Short Run SPC

Dr Juergen Ude Dr Juergen Ude

This article will discuss Short Run SPC, often hailed as revolutionary. Short Run SPC has been devised to overcome the difficulties of applying effective SPC to short production runs. One of the charts used is the nominal chart which records the deviation from a nominal reference point, e.g. specification target. This enables combining results of several product lines onto one chart, provided all products exhibit the same variability.

By combining data it is believed that process changes, such as drift and tool wear can be detected. For example, consider Figure 1a, where a well defined trend is visible even with random variation super imposed. If only a small subset of this data is selected and plotted, such as the points between A and B in Figure 1a, then this trend is no longer apparent, as shown in figure 1b. The belief in the converse that an increase in data helps to resolve trends and other patterns is therefore understandable.

Short Run SPC Diagram representing a well defined trend with random variation super imposed FIGURE 1(a): TREND
Short Run SPC Diagram with no trend visibile FIGURE 1(b): NO TREND WITH REDUCED DATA SET

However such an extension of the logic is not necessarily true. There is a serious flaw in the logic, which makes this “revolutionary” Short Run SPC method inefficient and ineffective, as a tool to detect sustained process changes. Combining data only works if the underlying trends are joined. Unfortunately this is practically impossible through the use of nominal values.

The best way to demonstrate this is by example. Consider the manufacture of three different products. The rate of tool wear i.e. slope is the same for all three products. Product A is manufactured, followed by product B, The trend for product B will continue from the RESET process average. The same applies to product C. The three products are plotted on a graph as shown in Figure 2a. Distinctive trends have been incorporated, which in practice may not be visible. When we then use the nominal chart and combine the data as shown in Figure 2b, a continuous trend is not shown. This shows that the uncombined plots are better, contrary to that alleged by proponents of the “revolutionary” method.

For this particular example the combined data falsely indicated a degree of cycling. In practice the degree of cycling depends on the number of points, degree of trend and variability about the trend.

Short Run SPC Diagram representing products with different targets FIGURE 2(a): 3 Products with different targets
Short Run SPC Diagram representing products combined using deviations from nominal FIGURE 2(b): Products combined using deviations from nominal

There are other problems. The process average for individual parts is unlikely to be centred on target, since only a few points are used to estimate the mean. Between time to time variation will therefore exist, which can result in inadequate control limits, if sub-group data was used to set these limits.


Using nominal values and combining data is not a revolutionary method that detects otherwise undetectable trends etc. It may be considered an ineffective application of SPC which can divert limited resources away from more beneficial applications.

There are no easy solutions for Short Run SPC, (as is the case for quality improvement in general). Options are to increase inspection frequency by reducing inspection time. Pre-control is a fast method. Although there are some who do not condone.

Another option is on-line automatic sensors.

A third option is increasing production run length. Short production runs stem from the concept of JIT. As with most concepts, there seems to be a blind adherence to the concept, instead of evaluating whether it is applicable to the situation. Reducing inventory may reduce tied up capital costs but may also increase other costs. Increasing production run length can release tied up capacity, reduce set up costs, reduce operational complexity and permit more effective quality control. Increasing run length MAY therefore be a viable alternative.

Download the Inferences APP, comprised of mainstream and machine-powered analytics for statistical analysis

Analytics as a Service (AaaS) for Quality

Drive quality improvement through actionable insights using analytics you can trust! Use up to 200 analytics tools downloadable through a suite of Apps!

FREE usage of the analytics Apps for quality improvement
  • Augmented with machine-powered smarts
  • Always updated with the latest tools and features
  • No licencing or fixed subscriptions - Pay ONLY for the analysis you run from 20 USD cents per analysis, billed monthly! Set a budget so you don't exceed!