What is Six Sigma?

The term sigma is a Greek alphabet letter (σ) used to describe variability.  In Six Sigma, the common measurement index is DPMO (Defects Per Million Operations) and can include anything from a component, piece of material, or line of code, to an administrative form, time frame or distance. A sigma quality level offers an indicator of how often defects are likely to occur, where a higher sigma quality level indicates a process that is less likely to create defects. Consequently, as sigma level of quality increases, product reliability improves, the need for testing and inspection diminishes, work in progress declines, cycle time goes down, costs go down, and customer satisfaction goes up.

To have a more comprehensive understanding about sigma quality level, it will be explained from two perspectives of process capability: short-term and long-term process capabilities.

Short-term process capability

A part or item is classified as defective if the desired measurement, denoted by X, is outside the customer-supplier specification limit (USL) or lower specification limit (LSL). In addition to specifying the USL and LSL, a customer would also specify a target value, which typically is the midpoint between the USL and LSL. From a short-term process capability view, after sampling data from the process, a six sigma process that produces the parts is normally distributed (see Figure 2.1). Table 2.1 displays short-term process capability in various sigma levels.

Table 2.1: Short-Term Process Capability at Various Sigma Quality Levels

Long-term process capability

Due to the nature of the process, when dealing with the situation of a long-term process, shifts and drifts in the mean of the distribution of a component value occur for a number of reasons as do changes in other parameters of the distribution: for example, tool wear is one source of a gradual drift, differences in raw material or change of suppliers can cause shifts in the distribution. A solution proposed by D.H. Evans (Statistical Tolerancing: The State of the Art Part III, Shifts and Drifts 1975) focuses on high production rates, and low cost components. Evans suggests that one should use 1.5s as the standard deviation to calculate the percentage of out of tolerance responses. Furthermore, research by M.J. Harry (The Nature of Six Sigma Quality 1988) has shown that a typical process is likely to deviate from its natural centering condition by approximately 1.5 standard deviations at any given moment in time. With this principle in hand, one can make a rational estimate of the long-term process capability with knowledge of only the short-term process capability (see Figure 2.2). Table 2.2 displays long-term process capability in various sigma levels.