How to Better Understand Your Business Processes by Plotting Their Run Chart and Frequency Distribution

Abstract:

Here is an explanation and example of how to analyse and improve business processes by charting existing historic data. An event run chart and the event frequency distribution are simple, fun, and insightful to plot with a spreadsheet. The two plots help you understand and learn why a process behaves as it does, and to consider what could be done to get better performance and results.

Keywords: process control charts, process data distribution, run chart, frequency distribution plot, big data analysis, process performance analysis, process continuous improvement

Organizational problems this article helps you to address:

      • Identify those business and operational processes with performance irregularities
      • Recognise the frequency of recurring performance problems
      • Propose new ideas for trial to address poor process results

If you are an executive, manager, or engineer who is wondering what is causing problems in a business or operational process, there are a couple of useful graphics—the run chart and the frequency distribution plot—that you can construct from historical data to help you explore what is happening ‘behind the scenes’ in the process.

The two images in this article are extracted from Chapter 17 of the Industrial and Manufacturing Wellness book. The first image is a run chart of ten years of production outages at an industrial site. The data comes from the operational performance report generated weekly for management review.

Using The Process Run Chart

Showing the run chart data points is done with a spreadsheet scatter plot. The Y-axis is the period of downtime during which the plant could not produce due to an outage. The X-axis is the date and time of an outage. The 1960’s and 70’s dates are fictious to protect the plant from being identified.

production-downtime-outage-run-chart

The run chart is the most basic of control charts and it is always the first plot developed in a process performance analysis. As famously said by Irving Burr in 1953, “The control chart is the process talking to us.” And there is a lot of information and insights being communicated to us in this outage 10-year history plot.

One observation is the data points are NOT randomly distributed. There are clusters of closely occurring downtimes that occasionally appear during the ten years. In-between the clusters, the occurrence of outages continually happens. There also is stratification evident in the durations. The spread of downtimes ranges massively from under an hour to over 160 hours. Each observation will have reasons for their cause, and if they were addressed the plant would become much more productive and profitable.

Using the Process Frequency Distribution Plot

The next process analysis chart is a frequency distribution plot of the production uptime periods when the plant was operational. The data in the first plot—the run chart—is of the outage periods. Between each outage is the uptime duration where the plant is running and producing. Thus, from the downtime data is inferred the uptime data used in the below ‘Frequency of Days with Uninterrupted Production (Uptime)’ control chart. The uptime distribution the organisation wants is the dashed curve, but what its corporate-wide processes are producing is the ‘hockey stick’ curve.

frequency-distribution-of-production-uptime

This frequency distribution plot also tells and informs us about the process’ behaviours. The uptime frequency distribution is mostly on a near straight line. A straight-line distribution is an ordered arrangement. The implication is the production process plant and equipment are being impacted by a variety of repeated, irregularly occurring events that cause the forced stoppages. Decisions are being made and actions taken that make the outages happen. This organization not only makes products, but it also makes forced outages as a product of running its operation.

Armed with the above insights from the charts, an investigation into the range of downtime causes and their removal can be instigated. A logical initiative to undertake would be to investigate the stratification of outages less than 10 hours, those lasting 10 to 20 hours, 20 to 60 hours, and more than 60 hours duration. The expectation is high that the new understandings gained about the causes producing the various stratifications of plant outages will lead to process continuous improvement projects with outstanding return on investment (ROI). After all, the outages are self-imposed and that means those decisions can be challenged and changed for the better.

Statisticians advise a minimum of 16 to 20 uninterrupted data points are necessary to plot a reliable run chart and distribution curve. There needs to be enough points to be representative of the typical performance of the process when in operation. More examples, and their spreadsheet templates, are available in the Industrial and Manufacturing Wellness book.

Mike Sondalini
PWWEAM System-of-Reliability
13 May 2022