The aim of Statistical Process Control charts is to improve the overall process. In this article, we will explore the signs of an unpredictable chart and determine how we might work to eliminate the cause of signals to make the metric more predictable over time.  

When met with an unpredictable chart, the goal is to figure out the cause of the signals so that we can make the metric more predictable. Once you've identified the signals and the metric becomes more predictable over time, it becomes much easier to focus on systematically improving the stable metric and even boosting the average line. 

In this example, lets' observe the unpredictable production output numbers below. As you can see, the average is nowhere near the goal line, and there are several places where signals are sporadically occurring. You might also observe the broad range in production output numbers between the upper and lower limit lines, which is another tell sign of unpredictable metrics.

Negative and Positive Spikes (Rule 1)

Some negative production output signals might be because of unexpected production downtime, power outages, or any other unusual situations that are temporary. One question that is worth asking when identifying negative rule 1 signals is whether or not the issue will resurface in the future and how we might be able to prevent it in the future.

‍

In the case where there is a positive production output signal above the upper limit, there might have missed an opportunity to sustain the improvement over time. A question you can ask when a positive spike is recorded is what might have caused the positive impact and whether or not it is a process that can be captured and distributed across the organization.

These signals (rule 1) are generally followed by an immediate return to the average range line, which could signify that the cause was identified and resolved, at least in the short term. 

When working with an unpredictable SPC chart, you should take the time to investigate and identify each signal that occurs. Over time, as you eliminate the cause of each signal, the metric should become more predictable. 

You can learn more about how to identify signals and the different rules here.

Identify Root Cause of Signals Using "The Five Whys"

One popular method of identifying root cause was popularized by Toyota and during the Lean Manufacturing movement, sometimes called "The Five Whys." The ideology of asking "Why" at least five times when an exceptional variation or signal is identified is to drill deeper into the root cause. The goal of asking "Why" is to get to a point where the countermeasure fixes the problem in the short term, but most importantly also the long term.

Below is an example of the five whys applied to find a root cause.

#1 Why did the machine stop?

Because the machine overheated and the fuse blew

#2 Why did the machine overheat?

Because the radiator is not cooling down the machine efficiently.

#3 Why is the radiator not cooling down the machine efficiently?

Because the coolant has not been replaced on the machine

#4 Why was the coolant not replaced in the machine?

Because there is no more coolant in the supply room

#5 Why was there no more coolant in the supply room?

Because the reorder level for coolant was not set frequently enough to sustain machine usage.

So the root cause of the machine overheating and a fuse blowing was a supply reordering issue.

‍

Below is an example of how an unpredictable metric SPC chart should look like over time as the signals are identified, and the metric becomes more predictable.

As you can notice, the average has shifted higher, and the upper and lower limits become more narrow. However, the average line has not yet hit the target goal; this sets up the team to start a systematic improvement process now that most signals have been identified and prevented.

In part 3, we'll go over how to improve a predictable SPC chart systematically.