Batch on-line analytics for every user

Process Analytical Technologies (PAT), in particular Principal Component Analysis (PCA) for fault detection and Projection to Latent Structures (PLS) for end of batch quality prediction, are seen as pivotal techniques for improving process operation. A number of software packages are available today for off-line analysis. The difficulty for control or production engineers is that these tools are not designed for on-line operation. Many PAT issues may be addressed by tightly integrating analytic tools with the production and control system. This special session examines basic design requirements associated with batch analytics applied for on-line operation. The presentation delivers an in-depth look at the data processing requirements, calculations and limitations for the application of on-line analytics to a batch process. It will show how to achieve proper data alignment for different batches, a key requirement for building good statistical models, and how to use the model for on-line analysis. A typical batch operation in the chemical industry and a running simulation will be used to illustrate the advantages of on-line process analytics over traditional monitoring and control techniques. Details will be presented on the approach taken to integrate analytic tools and results into a commercial control system with examples of control and production operation screens.