Advances in automation and software have increased the ability for Statistical Tolerance Analysis to be used in mechanical engineering. Performing a statistical tolerance analysis is easy and more desirable for increasing the quality and lowering the costs of manufacturing complex mechanisms.
Prior to technology improvements, single dimension or 1D tolerance stack-ups were used. A thorough engineer would perform multiple 1D stacks in a variety of directions to attempt to predict effect of tolerance value changes on overall product quality. By using statistics, engineers can understand how to optimize tolerances, especially in complex mechanisms. These optimizations can create more cost-effective, on-schedule products and reduce scrap, rework, and delay. With statistical tolerance analysis, manufacturers can better understand design intent and know how changing tolerances during production can impact product quality.
In order to balance assembly performance against manufacturing costs today’s engineers use optimization software to identify dimensions that do not affect performance. Tolerances can be traded off on critical surfaces. This tolerance optimization improves product quality without increasing costs. It also streamlines development cycles by requiring fewer physical prototypes.
For companies that make millions of a single assembly per year, such as a keyboard supplier to Dell or HP, having a statistical analysis can predict the number of defects per million and easily measure against the desired sigma value. For companies that make only a few assemblies per year, such as medical implants, the probability of a single failure can be computed.
Statistical tolerance analysis can predict the cost of warrantee, allow some manufacturers to manage warrantee programs and balance the marketing benefit of different warrantees compared with the predicted cost of warrantee work due to quality levels during manufacturing.
Similarly, liability costs associated with product quality can be reduced. In those industries where the cost of failure is high, such as defense and medical devices, statistical tolerances are used to determine where to tighten tolerances for an increase in quality. It is no wonder that organizations that build nuclear devices are heavy users of statistical tolerance analysis.
Because a basic tolerance stack-up does not account for all variations that can occur during the manufacturing process, and because randomized tolerance analysis, such as that used by Monte Carlo, are not precise enough to effectively predict outcomes in manufacturing, tolerance analysis needs to be performed as a statistical tolerance analysis.
Companies both large and small are using statistical tolerance analysis to increase the robustness of their assemblies. Companies knows for quality such as Raytheon, GE, Volvo, are all good examples of companies that have incorporated these more advanced methods in their overall quality programs.