Statistical process control refers to a statistical method used to separate variation produced b y special causes and varation produced by natural causes. This is done so that it is possible to eliminate the special causes and to establish and maintain consistency in the process, allowing the process to be improved.
A Process refers to everything that is done in a workplace. Multiple factors affect these processes and they are usually refered to as the Five M’s. The Five M’s are, the Machines employed, Materials used, the Methods(manuals for labour) provided, the Measurements taken, and the Manpower(workers) who operate the process. When all Five M’s are conformed(no misadjustments in the machines, no flaws in the material, work instructions are precisely followed and are totally accurate, people(manpower) following the work instructions properly and with full concentration), special causes are eliminated and the process will be in statistical control.
However, this does not mean the output from the process are 100% perfect. There is still natural variations which may affect the output of the process. Natural variation is expected to account for roughly 2,700 out-of-limits parts in every 1 million produced by a 3-sigma process, 63 out-of-limits parts in every 1 million produced by a 4-sigma process, and so on.
1.2 History of SPC
SPC(Statistical Process Control) originated as far back in 1931, when Dr Walter Shewhart wrote a book, The Economic Control of Quality of Manufactured Product. He is a statistician from Bell Laboratories which was the first to realise that data could be retrieved by industrial processes themselves. By using statistical methods, these data could then signal that the process is in control or affected by special causes(unnatural causes, predictable variation). (history of SPC)
1.3 Benefits of SPC
Implenting SPC is essential in a company’s manufacturing process. It benefits the process by reducing scrap, rework and warranty claims. SPC also improves productivity and still retains quality as natural and special variations are isolated. It also aids in continual improvement which ultimately reduce cost and increase profits.
2. Statistical Process Control Charts
There are various control charts to measure variables data(measured values) and attributes data(counted data).
Variable data charts:
Variable charts are used to measure counted data. It is done so by using collected data and analysis both refined to measure variability of parts and processes. It is usually used after coming up with an attributes chart(when products and processes have been prioritized to investigate)
Below are examples of charts used to measure variables data:
X-bar and R Chart
An X-bar and R Chart is used to show any chanes in mean value and dispersion of the process concurrently. This is very effective for identifying any defects in a process. The X-bar component shows changes in the mean value while the R component shows changes in the dispersion of the process.
Fig 2.1: X-bar and R chart sample
X-bar and s chart
X-bar and s chart is similar to the X-bar and R chart. However instead of showing changes in the dispersion of the process(represented by R), it uses standard deviation which is represented by the S portion. This helps provide a more effective measure of the process spread
Fig 2.2: X-bar and s(sigma) chart sample
Attributes data charts
Attributes data charts helps with the beginning of a quality improvement process by displaying data on defect rates. It can help one prioritize processes and products to investigate them later in detail with variables charts.
Below are various charts that measures attributes data:
A p-chart is used with data collected in subgroups that vary in sizes. Instead of the actual count, it exhibits a proportion on nonconforming items. It shows how the process changes with time. The process...