Speaking at the 2016 OSIsoft users conference, Tibor Komróczki showed how MOL is using the PI System in predictive refinery maintenance. MOL is aiming for ‘no unplanned downtime’ by integrating data from PI AF, PI EF and SAP PM via a new ‘integrator for business analytics.’ Use cases include pressure swing absorbers, chillers and heat exchangers. Data from the refinery’s Lims, PI Server, Opralog’s E-logbook and the refinery’s ‘Nice,’ information center are analyzed with a Python ML toolkit running in the Azure cloud.
The system is also used to optimize feedstock mixtures and to estimate diesel sulfur content. Lab data is compared with refinery performance and scored with predictive models. Microsoft’s cloud-based ML offering is now known as the Cortana Intelligence Suite.
© Oil IT Journal - all rights reserved.