Home / Guides / Data centers

Liquid-Cooling Flow Rate: Calculate L/min From kW

Calculate a first-pass water flow from heat load and temperature rise. Includes L/min and US gpm examples, glycol limits and OEM flow requirements.

US Liquid Cooling editorial · Reviewed · 4 min read

For a single-phase liquid-cooling loop, the heat-balance flow estimate comes from heat load, fluid heat capacity and the allowed temperature rise. For water using an approximate specific heat of 4.18 kJ/kg·K and density of 1 kg/L:

Flow in L/min ≈ 14.35 × heat load in kW ÷ coolant temperature rise in °C.

That is a useful starting calculation. It is not a substitute for the server's minimum flow, maximum pressure, cold-plate performance or the pump's actual operating point.

Start with the heat entering this liquid loop

Rack electrical power and liquid heat load are not always equal. If an illustrative 100 kW rack sends 80% of its heat to the liquid system, use 80 kW for that loop's heat balance. The remaining heat still needs an air-side path.

Use the equipment manufacturer's documented capture fraction and design conditions. Include auxiliaries that actually contribute heat within the chosen boundary. Do not add a percentage merely because a planning spreadsheet usually includes one.

Also distinguish a rack branch from a CDU group and the whole facility. A total system flow does not tell you whether every branch receives enough liquid.

Work through a simple water example

The underlying relationship is Q = mass flow × specific heat × temperature rise. For 100 kW and a 10°C rise, mass flow is approximately 100 ÷ (4.18 × 10), or 2.39 kg/s. Multiplying by 60 and using the stated density approximation gives about 144 L/min.

For the same assumed 100 kW water duty:

Temperature rise across the loadApproximate flowApproximate US gpm
5°C287 L/min76
10°C144 L/min38
15°C96 L/min25

These are calculated examples, not recommended operating points. They show that reducing the allowed temperature rise increases the required flow for a fixed heat load.

Which temperature difference belongs in the formula?

Use the difference between supply and return of the circuit whose heat load you are calculating. Do not substitute the CDU approach temperature, which compares temperatures on opposite sides of a heat exchanger.

For example, a technology loop supplied at 35°C and returning at 45°C has a 10°C rise. A separate facility loop may have its own supply temperature and rise. Calculate each using its own fluid and heat-transfer boundary.

For that distinction, see CDU approach temperature explained.

What changes with glycol or another fluid?

Density and specific heat change with fluid type, concentration and temperature. Viscosity also affects pressure losses and pumping requirements. Do not apply the water shortcut unchanged to a glycol mixture or a dielectric fluid.

The ASHRAE secondary-coolant properties chapter is a reference for those property differences. Use the actual approved fluid's technical data at the operating temperature when performing a design calculation.

The method still works: obtain heat capacity and density, calculate mass flow, and convert to the desired volumetric unit. Keep the unit system visible so a value in L/s does not accidentally become L/min.

Why can the heat balance disagree with an OEM flow requirement?

The heat balance only accounts for the energy transported by the liquid. A cold plate may need more flow to achieve the required local heat transfer or temperature distribution. Some assemblies also impose minimum branch flow and maximum pressure-drop limits.

A pump must provide the required flow against the complete circuit resistance. Adding manifolds, filters or disconnects changes that resistance. The CDU's headline maximum flow is not necessarily the flow available through your completed rack network.

Vertiv's CDU 100 planning guide illustrates how product selection depends on operating conditions rather than one nominal number.

A useful calculation handover

Record the heat boundary, normal and peak load, fluid specification, property temperature, supply and return targets, calculated flow and OEM constraints. Add the pressure-loss calculation and selected pump duty separately.

Check the numbers against measured load, temperatures and flow during commissioning, allowing for instrument uncertainty and changing workload. A mismatch is a reason to inspect assumptions and measurements before changing pump settings. For a broader overview of the equipment involved, read what a CDU does.

Sources & further reading

Use the manufacturer’s instructions for your exact equipment. Read our editorial policy or suggest a factual correction.

Explore the company directory → · All liquid-cooling guides →

01
Data centers

Assess whether data-center heat reuse fits your site

Evaluate the heat customer, temperature requirements, seasonal demand, connection costs, and operating responsibilities before promising heat reuse.

↗
02
Buying guides

Laser Chiller Buying Guide: What to Check Before Choosing

The right laser chiller depends on more than the wattage printed on the machine. Build a useful specification around heat load, flow, temperature, and support.

↗
03
Data centers

45°C AI Cooling: When Can a Data Center Skip Chillers?

Understand 45°C AI cooling, dry-cooler and CDU temperature budgets, and why chiller-less operation depends on supported hardware and local conditions.

↗

Directory matches are starting points for research. Confirm product fit and availability with the company.