How to analyse a control chart?
When you create a control chart, you have to analyse it :
- in real time to validate a series of measures,
- a posteriori, to study possible trends or problems.
Attention, this interpretation is only valid if the consecutive control results are obtained in reproductibility conditions.
(1) A oscillation in a random way around the target value
The results curve oscillates in a random way around the target value and within the control limit. Our process is under control.Your results are reliable and accurate.
Your process was under control and now you notice...
(2) A upper or lower trend
You notice that several consecutive points are above or below than the average. There is a reliability problem. There is a bias.
(3) A increasing or decreasing trend
You notice that several consecutive points are regularly increasing or decreasing. There is a deterioration of the reliability or/and the accuracy.There is a bias.
(4) One point is between the warning limits and the control limits
Be careful with the next control result.
(5) The last point is outside a control limit
Confirm the result and take remedial actions if it is necessary.
The preventive rules
Attention, this interpretation is only valid if the consecutive control results are obtained in reproductibility conditions.
- If 9 consecutive points* are on the same side of the average (upper or lower trend),
- If 6 consecutive points* increase or decrease (increasing or decreasing trend),
- If 2 points* in 3 are between a warning limit and a control limi,
- If a result is outside the control limit.
* these points can be repeatability averages due to different conditions (different day, different operators…).
It can show a drift, you must take remedial actions.