Process capability study in Chemical industry

Welcome to QFO!

Can you validate with test samples? Not for every batch, but as a part of design and process validation?

Industry experts I have read don't recommend the usual SPC methods for chemical manufacturing.

• Members of the University of Newcastle Department of Chemical and Process Engineering published the article Multivariate Statistical Process Control in Batch Process Monitoring in Science Direct. This article is long and looks thorough.
• A University of Manchester thesis paper titled Use of Multivariate Statistical Methods for Control of Chemical Batch Processes is also detailed and thorough. See page 42 for the section titled Overview of Classical Methods for Batch Process Control.
• The World Batch Forum's paper Statistical Process Monitoring of Industrial Batch Processes is shorter and easier for non-PhD candidates (like me ) to read.

I hope this helps!

 

Things to consider: SPC stands for Statistical Process Control. Most chemical processes are already controlled by electronic process controllers. Adding SPC would be redundant and non-value added. The electronic process controllers would typically create false alarms (e.g., runs, oscillations, etc.) on a control chart. In addition, chemical processes are typically autocorrelated, which creates major problems for subgrouped control charts.

 

SPC might help you in this case. you may have to think about what that means - you will probably need a "3D chart" which is designed for batch processes as well as strongly non-homogenous processes where you need to monitor more than 2 components of variation a free article for this is: The Three-Way Chart by Donald Wheeler.

You also will be able to perform a useful MSA. The variation has nothing to do with the usefulness of an MSA. Try reading the second example in my paper "MSA for for Functional Tests" in the "practical quality engineering resources" in the resources tab.

 

See my blog on R&R for Non-replicable Measurements, in addition to Bev's resources.

 

Student* did. The thing with t-tests is that they will detect non-homogeneity as a difference in means. Batch processes are notoriously non-homogenous. A homogenous process is one where the factors that determine the within subgroup (in your case within batch) standard deviation are the same factors that determine the average of each batch. You will likely determine that every batch has a statistically significantly different mean than the previous batch. That will not mean that the batches are “inconsistent”. You MUST understand how the math works and what the assumptions of these tests are.

A better approach is to use the 3D or 3 way SPC chart AND determine how much difference between batches can be tolerated. A t-test cannot tell you that.

*Student was the name that William Gosset used when he published his work that developed the t-test. Gosset used a nom de plume to keep his true employer secret and maintain a competitive advantage. Gosset was the chief brewer at Guinness. Happy St. Patrick’s day.