Development and implementation of on-line multivariate quality control was discussed at the 2005 AIChE meeting (Chiang and Colegrove, 2005) and long-term performance, model maintenance, and model leveraging was discussed at the 2006 AIChE meeting (Chiang and Colegrove, 2006)

To further satisfy customer application requirements a new analyzer was installed to monitor molecular weights for two key products. Representative samples from 7 products were collected for both molecular weight and product quality variable measurements. A partial least squares (PLS) model was built to predict product quality based on molecule weights. The PLS model provided accurate prediction of product quality for most time periods. Model discrepancy was observed for some time periods due to analyzer calibration problems. The new analyzer provided additional product quality information, but it was not feasible to build a PLS model to predict molecular weights based on product quality variables.

The project confirms that both multivariate quality control and molecular weights are needed for an effective product quality management program for these two key products.

Reference:

L. Chiang and L. Colegrove, Industrial implementation of on-line multivariate quality control, AIChE Annual Meeting, Cincinnati, OH, 2005. Paper 320a.

L. Chiang and L. Colegrove, Industrial implementation of on-line multivariate quality control, AIChE Annual Meeting, San Francisco, 2006. Paper 92d.