The Quality Improvement Model Define Process Select Measures Collect & Interpret Data Is Process Stable? Is Process Stable ? This module explains this model step Good point to use the quincunx to demonstrate variability Talk about “expected variability” of the quincunx Get class to understand that individual bead drops are unpredictable in an exact sense, but predictable within a range No Investigate & Fix Special Causes Purpose: Determine the stability of key measures of the product. Yes Is Process Capable ? No Improve Process Capability Yes Use SPC to Maintain Current Process

Types of Variation Common Causes Run Chart Causes that are inherent in the process over time, and affect all outcomes of the process. Ever-present Create small, random fluctuations in the process Lots of them The sum of their effects creates the expected variability Predictable Quality Characteristic Common causes create the “expected variability” Q: “What are the common causes in the quincunx?” - Pins Time

Types of Variation Special Causes Causes that are not present in the process all the time, but arise because of specific circumstances. Not always present in the process Can create large process disturbances, or sustained shifts Relatively few in number Pull the process beyond the expected level of variability Unpredictable Run Chart Quality Characteristic Special causes create additional variability in the process Will try to prevent them from happening again in the future What could be special causes in the quincunx May go back to the cause-effect diagram of the “Going to Work” process and diagnose the various causes as common or special - Some may be classified as both; don’t lead them to think that all special causes are just common causes gone really bad Time Control charts help identify the presence of special causes.

Control Chart Components Run chart of the data Center Line (CL) A line at the average of the data or target of the process Upper Control Limit (UCL) A line at the upper limit of expected variability Lower Control Limit (LCL) A line at the lower limit of expected variability 8 22 24 2 4 6 10 12 14 16 18 20 26 28 30 32 Run Order UCL CL LCL Control Chart Bracket common causes between the limits and special causes outside Points outside the limits indicate the process has changed Limits based on DATA Limits of “expected variability” on quincunx were derived by dropping beads - collected DATA Will not concern ourselves The control limits are based on data collected from the process.

Rules for Separating Common & Special Causes Two commonly used signals of special causes are: Rule 1: Any point above the Upper Control Limit (UCL) or below the Lower Control Limit (LCL) Rule 2: 8 points in a row on the same side of the center line (CL) A common practice is to circle the special cause on the control chart so others will know it has been recognized Additional rules are “beyond the scope of this course” Note: Additional rules do exist.

Process Stability Stable Process Unstable Process A process in which the key measures of the output from the process show no signs of special causes. Variation is a result of common causes only. Unstable Process A process in which the key measures of the output from the process show signs of special causes in addition to common causes. Variation is a result of both common and special causes. Special causes are the difference between stable and unstable processes

Process Stability

A Stable Process • o m n a u s e A l r t W k : B h v i R d M N y c T S f D P

Un-Stable Process L o k F r : • P i n t s O u d e h C l m S f y c R T g a b ( ) I V

Advantages of Stable Processes Are: Process output is predictable (Know the limits of process variation) Customers see consistent product Less time and effort spent firefighting Easier to make (and see) changes to the process Reduces complexity

Polymer Manufacturing Data 1 2 3 4 5 20 40 60 80 100 120 140 Avg=1.4 LCL=0.5 UCL=2.2 b* Sample Control Chart 1 2 3 4 5 b* Histogram LS US Circle the special causes on the control chart Histogram unable to reflect time order needed to assess stability Histogram may look typical, but process be highly unstable Note: b* is a measure of yellowness Histogram does not show whether the process is stable!

Histograms & Control Charts Plot past data Cannot tell if process is stable Only useful for prediction if the process is stable Control Charts Real-time evaluation Help identify presence of special causes Assess past and present stability of process Discuss the need for stability if histogram is to be used for prediction Both tools used in tandem tell you a great deal about what your process is doing and can do

Pump Maintenance Data Circle the special causes Target is 0 Is Process Stable ? Pump Maintenance Data 2 4 6 8 10 12 14 16 18 20 Number of Failures UCL=11.4 Circle the special causes Target is 0 NOTE: No Lower Control Limit Common when you are up against a boundary which is also your target Avg=4.8 LCL=None 2 4 6 8 10 12 14 16 18 20 22 24 Week Are there any signals of special causes? Circle them.