Improvement in Preform Weights from a 48-Cavity PET Injection Molder

John Jones (pseudonym), injection engineer for Corporation X, gave our student project group from Iowa State University the challenge of determining the most influential variables affecting weights of preform plastic beverage bottle produced by injection-molding machines. In addition to determining which variables are most important, we were charged with identifying settings for the variables that will produce optimal preform weights. We benchmarked the current injection-molding process to get an idea to typical performance. This allowed us to compare later experimental results to typical process output. From company experience with the process, the team and Mr. Jones identified injection time, injection pressure, hold time, and hold pressure as candidates for the variables most influential on preform weight. The team devised a matrix of experimental conditions to study that consisted of combinations of high, low, and medium values of these process variables. After experimentation with the injection-molding process, regression analysis was used to help determine which factors are most influential in determining preform weight. The group determined that hold time and hold pressure are the two most influential factors in determining mean preform weight. The group also determined that hold pressure has the primary influence on the consistency of preform weight across the cavities in the mold. Using regression equations for mean weight and a within-die standard deviation, new machine settings for hold time and hold pressure were recommended. Additional runs were then performed to validate our recommendations. Ten runs were made with hold pressure set to 1140 psi and hold time set to 3.95 s. The average weight of 50 preforms from the verification experiment was 23.41 g with a standard deviation of 07 g. These results compare well to the engineering specifications of 23.9–22.9 g set on preform weights. A subsequent comparison of routine process monitoring data collected after implementing our recommendations with historical process data also confirms the substantial improvement provided by our analysis.