Online Flooding Prognosis in Packed Columns by Monitoring Parameter Change in EGARCH Model

In the chemical industry, packed columns are commonly used operating units for separation. However, the flooding phenomenon often reduces the efficiency of packed columns and interferes with the performance of the system. Due to this reason, research on the real-time prognosis of flooding becomes a necessity in practice. Pressure drop is a key factor that indicates flooding phenomenon in packed columns. In this paper, the trajectory of pressure drop in each time window is modeled with an exponential generalized autoregressive conditional heteroskedastic (EGARCH) process. The onset of flooding is then implied by the parameter change of the model. To capture the change in an efficient manner, a nonparametric charting technique is adopted for statistical process control (SPC). The feasibility and efficiency of the proposed method are illustrated by the experimental results.