Treatment of bimodality in proficiency test of pH in bioethanol matrix

TL;DR

This paper addresses bimodality in proficiency testing of pH in bioethanol by applying statistical methods to improve data interpretation and ensure reliable measurement comparisons among laboratories.

Contribution

It demonstrates the effectiveness of kernel density and mixture models in overcoming bimodality in proficiency test data for bioethanol pH measurement.

Findings

Statistical treatments improved bimodality diagnosis.

Enhanced performance evaluation of laboratories.

Better data interpretation for bioethanol pH testing.

Abstract

The pH value in bioethanol is a quality control parameter related to its acidity and to the corrosiveness of vehicle engines when it is used as fuel. In order to verify the comparability and reliability of the measurement of pH in bioethanol matrix among some experienced chemical laboratories, reference material (RM) of bioethanol developed by Inmetro - the Brazilian National Metrology Institute - was used in a proficiency testing (PT) scheme. There was a difference of more than one unit in the value of the pH measured due to the type of internal filling electrolytic solutions (potassium chloride, KCl or lithium chloride, LiCl) from the commercial pH combination electrodes used by the participant laboratories. Therefore, bimodal distribution has occurred from the data of this PT scheme. This work aims to present the possibilities that a PT scheme provider can use to overcome the bimodality problem. Data from the PT of pH in bioethanol were treated by two different statistical approaches: kernel density model and the mixture of distributions. Application of these statistical treatments improved the initial diagnoses of PT provider, by solving bimodality problem and contributing for a better performance evaluation in measuring pH of bioethanol.