A new technique for analyzing electrical resistance measuring device data is described. A 5 percent chromium, ½ percent molybdenum corrosion probe was used to measure corrosion rates in a refinery thermal cracking unit. By using the statistical technique of correlation by regression analysis, it was possible to determine which operational variables significantly affected corrosion rates and to calculate an equation to predict the corrosion rates in terms of these operating variables.
The linear regression equation which resulted from this analysis contained only the simple variable of sulfur content times velocity to predict corrosion rates. The corrosion rates calculated using this equation were found to compare favorably with the measured corrosion rates.
The statistical analysis of the corrosion rates measured by the corrosion probe resulted in: (1) Determining which operating variables significantly affected the corrosion rate, (2) Calculating an equation which can be used to predict corrosion rates under varying operating conditions.
Later, when made available, a standard program for the IBM 650 computer was used to perform the statistical calculations. The program used is called Stepwise Multiple Regression. It provides the advantages of processing a larger number of variables than with a hand calculator and eliminating an extensive knowledge of statistics. 2.1.1
