Design of Experiments

 An experiment is a test or series of runs in which purposeful changes are

made to the input variables of a process or system so that we may observe
and identify the reasons for changes that may be observed in the output
response.
 Well-designed experiments can often lead to a model of system
performance; such experimentally determined models are called
empirical models.
 DoE is a structured, organized method that is used to determine the
relationship between the different factors affecting a process and the output
of that process.
Cont..
Advantages:
 Develop Better Products Faster.

 Improve Performance of Process or Product

 Reduce Experimental Effort and Time by 50 to 90%.

 Reduce time to design/develop new products & processes

 Improve performance of existing processes

Cont..
 In general, experiments are used to study the performance of
processes and systems.

FACTORS, LEVELS, RESPONSE

 Factors are input variable to the process
 Factors can be classified as either controllable or uncontrollable variables.
 Dependent variable: the variable we wish to predict or explain
 Independent variable: the variable used to explain the dependent variable
 Levels: Levels represent settings of each factor in the
experiments.
 Response: Response is output parameter or performance
parameter of the experiment
Major Approaches to DOE
 Factorial Design
 Taguchi Method
 Response Surface Design
Cont..
Strategy of experimentation that is used extensively in practice is the one
factor-at-a-time (OFAT) approach. In this study, One-Factor-at-a time
(OFAT) approach has been opted for analysing the effect of various
parameters. This approach involves a set of levels for each parameters and
successively varying each parameter within range, while keeping other
parameters constant.
The major disadvantage of the OFAT strategy is that it fails to consider any
possible interaction between the factors.

 The correct approach to dealing with several factors is to conduct a

factorial experiment. This is an experimental strategy in which factors are
varied together, instead of one at a time.
 The factorial experimental design concept is extremely important.
Guidelines for Designing Experiments
 Recognition and statement of the problem.
 Selection of the response variable
 Choice of factors, levels, and ranges
 Choice of experimental design
 Performing the experiment
 Statistical analysis of the data
 Conclusions and recommendations
 Cont..
 Interaction: Sometimes factors do not behave the same when they are
looked at together as when they are alone; this is called an interaction.
Interaction plot can be used to visualize possible interactions between two or
more factors
 Randomization: is a statistical tool used to minimize potential
uncontrollable biases in the experiment by randomly assigning material, people,
order that experimental trials are conducted, or any other factor not under the
control of the experimenter

 Blocking: is a technique used to increase the precision of an experiment by

breaking the experiment into homogeneous segments (blocks or clusters or
strata) in order to control any potential block to block variability

 Replication: is making multiple experimental runs for each experiment

combination. This is one approach to determining the common cause variation in
the process so that we can test effects for statistical significance
Factorial Design
 Full factorial experiments look completely at all factors
included in the experimentation.
 In full factorials, all of the possible combinations that are
associated with the factors and their levels are studied.
 The effects that the main factors and all the interactions
between factors are measured.
 If we use more than two levels for each factor, we can also
study whether the effect on the response is linear or if
there is curvature in the experimental region for each
factor and for the interactions.
 Full factorial experiments can require many experimental
runs if many factors at many levels are investigated.
 Cont..
 Factorial Design: 22 , 23 , 32 , 33

22 FD= 2 Factors, 2 Levels=4 runs.

23 FD= 3 Factors, 2 Levels=8 runs.
32 FD= 2 factors, 3 Levels=9 runs .
33 FD =3 Factors, 3 Levels=27 runs

Two Levels Full FD:

2 Factors: X1 and X2 (Independent variables)
2 levels : Low and High
Coding : (-1),(+1)
 Three level Full FD:
In three level factorial design,
3 factors: X1 , X2 and X3.
3 levels are use,
1) low(-1)
2) intermediate (0)
3) high (+1)
 Fractional Factorial method:

 Fractional factorials look at more factors with fewer runs.

 Using a fractional factorial involves making a major assumption - that
higher order interactions (those between three or more factors) are
not significant.
 Fractional factorial designs are derived from full factorial matrices by
substituting higher order interactions with new factors.
 To increase the efficiency of experimentation, fractional factorials
give up some power in analyzing the effects on the response.
 Fractional factorials will still look at the main factor effects, but they
lead to compromises when looking into interaction effects.
 This compromise is called confounding
Taguchi
 This gives much reduced " variance " for the experiment
with " optimum settings " of control parameters.
 "Orthogonal Arrays" (OA) provide a set of well balanced
(minimum) experiments serve as objective functions for
optimization.
RSM
 Response Surface Methodology is combination of statistical
and mathematical techniques, which are useful for modelling.
For example, there are some input x1, x2 which influences
the output y. Therefore,

Where, is error or noise observed in response y.

o If the response is well modeled by a linear function of the
independent variables, then the approximating function is the first-
order model
 If there is curvature in the system, then a polynomial of
higher degree must be used, such as the second-order
model
Regression
o The general approach to fitting empirical models is called regression
analysis.
o Regression methods are frequently used to analyze data from
unplanned experiments, such as might arise from observation of
uncontrolled phenomena or historical records.
o As a first approximation, we could try fitting a linear
model to the data.
o This is a multiple linear regression model with two
independent variables. We often call the independent variables
predictor variables or regressors.
Cont..
 where β0 and β1 are unknown parameters to be estimated and
ε is a random error term.The method often used to estimate the
parameters in a model such as this is the method of least
squares.
Cont..
 In general, suppose that there is a single dependent variable
or response y that depends on k independent or regressor
variables, for example, x1, x2, . . . xk. The relationship
between these variables is characterized by a mathematical
model called a regression model.
Regression Analysis
Estimation of the Parameters in Linear Regression Models

 In general, y is an (n * 1) vector of the observations, X is an (n* p) matrix

of the levels of the independent variables, β is a ( p * 1) vector of the
regression coefficients, and ε is an (n * 1) vector of random errors.
Regression Analysis
Estimation of the Parameters in Linear Regression Models
Regression Analysis
 Sixteen observations on the viscosity of a polymer (y) and two process
variables—reaction temperature (x1) and catalyst feed rate (x2)—are shown in
Table. We will fit a multiple linear regression model
Regression Analysis
Solution
The X’X matrix is

and the X’y vector is

Regression Analysis
Solution
The least squares estimate of is

The least squares fit, with the regression coefficients reported to two decimal
places, is
Cont..
 Predict the value of a dependent variable based on the value of at
least one independent variable
 Explain the impact of changes in an independent variable on the
dependent variable

 Dependent variable: the variable we wish to predict or explain

 Independent variable: the variable used to explain the dependent
variable