(Terzioğlu, 2020)
the Taguchi Method can be effectively used to aid in designing experiments and find the optimum
conditions by adopting Orthogonal Array, so time and effort can be reduced.
Some challenges will be encountered in experimental design when the number of parameters is large
(Yusoff et al., 2011) and the investigation will be time-consuming once all the experiments are
performed. Moreover, it is difficult to identify the influence of each parameter on physical phenomena
and interpret the results. The Taguchi method can provide a systematic approach for simplifying the
experimental design. Meanwhile, the Taguchi method has been widely employed to find the optimum
conditions of a system. However, to the authors’ knowledge, the Taguchi method has not been utilized
in the experiment design and in optimization of a TEG system yet.
The Taguchi method is used because it is a powerful tool to understand the sensitivity of factors on
physical behavior. Analysis of variance (ANOVA) was used to interpret the experimental results and
decide which level of the factor should be used. The significant factors with their respective levels were
identified and set to the optimum values. Eventually, the performance of the TEG system was maximized
using the Taguchi optimization method.
In fact, taking into account the physical test conditions, there are many factors and interactions that
affect the amount of power generated by the thermoelectric generator. The experimental study which is
conducted to reveal the effects of these factors on the system should be minimized in terms of time and
cost. In this study, the Taguchi experimental design was used to reduce the number of trials to be done,
reduce the cost and achieve the optimum result in a shorter time.
In this study, unlike the studies given above, with quite fewer experimental study steps, the optimum
levels of much more factors and levels were determined. The experimental design of the study was done
by Taguchi method. This method allows the detection and optimization of factor levels that give
optimum performance characteristics by testing the combinations of the orthogonal array instead of
testing each level of the factors in the experiments separately and in combination [33]. Because the
results can be obtained with high accuracy by a lesser number of experiments, it provides a significant
reduction in time and costs in experimental studies and is used by researchers for different applications
in many areas
Terzioğlu, H. (2020). Analysis of effect factors on thermoelectric generator using Taguchi method.
Measurement, 149, 106992.
Yusoff, N., Ramasamy, M., & Yusup, S. (2011). Taguchi's parametric design approach for the selection of
optimization variables in a refrigerated gas plant. Chemical Engineering Research and Design,
89(6), 665-675.
The difference between Classical Taguchi Methods, DOE, and RSM is elaborated in the following table.
Taguchi Methods Classical DOE RSM
· To develop a
· To develop a linear quadratic (2nd order)
(1st order) function to function to describe a
describe a product/process
· To estimate an product/process performance and use the
optimum setting for a performance and use the equation to set the
product/process, in equation to set the product/process at its
terms of its control product/process at its optimum setting.
factors and their design optimum setting. e.g., if y=f(x1,x2,x3),
levels. e.g., if y=f(x1,x2,x3), estimate the equation:
e.g., if y= f(A1/2/3, B1/2/3, estimate the equation: y=a0 +a11x12 +a22x22 +a33x32
C1/2/3), identify y=a0 +a1x1 +a2x2 +a3x3 +a12 +a1x1 +a2x2 +a3x3 +a12x1x2
combination of A,B & C x1x2 +a13x1x3 +a23x2x3 +a123 +a13x1x3 +a23x2x3 +a123x1x2x
at their levels 1/2/3 for x1x2x3 3
an optimum y · Based on the · Based on the
· The objective is objective of design, y is objective of design, y is
always to maximize the set at minima/maxima or set at minima/maxima or
Objecti S/N ratio of the its nominal value. its nominal value.
ve response.
Experi Fully saturated, Full factorial Factorial designs
mental fractional factorial designs. augmented by
Plans designs. These plans center points
The information space incorporate and axial points
for studying interaction information (Central
effects is utilized on space for Composite
studying additional studying two Designs – CCD)
factors within the same factors and three which offer
set of experimental runs factor flexibility of
Dr. Taguchi advocates interactions. sequential
the use of multi-levels (3, These are two experimentation
4 or 5 levels) to identify level designs for OR
the curvature in the estimating the Box-Behnken
response and identify the planar equation Designs for
near-optimum settings of (1st order) for a directly
the factors product/process estimating the
performance. 2nd order
quadratic
function for a
product/process
performance.
In the CCD, the
Taguchi Methods Classical DOE RSM
center point runs
are required to
detect the
presence of any
curvature of the
response
(indicating the
start of the
quadratic region)
and the axial
point runs are
required for
estimating the
coefficients of
the square
terms.
Factors could be both Factors could be Factors must be
Factors continuous or discrete continuous or discrete continuous variables!
Analysi Using the ‘Average Effect Using multiple linear In case of Central
s Plots’ the factors are set regression, the Composite Designs
at their respective levels mathematical model is (CCDs):
where the average-effect developed. Using the If the response
of the S/N ratio is equation the optimum could be
maximum. setting for the described by a
product/process is linear model, the
arrived at. “Method of
Steepest Ascent/
Descent” is
deployed.
On detecting a
presence of
curvature, the
quadratic model
is developed by
conducting the
additional runs
of axial points.
Box & Behnken Designs:
These designs are
deployed when the
product/process
Taguchi Methods Classical DOE RSM
performance is known to
follow a quadratic
equation. The quadratic
equation is directly
estimated using the
regression tool
