MM 323 MANUFACTURING SYSTEMS

QUALITY CONTROL (CONTROL CHARTS) IN MANUFACTURING SYSTEMS

Quality Control Systems

Quality Programs for Manufacturing

1. Quality in Design and Manufacturing 2. Traditional and Modern Quality Control

3. Process Variability and Process Capability 4. Statistical Process Control

5. 6. 7. 8. Six Sigma The Six Sigma DMAIC Procedure Taguchi Methods in Quality Engineering ISO 9000

Aspects of DESIGN Quality: Product

Features
---Characteristics that result from design ---Functional and aesthetic features that appeal to the customer

Design configuration, size, weight Function and performance Distinguishing features of the model Aesthetic appeal Ease of use Availability of options Reliability and dependability Durability and long service life Serviceability Reputation of product and producer

Aspects of MANUFACTURING Quality: Freedom

from Deficiencies
---Product does what it is supposed to do ---Product is absent of defects and out-of-tolerance conditions

Absence of defects Conformance to specifications Components within tolerance No missing parts No early failures

Quality Responsibilities
Product features are the aspect of quality for which the design department is responsible Product features determine to a large degree the price that a company can charge for its products Freedom from deficiencies is the quality aspect for which the manufacturing departments are responsible The ability to minimize these deficiencies has an important influence on the cost of the product These are generalities The responsibility for high quality extends well beyond the design and manufacturing departments

Traditional Quality Control

Widespread use of statistical quality control (SQC), in which inferences are made about the quality of the population of manufactured parts and products based on a sample Two principal sampling methods in SQC Control charts Graphical technique used to track measured variable of interest over time Acceptance sampling If the sample passes, the batch is accepted

HOW TO FORM THE Control charts? ----Process Variability

Manufacturing process variations are of two types:

1. Random variations result from intrinsic variability in the

process Process is operating normally Human variations from cycle to cycle, minor variations in starting materials, machine vibration

2. Assignable variations indicate an exception from

normal operating conditions (*SHOULD BE CORRECTED*) Operator errors, defective raw materials, tool failures, equipment malfunctions A PROCESS MUST HAVE ONLY RANDOM VARIATIONS TO BE IN STATISTICAL CONTROL.

WHY DO WE COLLECT SAMPLES AND CALCULATE ITS MEAN AND USE IT IN CONTROL CHARTS?
THE PROCESS (AND A SINGLE MEASUREMENT TAKEN FROM THE PROCESS) MAY HAVE ANY DISTRIBUTION TYPE BUT THE SAMPLE MEAN (LARGER THE SAMPLE SIZE IT IS BETTER) IS NORMALLY DISTRIBUTED. SO WE FORM A

CONTROL CHART FOR THE SAMPLE AVERAGE. ALSO WE FORM A

CONTROL CHART FOR THE VARIATION IN THE SAMPLE.

--WITH CONTROL CHARTS WE KEEP TRACK OF WHETHER THERE IS A STATISTICALLY SIGNIFICANT SHIFT IN THE MEAN AND THE VARIATION OF THE PROCESS.

WHAT IS A Control Chart?

A graphical technique in which statistics computed from measured values of a process characteristic are plotted over time to determine if the process remains in statistical control

Control Chart

UCL = + 3 LCL= - 3 (Center Line) = process mean = standard deviation of the process Assumptions: -Output is normally distributed -Steady state operation -Process is in statistical control

Quality based on Normal Distribution ----WHY 3 ?

Traditional metric for good process quality is 3 Includes 99.73% of population Defect rate = 2700 defects per million WE ARE 99.73% (ALMOST 100%) SURE THAT A SAMPLE MEAN OR THE VARIATION OF THE SAMPLE WILL FALL BETWEEN THE UPPER CONTROL LIMIT LIMIT AND LOWER CONTROL LIMIT. SO WHEN A SAMPLE MEAN FALLS OUT OF THIS RANGE WE CAN CONCLUDE THAT THE PROCESS IS NOT IN STATISTICALLY CONTROL ANYMORE. THERE IS AN ASSIGNABLE VARIATION WHICH SHOULD BE CORRECTED.

3 in the Normal Distribution

Normal distribution of process output variable, showing the 3 limits

Two Basic Types of Control Charts

1. Control charts for variables Require a measurement of the quality characteristic of interest Two principle types: (1) X-bar chart and (2) R chart 2. Control charts for attributes Require a determination of either fraction of defects in the sample or number of defects in the sample Two principle types: (1) p chart and (2) c chart

X-Bar Chart

R Chart

Interpreting the Control Charts

Most obvious sign is when sample mean or range are outside UCL or LCL Less obvious signs: Trends or cyclical patterns in the data Sudden changes in average values Points consistently near UCL or LCL Eight consecutive points that lie on one side of CL Six consecutive points in which each point is higher (or lower) than its predecessor

Control Charts
Look at each diagram and determine the

variation type.
Random Variations Assignable Variations (Special Causes)

Size

Diagram A
Upper Limit

Mean
Lower Limit

Sample

Size

Diagram B
Upper Limit

Size

Diagram C
Upper Limit

Mean

Mean
Lower Limit

Lower Limit

Sample

Sample

Process is in statistical control

Size
Diagram C
Upper Limit

Mean
Lower Limit

Sample
Most points are close to the mean and few near to the limits Comparable numbers of points are above and below the mean No pattern in the data exists (e.g., no trend up or down)

Control Chart used as Feedback Loop in Statistical Process Control

HOW TO IMPROVE QUALITY? ----Cause-and-Effect Diagram --------- can be used to identify which causes are most consequential and how
to take corrective action against them

Process Capability

A PROCESS CAN PRODUCE PARTS WITH 3 AROUND THE MEAN. SO 3 IS ALSO CALLED PROCESS CAPABILITY (PC).

Process Capability and Tolerances

Natural tolerance limits when tolerance is set = process capability Process capability index

UTL LTL PCI = 6

where PCI = process capability index, UTL and LTL = upper and lower tolerance limits, and 6 = range of natural tolerance limits