AHU Temperature Traverses: Weight Each Reading by Airflow Before Calculating Load

Original illustrative engineering schematic. The worked calculation below uses explicit example inputs.
An AHU leaving-air traverse can look reassuring on a temperature schedule while understating the temperature carried by most of the air. The problem is not necessarily a faulty thermometer. If the warmer parts of the face also carry more air, an equal-area temperature average gives those warm streams too little influence. The useful question for an energy balance is what temperature represents the transported energy, not which temperature represents an average position.
This article develops a dry, sensible-only calculation with paired local airflow and temperature readings. It is a data-reduction example, not a prescribed test-point layout or an acceptance standard. A competent testing team must choose suitable access, instrumentation, locations and stabilization criteria for the actual installation.
Start with the quantity that crosses each area
NASA Glenn describes mass flow as density multiplied by velocity and flow area. Apply that continuity relationship to small cells across a measurement plane. Use the local velocity component normal to the measurement plane. For cell i, mass flow is m_i = rho_i × v_i × A_i. Its contribution to sensible energy transport, using a common reference temperature and constant specific heat, is proportional to m_i × T_i. The representative temperature is therefore the sum of those contributions divided by the total mass flow.
T_flow = sum(m_i × T_i) / sum(m_i).
For equal cell areas and approximately equal density: T_flow = sum(v_i × T_i) / sum(v_i).
An arithmetic average is valid for this purpose only when the mass-flow weights are equal, or when their correlation with temperature happens to cancel.
Keep units consistent. Velocity in m/s times cell area in m² gives m³/s. Multiplying by density in kg/m³ gives kg/s. Temperature differences in kelvins and degrees Celsius have the same numerical magnitude, so either can be used consistently in a sensible heat calculation. Absolute Celsius temperatures can be used in this weighted average because the weights sum to one.

Original dry-air teaching example. Cell areas are each 0.50 m²; approximately uniform density is assumed.
Original four-cell example
Consider a 2.0 m² leaving-air plane divided into four equal 0.50 m² cells. The cells have velocities of 1.0, 2.0, 3.0 and 4.0 m/s, paired with temperatures of 10, 12, 14 and 16°C respectively. Assume each reading adequately represents its cell, steady flow, negligible leakage and approximately uniform density of 1.20 kg/m³. These are invented teaching inputs, not a measurement from a Nexora customer or a manufacturer selection.
Cell volume flows = 0.50, 1.00, 1.50 and 2.00 m³/s.
Total volume flow = 5.00 m³/s; total mass flow = 6.00 kg/s.
Equal-area temperature average = (10 + 12 + 14 + 16) / 4 = 13.0°C.
Flow-weighted temperature = (1×10 + 2×12 + 3×14 + 4×16) / 10 = 14.0°C.
The warmest cell carries 40% of the airflow but receives only 25% of the weight in the simple average. The coldest cell carries 10% of the airflow yet also receives 25% of the simple-average weight. The errors reinforce each other. The resulting one-degree difference is a consequence of the specified distribution, not an assumed instrument bias.
What that changes in a sensible-duty calculation
Now assume the entering-air plane is uniform at 26.0°C and that dry-air specific heat is approximated as 1.006 kJ/(kg·K). There is no condensation or humidity change in this example, and no fan heat or casing heat gain between the stated measurement boundaries. Use the same 6.00 kg/s mass flow on both sides.
Sensible duty using paired data = 6.00 × 1.006 × (26 − 14) = 72.432 kW.
Duty using the arithmetic temperature average = 6.00 × 1.006 × (26 − 13) = 78.468 kW.
Overstatement = 6.036 kW, or 8.33% of the paired-data result.
This does not establish actual coil performance to that precision. Real density variation, sensor uncertainty, spatial undersampling and time variation may be important. The extra decimals simply make the arithmetic reproducible. Report an appropriately rounded result and the measurement limitations rather than presenting a calculator output as an accuracy statement.
Keep the pairings when you move data into a spreadsheet
Give each cell a stable identifier, its represented area, velocity, temperature and measurement time.
Calculate local flow before combining temperatures; retain the original readings alongside derived values.
Confirm whether the velocity instrument reports actual or standard velocity and apply its documented density convention.
Record operating mode, fan speed, damper positions and water-side conditions so a later traverse is comparable.
Investigate missing, reverse-flow or unstable readings instead of replacing them silently with the overall average.
Averaging velocities and temperatures separately destroys the information about which temperatures belong to the largest flows. If the coil condition drifts during the traverse, even correctly paired spatial readings may not represent one operating state. Repeat or time-align the survey using a suitable procedure before drawing conclusions about maldistribution.
Where temperature weighting stops being enough
For a wet cooling coil, use a moist-air energy balance on a consistent dry-air mass basis and pair local humidity information with the relevant flow. Average specific enthalpy by dry-air mass flow, not relative humidity by area. An average RH combined with an average temperature does not generally reproduce average enthalpy. Condensate and other heat terms may also matter at the selected boundary.
The practical next step is small: add local-flow and weighted-temperature columns to the existing traverse record before comparing airside duty with a selection or a water-side check. The Air Balancing Toolkit and field handbook provide a related framework for organizing TAB readings and reports; this worked example does not imply that its workbook contains an automatic temperature-traverse calculation.
Continue the calculation review
Sources and calculation basis
Primary references checked on 24 October 2026. The worked examples and diagrams are original illustrative calculations, not manufacturer selections or project approvals. Apply the stated limitations and the requirements governing the actual project.

Comments