Christopher Drake, Statistician, US Army Douglas Ray, PhD, Lead Statistician, US Army
The US Army seeks to continuously improve the downrange target dispersion of 7.62mm sniper ammunition used in high-performance rifles. This interactive case study illustrates the use of several powerful analysis platforms in JMP to achieve this goal. This process begins with the use of the Custom DOE platform and design comparison utilities to select the optimal trade between the number of whole plots and prospective power given sample size constraints. The response data is downrange dispersion (measured as variation in the horizontal and vertical direction), which is inherently noisy. To analyze the response data, we used Loglinear Variance in JMP, a regression technique which effectively analyzes shot-by-shot variation data. Comparisons will be drawn between the Loglinear Variance-derived predictions and the more common least squares fit, using summarized target dispersion data. Using the DOE-based predictive model, Monte Carlo simulation-based sensitivity analyses were executed to compare best-case predicted design configuration, the worst-case, and other configurations of interest, in terms of target group size. These configurations were then compared to the existing ammunition design configuration in terms of customer requirements using the Graph Builder platform, allowing for the visualization of downrange group sizes versus relevant target sizes, and hit probabilities versus target distance. Pareto Frontier analysis enables visual decision support to determine tradeoff between downrange hit probability improvement versus cost of implementing manufacturing process changes associated with the optimal predicted design configuration.