More on Statistical Process Analysis and Control

Document Sample
More on Statistical Process Analysis and Control Powered By Docstoc
					More on Statistical Process
  Analysis and Control
 Engineering Experimental Design
           Winter 2003
          In Today’s Lecture
• Control charts
• Process capability
     What are Control Charts?
• A graphical method of evaluating process
  performance and comparing it to process
  specifications (requirements)
• A graphical method of distinguishing
  between common-cause (random, normal)
  and special-cause variation
• A graphical tool for deciding when you
  should intervene in a process
• A graphical representation of statistics
    Different Control Charts for
     Different Types of Data
• Variable (Continuous)   • Attribute (Discrete)
  Data                      Data
  – X-Bar chart              –   np-chart
  – R chart                  –   p-chart
                             –   c-chart
                             –   u-chart
         Variable or Attribute Data?
•   Assay of each batch made in a plat
•   Average assay of batches produced in a week
•   Number of printing defects on a pallet label
•   Number of typos per sales contract
•   Number of batches off-spec in monthly production
•   Percentage off-spec pounds in monthly production
•   Number of samples rejected per 100 tested
•   Time to close an “account receivable”
   Control Charts in this Course
• They are covered in Chapter 6
• You will construct and x-Bar and an R chart
  in HW 5
• I am planning a control chart project for
  later in the quarter
• No more lecture on control charts now
           Process Capability Ratios
            (Desired Performance) / (Actual Performance)
                            Note that average
                       performance is not centered
                         between the spec limits The shaded areas
 This curve is the                                  represent the
distribution of data                            percentage of off-spec
 from the process                                    production

                       Voice of Customer

                          Voice of Process
• The simplest process capability ratio
• Cp = (USL – LSL) / (6 s)
  – USL = upper spec limit
  – LSL = lower spec limit
  – s = estimate of process standard deviation
• No penalty for being off-target
• Combines variability and location
• Cpk = min((USL-x)/3s, (LSL-x)/3s)
  – x = estimate of process mean
  – s = estimate of process standard deviation
     • Note typo in book. On p. 255, both sigmas in the
       Cpl equation should be wearing hats
              Why Use Cp?
• Sometimes called “process potential”
• Gives you an idea of what the process could
  do, given its current variability, if it were
How is Cpk linked to Six-Sigma?
• Through the denominator, ± 3s
• If x ± 3s (a range of 6s centered on x)
  doesn’t fit inside the SLs, you are not Six-
  Sigma compliant

Shared By: