You are on page 1of 69

Outlines

Experimental
Introduction Design
01 02
Case Study 1 Case study 2
03 04
Introduction
1
What is
DOE?

DOE is a powerful
statistical technique for
improving product/process
designs and solving
process / production
problems
DOE makes controlled
variable changes to input
variables in order to gain
maximum amounts of
information on cause and
effect relationships with a
minimum sample size
Components of
Experimental Design
Factors Levels Response
(Variables, inputs) (Settings) (Outcomes)
Purpose of DOE?

● Achieving an Optimal
Process Output
● Comparing What are the necessary factors,
Alternatives and what are the levels of those
factors, to achieve the exact taste
we might want to compare
and consistency of Mom’s cake?
the results from two different
types of flour ● Minimizing, Maximizing, or
Targeting an Output
● Identifying the (Response)
Significant Inputs How can the cake be made as moist
(Factors) Affecting an as possible?
Output (Response
Why DOE?

● Improve reliability and


performance of products
● Reduce time to ● Achieve product & process
design/develop new robustness
products & ● Perform evaluation of
processes materials, design
alternatives, setting
● Improve component & system
performance of tolerances
existing processes
DOE Software Design Expert
www.statease.com

Features Advantage
This EASY TO USE software It has powerful graphical tools
has all the major experimental and it has been featured in
designs (general ANOVA, two- several well-known texts on
level full and fractional DOE.
factorials, three-level factorials, Apparently it has been WIDELY
several RSM designs, mixture USED IN INDUSTRY.
designs, and much more).
Experimental
Design
2
Building an Experimental Design

1 Define the 2 Define The design


Objective Variables/factors
Eg: Screening the important which will be manipulated
input, optimization? during the experiment and
their levels
3 Define The 4
Design the experiment
response
will be measured to Choose among the available
describe the outcome of the standard designs that is
experimental run compatible with the objective,
number of design variables and
has a reasonable cost
Building an Experimental Design
3
Case study 1
“Optimization of
quercetin extraction
Case method in
Dendrobium
study 1 officinale by
response surface
methodology”, Zhu
et al. (2019),
Heliyon, 5, e02374
STEP 1: DEFINE THE
OBJECTIVE of the
investigation
Objective 1
To screen the factors that significantly
affecting the quercetin extraction
Objective 2
To optimize the extraction process of
quercetin
Step 2: DEFINE THE DESIGN
VARIABLES

Objective 1: screening Objective 2: optimization


5 variables/ factors need to be 3 variables/ factors need to be
screened: optimized:
ethanol concentration (50-90%) ethanol concentration (50-90%)

solid-to-liquid ratio (40-60%) solid-to-liquid ratio (40-60%)

ultrasound power (140-160 W) ultrasound power (140-160 W)

time (30-60 min)


temperature (50-70⁰C)
Step 3: Define The response
For objective 1 (screening)
Variables/Factors

Ethanol What is the response for


Power objective 1?
concentration

Quercetin
content
Time

solid-to-
liquid
Temperature
For Objective 2 (optimization) Step 3: Define The response

Variables/factors
What is the response for
objective 2?
Ethanol
conc

Quercetin content S/L


Power
ratio
Step 4: Choose the design

Widely used design for screening are:

2-level full 2-level Fractional


factorial design factorial design
Step 4: Choose the design

Widely used design for optimization are:

Central Composite Response surface


Design Box-Behnken methodology
Start the program and click New Design
Stay with the Factorial choice, which comes up by default. You’ll be using the default
selection: Randomized Regular Two-Level Factorial.

Awesome
Words
Let’s get on with the case at hand – a full-factorial design. Click the white
square labeled 25 in column 5 (number of factors) in the Runs row labeled 32

Selecting a full,
two-level design
on FIVE factors
which produces
32 runs
Click the Next button. You can now enter the names, units of
measure, and levels for your experimental factors. Enter for each
factor (A, B, C and D) the Name, Units, Low and High levels shown on
the screen shot below .
Now click Next to bring up the Responses dialog box. In this case we
only need to enter a single response name (Quercetin content) and
units (ug/g) as shown below.

Response values entered!


Now click
Next and
Finish.

Next, key
in the
Response
Data
Design-Expert provides two methods of displaying the levels of the factors in a
design:
i. Actual levels of the factors.
ii. Coded as -1 for low levels and +1 for high levels.

To view the
design in
coded values,
click Display
Options on the
menu bar and
select Process
Factors -
Coded
How to
analyse?
The analysis…
Which variables show
significant effect?

Mass Diameter Surface


Mercury 0.06 0.38 0.38

Mars 0.11 0.53 0.38

Saturn 95.2 9.4 1.16

Venus 0.53 0.11 0.06


To really see the magnitude of the chosen
effects, Design-Expert displays them on an
ordered bar chart called the Pareto plot.
What is the Analysis of variance (ANOVA)
significance
of ANOVA?

P-values less than


0.0500 indicate
model terms are
significant.
In this case A, D, E,
AC, BE, CD, CE are
significant model
terms.
Analysis of variance (ANOVA)

Ideally the normal plot of


residuals is a straight line,
indicating no
abnormalities.

A good rule of thumb is


called the “fat pencil” test.
If you can put a fat pencil
over the line and cover up
all the data points, the
data is sufficiently normal
Analysis of variance (ANOVA)

The size of the residual


should be independent of
its predicted value (the
vertical spread of the
residuals should be
approximately the
same across all levels
of the predicted
values).

In this case, the plot looks


OK.
Examine Main Effects and
Any Interactions

Click the Model Graphs


tab
Examine Main Effects and
Any Interactions

Choose “term” to display


Examine Main Effects and
Any Interactions

Choose “term” to display


Main Effects (A, B, C, D, E)

Notice that A, D, E show


significant effect on
quercetin content.
2 Factor interactions

Rules of thumb: check the unparallel lines!

Unparallel
lines
indicate
interaction
between 2
factors
2 Factor interactions
2 Factor interactions
2 Factor interactions
2 Factor interactions
CONCLUSION (Screening)
3 significance factors are:

Ethanol Power NOW, YOU CAN


concentration
OPTIMIZE THE
EXTRACTION
PROCESS!
S/L ratio
Objective 2:
OPTIMIZATION
Objective 2:
OPTIMIZATION

Recall the objective 2:

To optimize the extraction process of quercetin


Let’s do
it!
Choose the design
and Key in the
factors and levels!
Key in the response
Key in the response data
The analysis…

click the Fit Summary tab


ANOVA

The model is
significant!
A, B, C, AB,
AC, A², B²,
C² are
significant
model
terms
ANOVA
R2>=0.94,
the model is 94% fitted to
the experimental data.
ANOVA

Model
equation
developed!
Diagnostics
Contour Plot

The shape of the


contour plot
indicates
whether the mutual
interaction
between the
variables are
significant or not
Contour Plot

Rule of thumb:

i. Circular form:
Interaction is
negligible/not
significant
ii. Elliptical form:
Interaction is
significant
3D Plot

Visual
interpretation of
the interaction
between 2 factors &
facilitated the
location of
optimum
experimental
conditions
Optimization

Click optimization
->numerical
Optimization

Set the goal of


optimization !
Optimization
Optimization
Optimization
Optimization

Click “solutions”

Target: Highest
Optimum points quercetin content
Optimization

needs to be
validated by
experiment!

Optimum condition calculated by the software


CONCLUSION (optimization)
Optimum conditions found at:
Ethanol Power =
concentration 156.8 W
= 81.67%

S/L ratio =41.39 %

that produce 2.58 ug/g quercetin.


4
Case study 2
“Optimization of Medium
Composition for the
Production of
Exopolysaccharides from
Case Phellinus Baumii Pilát in
study 2 Submerged Culture and
the Immuno-stimulating
activity of
Exopolysaccharides”, Luo
et al. (2009), Carbohydrate
Polymers, 78, 409-415

You might also like