SPC blog
Statistical Process Control
Optimizing and Controlling Processes
The Five M’s
All processes are affected by a multitude of factors:
● Machines ● Measurements ● Methods ● Materials ● Manpower
Statistical Process Control (SPC)
SPC does not eliminate all variation in the processes, but it does one thing that is essential if the process is to be consistent and if it is to be improved:
SPC allows workers to separate special causes of variation from natural variation.
After all special causes have been identified and eliminated, leaving only natural variation, the process is said to be in Statistical Control.
Statistical Process Control (SPC)
A statistical method for separating variation resulting from special causes from variation resulting from natural causes in order to eliminate the special causes and to establish and maintain consistency in the process, enabling process improvement.
The Rationale for Embracing SPC
1. Control of Variation 2. Continual Improvement 3. Predictability of Processes 4. Elimination of Waste 5. Product Inspection
Rationale: Control of Variation
Rationale: Control of Variation
Variation in any process is the enemy of quality.
Rationale: Continual Improvement
● The processes must be understood, plus the external influences that are special causes of variation.
● Special causes must be eliminated. ● Then, and only then, can the process be evaluated on its own
merits. ● Once the process is within control, SPC can be used to identify
natural causes of variation or flaws in the process that can be adjusted to reduce or eliminate variation.
Rationale: Predictability of Processes
Scenario: You’re the producer, and you get a big order. How do you know if you are capable of delivering the required amount/number of products? How do you answer the customer?
● Do you underestimate so you don’t fall short? (Under promise - over deliver?)
● Do you agree to the required amount if you’re not sure you can deliver?
What are the problems with each?
Rationale: Elimination of Waste
● Cost of goods is lowered. ● Quality is enhanced.
Phillip Crosby once said “Quality is free.” Why is this a conservative statement?
Quality pays dividends.
Rationale: Product Inspection
● Product inspection requires people with expertise, inspection equipment which costs money, or both.
● 100% inspection does not guarantee catching 100% of defects.
● Keeping processes in statistical control allows the manufacturer to “back off” of 100% inspection and do sampling.
Control Chart Development
Variables Data - measured values, such as height, weight, length, etc.
Attributes Data - counted values, such as number of defects, etc.
Control Chart Development
x - the mean (average) of samples over time
R - the range (variation) of each sampling over time
n - size of the subgroups
k - number of subgroups
Control Chart Development
1. Determine sampling procedure. 2. Gather initial data of 100 or so data points in k
subgroups of n amounts/numbers of samples. 3. Calculate the mean ( x ) of each subgroup. 4. Calculate the range (R) of each subgroup. 5. Calculate the average of the subgroup averages x.
This is the process average and will be the centerline for the x -chart.
Control Chart Development
6. Calculate the average of the subgroup ranges R. This will be the centerline for the R-chart.
7. Calculate the process upper and lower control limits, UCL and LCL respectively (using a table of factors, such as the one shown in Figure 6 ). UCL and LCL represent the +3 limits of the process averages and are drawn as dashed lines on the control charts.
Control Chart Development
8. Draw the control chart to fit the calculated values. 9. Plot the data on the chart.
Control Chart Development
● R and x charts are for tracking variables data.
Control Chart Development
● p-charts are for tracking attributes, such as defective parts.
Control Chart Development
● c-charts are for number of defects in one piece.
Inhibitors of SPC
● Capability in statistics ● Misdirected responsibility for SPC ● Failure to understand the target process ● Failure to have processes under control ● Inadequate training and discipline ● Measurement repeatability and reproducibility ● Low production rates