Hare, Lynne B.; de Jong, Sijmen (1988, ASQC) Thomas J. Lipton, Inc., Englewood Cliffs, NJ
Tea bag weight data taken during a routine audit showed bimodality. This seemed very unusual, given the configuration of the filling equipment. With the assistance of plant personnel, an experiment was designed to determine the source of erratic fills. The potential sources of variation were fill level of overhead hoppers, differences among the six machines and filler port-to-port (8 ports) variation of each machine. Production personnel had to develop a system for sampling at a higher frequency than ever accomplished before. They then sampled 240 consecutive bags from each machine, once when the hoppers were full and once when they were half full.The resulting weight data showed that four of the six machines performed with very little fill variation. The remaining two shoed different forms of erratic behavior. One showed highly variable fills while the other produced occasional low weights. Reasons for this behavior were found after much difficulty: each machine had been damaged, but in different ways.Machine repairs were made, and new data were taken. These data showed that there was still some significant but small periodicity; not enough to be of concern. Repairs results in more uniform products, reduced over- and under-pack and some financial savings. Some labor savings were realized also as a result of a simpler statistical process control system being implemented.This work was done using graphical techniques, rather than formal statistical tests, for diagnostic and communication purposes.