Block 4 - lecture 1 Flashcards
How do you know when a process is in control?
- special causes have been eliminated so that all subgroups are within control limits
- there’s a natural pattern of variation
is uniformity improvement possible when a process is in control?
no, the process would have to be improved
what percentage of subgroups are expected between each standard deviation?
0-1 = 34% 1-2 = 13.5% 2-3 = 2.5%
Where are control limits typically established?
3 standard deviations from the central line
What are the 2 types of error?
Type I error - good but outside limits (ie. over 3sd 0.27% of the time)
Type II error - assignable cause, but within limits
why are control limits often set to 3 standard deviations?
to balance cost of type I and II errors
Positives of a process being in control?
- more uniform so less rejections (scrap/rework)
- less samples needed, reducing cost of inspection
- process capability is 6sigma (stable and repeatable)
- trouble can be anticipated
- percentage in each range easy to predict
decisions to make about processes?
- spec requirements
- scrap vs rework
- loosening specifications and using selective matching
benefits from producers sharing Xbar and R charts with customers?
customer requires less checks, as they have more trust
two causes of variation?
- common causes eg. chance causes (stable)
- special causes eg. assignable causes (change)
what does ‘out of control’ show?
a change in the process due to an assignable cause
how can you view a subgroup that falls outside the control limits?
it’s as if it comes from a different population to the control limits, due to special causes
examples of patterns in control charts? (using bands of 1 standard deviation)
> 2 in 3 consecutive points outside 2sd
> 4 in 5 consecutive points outside 1sd
> 6 points increasing or decreasing
> 7 consecutive points on one side of the central line
examples of patterns in control charts? (using bands of 1.5 standard deviation)
> 2 consecutive points outside 1.5sd
> 6 points increasing or decreasing
> 7 consecutive points on one side of the central line
benefits of using 1.5 sigma bands for patterns vs 1 sigma bands
quicker and easier for operators
examples of patterns in control charts? (using the shape of plotted points)
- change/jump in level
- steady change in level
- recurring cycles
- two populations
- mistakes
possible causes of a jump in level in Xbar charts?
- change of settings (intentional or not)
- wrong setup
- process skipped
- new operator
- different material properties
- machine failure
possible causes of a jump in level in R charts?
- inexperienced operator
- sudden gear play
- material variation
possible causes of a steady change in level in Xbar charts?
- tool wear
- deteriorating equipment
- gradual change of environment
- viscosity breakdown in chemical processes
possible causes of a steady change in level in R charts?
- improving skill
- fatigue, boredom etc.
- material becoming more uniform
possible causes of a recurring cycle in Xbar charts?
- environment (seasonal or daily)
- daily/weekly event
- operator rotation
possible causes of a recurring cycle in R charts?
- operator fatigue after breaks
- lubrication cycles
danger of recurring cycles?
The inspection frequency might not pick up on a cycle if it only records one part of it (eg. same time every day)
possible causes of two populations in Xbar charts?
- large material differences
- multiple machines
- multiple methods
possible causes of two populations in R charts?
- multiple material suppliers
- different operators
possible causes of mistakes in Xbar charts?
- measuring calibration
- calculation errors
- equipment usage
- samples from different populations (different machines etc.)