Improved Boiler Performance using Targeted Sootblower Activation

A novel, targeted approach to sootblower activation has been demonstrated recently at a large, T-fired, coal unit. Targeted sootblowing techniques are designed to activate blowers based upon the conditions of the boiler rather than a traditional time and/or sequence based approach. Various approaches have been used in the past to perform targeted sootblowing including neural network based models, cleanliness based approaches, expert systems, and performance calculations. Recently, we have combined all of these approaches to provide a new technique for targeted sootblowing. At the core of this new approach is the use of neural network based models for predicting the changes in both the cleanliness factors and the heat duties of boiler heat transfer devices due to sootblower activation. To build such models, both the heat duty and cleanliness factor of each heat transfer surface in the boiler (i.e. division panels, final reheat, final superheat, etc.) are calculated using a series of performance calculations. Using the results of these calculations, a set of neural network based models are trained on historical data to predict the change in cleanliness factor and heat duty due to sootblower activation. The resulting neural network models are used in an optimization scheme to control sootblower activation throughout the boiler. In this optimization scheme, a set of candidate sootblowers that are eligible for activation is identified. Given the set of eligible sootblowers, an expert system is used in combination with the neural network models to identify and activate (if necessary) the sootblower with the greatest impact on boiler performance. This technique has been shown to improve boiler performance at the demonstration site (i.e. desired steam temperatures, improvement of boiler efficiency, etc.) while at the same time reducing overall sootblower activations.