Confirm which basis a handbook uses before mixing values from two sources.

Mass Flow Versus Volume Flow: A Four-Point Gap Between Two Bases

Every example above is a volume flow, V = Q / (ρ·cp·ΔT), because pumps are selected on volume flow. But heat transport depends on mass flow, ṁ = Q / (cp·ΔT), which involves specific heat only and is independent of density. Glycol is heavier than water, so the two are not equal:

The gap widens with concentration (4.0 → 6.2 → 7.8 points). Applying the mass-flow increase directly to pump selection oversizes the pump — the most common mix-up in this field.

Cross-checking published figures

① One published liquid-cooling flow table gives a "30% glycol flow adjustment factor of 1.18, an 18% increase". Back-calculating from its own cp values (water 4.18, 30% mix 3.52) gives 4.18 / 3.52 = 1.1875, which shows 1.18 is a mass flow ratio, not a volume one. Converted to volume flow it becomes 1.1875 × (998 / 1041) = 1.138, i.e. +13.8%. Sizing a pump on a genuine 18% volume increase overspecifies by about 4.2 points.

② A heat-exchanger selection guide gives the flow ratio as (cp_water / cp_mix) × (ρ_mix / ρ_water) — the density ratio is inverted. The correct form is (cp_water / cp_mix) × (ρ_water / ρ_mix). On its own 60/40 example the printed form yields 1.246 while the correct form yields 1.181, an overstatement of about 6.5 points; and the result is labelled "mass flow", whereas the true mass flow ratio is simply the cp ratio, 1.213 (+21.3%), not 25%. Three different numbers point to both a formula error and a basis error.

③ A commercial glycol-based electronics coolant lists, at 20°C: density 1058 kg/m³, specific heat 3480 J/(kg·K), thermal conductivity 0.420 W/(m·K), dynamic viscosity 2.96 mPa·s, chloride 3.0 mg/L, pH 8.5, minimum service temperature -26.5°C. Substituting those measured values into the same 30 kW / 8 K case gives a volume flow of 1.019 L/s, a 13.5% increase over water, whereas the 40% row here, built on an interpolated cp of 3700, gives 0.960 L/s (+6.95%) — a difference of about 6.3%. The cause is that specific heat falls slightly faster than linear interpolation suggests at high concentration, and handbook families disagree by up to 10% at 50%. The practical conclusion: the examples here establish the method; for a real design use the manufacturer's test certificate for the coolant you are actually buying.

Corroboration from operators. A published liquid-cooling facility white paper states that a 25% propylene glycol mix has cp ≈ 3.85 kJ/(kg·K) against 4.18 for water, and that the loop "needs roughly 8 percent more flow to carry the same heat". Back-calculated on the basis used here, 4.18 / 3.85 = 1.086, i.e. a mass flow increase of 8.6% — within 0.2 points of the 8.75% this article derives for the 25% case, an independent confirmation on the same basis. The same paper sets a conductivity target below 50 µS/cm (pH 8.5–9.5, molybdate inhibitor reserve above 200 ppm), which matches the control limit used in the conductivity example here exactly. The difference is that it states a target, while this article quantifies the consequence of exceeding it: aluminium loss rising from 2.21 g/year to 22.05 g/year. An engineering consultancy likewise frames the penalty in mass flow — about 7% more at 25% glycol and about 15% at 40% — the same basis and the same order of magnitude as the 8.75% / 13.16% derived here. A heat-exchanger manufacturer adds that "316L does not mean unlimited chloride resistance; chloride level, temperature, pH and oxygen ingress must be evaluated together", and that specifying a universal conductivity or pH limit is simply not correct — which corroborates both points this article keeps returning to: limits need a temperature attached, and a criterion has to state its basis.

FAQ

How does the addition of 25% glycol affect the specific heat and thermal conductivity of water?

Adding 25% glycol reduces the specific heat from 4187 to 3850 J/(kg K) and thermal conductivity from 0.598 to about 0.51 W/(m K).

What is the recommended chloride limit for 304 and 316L stainless steel in long-term service with glycol?

For long-term service, the chloride limit for 304 stainless steel should be around 50 ppm, and for 316L, it should be around 250 ppm.

How much does the flow rate need to increase when using 25% glycol for a 30 kW rack with a 8 K temperature difference?

The flow rate needs to increase from 0.897 L/s (pure water) to 0.94 L/s (25% glycol), which is an increase of about 4.8%.