Statistical Process Control Charts; SPC; Continuous Business Improvement
Continual Business Improvement using Control Charts
If you want to survive in business and make a profit then you need to have a planned process of continuous business improvement and use quality tools such as statistical process control charts (SPC). There are a number of quality tools around that have been shown to help continuous improvement, one of them is the SPC or Statistical Process Control chart.
The Statistical Process Control chart is a tool on which data regarding your process performance is recorded on a regular basis, in some cases hourly, others maybe daily. The reason is to compare the performance now against past performance and define if the data you have just taken is within the statistically expected range. If the data is beyond what is expected or a statistically unusual trend is observed then it could indicate that the process is out of control and that some form of special cause is present.
By being able to identify when things are changing it is often possible to prevent the creation of defective materials, rather than rely on inspection to remove defects.
Continuous Business Improvement using SPC Charts
Purpose of an SPC Chart
All processes have variation and this variation if stable can be predicted over time. Thus it is possible to take discrete measurements of either the process variables or the process output to monitor output and ensure that nothing unusual is occurring. The power of control charts is that they can often tell you that you have a problem long before your process begins to produce reject product.
The Statistical Process Control Chart or SPC Chart therefore is a very powerful quality tool for continuous Business Improvement.
Types of Statistical Process Control Charts
Control charts come in two main types;
- Variable charts that deal with variable data, that is data that will vary across a range of measurements such as temperature or length.
- Attribute charts, this is the data that we would normally record using a simple tally chart, so number of rejects would be controlled on an attribute chart.
SPC Charts Selection
What Sort of SPC Chart do I Need?
The diagram to the right is a simple breakdown of the purpose of each type of chart by which you can decide which specific chart you need to look at the process that you wish to monitor. With regard to attribute data, there is a distinction between defects and defectives. A defective is a unit that has defects present causing it to be a defective unit, whereas Defects are the number of specific faults that you count, so there could be more than one defect on a defective product.
X Bar R Chart Statistical Process Control
X Bar R Process Control Charts
The most commonly used chart for measuring variable data is the X bar R chart. This type of chart is very simple to use, the operator will take usually three or five samples from production and record them on the chart. They will then plot the average (arithmetic mean) of their sample (all of the samples added together and divided b y the number of samples) on the top portion of the chart. And record the range (difference between the largest and smallest of the samples) on the lower portion of the chart.
The chart will already have limits calculated from previous production runs drawn apon it, if the process is “in control” and no “special causes” are present then the results will fall within these limits falling randomly either side of the center line. Below are some simple rules to use to interpret the control charts.
Attribute Control Charts
Attribute Process Control Charts
An attribute chart will measure either defects of defectives, the data normally being recorded hourly on an hourly tally sheet. The operator will then plot the percentage of defects or defectives vs the number of products produced. Again the results should fall randomly within the control limits, the rules for interpretation being below.
Defectives being the actual number of products that are nonconforming while defects are the actual causes of the defectives, a reject could have several defects!
How to use SPC Charts
Statistical Process Control Chart Analysis
A rules of a control chart are based on the premise that the process is “under control”, that is that there are only the normal factors that influence variation and that nothing out of the ordinary effecting the process. Should a process become “out of control” then there are “special causes” of variation present that are potentially going to cause problems.
The rules for checking if a chart is out of control are as follows;
- Single point more than 3 sigma from centre line, outside of the limits.
- Nine consecutive points to one side of the centre line
- Six consecutive points that increase or decrease
- Two from three points more than 2 sigma to one side of the centre line
- Four from five points more than 1 sigma to one side of the centre line
- Fourteen consecutive points that alternate either side of the centre line
- Fifteen consecutive points within 1 sigma of the centre line
- Eight consecutive points more than 1 sigma from centre line
All of the above situations would indicate that something out of the ordinary has occurred within the process and that action should be taken.
Out of Control Points on an SPC Chart
Statistical Process Control Charts Video
Process Capability, CP and CPK
There are two measures by which we can measure a processes capability, capability being a processes ability to meet the stated specification.
The first is “CP”, CP is a direct comparison of the total variation of the process against the specification. So a process with a variation of + or – 3 sigma (standard deviations) that is the same as the specification would have a CP of 1. This process if centered correctly would be able to produce +/- 3 sigma of output within the specification. (around 99.8% of output)
The CPK is a similar measure but takes into account the actual process setting or mean. If the process above was perfectly centered on the mean then it would also have a CPK of 1, however should the process be either above or below the mean then the CPK would drop accordingly.
Six sigma capability would be asking for a CP/CPK of at least 2.
A term that is not always understood is that of "over-control" and is what sometimes happens if operators overreact from the signals that the receive from their process. This will result in more variation in the process rather than less.
An example could be that an operator takes a measurement each hour and then adjusts the process parameters acording to what measurement they take. So if the measurement is two points high they adjust the process down tow points, if the next measurement is one point down they adjust it up one point. This may at first glance seem sensible, however;
every process has variation that is normally distributed can be predicted.
Imagine throwing a standard die time after time.. you would get an average of 3.5 and the results would vary between 1 and 6.. Now imagine what would happen if you corrected each throw according to the ideas above. If we throw a 6 we need to take 2.5 from our next throw. If that were a 1 we would then make our adjustment creating an even greater amount of variation than would have occurred naturally. This is over-control and is surprisingly more common than people believe!
Continuous Business Improvement Tools
Continuous Business Improvement
Continuous business improvement has to be planned and managed. Without a continual process of business improvement you will always be chasing your competitors and your problems. If your business is to flourish then there are a number of quality tools that you can utilize to help you drive continuous business improvement, Statistical Process Control (SPC) is just one of those tools and is best used alongside all of the other quality tools to gain the best results.