Measurement uncertainty evaluation in analytical chemistry: Investigating the linearity assumption using an enhanced...

Area of Science:

• Analytical Chemistry
• Metrology
• Chemical Measurement

Background:

• Measurement uncertainty is crucial for conformity assessment and regulatory decisions in analytical chemistry.
• The Kragten spreadsheet method assumes local linearity, which may not hold for chemical models involving ratios of input quantities.
• Non-linearity can lead to underestimation of measurement uncertainty and inaccurate conformity assessments.

Purpose of the Study:

• To investigate the impact of non-linearity in chemical measurement models involving ratios of input quantities.
• To evaluate the performance of an enhanced Kragten spreadsheet method for detecting non-linearity.
• To compare uncertainty estimates from the Kragten method with Monte Carlo simulations for non-linear models.

Main Methods:

• Investigated three common measurement models with ratios of input quantities.
• Employed an enhanced Kragten spreadsheet to assess linearity assumptions by comparing sensitivity coefficients.
• Utilized Monte Carlo simulations to analyze output distributions and coverage intervals for non-linear models.

Main Results:

• Non-linear models were identified, showing deviations from the Kragten method's linear assumption.
• Monte Carlo simulations confirmed asymmetric output distributions and affected accuracy of 95% coverage intervals in non-linear models.
• Significant discrepancies between Kragten and Monte Carlo results were observed, especially with input uncertainties over 15%.

Conclusions:

• The enhanced Kragten approach effectively detects non-linearity in spreadsheet-based uncertainty evaluations.
• Accurate uncertainty estimation requires accounting for non-linearity, particularly in models with ratios of input quantities.
• This method enhances risk assessment and reliability of conformity decisions in analytical chemistry.