Calibration
What does the term mean?
Calibration gives a measuring system’s indication a documented relationship to known reference values. Indication and reference are compared under described conditions and their relationship evaluated, including uncertainties. Changing the instrument setting afterwards is a separate step: adjustment.
Relevance to process measurement
Indirect process measurement requires matching sensor readings and laboratory findings. Samples should cover the product states encountered in operation. A curve from a narrow subset does not establish suitability across the whole intended range.
How should a process calibration be prepared?
Define the target quantity, reference method and sampling procedure. Record the corresponding sensor reading and relevant conditions, including temperature and recipe, for every sample. Reference values need adequate reliability so their scatter does not dominate the assessment.
Use additional check samples that were not used to build the curve. Document applicable products and conditions. Installation changes, product changes or unusual deviations may call for a review. Add points that meaningfully extend coverage of actual operation.
Independently verify a calibration curve
Control samples that were not previously used to generate the measurement curve should be used to evaluate it. Otherwise, the assessment will primarily focus on how well the existing data is reproduced. Select control samples from the intended operating range and record the product, temperature, sampling time and reference method. A single matching point is not sufficient to confirm the correlation for all typical operating conditions. Similarly, an unusual individual value should first be investigated to determine its cause.
In the event of discrepancies, first compare the sampling procedure, the reference measurement and the timing. If repeated laboratory analyses of the same sample show significant variation, this variation must also be taken into account when evaluating the sensor. A calibration cannot demonstrate greater reliability than the references used for it permit. The intended use of the measured value determines which deviations are acceptable during operation.
Keep the valid version of the calibration curve and its scope of application together with the test records. Following an adjustment, it should be possible to trace which setting was changed and which subsequent control samples are available. In the event of a product change or a modification to the measuring point, this makes it possible to determine whether the previous calibration is still sufficiently validated.