Development of a supervisory control for a kraft pulp mill steam and power system using PCA and PLS
University of New Brunswick
Steam and power systems in kraft pulp mills can be complex and difficult to optimize. Supervisory controls can facilitate the optimization if the appropriate weightings and time delays are applied. The focus of this thesis is to: create models of the steam and power systems of a kraft pulp mill, to evaluate the models, and to develop a control system based on those models. A system involving a kraft recovery boiler, two turbo generators, a batch cooking plant and multiple process users is examined. Partial Least Squares (PLS) and Principal Component Analysis (PCA) are used to develop models that are utilized to develop factors that will be employed in the supervisory controls. Mass and energy balances are completed to validate measurements. A thorough understanding of the system, including the delays associated with changes to the variables, is required to ensure that the supervisory controls improve the responses. As a result, these objectives were satisfied resulting in an increase of 2.2 MWh (21%) of hog fueled power produced, no interruption of process steam users, and statistically significant reductions in steam venting.