Lisa M. Gill
Said Jahanmir, Lewis Ives
In 1992, NIST initiated a research program on ceramic machining and grinding as a consequence of a comprehensive survey of the U.S. ceramics industry, which confirmed that the high cost of machining is a primary impediment to the widespread use of advanced ceramics. NIST, wanting to assure that industry needs were properly addressed and to take advantage of expertise existing at other research institutions, established the Ceramic Machining Consortium with members from industry, academia, and government. The purpose of the Consortium is to provide measurement methods, data, and models for the assessment of these new, cost-effective, and innovative ceramics.
The Statistical Engineering Division (SED) has played an integral role in the Consortium since its inception. A major project of the Consortium is the study of fundamental mechanisms involved in the material removal process that occurs during grinding. The study focuses on the influence of grinding parameters, such as down feed, wheel grit size, grinding fluid and the role of microstructure and properties. SED assists the Consortium in designing new experiments, analyzing data, and consulting on issues as they arise.
An example of assistance is a discussion that SED led at the Consortium's bi-annual review meeting. A new suite of experiments was proposed by SED to characterize the effect of grinding parameters on a new material. To determine the effects of wheel grit, down feed, wheel speed, cross force, grit size and trueing and dressing, SED provided a 25-1 screening experiment, with center-points, design run in 3 replicates with trueing and dressing between replicates.
However, before the above experiment could be run, another interesting problem had to be resolved. A preliminary experiment had been conducted to assess the repeatability and reproducibility under one particular grinding condition. Six labs participated and each lab ran 3 tests (call them A, B, C). 10 bars were ground in each of the three tests. For the data from 4 of the 6 labs, graphs displayed an inherent pattern of A-high, B-low, C-high. SED needed to determine whether this pattern was ``real'' or just an artifact of the data. That is, while visually there appeared to be a pattern, quantitative tools were needed to determine whether there was a test (``A,B,C'') effect.
Using a probability model, SED determined that here were six possible patterns and the probability that four or more labs would exhibit a similar pattern was not statistically significant. With this conclusion, the Consortium could continue with their current schedule of experiments and not let this "confusing" pattern prevent further research. The experimental procedures and results that SED produces for the Consortium will be used to develop ASTM standard methodologies for the machining of advanced ceramics.
Figure 18: Surface grinding of silicon nitride with a resin-bond diamond wheel.
Figure 19: Plot of tests A, B, and C for 6 labs with a 7th confirmatory run. Each test consists of 10 ceramics bars. The mean and associated 95% confidence interval for each test are plotted.
Date created: 7/20/2001