Measurement errors can be undetected and highly significant. We had a great example of sensor radiation error recently.
I took a telephone call from a customer telling me my temperature sensors were inaccurate. He was building an AMS 2750D compliant Aerospace Furnace. He bought a number of thermocouples from Peak Sensors which were all UKAS calibrated.
There was a combination of Mineral Insulated Type N and Ceramic Insulated Type R (wire Ø 0.25mm) sensors. In four locations, a type N and a type R were installed adjacent to each other. One would expect them to read very similar temperatures. During pre commissioning trials it was noticed that at times there was up to 50°C difference between the pair. The largest errors were during initial warm up. The reading difference reduced when the furnace was held at temperature and disappeared during cool down.
Systems and instrumentation were double checked and confirmed as correct. Of course the sensors got the blame and the calibration was questioned.
The furnace was electrically heated by Kanthal Super Elements. These are intensely bright in operation. It became apparent that the direct line of sight between the elements and the sensors caused the larger type N sensors to collect significantly more radiant energy and therefore read higher. An extreme example of radiation error was identified.
The solution was to install a ceramic radiation shield around the sensor pairs. It totally solved the problem. Radiation error can be reduced by realigning a temperature sensor to reduce the sensor surface area visible to the radiation source.
Radiation Error Simply Explained
Direct radiation from a heat source will be absorbed by different targets at different rates. A sensor with radiation errors will therefore show a significantly different to the temperature of its ambient surroundings. A familiar example is a mercury or alcohol filled thermometer left in the sun. It will read much higher than the ambient air temperature. Stevenson Screens provide the solution to this radiation error.
Bigger Questions About Error
While it is great to solve a problem, it inevitably raises many questions. The customer would never have been aware of the problem if there was not two significantly different measuring systems. There are many other sources of temperature measurement error. If furnace operators rely on one sensor type and do not consider potential errors, how confident can they really be about their processes.