Sun, W., Beijing University of Chemical Technology
Chen, T., Beijing Key Lab of Membrane Science and Technology, Beijing University of Chemical Technology
Wang, J., Center for Process Simulation and Optimization, Beijing University of Chemical Technology
Heat exchanger is usually arranged between the feed and effluent of a continuous catalytic reforming unit to recover the energy from reaction product and to preheat the reactant stream. Platforming exchanger is well known in refinery industry for its well-designed structure layout to achieve high energy efficiency, but it also vulnerable to solid material carried in the stream, including catalyst dust, carboid residue, and ammonium salts, which could lead to the increase of flow resistance, i.e. pressure drop. As the effluent pressure is controlled to a pre-set value in practical operation, pressure drop between inlet and effluent will affect the heat exchanging efficiency and threat the equipment safety/service life. The increase of pressure drop doesnât happen suddenly, instead, it is built up by relatively long term improper operation or equipment failure, which will certainly be captured by historical data, i.e., it can be traced back if process information is properly extracted.
In this work, an industrial case of platforming exchanger in a continuous catalytic reforming (CCR) unit is investigated. The reactor effluent pressure drop of platforming exchanger slowly rose beyond the normal range after increasing production load, which is difficult to find the root cause of from the human operator's experience. Historical data from the CCR unit are collected for analysis. Through machine learning method, the relevant variables affecting reactor effluent pressure drop of platforming exchanger are selected to establish the prediction model, by which the root cause of abnormal changes in reactor effluent pressure drop are identified. The results show that the prediction model can alert the abnormal changes of reactor effluent pressure drop timely and accurately, and the root cause of abnormal changes is also supported by operation record.
