The most common and most frequently measurable variable in the process industry is temperature. When performing a temperature calibration, it is important to remember that the uncertainty of calibration is not the same as the accuracy of the device. Many factors influence the total uncertainty, and performing a calibration is not the least influencing factor. Using a dry block is a fast and efficient way to get accurate findings during a calibration out in the field.
Uncertainty is an estimate of the limits, at a given cover factor (or confidence level), which contain the true value. Uncertainty is evaluated according to either random error or systematic error. Random (Type A) involves the statistical analysis of a series of measurements and vary in an unpredictable manner. Systematic errors (Type B) or effects remain constant during the measurement. Examples of systematic effects include errors in reference value, set-up of the measuring, ambient conditions, etc.
Before you calculate the total uncertainty of a calibration performed with a dry block you should be aware of two things:
- The EURAMET guideline: (EURAMET cg-13, Version 2.0 (03/2011) this guideline outlines the normative way to calibrate dry blocks. Nowadays, most manufacturers include EURAMET guides with their product specifications.
- Various uncertainty components depend on whether you are using the dry block’s internal measurement or an external sensor as the reference. If the control engineer pushes for a tight performance accuracy of less than one degree, then an accurate external reference probe should be used. The typical dry block accuracy is anywhere from ±0.3ºF to ±1.0ºF, and it is good practice to have a 4:1 ratio of test equipment versus process measurement. In order to make a proper temperature simulation, a reference probe (RPRT or SPRT, reference or secondary primary resistance thermometers) along with an accurate PRT meter would both need to be utilized to achieve an improved measurement error of ±0.1ºF to ±0.2ºF. Note that this imposes a more significant investment in test equipment and it will require a higher cost of maintenance for the more accurate test equipment. For example, what if the quality engineer reports that an error of ±5ºF is all that is needed to make good product? Why impose an unnecessary burden on the instrumentation department? If the control engineer has no objection (along with input from reliability, safety, etc.), a practical approach would be to set a loop tolerance of ±2.0ºF, assuming the dry block is accurate to ±0.5ºF over the range of 50 to 250ºF. While not as accurate as the instrumentation in the loop, it is better than 2:1 for what is required to make a quality product and allows the calibration technician to utilize a simple combination of equipment.
Users need to define the error tolerance required to make good product and/or operate the most efficiently. This will determine the accuracy of the test equipment required. Based on the above discussion, an RPRT is required for uncertainties below ±1ºF, or an accurate dry block can make satisfactory measurements with an uncertainty above ±2ºF.