Every process has a sweet spot where everything seems to be running perfectly. Unfortunately, not only do processes have variation, they also have the tendency to drift. 

Let’s explore a little further what we mean by drift, what can cause it, how to monitor it, and how to prevent it.

Overview: What is drift? 

As your process performs over time, you will notice a tendency for the mean to change. This variation and change may be defined as a process shift where the mean can be higher or lower over time than the process expectation. In some cases, this shift is unidirectional and slowly moves in only one direction. This change is then described as a drift rather than a shift.

The signs of drift are often subtle and sometimes get masked if the overall variation is larger than the drift. Once the process has sufficiently drifted from its desired state, the process behavior will be noticeable enough to indicate something needs to be looked at. Process drift may occur as a result of aging of raw materials, wear and tear on equipment, decreased sensitivity of measurement equipment, or fatigue of process operators.

Control charts are a good tool for monitoring your process behavior over time. By properly selecting your samples and applying such rules as the Western Electric Rules, you should be able to spot any process drift before it becomes a problem. 

Rather than this more reactive strategy, though, you might consider a formalized process of recalibration, whether it be for the equipment or people. There is a natural degradation of mechanical equipment over time. Recalibration will bring the operating conditions back to its normal state before any negative impacts will occur. The same for people. Through retraining and reorientation, people can be recalibrated in a similar manner as equipment.

An industry example of drift 

A manufacturing manager was using control charts to monitor the key quality characteristic of one of the company’s primary products. As far as he knew, there was nothing changing in the process and everything was running at the required parameters and settings. 

It took him a while to notice, but his control chart indicated a drift of the quality characteristic to the low side. Once he noted this drift, he took corrective action and kept the product from violating any customer specs. If he hadn’t noticed the drift on the control chart, it would finally have shown up as more products being out of spec. Below is the control chart he was using.

Frequently Asked Questions (FAQ) about drift

1. What is process drift? 

An unintended deviation of the process from the desired setting and performance parameters. This may result in product or process characteristics being outside of documented requirements or specifications.

2. What is the difference between a process shift and a process drift? 

A process drift is usually in one direction while a process shift may be higher or lower than the original setting. 

3. What can cause a process drift? 

Drift may result from the natural wear and tear associated with equipment, fatigue of personnel, or degrading or aging of raw materials. Drift may be subtle and masked by the common cause variation of the process. Recalibration of both equipment and people may need to be done on an ongoing basis.

About the Author