(125b) Using Digital Process Twin Technology to Drive Operational Excellence in Utility Systems

Digital replicas of operating assets that combine plant data with high-fidelity process models are bringing a new level of decision support to operations of steam, condensate, fuel gas and power systems. They provide many benefits, from monitoring and ‘soft-sensing’ to advice on optimal set-points, diagnostics and prognostics such as maintenance interventions. At the heart of such shadowing systems is an always-current digital simulation model of the assets which updates itself periodically based on real-plant performance.

In the context of utility systems such as steam, condensate and power, digital twin technology is now being implemented. The associated benefits include: equipment monitoring to determine the real state of equipment; real-time ‘soft-sensing’ to provide up-to-date performance information which is either difficult or impossible to measure; forecasting to determine future performance such as maintenance intervals based on current equipment state and anticipated operation; operational optimisation to give advice to operators regarding set-points, diagnostics, prognostics and performance benchmarks; and finally ‘what-if’ analysis to anticipate how to operate for future/alternative operating scenarios.

Successful digital operating systems use high-fidelity predictive mathematical models of the assets to exploit redundancy between model prediction and plant data and maintain themselves in an always-current state. This means that ‘drift’ in key parameters such as gas turbine efficiency, steam pressure levels, equipment fouling and so on is taken into account in all monitoring, forecasting and optimisation calculations, to provide reliable information for decision support.

This paper describes how the combination of equation-based general-purpose modelling technologies and next-generation digital application frameworks provide an environment for easy construction and application of fast, robust online solutions based on high-fidelity utility system models. A real industrial example is presented where the digital twin of a steam system is used to improve the system’s performance. The twin is linked to plant data systems, updating itself through machine-learning capabilities, validating actual performance and, where appropriate, identifying departures from normal operation. Beneficial operating changes are highlighted. Visualisation of the twin is seen to be critically important to give operators greater insight and confidence to operate the process safely at the optimum point, thereby making a major contribution towards Operational Excellence.