Refinery Operation Scheduling Considering Both Varying Feedstocks and Operating Conditions: an Industrial Data-based Modeling Method
Xiaoyong Gao,Nan Ru,Yongheng Jiang,Tao Chen,Dexian Huang,Xin Zuo
DOI: https://doi.org/10.1016/b978-0-444-64241-7.50211-1
2018-01-01
Abstract:Due to the varying feedstocks and operating conditions in refinery routine operations, the current process models (i.e. outlets yield prediction and operating cost model) for scheduling mismatch much from the industrial reality. How to obtain the more accurate and feasible models to describe refining processes remains an open problem and is the key challenge for successful industrial application use. In this paper, an industrial data-based modelling method considering both varying feedstocks and operating conditions of process units is presented. In details, for a certain unit, feedstocks composition rates and outlets properties are introduced to indicate the varying feedstocks and operating conditions respectively. For catalytic reformer and diesel hydrotreater, outlets yield and operating cost models about feedstocks composition rates and outlets properties (here, specific properties variables are selected for each process) are determined by data-based regression. Furthermore, the resulted process models are embedded into the scheduling model. A computational case originated from a real refinery in China is solved by BARON in GAMS 24.0.2 using a CPU Intel Xeon E5-2609 v2 @2.5 GHz with RAM 32.0GB. Result shows that, (1) the operation schedule of units are well adjusted dynamically with the demand trend of different product oils, (2) the resulted schedules increase the outlet yield and reduce the operating cost of process units and thus, improve the economic profit of refinery.