Coronavirus and the critical need for equipment
With the outbreak of Coronavirus and the critical need we are going to have of equipment such as ventilators, a massive effort is going to be needed to ramp up production in such a short time. Techniques of quality and productivity improvement can help.
Which is the best technique?
Over the years there have been many canned techniques for quality improvement; 8D, DOTSTAR, Six Sigma, to name just a few. It can be daunting which is best, or are they the same?
There are two types of quality problems; chronic, and sporadic. Chronic quality problems generally present as a low level of rejects or non-conforming product, generally less than 5% of production, which over time most companies have learned to live with. Sporadic quality problems are huge spikes in rejects seemingly overnight. It is not unusual to fix a sporadic quality problem just to return to a chronic level of rejects. This is because the causes of the two problems are not the same.
To tackle sporadic quality problems, it sometimes can be best to use a process of elimination. In other words, something has changed in the system, with materials probably being the most frequent source.
Chronic quality problems
Chronic quality problems require a more systematic approach. My preferred approach, if the process is not in statistical control, is to start with a short-term machine capability study. This entails collecting 60 units of product evenly spaced out over a one-hour period of production during which no changes to the process occur. Each unit of product needs to be identified in its order of production and all the units need to be measured in a random order. Once all the measurements are entered into a spreadsheet or other program, the order of measurements needs to be examined for any nonrandom patterns that may be present; i.e. runs, trends, outliers, and or shifts in the mean. When it is certain that the measurements are good, the data can be examined in the order of production.
Again, looking first for nonrandom patterns. If there are no nonrandom patterns then group the data into subgroups of varying sizes; 3, 4, 5, 6, and 10. (This is why 60 units are a good number, as the data can be grouped into a greater number of evenly sized subgroups.) The purpose of putting the data into subgroups is the distribution of averages is more sensitive to shifts in the mean than individuals and any shifts would be more noticeable looking at averages.
Machine capability study
The reason to start with a short-term machine capability study, is there are four sources of variation; man, material, machine, and method. (For various reasons I don’t generally include measurement or environment as many others do.) The variation then in the short-term study is the least amount of variation there can be in the current process. Over a longer period of time, the additional sources of variation can only add to that variation, never reduce it.
Design of Experiments
Once the short-term variation is measured and understood steps can be taken to reduce it. Design of Experiments are particularly good for this task. There are many different types of DOE, some are best at identifying which factors from a list of all potential factors contribute the most to the variation in the process. Others are specifically used to measure that variation and how the different factors might interact. Still others are used to optimize the process settings for quality and productivity.
Recent Comments