Preliminary Dependence Analysis of Process Variables: A Case Study of a Bubbling Fluidized Bed Boiler

Data from a bubbling fluidized bed boiler was tentatively analyzed in order to find dependence among process variables. The data used was collected during normal operation of the power plant. Data was analyzed by using correlation analysis and the self-organizing map (SOM), which is a useful tool for visualization. SOM component planes and the U-matrix were utilized for visualization. Both correlation and SOM analysis showed similar results. For example, pressure differences across different parts of the flue gas duct had a practically linear correlation, flue gas temperature in stack requires further analysis because of surprisingly low correlation with other measurement points and flue gas SOx and NOx contents did not correlate. The visual presentation of SOM supported the results obtained from the correlation analysis. With the used methods, dependence among variables can be evaluated in a computationally inexpensive way and recognized both quantitatively and qualitatively.