Patterns of Variation Flashcards
Variation
- Lack of consistency
- It can introduce waste and errors into the process
Why measure Variation?
- Reliability - we want our customers to know that they’ll always get a certain level of quality for us. Also, we’ll often have a Service Level Agreement or similar in place. Consequently, every product needs to fit specific parameters.
- Cost - Variation costs money. So to lower costs, we need to keep levels low
Discrepancies occur when:
- There is wear and tear in machines
- Someone changes a process
- A measurement mistake is made
- Material quality or makeup varies
- The environment changes
- A person’s work quality is unpredictable
Six Elements in any Process
- Method
- Mother Nature, or Environmental
- Man or people
- Measurement
- Machine
- Materials
Process Spread vs Centering
Types of Variation
- Common Cause
- Special Cause
Common Cause Variation
- Happens in standard operating condition.
- Think about a factory. Fluctuations might occur due to:
- Temperature
- Humidity
- Metal quality
- Machine wear and tear
- Common cause variation has a trend that you can chart.
- In the factory, product differences might be caused by air humidity. You can chart those differences over time then compare that chart to weather bureau humidity data
Special Cause Variation
- Occurs in non-standard operating conditions.
- Factory example, disparities could occur if:
- a substandard metal was delivered
- one of the machines broke down
- a worker forgot the process and made a lot of unusual mistakes
- This type of variation does not have a trend that can be charted.
- Instead, you’ll see a departure from a trend
- Factory example, disparities could occur if:
Why is it important to differentiate Common and Special Cause Variations?
- Different factors affect them
- We should use different methods to counter each
- The wrong change can cause even more discrepancies
How to identify Common Cause Variation?
-
Run Charts
- Mark your median measurement
- Chart the measurements from your process over time
- Identify runs.
- These are consecutive data points that don’t cross the median marked earlier. They show common cause variation
How to identify Special Cause Variation?
-
Control Charts
- Mark your average measurement
- Mark your control limits.
- These are 3 standard deviations above and below the average
- Identify data points that fall outside the limits marked earlier.
- In other words, above the upper control limit or below the lower control limit. These show special cause variation
Variation Formula
Variation = SD2
Variation is the square of a sample’s standard deviation
How to find the cause of Variation?
Multi-Vari Chart
How to Counter Variation
- Counter Common Cause Variation using long term process changes
- Counter Special Cause Variation using exigency plans
- Extra or replacement processes
Combining Variation
- Rather than finding variation in a single sample, you might need to figure out combined variance in a data set.
- For example, a set of two different products. For this you’ll need the variance sum law
- First, look at whether the products have any common production processes
- Second, calculate the combined variance using one of the below formulas:
- No shared processes formula
- Shared processes formula
- For example, a set of two different products. For this you’ll need the variance sum law
No Shared Processes Formula
The two products don’t share any production processes? Great! Then you can use the simplest version of the variance sum law.
Variance(X + Y) = Variance(X) + Variance(Y) Variance(X - Y) = Variance(X) + Variance(Y)
Shared Processes Formula
The two processes do share some or all production processes? That’s OK, you’ll just need the dependent form of the variance sum law instead.
Variance(X + Y) = Variance(X) + Variance(Y) + Covariance(X,Y) Variance(X - Y) = Variance(X) + Variance(Y) - Covariance(X,Y)
Calculate covariance using the following formula.
Cov(X,Y) = Σ ((X-μ) * (Y-ν)) / n-1
where:
- μ is the mean value of X.
- ν is the mean value of Y.
- n = the number of items in the data set.