Where Temperature Errors Come From

Temperature is the most common measurement on a process plant, and one of the most trusted. That trust is often misplaced. A temperature reading passes through four stages before an operator or a control system uses it, and each stage can add error: the sensor, the installation, the signal path and the interpretation.

An intermediate technician or engineer must be able to look at a reading and ask which of the four stages is most likely to be wrong. Replacing the sensor is often the first reaction, and often the wrong one.

Worked example 1: the sensor was fine

A heat exchanger outlet temperature reads 8 degrees Celsius below what a heat balance predicts. The sensor is replaced twice with no change. An inspection finds that the thermowell is only 50 millimetres long in a 300-millimetre pipe, and the pipe around it is not insulated. The tip sits in the slower, cooler flow near the wall, and heat is lost through the well to the air. The fault was in the installation. A longer well and insulation remove the error.

Worked example 2: the interpretation was wrong

A storage tank has one temperature sensor near the bottom. The reading is used to correct the volume of product transferred. In summer, the top of the tank is several degrees warmer than the bottom. The single point does not represent the average temperature, so the corrected volume is wrong. The sensor and its signal are accurate; the problem is that one point was assumed to stand for the whole tank.