Early Identification of Process Deviation Based on the Spatial Correlation of Measurements
Ma, Fangyuan
Han, Xianyao
Lin, Dexi
Zhong, Jingtao
Wang, Jingde
Sun, Wei
Download PDF

How to Cite

Ma F., Han X., Lin D., Zhong J., Wang J., Sun W., 2019, Early Identification of Process Deviation Based on the Spatial Correlation of Measurements, Chemical Engineering Transactions, 74, 607-612.
Download PDF


Almost all equipment in chemical process are three-dimensional, where mass and heat transfer are conducted, sometimes reaction as well. The corresponding progress can be reflected by the measurements at different locations in a unit, such as composition, temperature and pressure. In a vertical fixed bed reactor, the inlet flow goes from top to bottom and reacts in the reactor. To ignore the difference at the horizontal direction at the same bed height, the measurements at vertical direction will be the only index of reaction progress. If these measurements are available, the reaction can be well monitored.
It is critical to make sure the reaction operate according to the design, i.e., the normal operating condition. The composition of key component is the direct measurement of it, but it is hard to obtain in real time, instead, the acquisition of pressure and temperature measurements is much easier. Based on the understanding from chemical engineering point of view, they cannot tell the reaction progress directly, but they are the consequence of it. As the reaction gets further along the vertical direction of the unit, the temperature also changes accordingly. The spatial correlation among temperature measurements can be extracted when the process is under normal operational condition. Once this spatial correlation changes, it indicates that the process has deviated from the normal operating state. Therefore, early identification of process deviation can be achieved by monitoring the easily-obtained variable spatial correlation, such as temperature, if a proper data analysis method is employed.
In this work, a pre-reforming reactor of hydrogen production units is investigated. The spatial correlation among temperature data is extracted by using a multivariate statistical method. The results show that the process deviation can be detected10 hours ahead of human operator.
Download PDF