Author’s Accepted Manuscript

Performance measurement of plate fin heat

exchanger by exploration: ANN, ANFIS, GA, and
SA

A.K. Gupta, P. Kumar, R.K. Sahoo, A.K. Sahu, S.

K. S.
www.elsevier.com/locate/jcde

PII: S2288-4300(15)30037-3
DOI: http://dx.doi.org/10.1016/j.jcde.2016.07.002
Reference: JCDE65
To appear in: Journal of Computational Design and Engineering
Received date: 27 November 2015
Revised date: 6 July 2016
Accepted date: 8 July 2016
Cite this article as: A.K. Gupta, P. Kumar, R.K. Sahoo, A.K. Sahu and S. K. S.,
Performance measurement of plate fin heat exchanger by exploration: ANN,
ANFIS, GA, and SA, Journal of Computational Design and Engineering,
http://dx.doi.org/10.1016/j.jcde.2016.07.002
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
 Performance measurement of plate fin heat exchanger by exploration: ANN, ANFIS, GA, and
SA
1*
A. K. Gupta, 2P. Kumar, 3R. K. Sahoo, 4 Dr. A. K. Sahu, 5Prof. S. K. S.
1*,2
Ph.D. Research Scholar, 4Assistant Professor, 3,5Professor
1*,2,3,5
Department of Mechanical Engineering, National Institute of Technology, Rourkela-
769008, Odisha, India
4
Department of Mechanical Engineering, J.K.I.E., Bilaspur, Chhattisgarh, India-495001
1
ajaythermal@gmail.com, 2pankajsingh14319@gmail.com, 3 rksahoo@nitrkl.ac.in,
4
anoop17212@gmail.com, 5sarangiskr@nitrkl.ac.in
(Communicating author: 1*ajaythermal@gmail.com)
Abstract:
An experimental work is conducted on counter flow plate fin compact heat exchanger
using offset strip fin under different mass flow rates. The training, testing, and validation set of
data has been collected by conducting experiments. Next, artificial neural network merged with
with Genetic Algorithm (GA) uitized to measure the performance of plate-fin compact heat
exchanger. The main aim of present research is to measure the performance of plate-fin compact
heat exchanger and to provide full explanations. An artificial neural network predicted simulated
data, which verified with experimental data under 10% - 20% error. Then, the authors examined
two well-known global search techniques, simulated annealing and the genetic algorithm. The
proposed genetic algorithm and Simulated Annealing (SA) results have been summarized. The
parameters are impartially important for good results. With the emergence of a new data-driven
modeling technique, Neuro-fuzzy based systems are established in academic and practical
applications. The neuro-fuzzy interference system (ANFIS) has also been examined to undertake
the problem related to plate-fin heat exchanger performance measurement under various
parameters. Moreover, Parallel with ANFIS model and Artificial Neural Network (ANN) model
has been created with emphasizing the accuracy of the different techniques. A wide range of
statistical indicators used to assess the performance of the models. Based on the comparison, it
was revealed that technical ANFIS improve the accuracy of estimates in the small pool and
tropical ANN.

Keyword: Plate Fin Heat Exchanger, Performance, Flow Rate, Methods

1. Introduction:
The compactness of heat exchanger is the index of progress in the present day scenario of
industrial growth [4]. Especially with increasing the need for developing the cryogenics field.
Usually, plate fin heat exchanger is suitable for numerous type of heat exchanger application for
a wide range of industry [21-31]. Plate fin units are normally arranged for counter flow heat
exchanger. Plate fin heat exchanger has thin corrugated fins or corrugated heat transfer surface of
the plates. Compact heat exchanger surface density is very high that means large surface area per
unit volume it could be as high as 1800m2/m3, the plate fin heat exchanger is suitable for a close
approach temperature as low as 2oC, Two or more streams can be used by changing the section.
Plate fin heat exchanger is significant nowadays and most widely used due to high heat transfer
rate. It is investigated that compact heat exchangers such as plain fin strip, offset fin, wavy fin,
perforated fin,etc the pressure drop decrease with respect to increasing the turbulence in working
fluid. Onwards 1942 by Norris and Spofford [1] provide the first experimental report they draw
out the effect of heat transfer coefficient on the basis of length, thickness and pitch of fins and
also reduced the friction factor and Colburn modules. As the practical demand of plate fin heat
exchanger has increased experimental studies, have been made by London and Shah[2] in 1967
they been brought to a conclusion that small offset spacing (length/dh), fin thickness and a large
number of fins per inch gives better heat transfer. In 1975 A.R.Wieting [3] set up a statically
relationship between the variables from earlier experimental heat transfer and fluid flow friction.
Data for a plate-fin heat exchanger of offset fin and by using this statically relationship (untested
offset fin geometries can be predicted realistically and accurately within the parameter range of
the correlation). So that one can predict virtually and correctly within the parametric range of
newest offset plate-fin heat exchanger having no previous tested data. Experimental validation of
numerical simulation and also provides a comparison of experimental result in computationally
obtained results from the effects of fin thickness and free stream turbulence. In 1977, a set of
experiments was performed by Sparrow [5] to observe the heat transfer for a mass flow rate with
varying the Reynolds number. The thickness ratio and the spacing ratio are the other factors
establish that the Nusselt number varies while changing the plate thickness and also come upon
after searching that it is not necessarily equal spacing, and length gives optimal results. Cru and
Sparrow [6] again in 1979 analysis the heat transfer effectiveness ‘ε’ of staggered plate arrange
is higher than in plate line channel. In 1985 Joshi and Webb [8] expressed an analytical
framework to predict the j and f factor for laminar and turbulent flow from experimental and
analytical work. In 1994 Hu and Herold [9] suggested a liquid coolant instead of using previous
air cooled models in an experimental set up to evaluate heat transfer and pressure drop of offset
fin heat exchanger. It shows that the liquid cooled apparatus Prandtl number has a large effect on
Nusselt number, and numerical analysis examines the surface temperature distribution. Related
to CFD work in 2007 Peng and Ling [10] calculates the Colburn factor j and friction factor f for
an Aluminum- oil-air Plate Fin Heat Exchanger (PFHE) with serrated fins at low Reynolds No.
(Between10-200) Both experimentally, with constant air flow rate and six different oil flow rates
and numerically, with 3D geometric analysis. One of the objectives of this paper is also to
propose a procedure for the ANFIS model and an Artificial Neural Network (ANN) model
alongside a few experiments so as to predict the performance of fins with the new configuration
in PFHE. Again in 2007 Peng and Ling[11]developed the successful utilization of Genetic
Algorithm (GA) combined with the Back Propagation (BP) algorithm of Artificial Neural
Network.That is more efficient and advanced than the traditional GA method for the optimal
design of PFHE and showed that this method is also applicable to various PFHEs. In the same
year, Xie and Wang [12] applied genetic algorithm to optimize the design of plain plate
triangular fin compact heat exchanger, where fins standards and offset strip design adapted as
refered Kays and London [4].
Almost through with their studies the various other fin geometries louvered fin, perforated
fin,etc. In 2009, Peng and Ling [13] set up an artificial neural network for prediction of j and f
factors from experimental data for five different types of plate fin heat exchangers.In 2009,
Misra et al.[14] developed optimization of cross flow PFHE using GA method and showed the
importance of design approach based on the second law of thermodynamics. The conclusion
drawn shows the effect of an additional constraint on the optimum solution and power
requirement regarding pressure drops.
Zhu and Li [15] 2009 carried out and investigated the three-dimensional numerical simulations
on the flow and heat transfer in the four types of fins after that Wang and Liu [16] 2008 carried
out a numerical study of plate-fin heat exchangers with plain fins and serrated fins.Regarding
work is done in the field of Artificial Neural Network (ANN), in 2009, Tan et al.[17] developed
an Artificial Neural Network (ANN).To represent the overall behavior of the heat exchanger
over the whole range of flow rates, inlet temperatures, liquid compositions and blockage ratios in
experiments. Thus demonstrating that an ANN was able to predict the overall heat transfer rate
between the liquid and air streams in a compact fin-tube heat exchanger with a high degree of
accuracy.In 2010, Sepehr Sanaye [18] purposed multi-objective optimization with the objective
function effectiveness and total cost using a genetic algorithm and on suggesting a close form
expression between the variables and the objective functions estimate the total annual cost and
effectiveness.
In recent years, work related to serrated plate fin heat exchanger was in 2011 by M.
Yusef and Darus[19] employed genetic algorithm with particle swarm optimization technique to
optimize the plate-fin heat exchanger design. Another paper in 2011 [20] applied Neural
Network Model (NNM) upon data collected by CFD simulation to measure the accuracy of j and
f factors of NNM. The result displyed that NNM embrace the accuracy in between 1.3% and 1%
which is higher than appliactions of other models (embace the accuracy in between 3.8% and
8.2%) for analysising the same data of CFD simulation. However, for a precise response neural
network has to be supplied with well- defined factors. Also in 2011, Kim et al. [20] proposed
new correlations for j & f factor for offset strip fins with blockage ratios of greater than 20%,
with the j correlations suggested as functions of the Prandtl number. Resulted in the
enhancement of j and f factor (by 24%) for the optimized offset strip fin compared to the
referenced non-optimized offset strip fin. To understand the uniform distribution in PFHE, Saad
et al. [21] in 2011 investigated the hydrodynamics of a single phase flow in offset strip fins
deducing new correlations for the friction factor (f) from laminar to turbulent ranges. Their
deductions of the new correlations were in agreement with the numerical results of that from
CFD simulations, thus enabling the experimental observation of uniform distribution that is
crucial to obtain high performance in compact heat exchangers. In an another paper from Saad et
al.[22] in 2011, they performed single-phase CFD simulations for the determination of pressure
drop characteristics in an offset strip fin H.E that achieved good agreement between
experimental data and numerical prediction of friction factor. They also showed that distribution
of two-phase flow in CHEs’ depends on gas and liquid superficial velocities concerning the
design of the distributor. The multi-objectives formulated problems always aid the industrial
sectors to solve their several problems and in case of multi-objectives problems, the criterion are
considered as objectives [32-36]. The authors too suggested the role of fuzzy logic applications
in industrial realms [27].
In 2012 Yosefi and Mohammadi [23] applied a different technique to optimize plate-fin heat
exchanger by using a competitive algorithm (ICA). Seven optimized variable exploit to minimize
the total weight and total annual cost. In 2013, Buyruk et al. [24] investigated ways to increase
the efficiency of PFHE by optimization of fin angles, finintervals, and heights, offsetting fins
along a horizontal direction that has a potential for direct application to heat exchanger design
data.
In an another paper from Peng et al. [25] in 2014, they investigated the flow and heat transfer
characteristics of an innovative offset strip fin both experimentally and numerically in the
Reynolds Number range of 500-5000.Results showed the dependence of fin length, fin pitch and
fin is the bent distance on the performance of the compact heat exchanger (CHE) and how these
results could directly be used to design CHEs’. Also in the same year, Aliabadi et al. [26]
performed experiments to compare between the seven common types of channels of PFHE. The
conclusions showed that better heat transfer obtained from the vortexgenerator, wavy, offset
strip, and pin, perforated, louvered, and plain channels, respectively. The same order followed
for the maximum ability to reduce the surface area of the PFHE in comparison to the plain one.
Along with the gathered information above we emphasis on other literature regarding
ANFIS.Model for PFHX about the prediction of heat transfer and pressure drop using adaptive
neuro-fuzzy inference system and results forecast. Average Nusselt number and dimensionless
pressure showed good agreement with the work available by Tahseen Ahmadin 2013 [31]
another in ANFIS related to thermal work. Moon [32] implemeted two logic such as ANFIS-
based (Adaptive Neuro-Fuzzy Inference System-based control) ANN-based (Artificial Neural
Network-based control) except artificial intelligence to determine that how much do the
buildings have temperature control systems.
Due to increasing demand of plate fin heat exchanger in industry and research
work every user is interested in high efficient plate fin heat exchanger and this objective can be
achieved with different approaches. Heat exchange efficiency, increase or decrease depending on
outlet cold and hot fluid temperature and which is reliant on other factors, i.e. by controlling
mass flow rate on a specified heat exchanger. However, as mass flow rate increase the pressure
drops also increases, so a sensible compromise is needed.
In this paper, the first purpose is to present a structured neural network model, produced based
on experimentally observed data and a simulated annealing (SA) algorithm to search the high-
quality optimal process parameter conditions.Simulated annealing optimization scheme is
schematic to solve the multi-objective formulation. The objectives to achieve maximum
efficiency are (i) cold fluid outlet temperature and (ii) hot fluid outlet temperature. The variable
parameters during experiments are

Flow rate ‘Q’ (lit/min),

Pressure at cold inlet ‘P1.'
Pressure at hot inlet ‘P2’,
Pressure drop cold fluid mm of Hg,
Pressure drop hot fluid mm of Hg
the inlet temperature of cold and hot fluid
These are usually the actual performance parameter to evaluate the outlet performance.

2. Experimental Investigation, Modeling and Learning Procedure:

A serialized set of experiments is conducted by Alur [28]. The experimental setup adopted for
the study of plate fin heat exchanger is the steady state experiment. Measurement of temperature
and mass flow rate in the two sides provides the required information to compute the heat
exchanger effectiveness for the flow rate of liter/min operating between 315k to 365k, the
calculations of the performance parameter has carried out by the effectiveness from Shah [2] are
given below.

Cc T2  T1  Ch T3  T4 
  (1)
Cmin T3  T1  Cmin T3  T1 
Where,
T1= Temperature at inlet of cold fluid
T2= Temperature at outlet of cold fluid
T3= Temperature at inlet of hot fluid
T4= Temperature at outlet of hot fluid
Based on the previous Experiment, we gleaned objective data; depicted in Table-1; it is decided
to assess the behavior of plate fin heat exchanger regarding flow properties before
recommending for Industrial applications. Experimental data were used to determine contact
between the parameters. The experimentconducted at different mass flow rates (5.7 g/s to 14.2
g/s) and different hot fluid inlet temperature between 315k to 365k. To study the variation of the
performance parameters and finally the data tabulated intotwo categories for training and testing
which areapplied for modeling and learning procedure:

Where,
Q (liters /min)is the flow rate
P1(kg/cm2) Pressure at cold inlet
P2(kg/cm2) Pressure at hot inlet
Pcd (mm of Hg) cold fluid Pressure drop
Phd(mm of Hg) hot fluid Pressure drop
T1(k) Cold fluid inlet temperature
T2(k) Cold fluid outlet temperature
T3(k) Hold fluid inlet temperature
T4(k) Hold fluid outlet temperature

3. Implementation of Neural Network Model:

The purpose of the developed is working purpose neural network. The network is a feed structure
after the first Levenberg-Marquardt propagation training algorithm applied presented by (Eq.2).
Network with neural network technology, part of the software MATLAB models. Unsubscribe
sigmoid transfer function is activated, all neurons:

x
log sin( x)  (2)
1  ex
Where x is the input signal. Thelearning set consisted of 24 cases of theplate-fin heat exchanger.
Results for outlet temperature of cold and hot fluid come from experiments. Many input
parameters can be considered to measure the effectiveness of heat exchanger. But, the problem
will be so complicated to be solved in case of many input parameters and experimental expenses
will also be more, in order to tide over this concerns, the authors conducted relevant literature
survey in the context of compuations problems pertain to exchanagers effectivess and elcted few
effectual inputs parameters i.e Flow rates are ‘Q’(l/min), the pressure of the cold water inlet ‘P1’
pressure at hot inlet ‘P2’, Pressure drop cold fluid and hot fluid mm of Hg and the inlet
temperature of cold and hot fluid.
The Network has an input layer, a hidden layer with 12 neurons and output layer produces
results. The structure of the simplified network given in Fig. 1.
 Fig.1. Simplified network structure designed for the purpose of optimization objective function
The data set was divided up between learning set for determination of network weights and
validation and testing data sets which give the independent measure for ability to generalize and
the network performance. The regression analysis is giving the information on network
performance presented in Fig.1.Network performance seems good accuracy. It is a matter to
optimize the parameters of the systematic changes with learning and processing of the data is
ready to shape. An algorithm based on empirical data, primarily designed. The network develops
its simplest form. The (Fig. 2a, b) depicted that ANN’s accuracy is valid.

Fig. 2 (a). Network error-1 during learning process

 Fig. 2 (B): Network error-2 during learning process

This Plate Fin Heat Exchanger (PFHX) ANN network performance can be obtained directly
from the input information that can be seen. In other words, given the mass flow rate in the
construction of a plate heat exchanger that is, on the inlet and outlet temperature and the
temperature difference and both cold and hot sides of the ribs geometries. Engineers or designers
have limited experimental data so to predict the performance Plate-fin heat exchangerANN
approach is useful and convenient. This heat transfer and flow characteristics can express by the
mathematical formulas that are a very complex phenomenon, which does not require an
understanding of ANN approach.

4. Implementation of ANFIS Model:

Adaptive neuro-fuzzy inference system (ANFIS) is a type Artificial Neural Network (ANN),
which based on fuzzy interference system. The name of the fuzzy production system (FIS) is
used to provide ANFIS with the start of a member of the training activities.
Are shown in Table. 3. As the raw data collected by the MATLAB software with a specific
structure and format converted into a data file. Model to produce accurate information about the
system (neural model input and output parameters mentioned) is important to the quality of the
training database. This data is not so easy for the manual pre-treatment, therefore, a fuzzy
clustering method without human intervention used for this task.
For Plate Fin Heat Exchanger (PFHX) two separate fuzzy system have been generated for T2
and T4 i.e. outlet temperature of cold and hot fluid separately with the same inputs, and as shown
in Fig. 3 (a-d).
Fig. 3 (a): ANN shows that output changes as changes in input variables
Fig. 3 (b): Surface viewer-1 show that output changes as changes in input variables

Fig. 3 (C): ANFIS shows that relationship between input and output variable
 Fig. 3 (d): Surface viewer-2 show that relationship between input and output variable

Fig: 3 (a, d), both controllers (ANN and ANFIS) is used to evaluate the performance of
the plate fin heat exchanger. ANFIS and ANN both depicted the corresponding results. In Fig. 3
(a), ANN showed that output changes as changes in input variable, while in Fig. 3 (C), ANFIS
showed the relationship between input and output variable.
On other hand, it is found by Fig. 3 b and d showing surface viewer, established the relationship
between input and output variable (output changes as changes in input variables). In Fig. 3 (b),
surface viewer-1 showed that output changes as changes in input variable, while in Fig. 3 (d),
surface viewer-2 showed that relationship between input and output variable

5. Implementation of Optimization Algorithm:

The simulated annealing has been applied to reduce labor represented by neural network
algorithm. Annealing is a way to solve the problems presented limited and unconstrained and
bound. Process models the physical process of heating and then slowly lowering the temperature
to reduce defects, whereby the energy system reduced. The way each simulated annealing
algorithm generates a new random point. The distance from the new location at the moment or
the scope of the search based on the probability distribution of a weight proportional to
temperature. Usability has accepted all the points in the new law for the purpose, but also the
probability of certain, pointing upwards determined. By the dots to the purposes of the
assumption that the supply to avoid trapped in local minima, which is possible in the position to
solve the whole world to explore more. Annealing schedule was chosen in order to reduce the
logarithm temperature statement systematically. As the temperature decreases, the algorithm
reduces the scope of the search to be together at least. Overall, the following elements of the
Simulated Annealing algorithm may be distinguished. The representation of possible solutions,
-The generator of random values in the solution,
-Solution assessment function-target function,
-Cooling or annealing methodthe starting temperature and the rules of its lowering in the process
of target function minimum searching. When the annealing algorithm is working it is necessary
to adopt a method reducing the probability of transition to the state of worse parameters. Such a
rule is called the cooling schedule. For the definition of the cooling schedule the following must
be given:
-Starting temperature T0,
-Final temperature or the alloy criteria,
-The length of the Markov's chain (depending on the number of variables),
-The rule for the temperature is decreasing.

(i) For cold fluid outlet temperature:

T2 = 5.29036 - 0.00302818 Q + 13.9982 P1 - 16.8278 P2 - 0.0849 Pcd + 0.0800369Phd +
0.91353 T1 + 0.0777416 T3

(ii) For hot fluid out let temperature:

T4 = -0.138507 + 0.022083 Q - 10.2438 P1 - 14.6291 P2 - 0.150565 Pcd +0.113141 Phd +
0.120165 T1 + 0.863851 T3

6. Result and Discussions:

 The result of conducted research has summarized below:
The Table (1) predicted data through modeling are intimating near the experimental values. The
error shows in Table (2) is within the acceptable circumscribe. Table (3) represented the errors in
prediction of responses. The Fig (3) recapitulates the summarized view of validation and
experimental result briefly after the construction the process model conveniently simulated
annealing is used for obtaining optimization result. The two objectives of the present study are
the maximization of temperature at the coldoutlet, and minimization of hot outlet temperature.
Table (4) represents the optimal fitness value for the simulated annealing. Similarly, Table (5)
represents the optimal fitness value for the genetic algorithm.

Table 1: Heat exchanger dataset for the Neural Network Model

S.No. Q P1 P2 Pcd Phd T1 T2 T3 T4 Eff

1. 300 0.08 0.06 9 6 315.24 360.22 368.96 321.1 0.890916
2. 400 0.14 0.12 15 12 311.35 359.94 367.91 316.95 0.90099
3. 500 0.2 0.17 25 22 311.93 361.38 368.88 317 0.910975
4. 550 0.24 0.2 30 26 312.82 361.71 369.45 317.35 0.920007
5. 588 0.28 0.24 31 27 313.41 361.33 368.96 317.86 0.919892
6. 650 0.32 0.26 40 35 314.16 360.74 368.72 318.08 0.928152
7. 300 0.09 0.06 12 10 313.94 352.08 358.83 319.3 0.880597
8. 400 0.14 0.1 15 13 313.6 352.88 358.86 318.43 0.893283
9. 500 0.2 0.16 24 20 312.7 353.05 358.69 317.35 0.898891
10. 550 0.24 0.19 30 26 315.08 353.06 358.86 318.99 0.91069
11. 588 0.28 0.23 34 31 316.55 353.16 358.83 320.3 0.911306
12. 650 0.34 0.28 38 35 315.75 352.39 358.32 319.06 0.922246
13. 300 0.08 0.06 8 7 313.32 343.27 348.86 317.78 0.874508
14. 400 0.13 0.11 15 13 314.13 344.11 348.98 317.85 0.893257
15. 500 0.2 0.16 23 21 316.18 344.66 348.88 319.5 0.898471
16. 550 0.24 0.19 30 26 316.1 344.52 348.71 319.44 0.897577
17. 588 0.28 0.24 33 31 316.62 344.63 348.88 319.59 0.907936
18. 650 0.34 0.28 39 34 316.6 344.16 348.8 319.18 0.919876
19. 300 0.08 0.06 8 6 313.92 335.01 339.31 316.94 0.881056
20. 400 0.14 0.11 16 14 315.77 335.86 339.26 318.45 0.885909
 21. 500 0.2 0.16 24 22 312.51 335.42 338.9 315.55 0.884805
22. 550 0.24 0.19 30 26 316.46 336.01 338.83 318.86 0.892713
23. 588 0.28 0.23 33 31 312.99 335.34 338.8 315.57 0.900039
24. 650 0.34 0.28 37 34 315.72 335.67 339.16 317.93 0.905717

Table 2: Comparison of the developed model with experimental data

Experimental Predicted
S.No. Q T1 T3
ANN ANFIS
T2 T4 T2 T4 T2 T4
1 588 312.99 338.80 335.34 315.57 334.84 315.37 335 315
2 650 316.6 348.80 344.16 319.18 344.76 318.98 344 319
3 650 315.72 339.16 335.67 317.93 335.12 317.72 339 318
4 650 314.16 368.72 360.74 318.08 360.79 317.92 361 318
5 650 315.75 358.32 352.39 319.06 352.44 318.88 352 319

Table 3: Errors in prediction of responses

% Error in prediction of T2 % Error in prediction of T4
ANN ANFIS ANN ANFIS
0.5 0.2
0.34 0.57
-0.6 0.16 0.2 0.18
0.55 -3.33 0.21 -0.07
-0.05 -0.26 0.16 0.08
-0.05 0.39 0.18 0.06

Table 4: Optimization results with best fitness values for simulated annealing

Best Q P1 P2 Pcd Phd T1 T3

Output fitness
response function
value
T2 312.34 418.19 0.096 0.273 36.779 6.026 315.37 335.85
T4 322.52 300.49 0.34 0.28 33.656 7.314 315.24 335.85

Table 5: Optimization results with best fitness values for genetic algorithm
 Best fitness Q P1 P2 Pcd Phd T1 T3
Output
function
response
value
T2 311.88 438.554 0.081 0.28 36.499 6.188 315.252 335.71
T4 321.86 300 0.34 0.28 37 6 315.24 335.67

7. Conclusion:
The purpose of conducted research work is to expose the modern models: ANN and ANFIS,
which investigated as a best modern model to prediction the results of compact heat exchanger
on input parameters. The ANN and ANFIS modern models have been found valid corresponding
to traditional other models i.e Genetic Algorithm and Simulated Annealing.
The first part of paper precede a multi-input output ANN and ANFIS based predictive model for
the anticipation of performance parameter such as (i) cold fluid outlet temperature and (ii) hot
fluid outlet temperaturefor experimental studies on plate fin heat exchanger. The model served as
a tool to calculate the performance parameter based on the variation of process parameters.
Another part gives an optimal result with best-fit values using simple genetic algorithm
and Simulated Annealing, which shown in Table 4 and Table 5. It is an experimental
investigation into the various parameters affecting these two algorithms and adapting them to our
problem.

REFERENCES:
1. R. H. Norris and W. A. Spofford, High performance fins for heat transfer, Trans. Am.
Sot. Mech. Engrs 64, 489-496 (1942).
2. R. K. Shah and A. L. London, Offset rectangular platefin surfaces-heat transfer and flow
friction characteristics, J. Engng Power 90,218-228, 1968.
3. R. Wieting, Empirical correlation for heat transfer and flow friction characteristics of
rectangular offset-fin platefin heat exchangers, J. Heat Transfer 97,488-490, 1975.
4. W. M. Kays and A. L. London, Compact Heat Exchangers, 2nd edn. McGraw-Hill, New
York (1964).
5. E. M. SPARROW and C. H. LIU, Heat-Transfer, Pressure-Drop And Performance
Relationships For In-Line, Staggered, And Continuous Plate Heat Exchangers. Vol.22,
1013-1025, 1979.
6. N. CUR and E. M. SPARROW Experiments on Heat Transfer and Pressure Drop For a
Pair of Collinear, Interrupted Plates Aligned With the Flow Heat and Mass Transfer.
Vol.21, 1069-1080, 1977.
7. K. Suzuki, E. Hirai and T. Miyake Numerical And Experimental Studies on A Two-
Dimensional Model of An Offset-Strip-Fin Type Compact Heat Exchanger Used At Low
Reynolds Number. Heat Mass Transfer. Vol. 28, 823-836, 1985.
8. Himanshu M. Joshi and Ralph L. Webb, Heat transfer and friction in the offset strip fin
heat exchanger, Heat Mass Transfer Vol. 30. 69- 84, 1987.
9. Sen Hu and Keith E. Herold, Prandtl Number Effect on Offset Fin Heat Exchanger
Performance: Experimental Results. Heat Mass Transfer Vol. 38. 1053-1061, 1995.
10. Hao Peng, Xiang Ling, Numerical modelling and experimental verification of flow and
heat transfer over serrated fins at low Reynolds number
11. HAO PENG and XIANG LING, Optimal design approach for the plate-fin heat
exchangers using Neural Networks cooperated with genetic algorithms, Applied Thermal
Engineering 28, 642-650, 2008.
12. Xie and Wang Optimization of compact heat exchangers by a genetic algorithm Applied
Thermal Engineering 28, 895-906, 2008.
13. Hao Peng, Xiang Ling, Neural networks analysis of thermal characteristics on plate-fin
heat exchangers with limited experimental data. Applied Thermal Engineering 29, 2251-
2256, 2009.
14. MANISH MISHRA, P.K. DAS and Sunil Sarangi, Second law based optimisation of
cross flow plate-fin heat exchanger designusing genetic algorithm, Applied Thermal
Engineering 29, 2983-2989, 2009.
15. Yinhai Zhu Yanzhong Li, Three-Dimensional Numerical Simulation on the Laminar
Flow and Heat Transfer in Four Basic Fins of Plate-Fin Heat Exchangers, Journal of Heat
Transfer, Vol. 130, 2008.
16. Yong-Qing Wang, Qi-Wu Dong, Min-Shan Liu and Dan Wang Numerical Study on
Plate-Fin Heat Exchangers with Plain Fins and Serrated Fins at Low Reynolds Number,
Chem. Eng. Technol., 32, No. 8, 1219-122, 2009.
17. C.K. Tan , J. Ward, S.J. Wilcox, R. Payne, Artificial neural network modelling of the
thermal performance of a compact heat exchanger, Applied Thermal Engineering 29,
3609-3617, 2009
18. Sepehr Sanaye , Hassan Hajabdollahi, Thermal-economic multi-objective optimization of
plate fin heat exchanger using genetic algorithm Applied Energy 87, 1893-1902, 2010.
19. M. Yusefi and Darus An imperialist competitive algorithm for optimal design of plate-fin
heat exchangers Heat and Mass Transfer 55, 3178-3185, 2012.
Min-Soo Kim, Jonghyeok Lee, Se-JinYook, Kwan-Soo Lee, Correlations and
optimization of a heat exchanger with offset-strip fins, School of Mechanical
Engineering, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul 133-791,
Republic of Korea.
20. Selma Ben Saad , Patrice Clément , Caroline Gentric , Jean-François Fourmigué , Jean-
Pierre Leclerc , Experimental distribution of phases and pressure drop in a two-phase
offset stripfin type compact heat exchanger, International Journal of Multiphase Flow 37,
576-584, 2011.
21. Selma Ben Saad , Patrice Clément , Caroline Gentric , Jean-François Fourmigué , Jean-
Pierre Leclerc, Single phase pressure drop and two-phase distribution in an offset strip
fincompact heat exchanger, Applied Thermal Engineering 49, 99-105, 2012.
22. Yosefi and Mohammadi An imperialist competitive algorithm for optimal design of
plate-fin heat exchangers Heat and Mass Transfer 55, 3178-3185, 2012.
23. ErtanBuyruk , KorayKarabulut, O merOnurKarabulut , Three-dimensional numerical
investigation of heat transferfor plate fin heat exchangers, Springer-Verlag Berlin
Heidelberg, 2013.
24. Hao Peng, Xiang Ling , Juan Li, Performance investigation of an innovative offset strip
fin arraysin compact heat exchangers, Energy Conversion and Management 80, 287-297,
2014.
25. M. Khoshvaght-Aliabadi , F. Hormozi , A. Zamzamian, Role of channel shape on
performance of plate-fin heat exchangers:Experimental assessment, International Journal
of Thermal Sciences 79, 183-193, 2014.
 26. J.E. Hesselgreaves, Compact Heat Exchangers; Selection, Design and Operation;
PERGAMON, 2001.
27. Sahu A. K., Sahu, N. K., & Sahu, A. K, Benchmarking CNC machine tool using hybrid
fuzzy methodology a multi indices decision making approach, International Journal of
Fuzzy System Applications, 4, 32-50, 2015a.
28. W.M. Kays, The basic heat transfer and flow friction characteristics of sixcompact high-
performance heat transfer surfaces, Journal of Engineering for Power, 27-34, 1960
29. J.V. Wolfersdorf, E. Achermann, B. Weigand, Shape optimization of coolingchannels
using genetic algorithms, ASME Journal of Heat Transfer 119, 380–38, 1997.
30. Tahseen Ahmad Tahseen , M. Ishak , M.M. Rahman Performance predictions of laminar
heat transfer and pressure drop in an in-line flat tube bundle using an adaptive neuro-
fuzzy inference system (ANFIS) model, Heat and Mass Transfer 50, 85-97, 2014.
31. Jin Woo Moona, Sung Kwon Jung b, Youngchul Kim b, Seung-Hoon Han, Comparative
study of artificial intelligence-based building thermal control methods - Application of
fuzzy, adaptive neuro-fuzzy inference system, and artificial neural network, 31, 2422-
2429, 2011.
32. Sahu A. K., Sahu, N. K., & Sahu, A. K. (2016c). Application of Modified MULTI-
MOORA for CNC Machine Tool Evaluation in IVGTFNS Environment: An Empirical
Study. International Journal of Computer Aided Engineering and Technology, Volume 8,
Number 3, 234–259, 2016a.
33. Sahu, A. K., Sahu, N. K., & Sahu, A. Appraisal of CNC machine tool by integrated
MULTI MOORA-IGVN circumstances: An empirical study. International Journal of
Grey Systems: Theory and Application, Volume 4, Number 1, 104-123, 2014.
34. Sahu, N. K., Sahu A. K., & Sahu, A. K, Appraisement and benchmarking of third party
logistic service provider by exploration of risk based approach, Cogent business and
management, 2, 1-21, 2015b
35. Sahu A. K., Sahu, N. K., & Sahu, A. K, application of integrated TOPSIS in ASC index:
partners benchmarking perspective. Benchmarking: An International Journal, 2, 540-563,
2016b.
36. Sahu A. K., Sahu, N. K., & Sahu, A. K, Appraisal of Partner Enterprises under GTFNS
Environment in Agile SC. International Journal of Decision Support System Technology
(IJDSST), Volume 8, Number 3, 2016c,

Highlights
1. Performance of compact plate fin heat exchanger has been measured.
2. Predicted data given by ANN has been verified by simulation data and the experimental
data.
3. Depicted the applications of optimization methods upon fin heat exchanger.
GRAPHICAL ABSTRACT