- Hybrid machine learning and process simulation models have been used to reduce emissions on a crude distillation unit at a European oil refinery.
- Two years of historical operating data were simulated and optimised using a rigorous first principles model.
- The optimisation results were used as synthetic data to train machine learning models.
- The machine learning models were used, on-line in real-time, to set the targets for the multivariable predictive controller.
- The result was a reduction in emissions through reduced fuel gas and steam use and an increase in the unit gross margin.
Simon Rogers
Vice President, Digital Solutions at Yokogawa
Helping the Energy and Chemical industries to address sustainability and the energy transition by using data driven models to optimize and automate value chains and operating assets. Experience of planning, scheduling, simulation, real-time optimization, advanced control and automation with cloud, big data, AI and IIoT to help transform operations making them safer, more sustainable, more reliable and more profitable. Enabling safe and sustainable autonomous operations with a combination of first principle models, machine learning and semantic web technologies to automate and integrate business processes.