Statistical Process Control

SPC

QUALITY PRODUCTIVITY MANAGEMENT CONSULTANCY (QPMC)

STATISTICAL PROCESS CONTROL (SPC)

Synopsis:
Statistical Process Control (SPC) is the technique, adopted by most of the companies; which helps to predict variation, using sample measures by determining the amount of common cause & detect special / assignable causes of variation in the process through control chart. SPC should be used as an operators tools to make quality visible (as for example, through capability indices, control chart, data display, using different types of light / color code, pictorial diagram etc.) for continuous improvement by documenting statistical evidence, in Preventive-based-System throughout the organization. Control chart helps to reflect the performance of the process & monitor output of the process, and give a good indication of whether any problem are likely to be correctable locally or will require action on the system. A machine or process is under statistical control if it demonstrates stability and consistency over a period of time, when all opportunities for special / assignable causes of variation have had a chance to occur. Process capability is an indication of prediction of the level of quality (or a measure of the inherent uniformity) a machine ( or manufacturing process) can produce. Case study on control chart and Process Capability clarify how to take sample and get a clue of the problem systematically.

Purpose
The purpose of this workshop is to provide individuals and teams with the knowledge and skills to X and R statistical process control chart, Process capability indices. The training program covers defining variation, source of variation, review of methods of data gathering, organizing and analyzing and construction of X bar and R chart.

Objectives:
After completion of this workshop, participant will be able to: 1. Explain the importance of thinking. 2. Understand variability. 3. Understand and apply the Control charts. 4. Able to calculate Cp and Cpk 5. Able to take decision on process variation.

Precaution:
The participant manual was designed for use in the context of an instructor-led workshop and as a reference outside the workshop. All graphical presentations are in transparent sheet material. It was not designed as a self-study text. This material should not be photocopied without the permission of author. Also this material should not be shown or given to any Consultant or Consultancy Organization.

Dr. S Bhuniya
2

STATISTICAL PROCESS CONTROL (SPC)

FOUNDATION CONCEPT What is quality?
Quality is fitness for use. - JURAN Quality should be aimed at the needs of the customer at present and future. - DEMING Quality is the totally of features and characteristics of a product or service that bear on its ability to satisfy stated or implied needs of the customer(s). - ISO 9000 Quality is the loss (from function variation and harmful effects) a product causes to society after being shipped, other than any losses caused by its intrinsic functions. - Taguchi

What are the different dimensions of quality?

Dimension of Product / Service 1 2 3 4 5 6 7 8 9 Performance Features Reliability Durability Serviceability Response Aesthetics Reputation Cost Meaning Primary product or service characteristics Added touches, bells and whistles, secondary characteristics Consistency of performance over time Useful life Resolution of problems and complaints Characteristics of human-tohuman interface (timeliness, courtesy, professionalism etc.) Sensory characteristics (e.g. Sound, feel, look etc.) Past performance and other intangibles Competitive cost compared to other competitors similar type of product(s) Measures Product Example Signal-to-noise ratio, power Remote control Mean time to failure Useful life ( with repair) Ease to repair Courtesy of dealer Ok-finished cabinet Public opinion, Advertisement Moderate cost Service Example Time to process customer requests Automatic bill paying Time variability to process requests Keeping pace with industry trends Resolution of errors Courtesy of teller Appearance of bank lobby Past Experience Reasonable processing time

What customer will expect from TV manufacturer?

Customer expects the manufacturer of the TV to do the best possible job of all these things so when we decide to watch the set, it will function the way we expect.

What is Quality Control?

Quality control is the process through which we measure actual quality performance, compare it With a standard, and act on the differences. - Taken from quality Planning and Analysis by Juran

Product Control Vs Process Control

(1) Product Control Keep product characteristics within specification Process Control Identify those process parameter, which are affecting variation in product characteristics; control them on target with minimum variation . Maintain continuous quality and productivity improvement Prevent production of nonconforming units

(2)

Maintain outgoing product

(3) Detect nonconforming units in final stage

Why should we now be concerned with quality?

Customer satisfaction To improve productivity To improve competitive position To stay in business

Variation
The purpose of using capability analysis is to measure / predict the amount of variation. But, in order to do that, we must understand variation. There is always variation in everything we do, even though it may be slight. No two tools are sharpened the same. We would like to completely get rid of variation but we know that cant be done. We have to predict how big it is, whether it is bigger than, what we can tolerate and what to do about if it is too big.

Types of Variation
Natural Variation (Common Causes): Variation comes from inherent variation or from common causes. Assignable Variation (Special Causes): Special cause is something has gone wrong in the process something that isnt normally there. The important thing now are that variation always exists, variation comes from common causes and special causes, common causes variation can be changed only by changing the process, and special cause variation can be eliminated by fixing the process. Process improvement can only take place when the special causes of variation are eliminated and one by one the common causes or variation reduced.

Difference between Common & Assignable Cause

SL Common Cause Inherent Variability & remain constant over time. Example: i) Output from a single established machine Only a change in the system will reduce that variability. Only management has the ability changes.

Assignable Cause
They occur infrequently in an unpredictable manner. Example: Between operator /machines / different batches of raw material. ii) Due to Tool wear / set up. They can be found, measured & eliminated. The machine operator is best able to discover and make to make changes.

1 2

3 4

Sources of Variation
MAN (Training, Skill, Motivation, Communication, Attitude, Experience, Trust etc.) MACHINE (Fixture, Gages, Tooling etc.) MATERIAL (Raw material, Hardness, Coolant etc.) METHOD (Process Docs., Speed, Feed etc.) MEASUREMENT (Measuring instrument, Procedure, Gage R & R etc.) ENVIRONMENT (Cleanliness, Noise, Humidity, Timeliness, Central coolant etc.)

Decision Makers
Gather the facts Organize the facts Interpret the facts Make a decision

Data Gathering
* Types of data: Variable & Attribute * Reasons for collecting Data: Primarily for decision making

100 % Inspection & Sampling

100% inspection is only may be 90% effective. It - Does not ensure good parts. - Can be very fatiguing. - Can be expensive. Is non-value-added work. Sampling is easier, faster, and less expensive and can generate better results.

Types of Data: Variable & Attribute

Variable: Data that can be measured and can be put on a scale Ex. Distance, Length, Dia etc. measuring by Scale, Micrometer, Dial indicator Attribute: Non-measurable data that cannot be put on a scale Ex. Color, Plug Gage (go / no-go)
5

Reason for Collecting Data

* To understand an actual situation: You need correct data to better understand what is happening to a process or method at a certain point in time. Correct data are required to determine if and what action needs to be taken. * For analysis: After data are collected, they are analyzed and information is extracted through the use of statistical methods. Therefore, data should be collected in such a way as to simplify later analysis. These data are used in experiments and problem solving. * Data for process control Acceptance or Rejection of lot: Data are collected and placed in a control chart depends on the relationship between what the process is doing and what is supposed to do.

Purpose of Data gathering

Data gathering means to gather information on a specific topic or problem that you are working on. Data is another word for information. Gathering and organizing data can make better decisions. There are following benefits to data gathering - Improved work flow - Reduction in costs - Prevention oriented - Development of a quality consciousness - Work with facts rather than opinions - Improve decision making We use this data to capability analysis. The purpose of doing capability analysis is to answer these questions. 1. Can we make it right? (Machine / Process Capability) 2. Are we making it right? (Process Control) 3. Have we made it right? (Product Audit) 4. Can we make it better? (Process Improvement)

Organize the facts

Basic Problem Solving Tool
A type of chart to identify problems is a trend chart. It is a simple and basic tool that will show period-to-period picture. The trend chart can also be used to monitor what is happening on certain operations and be used to predict what will happen in the future. This can be used to see if improvements are being made and how much they are affecting the process. Histogram serves many purposes. They are graphic displays of data to show a condition as it currently exists. The histogram is readily and simply understood and therefore, it is often used as a simple and common communication tool. A histogram is one type of Frequency distribution.
6

Construction of Histogram
Step 1: Gather the data or information. Step 2: Rank order and tally data
[ (i) R = max value min value (ii) Choose approx. no. of class intervals between 5 and 15. Use formula k = 1 + 3.222 log 10 N, N= total frequency (iii) C = (R + W)/ k, W = smallest unit of measurement (iv) Lower limit = X min 0.5W ]

Step 3: Prepare the vertical and horizontal axis of the histogram. Step 4: Fill in numeric values on horizontal axis Step 5: Fill in frequency values on vertical axis Step 6: Draw each frequency bar corresponding to each value. Introduction to Basic Statistics To understand how to analyze and use data it must be put into a consumable form. When data is presented graphically, it is more understandable or consumable. The purpose of basic statistics is to take data and put it into a form that makes it easy to understand and easy to communicate. Total for Member 2000 Mean (Average) = ------------------= ------ = 400 Number of Members 5 Median (Middle Value): Median value in a group when arranged from lowest to highest. Mode (Most Occurring Value): Mode is the most of the most frequent occurring value in a group. When the data is arranged into a histogram, the highest bar represents the mode. Range : Difference between the lowest and greatest value. \Standard Deviation: = (X - X)2 /n (Population SD)
2

s = (X - X) /n-1 (Sample SD)

Normal Distribution is bell shaped. Statistical Process Control Statistical means having to do with numbers or more particularly with drawing conclusions from numbers. Process means a set of conditions or causes which work together to produce a given result. Control means to keep something within boundaries or to make something behave the way we want it to behave. Why we use SPC? * Prevention Vs. Detection * To Document Statistical Evidence * A Tool for Operator * To Make Quality Visible * Continuous Improvement Tool

Operator Controllability Three elements are required in order to provide operators an environment of self-control * Knowing what they are supposed to do. * Knowing what they are actually doing. * Having the ability to regulate the process. Quality Visibility * Through Capability Indices (M/C & Process) * Through Control Charts * Through Data Display / Pictorial Diagram * Using Different Type of Light / Color Code How SPC Helps * Using sample measures to predict variation by determining the amount of common cause and helping to detect Special causes of variation in the process. Capability Analysis * Communicate capability requirements to your machine supplier. * Qualify blanks, or raw parts. * Assure gage capability. * Assure proper machine / equipment set-up, tooling, and adjustments. * Begin capability analysis at supplier end. * If acceptable, repeat process in in-house. Conditions for Capability Study * New or rebuilt equipment * On-going capability assessment for existing equipment * Actions to be taken if equipment is not capable: - Work to improve capability of the equipment and repeat capability evaluation. - Change engineering specifications / tolerances to improve capability index to desired value. - Accept equipment as it is, with agreement to employ a documented process control plan which will insure that non-conforming product cannot be passed on the customer. Types of Capability Studies There are different types of studies that determine capability: * Short Term Machine Capability Study (Small sample of consecutive pieces) * Long Term Machine Capability Study (Variation over brief time period - 8 hours) * Process Capability Study (Meeting customer requirements over long time period) Process Capability Capability refers to the quantification of common cause of variation. The presence of special or Assignable cause variation make the prediction of capability impossible and the meaning of a capability index unclear . - Victor E Kane, Process Capability Indices, JQT, Jan 1986

Cpk : what is it?:-Measure the how much process is shifted from the mean line of UCL & LCL A prediction, of the level of quality (or a measure of the inherent uniformity) a machine (or manufacturing process) can produce. Cp: Measure the how much variation in process Statistical Control A machine or process is in control when it has demonstrated stability and consistency over a period of time, when all opportunities for special / assignable causes of variation have had a chance to occur. Control Development Steps 1. Choose a control subject. 2. Choose a unit of measure. 3. Set a standard of goal for the control subject. 4. Choose a sensing device. 5. Measure and record performance. 6. Interpret difference between actual and standard. 7. Take action on difference. Control Chart helps to - Reflects the performance of the process. - Give a good indication of whether any problems are likely to be correctable locally or will require action on the system. - Monitor output and communicates to users. - Separate special causes from common causes. - Determine capability of production process. - Determine process stability. - Predict future output. - Problem solving tool. - Historical document. - Minimize the net economic loss from overadjustment and underadjustment (Dr. Deming). CONTROL CHART (VARIABLE & ATTRIBUTE) Answers questions before choosing a control chart ??? * What characteristic is to be investigated? * What testing devise of gauges will be required? * What is the purpose of the control chart? * Which chart will most effectively accomplish the purpose: A: Variable Control Chart [Moving Range, Average Range ( X & R), Average SD.( X & S), Moving Average, Median Range, Pre-Control] B: Attribute Control Chart [Defective ( p, np), Defect (c,u)] * What sample size should be drawn? (Risk, Sample to sample variation, cost, within sample variation) * How often sample should be drawn? * How sample will be drawn? (Consecutive, Randomly)
9

How to decide which is important characteristic? * Customer Critical * Product Critical * Process Critical * Under Study for Quality Improvement Control Chart Selection
CONTROL CHART SELECTION
Type of Data Variable n Size Defects
n=1

Attribute Defective
n size Equal n size Unequal

n<7 n>1

n>7

X&R

X&s

X & Rm
n size equal
1 defect per chart Several defects per chart

np chart p chart

p chart n size unequal

Several defects per chart

1 defect per chart

3/1/99

DSB

Average & Range chart (X & R) Step 1: Select subgroup period & sample size Step 2: Record the data on the control chart format. Step 3: Calculate Mean (X bar) and Range (R) for each sample. Step 4: Test distribution for Normality Step 5:Calculate the central lines R = R1 + R2 +.. + Rk k X = X1 + X2+....+ Xk
k

Step 6: Calculate the control limits UCL R = D4 R LCLR= D3R UCL X = X+ A2R LCL X = X - A2R Step 7: Select vertical & horizontal axis Step 8: Draw control limits Step 9: Plot ranges and averages Step 10: Analyze the range chart Step 11: Analyze the X chart
10

Step 12: Locate & Correct special causes Step 13 : Estimate Standard deviation. ^ = R /d2 Step 14: Cp = USL - LSL 6 ^ Cpk = min{ USL-X , X-LSL } 3 ^ 3 ^ Step 15 : Comment on Process Behavior. Review for - Points beyond the control limits - Distance of points from the process average - Runs - Trends Cycles Attribute Control Chart Step 1: Select subgroup period & sample size. Step 2: Record each subgroup number of nonconforming units (defective) or nonconformities on the chart. Step 3: Depending upon the type of chart, record each subgroups. a) Proportion of nonconforming units (p); or b) Number of nonconforming units (np); or c) Number of nonconformances (c); or d) Number of nonconformitites per unit (u) Step 4:Calculate the appropriate process average ( p, or np or c or u) Step 5: Draw control limits P np u c Process Average p np u c 2 {p(1-p)/n} {np(1-p)} u/n c UCL p + 3{p(1-p)/n} LCL p 3{p(1-p)/n}

np + 3{np(1-p)} np 3{np(1-p)} u+ 3{u/n} c 3c u 3{u/n} c 3c

Step 6: Select vertical & horizontal scale. Step 7: Draw control limits and process average. Step 8: Plot the subgroup values on the control chart. Step 9: Interpret the chart for process control * Points beyond the control limits * Distance of points from the process average * Runs * Trends * Cycles Step 10: Locate and Correct special causes

11

Control Chart will not Make processes capable or Solve process problems but they can give clues to possible causes. Process Capability & Control If control has been established. Then Capability can be assessed. Data Tool Parameter Evaluation Process Control Average and Ranges ( X bar & R) Control Chart Control Limits Stability of the process over time Capability Individual X Histogram and Normal Curve Specification Limits Ability of the process to produce parts within specification

Process Capability Classification: Process Classification Incapable Marginally Capable Capable Cpk < 1.0 1.0 to 1.33 > 1.33 Evaluation Criteria Defect Rate > 0.3 % 0.3 % to 0.003% < 0.003 %

Product- Process Compatibility Classification: Product Critically Classification Incapable Process Capability Classification Marginal Capable

Critical Major Minor

V IV III

IV III II

II II I

V Not acceptable for production IV - 100 % inspection III SPC II - Frequency Check I - Audit only

12

Process Capability & Process Control NOT CAPABLE Inconsistent shape of distribution process center Inconsistent process spread Exceeds specification limits Not predictable Use SPC Perform experimental design Investigate specifications Change process Consistent shape of distribution OUT OfConsistent process center CONTROL Consistent process spread Exceeds specification limits Predictable Use SPC Perform experimental design Investigate specifications Change process
OUT OfInconsistent CONTROL

CAPABLE
OUT of Inconsistent CONTROL

Inconsistent shape of distribution process center Inconsistent process spread Consistently within spec. limit Not predictable Use SPC

IN Consistent CONTROL

Consistent shape of distribution process center Consistent process spread Consistently within spec. limits Predictable Use SPC

Process Capability & Process Control

OUT OF CONTROL

NOT CAPABLE CAPABLE 4

IN 3 CONTROL

2 1

No. 1 2 3 4

ACTION Continue to monitor process Must reduce special causes Must improve process by reducing common causes Must take action to protect customer

Impact on Customers Low Risk Marginal Risky High Risk

13

SET UP VERIFICATION: This has to be done before starting the mass production in the press shop machine mainly. Run the machine after die, tool and fixture proper adjustment, machine maintenance and setting of machine parameter at their working recommended levels. See the trail sample and try to set the above things until the desired output comes at middle level. Calculate the control limits for special characteristics parameters, if process is stable and under statistical control. For next time onwards, after doing die, tooling, fixture adjustment, machine maintenance and after setting of machine process parameter; collect minimum two trail pieces ; see whether points fall within central one-third control limits. If it is so, set up can be approved for mass production. If the results fall in outer two-third zone of control limit, then a second set of parts should be manufactured, measured. If point falls in the same outer two-third control zone, then setup should be adjusted and this sequence to be repeated; unless consecutive minimum two parts fall within one-third control zone. SIX Sigma

14

Relationship between Cpk and percent Nonconforming Cpk 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.33 1.40 1.50 1.60 1.67 1.70 1.80 1.90 2.00 Sigma Expected Percent Nonconforming 76.4177 54.8506 36.8120 23.0139 13.3614 7.1861 3.5729 1.6395 0.6934 0.2700 9.6685 X 10 -2 3.1822 X 10 -2 9.6193 X 10 3 6.3342 X 10 3 2.6691 X 10 3 6.7953 X 10 4 1.5867 X 10 4 5.7330 X 10 5 3.3965 X 10 5 6.6641 X 10 6 1.1981 X 10 6 1.9732 X 10 7 Expected yield % 23.5823 45.1494 63.1880 76.9861 86.6386 92.8139 96.4271 98.3605 99.3066 99.7300 99.9033 99.9682 99.9904 99.9937 99.9973 99.9993 99.9998 > 99.9999 > 99.9999 > 99.9999 > 99.9999 > 99.9999 Expected no. of parts out of spec 764.2 PPT 548.5 PPT 368.1 PPT 230.1 PPT 133.6 PPT 71.9 PPT 35.7 PPT 16.4 PPT 6.9 PPT 2.7 PPT 966.8 PPM 318.2 PPM 96.2 PPM 63.3 PPM 26.7 PPM 6.8 PPM 1.6 PPM 573.3 PPB 339.7 PPB 66.6 PPB 12.0 PPB 2.0 PPB

3 sigma

4 sigma

5 sigma

6 sigma

Assumptions underlying the construction of this table: The process is in control The characteristic being assessed is on target at the midpoint of the specification limits. The characteristic being assessed is normally distributed

15