Tip 1: use a probe whose length, in addition to the length of the sensing element inside, is at least 20 times the diameter. This should ensure that the immersion error is no more than 0.1 % of the difference between the liquid temperature and the outside environment temperature. The error will be somewhat larger for static liquids, though some natural convection is always likely to be present.

Tip 2: where possible allow extra length: an increase of 50% in the immersion depth could reduce the error by as much as a factor of ten.

Tip 3: under steady conditions, test for immersion errors by withdrawing the sensor by 2, 4, 6, etc, cm to find out how the measurements are affected. Wait for long enough (several minutes) at each point for the new conditions to settle, and make a further measurement when the probe is reinserted.

The error will be larger if the probe has to be inserted into an additional protective tube; for example, if it is too short and the cable is not compatible with the liquid. In many cases, such as in pressurised chemical reactors, the probe is inserted into a steel ‘thermowell’ which is built into the vessel wall and from which the probe can be removed without affecting the process inside. In this situation the thermowell, and the inevitable air gap between it and the probe, are both additional barriers to heat transfer, which means that deeper immersion is needed.

Tip: the recommended immersion depth should be at least 20 times the diameter of the thermowell, not just the probe. Again, allow extra length where possible. In confined geometries, such as pipes, extra length can be gained by locating the thermowell at bends or elbows, or inserting it at an oblique angle. Test for immersion errors as suggested above.