Statistical Process Control (SPC) CONTROL
Type 1 Error
Alpha, rejection of the hypothesis when it is true. Occur when a behavior treated as a special cause has no effect in the process.
What are control charts used for
Attain statistical Control Monitor a process Determine Capability NOT to reduce variation, other techniques use control charts as reference for that
U chart
Attribute Control Charts designed for tracking the number of defects per unit for discrete data. Used when defects rather than defectives are counted. Scratches, bubbles, chips, etc. Defects is denoted by u. As in u suck. Fraction is calculated and plotted, # defects/sample size.
What is different between making a variable chart and an attribute chart
Attribute charts use much larger sample size, 50
Type 2 Error
Beta, acceptance of the hypothesis when it is false. Occur when special causes affecting the process are not addressed.
SPC Objective
Center process monitor process for change When or not to take action reduce variation improve understanding monitor and improve changes operating as expected?
How do control charts maintain statistical control and provide traceable evidence?
Collection Control (calculate limits) Capacity
Where does variable data come from
Continuous and come from scales or measures
SPC Tactics
Design of Experiments, multi-var analysis, tolerance band target zone.
Where does attribute data come from
Discrete - come from indicators, % defective, need 50 + samples
Where does common cause variation reside?
In processes within statistical control and can be characterized by location (process average), spread (piece to piece var), and shape (distribution).
Success of Control Charting depends on?
Rational selection of subgroups. Should be Homogeneous as possible. To eliminate process change smaller subgroups (5 subgroups of 5 measurements) are more useful.
When Does SPC monitor the process
Real Time collection and analyzing of data
SPC Theory
Shewhart, variation in manufacturing could be attributed to inherent (random or Common - Noise) variation and intermittent (assignable or Special) variation.
What is the basis of the control portion of 6 sigma?
The correct application and interpretation of control charts, Statistical Process Control. Control Charts separate assignable causes from random variation.
What is the difference between a u chart and a c chart
U chart graphs factional percent defects (not defectives) and can have non consistent sample sizes. C chart graphs integer defects (count), sample size must be constant.
How to calculate control limits for Xbar and R
UCLxbar=Xbarbar +A2*Rbar LCLxbar=Xbarbar - A2*Rbar UCLrange=D4*Rbar LCLrange=D3*Rbar A and D are CC constants dependent on sample size
Individual and moving range chart
When sample size is 1. Moving range is calculated by taking the |difference between measurements|, absolute value. Constants again change. First moving R entry will be blank, duh.
Common Variable Charts
Xbar - and R, Xbar and s, individuals and moving range, and median charts
When to use and Xbar and R Chart and When to use an MXbar-MR (moving average, moving range) are variable charts
Xbar R = data readily available MXbar-MR = limited data
What are the most common type of control charts
Xbar and R charts. Generally 4 - 10 samples
I/XmR and Xbar s charts are variable charts
XmR=limited data, INDIVIDUAL moving range Xbar and s charts = sigma available
Is a process behavior chart another name for a control chart and SPC?
Yes developed by Shewhart (father of quality control). 30 different types of charts. Typical 6 or 7 types are used.
operating characteristics (OC) curve
a graph that describes how well an acceptance plan discriminates between good and bad lots, highlights the risk of type II error in the sample.
Run Charts and Control Charts
a run chart is a line graph with data plotted over time (single-point data); control charts include control limits.
c Chart
control chart for attributes, measures the # of defects within a sample. Sample size must be consent and integer not fraction is charted.
What is Statistical Process Control
data collection and analysis method used to support decisions with evidence, values in CC may be continuous (variables) or counts of occurrences (attributes)
How to establish overall mean and range
data collection during a base period, against which results can be compared to determine process behavior.
np chart
measures the number of rejected items in a sample as integer, not proportion. Sample sizes must be constant!!! NOTE: when the formula for LCL produces a neg number, no LCL is used. Attribute chart
What is the goal of any quality activity
meet needs of customer
Control Charts for Attributes (defects, pass/fail,%)
p-charts=defectives-varying sample size np-charts=defectives-fixed sample size c-charts=Defect-sample size fixed u-charts=Defects-sample size varies short run charts for all
Median Control Charts
plot the median rather than the average, use when outliers are expected. Different because all data points are plotted and and range gage is used to spot outliers, page 396.
P Control Chart, p-chart
proportion of defective parts within the group under review. Either conform are are rejected. Control Limits are calculated based on formulas that proportion discrepancies to sample sizes and sample size rations. Page 397. Attribute Chart
Process Behavior Charts or SPC
realtime Picture of process variation early warning, ensure process is stable and operating within process boundaries Attributes data are discrete Use UCL and LCL from a table, being in control refers to being between these 2 limits.
Xbar and s CC
sigma calculated from ALL values in subgroup. When subgroup is small < 5 is small, range and sigma charts are similar.
Statistical Process Control (SPC)
the application of statistical techniques for measuring and analyzing variation
Statistical Process Display (SPD)
the misapplication of statistical techniques to pretend that info is being used to run a process.
When do you use an np chart
when sample sizes are consistent and you want the actual number of rejected items not proportion.
Is Calculation similar between Xbar and R and Xbar and s?
yes, except different constants are used and SIGMAbar is substitued for Rbar.
Attribute Control Charts
Poisson or binomial Dist. Require 50+ samples in subgroup.
How many subgroups are need to attain statistical control for Continuous
25 (measurements 4-10), also log of process changes during data collection, trial control limits, charting data for each subgroup, eliminating assignable causes
Difference between Defects and Defectives
A defect is an item that exhibits departure from specification (nonconformities), doesn't mean the item can't be used. A defective is completely unacceptable for use.
Control Process
A feedback loop through whcih we measure actual performance, compare it with a standard and act on the difference
What are the 3 types of variables (identify through QFD, measurable and important to customer)
KPIVs-Key Process Input Variables KPOV-Key Process Output Variables DOE & ANOVA methods identify variables
After Special or Assignable Causes are removed, what next
New Control Limits calculated from 25 new subgroup averages and ranges narrow the process capability
What is the difference between a p and an np chart
P chart graphs proportion of defectives (rejects) as fractional percent, while np chart charts integer count of rejects, must have constant sample size.
Which type of variation should be discovered and removed first
Special or Assignable - Causes are usually related to some defect in the system or method
What causes a process to go out of control?
Special or assignable causes.
Specification limits and Process Variation
Specification Limits should reflect VOC and Process Variation reflect VOP.
Statistical Quality Control (SQC)
Statistical techniques for measuring and improving the quality of processess
What are control limits based on
Statistical variation of the process and can be established at +- 3 stdDev from the mean
What is Rational Subgrouping
Subgroups chosen so that units(measurements) have very close alignment and units between subgroup show a difference. Must be in control.
Assignable Cause
The observed change is caused by the behavior of a major variable in the process.
Chance or unassignable cause
The observed change is caused by the interplay of multiple minor variables in the process
What are control charts used for
attain statistical control monitor a process determine process capability