Edelweiss Applied Science and Technology 
ISSN: 2576-8484 
Vol. 9, No. 4, 3082-3093 
2025 
Publisher: Learning Gate 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

© 2025 by the author; licensee Learning Gate 
History: Received: 10 February 2025; Revised: 22 April 2025; Accepted: 25 April 2025; Published: 30 April 2025 
* Correspondence:  kermezli.tayeb@univ-medea.dz 

 
 
 
 
 

Prediction by artificial neural network of insulation performance of eco-
treated cork stoppers: Experimental measurement, modeling and 
optimization 

 
Tayeb Kermezli1*, Mohamed Announ1, Aboubakr Boukrida1, Mustapha Douani2 
1Materials and Environment Laboratory, Faculty Tech. Medea University, Algeria; kermezli.tayeb@univ-medea.dz (T.K.)  
announ.mohamed@univ-medea.dz (M.A.) boukridaaboubakr6@gmail.com (A.B.) 
2LCVVE, Faculty Tech. Univ.HB, Chlef, Algeria; douani_mustapha@yahoo.com (M.D.)  

 

 

Abstract: This study aims to predict by artificial neural networks (ANN) the improvement in mass 
insulation of cork stoppers treated by high temperature thermal (HTT) and/or boiling. Experimental 
tests have shown that the desorption kinetics are more favorable for smaller molecules DKCl < DNaCl. 
The results validated the developed mathematical model, which accounted for the actual cylindrical 
shape of the stopper, and quantified the improvement in apparent diffusion coefficients as a function of 
the maximum temperature of the treatment cycle: D105°<D200°<D350° ≈ D450° and the protocol type: 
DA<DB<DC. The results revealed a positive correlation between temperature and the diffusion 
phenomenon, with a significant influence observed up to 350 °C. Furthermore, to enhance the accuracy 

of Dapp, the Bat Algorithm optimization method was applied, achieving a precision of the order of 10⁻⁵. 
An experimental database, composed of 3864 points, was previously optimized to be integrated into an 
artificial neural network (ANN) model with a specific architecture (5-5-6-1). The model thus developed 
demonstrated remarkable reliability, displaying a coefficient of correlation R² of 0.9997 and an 

extremely low root mean square error (RMSE), evaluated at 7.38 × 10⁻¹⁴. These performances underline 
the robustness and accuracy of the proposed model for the prediction of the studied phenomenon. 

Keywords: ANN, Cork, Mass diffusion, Modeling, Optimization, THT. 

 
1. Introduction  

The study of fluid flow in heterogeneous and anisotropic porous media, such as cork stoppers, 
highlights the importance of material management and selection for insulation purposes [1-3]. 
Traditionally, this management has focused on the mechanical requirements of the materials [4]. This 
research concentrates on improving the insulation performance of cork stoppers through an eco-friendly 
high-temperature and boiling treatment, to make them competitive with other polymeric materials [3]. 
Cork is primarily extracted from the bark of the cork oak (Quercus suber), from Mediterranean tree 
species. This natural material has diverse applications in many industrial sectors, including the wine 
industry (bottle stoppers [5]), construction (thermal and acoustic insulation [6, 7]), fashion (accessories 
and footwear [8]), and the manufacture of decorative items and everyday objects. Its versatility and 
unique properties make it a sought-after renewable resource, meeting the demands of ecology, 
sustainability, and innovation in these fields. 

Cork is characterized by marked anisotropy, making it a highly complex material. Furthermore, it 
exhibits a hygroscopic nature: its water content changes dynamically depending on humidity conditions, 
temperature, environment, and soil properties, thus influencing its physicochemical properties [9]. 
These variations in water content can also affect its mechanical performance and long-term behaviour, 
particularly in structural or insulation applications. 



3083 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

Optimizing the properties of certain porous biomaterials, as well as their preservation, relies 
primarily on chemical processes [10]. These methods also include high-temperature heat treatments, 
which are increasingly important in the fields of thermal and acoustic insulation, particularly with a 
view to energy efficiency and environmental sustainability. It should be noted that applying a high-
temperature heat treatment to a porous biomaterial, such as cork, induces significant changes in its 
physical and chemical properties. These transformations give the material increased biological 
resistance to decay, while potentially improving its mechanical properties and dimensional stability. 
Furthermore, these heat treatments can also modify the microstructure of cork, reducing its porosity 
and increasing its bulk density, which may influence its performance as an insulating material. These 
modifications thus open up interesting prospects for the use of cork in innovative applications, while 
meeting growing demands for sustainability and environmental friendliness. 

The enhancement is quantified by determining the apparent diffusion coefficient (Dapp) for different 
chemical species (NaCl or KCl) using conductometric methods [11] through various treatment cycles: 
105°C, 200°C, 350°C, and 450°C, in contrast to other studies using pure water or gas [12]. 
Additionally, this study aims to extend the work of Seibert, et al. [13] and Brazinha, et al. [12] by 
comparing the natural and treated states of cork. 

The goal is to develop a competitive biomaterial for insulation by improving its mechanical 
structure through various parameters, including treatment temperature, treatment protocol, and the 
nature of the solution (NaCl and KCl) [1] as there are other parameters, such as the cork plantation 
area, to consider [14, 15]. By employing conductometric techniques, experimental results in transient 
modes are calibrated using a developed mathematical model to determine the apparent diffusion 
coefficient (Dapp), according to the algorithm in figure 1. It should be noted that the effect of the 
treatment on cork is directly related to the value of Dapp. Given the numerous parameters influencing 
the insulating properties of the material, the analysis is refined using an artificial neural network (ANN) 
model, which will enable the prediction of various material properties without requiring new practical 
experiments for each sample [16]. 
 

Selection of samples according to 

class, form and direction

Choice of protocol (THT with boiling) 

and treatment cycle 

Experimental 

conductimetric method
Modeling of 

reduced mass

Determination of 

reduced mass

Model calibration with experimental 

data for minimal error

 Determination of Dapp

Select the nature of the solution 

NaCl or KCl

 
Figure 1. 
Organigramme de détermination du Dapp. 



3084 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

2. Materials and Methods 
Cork stopper samples were collected from the "EPE Jijel. Cork Waterproofing SPA", Skikda region 

in Algeria. They were cut from the same cork slab in the same direction (tangential) and to the same 
dimensions (Figure 2).  

The samples were treated by boiling and/or high temperature treatment (HTT). The THT is under 
inert gas (argon). A programmable furnace of the NABERTHERM type “MORE THAN HEAT 30-
3000°C” from the mechanical laboratory “Military Polytechnic School” in Bordj el-Bahri, Algiers, 
Algeria was used, according to the cycles described in Figure 3, following the protocols specified 
according to Table 1.  

The structural modifications of cork were characterized by Fourier transform infrared spectroscopy 
(FTIR), carried out in the materials and environment laboratory of the University of Medea [17] as 
well as by scanning electron microscopy (SEM) using a Quanta 250 device from the FEI company, 
available in the Scientific and Technical Research Center in Physico-Chemical Analysis EXPERTISE, 
located in Bou-Ismail, Tipaza, Algeria [18, 19]. These techniques made it possible to obtain detailed 
information on the chemical composition and surface morphology of cork, thus facilitating the analysis 
of structural modifications induced by thermal treatments [20, 21]. 
 

Longitudinal

Radial

Base surface

 
Figure 2. 
Photo of the caps used. 

 

0 25 50 75 100 125 150 175

0

50

100

150

200

250

300

350

400

450

500

550

Heating speed at 5°C.min-1

450°C

350°C

200°C

T
H

T
 c

y
cl

e 
te

m
p
er

at
u

re
 (

°C
)

Time (min)

 Profile of the THT cycle

 THT cycle at 105°C

 THT cycle at 200°C

 THT cycle at 350°C

 THT cycle at 450°C

105°C

Heating speed at 2°C.min-1

Thermal homogenization

 
Figure 3. 
Different THT cycles at different heating rates. 

 



3085 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

Table 1. 
Different variations of the experimental protocol. 

Treatment Protocol A Protocol B Protocol C 

1st step Boiling THT THT 

2nd step THT Boiling // 

 
The conductimetric method was used to monitor the reduction in mass of the samples, while the 

desorption kinetics were tracked through conductimetric analysis of aqueous solutions. Factors such as 
treatment temperature, treatment protocol, and the nature of the diffusing species were considered for 
each test. 
 
2.1. Modeling 

To model the diffusion through the lateral surface of the cylindrical sample simplifying assumptions 
are adopted to facilitate the mathematical description of the phenomenon. This modelling is based on 
the establishment of a mass balance of the diffusing chemical species (i) through a differential volume 
element (dv) of the composite material. The mass balance is expressed after abstraction of any chemical 
reaction as well as the negligible contribution of convection, the mass balance equation is reduced to: 
  

      (1) 
who gives 

                                                           (2) 

The partial differential equation from the mass transfer balance is solved by adopting simplifying 
assumptions, as well as boundary and initial conditions adapted to the geometry of the cylinder sample 

characterized by a shape factor β=1. This approach allows the equation to be reduced to a system 
expressed in cylindrical coordinates. The final expression of the diffusion model, taking into account the 
real cylindrical shape of the sample, is given by: 

                (3) 

Where: mt is the mass of the substance released at time t, m∞ is the mass of the substance transferred 
after complete desorption of the cork at infinite time, R is the radius of the stopper along the diffusion 
direction, and Diapp is the apparent diffusion coefficient of the chemical species. 

The mass diffusivity DX (cm2s-1) of the ion and cation X in water can be calculated using the Nernst-
Einstein equation [22]: 

For potassium chloride (KCl) :    (4) 

For sodium chloride (NaCl) :      (5) 

Where: R = 8.3143 [J(mole.K)-1] is the ideal gas constant, F = 96488 Coulomb is the Faraday 

constant,   is the limiting equivalent conductivity of the anion and/or cation in water at 25°C, T is the 
temperature of the water (in Kelvin), and ZX is the absolute value of the algebraic charge of the anion 
and/or cation. 

This observation confirms the findings of Langford, et al. [23] which highlight the significant 
influence of molecular size on diffusion rates. More specifically, it highlights that the solute KCl, with a 
smaller molecular size, has a higher diffusion rate than NaCl. This difference can be explained by the 

smaller ionic radius of potassium (K⁺) compared to that of sodium (Na⁺), thus favoring an increased 



3086 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

mobility of KCl in the diffusion medium [24]. These conclusions are in perfect agreement with the 
theoretical predictions described by equations 2 and 3, which confirm the inverse relationship between 
the size of the ions and their diffusion rate. 
 
2.2. Implementation of Artificial Neural Networks (ANN) 

In this study, we developed a methodology to estimate the diffusion rate in cork under the 
operational conditions detailed in the experimental section, employing a statistical approach based on 
artificial neural networks (ANN). This predictive technique was chosen for its balance between 
simplicity and efficiency, as highlighted by Zhang, et al. [25]. To achieve this goal, we constructed an 
extensive experimental database comprising 3864 data points that encapsulate the primary operating 
parameters. However, this size must meet the requirements of the theoretical relationships given by 
Shalev-Shwartz and Ben-David [26] and Aggarwal [27]. 

The minimum database size required to train an artificial neural network (ANN) is estimated based 
on the number of model parameters. The total number of parameters Nparam is given by: 

      (6) 

Where ni represents the number of neurons in layer i. A rule of thumb suggests that the database size 
Ndata should satisfy: 

                    (7) 

With k generally between 5 and 10. This approach minimizes the risk of overfitting while ensuring 
effective model generalization. 

These parameters include the time, the the nature of the diffusing species, the treatment protocol, 
and the maximum temperature of the thermal treatment (THT) cycle. This database will serve as a basis 
for the design of an optimal ANN architecture by determining the most appropriate number of hidden 
layers and neurons to accurately predict the apparent diffusion coefficient Dapp. 
 

3. Results and Discussion 
Figures 4 and 5 illustrate that the proposed model effectively captures the experimental behavior of 

the reduced mass under the influence of various parameters, including the maximum temperature of the 
thermal treatment cycle, the treatment protocol and the nature of the aqueous solution. 

Figure 4 reveals a significant dependence of the apparent diffusion coefficient (Dapp) on the 
maximum temperature of the thermal heat treatment (THT), particularly up to 350 °C. Beyond this 
threshold, however, the temperature no longer exerts a measurable influence on the diffusion kinetics of 
the chemical species (NaCl), as evidenced by the convergence of Dapp values: D105°<D200°<D350° ≈ D450°. 

This temperature-dependent behavior is further corroborated by the infrared (IR) spectroscopy 
analysis presented in Figure 6, which highlights the structural modifications induced by the thermal 
treatment. Specifically, the observed shifts in peak intensities in the IR spectra provide direct evidence of 
changes in the chemical structure of the material, consistent with the alterations in diffusion properties. 

Additionally, the nature of the diffusing species (NaCl and KCl) plays a pivotal role in determining 
the mass transfer rate, with the degree of influence directly correlated to molecular size of the chemical 
species. This relationship is illustrated in Figure 5, where the DKCl for KCl consistently exceeds that 
DNaCl of NaCl, reflecting the differential mobility of the two species. 

Furthermore, the timing of the THT protocol—whether applied before or after boiling, or as a 
standalone treatment significantly affects the mass isolation efficiency. As shown in the corresponding 
figure, the protocol involving THT prior to boiling (Protocol B) yields the highest Dapp value, with the 
following hierarchy: DB > DC > DA. These findings underscore the critical importance of optimizing 
treatment protocols to achieve enhanced mass transfer performance. 
 



3087 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

0 20 40 60 80 100 120 140 160

0,00

0,05

0,10

0,15

0,20

0,25

0,30

0,35

0,40

0,45

0,50

0,55

0,60

D450°C
=5,81.10-13m2s-1

D350°C
=5,97.10-13m2s-1

D200°C
=2,17.10-12m2s-1

C
on

ce
nt

ra
ti

on
, C

i, 
t (

m
M

)

Time (min)

 Experimental data at 105°C)

 Model at 105°C

 Experimental data at 200°C

 Model at 200°C

 Experimental data at 350°C

 Model at 350°C

 Experimental data at 450°C

 Model at 450°C D105°C
=5,41.10-12m2s-1

 
Figure 4. 
Validation of the diffusivity model with experimentation under the 
effect of THT temperature. 

 

 
Figure 5. 
Validation of the diffusivity model with experimentation under 
the effect of the nature of the solute. 

 



3088 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

 
Figure 6. 
IR spectra of treated and native cork stopper. 

 
Following model calibration using experimental data on reduced mass obtained by the 

conductometric method, the resulting apparent diffusion coefficients (Dapp) were optimized using the Bat 
algorithm [28]. This optimization process achieved a high level of accuracy, with a margin of error of 

the order of 10⁻⁵, as detailed in Table 2. The Bat algorithm, inspired by the echolocation behavior of 
bats, was chosen for its efficiency in solving complex optimization problems and its ability to converge 
quickly to optimal solutions. The strong agreement between model predictions and experimental data 
underlines the robustness of the proposed approach to accurately describe the diffusion phenomena 
studied. Furthermore, the minimal margin of error highlights the reliability of the calibration process 
and the suitability of the Bat algorithm for refining the model parameters. These results not only 
validate the theoretical framework but also provide a solid basis for further investigations into the 
influence of other variables on mass transfer processes. The optimized Dapp values, presented in Table 
2, provide an essential reference for further analyses and applications in related fields. 
 
Table 2. 
Diffusion coefficients in a cork stopper under the effect of the protocol, the temperature of the THT and the nature of the 
solute. 

Protocol A B C 
Protocol with NaCl 

 
5,41.10-12 6,05.10-12 5,80.10-12 

 
2,17.10-12 2,63.10-12 2,91.10-12 

 
5,9.10-13 8,33.10-12 7,14.10-12 

 
5,81.10-13 7,42.10-12 6,23.10-12 

Protocol with  
KCl 

 
8,12.10-12 8,97.10-12 8,61.10-12 

 
2,9.10-12 3,71.10-12 3,41.10-12 

 
7,08.10-13 9,32.10-12 9,90.10-12 

 
7,12.10-13 9,04.10-12 7,39.10-12 

 
The learning model developed in this study is based on a multi-layer perceptron (MLP) 

architecture, designed to accurately predict Dapp based on the various experimental parameters in Table 



3089 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

2. To identify the optimal neural network architecture, a series of systematic tests was performed, as 
detailed in Table 3. These tests consisted of evaluating different network configurations, varying the 
number of hidden layers (from one to three) and the number of neurons per layer (from 5 to 15). This 
methodical approach allowed us to explore a wide range of combinations to determine the structure 
offering the best predictive performance while avoiding overfitting. 

Table 3 summarizes the key parameters involved in identifying the optimized model, obtained using 
ANN. This configuration made it possible to capture the nonlinear relationships between experimental 
parameters and Dapp values, while minimizing prediction errors. 
 
Table 3. 
Proposed ANN model structure. 

Network type Algorithm Input layer Hidden layer de (1 à 3) Output layer 
Feed-Forward Backpropagation  

Levenberg-Marquardt 
(trainlm) 

Neurons 
number 

Neurons Activation 
function for 

each 

Neurons 
Activation 
function Neural Network  

Number for 
each 

number 

(FFFBP NN, newff) 5 For 5 to 15 
ReLU 

(poslin) 
1 purelin 

 
The graphs presented in Figure 7 illustrate that the implementation of the artificial neural network 

(ANN) is both satisfactory and robust.  
This performance is attributable to the use of a substantial database, comprising 3864 points, which 

represents 529.3% of the recommended minimum size of 730 points calculated from equations 3 and 6 
for the specific architecture (5-5-6-1), illustrated in Figure 8.  

This conclusion is supported by correlation coefficients close to 1 (0.9998) for the different data 
analysis phases (training, validation, and testing), as well as by an extremely low root mean square error 

(RMSE) of around 7.38 × 10⁻¹⁴. 
 



3090 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

 
Figure 7. 
Regression performance output vs. Target with linear fit for different data subsets. 

 

 
Figure 8. 
Architecture of the neural network. 



3091 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

These results demonstrate that the model is capable of capturing the complex relationships between 
input and output variables with high accuracy, while generalizing effectively to new data. Furthermore, 
the model performance, characterized by such a low RMSE, highlights the effectiveness of the 
optimization of the RNA architecture and the relevance of the selected parameters. These elements 
confirm the reliability of the proposed model and its potential for practical applications requiring high 
predictive accuracy. 

Figure 9 illustrates that the gradient decreases globally over the course of iterations (epochs), 
indicating that the optimization process is converging toward a minimum of the cost function. The 

variations in μ reflect the dynamic adaptation of the algorithm. When μ decreases, the algorithm places 
greater trust in the local gradient trends, suggesting improved convergence. Furthermore, the absence 
of fluctuations in the validation phase demonstrates the network's strong generalization capability, even 
with a relatively low number of epochs (15). 
 

 
Figure 9. 
Evolution of Gradients, Learning Rate (mu), and Validation Values per Epoch. 

 

4. Conclusion  
This study highlighted the significant impact of high-temperature thermal treatments (THT) and 

boiling processes on the mass insulation properties of cork stoppers. We successfully modeled the 
desorption kinetics, with experimental results validating the developed model, showing improved 
diffusion coefficients with increasing treatment temperatures, particularly beyond 350°C: 
D105°<D200°<D350°≈ D450°=5,81.10-13m2s-1.  The findings confirmed that molecular size affects diffusion, 
with DKCl > DNaCl. Additionally, the study emphasized the crucial role of the treatment protocol on 
insulation performance, demonstrating that the sequence of THT and boiling directly influences mass 
insulation: DA<DC<DB. The Dapp values were optimized using the Bat Algorithm, achieving a precision 



3092 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

of 10⁻⁵. Through these improvements, including an innovative treatment protocol and optimized cycle 
temperature, the final product (cork stopper) becomes highly competitive in the market. 

The database used for neural network modeling allowed the definition of an optimal architecture (5-
5-6-1), achieving exceptional accuracy, with a coefficient of determination R² of 0.9997 and an extremely 

low root mean square error (RMSE) of 7.38 × 10⁻¹⁴. This remarkable performance is partly explained by 
the substantial size of the experimental database, comprising 3864 data points, or 529.3% of the 
recommended size limit for this specific architecture. The results obtained confirm the reliability and 
effectiveness of the developed model, which positions itself as a powerful predictive tool for analyzing 
diffusion properties as a function of various parameters. 

This study not only contributes to a better understanding of cork stopper processing processes, but 
also establishes a solid methodological framework for future work aimed at optimizing material quality 
and production methods in the cork industry. The successful application of a neural network in this 
context highlights its potential for predicting complex material behaviors, opening up innovative 
perspectives in materials science and quality control. These advances underscore the importance of 
artificial intelligence-based approaches to addressing current industrial and scientific challenges. 
 

Transparency:  
The authors confirm that the manuscript is an honest, accurate,  and  transparent  account  of  the  
study; that  no  vital  features  of  the  study  have  been  omitted;  and  that  any  discrepancies  from  
the  study  as planned have been explained. This study followed all ethical practices during writing. 
 

Copyright: 
© 2025 by the authors. This open-access article is distributed under the terms and conditions of the 
Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). 
 

References 
[1] C. Leite and H. Pereira, "Cork-containing barks—A review," Frontiers in Materials, vol. 3, p. 63, 2017.  

https://doi.org/10.3389/fmats.2016.00063 

[2] J. A. Paulo and D. I. Santos, "Virgin cork colour and porosity as predictors for secondary cork industrial quality," 

Industrial Crops and Products, vol. 205, p. 117513, 2023.  https://doi.org/10.1016/j.indcrop.2023.117513 

[3] M. E. Rosa and M. A. Fortes, "The cellular structure of cork," Journal of Materials Science, vol. 23, no. 3, pp. 879–885, 

1988.  https://doi.org/10.1007/BF01153981 

[4] H. Pereira, Cork: Biology, production, and uses. Netherlands: Elsevier, 2007. 

[5] M. S. Pino, I. M. Alconchel, and P. V. Castro, "Manufacture of sustainable cork stoppers for spirits through 

compression molding," presented at the International Joint Conference on Mechanics, Design Engineering & 

Advanced Manufacturing, Cham: Springer Nature Switzerland, 2024. 

[6] Ö. Yay, M. Hasanzadeh, S. F. Diltemiz, and S. Gürgen, Cork agglomerates in acoustic insulation. In Cork-Based Materials 

in Engineering: Design and Applications for Green and Sustainable Systems. Cham: Springer Nature Switzerland, 2024. 

[7] Ö. Yay, M. Hasanzadeh, S. F. Diltemiz, M. C. Kuşhan, and S. Gürgen, Thermal insulation with cork-based materials. In 

Cork-Based Materials in Engineering: Design and Applications for Green and Sustainable Systems. Cham: Springer Nature 

Switzerland, 2024. 

[8] P. Burgos Pintos, P. Marzo Gago, N. Fernández Delgado, M. Herrera, A. Sanz de León, and S. I. Molina, 

"Sustainable product design by large format additive manufacturing of cork composites," Virtual and Physical 

Prototyping, vol. 19, no. 1, p. e2386106, 2024.  https://doi.org/10.1080/17452759.2024.2386106 

[9] M. J. Kwak, S. H. Lee, and S. Y. Woo, "Growth and anatomical characteristics of different water and light intensities 

on cork oak (Quercus suber L.) seedlings," African Journal of Biotechnology, vol. 10, no. 53, p. 10964, 2011.  

https://doi.org/10.5897/AJB11.2846 

[10] L. Liibert, T. Andreas, and P. Meier, "The fixation of new alternative wood protection systems by means of oil 

treatment," Materials Science, vol. 17, no. 4, pp. 402-406, 2011.  https://doi.org/10.5755/j01.ms.17.4.777 

https://creativecommons.org/licenses/by/4.0/
https://doi.org/10.3389/fmats.2016.00063
https://doi.org/10.1016/j.indcrop.2023.117513
https://doi.org/10.1007/BF01153981
https://doi.org/10.1080/17452759.2024.2386106
https://doi.org/10.5897/AJB11.2846
https://doi.org/10.5755/j01.ms.17.4.777


3093 

 

 

Edelweiss Applied Science and Technology 
ISSN: 2576-8484 

Vol. 9, No. 4: 3082-3093, 2025 
DOI: 10.55214/25768484.v9i4.6738 
© 2025 by the authors; licensee Learning Gate 

 

[11] A. Zemirline, T. Kermezli, M. Announ, and M. Douani, "Investigation of the effect of cutting directions on the 

improvement of mechanical parameters of treated cork by tht: experimental measurement, modelling and 

optimization of mass transfer.," Mechanics, vol. 28, no. 3, pp. 204-210, 2022.  https://doi.org/10.5755/j02.mech.26956 

[12] C. Brazinha, A. P. Fonseca, H. Pereira, O. M. Teodoro, and J. G. Crespo, "Gas transport through cork: Modelling gas 

permeation based on the morphology of a natural polymer material," Journal of Membrane Science, vol. 428, pp. 52-62, 

2013.  https://doi.org/10.1016/j.memsci.2012.10.019 

[13] D. Seibert, H. P. Felgueiras, A. Módenes, F. Borba, R. Bergamasco, and N. C. Homem, "Application of cork as 

adsorbent for water and wastewater treatment using ciprofloxacin as pharmaceutical model," International Journal of 

Environmental Science and Technology, pp. 1-18, 2024.  https://doi.org/10.1007/s13762-024-05836-w 

[14] F. Benmeriem, M. Douani, C. Fares, T. Kermezli, and M. Announ, "Contribution to the analysis of the influence of 

climatic and geographic factors on the average mechanical parameters of cork growing in North Africa: The Case of 

Algeria," Letters in High Energy Physics, pp. 6516–6525, 2024.  https://doi.org/10.31526/lhep.2024.651 

[15] J. J. Camarero, Á. Sánchez-Miranda, M. Colangelo, and L. Matías, "Climatic drivers of cork growth depend on site 

aridity," Science of the Total Environment, vol. 912, p. 169574, 2024.  https://doi.org/10.1016/j.scitotenv.2023.169574 

[16] H. Pereira, "The rationale behind cork properties: A review of structure and chemistry," BioResources, vol. 10, no. 3, 

pp. 437–447, 2018.  https://doi.org/10.15376/biores.10.3.437-447 

[17] F. Kačík et al., "Impact of thermal treatment and accelerated aging on the chemical composition, morphology, and 

properties of spruce wood," Forests, vol. 16, no. 1, p. 180, 2025.  https://doi.org/10.3390/f16010180 

[18] H. Pereira and M. E. Rosa, "Scanning electron microscopy in cork structure evaluation," Journal of Cellulosic Materials, 

vol. 7, no. 1, pp. 34–42, 2009.  

[19] D. Pérez-Terrazas, J. R. González-Adrados, and M. Sánchez-González, "Qualitative and quantitative assessment of 

cork anomalies using near infrared spectroscopy (NIRS)," Food Packaging and Shelf Life, vol. 24, p. 100490, 2020.  

https://doi.org/10.1016/j.fpsl.2020.100490 

[20] M. A. Fortes, L. F. Nunes, and P. Oliveira, "Thermal degradation and structural analysis of cork," Journal of Applied 

Polymer Science, vol. 92, no. 2, pp. 945–951, 2004.  https://doi.org/10.1002/app.20045 

[21] T. Kermezli, A. Bensmaili, K. Amokrane, and A. Gacemi, "Mechanical behavior of cork modified by heat treatment at 

high temperature," Mechanics, vol. 19, no. 4, pp. 424-428, 2013.  https://doi.org/10.5755/j01.mech.19.4.5053 

[22] X. Lu, "Application of the Nernst-Einstein equation to concrete," Cement and Concrete Research, vol. 27, no. 2, pp. 293-

302, 1997.  https://doi.org/10.1016/S0008-8846(97)00005-6 

[23] J. F. Langford, M. R. Schure, Y. Yao, S. F. Maloney, and A. M. Lenhoff, "Effects of pore structure and molecular size 

on diffusion in chromatographic adsorbents," Journal of Chromatography A, vol. 1126, no. 1–2, pp. 95–106, 2006.  

https://doi.org/10.1016/j.chroma.2006.05.042 

[24] M. Liu, H. Guo, J. Luo, X. Gui, Y. Xing, and Y. Cao, "Investigation on the effect of metal cation radius on 

montmorillonite hydration: Combining experiments with molecular dynamics simulation," Separation and Purification 

Technology, vol. 353, p. 128474, 2025.  https://doi.org/10.1016/j.seppur.2024.128474 

[25] G. Zhang, B. E. Patuwo, and M. Y. Hu, "Forecasting with artificial neural networks:: The state of the art," 

International Journal of Forecasting, vol. 14, no. 1, pp. 35-62, 1998.  https://doi.org/10.1016/S0169-2070(97)00044-7 

[26] S. Shalev-Shwartz and S. Ben-David, Understanding machine learning: From theory to algorithms. Cambridge University 

Press: United Kingdom, 2014. 

[27] C. C. Aggarwal, Neural networks and deep learning. Cham: Springer, 2018. 

[28] X.-S. Yang and X. He, "Bat algorithm: literature review and applications," International Journal of Bio-Inspired 

Computation, vol. 5, no. 3, pp. 141-149, 2013.  https://doi.org/10.1504/IJBIC.2013.055093 

 

https://doi.org/10.5755/j02.mech.26956
https://doi.org/10.1016/j.memsci.2012.10.019
https://doi.org/10.1007/s13762-024-05836-w
https://doi.org/10.31526/lhep.2024.651
https://doi.org/10.1016/j.scitotenv.2023.169574
https://doi.org/10.15376/biores.10.3.437-447
https://doi.org/10.3390/f16010180
https://doi.org/10.1016/j.fpsl.2020.100490
https://doi.org/10.1002/app.20045
https://doi.org/10.5755/j01.mech.19.4.5053
https://doi.org/10.1016/S0008-8846(97)00005-6
https://doi.org/10.1016/j.chroma.2006.05.042
https://doi.org/10.1016/j.seppur.2024.128474
https://doi.org/10.1016/S0169-2070(97)00044-7
https://doi.org/10.1504/IJBIC.2013.055093