Process Capability for Prototype and Early Production Parts
When prototype parts are acquired, whether through purchase or made in the company's internal factories, the following methodology is recommended for process capability calculations;
New parts that are very similar to current parts, or made in the same production line or process, can assume the current part process capability. Examples would be fabricated and assembled PCBs. Process capability can be derived from existing manufacturing statistical control data.
2.For parts new to the company, either purchased from the supply chain or locally manufactured, the sampling plan of Table 5.4 can be used for high-volume manufacturing.
3.For low-volume manufacturing, use smaller sample sizes, including the moving range method. Use the statistical techniques of t and x2 distributions as well as sample size determination, discussed in this chapter, to determine the ranges of population average μ and standard deviation CT. Use the confidence limits to deter- mine the worst-case process capability.
4.To determine the specification limits, especially for six sigma design, ensure that the specifications are related to the customer wishes, and that the average and population standard deviations are within the six sigma limits of design.
5.The six sigma or the Cpk quality level target can be altered for the short versus the long term. In some case including prototype an early production, close attention is given to the parts and manufacturing process by the design team and manufacturing engineers in the short term. As production ramps up, more parts are made with newer and less-skilled operators, resulting in poor quality, even if a good control system is in place. In the long term, with good use of corrective action processes and TQM, as well as increased operators' skills through the learning curve, the part’s quality levels will increase. Considering the previous arguments, it is advisable to set a higher quality level in the early production phase in order to counteract the problems when production ramps up. An example would be to set quality for early production runs to Cpk = 1.67 (five sigma), then back off to Cpk = 1.33 (four sigma) in the long term when the product matures. In Figure 5.9, the standard deviation used is the combined CT based on the prototype and production runs.
6. The formulas for combining s (small samples)or a (large samples) from two distinct samples with varying sample sizes and n2) follow. For large samples (>30) of standard deviationσ1, σ2 and sample sizes n1, n2：
To compare large samples to see if the differences between sample averages are significant, a test statistic z is generated: