Emissions From a Sintering Plant Using Statistical Models

摘要

Statistical models based on linear regressions, factor analysis and analysis of variance were used to predict the concentrations of total particulate matter, sulfur dioxide and nitrogen oxides from a sintering plant chimney. A data bank with more than 2,000,000 values of production factors, raw materials and operating variables was used in conjunction with minute based emissions monitoring values to extract relationships among concentrations and basic measurable variables. Stepwise regressions obtained with one-year history show R squared approximately 0.60, good residuals behavior and explanation factors rotated by using varimax method in good agreement with the results expected by master operators. The tests performance demonstrated the usefulness of the method which allows the prevision of emissions and the necessary selection of raw materials, operating procedures and set points in order to avoid high concentrations in the chimney exhaust gases.