From Crude Oil to Steam Cracker: How Petrochemical Naphtha Quality Drives the Simultaneous Refinery and Petrochemicals Optimization

MOL Group is an integrated, international oil and gas company headquartered in Budapest, Hungary, with lead position in its core markets within Central and Eastern Europe. Its northern region downstream business consists of two complex refineries and three steam crackers at two locations. To be able to exploit synergies thus to maximize potential profitability of assets via transfer connections, current downstream planning activities are performed on group level in a multi-site linear planning model resulting in simultaneous optimization possibility of petrochemical and refinery sites. In such a planning model system, next to internal asset constraints, specific utility costs and market purchase or sales conditions drive the optimization result. It does not only support optimal feedstock selection for both refinery (different crude oils provide different yield structure) and petrochemicals (LPG against naphtha feedstock) together with their most profitable asset configuration, but it also enables proper setting of refinery product portfolio against polimer production. In current MOL Group solution, an additional aspect has been implemented for improving the refinery-petrochemical connection: next to its volumetric effect, specific quality of produced petrochemical naphtha feedstock is also considered within the optimization as this information is channeled into the olefin plants in order to better estimate monomer yields of the steam cracker. It provided another milestone of this simultaneous optimization challenge thus improved on its delivered results. The paper focuses on our current best practices by demonstrating how above planning system features support MOL Group Downstream to derive right business decisions. It furthermore explains how our development roadmap puts emphasis on further refinement of the refinery-petrochemicals connection, utilizing state-of-the-art technologies like upgrading our linear optimization models with non-linear correlations.