2015 - Users Conference - San Francisco - Pulp and Paper
Using the PI System to Support Prediction and Advanced Control of Reel to Reel Paper Quality
PI Server data is used to support Multivariate (MV) methods for characterizing process variations and monitoring statistical process control (SPC). These modeling methods can also be used for predictive purposes to estimate future trajectories of important process parameters or final conditions of batch type processes. Control of batch type systems is often performed at two levels consisting of maintaining important process parameters along a desired trajectory and a higher level of supervisory control to determine the best trajectory for these variables in order to optimize the final batch quality. Paper machines are run as a continuous process yet quality is determined for each reel. Considering each reel as a batch, batch modeling methods can be applied. In this session a hierarchical approach using latent variable methods (PCA/PLS) is presented to provide these two levels of control and applied to a continuously operated paper machine. The opportunity for using contemporary PI Clients to parse data into event frames will be discussed.
Tim Michaelson is a Senior ME Consultant at International Paper.
Chris McCready is a chemical engineer by training with degrees from University of Waterloo and McMaster University. He has worked in an advanced automation with Petro-Canada setting up and maintaining advanced control strategies, and at UTC Power developing mechanistic models of fuel cell systems for systems level dynamic analysis. Currently Chris is a Principal Engineer with Umetrics, where he is involved with many QbD activities including DoE, implementation of multivariate monitoring and development of advanced control strategies.