Process control perspectives
A time-series graph can help you to understand the past, present and future of your system. A time-series graph is essentially any graph that has time, or a proxy for time, as the x-axis. No matter what your system or product, aspects of the performance will change over time as components or workers wear and tire. In this topic, we’ll explore control charts. In manufacturing, process and systems engineering, statistical bounds are placed on a variable of interest, and understanding any deviation outside of those statistical bounds can indicate that there’s a problem in your system.
Example applications
Consider that you are running a process that filled a bottle of expensive hand lotion. The advertised volume is 250ml, with a tolerance of ±5ml. This tolerance, due to the machinery, has been established over many thousands of runs and is acceptable to the client. The bottle is weighed before and after it is filled, and then recorded. When a sample comes back that is outside of the bounds (below 245ml or above 255ml), it is removed. If the behaviour is repeated, this is a sign that there is a problem with your process.
Steps
- establish the mean value. This is probably the desired value - in the example above, 250ml
-
determine the upper and lower bounds. This in probably an agreed tolerance, but could be determined statistically at first at a value that covers the majority of samples (e.g. 2 standard deviations from the mean)
- set up a repeatable and recordable experiment to measure the variable of interest (e.g. measuring weight as a proxy for volume)
Key concepts
- an explanation of statistical process control, and advice on setting up the upper and lower critical limits
- an example that walks through a relevant situation using statistical process control
- advice to the student engineer on how to interpret the results of a statistical process control analysis
Core resources
- Control charts take a statistical approach to looking for variation in a process over time - see Statistical
Similar tools…
Another similar approach could be to examine other time-series graphs. For example, the charging and discharging of a battery.