"An investigation was undertaken to statistically analyze the data obtained on X-40 (Stellite 31) production blends which were tested to the requirements of AMS 5382. For our purpose a blend is one or more furnace heats, the shot of which has been thoroughly mixed to produce greater uniformity of product. The mixed shot is charged to the arc furnaces and cast into parts, and test specimens. A great range of test results have been obtained in testing blends, and it was felt that a statistical analysis of a few of the variables might yield information which could be applied toward improving the general quality and performance of this alloy, particularly with respect to stress rupture time.
While blended shot produces more uniformity of product than unblended shot, the spread of values obtained, particularly in stress rupture testing, is still considerable. Factors such as variation in composition, grain size and carbide spacing, were felt to be likely quantities which could possibly be found to influence the properties of Stellite 31, if the masking effects of the many variables present could be reduced or eliminated. To do this by a test program would be nearly impossible and prohibitive in cost. Statistical analysis would accomplish the purpose of indicating trends and tell us which basic factors should be concentrated upon, and perhaps permit a more scientific approach after the variables of little or no significance were eliminated.
Some of these complex factors involve variations in processing, and some involve an almost infinite number of minor variations in relative chemical composition. In fact, no two production blends have been observed to date which permit direct comparison of one blend to another. In any alloy containing some ten chemical elements phase relationships may become very sensitive to slight composition shifts of any single element. Phase stability, of course, often reflects on the properties obtained.
Since statistics in metallurgy is a tool and not an end in itself the statistical methods used are necessarily of a rudimentary nature. In this work a combination of arithmetic and graphical methods has been used, the arithmetic mean and value of the standard deviation being determined mathematically for the purpose of eliminating guess work in determining the slope and location of the distribution line. The graphical method has been used to plot the test data, confidence limits as calculated, and to illustrate the variations in accordance with the method described by L.R. Hill and P.L. Schmidt and by P.L. Schmidt"--Introduction, pages 1-2.
Materials Science and Engineering
M.S. in Metallurgical Engineering
Missouri School of Mines and Metallurgy
ii, 86 pages
© 1957 Edward P. Patterson, All rights reserved.
Thesis - Open Access
Cobalt alloys -- Analysis
Heat resistant materials -- Testing
Corrosion and anti-corrosives
Print OCLC #
Electronic OCLC #
Link to Catalog Record
Patterson, Edward P., "Statistical analysis of X-40 (Stellite-31) production blends." (1957). Masters Theses. 2561.