energies

Article
Thermal Degradation Studies and Machine Learning Modelling
of Nano-Enhanced Sugar Alcohol-Based Phase Change
Materials for Medium Temperature Applications
Ravi Kumar Kottala 1,2 , Bharat Kumar Chigilipalli 3 , Srinivasnaik Mukuloth 4 , Ragavanantham Shanmugam 5 ,
Venkata Charan Kantumuchu 6 , Sirisha Bhadrakali Ainapurapu 7 and Muralimohan Cheepu 8,9, *

1 Department of Mechanical Engineering, National Institute of Technology,

Tiruchirappalli 620015, Tamil Nadu, India
2 Department of Mechanical Engineering, M V G R College of Engineering (A),
Vizianagaram 535005, Andhra Pradesh, India
3 Department of Mechanical Engineering, Vignan’s Institute of Information Technology (A),
Visakhapatnam 530049, Andhra Pradesh, India
4 Department of Mechanical Engineering, Chaitanya Deemed to be University,
Warangal 506001, Telangana, India
5 School of Engineering, Math and Technology, Navajo Technical University, Crownpoint, NM 87313, USA
6 Electrex Inc., Hutchinson, KS 67501, USA
7 Department of Mechanical Engineering, Aditya Engineering College (A),
Surampalem 533437, Andhra Pradesh, India
8 Department of Materials System Engineering, Pukyong National University, Busan 48547, Republic of Korea
9 STARWELDS Inc., Busan 46722, Republic of Korea
* Correspondence: [email protected]

Abstract: Thermogravimetric analysis (TGA) was utilised to compare the thermal stability of pure
phase change material (D-mannitol) to that of nano-enhanced PCM (NEPCM) (i.e., PCM containing
0.5% and 1% multiwalled carbon nanotubes (MWCNT)). Using model-free kinetics techniques, the
kinetics of pure PCM and NEPCM degradation were analysed. Three different kinetic models such
Citation: Kottala, R.K.; Chigilipalli,
B.K.; Mukuloth, S.; Shanmugam, R.;
as Kissinger-Akahira-Sunose (KAS), the Flynn-Wall-Ozawa (FWO), and the Starink were applied
Kantumuchu, V.C.; Ainapurapu, S.B.; to assess the activation energies of the pure and nano-enhanced PCM samples. Activation energies
Cheepu, M. Thermal Degradation for pure PCM using the Ozawa, KAS, and Starink methods ranged from 71.10–77.77, 79.36–66.87,
Studies and Machine Learning and 66.53–72.52 kJ/mol, respectively. NEPCM’s (1% MWCNT) activation energies ranged from
Modelling of Nano-Enhanced Sugar 76.59–59.11, 71.52–52.28, and 72.15–53.07 kJ/mol. Models of machine learning were utilised to predict
Alcohol-Based Phase Change the degradation of NEPCM samples; these included linear regression, support vector regression,
Materials for Medium Temperature random forests, gaussian process regression, and artificial neural network models. The mass loss of
Applications. Energies 2023, 16, 2187. the sample functioned as the output parameter, while the addition of nanoparticles weight fraction,
https://doi.org/10.3390/en16052187
the heating rate, and the temperature functioned as the input parameters. Experiment-based TGA
Academic Editor: Chi-Ming Lai data can be accurately predicted using the created machine learning models.

Received: 21 January 2023

Keywords: thermal degradation kinetics; multi walled carbon nanotubes; phase change materials;
Revised: 17 February 2023
thermal storage; machine learning models
Accepted: 21 February 2023
Published: 24 February 2023

1. Introduction
Copyright: © 2023 by the authors. During the past few decades, with urbanisation and the fast development of popu-
Licensee MDPI, Basel, Switzerland. lations, the energy demand for space cooling and heating in the construction sector has
This article is an open access article dramatically increased. It is estimated that around one-third of the world’s energy pro-
distributed under the terms and duction is used by the heating industry. As the majority of energy is required for space
conditions of the Creative Commons
heating, nations with cold climates are facing substantial energy constraint problems [1].
Attribution (CC BY) license (https://
The average ambient temperature has been raised with the rise in emissions of greenhouse
creativecommons.org/licenses/by/
gases, which eventually affects the comfort levels of inside building thermal levels. Hence,
4.0/).

Energies 2023, 16, 2187. https://doi.org/10.3390/en16052187 https://www.mdpi.com/journal/energies

Energies 2023, 16, 2187 2 of 24

passive thermal heating methods are more effective than active thermal methods [1–3].
Integrating phase change material (PCM) is one of the best methods of passive heating,
which can either reduce or control the building’s interior temperature throughout the day
and night. This will help save costs by reducing heat pump running times, improving
energy efficiency, etc [4,5]. This helps in finding a solution for the growing energy demands,
and the depletion of fossil fuels necessitates the search for such alternate energy sources.
However, solar-integrated heat pump systems are becoming the most promising option for
space heating applications. However, the imbalance between energy supply and demand
has widened due to uneven solar radiation distribution. The most promising approach to
closing this gap is to incorporate thermal energy storage (TES) systems into solar thermal
systems [6]. Anand et al. [2] conducted an experiment to investigate the effect of the loop
heat pipe thermal management system technology on the regulation of the temperature
of low-concentrated solar photovoltaic cells. Because it is a passive thermal management
approach that does not require the use of any electricity, the phase-change material (PCM)-
based solar photovoltaic thermal management method has attracted great interest from
researchers [2].
A thermal energy storage (TES) system is a significant method for ensuring a constant
supply of energy, which ultimately results in an improvement in both the system’s per-
formance and its consistency. The sensible energy storage system (SESS) and the latent
energy storage system (LESS) are the two primary subcategories that make up the TES
system. The SESS method occupies more space due to the fact that it mostly depends on
the specific heat capacity and mass of the material, whereas the LESS method occupies
less volume and majorly depends on the phase transformation enthalpy. Solid–liquid and
solid–solid PCM make up the majority of the materials used in LESS because of the fact
that they require lower store energy density per unit volume and a higher energy density
than liquid–gas phase change materials [7]. Both organic and inorganic PCM materials
hold the same principles for energy storage. However, inorganic PCM has superior thermal
stability and PCM does not degrade when used in higher temperature applications [8,9].
The two types of PCM materials are organic and inorganic. Organic PCMs are further
subdivided into paraffin wax, polymer, alkane, fatty acid, and their eutectic compositions.
Specifically, these PCMs excel at storing low-temperature thermal energy (below 120 ◦ C).
However, salt hydrates, nitrates, carbonates, chlorides, and metals are all examples of inor-
ganic PCMs that are often used at temperatures below 300 ◦ C (suitable for high temperature
applications) [10–12]. There are two major classes of inorganic phase change materials: salt
hydrates and metals. The volumetric latent heat density, cyclic stability, and thermal con-
ductivity of these materials are all extremely high compared with the organic phase change
materials. However, phase segregation and sub-cooling are common in salt hydrates. Since
inorganic salts are extremely corrosive, containers must be made of high-quality materi-
als, resulting in a higher cost [13–15]. These inorganic PCM salts should offer improved
thermal conductivity in particular uses. Inorganic PCMs based on carbon can be utilised
to address this issue. The thermal conductivity of inorganic PCM materials can often be
enhanced through the use of carbon additive elements. In the past few years, scientists
have experimented with a wide range of carbon structured materials, such as nanoparticles
made of metal, single- and multi-walled carbon nanotubes [16], and expanded graphite [17].
Space heating, electronic device cooling, refrigeration systems, solar heating, vehicles, food
processing and storage, and the space industry are just a few of the many areas where
organic latent heat storage materials have been successfully tested [18]. Organic PCMs are
advantageous because they do not undergo phase separation; they have a low corrosion
resistance with metal containers, and their latent heat of fusion does not change over
time [19,20]. Organic PCMs are advantageous in low-temperature applications due to their
low melting point. Since organic PCMs have poor thermal conductivity, a greater surface
area is required to achieve the necessary heat transfer rate. Since organic PCM has poor
thermal conductivity, its application has been restricted. PCM’s thermal conductivity can
be improved by incorporating nanometre-sized particles into the base material. Increased
Energies 2023, 16, 2187 3 of 24

thermal conductivity shortens the charge/discharge time period by increasing heat transfer
exchange [21]. Because of this, the charge/discharge rate could be increased by applying
a minimal quantity of heat. In order to enhance the crystallisation temperature of PCMs,
multi-walled carbon nanotubes with high specific surface areas have been proposed as a
type of heterogeneous nucleation centre [22–25]. Among all organic PCM samples, sugar
alcohols such as erythritol, xylitol, D-mannitol, Myo-inositol, etc., have become a most
promising option for medium temperature applications due to high latent heat of fusion.
Jame L. et al. [26] dispersed graphene oxide nanoparticles onto D-mannitol phase change
materials to improve their thermal conductivity. The prepared nano-PCM composite could
be incorporated into a thermoelectric system for power generation.
Furthermore, a PCM’s thermal stability is an important aspect to consider when using
it in thermal energy storage systems. The thermal stability of the PCM can be evaluated
with the help of thermogravimetric analysis. Two of the most crucial factors in establishing
PCM thermal stability are the rate of degradation and the temperature at which degrada-
tion occurs. Experimental TGA measurements, such as temperature versus weight loss at
different heating rates, can be used to calculate the degradation kinetics of the chosen PCM
samples. TGA studies have shown that temperature plays a significant role in the reaction
rate of any given composition. Changes in reaction rate as a function of temperature can be
accessed through the activation energy. However, this evaluation is essential for forecasting
the thermal stability of nano-enhanced PCMs. Many different kinetic models can be used
to study the thermal degradation of organic and inorganic materials using non-isothermal
TGA data. There are two primary ways to study kinetics: model fitting and model-free. One
or more kinetic models are fitted, and from those models the best statistical fit is determined.
The activation energy and pre-exponential factors may be calculated using the kinetic model
that best fits the data. It is common practice to use these models to determine the kinetic
properties of specific samples using only a single heating rate. Iso-conventional kinetic
methods can be used to address the fundamental drawback of this method, which is that it
does not take into account variations in activation energy as a function of the heating rate
(also known as model-free kinetic methods). The response model may be estimated with a
small number of errors using these methods, and they are also quite easy to implement.
Degradation of phase transition materials has been recently studied using iso-conventional
kinetic techniques [27]. The thermal degradation kinetics of nano-enhanced solid-solid
PCM materials were studied by Venkitaraj et al. [28] using the KAS model. According
to kinetic analyses, the activation energy of PCM (Pentaerythritol) lowers as the loading
of nanoparticles increases (Al2 O3 ). Pure PCM and nano-PCM composites with 0.1 and
1% Al2 O3 each had an average activation energy of 123, 116, and 98 kJ/mol, respectively.
Sun et al. [29] developed the microencapsulated PCM using in situ polymerisation. The
pentaerythritol and ammonium polyphosphate (PCM) sample is protected within a shell
material in this encapsulated PCM material (i.e., melamine-formaldehyde). The kinetics of
deterioration of the encapsulated materials were evaluated using the Kissinger and FWO
models. The prepared sample had a maximum activation energy of 151.7 kJ/mol. To inves-
tigate the thermal stability of the bio-based PCM, Fabiani et al. [30] used iso-conventional
techniques (the Starink and Miura-Maki procedures) (Palm Oil). The computed activation
energy for both models is between 48.5–127 kJ and 48.1 −126.7 kJ, respectively. Kinetic
analysis suggests the two-dimensional phase boundary reaction model best describes palm
oil oxidation. Xiang et al. [31] fused various hollow fibre membranes (polypropylene,
polysulfone, and polyvinylidene fluoride) to construct shape-stabilised PCMs. Selected
PCM samples’ degradation kinetics were analysed using the FWO technique. At a 50%
conversion rate, the activation energy of polypropylene-based PCM was measured to be
99.30 kJ/mol, making it the most energy-intensive PCM.
To achieve certain objectives, such as good thermal stability, less degradation, higher
peak holding times, etc., it is important to optimise the relevant parameters during PCM
preparation. In addition, the empirical method is commonly employed in the industrial
sector to decrease experimental effort [32,33]. Modelling and predicting the performance
Energies 2023, 16, 2187 4 of 24

of a variety of engineering systems has been accomplished by researchers making use of

Artificial Intelligence (AI) techniques. Therefore, in order to predict PCM degradation, it
was required to undertake an extensive examination of the thermogravimetric analysis of
PCM materials, which helps in determining the thermal stability of the material. Therefore,
to eliminate the repeated experimentation work on various input conditions, several
machine learning models have been effectively used for the prediction of thermal stability
of PCM samples [16]. The mass loss of the sample is majorly dependent on the addition
of nanoparticles weight fraction, the heating rate, and the temperature functioning as the
input parameters [17]. As a result, the overall purpose of employing prediction tools helps
in achieving an ideal set of parameters [34].
Support vector regression (SVR), Random Forest (RF), and Artificial Neural Network
(ANN) are some of the most popular artificial intelligence models that are used to anticipate
output characteristics with a high level of accuracy. These machine learning models have
been efficiently employed in various engineering applications to create the nonlinear
correlations between the output and input features. [27,28]. Ravi et al. studied DSC results
of multi-walled carbon nanotubes (MWCNT) [16], binary eutectic/expanded graphite
composite nano-PCMs [17], and nano-encapsulated phase change material (NEPCM) [35]
to forecast the thermophysical characteristics of various PCM materials behaviour using an
artificial neural network. Recently, machine learning models are effectively implemented to
predict the TGA plots of the pyrolysis of the bio mass samples [36]. Wang et al. [37] studied
the pyrolytic kinetics of various feedstocks using machine learning approaches. This helped
in studying thermal decomposition, a model-free pyrolytic activation energy, using RF
machine learning methods. Balasore et al. [38] used a machine learning approach for the
prediction of biomass pyrolysis kinetics. Wei et al. [39] studied predicting co-pyrolysis of
coal and biomass using machine learning approaches.
From the above analyses of the existing literature, it is apparent that researchers have
not yet undertaken a thorough analysis into the thermal stability of the nano-enhanced
D-mannitol PCM sample. The main objective of this research is to use various model-free
integral kinetic models to evaluate the kinetic characteristics of pure PCM and nano-
enhanced PCM samples for medium temperature thermal storage systems. This can be
accomplished by performing non-isothermal TGA experiments on pure PCM and nano-
enhanced PCMs at different heating rates of 10, 20, 30, and 40 ◦ C/min. This current research
is being expanded to include the development of various machine learning models for
predicting the stability of pure PCM and nano-enhanced PCMs. The type of PCM, heating
rates, and temperature at corresponding conversion rates are all considered as input data.
As output data, the mass loss of the corresponding input data set is used. These machine
learning models effectively predict experimental outcomes with the highest R2 values.

2. Materials and Methodology

2.1. Sample Preparation and Chemical Stability of Nano PCM Composites
D-Mannitol phase change material, with a melting point in the range of 165–168 ◦ C,
was purchased from I lab tech scientific solutions Pvt Ltd. in Trichy, India. The multi-walled
carbon nanotube (CNT) was purchased from Royal Scientific Solutions in Trichy, India. It
had a nominal diameter of around 4–25 nm, a purity of approximately 98.5%, a specific
surface area of approximately 360 m2 /g, and an approximate length of approximately
1–15 µm. In order to analyse the crystal structure with the Rigaku Ultima IV, X-ray diffrac-
tion was utilised as a technique. The Debye-Scherer relation was used to compute the
average crystalline diameter of pure PCM, and the results came out to be 43.1 nanometres.
The polymorphic form of pure PCM was shown to be in its b phase when typical peaks were
observed at 23.715◦ and 19.08◦ (22–1797 JCPDS). For the preparation of nano-enhanced
PCM samples, an innovative approach consisting of two steps was chosen. The sample
was manufactured by adding MWCNT at various mass fractions (i.e., 0.5 and 1%) using a
ball milling technique (High Energy Ball Mill Emax). Sample holders were outfitted with
stainless balls, which, due to the centrifugal motions of the balls, ensured consistent mixing
 pute the average crystalline diameter of pure PCM, and the results came out to be 43.1
nanometres. The polymorphic form of pure PCM was shown to be in its b phase when
typical peaks were observed at 23.715° and 19.08° (22–1797 JCPDS). For the preparation
of nano-enhanced PCM samples, an innovative approach consisting of two steps was cho-
sen. The sample was manufactured by adding MWCNT at various mass fractions (i.e., 0.5
Energies 2023, 16, 2187 5 of 24
and 1%) using a ball milling technique (High Energy Ball Mill Emax). Sample holders
were outfitted with stainless balls, which, due to the centrifugal motions of the balls, en-
sured consistent mixing of the sample. Following the ball milling procedure, the nano-
of the sample.
composites wereFollowing
subjectedthe
to ball milling
thirty minutes procedure, the nanocomposites
of vigorous were subjected
ultra-sonication using a probe-
to thirty minutes of vigorous ultra-sonication using a probe-type sonicator.
type sonicator. The samples were placed on a temperature-controlled flat plate Theheater
samples
so
were placed on a temperature-controlled flat plate heater so that ultra-sonication
that ultra-sonication could be performed while the substances were still liquid. The ther- could be
mostat on the heater◦was set to a temperature of 200 °C, which is considerably higher thana
performed while the substances were still liquid. The thermostat on the heater was set to
temperature
the of 200
melting point C, which
of pure PCMis(165–168
considerably higher
°C). After than thesonification,
extensive melting point of pure PCM
well-dispersed
(165–168 ◦ C). After extensive sonification, well-dispersed homogeneous nano-enhanced
homogeneous nano-enhanced PCM samples were generated. The XRD peaks of the pure
PCM samples were generated. The XRD peaks of the pure PCM, MWCNT, and nano-
PCM, MWCNT, and nano-enhanced PCM samples are shown in Figure 1. The XRD test
enhanced PCM samples are shown in Figure 1. The XRD test of the 0.5% MWCNT-based
of the 0.5% MWCNT-based PCM sample was carried out, and the results showed that
PCM sample was carried out, and the results showed that there were neither any new
there were neither any new peaks nor any notable peak shifts when compared to the XRD
peaks nor any notable peak shifts when compared to the XRD spectra of pure PCM. This
spectra of pure PCM. This leads one to believe that there is no significant change in the
leads one to believe that there is no significant change in the crystal structure of pure PCM
crystal structure of pure PCM following the dispersion of MWCNT.
following the dispersion of MWCNT.

Figure 1. XRD peaks of pure PCM, MWCNT, and nano-enhanced PCM samples.
Figure 1. XRD peaks of pure PCM, MWCNT, and nano-enhanced PCM samples.
The FT-IR was used to investigate how well pure PCM and MWCNT were compatible
The FT-IR was used to investigate how well pure PCM and MWCNT were compati-
from a chemical perspective (Make: FT-IR, NIR, and FIR Spectroscopy, PerkinElmer Fron-
ble from a chemical perspective (Make: FT-IR, NIR, and FIR Spectroscopy, PerkinElmer
tier). The FT-IR spectra of PCM nano composites are shown in Figure 2. After analysing
Frontier). The FT-IR spectra of PCM nano composites are shown in Figure 2. After analys-
the spectra, it was discovered that the characteristic bands of pure PCM were located
ing the spectra,
at 3280 it wascm
cm−1 , 2942 discovered
−1 , 1386 cmthat
−1the characteristic
, 1078 cm−1 , 942bandscm−1 ,of876purecmPCM were
−1 , and 723located
cm−1 .
at 3280 cm −1, 2942 cm−1, 1386 cm−1, 1078 cm−1, 942 cm−1, 876 cm−1, and 723 cm−1. The stretch-
The stretching frequency of O-H is denoted by the vibrational spectra at a wavelength of
ing
3280frequency
cm−1 , with of O-H is denoted by
a transmittance the vibrational
percentage spectra
of 58.49%. at a wavelength
A vibrational peak of at 3280
2942 cmcm−,1
−1

with
with aa transmittance
transmittancepercentage
of 74.85% wasof 58.49%.
assignedA vibrational peak at
to the stretching 2942 cm also
frequency,
−1 with a trans-
known as
mittance of 74.85% was assigned to the stretching frequency, also
the strong band, for the CH3 , CH2 , and C-H functional groups. The bending vibration known as the strong
band,
of CHfor the CH3, CH2, and C-H functional groups. The bending vibration of CH3, CH2
3 , CH2 (medium), and stretching bands were observed in the finger print region at
(medium),
1386 cm 1 and
− and stretching
1078 cm−1bands were observed
, respectively. At 723 in
cmthe
−1 , finger
876 cmprint region
−1 , and 942 at
cm1386 cm−1 and
−1 , there was
1078 cm −1, respectively. At 723 cm−1, 876 cm−1, and 942 cm−1, there was evidence of C-H
evidence of C-H rocking out of plane, =C-H bending, and C-H bending, respectively. The
rocking
spectra of outnano-enhanced
of plane, =C-HPCM bending, andshowed
samples C-H bending,
the same respectively. The spectra
distinctive peaks as those ofofnano-
pure
enhanced
DM, withPCM samples showed
the exception of a verythe same
small distinctive
shift peaks as value.
in transmittance those of It pure DM, with
is possible that the
the
exception of a very small shift in transmittance value. It is possible that
physical mixing of the additives in the pure PCM caused these modifications. In addition, the physical mix-
ing
the of the additives
spectra in the pure PCM
of the nano-augmented caused
PCM theseonly
showed modifications.
modest shifts In throughout
addition, the thespectra
whole
spectrum, which demonstrates that there was no chemical disintegration in the PCM as a
result of the addition of MWCNT. It was clear from the comparison of the FTIR spectra that
the composites were chemically inert and could be utilised in LHES on a long-term basis.
 of the nano-augmented PCM showed only modest shifts throughout the whole spectrum,
which demonstrates that there was no chemical disintegration in the PCM as a result of
Energies 2023, 16, 2187 the addition of MWCNT. It was clear from the comparison of the FTIR spectra that 6theof 24
composites were chemically inert and could be utilised in LHES on a long-term basis.

FTIR
Figure2.2.FTIR
Figure spectrum
spectrum ofof purePCM
pure PCMand
andnano-PCM
nano-PCMsamples.
samples.

2.2. Thermogravimetric Analysis of Nano-PCM Samples

2.2. Thermogravimetric Analysis of Nano-PCM Samples
2.2.1. Model-Free Kinetic Methods
2.2.1. Model-Free Kinetic Methods
A single-step kinetic equation is commonly used to examine the thermal degradation
A single-step
of solid-state kinetic equation is commonly used to examine the thermal degradation
reactions.
of solid-state reactions. dα
= k( T )· G (α) (1)
dt
= 𝑘(𝑇) · 𝐺(𝛼) (1)
where, α is denoted as conversion of reaction rate, k ( T ) is known as reaction rate constant,
and Gα(αis
where, ) is definedasasconversion
denoted a reactionof
model relevant
reaction rate, to𝑘 (𝑇)
the is
reaction
knownmechanism. Here,
as reaction rate the
con-
conversion of reaction rate can be determined by following Equation (2).
stant, and 𝐺(𝛼) is defined as a reaction model relevant to the reaction mechanism. Here,
α
the conversion of reaction rate α can be determined by following Equation (2).
(m − mα )
α = ( i )  (2)
α = mi − m f (2)

where𝑚mi is
where is known
known as
as the
the initial
initial mass
mass of
of the
the nano-PCM
nano-PCM sample,
sample, m𝑚 mass
α
massofofthe
thesample
sampleat
atreaction
reactiontemperature,
temperature,andandm 𝑚is known
is known as the final residual mass of the PCM sample
as the final residual mass of the PCM sample after
f
after complete
complete decomposition.
decomposition. The following
The following is theisArrhenius
the Arrhenius expression
expression stated stated as a ki-
as a kinetic rate
netic rate equation with respect to conversion
equation with respect to conversion accounting: accounting:

𝐾(𝑇) A·𝑒 ( −E. a ) (3)

K ( T )== A ·e Rg .T (3)
Equation (4) can be obtained by substituting Equation (3) into Equation (1)
Equation (4) can be obtained by substituting Equation (3) into Equation (1)
𝑑𝛼
= 𝐴 · 𝑒 −×Ea · 𝐺(𝛼) (4)
𝑑𝑡dα = A·e Rg ×T · G (α) (4)
dt
where the T, 𝐸 , and 𝑅 are known as absolute temperature (K), activation energy
where the
(kJ/mol), 𝛽 T,
is Eknown
a , and R are known
asg heating rateas(K/min)
absolute temperature
and universal gas (K), constant
activation(Jk
energy
−1mol−1(kJ/mol),
).
β isAknown
linear as heatinginrate
increase (K/min) and
temperature withuniversal
respect togastimeconstant (Jk−1 mol
is observed −1 ).
in non-isothermal
A linear increase in temperature with respect to time is observed in non-isothermal
conditions;
conditions;
𝑑𝑇dT= 𝛽
=β (5)(5)
𝑑𝑡
dt
Equations
Equations(4) (4)and
and(5)(5)can
canbebecombined
combinedtotoobtainobtainthe thefollowing
followingEquation
Equation(6) (6)

dα A − Ea
= · e Rg ×T · G (α) (6)
dT β
Energies 2023, 16, 2187 7 of 24

The above equation can be expressed as an integral, as follows:

Z T − Ea
dα A AEa
Z α
f (α)= = e Rg .T
dT = . p(z) (7)
0 G (α) β To βR g

p(z) is a temperature integral that is approximated by the numerical approach.

Model fitting and iso-conversional methods are both used to estimate kinetic pa-
rameters in non-isothermal thermogravimetric analysis. In order to estimate the kinetic
parameters, the model-free method is the most commonly employed approach. The iso-
conversional approach yields kinetic parameters with a constant degree of conversion at a
constant heating rate. As a result, the iso-conversional method did not require any prior
knowledge of kinetic reactions, in which the rate of reaction rate is solely dependent on the
degradation temperature of the PCM sample. Differential or integral approaches can be
used to obtain the iso-conversional method. FWO, KAS, and Starink methods are the most
often used integral methods for determining kinetic parameters.

2.2.2. Kissinger–Akahira–Sunose (KAS)

Kissinger (Kissinger 1957), as well as Akahira and Sunose (Akahira and Sunose
1971), proposed an iso-conversional integral model-free method known as the KAS [40].
KAS employs a simple approximation of the exponential integral, also known as Doyle’s
approximation, i.e., p(z) = z−2 × e−z .

A × Ea
   
β Ea
ln = ln − (8)
T2 R g × g(α) Rg × T
 
Plotting ln T2 Vs T1 on a graph can be used to determine the activation energy (Ea )
β

at each conversion. The slope of the plot at each conversion can be used to estimate the
activation energy of the prepared samples.

2.2.3. Flynn-Wall-Ozawa (FWO)

Ozawa (Ozawa 1965) and Flynn and Wall (Flynn and Wall 1966) discovered the FWO
kinetic method [41]. It is an integral iso-conversional technique for the computation of
activation energy that uses the linear Doyle approximation instead of the temperature
integral (Equation (9)), i.e., ln( p(z)) ≈ −5.331 − 1.052z.

A × Ea
 
Ea
ln( β)= ln −1.0516 × (9)
R g × g(α) Rg × T

The activation energy (Ea ) can be determined by plotting the graph of ln( β) versus T1
at each conversion. The activation energy of the prepared samples can be calculated from
the plot slope at each conversion.

2.2.4. Starink Method

The activation energy is estimated using the Starink approach [42], which is similar to
the KAS and FWO methods. The activation energy is represented as follows.
 
β Ea
ln 1.8
= Cs −1.0037 × (10)
T R g×T
 
Plotting the graph of ln T1.8 versus T1 at each conversion can be used to determine
β

the activation energy (Ea ). The plot slope at each conversion can be used to estimate the
samples’ activation energy.
Energies 2023, 16, 2187 8 of 24

2.3. Description of Machine Learning Models

2.3.1. Linear Regression
Linear regression is an old technique for determining the degree of association between
two or more characteristics. Once the features (input) and the goal (output) are established,
the learning process is executed to minimise the loss function value (such as Mean Squared
Error). In regression, the optimal parameters are those that result in the smallest loss. This
model is reliable and simplistic nature. Multiple linear regression models can be described
in their broadest sense by the expression.
m
ŷ p = bo + ∑ j=1 x j b j (11)

where ŷ p is the predicted outcome, x j is the selected features, and bo and b j are denoted as
coefficients of the regression model.
The model is used to find good fits for several linear equations relating the mass
loss of the nano-enhanced PCM sample to the specified parameters. The connection
between characteristics and mass loss of the PCM sample is convoluted and nonlinear. For
this reason, original characteristics of varying polynomial degrees are used to generate
polynomial features with the goal of improving the prediction accuracy of the linear
regression model [43].

2.3.2. Support Vector Machine

Extending the functionality of support vector machines to regression issues with
non-conventional kernels, including polynomial, linear, and radial basis functions. The
kernel function is a technique for processing raw data and transforming it into the desired
format for further analysis. The radial basis function is a generic kernel that can be applied
to any dataset without any prior information. It is critical in small-sample learning that
the model’s error margin be minimised, while its generalisation potential be maximised;
the SVR approach can accomplish both of these goals. The error-fitting threshold SVR is
constructed as follows:
1 1 m
RSV M (ω, a, ϕ1 , ϕ2 )= ||ω ||2 +CSV M ∑i=1 L( ϕ1 , ϕ2 ) (12)
2 N


y j − ωτSV M
x j − a ≤ σ + ϕ1
Subjected to ωτSV M x j + a − y j ≤ σ + ϕ2
ϕ1 , ϕ2 ≥ 0 j = 1 . . . N


where ||ω || is represented as a regularization term, τSV M is termed as a high dimensional

space. The error penalty factor known as CSV M is utilised in the process of determining
the optimal trade-off between the regularisation term and the empirical risk. The ϕSV M
loss function is the one that is equivalent to the approximation precision of the training
data point. ϕ1 and ϕ2 are known as loss functions at plane 1 and plane 2. N represents the
number of data points. σ denotes the variance of the data points. x j and y j are termed as
input and output parameters.

2.3.3. Random Forest Regression

Random forest (RF) is an ensemble classifier, put to use in the process of conducting
analysis on TGA data. The training step of random forest involves the creation of many
decision trees, which are then followed by the generation of class labels based on which
options received the greatest number of votes. A decision tree stands for a collection of
constraints or conditions that are hierarchically organized and carried out in the correct
order, beginning at the tree’s origin and progressing outward until they reach the tree’s
final node or leaf. When compared with artificial neural networks, the most significant
benefit of utilising a hierarchical tree structure is that the structure is visible, which makes
it much simpler to understand. To construct the DT from a collection of input datasets, an
Energies 2023, 16, 2187 9 of 24

evaluation measure is applied to each evidential attribute to boost the internode’s level
of variability. To obtain the DT, multiple regressions on the dataset, as well as recursive
partitioning of the dataset, are utilised. The technique for data splitting is carried out
in an iterative fashion, beginning at the root node and continuing until an established
termination condition is met. Only specified terminal nodes or leaves are included in
the scope of application for a simple regression model. To boost the tree’s generalisation
capacity, reduce its structural complexity. When pruning, the total occurrences in a node
can be considered. The method known as random forest is one that gathers the results of
multiple decision trees into a collective pool and then chooses the most optimal one from
among those pools.
RF achieves a high level of regression accuracy and is able to deal with outliers and
noise in the data. The use of RF is justified by the fact that it is robust to overfitting and
has shown promising regression outcomes in previous studies [44]. The random forest
regression is given a sample that has already undergone pre-processing consisting of n
samples. Through the utilisation of numerous feature subsets, RF generates n distinct trees.
Each tree generates a regression result, and the outcome of the overall model is dependent
on the vote of the majority of the trees. The class with the highest total number of votes will
be the one to receive the sample. From the literature [44], it is clear that the RF regression
model outperforms the decision tree model for such data. The RF has a higher accuracy
than the decision tree and there is less of a danger of overfitting the data. Furthermore,
it has outperformed other classifiers because it is easy to interpret, non-parametric, and
performs well on a large dataset.

2.3.4. Gaussian Process Regression

For nonlinear problems, the non-parametric Gaussian process method is used. It treats
all input and output data as though they came from a Gaussian distribution by employing
random variables. The Gaussian process, when fed into the training dataset, provides
distributions for all realisable functions. Therefore, there is no upper limit on the number
of variables that can be employed in a Gaussian process, and this number can be expanded
in proportion to the number of datasets used for training. Definition: GPR is a Gaussian
.
. .

process with the mean function m z and the kernel function k z, z0 .
.
.
. .


GR z = GPR m z , k z, z0 (13)
.

The central tendency of GR is represented by
the variable m z . The following is a
. .

relationship between the values of the test input x and the test output y p :
. .

y p = GR z + χ (14)

here, χ is known as the noise of the independent variable. It is represented by a distribution

with a mean equal to zero and a standard deviation of ϕn , and its definition is as follows:
 
χ = Y 0, ϕ2n (15)

A minimal confidence formula is applied to the dataset that is used as a sample. The
formula is as follows:  
J(y| I )= Z y I, ϕ2n K (16)
. h. . . . iT
Y= Y 1 , Y 2 , Y 3 , . . . . . . . . . Y n (17)
 .
.
.
.
 T
g = GP z1 , GP z2 , GP z3 . . . . . . . . . GP zn (18)
The distribution that the predictive dataset follows can be expressed as:
. . . .  
J Y s z, z0 , Y = Z σs , ϕ2s (19)
Energies 2023, 16, 2187 10 of 24
Energies 2023, 16, x FOR PEER REVIEW 10 of 25

  −1 .
σs = {s,n {n + ϕ2n I Y (20)
J 𝑌 |𝑧, 𝑧 , 𝑌 =Z(𝜎 , 𝜑 )−1 (19)
ϕ2s = {ss −{s,n {n + ϕ2n I { M,s (21)
𝜎 = ∁ , (∁ + 𝜑 𝐼) 𝑌 (20)
where σs indicates the mean value of the posterior Gaussian process, ϕ2s the covariance
matrix of the prediction data set, 𝜑 { =n ∁= k−(x, ∁ x), {s,n = k (x ∗ , x), { = k (x, x ∗ ), and {s,s(21) =k
, (∁ + 𝜑 𝐼) ∁ , M,s
(x ∗ , x ∗ ). The terms { are known as the covariance matrix that connected the training and
where
test data, 𝜎 and .
indicates
x is thethe mean value
training dataset.of the
Theposterior
matrix MGaussian process, 𝜑 bythe
× M is represented I. covariance
matrix of the prediction data set, ∁ = k (x, x), ∁ , = k (𝑥 ∗ , x), ∁ , = k (x, 𝑥 ∗ ), and ∁ , =
(𝑥 ∗ , ANN
k2.3.5. 𝑥 ∗ ). The terms ∁ are known as the covariance matrix that connected the training
Modelling
and test data, and
Simplified 𝑥 is thecomputational
nonlinear training dataset. andThe matrix M ×models,
mathematical M is represented by I. neural
such as artificial
networks, are capable of resolving a wide range of engineering challenges. When it comes
2.3.5. ANN Modelling
to predicting experimental outcomes, neural networks have become the most common
strategy, with greater
Simplified performance
nonlinear than older
computational methods. An ANN
and mathematical models, modelsuchmainly consists
as artificial neu-of
an input
ral networks, layer,area hidden
capablelayer, and an output
of resolving a wide layer.
rangeThe number of neurons
of engineering challenges. in the
Wheninputit
layer and
comes output layer
to predicting is equal to the
experimental numberneural
outcomes, of input and output
networks haveparameters.
become theConnecting
most com-
processing
mon strategy, elements via weighted
with greater performance linkages
thanisoldercommon
methods. in neural
An ANN networks. A weighted
model mainly con-
incoming
sists of an connection
input layer,isa usedhidden forlayer,
each processing
and an output element.layer.The
Theweights
numberand biases ofineach
of neurons the
neuron are changed to train the network structure in an appropriate
input layer and output layer is equal to the number of input and output parameters. Con- learning approach. The
training processing
necting continues until the root
elements viamean squarelinkages
weighted error is atisits lowest. Changing
common in neural the network’s
networks. A
weights and bias will help lessen the discrepancy between
weighted incoming connection is used for each processing element. The weights and predicted values and actual
bi-
values.
ases To construct
of each neuron are thechanged
network,tothe equation
train shownstructure
the network below is in effectively
an appropriateemployed.
learning
approach. The training continues until the root mean n
square error is at
 its lowest. Chang-
∑ j=0the ijdiscrepancy


F ( H n )= Logsig
ing the network’s weights and bias will help lessen ( or ) Tansig W × Z i + b j between predicted (22)
values and actual values. To construct the network, the equation shown below is effec-
tivelyInemployed.
the current ANN structure, the input neuron receives the total amount of infor-
mation from the input data, which is denoted by the letter Zi . According to the equation,
Wji represents the connection 𝐹(𝐻 ) =weight Logsigbetween
(or) Tansig input 𝑊
the (∑ layer× and
𝑍 + the𝑏 )hidden layer, and (22)
bj
represents the bias of each hidden neuron. W jk and b k , on the
In the current ANN structure, the input neuron receives the total amount of infor- other hand, denote the weight
connections
mation frombetween
the inputthe hidden
data, whichlayer and the by
is denoted output layer, 𝑍and
the letter the bias, respectively.
. According The
to the equation,
basic structure of an ANN network is shown in Figure 3.
𝑊 represents the connection weight between the input layer and the hidden layer, and
𝑏 represents the bias of each hidden neuron. n  𝑊 and  𝑏 , on the other hand, denote the
weight connections between theOhidden k = ∑ layer Wjk × andHnthe+outputbk layer, and the bias, respec- (23)
k = 0
tively. The basic structure of an ANN network is shown in Figure 3.

Figure 3. The basic structure of an artificial neural network.

Figure 3. The basic structure of an artificial neural network.
The anticipated output of each output neuron (Ok ) may be determined using
Equation (23). The Tansig or Logsig activation function is used between the input and
hidden layers, and the purelin activation function is used between the output and hidden
Energies 2023, 16, 2187 11 of 24

layers. The number of neurons present in the hidden layer is critical for the predictability
of the network. If there are too few neurons in the hidden layer, the precision of the results
will be poorer, whereas if there are too many neurons, the results would be unfair. As a
result, it is necessary to optimise the number of neurons in the hidden layer. The optimal
neurons are calculated with the help of Equation (24) [16]. The optimal neurons found
by solving this equation cannot be considered to be the ultimate solution. The neuronal
independence test is carried out with a sufficient number of neurons, falling somewhere
within the optimal range.

Input + Output √
Hn = + Train data (24)
2
Normalisation of input and output data is essential to increase the accuracy of results.
With this normalisation, accuracy and convergency time could be improved. Within the
range of 0 to 1, the input and output data are normalised. Use the following equation to
determine data normalisation.
( Zi − Zmin )
Z= ( H − L) + L (25)
( Zmax − Zmin )

The L and H denote the normalized data set’s lower and higher limit values. Zmin and
Zmax are known as the minimum and maximum values of the dataset. Zi is known as the
selected data point.

3. Results and Discussion

3.1. Thermogravimetric Analysis of Prepared Nano-Enhanced PCM Samples
Figure 4a–c show the thermogravimetric curves of D-mannitol PCM with differ-
ent multi-walled carbon nanotubes weight fractions (0, 0.5, and 1%) at heating rates of
5 ◦ C/min, 10 ◦ C/min, 15 ◦ C/min, and 20 ◦ C/min. The results of the TGA are presented as
a percentage of the PCM samples’ weight loss as a function of the temperature. The PCM
samples were heated from 40 to 700 ◦ C at a variety of different heating speeds. Experiments
were conducted with an initial sample size of 5 ± 0.5 mg in an alumina sample pan under a
nitrogen environment, with the gas flow rate set at 60 mL/min. From the results, it can be
seen that no samples were degraded, and the overall weight loss was found to be negligible
(less than 2%) during the phase transition zone. The temperature area of D-mannitol (below
200 ◦ C) indicates that D-mannitol was thermally stable for the application of energy storage
in the temperature range of its phase transition. When the pace of heating was increased,
the experiment revealed that the temperature at which the decomposition process begins
was significantly elevated. As the temperature rises, an examination of the TG curves
revealed a progressive decomposition pattern that occurred in a single step at each new
temperature. The melting process that resulted from heat degradation was responsible for
this decomposition, which may be traced back to the original substance. The rate of weight
loss also varied, but it was often slower at the beginning stages of decomposition. The
rate of weight loss was also variable. Chemical reactions, the release of species that were
adsorbed, and breakdown are the three processes that may be singled out as the causes of
the weight loss that occurs during thermal cycling.
The thermogravimetric analysis (TG) results of pure D-mannitol are displayed in
Figure 4a, which compares the effects of several heating rates. When heated at a rate of
5 ◦ C per minute, the pure PCM sample began to degrade, albeit at a slower pace at first;
however, at about 248.87 ◦ C, there was only a weight loss of less than 5%. This initial
weight loss was primarily attributable to the loss of PCM as it evaporated. The rate of
weight loss accelerated fast above 260 ◦ C, showing weight losses of 20%, 40%, 60%, and
80% at temperatures of 271.06 ◦ C, 284.99 ◦ C, 300.94 ◦ C, and 312.83 ◦ C, respectively. At
temperatures of 285.33 ◦ C, 297.21 ◦ C, 317.25 ◦ C, and 329.14 ◦ C, respectively, weight losses
of 20%, 40%, 60%, and 80% were observed when the heating rate was increased to 10 ◦ C per
minute. In the situation of TG that was carried out at a heating rate of 15 ◦ C/min, weight
 Energies 2023, 16, 2187 12 of 24

losses of 20%, 40%, 60%, and 80% were seen at temperatures of 295.538 ◦ C, 311.49 ◦ C,
Energies 2023, 16, x FOR PEER REVIEW331.53 ◦ C, and 347.52 ◦ C, respectively. Again, when the rate of heating was increased to13 of 25
20 ◦ C per minute, these weight losses were recorded at temperatures of 311.87 ◦ C, 327.83 ◦ C,
349.93 ◦ C, and 365.87 ◦ C, respectively.

Figure 4. TGA plots (a) pure PCM (b) PCM + 0.5% MWCNT (c) PCM + 1% MWCNT.
Figure 4. TGA plots (a) pure PCM (b) PCM + 0.5% MWCNT (c) PCM + 1% MWCNT.
Figure 4 displays the thermogravimetric data that were collected for PE combined
with As0.5% MWCNT
an energy at a variety
barrier, of heating
activation energy rates
works(b). against
When the thesample
degradingwas reaction.
heated at Asa seen
◦
inrate of 5 5,C/min,
Figure it was seen
PCM samples that athe
show addition
wide rangeofof 0.5% of the weight
conversion ratesfraction of MWCNT
(𝛼). Activation energies
had very
of pure PCM, little effect on the deterioration at the beginning of the process.
PCM + 0.5% MWCNT, and PCM + 1% MWCNT were estimated in this ex- At a temperature
of 229.91 ◦ C, there was still a loss of weight of less than 5%. However, at temperatures
periment using iso-conversional methods described by Kissinger Akahira-Sunose, Flynn-
higher than 260 ◦ C, the degradation of D-mannitol plus 0.5% MWCNT occurred at a pace
Wall-Ozawa, and Starink kinetic models. The activation energy (E) for complete PCM de-
that was slightly faster than that of pure PE. The thermogravimetric analysis (TG) results
composition was
of D-mannitol combined determined for various
with 0.5% MWCNT conversion valueslosses
indicated weight (0.2 toof0.8).
20%,KAS, FWO, and
40%, 60%,
Starink
and 80% kinetic models are
at temperatures 264.80 ◦ C, by
of analysed ◦ C, 293.54
plotting
279.17 ◦ C, and
the plots 305.86 ◦ C,
between ln respectively.
and , ln 𝛽
◦
The heating rate used was 5 C/min. When the rate of heating was increased to 10 C per ◦
and , ln . , and , respectively. The R2 value of the linear fitted lines of developed
minute, weight losses of 20%, 40%, 60%, and 80% were seen at temperatures of 279 ◦ C,
kinetics
293.54 ◦models
C, 309 ◦ C,of and
prepared
324 ◦ C,samples are In
respectively. more than 0.92.
the situation of Figures
TG carried 6–8outpresent these plots
at a heating
forrate
pure
of 15PCM,
◦ PCM
C/min, + 0.5%
weight MWCNT,
losses of 20%,and
40%,PCM60%, + 1%80%
and MWCNT.
were seen Asatpreviously
temperatures discussed,
of
the
289.44 ◦ C, 305.86
activation ◦ C, 324.34
energy ◦ C, and 338.70
was obtained ◦ C, respectively.
by determining the slope
When of thethe
rateplots. The calculated
of heating was
activation to 20 ◦ C per
increased energies forminute,
various these weight losses
conversion valueswere forseen
pureat temperatures
PCM, PCM of ◦ C,
299.70MWCNT,
+ 0.5%
322.28 ◦ ◦ C, and 363.34 ◦ C, respectively.
and PCMC,+342.811% MWCNT are shown in Table 1. The average activation energy calculated
using the KAS, OFW, and Starink methods for pure PCM sample is 76.12, 81.63, and 76.81
kJ/mol, respectively; for the PCM + 0.5% MWCNT sample, it is 74.93, 80.38, and 75.61
kJ/mol, respectively; for the PCM + 1% MWCNT sample, it is 62.45, 68.40, and 63.16
kJ/mol, respectively. It was found that the estimated activation values for the different
models fall into the following order: OFW > Starink > KAS. The results make it abundantly
evident that the activation energy calculated using the OFW model is significantly higher
than that computed using the other models. From the results, it is also seen that activation
energies are changed with respect to each conversion of decomposition. The main reason
for the variation in activation energy is due to the non-homogeneous structure of the se-
lected samples, or may be due to the complex reaction mechanism. In the case of pure
Energies 2023, 16, 2187 13 of 24

Figure 4c illustrates the thermogravimetric (TG) curves of D-mannitol with 1% MWCNT

at a variety of heating rates. According to the findings, the addition of MWCNT at a weight
fraction of 1% caused the sample to undergo degradation a great deal sooner than when
it was treated with pure D-mannitol or D-mannitol combined with 1% MWCNT. This
is because a higher concentration of MWCNT nanoparticles allows for a more efficient
transfer of heat, which led to the observed phenomenon. At a temperature of 244.18 ◦ C,
the thermogravimetric analysis (TG) results showed that the sample of pure D-mannitol
had lost nearly 10% of its weight. A weight loss of 20%, 40%, 60%, and 80% was recorded
at temperatures of 256.58 ◦ C, 271.24 ◦ C, 281.81 ◦ C, and 294.41 ◦ C, respectively. This
was also observed when the rate of heating was increased to 10 ◦ C per minute; weight
losses of 20%, 40%, 60%, and 80% were seen at temperatures of 268.81 ◦ C, 283.44 ◦ C,
296.05 ◦ C, and 308.67 ◦ C, respectively. In the situation of TG carried out at a heating
rate of 15 ◦ C/min, weight losses of 20%, 40%, 60%, and 80% were seen at temperatures
of 285.61 ◦ C, 301.31 ◦ C, 316.97 ◦ C, and 330.72 ◦ C, respectively. Again, when the rate of
heating was increased to 20 ◦ C per minute, these weight losses were seen at temperatures of
297.27 ◦ C, 322.13 ◦ C, 336.73 ◦ C, and 357.49 ◦ C, respectively. Because of this, it is clear from
the previous examination of the TG results that an increase in the heating rate influences
both the location of the temperature gradient (TG) curve and the rate of breakdown. The
maximum points on the TG curves go upwards toward higher temperatures as a result of
the rise in the heating rate.
As an energy barrier, activation energy works against the degrading reaction. As
seen in Figure 5, PCM samples show a wide range of conversion rates (α). Activation
energies of pure PCM, PCM + 0.5% MWCNT, and PCM + 1% MWCNT were estimated in
this experiment using iso-conversional methods described by Kissinger Akahira-Sunose,
Flynn-Wall-Ozawa, and Starink kinetic models. The activation energy (E) for complete
PCM decomposition was determined for various conversion values (0.2 to 0.8). KAS, FWO,
and Starink kinetic models are analysed by plotting the plots between ln T2 and 1000
β
T , ln β
and 1000 1000 2
β
T , ln T 1.8 , and T , respectively. The R value of the linear fitted lines of developed
kinetics models of prepared samples are more than 0.92. Figures 6–8 present these plots
for pure PCM, PCM + 0.5% MWCNT, and PCM + 1% MWCNT. As previously discussed,
the activation energy was obtained by determining the slope of the plots. The calculated
activation energies for various conversion values for pure PCM, PCM + 0.5% MWCNT, and
PCM + 1% MWCNT are shown in Table 1. The average activation energy calculated using
the KAS, OFW, and Starink methods for pure PCM sample is 76.12, 81.63, and 76.81 kJ/mol,
respectively; for the PCM + 0.5% MWCNT sample, it is 74.93, 80.38, and 75.61 kJ/mol,
respectively; for the PCM + 1% MWCNT sample, it is 62.45, 68.40, and 63.16 kJ/mol,
respectively. It was found that the estimated activation values for the different models fall
into the following order: OFW > Starink > KAS. The results make it abundantly evident
that the activation energy calculated using the OFW model is significantly higher than
that computed using the other models. From the results, it is also seen that activation
energies are changed with respect to each conversion of decomposition. The main reason
for the variation in activation energy is due to the non-homogeneous structure of the
selected samples, or may be due to the complex reaction mechanism. In the case of pure
PCM, the estimated activation energies varied in ranges of 65.89–71.79, 71.52–77.77, and
66.53–72.52 kJ/mol for the KAS, OFW, and Starink methods, respectively. For the PCM +
0.5% MWCNT sample, the estimated activation energies varied in ranges of 79.36–66.87,
84.14–73.13, and 79.97–67.62 kJ/mol for the KAS, OFW and Starink methods, respectively.
However, in the case of the PCM + 1% MWCNT sample, the estimated activation energies
varied in ranges of 71.52–52.28, 76.59–59.11, and 72.15–53.07 kJ/mol for the KAS, OFW and
Starink methods, respectively.
Energies 2023, 16, x FOR PEER REVIEW 15 of 25
Energies 2023, 16, 2187 14 of 24

Energies 2023, 16, x FOR PEER REVIEW 15 of 25

Figure5.
Figure 5. Conversion
Conversion plots
plots(a)
(a)pure
purePCM
PCM(b)
(b)PCM
PCM++ 0.5%
0.5% MWCNT
MWCNT (c)
(c) PCM
PCM ++ 1%
1% MWCNT.
MWCNT.
Figure 5. Conversion plots (a) pure PCM (b) PCM + 0.5% MWCNT (c) PCM + 1% MWCNT.

β 1000 1000 β 1000

Figure6.6.Pure
Figure PurePCM
PCMsample
sample (a)(a)
ln ln
T2
vs.vs.
T plots (b) ln
plots (b)β and T plots (c)
ln 𝛽 and ln T(c)
plots 1.8 vs.
ln T. vs. (d) Activa-
(d)
𝛽 1000 1000 𝛽 1000
tion
Figure energy
6. Pure with
PCM respect
sampleto various
(a) ln conversion.
vs. plots
Activation energy with respect to𝑇 2various𝑇 conversion.(b) ln 𝛽 and plots (c) ln vs. (d)
𝑇 𝑇 1.8 𝑇
Activation energy with respect to various conversion.
 Energies 2023, 16, x FOR PEER REVIEW 16 of 25

Energies 2023, 16, x FOR PEER REVIEW 16 of 25

Energies 2023, 16, 2187 15 of 24

𝛽 1000 1000
Figure 7. Pure PCM with 0.5 % MWCNT (a) ln vs. plots (b) ln 𝛽 and plots (c)
𝑇2 𝑇 𝑇
𝛽 1000
ln 𝑇 1.8 7.
vs.Pure (d) Activation energy with respect
β to various(b)
1000 conversion.1000 ln T1.8 vs. 1000
β
Figure 𝑇 PCM with 0.5 % MWCNT (a) ln 2 vs. T 𝛽 plots
1000 ln β and T plots (c) 1000
Figure 7. Pure PCM with 0.5 % MWCNT T(a) ln vs. plots (b) ln 𝛽 and plots T(c)
(d) Activation 𝑇2 𝑇 𝑇
𝛽 1000 energy with respect to various conversion.
ln 1.8 vs. (d) Activation energy with respect to various conversion.
𝑇 𝑇

Figure 8. Pure PCM with 1% MWCNT (a) ln T2 vs. 1000 1000

vs. 1000
β β
T plots (b) ln β and T plots (c) ln T 1.8 T
(d) Activation energy with respect to various conversion.
Energies 2023, 16, 2187 16 of 24

Table 1. Activation energy values of various PCM samples for developed kinetic models.

Activation Energy (Ea ) (kJ/mol)

Conversion (α)
KAS Method OFW Method Starink Method
Pure PCM
0.2 65.89 71.10 66.53
0.3 78.76 83.74 79.40
0.4 76.52 81.74 77.17
0.5 78.36 83.60 79.02
0.6 80.80 86.08 81.47
0.7 74.51 80.25 75.22
0.8 71.79 77.77 72.52
PCM + 0.5% MWCNT
0.2 79.36 84.14 79.97
0.3 86.36 90.97 86.97
0.4 77.64 82.82 78.29
0.5 75.47 80.92 76.15
0.6 70.06 75.90 70.77
0.7 68.74 74.79 69.47
0.8 66.87 73.13 67.62
PCM + 1% MWCNT
0.2 71.52 76.59 72.15
0.3 70.88 76.17 71.54
0.4 68.74 74.27 69.41
0.5 60.35 66.43 61.07
0.6 57.46 63.78 58.20
0.7 55.90 62.43 56.67
0.8 52.28 59.11 53.07

3.2. Kinetic Characterization of Nano Enhanced PCM Samples

D-mannitol PCM degradation kinetics after MWCNT nano-particle addition was in-
vestigated using thermogravimetric analysis data. Determining D-mannitol degradation
kinetics using a non-isothermal method was carried out because it more closely mimics
actual industrial processing conditions. The PCM samples were heated from room tempera-
ture to 700 ◦ C at continuous scanning speeds of 5, 10, 15, and 20 ◦ C/min for non-isothermal
measurements, and the resulting percentage weight loss was recorded. In this work, the
thermal degradation kinetics of D-mannitol with added MWCNT nanoparticles was in-
vestigated using iso-conversional techniques. Using the relationship between activation
energy and conversion degree, a kinetic analysis of thermal decomposition was performed.
Lower activation energy indicates that the least amount of energy is necessary to
break the PCM molecules, which in turn speeds up the rate at which the reaction occurs.
According to the findings, it can be shown that the rate of decomposition of PCM is initially
lower (that is, at 20% conversion), but then it quickly rises to 60%. This is because the
calculated activation energy of PCM has increased from 10% to 60% of the conversion
range. The rate of decomposition continues to gradually increase until it reaches a state
of rest, at which point it can reach up to 80% conversion. This is due to a slight decrease
in the activation energy value. From Table 1, it can be seen that the rate of decomposition
of nano-enhanced PCM samples are initially higher (that is, at 20% conversion), but then
Energies 2023, 16, 2187 17 of 24

it sharply decreases up to 80%. The pure PCM sample was analysed, and the activation
energy was determined to have an average value of 76.12 kJ/mol. Whereas, the calculated
average activation energy values (i.e., using the KAS model) for PCM samples containing
0.5% MWCNT and 1% MWCNT were 74.93 and 62.45 kJ/mol, respectively, which was
1.56% and 17.96% smaller than the activation energy values for the pure PCM sample.
Though this drop is rather small at low MWCNT nano-particle weight fractions, it becomes
extremely noticeable at large nano-particle weight fractions. These findings demonstrate
that MWCNT addition reduces the activation energy of pure PCM. MWCNT nanoparticles
not only operate as a textural support that improves surface area, but the are also involved in
catalysis, as evidenced by the reduction in activation energy values upon their incorporation.
As a result, MWCNT nanoparticles catalyse pure PCM samples, reducing the latter’s
activation energy. Chemical reactions simultaneously occur with evaporation, reducing the
activation energy required for the former process. Condensation appears to take place in
pure PCM, leading to the loss of water and, consequently, mass, as well as the creation of a
low volatility condensation product. Thermogravimetric analysis (TG) reveals that in the
RT−200 ◦ C operating temperature range, pure PCM and PCM added with 0.5 and 1 wt.%
of MWCNT show no appreciable weight loss. At these temperatures, evaporation hardly
contributes to the thermal degradation of the samples.

3.3. Machine Learning Modelling of Thermal Degradation of Nano Enhanced PCM Samples
In the present study, the mass loss of the samples of pure PCM and nano-enhanced
PCM was anticipated by making use of three distinct characteristics: the weight percent
of MWCNT, the heating rate, and the temperature. Python 3.10.5 software was used to
construct machine learning models on the basis of experimental data in order to make a
prediction about the mass loss of the PCM sample.

3.4. Hyper-Parameter Tuning of Developed Machine Learning Models

Finding the exact values of the hyperparameters is a vital component of developing
a successful model, especially for a parameter-sensitive algorithm like a random forest,
support vector machine, Gaussian process regression, or ANN model. To find the machine
learning model that yields the best results, various hyperparameter settings can be tested.
The optimal settings for each model’s hyperparameter values are listed in Table 2. The ANN
model makes use of three layers of feed-forward back propagation in its computations.
The output neurone reflects the mass loss of the PCM sample, while the input neurones
are substituted in for the various factors that were investigated in the experiment (such as
weight fraction of the MWCNT, heating rate, and temperature). As an activation function
that lies between the input layer and the hidden layer, the hyperbolic tangent sigmoid
(tanh) function is utilised. From the literature [16], the tanh function has shown significantly
better performance than the sigmoid function. On the other hand, the purelin function
is employed in the role of an activation function between the input layer and the hidden
layer. The method of finding the optimal number of hidden neurones in the hidden layer
consisted of multiple rounds of trial and error. In order to establish which neural network is
the most effective for this purpose, a neural independence test was carried out. The results
showed that the RMSE and R2 values were at their lowest and highest levels, respectively,
when there were 33 hidden neurones.
 Energies 2023, 16, 2187 18 of 24

Table 2. Hyper-parameter tuning values of developed machine learning models.

Model Parameter Value

kernel rbf
Support vector machine
degree 3
n-estimators 55
Random Forest Min-samples-split 3
Regression Min-samples-leaf 1
Max-depth 9
Basis function 0
Gaussian Process Regression Sigma 0.143
Kernel function Isotropic matern 3/2
Number of hidden layers 1
ANN Number of hidden neurons 33
Activation functions Tanh, Purelin

3.5. Comparison of Model Accuracy of Developed Machine Learning Models

Energies 2023, 16, x FOR PEER REVIEW 19 of 25
The variation of MSE value with respect to the iterations at optimal neural network is
shown in Figure 9. Figure 10 indicates the regression plots of optimal network topology.
The correlation coefcient (R) values for all data sets are nearly equal to one, as shown in the
distribution of points that
plots. This indicates out the
of all the models
developed because
optimum it has the
network’s least
data amount of
predictions aredeviation
consistent
around the diagonal line.
with the experimental data.

Figure 9.9.
Figure Variation
VariationofofMSE
MSEvalue
valuewith
withrespect
respecttotoepochs
epochsatatoptimal
optimalneural
neuralnetwork.
network.

A scatter plot is an accurate representation of the accuracy of the machine learning

models that were generated. The scatter distribution of both experimental and predicted
values of various machine learning models is depicted in Figure 11. Alongside the an-
ticipated values of mass loss, a regression diagonal line is presented that connects the
experimental data to the expected values. If the scattered points are grouped together in
close proximity to the diagonal line, then the model has a high R2 value and provides a
strong fit. On the other hand, if the points are scattered in a direction that is not perpendic-
ular to the diagonal line, then the model has a lower level of goodness of fit and R2 values.
When compared to the scatter distribution of the other machine learning models, as shown
in Figure 12, the scatter distribution of the Gaussian process regression and ANN machine
learning models are more concentrated near the centre line. This can be seen by looking
at the distance between the centre line and the scatter distribution. Around the diagonal
line, the scatter point distribution of the other machine learning models demonstrates a
broad or foggy dispersion. The Gaussian process regression model has the best scatter
distribution of points out of all the models because it has the least amount of deviation
around the diagonal line.
Energies 2023, 16, 2187 19 of 24
Figure 9. Variation of MSE value with respect to epochs at optimal neural network.

Energies 2023, 16, x FOR PEER REVIEW 20 of 25

Figure 10. Coefficient of correlation plot for optimal neural network.

Figure 10. Coefficient of correlation plot for optimal neural network.

Figure 11. Regression plots of trained data for developed machine learning models (a) Random
Figure 11. Regression plots of trained data for developed machine learning models (a) Random For-
Forest (b) Linear est
regression
(b) Linear(c) SVM (d)(c)
regression Gaussian
SVM (d) process
Gaussianregression (e) ANN.
process regression (e) ANN.
 Energies 2023, 16, x FOR PEER REVIEW 21 of 25

Energies 2023, 16, 2187 20 of 24

Figure
Figure 12.12.Regression
Regressionplots ofof
plots tested data
tested for
data fordeveloped
developedmachine
machine learning models
learning (a)(a)
models Random Forest
Random Forest
(b)(b)
Linear regression (c) SVM (d) Gaussian process regression (e)
Linear regression (c) SVM (d) Gaussian process regression (e) ANN. ANN.

3.6. Sensitivity Analysis

3.6. Sensitivity Analysis
The root-mean-square error (RMSE) and the correlation coefficient (R2 ) were2employed
The root-mean-square error (RMSE) and the correlation coefficient (R ) were em-
as performance indicators to measure the model’s accuracy. The best neural structure was
ployed as performance indicators to measure the model’s accuracy. The best neural struc-
found to have an RMSE of 0.8437 and an R2 of 0.9998. Similar to how RMSE and R2
ture was found to have an RMSE of 0.8437 and an R2 of 0.9998. Similar
values were used to evaluate the accuracy of the aforementioned models’topredictions,
how RMSEthe and
R 2 values were used to evaluate the accuracy of the aforementioned models’ predictions,
remaining machine learning models were also evaluated. The relative RMSE and R2 are
the remaining
shown in Table 3. machine
Table 3learning models
shows that were
all of the also evaluated.
created The relative
ML models, with theRMSE and R2
exception
ofare
theshown
linear in Table 3. Table
regression 3 shows
model, have Rthat all of above
2 values the created
0.95. ML
ANN models, with the exception
and Gaussian process
of the linear regression model, have R 2 values above 0.95.2
regression models in machine learning have the highest R and the lowest ANN and Gaussian process
RMSE. When
regression models in machine learning have the highest R 2 and the lowest RMSE. When
compared to the ANN model, the results from the Gaussian process regression model are
compared
more to theFor
promising. ANNthemodel, the results
regression modelfrombased theonGaussian process
the Gaussian regression
process, model are
the obtained
moreand
RMSE promising.
R2 valuesFor the0.8871
were regression modelrespectively.
and 0.9998, based on theAccording
Gaussiantoprocess,
the datathe obtained
in Table 3,
RMSE and R 2 values were 0.8871 and 0.9998, respectively. According to the data in Table
the Gaussian process model best describes the results of the experiments. To a high
degree, the results of the Gaussian process regression model match with the results of the
 3, the Gaussian process model best describes the results of the experiments. To a high
degree, the results of the Gaussian process regression model match with the results of the
experimental testing dataset. Thus, the non-linear relationship between the study’s output
Energies 2023, 16, 2187 and input parameters can be accurately predicted by employing the Gaussian process re-
21 of 24
gression technique.

Table 3. Sensitivity
experimental analysis
testing of developed
dataset. Thus, themachine learning
non-linear models. between the study’s output
relationship
and input parameters can be accurately predictedTesting
Training by employing the Gaussian
Validationprocess
Model
regression technique.
RMSE R2 RMSE R2 RMSE R2
Random Forest Re-
Table 3. Sensitivity analysis 2.6325
of developed machine learning
0.9983 4.2996 models.
0.9957 19.9326 0.9029
gression
Linear Regression 18.6926 Training 19.4342
0.9162 Testing
0.9154 20.0745Validation
0.8854
Model
Support vector ma- RMSE R 2 RMSE R2 RMSE R2
4.7426 0.9946 4.9398 0.9944 18.9254 0.9113
chine
RandomRegression
Forest Regression 2.6325 0.9983 4.2996 0.9957 19.9326 0.9029
ArtificialLinear
Neural Net-
Regression 18.6926 0.9162 19.4342 0.9154 20.0745 0.8854
0.8437 0.9998 1.0541 0.9997 20.1265 0.9010
Support
workvector machine 4.7426 0.9946 4.9398 0.9944 18.9254 0.9113
Regression
Gaussian Process Re-
Artificial Neural Network 0.8871 0.9998 0.9998
0.8437 1.000 1.05410.99980.9997 1.0632
20.1265 0.9997
0.9010
gression
Gaussian Process Regression 0.8871 0.9998 1.000 0.9998 1.0632 0.9997

3.7. Model Accuracy Test

3.7. Model Accuracy Test
The prediction performance of the developed machine learning models can be deter-
The prediction performance of the developed machine learning models can be deter-
mined with the help of the known TGA experimental dataset. The PCM with the 0.5%
mined with the help of the known TGA experimental dataset. The PCM with the 0.5%
MWCNT sample was heated at 7.5 °C·min−1 using TGA equipment, and the experimental
MWCNT sample was heated at 7.5 ◦ C·min−1 using TGA equipment, and the experimental
outcomes
outcomes of this sample were considered as the
of this sample were considered as validation dataset.
the validation dataset. TheThe comparison
comparison of of
validated
validated experimental TGA data and machine learning models predicted valuesvalues
experimental TGA data and machine learning models predicted are
are shown
shown in 13.
in Figure Figure
From13.Figure
From13Figure
shows13that
shows that the Gaussian
the Gaussian regressionregression model
model is well is well
predicted
predicted with the lowest RMSE value of 1.0632 and the highest R
with the lowest RMSE value of 1.0632 and the highest R2 value of 0.9997.
2 value of 0.9997.

Figure 13.
Figure Comparison
13.Comparison of experimental
of experimental TGA curves
TGA curves with developed
with developed predictedpredicted machine
machine learning
learning models.
models.

4. Conclusions
4. Conclusions
In this study, TGA analysis of samples of pure PCM and nano-enhanced PCM samples
In this study, TGA analysis of samples of pure PCM and nano-enhanced PCM sam-
were carried out using a variety of heating rates. In order to evaluate the thermal degrada-
ples were carried out using a variety of heating rates. In order to evaluate the thermal
tion kinetics of the proposed PCM samples, various degradation kinetic models such as
degradation kinetics of the proposed PCM samples, various degradation kinetic models
KAS, FWO, and Starink models were employed. Using these kinetic models, the activation
such as KAS, FWO, and Starink models were employed. Using these kinetic models, the
energy of the prepared samples was determined. The Ozawa, KAS, and Starink techniques
all produced activation energies for pure PCM that varied between 66.53 to 72.52 kJ/mol,
79.36 to 66.87 kJ/mol, and 71.10 to 77.77 kJ/mol, respectively. The activation energies
of NEPCM (1% MWCNT) were 76.59 to 59.11, 71.5 to 52.28, and 72.15 to 53.07 kJ/mol.
According to the results, it could be observed that the prepared nano-enhanced PCM sam-
ples were chemically and thermally stable. In addition, several machine learning models
were developed to predict the thermal degradation of the chosen PCM samples. These
Energies 2023, 16, 2187 22 of 24

included linear regression, support vector regression, random forest, Gaussian process
regression, and artificial neural network models. The generated models could accurately
predict the experimental TGA dataset of PCM samples across a wide range of operating
conditions. In summary, the findings of this work add to a better understanding of the
thermal degradation that nano-enhanced PCM undergoes when it is used for thermal
storage applications.

Author Contributions: R.K.K.—Conceptualization, methodology, software, validation, formal anal-

ysis, investigation, resources, data curation, writing—original draft preparation, writing—review
and editing, visualization, supervision, project administration. B.K.C.—Investigation, resources,
data curation, writing—original draft preparation, writing—review and editing, visualization.
S.M.—Conceptualization, resources, data curation, writing—review and editing, visualization.
R.S.—Investigation, Resources, Software, writing—review and editing, visualization, formal anal-
ysis. V.C.K.—Conceptualization, resources, data curation, writing—review and editing, visualiza-
tion. S.B.A.—Methodology, software, formal analysis, data curation, writing—review and editing.
M.C.—Conceptualization, methodology, software, validation, Investigation, resources, data curation,
writing—original draft preparation, writing—review and editing. All authors have read and agreed
to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Jinshah, B.S.; Ramaraj, B.K.; Kottala, R.K.; Sivapirakasam, S.P. A complete numerical model for low temperature composite
form-stable phase change material slab based on dynamically simplified temperature transforming method. Energy Sources A
Recovery Util. Environ. Eff. 2021, 1–26. [CrossRef]
2. Balasubramanian, K.R.; Jinshah, B.S.; Ravikumar, K.; Divakar, S. Thermal and hydraulic characteristics of a parabolic trough
collector based on an open natural circulation loop: The effect of fluctuations in solar irradiance. Sustain. Energy Technol. Assess.
2022, 52, 102290. [CrossRef]
3. Jinshah, B.S.; Kottala, R.K.; Balasubramanian, K.R.; Francis, A. Experimental analysis of phase change material integrated
single-phase natural circulation loop. Mater. Today Proc. 2021, 46, 10000–10005. [CrossRef]
4. Jinshah, B.S.; Balasubramanian, K.R.; Kottala, R.; Divakar, S. Influence of power step on the behavior of an Open Natural
Circulation Loop as applied to a Parabolic Trough Collector. Renew. Energy 2022, 181, 1046–1061. [CrossRef]
5. Marudaipillai, S.K.; Karuppudayar Ramaraj, B.; Kottala, R.K.; Lakshmanan, M. Experimental study on thermal management and
performance improvement of solar PV panel cooling using form-stable phase change material. Energy Sources A Recovery Util.
Environ. Eff. 2020, 45, 160–177. [CrossRef]
6. Senthilkumar, M.; Balasubramanian, K.R.; Kottala, R.K.; Sivapirakasam, S.P.; Maheswari, L. Characterization of form-stable
phase-change material for solar photovoltaic cooling. J. Therm. Anal. Calorim. 2020, 141, 2487–2496. [CrossRef]
7. Goud, M. A comprehensive investigation and artificial neural network modeling of shape stabilized composite phase change
material for solar thermal energy storage. J. Energy Storage 2022, 48, 103992. [CrossRef]
8. Goud, M.; Raval, F. A sustainable biochar-based shape stable composite phase change material for thermal management of a
lithium-ion battery system and hybrid neural network modeling for heat flow prediction. J. Energy Storage 2022, 56, 106163.
[CrossRef]
9. Peter, R.J.; Balasubramanian, K.R.; Kumar, K.R. Comparative study on the thermal performance of microencapsulated phase
change material slurry in tortuous geometry microchannel heat sink. Appl. Therm. Eng. 2023, 218, 119328. [CrossRef]
10. Luo, J.; Zou, D.; Wang, Y.; Wang, S.; Huang, L. Battery thermal management systems (BTMs) based on phase change material
(PCM): A comprehensive review. J. Chem. Eng. 2022, 430, 132741. [CrossRef]
11. Karthick, A.; Manokar Athikesavan, M.; Pasupathi, M.K.; Manoj Kumar, N.; Chopra, S.S.; Ghosh, A. Investigation of inorganic
phase change material for a semi-transparent photovoltaic (STPV) module. Energies 2020, 13, 3582. [CrossRef]
12. Ye, R.; Fang, X.; Zhang, Z. Numerical Study on Energy-Saving Performance of a New Type of Phase Change Material Room.
Energies 2021, 14, 3874. [CrossRef]
13. Hua, W.; Xu, X.; Zhang, X.; Yan, H.; Zhang, J. Progress in corrosion and anti-corrosion measures of phase change materials in
thermal storage and management systems. J. Energy Storage 2022, 56, 105883. [CrossRef]
14. Bell, S.; Jones, M.W.M.; Graham, E.; Peterson, D.J.; van Riessen, G.A.; Hinsley, G.; Will, G. Corrosion mechanism of SS316L
exposed to NaCl/Na2 CO3 molten salt in air and argon environments. Corros. Sci. 2022, 195, 109966. [CrossRef]
Energies 2023, 16, 2187 23 of 24

15. Moreno, P.; Miro, L.; Sole, A.; Barreneche, C.; Sole, C.; Martorell, I.; Cabeza, L.F. Corrosion of metal and metal alloy containers
in contact with phase change materials (PCM) for potential heating and cooling applications. Appl. Energy 2014, 125, 238–245.
[CrossRef]
16. Ravi Kumar, K.; Balasubramanian, K.R.; Jinshah, B.S.; Abhishek, N. Experimental analysis and neural network model of MWCNTs
enhanced phase change materials. Int. J. Thermophys. 2022, 43, 11. [CrossRef]
17. Kottala, R.K.; Ramaraj, B.K.; BS, J.; Vempally, M.G.; Lakshmanan, M. Experimental investigation and neural network modeling of
binary eutectic/expanded graphite composites for medium temperature thermal energy storage. Energy Sources A Recovery Util.
Environ. Eff. 2022, 1–24. [CrossRef]
18. Png, Z.M.; Soo, X.Y.D.; Chua, M.H.; Ong, P.J.; Suwardi, A.; Tan, C.K.I.; Zhu, Q. Strategies to reduce the flammability of organic
phase change Materials: A review. Sol. Energy 2022, 231, 115–128. [CrossRef]
19. Saleel, C.A. A review on the use of coconut oil as an organic phase change material with its melting process, heat transfer, and
energy storage characteristics. J. Therm. Anal. Calorim. 2022, 147, 4451–4472. [CrossRef]
20. Tyagi, V.V.; Chopra, K.; Sharma, R.K.; Pandey, A.K.; Tyagi, S.K.; Ahmad, M.S.; Kothari, R. A comprehensive review on phase
change materials for heat storage applications: Development, characterization, thermal and chemical stability. Sol. Energy Mater.
Sol. Cells 2022, 234, 111392. [CrossRef]
21. Ma, C.; Wang, J.; Wu, Y.; Wang, Y.; Ji, Z.; Xie, S. Characterization and thermophysical properties of erythritol/expanded graphite
as phase change material for thermal energy storage. J. Energy Storage 2022, 46, 103864. [CrossRef]
22. Ramaraj, B.K.; Kottala, R.K. Preparation and characterisation of binary eutectic phase change material/activated porous bio
char/multi walled carbon nano tubes as composite phase change material. Fuller. Nanotub. Carbon Nanostructures 2022, 31, 75–89.
[CrossRef]
23. Yan, X.; Zhao, H.; Feng, Y.; Qiu, L.; Lin, L.; Zhang, X.; Ohara, T. Excellent heat transfer and phase transformation performance of
erythritol/graphene composite phase change materials. Compos. B Eng. 2022, 228, 109435. [CrossRef]
24. Wang, K.; Sun, C.; Biney, B.W.; Li, W.; Al-shiaani, N.H.; Chen, K.; Guo, A. Polyurethane template-based erythritol/graphite foam
composite phase change materials with enhanced thermal conductivity and solar-thermal energy conversion efficiency. Polymer
2022, 256, 125204. [CrossRef]
25. Lin, W.; Lai, J.; Xie, K.; Liu, D.; Wu, K.; Fu, Q. D-Mannitol/Graphene Phase-Change Composites with Structured Conformation
and Thermal Pathways Allow Durable Solar–Thermal–Electric Conversion and Electricity Output. ACS Appl. Mater. Interfaces
2022, 14, 38981–38989. [CrossRef]
26. Jame, L.; Perumal, S.; Moorthy, M.; Kumar, J.R. Nanocomposites of GO/D-Mannitol Assisted Thermoelectric Power Generator
for Transient Waste Heat Recovery. J. Nanomater. 2022, 2022, 2460364. [CrossRef]
27. Balasubramanian, K.R.; Ravi Kumar, K.; Sathiya Prabhakaran, S.P.; Jinshah, B.S.; Abhishek, N. Thermal degradation studies
and hybrid neural network modelling of eutectic phase change material composites. Int. J. Energy Res. 2022, 46, 15733–15755.
[CrossRef]
28. Venkitaraj, K.P.; Suresh, S. Experimental thermal degradation analysis of pentaerythritol with alumina nano additives for thermal
energy storage application. J. Energy Storage 2019, 22, 8–16. [CrossRef]
29. Sun, L.; Qu, Y.; Li, S. Co-microencapsulate of ammonium polyphosphate and pentaerythritol and kinetics of its thermal
degradation. Polym. Degrad. Stab. 2012, 97, 404–409. [CrossRef]
30. Fabiani, C.; Pisello, A.L.; Barbanera, M.; Cabeza, L.F. Palm oil-based bio-PCM for energy efficient building applications:
Multipurpose thermal investigation and life cycle assessment. J. Energy Storage 2020, 28, 101129. [CrossRef]
31. Xiang, L.; Luo, D.; Yang, J.; Sun, X.; Qi, Y.; Qin, S. Preparation and comparison of properties of three phase change energy storage
materials with hollow fiber membrane as the supporting carrier. Polymers 2019, 11, 1343. [CrossRef]
32. Cheepu, M. Machine Learning Approach for the Prediction of Defect Characteristics in Wire Arc Additive Manufacturing. Trans.
Indian Inst. Met. 2022, 13, 1–9. [CrossRef]
33. Kumar, C.B.; Anandakrishnan, V. Experimental investigations on the effect of wire arc additive manufacturing process parameters
on the layer geometry of Inconel 825. Mater. Today Proc. 2020, 21, 622–627. [CrossRef]
34. Chigilipalli, B.K.; Veeramani, A. An experimental investigation and neuro-fuzzy modeling to ascertain metal deposition parame-
ters for the wire arc additive manufacturing of Incoloy 825. CIRP J. Manuf. Sci. Technol. 2022, 38, 386–400. [CrossRef]
35. Kumar, K.R.; Balasubramanian, K.R.; Kumar, G.P.; Bharat Kumar, C.; Cheepu, M.M. Experimental Investigation of Nano-
encapsulated Molten Salt for Medium-Temperature Thermal Storage Systems and Modeling of Neural Networks. Int. J.
Thermophys. 2022, 43, 145. [CrossRef]
36. Ozveren, U.; Kartal, F.; Sezer, S.; Ozdogan, Z.S. Investigation of steam gasification in thermogravimetric analysis by means of
evolved gas analysis and machine learning. Energy 2022, 239, 122232. [CrossRef]
37. Wang, S.; Shi, Z.; Jin, Y.; Zaini, I.N.; Li, Y.; Tang, C.; Yang, W. A machine learning model to predict the pyrolytic kinetics of
different types of feedstocks. Energy Convers. Manag. 2022, 260, 115613. [CrossRef]
38. Balsora, H.K.; Kartik, S.; Dua, V.; Joshi, J.B.; Kataria, G.; Sharma, A.; Chakinala, A.G. Machine learning approach for the prediction
of biomass pyrolysis kinetics from preliminary analysis. J. Environ. Chem. Eng. 2022, 10, 108025. [CrossRef]
39. Wei, H.; Luo, K.; Xing, J.; Fan, J. Predicting co-pyrolysis of coal and biomass using machine learning approaches. Fuel 2022, 310,
122248. [CrossRef]
40. Kissinger, H.E. Reaction kinetics in differential thermal analysis. Anal. Chem. 1957, 29, 1702–1706. [CrossRef]
Energies 2023, 16, 2187 24 of 24

41. Ozawa, T. A new method of analyzing thermogravimetric data. Bull. Chem. Soc. Jpn. 1965, 38, 1881–1886. [CrossRef]
42. Starink, M.J. A new method for the derivation of activation energies from experiments performed at constant heating rate.
Thermochim. Acta 1996, 288, 97–104. [CrossRef]
43. Mavromatidis, L.E.; Bykalyuk, A.; Lequay, H. Development of polynomial regression models for composite dynamic envelopes’
thermal performance forecasting. Appl. Energy 2013, 104, 379–391. [CrossRef]
44. Ahmad, M.W.; Reynolds, J.; Rezgui, Y. Predictive modelling for solar thermal energy systems: A comparison of support vector
regression, random forest, extra trees and regression trees. J. Clean. Prod. 2018, 203, 810–821. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.