1. Operational decision making based on SPC (actual vs interpretation
1.1. Common Cause
1.1.1. Interpreted as
1.1.1.1. Common Cause
1.1.1.2. Special Cause
1.1.1.2.1. Type 1 error
1.1.1.2.2. Chasing ghost
1.2. Special Cause
1.2.1. Interpreted as
1.2.1.1. Special Cause
1.2.1.2. Common Cause
1.2.1.2.1. Type 2 error
1.2.1.2.2. False -
1.2.1.2.3. Missing the boat
2. In business Cp≈ 1.33 is good enough
3. Change in the distribution itself
3.1. centrality changes(mean)
3.2. spread changes(std. deviation)
3.3. Distribution shape changes
4. Whats big deal (the Unit)
4.1. Ordinary run charts are inadequate as they given guidance for interpretation in movement
5. Big Ideas
5.1. Control limits vs. Spec Limits
5.1.1. Definition => -/+3 σ
5.1.2. Objective => in control vs produces defects
5.1.3. Tool => control chart vs. histogram/box chart
5.1.4. Change trigger => process chg vs customer need
5.1.5. phylosophy => Voice of the Process vs. VOC
5.2. Causess
5.2.1. Common
5.2.1.1. Random variation within the process
5.2.1.2. Hard to remove
5.2.1.2.1. usually involves process redesign
5.2.1.3. Predictable
5.2.2. Special
5.2.2.1. Specific/ assignable events
5.2.2.2. Relatively easy to eliminate
5.2.2.3. Unpredictable
5.3. Stable Process
5.3.1. All data falls under +/-3 sd
5.3.1.1. I.e. between UCL and LCL
5.3.2. 99.74% values
5.3.2.1. I.e. 3 in 1000 items
5.4. Capable Process
5.4.1. All data falls under USL and LSL set by the customer
6. Chart Type
6.1. Continuous
6.1.1. Single observations
6.1.1.1. I(individuals)
6.1.1.2. Moving Range
6.1.1.2.1. Onlyoutside control limit rule applies
6.1.2. Averages
6.1.2.1. X Bar
6.1.2.2. Range
6.2. Discrete
6.2.1. Data in categories
6.2.1.1. p(roportions)
6.2.1.1.1. Usage
6.2.1.1.2. Assumption
6.2.1.1.3. Type
6.2.1.1.4. Application
6.2.2. average conforming and non conforming results > 5
6.2.2.1. np>5
6.2.2.2. n(1-p)>5
6.2.3. Occurrences being counted
6.2.3.1. c
6.3. Assumption
6.3.1. data is ordered by time
7. SPC Method
8. Checking for Special Causes
8.1. Extreme Points outside ULC and LCL
8.2. Shift
8.2.1. 8 or 8+ points above or below the CL
8.3. Trend
8.3.1. 8(7) or 8(7)+ points up or downward
8.4. Too quiet
8.4.1. 15 consecutive points falling within Zone B on either side of the CL
8.5. Too noisy
8.5.1. 2 or 3 points above and below the CL
8.5.1.1. I.e. too much variability
8.6. Zone rules
8.6.1. Zones
8.6.1.1. Zone A
8.6.1.1.1. 2 sigma warning limits
8.6.1.2. Zone B
8.6.1.2.1. 1 sigma warning limits
8.6.2. Rules
8.6.2.1. Zone A analysis
8.6.2.1.1. 2 of 3 consecutive points falling above +2 sigma or falling below -2 sigma
8.6.2.2. Zone B Analysis
8.6.2.2.1. 4 of 5 consecutive points falling above +1 sigma or falling below -1 sigma
9. Real time monitoring
9.1. Tampering by changing process for quick wins
9.1.1. increases or upsets natural variability within the process
9.1.2. micromanaging
9.1.2.1. process looses equillibrium
9.1.2.2. looses capability
10. Process Capability
10.1. What
10.1.1. Process capability Ratio (Cp)
10.1.1.1. how many times the allowable tolerance between USL & LSL covers for the natural process variation (UCL-LCL=> 3+3+ => 6 sigma)
10.1.1.2. Cp=(USL-LSL)/6s
10.1.1.3. Application Limitation
10.1.1.3.1. process (UCL, LCL) is centred around the mean
10.1.1.3.2. Pg. 4-24/25
10.1.2. Process Capability Index (Pck)
10.1.2.1. Pck=min(Cpl, Cpu)
10.1.2.1.1. I.e. which specification limit is likely to be breached easily?
10.1.2.1.2. Cpl=(Xbar-LSL)/3s
10.1.2.1.3. Cpu=(USL-Xbar)/3s
10.1.2.1.4. Cpu=Cpl=1 when process is centred
10.2. What for?
10.2.1. process vs customer expectation (specification)
10.2.1.1. measure is around the "variability"
10.3. So What
10.3.1. is the process good enough to meet customer expectations?
10.3.2. Are we overdoing
10.3.2.1. I.e. is the process is too good compared to customer specification
10.3.2.2. USL<UCL
10.3.2.3. LSL>UCL
10.4. Link to other units
10.4.1. Proportion outside USL, LSL
10.4.1.1. NORM.DIST
10.5. E.g.
10.5.1. P.25 is great example