Understanding Manufacturing Process Variability

1

• Understanding Manufacturing Process Variability
• IE 285
• November 14, 2002
• Presented by
• Toni L. Doolen

2
Who I am

• Education
• BS in BS in Material Science and Engineering and
  Electrical Engineering at Cornell University
• MS in Manufacturing Systems Engineering at
  Stanford University
• PhD in Industrial Engineering at Oregon State
  University
• Work Experience 13 years at Hewlett Packard Co
• Fab (semiconductor manufacturing) process
  engineer
• Senior member of technical staff working on
  process control implementation for inkjet
  cartridge assembly lines world-wide
• Systems engineering manager responsible for
  hardware and software development associated with
  the design, build and qualification of customized
  automated equipment

3
Who I am

• Research interests include manufacturing systems
  human factors engineering and statistical
  analysis
• Joined OSU Industrial and Manufacturing
  Engineering faculty in June 2001
• Outside of work
• Married with 3 children (ages 9, 7, and 7). My
  husband is a manager at Hewlett Packard
• Golf (whenever I get the chance)
• Reading, cooking, violin (and going to lots of
  soccer games and gymnastic meets ?)
• Active in Society of Women Engineers
  particularly in outreach to encourage girls to
  consider engineering careers

4
Processes, Variation, Measurement
5
Process

• A process is a set of related activities that are
  based on a set of inputs and result in outputs
  that have added value. A process includes
  people, equipment, materials, methods, and
  environment that work together to produce output.
  A process is how we create products and services.

6
A Simple Process Example

• Making spaghetti
• What are the related activities?
• What are some inputs?
• What are the outputs?
• What is the added value of the process?

7
Variation

• In the industrial and business world, no two
  things are ever exactly alikethis is why
  engineers include tolerances in specifications
• Variation exists in products, services, and the
  processes used to create them
• In trying to understand the causes of variation
  and predict the occurrences of variation, it is
  necessary to measure variation

8
A Simple Process Example

• Making spaghetti
• What are some of things that might vary from one
  batch of spaghetti to the next?
• What are some of the reasons for this variation?

9
Processes and Performance Measurement

• "You can't control what you don't measure".
• (Deming, W.E. Out of the Crisis. Cambridge,
  MA MIT, 1986. )
• Without measurement there is no way to know how a
  process is performing therefore there is no way
  to improve it. By measuring the voice of the
  customer and the voice of the process, gaps can
  be identified between the two. This information
  gives us direction in our improvement efforts as
  we begin closing the gap.

10
A Simple Example

• Making spaghetti
• What performance measures are related to making
  spaghetti?
• What is the voice of the customer?
• What is the voice of the process?

11
Process Control
12
Measures of location

• The average and mean are the same quantity. The
  average (mean) of 6, 9, 10, 11, and 13
  is(69101113)/5 9.8
• The median is found by listing data from high to
  low and finding the value that is the middle.The
  median of 6, 9, 10, 11, 13 is 10

13
Variation

• The difference in the reproducibility of a
  particular action. The difference between a
  particular action and the target outcome
• Random variation stable consistent patterns of
  variation over time aka controlled variation,
  natural variation
• Nonrandom variation patterns of variation that
  change over time aka special or assignable cause
  variation

14
Deming on Variation

• If I had to reduce my message to management to
  just a few words, Id say it all had to do with
  variation
• Deming (1982)

15
Measures of variation

• The range of 6, 9, 10, 11, and 13 is the highest
  value minus the lowest value ? 13 - 6 7
• The variance is calculated by looking at the
  average difference between each value and the
  overall average of the data
• (6 9.8)2 (9 9.8)2 (10 9.8)2 (11
  9.8)2 (13 9.8)2
• /(5-1) 6.7
• The standard deviation is the square root of the
  variance?2.59

16
Inspecting in Quality

• A traditional approach in manufacturing is to
  depend on production to make a product that meets
  customer requirements and to use inspection of
  the final product as a gate to make sure the
  customer gets what they want.
• This approach is wasteful since itallows time
  and material to beinvested even when the product
  orservice might not be usable

17
Prevention

• Process control methodologies such as design of
  experiments, control charts, and gage studies
  enable us to study, characterize, optimize, and
  understand our processes and gain confidence that
  our process is producing a product or service
  that will meet the customer requirements

18
Designed Experiments

• In studying existing processes or developing new
  processes, we need to understand the relationship
  between inputs and outputs.
• Statistically designed experiments provide us
  with a structured procedure for obtaining the
  most information possible on these relationships
  with a minimum sized experiment.
• Statistically designed experiments allow us to
  construct experiments to test the relationships
  and relative impact of multiple process
  characteristics with process outputs

19
Control Charts

• The goal of using control charts to monitor key
  attributes of a process is to determine when the
  process is operating in control (only random
  variation) vs. when we need to take action
  because special or assignable cause variation is
  present
• If we can identify and eliminate special causes,
  the process will be in control and we can use
  statistical analyses to predict its behavior so
  we will know whether or not we can meet customer
  requirements

20
Gage Studies
21
Gage Studies

• A gage is a measurement tool or system
• In studying, characterizing, optimizing, and
  controlling our processes, we measure important
  characteristics or attributes about our process
• We use gages to measure these attributes
• Gages may be simple (a ruler) or extremely
  complex (ellipsometer), but all gages share one
  thing variation

22
Measurement System Components

• A measurement system typically includes the
  following components
• An operator
• A reference (often called a standard)
• A procedure
• A gage
• An environment

23
Variability in Measurement Systems

• As with any other manufacturing process, a
  measurement system is subject to variability
  which can be either random or special cause in
  nature.
• Examples of special causes might be
• Untrained operators
• Uncalibrated gages
• Lack of procedures

24
Measurement System Analysis

• The goal of a measurement system analysis is to
  detect and eliminate sources of variation that
  are a result of the system used to measure
  product or process attributes.
• We strive to have measurements systems that are
• Accurate
• Precise
• Capable

25
Definitions

• Accuracy The difference between the observed
  average value of measurements and the true value
• Precision Degree of agreement between individual
  measurements on a specific sample composed of
  repeatability and reproducibility
• Capability The total variability of
  measurements measured using a Precision/Tolerance
  (P/T) ratio

26
Definitions

• Repeatability Variation in measurements obtained
  with one gage when used several times by one
  operator while measuring identical
  characteristics on the same parts.
• Reproducibility Variation in the average of the
  measurements made by different operators using
  the same gage when measuring identical
  characteristics on the same parts.
• P/T Ratio Ratio between the precision and the
  tolerance (specification window) for the
  characteristic being measured

27
Assessing Gage Acceptability

• P/T Ratio less than 10 -- gage is acceptable
• P/T Ratio between 10 and 30 -- gage may be
  acceptable, depending on importance of
  application, cost of gage, cost of repairs, etc.
• P/T Ratio over 30 -- Gage system needs
  improvement or to be replaced

28
Ishikawa Diagrams
29
Sources of Variation

• Materials
• Methods
• Machines
• Measurements
• Humans
• Environment

30
Ishikawa diagrams

• Ishikawa diagrams are also called cause/effect
  diagrams and fishbone diagrams
• This type of diagram can provide a foundation to
  break down a complex process into manageable
  factors. You can then generate ideas for
  potential sources of variation
• The basic diagram looks like a fish skeleton,
  with a main idea forming the backbone and
  connecting ideas forming the smaller bones.

31
Fishbone diagram example
32
Fishbone diagram how-tos (1)

• Clearly define the effect or symptom for which
  the causes must be identified.
• Place the effect or symptom being explored at the
  right, enclosed in a box.
• Draw the central spine as a thick line pointing
  to it from the left.
• Brainstorm to identify the "major categories" of
  possible causes using the 6 sourcesof
  variation

33
Fishbone diagram how-tos (2)

• Place each major category in a box and connect it
  to the central spine.
• Within each major category, ask "Why does this
  happen? Why does this condition exist?"
• Continue to add clauses to each branch until the
  fishbone is completed.
• Once all the bones have been completed, identify
  the likely, actionable root cause.

34
A Simple Process Example

• Making spaghetti
• Create an Ishikawa diagram to identify possible
  sources of variation that may lead to bad
  spaghetti.