Tech Directions - March 2009 - (Page 20) is run to one side of the tolerance to compensate for die wear. As the tool wears, the dimension will change with it. The second set of samples, random die cast parts, represented in Fig. 3, shows a bimodal process. Each of the four cavities may differ in size due to pressures, tool size, but also interpreting it as an engineer working in industry would. Benefits and Cautions Students leave this activity with the ability to make informed decisions about a process and its ability to make products. Challenging students to evaluate their own process gives them the opportunity to make changes and learn with continual improvement. For students to perform this activity correctly, several steps need to take place to ensure that various forms of error do not bias results. Process capability “results from sampling of a continuous process over planned intervals” (QC Inspection Services, 2008). The students in this activity are performing tasks that a quality engineer would Fig. 3—Process capability analysis of 100 die cast parts from a four-cavity mold cooling time, and/or mold temperature. As a result, students are not able to use the process capability to determine process robustness. Samples from each cavity need to be reviewed individually. Students are encouraged to segregate samples by cavity identification and revaluate the capability. The third group of samples, machined aluminum parts, shown in Fig. 4, yielded a high process capability of 2.27. The process is robust and could be released for production. With a high process capability, the number of parts per million was set at zero. The following answers were developed as a result of the data analysis of the example samples: • Are the processes capable?—Samples from groups 1 and 2 were not capable. Once the samples in group 2 are sorted by cavity and evaluated, they may be found acceptable. • Is the process shifted to one side of the tolerance? If so, why?—In these samples, none of the charts show a process that runs towards one side. If they did, typically they are driven to one side in anticipation of tooling wear. • Is more information needed to determine process capability?—In some cases, more data or details are required. In sample group 2, the students needed to evaluate parts from one cavity only. • Does the process always need a high process capability to be capable?—A process with a high process capability index may be accepted. Sometimes, it makes sense to offset the sample mean from the center for assembly or design reasons. At the end of the activity, students write a report that summarizes their answers to the questions. This gives them experience with not only collecting important data Fig. 4—Process capability analysis of 100 aluminum parts perform. If process capability were being performed to monitor a production process, the engineer would establish routine sampling times. For example, five samples of a particular diameter might be checked every two hours. In this activity, the samples have been preselected and numbered 1 through 100. If an interruption occurs, there is increased risk that you’ve introduced a special cause into your capability study (QC Inspection Services, 2008). Any form of interruption, such as a process change, lunch breaks, or change in operators and/or inspectors can attribute to variation. Even the distraction of casual conversation can result in a slight change and variation. As a result, students should work diligently in class to finish measuring samples from each population. Measurement error can also greatly affect the results in determining capability. Inspection devices with inadequate resolution and repeatability can bias variable data and your end result before process variation is even introduced. If more than 25% of the specification tolerance is consumed by gauging error, the instrument is unacceptable. In this situation, you will most likely show high variation and a low capability as a result of moving data points. If necessary, a calibration lab or inspection equip- 20 techdirections ◆ MARCH 2009
For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.