(38a) Risk Analysis for Count Data

The methodology for performing risk analysis calculation in the presence of process and measurement variation has been well established in the literature [1-4]. However current methods are limited to continuous distribution, specifically normal distribution. Even though a majority of quality measurements made within Dow are continuous in nature, there are still discrete measurements such as counts of defects, grade level assignments to name a few. In addition to this, the advances made in the field of metrology have also allowed higher resolution measurements be made on the process or products. One such capability is the ability to count low-level defects continuously on product, and such count data typically do not fit a normal distribution. The work here is to extend the risk analysis on low level defect count data which can be best described using the Poisson distribution.

[1] Burdick, R. K.; Borror, C. M.; and Montgomery, D. C. (2003). A Review of Methods for Measurement Systems Capability Analysis. Journal of Quality Technology, 35 (4), pp. 342–354.

[2] Mader, D. P.; Prins, J.; and Lampe, R. E. (1999). The Economic Impact of Measurement Error. Quality Engineering 11(4), pp. 563–574.

[3] Mottonen, Matti; Belt, Pekka; Harkonen, Janne; Haapasalo, Harri; Kess, Pekka; Manufacturing Process Capability and Specification Limits, The Open Industrial and Manufacturing Engineering Journal, 2008, 1, 29-36.

[4] Basnet, Chuda; Case, Kenneth E. (1992). A Review of Methods for Measurement Systems Capability Analysis. Quality Engineering, 4 (3), pp. 383–397.