Combustion monitoring for biomass boilers using multivariate image analysis
|Abstract:||Biomass is increasingly used in the process industry, particularly in utility boilers, as a low cost source of renewable, carbon neutral energy. It is, however, a solid fuel with some degree of moisture which feed rate and heat of combustion is often highly variable and difficult to control. Indeed, the variable bark properties such as its carbon content or its moisture content have an influence on heat released. Moreover, the uncertain and unsteady bark flow rate increases the level of difficulty for predicting heat released. The traditional 3-element boiler control strategy normally used needs to be improved to make sure the resulting heat released remains as steady as possible, thus leading to a more widespread use biomass as a combustible. It has been shown in the past that the flame digital images can be used to estimate the heat released by combustion processes. Therefore, this work investigates the use of Multivariate Image Analysis (MIA) of biomass combustion images for early detection of combustion disturbances. Applied to a bark boiler operated by Irving Pulp & Paper Ltd, it was shown to provide good predictions, 2.5 minutes in advance, of variations in steam flow rate (R2fit=93.6%, R2val=70.1%) when information extracted from images were combined with relevant process data. This project is the first step in the development of a new automatic control scheme for biomass boilers, which would have the ability to take proactive control actions before such disturbances in the manipulated variable (i.e. bark flow and bark properties) could affect steam production and steam header pressure.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||16 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.