Use of multivariate methods and DOE to improve industrial‐scale production quality of a cellulose derivative

The manufacturing process of a cellulose derivative, ethyl-hydroxy-ethyl-cellulose (EHEC), was improved substantially regarding the amount of (water-)insoluble matter (gel particles) in the final product. The existing process was improved through multivariate follow-up of one of the full-scale reactors and designed experiment investigations, DOE (design of experiments), in a pilot plant reactor. The multivariate follow-up of the full-scale reactor was evaluated through PLS modelling and resulted in models with low predictability. These models did however turn out to be very useful, as they gave guidelines regarding the difference in the process parameter settings when high-gel batches were produced compared to those with a lower gel level. Designed pilot plant investigations were made and evaluated with the use of MLR. The investigations showed that two of the examined process parameters had a significant influence on the gel level of the manufactured EHEC. These parameters were changed in the full-scale process in accordance with the findings in the pilot plant. By utilizing the above-mentioned methods, it was possible to obtain process improvements, resulting in an EHEC product with a substantially lower content of gel particles. Copyright © 2000 John Wiley & Sons, Ltd.