Technique for reducing erosion in large-scale circulating fluidized bed units

This paper presents a methodology, implemented for a real industrial-scale circulating fluidized bed boiler, to mitigate the risk of heating surfaces exposed to an intensive particle erosion process. For this purpose, a machine learning algorithm was developed to support the boiler reliability management process. Having a tool that can help mitigate the risk of uncontrolled power unit failure without expensive and technically complex modernization is desired. A virtual procedure can be seen as a milestone towards the application of digital models to the diagnostic procedure of large power units, providing answers for many scenarios that cannot be normally studied during boiler operation. The predictive model developed in this work allows us to provide the requested feedback to the unit control systems regarding possible changes in boiler operating conditionsand reduce the erosion effect. The functionality of the discussed methodology is investigated via application of the developed multiphase computational model.

Automated summary

This paper presents a methodology to mitigate the risk of heating surfaces exposed to particle erosion in a circulating fluidized bed boiler. A machine learning algorithm is developed to support boiler reliability management. The virtual procedure using digital models provides answers for scenarios that cannot be studied during boiler operation. The predictive model developed in this work provides feedback to the control systems regarding changes in operating conditions and reduces erosion effect.