Goodson, Richard M. (1992, ASQC) Johnson Controls, Inc., Watertown, WI 53094
Johnson Controls, a manufacturer of products used in natural and LP gas appliances, received a customer request to meet a new upper limit of 10 cc/hr for one of their standard product specifications.
Using the Weibull distribution to determine if their process could meet the requested product specification, Johnson Controls plotted data on Weibull Probability Paper. They studied the resulting statistics using the Weibull cumulative function to estimate the percentage of the population exceeding the specified limits.
Normal probability paper for a graphical representation or a statistical test of normality, such as the Chi Square Goodness of Fit test, are useful tools for determining if a population is normally distributed. When a process is normally distributed, the standard normal deviate (also known as the Z-score) can be used to estimate the percentage of the population exceeding the specified limit.
Although the Weibull distribution method is used mainly in reliability prediction, it is also a flexible tool for representing a number of distributions from the exponential to normal and skewed distributions. Consequently, the Weibull distribution is a good choice for process characterization.
Goodness of fit,Deviation,Manufacturing,Standard deviation,Weibull analysis,Scoring methods