Digital Twinning for condition monitoring of Marine Propulsion Assets

Abstract

Digital twinning approaches for high-value propulsion system assets is a growing trend, currently seen in renewables, marine and aviation markets. The benefits of this approach, involving "living-learning" models for diagnostic, prognostic and system optimizations are suggested in literature as many. When deploying such an approach, there are many discussions and decisions around modelling techniques to deploy, and the required fidelity of such models. It can be assumed though that the greater the potential to gain real measurement data, the greater the opportunity to improve the overall system model accuracy.This paper develops a model of a previously overlooked but essential part of the engine control system - the measuring chain relating to the closed-loop engine combustion controller. This part of the system performs an essential role to provide real-time data for the engine control loop, to be able to optimize the engine performance on a cylinder-by-cylinder, cycle-by-cycle basis. However, failure of this part of the system can often be impossible to distinguish from an engine fault when there is no knowledge of the system health of the measuring chain. Therefore, performance monitoring of this sub-system is of high value to the end-user, to fully optimize and potentially decarbonize their engine system, in combination with digital twin methods.