400678 Dynamic Real Time Optimization of Oil Refining Processes

Wednesday, April 29, 2015: 10:15 AM
13B (Austin Convention Center)
Stig Strand, R&D Process Control, Statoil, Trondheim, Norway and Arne Aase, Refining, Statoil, Mongstad, Norway

Statoil has implemented dynamic real-time optimization (DRTO) of several oil refining process units during the last decade. The DRTO applications are in fact model predictive control (MPC) applications, utilizing the features of the in-house MPC software SEPTIC (Statoil Estimation and Prediction Tool for Identification and Control). The basic idea is to model what is needed to solve the problem, and nothing more. Typically, the dynamic models are hybrids of fundamental and experimental parts. The fundamental parts of the hybrid models may come from conservation laws, and quality propagation and mixing of process streams, in a time-varying process structure. While the experimental parts of the hybrid models may capture process responses that are sufficiently isolated and represented by traditional MPC types of models.

Experiences will be shared from RCC, CDU and product blending processes. The applications execute at least once a minute and the models may be updated in dynamic transients. This means that disturbances with economic impact, and a relatively high frequency compared to the process dynamics, may be considered and brought into a new optimization in a proper and timely manner. The panel operators are familiar to the MPC technology through several lower control layer applications, and have the responsibility of running the DRTO applications as well. So far, no convergence nor maintenance problems have been observed, as have been reported for more traditional, highly rigorous, steady-state type of real-time optimization applications.


Extended Abstract: File Uploaded