Application of Digital Twin and Machine Learning for Optimization of Wood Chips Combustion in Boilers

Summary

No reviewed summary yet. Read the abstract below or open the original publication.

Abstract

The article examines the shortcomings of existing control systems for boilers operating on wood chips. A key factor is the moisture content in wood chips, which significantly affects the combustion process efficiency, boiler performance, and the level of harmful emissions, particularly carbon monoxide. A detailed analysis of current methods for optimizing the combustion process in wood chip boilers has been carried out. It is noted that traditional control systems are often unable to account for real-time changes in fuel properties, limiting their effectiveness. The need for using digital twins and machine learning algorithms to predict changes in moisture content in wood chips and ensure optimal fuel-to-air ratios has been highlighted, as this allows for improved fuel efficiency and reduced emissions. The article proposes the development of a digital twin to optimize the combustion process in the boiler, based on cloud services integrated with existing boiler automation systems. It is noted that cloud services provide the advantages of scalable and flexible architecture for building the digital twin. A comparative analysis of Microsoft Azure and Amazon Web Services cloud platforms was conducted, with the most suitable solution selected for implementing the digital twin functions. The architecture of the digital twin has been developed, which includes the use of the OPC UA protocol to ensure reliable data transmission. The implementation of an automatic digital twin model definition based on input data is also proposed, simplifying the process of its creation. The results of moisture content analysis in wood chips at the Koriukivka power station are presented. Based on the findings, an additional system for regulating moisture content in wood chips via water spraying has been proposed. This system allows maintaining the moisture content at the target level within the recommended 20…30 % range, in cases of excessively dry fuel. To implement the system and ensure more accurate moisture measurements, as well as to address the issue of fuel heterogeneity, an additional sensor is installed above the fuel feed conveyor belt leading to the boiler hopper.

Abstract as supplied by the source. Full text may be restricted by the publisher.