Master of Science (MS)
Industrial and Systems Engineering
H. S. Jacob Tsao
Statistics.; Engineering, Industrial.; Computer Science.
The Dodge-Romig plans and the Military Standard 105 Plans are the most popular published sampling plans covered in detail in many standard textbooks on statistical quality assurance such as Introduction to Statistical Quality Control by Douglas C. Montgomery. In this thesis, we focus on Dodge-Romig AOQL plans. The Dodge-Romig AOQL plans are designed to minimize the average total inspection (ATI) for a given AOQL and a specified process average. It is argued that the Dodge-Romig plans as currently tabulated (particularly in the standard publications or textbooks on acceptance sampling), are not useful. However, they can be made useful if the plans focus on ranges of the (incoming) process-average values that are greater than the target AOQL. Whether one knows the exact value of process average p or just the exact range of the process averages containing an uncertain p, one can use the pair of (n, c) listed under the corresponding range to ensure the target AOQL with the minimum ATI. The thesis provides example expansions of some of the published Dodge-Romig AOQL plans. A new concept of certainty line is also developed which is a measure of process-average (p) associated with gamma risk (gamma) whose numerical value is 0.000033. For any given lot size if the calculated or knowledge base process-average of a manufacturer falls below or equal to the value of the certainty line, then one would not be required to consider any sampling plan to achieve a worst case outgoing quality.
Ganguly, Avik, "Dodge-Romig sampling plans: Misuse, frivolous use, and expansion for usefulness." (2009). Master's Theses. Paper 4002.