Chapter 17. What Tolerance Is Really Required?
What you will learn in this chapter is that most tolerances have not been determined by application needs. Tolerances often have little to do with what is really required. Therefore, tolerances can and should be questioned. You will also learn the RSS (root sum-of-squares) approach to calculating stacked tolerances. Tolerances apply to the Improve and Control steps in DMAIC. This chapter is primarily for those involved in design, manufacturing, or process work.
The initial tolerance on a part is often based on the machine that the designer believes will make the part. The tightest tolerance the machine can achieve is then used. Or, the designer just copies the tolerance from another similar part. An analysis of the real need is seldom done.
Sometime later, the customer may have some issue with the part and complain. Then the tolerance is tightened—whether or not the issue involved the tolerance. Because most tolerances have evolved in this manner, there is large savings potential in determining what is really required in a tolerance.
Case Study: Required Tolerances on MoldsEight sets of interchangeable molds were used on several high-speed machines. These mold sets were very expensive ($100,000 per set), so they tended to be kept for many years. However, the customer complained that the "old" mold sets were making product with too much variation and wanted the old mold sets replaced. When an attempt was made to correlate product variation with each mold set, however, it was shown that the age of a mold set had no correlation with product variation. Instead, it was concluded that several mold sets were made with excessive variation from the start. Confidence tests, as covered in Chapter 12, confirmed this. The mold shape was very complex and the plant purchasing the molds was not capable of validating the mold dimensions. It had to use an outside firm with three-dimensional measuring capability to check the dimensions. When molds from each set were measured, it was found that all sets had some molds outside specifications. The problem mold sets just had more molds further out of tolerance than the other sets! Over the years, in response to complaints, the mold tolerances had been continually tightened until the supplier was not capable of making molds within specifications. However, since the plant kept buying molds, the supplier kept selling them. The mold tolerances became meaningless. Molds that were making acceptable product were studied and it was found that the supplier was capable of making molds that made acceptable product, even though many of these molds were outside the specified tolerances. The tolerances were then doubled to reflect the dimensions of the molds that made acceptable product, based on this real-need criterion. Inspection procedures were put in place to verify that all future molds were within the new specifications. Molds beyond the revised tolerances were taken out of production and excess product variation was eliminated. With the more realistic mold specifications, the plant was able to find alternate suppliers and the mold costs were reduced 50%. Total savings were in excess of $100,000 per year. |