As already mentioned in my post https://r1132100503382-eu1-3dswym.3dexperience.3ds.com/#community:26/post:lk-NdD8lS7KjzpieKN91eQ
there are several levels of analytics in general. But even with the simplest process intelligence, the Statistical Process Control (SPC), we are asked by many customers on the best practical approach to improve a process. Herewith I would like to give some food for thought.
The success of an improvement on a given process is not based upon the successful implementation of the selected solutions, but instead when the process measurements (KPIs) have improved and this has been validated with appropriate statistical techniques (i.e. graphs, hypothesis testing, etc.).
Statistical Process Control (SPC) charts are essentially a sophisticated form of Time Series plot that enable the stability of the process and the type of variation involved to be understood.
An example of an SPC chart is shown below. It plots the performance of a process over time and shows control limits (not specification limits) which the results will fall between if the process is stable and “in control”.
Deploying SPC charts will not mean that processes will suddenly become “in-control”. What they can do however, is help you to measure and understand processes, and provide a rigorous approach for deciding when a process has changed and/or needs intervention.
What do SPC charts detect?...... Changes!
- Changes in process average
- Changes in process variation
- On off events such as special causes
There are many types of changes that can occur in a process, but they can usually be characterized as on of (or a combination of) the following three types of change:
A change in process average: The time series chart shows a clear increase in the process average. Whilst this example is visually very obvious, SPC charts can detect much smaller changes that wouldn’t normally be obvious to the human eye.
This might represent an uncontrolled change in the process or be the result of a deliberate improvement to the process.
A change in process variation: The time series chart shows a clear increase in the amount of variation in the process (note that the average doesn’t necessarily change). Again, this example shows a very marked change, but SPC charts can detect much smaller changes in variation that wouldn’t normally be obvious.
One off events (Special Causes): The time series plot to the right appreas to be relatively stable (and “in-control”), with the exception of the two points that are significantly higher than the rest. These two points are known as special causes because they fall outside of the expected variation range of the process, and are therefore likely to be as a result of a specific “special cause”.
SPC charts help to detect theses special causes, which can then investigated to identify the root cause.
So, in order to improve your process, on what should you work on? Improve the average, reduce the variance, avoid special causes or all together in one?
The theory as well as the practice show, to improve your process you shall take sequenced improvements actions to
- Reduce your variance to get a more stable and repeatable process
- Improve your process average for better performance
- Eliminate opportunities to avoid special causes
You can see, with some simple usage of Manufacturing Process Intelligence, you will be able to tackle a big part of your process issues in a simple way. It is not difficult to understand and with the basic training of the above to your production team, will improve your processes significantly.
