Subject Code

18MEO113T - Design of Experiments

Handled by
A.C. Arun Raj,
Assistant Professor,
Department of Mechanical Engineering,
SRM IST, Kattankulathur.

Disclaimer
The content prepared in the presentation are from various sources, only used for education
purpose. Thanks to all the sources.
Unit 2

2
 Unit 2

1: Need for DOE methodology.

2: Barriers in the successful application of DOE

3: Practical methodology of DOE and Analytical tools for DOE

4: The confidence interval for the mean response

5: Introduction to Screening design

6: Geometric and non-geometric P-B design

7: Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

8: Solving Case studies on Full factorial design with statistics software

9: Solving case studies on Full factorial design with statistics software

 Need for DOE methodology

Experimental design methods are also of fundamental importance in

engineering design activities, where new products are developed and
existing ones improved. Some applications of experimental design in
engineering design include
1. Evaluation and comparison of basic design configurations

2. Evaluation of material alternatives

3. Selection of design parameters so that the product will work well under a wide variety

of field conditions, that is, so that the product is robust

4. Determination of key product design parameters that impact product performance

5. Formulation of new products.

4
 Need for DOE methodology

• The use of experimental design in product realization can result in products

that are easier to manufacture and that have enhanced field performance
and reliability, lower product cost, and shorter product design and
development time.

• Designed experiments also have extensive applications in marketing, market

research, transactional and service operations, and general business
operations.

5
 Need for DOE methodology

• Examples for need

– Characterizing a process

– Optimizing a process

– Designing a product

– Formulating a product

– Designing a web page

6
 Need for DOE methodology

• Characterizing a process

7
 Need for DOE methodology

• Optimizing a process

8
 Need for DOE methodology

• Optimizing a process

9
 Need for DOE methodology

• Designing a Product

10
 Need for DOE methodology

• Designing a Product

11
 Need for DOE methodology

• Formulating a Product

12
 Need for DOE methodology

• Designing a Web Page

13
 Need for DOE methodology

• Designing a Web Page

14
 Barriers in the Successful Application of DOE

The ‘effective’ application of DOE by industrial engineers is limited in many

manufacturing organisations.

Some noticeable barriers are as follows:

• Educational barriers

• Management barriers
• Cultural barriers

• Communication barriers

• Other barriers

15
 Barriers in the Successful Application of DOE

• Educational barriers
– The word ‘statistics’ invokes fear in many industrial engineers.

– The fundamental problem begins with the current statistical education for the engineering
community in their academic curriculum. The courses currently available in ‘engineering
statistics’ often tend to concentrate on the theory of probability, probability distributions and
more mathematical aspects of the subject, rather than practically useful techniques such as
DOE, Taguchi method, robust design, gauge capability studies, Statistical Process Control
(SPC), etc.

16
 Barriers in the Successful Application of DOE

• Management barriers
– Managers often don’t understand the importance of DOE in problem solving or don’t
appreciate the competitive value it brings into the organisation. In many organisations,
managers encourage their engineers to use the so-called ‘home-grown’ solutions for
process and quality-related problems.

– These ‘home-grown’ solutions are consistent with the OVAT approach to experimentation,
as managers are always after quick-fix solutions which yield short-term benefits to their
organisations.

17
 Barriers in the Successful Application of DOE

• Management barriers
– Responses from managers with high resistance to change may include the following
• DOE tells me what I already know.

• It sounds good, but it is not applicable to my job.

• I need to make additional effort to prove what I already know.

– Many managers do not instinctively think statistically, mainly because they are not
convinced that statistical thinking adds any value to management and decision-making.

– Managers in organisations believe that DOE is very demanding of resources.

18
 Barriers in the Successful Application of DOE

• Cultural barriers
– Cultural barriers are one of the principal reasons why DOE is not commonly used in many
organisations.

– Many organisations are not culturally ready for the introduction and implementation of
advanced quality improvement techniques such as DOE and Taguchi.

– The best way to overcome this barrier is through intensive training programs and by
demonstrating the successful application of such techniques by other organisations during
the training.

19
 Barriers in the Successful Application of DOE

• Communication barriers
– Research has indicated that there is very little communication between the academic and
industrial worlds.

– Moreover, the communication among industrial engineers, managers and statisticians in

many organisations is limited.

– For the successful initiative of any quality improvement programme, these communities
should work together and make this barrier less formidable.

20
 Barriers in the Successful Application of DOE

• Other barriers
– Negative experiences with DOE may make companies reluctant to use DOE again. The
majority of negative DOE experiences can be classified into two groups. The first relates to
technical issues and the second to non-technical issues.

• Technical issues include:

– choosing unreasonably large or small designs;

– inadequate or even poor measurement of quality characteristics;

– not choosing the appropriate levels for the process variables, etc. Non-linearity or curvature effects of process variables
should be explored to determine the best operating process conditions;

– assessing the impact of ‘uncontrolled variables’ which can influence the output of the process. Experimenters should try to
understand how the ‘uncontrolled variables’ influence the process behaviour and devise strategies to minimise their impact as
much as possible;

– Lacking awareness of assumptions: data analysis, awareness of different alternatives why they are needed, etc.

21
 Barriers in the Successful Application of DOE

• Non-Technical issues include:

– lack of experimental planning;
– executing one-shot experimentation instead of adopting sequential, adaptive and iterative nature of
experimentation
– not choosing the right process variables or design variables for the experiment in the first round of
experimentation, etc.

22
 Barriers in the Successful Application of DOE

• Educational barriers
– Current statistical education for the engineering community in their academic curriculum is not sufficient

• Management barriers
– Managers encourage their engineers to use ‘home-grown’ solutions for process-and quality-related problems.

• Cultural barriers
– Principal reason; reluctant and fear of embracing the DOE

• Communication barriers
– Gap between academia and industry ; lack of knowledge in engineering and statistics

• Other barriers
– Technical and non-technical issues might bring negative experiences with DOE; makes engineers reluctant
to use DOE

23
 Barriers in the Successful Application of DOE

• Commercial software systems and expert systems in DOE provide no

guidance whatsoever in classifying and analysing manufacturing process
quality-related problems from which a suitable approach (Taguchi, Classical
or Shainin’s approach) can be selected.

24
 Barriers in the Successful Application of DOE

• The selection of a particular approach to experimentation (i.e. Taguchi,

Classical or Shainin) is dependent upon a number of criteria:
– the complexity involved

– the degree of optimisation required by the experimenter

– the time required for completion of the experiment

– cost issues associated with the experiment

– the allowed response time to report back to management, etc.

25
 A Practical Methodology for DOE

The methodology of DOE is fundamentally divided into four phases.

These are:

1. planning phase

2. designing phase
3. conducting phase

4. analysing phase.

26
 Planning Phase
• Many engineers pay special attention on the statistical details of DOE and
very little attention to the non-statistical details.

• Experimental studies may fail not only as a result of lack of technical

knowledge of the process under study or wrong use of statistical techniques
but also due to lack of planning.

• It is the responsibility of the senior management team in the organisation to

create an environment that stimulates a culture of using experimental design
techniques for process optimisation problems, product and process
development projects, improving process capability through systematically
reducing excessive variation in processes, etc.
27
 Planning Phase

The planning phase is made up of the following steps.

• Problem Recognition and Formulation

• Selection of Response or Quality Characteristic

• Selection of Process Variables or Design Parameters

• Classification of Process Variables

• Determining the Levels of Process Variables

• List All the Interactions of Interest

28
 Planning Phase

1. Problem Recognition and Formulation

• A clear and succinct statement of the problem can create a better
understanding of what needs to be done. The statement should contain an
objective that is specific, measurable and which can yield practical value
to the company (Kumar and Tobin, 1990). The creation of a
multidisciplinary team in order to have a shared understanding of the
problem is critical in the planning phase. The multidisciplinary team should
be led by someone with good knowledge of the process (a DOE specialist),
good communication skills, good interpersonal skills and awareness of team
dynamics.
• Other team members may include process engineers, a quality
engineer/manager, a machine operator, a management representative and
manufacturing/production engineers/managers. Sharing experiences and
individual knowledge is critical to assure a deeper understanding of the
process providing more efficient ways to design experiments (Romeu, 2006).
29
 Planning Phase

1. Problem Recognition and Formulation

• Some manufacturing problems that can be addressed using an experimental
approach include
– development of new products; improvement of existing processes or
products;
– improvement of the process/product performance relative to the needs
and demands of customers;
– reduction of existing process spread, which leads to poor capability.

• The objective of the experiment must be clearly specified and has to be

measurable.
• Objectives can be either short term or long term. A short-term objective
could be to fix a problem related to a high scrap rate.

30
 Planning Phase

2. Selection of Response or Quality Characteristic

• The selection of a suitable response for the experiment is critical to the
success of any industrially designed experiment.

• The response can be variable or attribute in nature.

• Variable responses such as length, thickness, diameter, viscosity,

strength, etc. generally provide more information than attribute responses
such as good/bad, pass/fail or yes/no.

• Moreover, variable characteristics or responses require fewer samples than

attributes to achieve the same level of statistical significance.

31
 Planning Phase
2. Selection of Response or Quality Characteristic
• Experimenters should define the measurement system prior to performing
the experiment in order to understand what to measure, where to measure
and who is doing the measurements, etc. so that various components of
variation (measurement system variability, operator variability, part
variability, etc.) can be evaluated.

• Defining a measurement system, including human resources, equipments

and measurement methods, is a fundamental aspect in planning
experimental studies.

• It is important to ensure that equipment exists and is suitable, accessible

and calibrated. 32
 Planning Phase

2. Selection of Response or Quality Characteristic

• The quality of a measurement system is usually determined by the statistical
properties of the data it generates over a period of time which captures both
long- and short-term variation.

• Experimenters should be aware of the repeatability, reproducibility and

uncertainty of the measurements prior to the execution of industrial
experiments (Launsby and Weese, 1995).

• It is advisable to make sure that the measurement system is capable,

stable, robust and insensitive to environmental changes.

33
 Planning Phase

3. Selection of Process Variables or Design Parameters

• Some possible ways to identify potential process variables are the use of
engineering knowledge of the process, historical data, cause-and-effect
analysis and brainstorming.

• It is a good practice to conduct a screening experiment in the first phase of

any experimental investigation to identify the most important design
parameters or process variables.

34
 Planning Phase

4. Classification of Process Variables

• After identification of process variables , classify them into controllable and
uncontrollable variables

• Control variables are those which can be controlled by a process

engineer/production engineer in a production environment.

• Uncontrollable variables (or noise variables) are those which are difficult or
expensive to control in actual production environments.

35
 Planning Phase

4. Classification of Process Variables

• Variables such as ambient temperature fluctuations, humidity fluctuations,
raw material variations, etc. are examples of noise variables.

• The effect of such nuisance variables can be minimized by the effective

application of DOE principles such as blocking, randomization and
replication.

36
 Planning Phase

5. Determining the Levels of Process Variables

• A level is the value that a process variable holds in an experiment.

• For example, a car’s gas mileage is influenced by such levels as tire

pressure, speed, etc.

• The number of levels depends on the nature of the process variable to be

studied for the experiment and whether or not the chosen process variable
is qualitative (type of catalyst, type of material, etc.) or quantitative
(temperature, speed, pressure, etc.).

37
 Planning Phase

5. Determining the Levels of Process Variables

• For quantitative process variables, two levels are generally required in the
early stages of experimentation.

• However, for qualitative variables, more than two levels may be required. If
a non-linear function is expected by the experimenter, then it is advisable to
study variables at three or more levels.

• This would assist in quantifying the nonlinear (or curvature) effect of the
process variable on the response function.

38
 Planning Phase

6. List all the Interactions of Interest

• Interaction among variables is quite common in industrial experiments.

• In order to effectively interpret the results of the experiment, it is highly desirable to have a
good understanding of the interaction between two process variables (Marilyn, 1993).

• The best way to relate to interaction is to view it as an effect, just like a factor or process
variable effect.
• In the context of DOE, we generally study two-order interactions.

• The number of two-order interactions within an experiment can be easily obtained by using
a simple equation:

Where n is the number of factors.

39
 Planning Phase

6. List all the Interactions of Interest

• For example, if you consider four factors in an experiment, the number of
two order interactions can be equal to six.

• The questions to ask include

– ‘Do we need to study the interactions in the initial phase of experimentation?’

– ‘How many two-order interactions are of interest to the experimenter?’

• The size of the experiment is dependent on the number of factors to be

studied and the number of interactions, which are of great concern to the
experimenter.

40
 Designing Phase

• During the design stage, it is quite important to consider the confounding

structure and resolution of the design.
– Confounding
• A concept that basically means that multiple effects are tied together into one parent effect and cannot be separted.
For example,

• Two people fipping two different coins would result in the effect of the person and the effect of the coin to be
confounded.

• It is good practice to have the design matrix ready for the team prior to
executing the experiment.

41
 Designing Phase

• The design matrix generally reveals all the settings of factors at different
levels and the order of running a particular experiment.

• Experimenters are advised to carefully consider the three principles

(randomisation, replication and blocking) of experimental design prior to
conducting the real experiment.

42
 Conducting Phase

• This is the phase in which the planned experiment is carried out and the
results are evaluated.

• Several considerations are recognized as being recommended prior to

executing an experiment, such as
– selection of a suitable location for carrying out the experiment. It is important to ensure that
the location is not affected by any external sources of noise (vibration, humidity, etc.);

– availability of materials/parts, operators, machines, etc. required for carrying out the
experiment;

43
 Conducting Phase

• The following steps may be useful while performing the experiment in order
to ensure that it is performed according to the prepared experimental design
matrix (or layout).
– The person responsible for the experiment should be present throughout the experiment. In
order to reduce the operator-to-operator variability, it is best to use the same operator for
the entire experiment.

– Monitor the experimental trials. This is to find any discrepancies while running the
experiment. It is advisable to stop running the experiment if any discrepancies are found.

– Record the observed response values on the prepared data sheet or directly into the
computer.

– Any experiment deviations and unusual occurrences must be recorded and analysed.

44
 Analysing Phase

• Having performed the experiment, the next phase is to analyze and interpret
the results so that valid and sound conclusions can be derived.

• The following are the objectives to be achieved ,

– Determine the design parameters or process variables that affect the mean process
performance.

– Determine the design parameters or process variables that influence performance

variability.

– Determine the design parameter levels that yield the optimum performance.

– Determine whether further improvement is possible.

45
 Analysing Phase
• Statistical methods should be used to analyze the data.
• Analysis of variance is widely used to test the statistical significance of the effects through F-
Test.

• Confidence interval estimation is also part of the data analysis. empirical models are
developed relating the dependent (response) and independent variables (factors).

• Residual analysis and model adequacy checking are also part of the data analysis
procedure.

• Statistical analysis of data is a must for academic and scientific purpose.

• Graphical analysis and normal probability plot of the effects may be preferred by Industry

46
 Analytical Tools of DOE
Main Effects Plot

• A main effects plot is a plot of the mean response values at each level of a
design parameter or process variable.

• One can use this plot to compare the relative strength of the effects of
various factors.

• The sign and magnitude of a main effect would tell us the following:
– The sign of a main effect tells us of the direction of the effect, that is,
whether the average response value increases or decreases.

– The magnitude tells us of the strength of the effect.

47
 Analytical Tools of DOE
Main Effects Plot

• If the effect of a design or process parameter is positive, it implies that the

average response is higher at a high level rather than a low level of the
parameter setting.

• In contrast, if the effect is negative, it means that the average response at

the low-level setting of the parameter is more than at the high level.

48
 Analytical Tools of DOE
Main Effects Plot

Effect of a process

49
 Analytical Tools of DOE
Interactions Plots
– An interactions plot is a powerful graphical tool which plots the mean
response of two factors at all possible combinations of their settings.

– If the lines are parallel, this indicates that there is no interaction between
the factors.
– Non-parallel lines are an indication of the presence of interaction
between the factors.

50
 Analytical Tools of DOE
Interactions Plots

Synergistic Interaction : lines nearly Antagonistic Interaction : lines intersect and

parallel – negligible or no interaction cross each other– strong interaction exists
exists

51
 Analytical Tools of DOE
Cube Plots
– Cube plots display the average response values at all combinations of
process or design parameter settings.
– One can easily determine the best and worst combinations of factor
levels for achieving the desired optimum response.
– A cube plot is useful to determine the path of steepest ascent or descent
for optimisation problems.

52
 Analytical Tools of DOE
Cube Plots
– Figure below illustrates an example of a cube plot for a cutting tool life optimisation study with
three tool parameters: cutting speed, tool geometry and cutting angle.
– The graph indicates that tool life increases when cutting speed is set at low level and cutting angle
and tool geometry are set at high levels.
– The worst condition occurs when all factors are set at low levels.

53
Analytical Tools of DOE

54
 Analytical Tools of DOE
Pareto Plot of Factor Effects
– To detect the factor and interaction effects that are most important to
the process or design optimisation
– It displays the absolute values of the effects, and draws a reference
line on the chart.
– Any effect that extends past this reference line is potentially important.

55
 Analytical Tools of DOE
Pareto Plot of Factor Effects
– The graph shows that factors B and C
and interaction AC are most important.
– It is always a good practice to check the
findings from a Pareto chart with
Normal Probability Plot (NPP) of the
estimates of the effects

- 80:20 principle.

56
 Analytical Tools of DOE
Pareto Plot of Factor Effects
• Taste of the Food
– Let us take an example, where we need
• Quality of the food
to prepare a chart of feedback analysis
• Price
for XYZ restaurant, as per the reviews
• Presentation
and ratings received from the
customers. Here the customers are
given a checklist of four points based on
which they have to rate the restaurant
out of 10. The four points are:

57
 Analytical Tools of DOE
Pareto Plot of Factor Effects

58
 Analytical Tools of DOE
NPP (Normal Probability Plot) of Factor Effects
– For NPP, the main and interaction effects of factors or process (or design)
parameters should be plotted against cumulative probability (%).
– Inactive main and interaction effects tend to fall roughly along a straight
line, whereas active effects tend to appear as extreme points falling off
each end of the straight line (Benski, 1989).

59
 Analytical Tools of DOE
NPP of Factor Effects
– These active effects are
judged to be statistically
significant.
– The results are absolutely
identical to that of a Pareto plot
of factor/ interaction effects.

60
 Analytical Tools of DOE
NPP of Residuals
– One of the key assumptions for the statistical analysis
of data from industrial experiments is that the data
come from a normal distribution.
– In order to check the data for normality, it is best to
construct an NPP (Normal Probability Plot ) of the
residuals.
– NPPs are useful for evaluating the normality of a data
set, even when there is a fairly small number of
observations.
– Here residual is the mean difference between the
observed value (obtained from the experiment) and
the predicted or fitted value. 61
 Analytical Tools of DOE
NPP of Residuals
– If the residuals fall approximately along a straight line, they are then
normally distributed.
– In contrast, if the residuals do not fall fairly close to a straight line, they
are then not normally distributed and hence the data do not come from a
normal population.
– In the graph , points fall fairly close to a straight line, indicating that the data
are approximately normal.

62
 Analytical Tools of DOE
NPP of Residuals

63
 Analytical Tools of DOE
Response Plots

• Response surface plots(display 3D views) such as contour and surface plots

are useful for establishing desirable response values and operating
conditions.

• In a contour plot, the response surface is viewed as a two- dimensional plane

where all points that have the same response are connected to produce
contour lines of constant responses.

• A surface plot generally displays a three-dimensional view that may provide a

clearer picture of the response.

64
 Analytical Tools of DOE
Response Contour Plots
• In this Figure, the tool life increases with an
increase in cutting angle and a decrease in
cutting speed.
• If the regression model (i.e. first-order model)
contains only the main effects and no interaction
effect, the fitted response surface will be a plane
(i.e. contour lines will be straight).
• If the model contains interaction effects, the
contour lines will be curved and not straight.
• The contours produced by a second-order
model will be elliptical in nature.

65
 Analytical Tools of DOE
Response Surface Plots

Surface plots are diagrams of three-dimensional data. Rather

than showing the individual data points, surface plots show a

functional relationship between a designated dependent

variable (Y), and two independent variables (X and Z). The plot

is a companion plot to the contour plot.

66
 Analytical Tools of DOE
Response Surface Plots
– Moreover, we have used a fitted surface in latter figure to find a direction of
potential improvement for a process.

– A formal way to seek the direction of improvement in process optimisation

problems is called the method of steepest ascent or descent (depending on the
nature of the problem at hand, i.e. whether one needs to maximize or minimize
the response of interest).

67
 Model for Predicting Response Function
• Use of this regression model is to predict the response for different
combinations of process parameters (or design parameters) at their best
levels.

• In order to develop a regression model based on the significant effects (either

main or interaction), the first step is to determine the regression
coefficients.
Regression coefficients are estimates of the
unknown population parameters and describe the relationship between
a predictor variable and the response. In linear regression, coefficients are the
values that multiply the predictor values. Suppose you have the following
regression equation: y = 3X + 5. In this equation, +3 is the coefficient, X is
the predictor, and +5 is the constant.
68
 Model for Predicting Response Function
• Regression coefficients
– The sign of each coefficient indicates the direction of the relationship between a predictor
variable and the response variable.

– A positive sign indicates that as the predictor variable increases, the response variable also
increases.

– A negative sign indicates that as the predictor variable increases, the response variable
decreases.

Example: The coefficient value represents the mean change in the response
given a one unit change in the predictor. For example, if a coefficient is +3, the
mean response value increases by 3 for every one unit change in the predictor.

69
Overview of Linear Models

70
Least Squares Regression

71
Introduction to Regression Analysis

72
Linear Regression Model

73
Simple Linear Regression Model

74
 Simple Linear Regression Model
Dependent variable (Y)

Independent variable (X)

Sum of Squares of Error should be minimal, then only we can say that it is best regression line.
75
Simple Linear Regression Model

Line of best fit: Best fit means that the

sum of the squares of the vertical
distance from each point to the line is
at minimum.

76
Simple Linear Regression Model

77
Regression Model

78
Regression Model

79
 Model for Predicting Response Function
• For factors at 2-levels, the regression coefficients are obtained by
dividing the estimates of effects by 2.

• A regression model for factors at 2-levels is usually of the form

• where β1 and β2 are the regression coefficients and β0 is the

average response in a factorial experiment.

• The term ‘ε’ the random error component which is approximately normal and
independently distributed with mean zero and constant variance σ2.

• The regression coefficient β12 corresponds to the interaction between the process parameters x1
and x2

80
 Model for Predicting Response Function

• For example, the regression model for the cutting tool life optimisation study is
given by

• The response values obtained from above Eq. are called predicted values and the
actual response values obtained from the experiment are called observed values.

81
 Model for Predicting Response Function

• Residuals can be obtained by taking the difference of observed and

predicted (or fitted) values.

• We can predict the cutting tool life for various combinations of these tool
parameters.

• For instance, if all the cutting tool life parameters are kept at low level settings,
the predicted tool life then would be

82
 Model for Predicting Response Function
• The observed value of tool life (refer to cube plot) is 26 h.
• The difference between the observed value and predicted value (i.e. residual) is
− 1.332.

• Similarly, if all the cutting tool life parameters are kept at the optimal condition (i.e.
cutting speed = low, tool geometry = high and cutting angle = high), the predicted
tool life would then be

83
 Model for Predicting Response Function

• Once the statistical analysis is performed on the experimental data, it is important to

verify the results by means of confirmatory experiments or trials.

• The number of confirmatory runs at the optimal settings can vary from 4 to 20
(4 runs if expensive, 20 runs if cheap).

84
 Confidence Interval for the Mean Response

• The statistical confidence interval (CI) (at 99% confidence limit) for the mean
response can be computed using the equation

where
y = mean response obtained from confirmation trials or runs
SD = standard deviation of response obtained from confirmation
trials
n = number of samples (or confirmation runs).

85
 Confidence Interval for the Mean Response

• For the cutting tool life example, five samples were collected from the process at the
optimal condition (i.e. cutting speed = low, tool geometry = high and cutting angle =
high).
• Confirmation trials
Y(bar) = 53.71h SD = 0.654 h

86
 Confidence Interval for the Mean Response

• 99% CI for the mean response is given by:

• As the predicted value (54.384) based on the regression model falls within the
statistical CI, we will consider our model good.

87
 Confidence Interval for the Mean Response

• If the results from the confirmation trials or runs fall outside the statistical CI, possible causes must be
identified. Some of the possible causes may be

– incorrect choice of experimental design for the problem at hand

– improper choice of response(s) for the experiment
– inadequate control of noise factors, which cause excessive variation
– omission of some important process or design parameters in the first rounds of experimentation

– measurement error
– wrong assumptions regarding interactions

– errors in conducting the experiment, etc.

88
 Confidence Interval for the Mean Response

• If the results from the confirmatory trials or runs are within the CI, then improvement action on the
process is recommended.

• The new process or design parameters should be implemented with the involvement of top
management.

• After the solution has been implemented, control charts on the response(s) or key process
parameters should be constructed for constantly monitoring, analyzing, managing and improving the

process performance.

89
 Screening Designs

• In many process development and manufacturing applications, the number of potential process or

design variables or parameters (or factors) is large.

• Screening reduces the number of process or design parameters (or factors) by identifying the

key factors affecting product quality or process performance.

• This reduction allows one to focus process improvement efforts on the few really important factors, or

the ‘vital few’.

90
Types of Factorial Design

91
Full Factorial Design

92
Full Factorial Design

93
Full Factorial Design

94
Fractional Factorial Design

95
96
97
98
99
100
101
Resolution Design

102
Resolution Design

103
104
 Screening Designs

• Screening designs provide an effective way to consider a large number of process or design

parameters (or factors) in a minimum number of experimental runs or trials (i.e. with minimum

resources and budget).

• The purpose of screening designs is to identify and separate out those factors that demand further

investigation.

• For screening designs, experimenters are generally not interested in investigating the nature

of interactions among the factors (Antony, 2002).

105
 P-B design

• Focus on - Screening Designs expounded by R.L. Plackett and J.P. Burman in 1946 – hence the

name Plackett–Burman designs (P–B designs).

• P–B designs are based on Hadamard matrices in which the number of experimental runs or trials is

a multiple of four, i.e. N = 4, 8, 12, 16 and so on, where N is the number of trials/runs (Plackett

and Burmann, 1946).

• P–B designs are suitable for studying up to k = (N−1)/(L−1) factors, where L is the number

of levels and k is the number of factors.

• For instance, using a 12-run experiment, it is possible to study up to 11 process or design parameters

at 2-levels. 106
 P-B design
• One of the interesting properties of P–B designs is that all main effects are estimated with the

same precision. This implies that one does not have to anticipate which factors are most likely to

be important when setting up the study.

• Resolution III Design🡪 No main effects are confounded (aliased) with any other main effect, but

main effects are confounded with two-factor interactions.

107
 P-B design
• The aim of P-B design is to study as many factors as possible in a minimum number of trials and to

identify those that need to be studied in further rounds of experimentation in which interactions can be

more thoroughly assessed.

• Thus, they can determine which factors are important. i.e., the main effects are
interested (not interaction effects)

• However, they have the advantage of requiring fewer experimental runs.

• P-B design is equivalent to fractional factorial design (are multiples of four but are not
powers of two. )

• It uses standard orthogonal arrays.

• Here, k = N-1, there will be no degrees of freedom available for error to be estimated.
108
 P-B Designs

Advantages:
• Limited number of runs to evaluate large number of factors.
• Important main effects can be selected for more in- depth study

109
 Disadvantage of P-B Designs
• Their disadvantage is their complexity.

• As P-B designs can’t be represented as cubes, they are sometimes called

nongeometric designs.

• Assumption required to evaluate up to k = (N − 1) factors in N runs.

• One can study fewer than (N − 1) factors in N runs.

• Geometric P–B designs are resolution III designs and therefore these

designs can be folded over to achieve a design resolution IV.

110
P-B Design vs. Fractional factorial

111
 P-B Designs
• Geometric
– 2N (N = 4, 8, 16, 32, etc.)
– Geometric designs are identical to fractional factorial designs in which one may be able
to study the interactions between factors.

– If the process is suspected to be highly interactive

• Non-geometric
– are multiples of four but are not powers of two. E.g. have runs of 12, 20, 24, 28, etc.

– These designs do not have complete confounding of effects.

– if interactions are of no concern to the experimenter

112
P-B Designs

113
P-B Designs

114
 P-B Designs
• Step 1: Assign the levels for a first factor as per P-B design generator table
• Step 2: Keep the last run/trail at a low level for all factors.
• Step 3: To make levels for consecutive factors follow the cyclic manner as represented.

Generator
vectors

2
 P-B Designs
• Step 1: Assign the levels for a first factor as per P-B design generator table
• Step 2: Keep the last run/trail at a low level for all factors.
• Step 3: To make levels for consecutive factors follow the cyclic manner as represented.
P-B Designs

117
 Ex 1. Paperboard product

• This example is based on the manufacturing process of a

paperboard product.
• The objective of the experiment was to increase the puncture resistance
of this paperboard product.
• The response or quality characteristic of interest to the team
conducting the experiment was the force required to penetrate the
material.

• The objective was to maximize the mean force required to

penetrate the material. 118

Ex 1. Paperboard product

Number of factors: 7
Number of levels: 2
Hence, it is 27 designs.

Experimental design
27 = 128 runs

P-D = 8 runs

119
Ex 1. Paperboard product

120
 Ex 1. Paperboard product
Finding key main effects
• Minitab software – used for plotting
• To identify the key main effects that were most influential on the response (i.e.
force).
Active Effects

121
 Ex 1. Paperboard product
Pareto chart –Substantiation

C
E
B

122
 Ex 1. Paperboard product
Main effects
Conclusion: Main effects C (press roll pressure), E (paste type) and B (amount of
additive) are found to have significant impact on the mean puncture resistance
(i.e. the force required to penetrate the paper board).

123
 Ex 1. Paperboard product
Design Matrix
In order to analyze the factors affecting variability in force, we need to
calculate the SD of observations at each experimental design point.

124
 Ex 1. Paperboard product
Normal plot: Effects on SD
The normal plot indicates that only factor F (cure time) influenced the variation in
the puncture resistance (i.e. force). Further analysis of factor F has revealed that
variability is maximum when cure time is set at high level (i.e. 5 days).

125
 Ex 1. Paperboard product
Conclusion
• Factors C, B and E have a significant impact on process average, whereas
factor F has a significant impact on process variability.
• Other factors such as A, D and G can be set at their economic levels since
they do not appear to influence either the process average or the process
variability.
• The next stage of the experimentation would be to consider the interaction
among the factors and select the optimal settings from the experiment that
yields maximum force with minimum variability. [ Full or fractional factorial with
resolution 4]
126
 Ex 2. Plastic Extrusion Process

• Problem : Affection of porosity of plastic parts.

• Eight process parameters might have some impact on porosity.
• Each factor was studied at 2-levels.
• As the total degrees of freedom for studying eight factors at 2-levels is equal to
8, it was decided to Choose a non-geometric 12-run P–B design with 11
degrees of freedom.
• The extra 3 degrees of freedom can be used to estimate experimental error.

127
 Ex 2. Plastic Extrusion Process

• Design Matrix

128
 Ex 2. Plastic Extrusion Process

• Effects of the experiment

129
 Ex 2. Plastic Extrusion Process

• Main Effects

130
 Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

• It is widely accepted that the most commonly used experimental designs in

manufacturing companies are full and fractional factorial designs at 2-levels

and 3-levels.

• Factorial designs would enable an experimenter to study the joint effect of the

factors (or process/design parameters) on a response.

131
 Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

• A full factorial design consists of all possible factor combinations in a test,

and, most importantly, varies the factors simultaneously rather than one factor

at a time.

• Using this approach, the tester can examine both main effects (effect of the

independent variables on the dependent variable) and interactions (effect of the

interaction between independent variables on the dependent variable)

associated with both categorical and continuous factors.

132
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

133
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

134
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

All the experimental design should be balanced

and orthogonal.

Balanced: in terms of factors

Orthogonal: in terms of interactions.

Factorial Design always balanced and orthogonal.

Also, it should be Unbiased.

135
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

136
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

137
Introduction of full factorial design, Basic concepts of 22, 23 and 2k designs

138
 22 factorial designs

F value = variance of the group means (Mean Square Between) / mean of the within group variances (Mean Squared Error)

139
Example 22 design

140
Example 22 design

141
Example 22 design

142
22 factorial designs

143
F-Distribution Table (alpha = 0.05) for Critical Value

144
F-Distribution Table (alpha = 0.025) for Critical
Value

145
F-Distribution Table (alpha = 0.1) for Critical Value

146
 Example 22 design

• As an example, consider an investigation into the effect of the concentration of the reactant
and the amount of the catalyst on the conversion (yield) in a chemical process. The
objective of the experiment was to determine if adjustments to either of these two factors
would increase the yield. Let the reactant concentration be factor A and let the two levels of
interest be 15 and 25 percent. The catalyst is factor B, with the high level denoting the use
of 2 pounds of the catalyst and the low level denoting the use of only 1 pound. The
experiment is replicated three times, so there are 12 runs.

147
22 factorial designs

148
Example 22 design

149
Example 22 design

150
 22 factorial designs

Example:

SST = [282 +362 +182 +312 +252 +322 +192 +302 + 272 +322 +232
+292] – [80+100+60+90]2/4n

151
Example 22 design

152
 Factor A: F Critical Value (1,8) = 5.3177
Example 22 design Factor B: F Critical Value (1,8) = 5.3177
Factor AB: F Critical Value (1,8) = 5.3177

Factor A: F Value > F- Critical Value, P<0.05

Factor B: F Value > F- Critical Value, P<0.05
Factor AB: F Value < F- Critical Value, P>0.05
If F value is greater than the F- Critical Value, then significant difference exists

153
F-Distribution Table (alpha = 0.05) for Critical Value

154
 Example (2) 22 design
An article in the AT&T Technical Journal describes the application of two-level
factorial designs to integrated circuit manufacturing. A basic processing step in
this industry is to grow an epitaxial layer on polished silicon wafers. The wafers
are mounted on a susceptor and positioned inside a bell jar. Chemical vapors are
introduced through nozzles near the top of the jar. The susceptor is rotated, and heat
is applied.
A deposition time and B arsenic flow rate.
two levels of deposition time are short (-) and long (+)
two levels of arsenic flow rate are 55% (-) and 59%(+)
n=4 replications

155
Example (2) 22 design

156
 Example (2) 22 design

Effect of A, B, AB..? SS of A, B, AB..?

157
 23 factorial design

• The experiment consists of k factors, each factor consists of 2 level (+,-)

• For example k=3;

158
 23 factorial design

• Main and Interaction Effects

159
 23 factorial design

• Main and Interaction Effects

The value in the brackets are

“Contrast”

160
23 factorial design

161
Example: 23 Factorial design

162
Example: 23 Factorial design

163
Thank you

164