Batch process is essential to today's drug production in pharmaceutical industries. Batch consistency and end-point timing lead to production quality and efficiency. In this paper, mid-infrared (FTIR) and calorimetry based principal component analysis (PCA) and dynamic principal component analysis (DPCA) were used for identification of batch reactions. We studied two types of triflation reactions and performed many batch processes including six production batches and acquired thousands of (FTIR) measurements directly from each batch. PCA were able to reduce this multi-dimensional data matrix into a few dimensional score plots for visualizing batch concentrations, end point and other batch variations with only limited calibration efforts. The prediction sensitivity was as low as < 1 PA% of starting R-OH residues. Combining process calorimetry data together with FTIR data, we were able to make PCA models more robust. This paper will also briefly discuss the risks and benefits of using PCA monitoring batch process operated in an inconsistent or a faulty mode.

Key word: Batch process, PCA, DPCA, MIR, Calorimetry