Browsing by Subject "control charts"

The modern definition of quality states that quality is inversely proportional to variability. According to this view, understanding and reducing process variation are key components of process improvement. Consequently, it is no wonder that statistical methods play such an essential role in modern quality management systems. Over the last decades, there has been a growing interest in statistical process improvement methods in different industries. The purpose of this thesis was to examine how statistical methods could be used in quality monitoring and improvement of bottling process. The literature review dealt with the role of statistics in quality management and statistical techniques, especially statistical process control tools, which are most commonly used in process improvement. The role of material losses in process quality was also reviewed. The aim of the experimental part was to evaluate the appropriateness of current in-process sampling methods, measurement methods and specifications of opening torque, reclosing torque and fill volume. Additionally, the aim was to develop a method for measuring liquid and packaging material losses of the bottling process. Retrospective control chart analysis and process capability study were used to evaluate the appropriateness of specification limits. Optimal sample size for process monitoring was determined based on statistical power analysis. Destructive Gage R&R study was performed to assess the capability of measurement system and to identify the sources of measurement variation. Liquid losses were calculated for eight different liquids from years 2014–2016 and packaging material losses were calculated in respect of capsules over a period of three months. The process capability study indicated that it was possible to tighten the specifications of opening torque and fill volume. The opening torque specifications were also relatively wide in relation to general recommendations, and therefore, it was advisable to update the specifications. The power analysis showed that the present sample size used for process monitoring was inadequate. The optimal sample size was approximately five bottles. Study considering liquid losses revealed that the calculation of bottling batch specific liquid losses was not possible in most of the cases. In order to benefit from measuring liquid losses, monitoring should be conducted at bottling batch level. Flow meters would be required for this purpose. It was also discovered that the use of packaging material losses as a quality indicator of bottling process was not meaningful, since the attainable precision for loss calculation was not sufficient