Ultrasonic Flow Meters

20+ Years Manufacturing Experience

CDU Cooling Capacity Calculation: Formula, Flow Meter Accuracy & Measurement Requirements

ENGINEERING REFERENCE

CDU Cooling Capacity Calculation: Formula, Flow Meter Accuracy & Measurement Requirements

How to compute heat rejection from a Coolant Distribution Unit, where measurement error comes from, and what flow meter accuracy you actually need for billing-grade CDU metering.

Every CDU datasheet lists a "cooling capacity" in kW. But that number is not magic — it comes from a simple thermodynamic equation that combines flow rate, temperature difference and fluid properties. In this article we break down the formula, show how flow meter accuracy propagates into total heat uncertainty, and list the sensor requirements you should write into your CDU specification.

1. The Cooling Capacity Formula

The heat rejected by a CDU (or absorbed by the IT load) is calculated as:

Q = ṁ × cp × ΔT

Where:

  • Q = heat transfer rate (kW)
  • ṁ = mass flow rate (kg/s) — derived from volumetric flow × fluid density
  • cp = specific heat capacity of the coolant (kJ/kg·K)
  • ΔT = temperature difference between supply and return (K)

For water at 20–40 °C, cp ≈ 4.18 kJ/kg·K. A common engineering shorthand:

Q (kW) ≈ 500 × Flow (GPM) × ΔT (°F)   — imperial units, water

Q (kW) ≈ 4.18 × Flow (L/s) × ΔT (K)   — metric units, water

Worked example: A secondary loop flows 10 L/s with a supply of 35 °C and return of 45 °C. Q = 4.18 × 10 × 10 = 418 kW of heat rejected to the CDU.

2. How Flow Meter Accuracy Drives Heat Uncertainty

Because Q is directly proportional to flow, any error in the flow meter translates linearly into error in calculated cooling capacity. Temperature sensors and specific heat also contribute, but flow is usually the dominant uncertainty term.

Flow Meter Accuracy Flow Contribution to Q Error Typical Application
±2.0% of reading ±2.0% Internal monitoring, alarm thresholds
±1.0% of reading ±1.0% Commissioning, trend analysis
±0.5% of reading ±0.5% Performance verification, energy reporting
±0.2% of reading ±0.2% Billing-grade, district cooling interfaces

Add a temperature sensor uncertainty (typically ±0.1–0.3 K) and a cp uncertainty (from fluid composition), and a well-installed CDU metering loop typically lands at ±1.5–2.5% total heat uncertainty with a ±1% flow meter. For PUE-critical reporting, that is usually acceptable; for billing, specify ±0.5% flow and calibrated RTDs on both supply and return.

3. Sensor Requirements for CDU Heat Measurement

1.Flow sensor: installed on the secondary loop supply (or return — both work if the loop is closed). Prefer a meter with 4–20 mA or Modbus output integrated into the BMS. Clamp-on ultrasonic is acceptable for retrofit; inline magnetic is preferred for permanent billing-grade metering.
2.Temperature sensors: matched pair of PT100/PT1000 class A RTDs in deep-well thermowells on both supply and return. The pair must be calibrated together — relative error matters more than absolute accuracy.
3.Sampling rate: minimum 1 sample per second for dynamic load; 10-second averages are fine for steady-state reporting.
4.Data integration: Q must be computed in the BMS / DCIM platform, not in the flow meter itself. Most CDU PLCs expose flow and temperature as separate Modbus registers.

4. Common Sources of Error

Air in the loop

Even small bubbles reduce ultrasonic signal strength and bias magnetic readings. Vent the loop fully before commissioning heat measurement.

Incorrect fluid properties

Using water cp on a 30% glycol mix introduces 2–4% bias. Enter actual glycol concentration and temperature in the BMS calculation.

Poor straight run

Turbulence immediately after an elbow or pump skews the flow profile. Respect upstream/downstream straight-run rules or use a flow conditioner.

Frequently Asked Questions

Q1. Should I measure flow on supply or return?

In a closed secondary loop, flow is the same on supply and return. Pick whichever pipe gives the best straight run and easiest sensor access. Do not assume primary-side flow equals secondary-side flow — the plate heat exchanger creates a hydraulic break.

Q2. How often should flow meters be recalibrated?

Clamp-on ultrasonic meters should be field-verified annually (compare against a reference or check paired supply/return balance). Inline electromagnetic meters can go 2–3 years between calibrations, unless the process fluid is abrasive or heavily scaled.

Q3. What ΔT is typical for liquid-cooled GPU racks?

Direct-to-chip loops typically run ΔT of 8–15 °C across the rack. Immersion cooling runs smaller ΔT (3–8 °C) at much higher total flow. Always use the actual measured ΔT, not the design value, in your Q calculation.

Q4. Can I derive IT load from CDU cooling capacity?

Yes — for a closed secondary loop, heat rejected equals IT heat generated (minus small pump heat contributions). This is the basis of liquid-side PUE measurement and is more accurate than airflow-side methods.

Kathy Liang, International Sales at Lanry Instruments

TALK TO OUR ENGINEER
Kathy Liang · International Sales

A core member of Lanry Instruments' global sales team. She focuses on ultrasonic flowmeter solutions and supports international clients to solve complex industrial measurement challenges with tailored proposals and reliable project assistance. Feel free to contact my WhatsApp.


Post time: Oct-28-2026

Send your message to us: