(79b) Digital Twin Distillation Operations Training

The safe and on-specification control of distillation columns is one of the challenging aspects of operating a modern-day chemical production facility. In comparison to the previous decades, today many distillation columns are operated at tighter product specifications and at higher product and process recovery rates. And as we transition to more renewables, operation during increasing variability in energy inputs due to demand response and supply upsets is increasingly likely. As a result, even experienced process operators and engineers need to think about the complex interactions between thermodynamics, internal hydraulics, mass and energy balance limitations and controllers when making changes. Considering these shifts and the inevitable loss of experienced operators and engineers there is a need for continuing education and competence building to be carried out in the domain of distillation process control.

Traditional continuing education and operator trainers even when built from dynamic process simulators are often based on local dynamic models based on simple linear but abstract transfer function models or resistance capacitance network models. These local models are useful for simulation of relatively mild process upsets or more carefully curated upsets such as start-up and shutdown that are pre-programmed. We propose and advocate the use of dynamic digital twins based on a combination of mechanistic and data driven modelling for a wider range of operational upsets, e.g., such as recovery from significant and rapid power turndown for demand response. In this presentation we describe the key concepts and advantages of our real-time approach to operations training simulation over the aforementioned traditional approaches with the use of a validated dynamic digital model of a real methanol refining column.