Evaluation of electrically boosted natural gas fired glass furnace performance by using data reconciliation method

Abstract

The purpose of this study is to address the challenges posed by errors in sensor measurements and unmeasured variables in glass-melting furnaces, which can lead to misleading information regarding furnace performance. We implemented the Data Reconciliation Methodology to filter errors and estimate unmeasured variables, aiming to achieve accurate and reliable furnace characteristics. This task involved generating a dataset from measured furnace variables, and conducting observability and redundancy checks. By applying the data reconciliation method, gross errors were detected and removed, and the database was filtered for noise. Additionally, we estimated the necessary unmeasured variables. The results demonstrated the effectiveness of our approach. With accurate data, the energy efficiency, regenerator efficiency, and specific energy consumption of the furnace were found to be 38.63 %, 62.72 %, and 4159.84 $\frac{kJ}{kgglass}$ , respectively. The difference (before and after data reconciliation) between the raw and reconciled values of energy efficiency, regenerator efficiency, and specific energy consumption were around 0.09 %, 1.58 %, and 0.86 $\frac{kJ}{kgglass}$ , respectively. These findings underscore the importance of accurate data and the implementation of data reconciliation methods in the glass industry, providing valuable insights for improving furnace performance and energy efficiency.

Keywords: Data reconciliation; Energy efficiency; Glass furnaces; Gross error detection; Mass and energy balance; Parameter estimation.