You are on page 1of 71

Statistics In Validation

Sampling Plans that Result in


Statistical Confidence
Jacalyn L. Schroeder
Statistician, Validation/Quality
Engineering Specialist

Medical Products

Strong Presence in Many Segments


PURIFICATION

PHARMACEUTICALS

HEALTH INFORMATION SYSTEMS

DENTAL

DRUG DELIVERY
SYSTEMS

PERSONAL CARE

AGENDA
Acceptance Sampling Plan Basics
Sampling Plans to Meet Validation
Objectives
Sampling Plans and Statistics
Throughout Validation Lifecycle
Interactive Exercises (Throughout)
J.L. Schroeder , 3M, March 2014

Acceptance Sampling
Plan Basics

J.L. Schroeder , 3M, March 2014

Sampling Plans
A sampling plan is a
procedure for making
accept/reject decisions:
Take 300 samples from a lot
and accept if there are zero
defects

Includes both a sample size


and acceptance criteria
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Basic Sampling Plan


(For Percent Defective or Proportion Nonconforming)
Take a representative
sample of n units

Count the number


that are defective

Count a

ACCEPT

Compare
count to
accept
number
a

Count > a

REJECT

J.L. Schroeder, 3M, March 2014

N=100

Sampling Plan Selection


1. Determine sampling plan
purpose
2. Determine requirements to inspect for
Free of Red Stripes
Specifications Must Define

3. Determine appropriate statistical


performance property (i.e. percent
defective, defects per quantity, etc)
4. Choose sampling risks and acceptable quality
level and reject performance level
J.L. Schroeder , 3M, March 2014

Sampling Plans
Devised to serve different purposes:

Neubauer Acceptance
Sampling Facts

J.L. Schroeder , 3M, March 2014

OC Curve and Two Points


(AQL, 1-)

(LTPD, )

J.L. Schroeder , 3M, March 2014

Specify a Plan
Two points on OC Curve (AQL, 1-) (LTPD, )
Four values may be used to characterize the
OC curve
AQL = Acceptable Quality Level
RQL or LTPD = Reject Quality Level or Lot
Tolerance Percent Defective
= Producers Risk
Sampling Risks
= Consumer Risk

J.L. Schroeder , 3M, March 2014

Every Sampling Plan Has Its Own OC


Curve
Probability Pass

1
Single: n=30, a=0
Single: n=300, a=0
0.5

Single: n=3000, a=0

0
0

10

Percent Defective
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class
- 2008

J.L. Schroeder, 3M, March 2014

Determine the chances


of passing at different
defective rates.

Every Sampling Plan Has Its Own OC


Curve

0.18

~25%

J.L. Schroeder , 3M, March 2014

Sampling Plans and Risk


Sampling involves risk that the sample will not adequately
represent the entire population

Producers =0.05

Consumers

= 0.05

(AQL, 0.95)

(LTPD, 0.05)

J.L. Schroeder, 3M, March 2014

Sampling Risk

J.L. Schroeder , 3M, March 2014

Summary
Purpose of a sampling plan is to make
accept/reject decisions
Sampling Plans can serve many different
purposes (i.e. consumer/producer
protection, compliance sampling)
Every sampling plan has its own OC curve
that defines protection
OC Curve Defined by Two Points
(AQL, 1-)

(LTPD, )
J.L. Schroeder, 3M, March 2014

Sampling Plans to Meet


Validation Objectives

Sampling Plans for Validation


Accept = Pass validation study
and go into production

Reject = Fail validation study and


must improve the product or
process
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

J.L. Schroeder , 3M, March 2014

Regulations and Sampling Plans


Per 21 CFR 820.250 Sampling plans, when used,
shall be written and based on a valid statistical
rationale
Sampling Plans must appear in writing
Assumptions on which sampling plans are based
shall be verified
We should understand how the sample is taken
(random, stratified, periodic) to represent lot
For validation we want a sample to represent
production
J.L. Schroeder , 3M, March 2014

Representative Sample
Obtaining a representative sample for
validation is more difficult than obtaining a
representative sample in manufacturing
Want sample that is representative of all
, not just one lot:
3 lots
Manufactured under what regulations call
anticipated conditions (material lots,
operators, setups, tooling, )
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Objective of Validation
Demonstrate that the product or process
consistently meets pre-determined
performance requirements
A sampling plan can be used to demonstrate
that a required performance level is met with
certain confidence (AQL, LTPD, alpha, beta)

J.L. Schroeder , 3M, March 2014

What is AQL?
What is routinely accepted

AQL

The AQL (Acceptable Quality Level) of a


sampling plan is a level of quality (percent
defective, defects per hundred units, etc.)
routinely accepted by the sampling plan
Products or processes at or better than the
AQL are passed at least 95% (1-)% of the
time
J.L. Schroeder , 3M, March 2014

What is LTPD (AKA RQL)?


What is routinely rejected

.05 or .10

LTPD

The LTPD (Lot Tolerance Percent Defective) of


a sampling plan is a level of quality (percent
defective, defects per hundred units, etc.)
routinely rejected by the sampling plan
Products or processes at or worse than the
LTPD fail most of the time
Determine confidence statements
J.L. Schroeder , 3M, March 2014

Confidence Statements
Validation sampling plans are designed to
demonstrate that the product or process
meets a specified performance level with
certain confidence:
Confidence Statement General Form:
With 95% confidence the requirement is
met
Confidence Levels must be documented in validation procedures
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Performance Levels [LTPD.05] for PQ


Commonly used in pharmaceutical and
medical device industries for disposables
are:
For
Defect (Health and Safety)
0.065%, 0.1%
For
Defects (Functional)
0.25%, 0.4%, 0.65%, 1.0%
For
Defects (Cosmetic)
2.5%, 4.0%
Performance Levels must be documented in validation procedures
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Summary
A sampling plan can be used to
demonstrate that a required
performance level is met with certain
confidence
Passing the plan n=300, a=0 allows one to
state: With 95% confidence the defect
level is below 1% defective.

PV SOP procedure to define


recommended performance and
confidence levels
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Summary
(AQL, 0.95)

= =0.05

95% Confidence statement


(LTPD, .05)

AQL=.72

LTPD.05=9.1

J.L. Schroeder , 3M, March 2014

Sampling Plans and


Statistics
Throughout the
Validation Lifecycle

General Validation Roadmap

Design/Process
Development
Validation Plan

Installation Qualification
Document (IQ)
Operational Qualification
Document (OQ)

Statistically Designed
Experiments

Performance Qualification
Document (PQ)

Cont. Process
Verification

Design
Qualification
Risk
Assessment
Equipment
Qualification
Design
Verification
- PPU
Process
Validation
Control

Validation Sampling Plans


Steps for sampling plan selection and
execution
1. Determine Protection Required
2. Select Best Sampling Plan
3. Document Sampling Plan and Analysis
Methods in Protocol
4. Execute Sampling Plan given in Protocol
5. Document Results and Conclusion in Final
Report
Resource: Taylor Enterprises
Statistical Procedures For
Medical Device Industry 2013

J.L. Schroeder , 3M, March 2014

Step 1:

Determine Plan Protection


(Document in SOP)
Validation Sampling Risks
Recommend: =0.05, =0.05

Performance Levels at LTPD (LTPD.05)


Performance levels should be commensurate with
the risk associated with the characteristic

J.L. Schroeder , 3M, March 2014

Performance Levels at LTPD.05

Risk
Critical
Major
Minor

Resource: Taylor Enterprises


Statistical Procedures For
Medical Device Industry 2013

OQ
5% (95% Conformance)
10% (90%
Conformance)
20% (80%
Conformance)

PQ
1% (99% Conformance)
3% (97% Conformance)
5% (95% Conformance)

Step 1: Determine Plan Protection

Make Confidence Statement


Make Confidence Statement based on chosen
plan protection
If validation passes, one can be 95% (1-)
confident that process is below the LTPD0.05
(performance level)

J.L. Schroeder , 3M, March 2014

Key Points
In order to pass a validation study
,
the process must be significantly
better than the validation performance level

Two Questions You Should Always Ask:


Answered by LTPD:
If I pass the validation study, what confidence
statement can I make?

Answered by AQL:
If my process is good, what are the chances of
passing the validation study?
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

J.L. Schroeder , 3M, March 2014

Interactive Excercise
The table shows how the sample size depends on the
selected performance level and confidence level
IN CLASS EXERCISE ANSWER THE FOLLOWING:
With 95% confidence the defect rate is below 5% defective. How
many samples free of defects does this require?
With 90% confidence the defect rate is below 1% defective. How
many samples free of defects does this require?

This table is for attribute data, which gives larger sample


sizes. Use of variables sampling plans can result in lower
sample sizes.
J.L. Schroeder , 3M, March 2014

Single Sampling Plans With a=0

Resource:
Dr. Wayne
Taylor
Validation
Sampling
Plan Class 2008

Performance Level

90% Confidence Meet


Performance Level

95% Confidence Meet


Performance Level

20%

n = 10, a = 0

n = 14 , a = 0

10%

n = 22, a= 0

n = 28 , a = 0

6.5%

n = 34, a = 0

n = 45 , a = 0

5%

n = 45 , a = 0

n = 60 , a = 0

4%

n = 56 , a = 0

n = 75 , a = 0

3%

n = 75 , a = 0

n = 100 , a = 0

2.5%

n = 90 , a = 0

n = 120 , a = 0

1.5%

n = 150 , a = 0

n = 200 , a = 0

1%

n = 230 , a = 0

n = 300 , a = 0

0.65%

n = 350 , a = 0

n = 460 , a = 0

0.4%

n = 575 , a = 0

n = 750 , a = 0

0.3%

n = 770, a = 0

n = 1000 , a = 0

0.25%

n = 920 , a = 0

n = 1200 , a = 0

0.15%

n = 1530 , a = 0

n = 2000 , a = 0

0.1%

n = 2300 , a = 0

n = 3000 , a = 0

0.065%

n = 3540 , a = 0

n = 4600 , a = 0

Step 2:

Select Sampling Plans


Choose Attribute or Variables Plans

Digress

Use Validation Sampling Plan Tables


(www.variation.com)
J.L. Schroeder , 3M, March 2014

Attribute Sampling Plans


Choose between single or multiple
sampling plans at same performance
level
Multiple sampling plans were invented
to give questionable populations
another chance in order to better
understand the true proportion
defective
J.L. Schroeder , 3M, March 2014

Double Sampling Plan


(For Percent Defective or Proportion Nonconforming)

START
Take a representative
sample of n1 units and
count number of defectives

Count a1

Compare
count

ACCEPT

Count r1

Take second representative


sample of n2 units and
count the total number of
defectives in both samples

REJECT

Continue To Second Stage


Count a2
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

ACCEPT

Compare
count

Count > a2

REJECT

Interactive Exercise
Which LTPD.05 Plan is Best?
1

Pa

AQL HigherChances of
Passing Are Better
0.5

0
0

Percent Defective

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

Defective/Nonconforming Units
Single: n=300, a=0
Single: n=630, a=2
Double: n1=325, a1=0, r1=3, n2=580, a2=3
J.L. Schroeder , 3M, March 2014

LTPD = 1%
(Attribute 95% Confidence)

Sampling Plan

AQL

LTPD0.05

n=300, a=0

0.02%

1%

n=470, a=1

0.08%

1%

n1=325, a1=0, r1=2, n2=400, a2=2

0.09%

1%

n=630, a=2

0.13%

1%

n1=325, a1=0, r1=3, n2=580, a2=3

0.15%

1%

n=770, a=3

0.18%

1%

n1=325, a1=0, r1=3, n2=720, a2=4

0.18%

1%

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

Variables Sampling Plan


Variables data:
Coat Weight
Seal Strength

Assumptions:
Representative Sample
Measurements are Normally Distributed:
Should formally test for normality
Multiple lanes, multiple positions, may cause
non normal data sublott by position
Transform data to normal or find best non
normal distribution
J.L. Schroeder , 3M, March 2014

Variables Sampling Plan


Use normal distribution to predict the
percentage of units outside of spec
USL

LSL

2.3%
Out of Spec

50

51

52

53

54

Fill Volume (ml)


Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

J.L. Schroeder, 3M, March 2014

55

Estimated Process Capability


Voice of Customer Nearest Spec Ave
Ppk =
=
3
Voice of Pr ocess

Numerator
LSL
Denominator

Target
J.L. Schroeder , 3M, March 2014

USL

Interpreting Ppk
Ppk

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2007

Defect Level)

1.5

Distance From
Spec
4.5

1.33

32 dpm

1.167

3.5

233 dpm

0.14%

0.833

2.5

0.62%

0.78

2.3

0.89%

0.67

2.3%

0.5

1.5

6.7%

J.L. Schroeder, 3M, March 2014

3.4 dpm

Procedure For Variables Sampling Plans


START
Take a representative sample of n units,
measure each unit and calculate
average (X) and standard deviation (S)
Calculate Ppk

Is Ppk Ppk

ACCEPT
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Yes

No

J.L. Schroeder, 3M, March 2014

REJECT

Normality Assumption
Requires separate normality test
Normal probability plot or histogram alone
are not enough.
Data must be in control for future process
predictions.

J.L. Schroeder , 3M, March 2014

Step 2:

Select Sampling Plans


Choose Attribute or Variables Plans
Use Validation Sampling Plan Tables
(www.variation.com)
Select BEST Sampling Plan
Compare samples sizes and AQLs to determine
best choice at a specified performance level

J.L. Schroeder , 3M, March 2014

Step 2:

Select Sampling Plans


Choose Attribute or Variables Plans
Use Validation Sampling Plan Tables
(www.variation.com)

J.L. Schroeder , 3M, March 2014

Resource: Dr. Wayne Taylor


Statistical Procedures for the
Medical Device Industry
2013/2014
www.variation.com

Step 2:

Select Sampling Plans


Choose Attribute or Variables Plans
Use Validation Sampling Plan Tables
(www.variation.com)
Select BEST Sampling Plan
Compare samples sizes and AQLs to determine
best choice at a specified performance level

J.L. Schroeder , 3M, March 2014

Attribute Plan Selection


Choose sample size by
Comparing AQL with expected
performance of process
The lower the process percent defective,
the lower the sample size (n) and accept
number (a) can be

Why is AQL important


Effects chances of passing
Effects overall cost of validation
J.L. Schroeder , 3M, March 2014

Attribute Plan Selection


Why is AQL important
Effects chances of passing (Higher defective, need higher
AQL to pass)
Effects overall cost of validation

Example of LTPD0.05 = 1% Plans


All these plans allow the same confidence statement to be
made
They have different chances of passing (AQL)
Sampling Plan

AQL

LTPD0.05

n=300, a=0

0.017%

1.0%

n=470, a=1

0.076%

1.0%

n=630, a=2

0.13%

1.0%

n=770, a=3

0.18%

1.0%

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

LTPD0.05 = 1% Plans

Pa

Defective/Nonconforming Units
Single: n=300, a=0
Single: n=470, a=1
Single: n=630, a=2
Single: n=770, a=3

0.5

0
0

Percent Defective

All these plans have LTPD0.05 = 1%


Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Variables Plan Selection


Choose sample size by
Comparing AQL with expected performance of process.
AQL estimated from process Ppk during design verification
studies (PPU or OQ)

Example of LTPD0.05 = 1% Plans (1-Sided)


Sampling Plan

AQL

LTPD0.05

n=15, Ppk=1.17

0.00017% (Ppk=1.55)

1% (Ppk=0.78)

n=20, Ppk=1.10

0.0012% (Ppk=1.41)

1% (Ppk=0.78)

n=30, Ppk=1.02

0.008% (Ppk=1.26)

1% (Ppk=0.78)

Sample Size &


Ppk to pass plan

Process Ppk
estimate (accept
95% of time)

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

Ppk value
that will fail

LTPD = 1%
(Variables 1-sided 95% Confidence)
Sampling Plan

AQL

LTPD0.05

n=15, Ppk=1.17

0.00017% (Ppk=1.55)

1% (Ppk=0.78)

n=20, Ppk=1.10

0.0012% (Ppk=1.41)

1% (Ppk=0.78)

n=30, Ppk=1.02

0.008% (Ppk=1.26)

1% (Ppk=0.78)

n=40, Ppk=0.98

0.02% (Ppk=1.18)

1% (Ppk=0.78)

n=50, Ppk=0.95

0.04% (Ppk=1.12)

1% (Ppk=0.78)

n=60, Ppk=0.94

0.05% (Ppk=1.10)

1% (Ppk=0.78)

n=80, Ppk=0.91

0.09% (Ppk=1.04)

1% (Ppk=0.78)

n=100, Ppk=0.89

0.13% (Ppk=1.01)

1% (Ppk=0.78)

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

LTPD = 1%
(Variables 2-sided 95% Confidence)
Sampling Plan

AQL

LTPD0.05

n=15, Ppk=1.17, Pp=1.17

0.00016% (Ppk=1.55)

1% (Ppk=0.78)

n=20, Ppk=1.11, Pp=1.13

0.001% (Ppk=1.42)

1% (Ppk=0.78)

n=30, Ppk=1.03, Pp=1.07

0.007% (Ppk=1.27)

1% (Ppk=0.78)

n=40, Ppk=0.99, Pp=1.04

0.018% (Ppk=1.19)

1% (Ppk=0.78)

n=50, Ppk=0.96, Pp=1.02

0.033% (Ppk=1.13)

1% (Ppk=0.78)

n=60, Ppk=0.95, Pp=1.01

0.044% (Ppk=1.11)

1% (Ppk=0.78)

n=80, Ppk=0.92, Pp=0.99

0.08% (Ppk=1.05)

1% (Ppk=0.78)

n=100, Ppk=0. 90, Pp=0.97

0.11% (Ppk=1.02)

1% (Ppk=0.78)

Resource: Dr. Wayne Taylor


Validation Sampling Plan Class 2008

Interactive Exercise
Suppose:
2-sided
LTPD.05 = 1%
Have data
where
Ppk = 1.06

Sampling Plan

AQL

LTPD0.05

n=15, Ppk=1.17, Pp=1.17

0.00016% (Ppk=1.55)

1% (Ppk=0.78)

n=20, Ppk=1.11, Pp=1.13

0.001% (Ppk=1.42)

1% (Ppk=0.78)

n=30, Ppk=1.03, Pp=1.07

0.007% (Ppk=1.27)

1% (Ppk=0.78)

n=40, Ppk=0.99, Pp=1.04

0.018% (Ppk=1.19)

1% (Ppk=0.78)

n=50, Ppk=0.96, Pp=1.02

0.033% (Ppk=1.13)

1% (Ppk=0.78)

n=60, Ppk=0.95, Pp=1.01

0.044% (Ppk=1.11)

1% (Ppk=0.78)

n=80, Ppk=0.92, Pp=0.99

0.08% (Ppk=1.05)

1% (Ppk=0.78)

n=100, Ppk=0. 90, Pp=0.97

0.11% (Ppk=1.02)

1% (Ppk=0.78)

What sample size should be selected to


provide 95% of passing?
What is corresponding AQL?
Resource: Dr. Wayne Taylor
Validation Sampling Plan Class 2008

Step 3:
Document Sampling Plan
and Analysis Methods in Protocol

Document in Protocol
Test method, products to be tested, and specifications
(variables or attribute data)
Worst case product(s)

Estimated process Ppk or estimated AQL


To determine sample size and acceptance criteria

Risk based selection of Performance Level (LTPD.05)


Corresponding Confidence Statement

Validation lot specifics and instructions for ensuring a


representative sample
Method of Analysis (LCR, Grouped, Sublotted, Statistical
Package)
J.L. Schroeder , 3M, March 2014

Step 3:
Document Sampling Plan
and Analysis Methods in Protocol

Confidence Statement Examples to Meet


Validation Objectives
Per 95% Confidence Sampling Plan for LTPD.05 =
1%, the validation will pass for fill volume if the Ppk
is 1.11 or greater.
If this validation passes one can state with 95%
confidence that the process for controlling fill
volume is less than 1% defective.
J.L. Schroeder , 3M, March 2014

Example

J.L. Schroeder , 3M, March 2014

Interactive Exercise
LTPD = 1%
Attribute Plans with 90% Confidence

Attribute Plans with 95% Confidence

Type

Parameters

AQL

LTPD0.1

Type

Parameters

AQL

LTPD0.05

Single

n=230, a=0

0.02%

1%

Single

n=300, a=0

0.02%

1%

Single

n=390, a=1

0.09%

1%

Single

n=470, a=1

0.08%

1%

Double

n1=250, a1=0, r1=2, n2=400, a2=2

0.11%

1%

Double

n1=325, a1=0, r1=2, n2=400, a2=2

0.09%

1%

Single

n=530, a=2

0.15%

1%

Single

n=630, a=2

0.13%

1%

Double

n1=250, a1=0, r1=3, n2=560, a2=3

0.18%

1%

Double

n1=325, a1=0, r1=3, n2=580, a2=3

0.15%

1%

Single

n=670, a=3

0.20%

1%

Single

n=770, a=3

0.18%

1%

Double

n1=250, a1=0, r1=3, n2=720, a2=4

0.21%

1%

Double

n1=325, a1=0, r1=3, n2=720, a2=4

0.18%

1%

Variables 1-sided 95% Confidence

Variables 2-sided 95% Confidence

Parameters

AQL

LTPD0.05

Parameters

AQL

LTPD0.05

n=15, P pk =1.17

0.00017% (Ppk =1.55)

1% (Ppk =0.78)

n=15, Ppk =1.17, Pp =1.17

0.00016% (Ppk =1.55)

1% (P pk=0.78)

n=20, P pk =1.10

0.0012% (Ppk =1.41)

1% (Ppk =0.78)

n=20, Ppk =1.11, Pp =1.13

0.001% (P pk =1.42)

1% (P pk=0.78)

n=30, P pk =1.02

0.008% (P pk =1.26)

1% (Ppk =0.78)

n=30, Ppk =1.03, Pp =1.07

0.007% (P pk =1.27)

1% (P pk=0.78)

n=40, P pk =0.98

0.02% (P pk =1.18)

1% (Ppk =0.78)

n=40, Ppk =0.99, Pp =1.04

0.018% (P pk =1.19)

1% (P pk=0.78)

n=50, P pk =0.95

0.04% (P pk =1.12)

1% (Ppk =0.78)

n=50, Ppk =0.96, Pp =1.02

0.033% (P pk =1.13)

1% (P pk=0.78)

n=60, P pk =0.94

0.05% (P pk =1.10)

1% (Ppk =0.78)

n=60, Ppk =0.95, Pp =1.01

0.044% (P pk =1.11)

1% (P pk=0.78)

n=80, P pk =0.91

0.09% (P pk =1.04)

1% (Ppk =0.78)

n=80, Ppk =0.92, Pp =0.99

0.08% (P pk =1.05)

1% (P pk=0.78)

n=100, Ppk =0.89

0.13% (P pk =1.01)

1% (Ppk =0.78)

n=100, Ppk =0. 90, Pp=0.97

0.11% (P pk =1.02)

1% (P pk=0.78)

Tables Developed from Validated Sampling Plan Analyzer Version 2.0


Copyright 2001 Taylor Enterprises Inc.
J.L. Schroeder, 3M, March 2014

L C R

Step 4 & 5
Execute Sampling Plan given in Protocol
Document Results and Conclusion in Final Report

EXAMPLE

J.L. Schroeder , 3M, March 2014

PQ Analysis
Process Capability Sixpack of Lot 1 Grouped
Sample M ean

Xbar Chart

Capability Histogram
UCL=1.2199

LSL

Specifications
LSL 1

1.20
_
_
X=1.1483

1.15
1.10

LCL=1.0766
1

11

13

15

17

19

1.035 1.080 1.125 1.170 1.215

Sample Range

R Chart

Normal Prob Plot

1. Is Normality an
issue with the Grouped
Data?

A D : 1.746, P : < 0.005

0.2

UCL=0.1804

0.1

_
R=0.0701

0.0

LCL=0
1

11

13

15

17

19

1.0

Within
StDev 0.04139
Cp
*
Cpk
1.19
PPM
170.42

1.15
1.10
5

10

1.1

1.2

1.3

Capability Plot

Last 20 Subgroups
1.20

V alues

LCR Grouped Analysis


In Minitab

15

20

Within
O verall

Overall
StDev 0.03836
Pp
*
Ppk
1.29
Cpm
*
PPM
55.65

Specs

Sample

2. Is Ppk for Grouped


data Ppk > 0.95?

Distribution
Analyzer at
www.variation.com

J.L. Schroeder, 3M, March 2014

Distribution
Analyzer

PQ Analysis (Steps 4 & 5)


Process Capability Sixpack of Lot 2 Grouped
Xbar Chart

Capability Histogram

Sample M ean

UCL=1.1797

LSL

S pecifications
LS L 1

1.16
_
_
X=1.1235

1.12
1.08

LCL=1.0673
1

11

13

15

17

19

1.00

1.04

Sample Range

R Chart

1.08

1.12

1.16

1.20

Normal Prob Plot

1. Is Normality an
issue with the Grouped
Data?

A D : 0.597, P : 0.116

0.16
UCL=0.1414
0.08

_
R=0.0549

0.00

LCL=0
1

11

13

15

17

19

1.0

Last 20 Subgroups

1.15
1.10
5

10

15

20

Sample

2. Is Ppk for Grouped


data Ppk > 0.95?

Distribution
Analyzer at
www.variation.com

1.1

1.2

Capability Plot
Within
StDev 0.03245
Cp
*
Cpk
1.27
PPM
70.59

1.20

V alues

LCR Grouped Analysis


In Minitab

Within
Ov erall
Specs

Overall
StDev 0.03598
Pp
*
Ppk
1.14
Cpm
*
PPM
299.16

Distribution
Analyzer

PQ Analysis (Steps 4 & 5)


Process Capability Sixpack of Lot 3 Grouped
Sample M ean

Xbar Chart

Capability Histogram
UCL=1.2212

LSL

1.20

S pecifications
LS L 1

_
_
X=1.1293

1.12

1.04

LCL=1.0374
1

11

13

15

17

19

1.00

1.05

1.10

Sample Range

R Chart

1.15

1.20

1.25

Normal Prob Plot


A D : 0.678, P : 0.073

UCL=0.2313
0.2
_
R=0.0898

0.1

0.0

LCL=0
1

11

13

15

17

19

1.0

Last 20 Subgroups
V alues

LCR Grouped Analysis


In Minitab

1.20
1.12
1.04
10

15

20

Sample

2. Is Ppk for Grouped


data Ppk > 0.95?
Investigate
Larger than
expected
Variation.

Distribution
Analyzer at
www.variation.com

1.2

1.3

Capability Plot
Within
StDev 0.05306
Cp
*
Cpk
0.81
PPM
7423.40

1.1

1. Is Normality an
issue with the Grouped
Data?

W ithin
O v erall

Overall
StDev 0.05365
Pp
*
Ppk
0.80
Cpm
*
PPM
7983.88

Specs

Distribution
Analyzer

Step 5
Results: Validation Report
Data
Analysis and Conclusions
Was acceptance criteria met?
Did validation pass?
What confidence statement can be made if
criteria was met?

J.L. Schroeder , 3M, March 2014

Step 5
Results: Validation Report Example

ABC

J.L. Schroeder , 3M, March 2014

XYZ

Questions?

J.L. Schroeder , 3M, March 2014

Contact Information
Jacalyn L. Schroeder 3M Company
Title: Statistician, Process Validation/Quality
Specialist
E-mail: jlschroeder2@mmm.com
Work Phone: 605-696-1355

Dr. Wayne Taylor, Taylor Enterprises


Website: www.variation.com
J.L. Schroeder , 3M, March 2014

References
1. Dr. Wayne A. Taylor, Guide to Acceptance Sampling,
version 1, www.variation.com, Taylor Enterprises,Inc.
Note: Version 2 of new book 2013-2014.
2. GHTF/SG3/N99-10:2004 (Edition 2) , Quality
Management Systems - Process Validation Guidance,
January 2004
3. Acceptance Sampling in Quality Control, 2nd edition,
Schilling & Neubauer
4. Validation Sampling Plan Course, Wayne Taylor, March
2008
5. Acceptance Sampling and Control Class, Neubauer, 2012
6. Standards: ANSI Z1.4, ANSI Z1.9, ISO 2859-1, ISO 39511
7. Schilling/Johnson, Tables for the Construction of
Matched Single, Double, and Multiple Sampling Plans
with application to MIL-STD-105D, Journal Quality
Technology, October, 1980

You might also like