Statistical Process Controls for Manufacturing

Our client had several requirements:

• they wanted better oversight and control of their manufacturing processes to reduce non-conformance and waste; and

• they wanted to take a proactive approach to managing and monitoring their processes so that root causes of a system upset could be found early, and  corrective action could be taken early.

Manufacturing occurred at 12 different plants around the country, each with its own set of processes and operating conditions.

 

Our strategy was to implement a multivariate statistical process control system to diagnose and isolate the source/cause of a non-conforming batch. This approach serves several purposes:

1. it gives an understanding of the current capability of the manufacturing process at each plant;

2. it gives an understanding/measure of how far from industry specification the current system capability is; and

3. it provides monitoring and feedback on the process -- engineers can see when the process is changing, or can see how the process is changing when alterations are deliberately made, which is useful for maintaining current capability or manipulating the process into a new state (which is in line with industry specification).

 

We developed a desktop app in R and Python whose pipeline automated the cleaning, curation, analysis, and management of our client's manufacturing data, and provided a suite of quantitative and visual diagnostics of the manufacturing process. The data management system was designed to facilitate further analysis for determining additional factors that influence changes in the process and output, and to implement effective service and maintenance strategies.

​

If you don't have a statistical understanding of your processes and operations, then you don't know what is "normal" or "expected". Take corrective action with complete information. Come and talk to us if you're interested in statistical process controls for your operations and quality control.