Application of Artificial Intelligent Approach to Predict the Normal Boiling Point of Refrigerants

Liu, Bo; Karimi Nouroddin, Maryam

doi:https://doi.org/10.1155/2023/6809569

International Journal of Chemical Engineering

On this page

Abstract Introduction Results and Discussion Conclusions Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article | Open Access

Volume 2023 | Article ID 6809569 | https://doi.org/10.1155/2023/6809569

Application of Artificial Intelligent Approach to Predict the Normal Boiling Point of Refrigerants

Bo Liu¹and Maryam Karimi Nouroddin²

Academic Editor: Abdolreza Kharaghani

Received22 Aug 2022

Revised10 Nov 2022

Accepted22 Nov 2022

Published02 Jan 2023

Abstract

Due to the desirable and interesting applications of refrigerants in organic Rankine cycles, heat pumps, and refrigeration, engineers and researchers are becoming more interested in refrigerant properties. One of the most dominant thermophysical properties of these fluids is their normal boiling point (T_b). In the current study, a novel extreme learning method (ELM) and ensemble decision tree boosted algorithm (EDT Boosted) are proposed to forecast the normal boiling point from 16 different molecular groups and one topological index. To this end, a total of 334 data points of T_b are gathered to prepare and test ELM and EDT boosted algorithms. The visual and mathematical comparisons of model outputs and real T_b express that proposed models have great potential to predict T_b of refrigerant. Moreover, sensitivity analysis is applied to explain the effectiveness of input parameters on the determination of T_b for refrigerants.

1. Introduction

The fundamental of Organic Rankine Cycle (ORC), heat pump, and refrigeration system investigations is the study of refrigerants [1]. The studies on refrigerants have been highlighted recently [2–4]. The accuracy of thermophysical properties of materials is known as a main parameter in process design [2, 5, 6]. However, a large number of refrigerants’ physical properties have been reported in the previous research studies [7–9]. Developing alternative desirable refrigerants is necessary because of increasing attention to the greenhouse effect and depletion of the ozone layer [10]. In the computer-aided molecular design (CAMD) process for refrigerants, predicting approaches for estimation of their properties are important, so highly accurate prediction models are needed for engineers and scientists working on this issue [11–15]. One of the important thermal properties of refrigerants is their normal boiling point (T_b), which has applications in the prediction of other thermal properties. The temperature at which the vapor pressure of a liquid is equal to atmospheric pressure is called the normal boiling point. There are several approaches to estimate T_b [16]. Joback and Reid suggested a group contribution approach that estimates an approximation of T_b for aromatic and aliphatic hydrocarbons [17]. This property can be predicted by the summation of the contributions of all molecular groups that exist in material structures. This approach does not perform accurately, but it is an acceptable manner for the preliminary determination of T_b. Then, Devotta and Pendyala upgraded the Joback approach to estimate halogenated mixtures’ normal boiling point more accurately [18]. Later, Constantinou and Gani proposed a new group contribution approach involving UNIQUAC functional-group activity coefficients (UNIFACs) group [19]. Marrero−Morejón and Pardillo−Fontdevila presented a group interaction contribution method [20]. After that, Wang and workers estimated the T_b for organic materials by position group contribution [21]. Abooali and Saboti improved some group contribution prediction methods by applying molecular descriptors to them [22]. Deng et al. used the artificial neural network to estimate the T_b of refrigerants based on a topological index and 16 molecular groups [23].

The prediction of normal boiling points has vital importance on modeling of different processes in refrigerants. The needed time and cost of experimental studies make computational and modeling studies valuable. On the other hand, artificial intelligence methods have shown interesting performance in different topics. In the present study, two novel extreme learning methods (ELM) and ensemble boosted trees (EDT Boosted) are suggested to forecast the normal boiling point of 16 different molecular groups and one topological index. Furthermore, sensitivity analysis is applied to explain the effectiveness of input parameters on the determination of T_b for refrigerants.

2. Methodology

2.1. Experimental Dataset Collection

In this work, a comprehensive dataset of normal boiling points for 334 compounds, which consist of refrigerant components, is gathered from previous works. The data points are collected from three reliable sources, namely, SciFinder, Molbase, and Chemical Abstracts Service, which are provided in Deng et al.’s paper [23]. In some research studies, in order to achieve suitable properties for refrigerants, a classification for them was suggested, that is, alcohols, amines, halogenated hydrocarbons, ethers, organic compounds-alkanes, and alkenes [11, 24]. In order to better classify molecules, the functional groups are selected on the basis of classification (see Figure 1).

2.2. Ensemble Boosted Trees (EDT Boosted)

Quinlan proposed decision trees (DTs) [25] is one of the popular machine learning methods which has the potential to solve many actual problems, such as a short-term photovoltaic power prediction, landslide spatial estimation, flash flood forecasting, and risk factor determination for using drugs [26–29]. The fundamental of the DT is constructed based on the utilization of a series of rules for the identification of regions which have the most homogenous outputs to inputs on which a value is fitted to each one. There are some advantages to the DT as given as follows:(i)This method expresses information in an easy and intuitive manner for visualization(ii)It is a reliable tool for mining interactions and nonlinear effects of various variables(iii)No mathematical assumption is required between input and output variables(iv)It has the ability to handle outliers and suspected values

On the other hand, there are some drawbacks to the DT as given as follows:(i)There are some difficulties in modeling smooth functions(ii)This method is highly sensitive to training data, so that a small alteration in training data can obtain different outputs(iii)It has low bias and high variance [30, 31]

Hence, many strategies have been suggested to improve the predictive ability of the DT, such as ensemble boosted trees (EDT Boosted).

EDT Boosted represents an additive regression algorithm which has a simple tree individual term. The combination of boosting and regression trees is known as an ensemble method, which uses recursive binary splits to connect outputs to input variables. EDT Boosted uses the advantages of DT-based approaches and also overcomes their disadvantages [31, 32]. EDT Boosted has been used in different issues, such as medical [33], ecology [34], and banking [32]. In the current work, EDT Boosted is used for the estimation of the normal boiling point of refrigerants for the first time.

By supposing x = (x₁, x₂, …, x_n) as predictor vector, y represents response. The EDT boosted model can be trained by the following formulation [35]:in which , , and are weights of nodes, split variables, and single decision trees, respectively.

2.3. Extreme Learning Machine

All neurons of the single hidden layer feedforward neural networks (SLFNs) are located in an input layer, an output layer, and a hidden layer by considering the applied function. These algorithms work by first connecting the input layer to the input mode and then switching it to the hidden layer. Thus, it is worthy to point out that in ANN, hidden layers can be considered as processors and then the output layer constructs the output mode [36–39].

The requirement of SLFN algorithms to differentiable activation functions and layered base of these algorithms converts them to complex and inefficient algorithms. Figure 2 gives information about the scheme of ELM, which includes the three aforesaid layers.

By considering {(x_i, y_i),i = 1, …, N} in Rⁿ × R^m which the target and training values are shown by y_i and x_i, respectively, and also dimensions of output and input data are illustrated by m and n, respectively. For the ELM algorithm, if the number of nodes is assumed to be L, this algorithm can be written as follows:in which β = [β₁, …, β_L] and h(x) = [h₁(x),…h_L(x)] are known as weight vector and nonlinear feature mapping. denotes the estimated value. There are different kinds of functions which have the potential to be utilized in neurons of the hidden layer, such as multiquadric, cosine, hyperbolic tangent, and hard limit. h_i(x) for real conditions is written as follows:in which b_iR, a_iR^d, and are nonlinear differential functions in terms of a and b, which are hidden node parameters for the ELM estimation process. This approach is constructed based on a two-step learning system of preparation of SLFN, including random feature mapping and linear parameter solution. As a first step, ELM uses input weights and hidden biases to estimate the mapping matrix of the hidden layer of transferring feature space. As a second step, the weights of the hidden layer should be connected to outputs, which is done by β. These parameters will be determined by the minimum squared error as follows:in which u and ||.|| point to the matrix of hidden layer outputs and Frobenius norm. “u” is defined as follows:Also, y points to the target matrix, which is expressed as follows:

Equation (4) can be solved by the following formulation:in which denotes the Moore−Penrose inverse of u which can be rewritten with substitution by as follows [40, 41]:in which u^T is the transpose of u.

3. Results and Discussion

In this study, ELM and EDT boosted algorithms were developed to estimate T_b for different refrigerants based on the molecular groups and one topological index. To this end, the activation function of the ELM algorithm is chosen as the sigmoid function and also the initial weights of inputs are made randomly in the range of −1 to 1. Moreover, the cross-validation method is used to determine the number of hidden layer neurons [42]. This parameter is equal to 10 for normal boiling point calculation. The general performance of ELM and EDT boosted algorithms in the prediction of normal boiling point is evaluated by the following error measurements [43, 44]:

The results of the abovementioned error measurements are inserted in Table 1. In the prediction of T_b, R² = 0.995, MRE = 1.08, MSE = 25.96, RMSE = 5.16, and STD = 3.76 are determined for the overall process of the ELM. Furthermore, the details of the training and testing phases are reported in this table. This error analysis shows the high ability of ELM in estimation of the refrigerant’s normal boiling point. Another powerful approach for assessing the suitability of the ELM in the estimation of T_b is visual comparison of model outputs and real T_b values, as shown in Figure 3.

(a)

(b)

Moreover, the cross plot of actual normal boiling point versus experimental value is depicted in Figure 4 which illustrates that T_b points lie on the bisector line.

(a)

(b)

On the other hand, predicted T_b values have great agreement with their real ones. On the one hand, relative deviations between model outputs and experimental normal boiling points are calculated (see Figure 5). It is shown that determined values are near the x axis, so the proposed models have interesting potential in the prediction of normal boiling point. The accuracy of models shows their potential to be used in the estimation of the normal boiling point of refrigerants. These models can be employed easily; therefore, the construction of accurate software based on these models is possible and reasonable.

(a)

(b)

In the current work, a new analysis called sensitivity is used to show effects of different molecular descriptors on the normal boiling point of refrigerants. In this method, the relevancy factor (r) is defined for each molecular group as follows [45]:where , , ,and are experimental T_b, the average of T_b, kth of input, and the average of inputs. Figure 6 illustrates the value of r for each molecular descriptor. In this method, the negative value of r expresses that as the number of associated molecular group increases in refrigerant molecules, the normal boiling point will decrease.

Furthermore, the range of this parameter lies between −1 and 1. The absolute value of r represents the intensity of the effect of the molecular group on the normal boiling point.

The correctness of the predicting model is function of the accuracy of the utilized real data [46, 47]. The present study employs some experimental data related to the normal boiling points of refrigerants, so they may have some measurement errors. For the abovementioned reasons, identification of suspected experimental data becomes necessary [48]. The leverage method is one of the applicable approaches to this problem. This method uses a matrix called Hat to obtain the following criteria [49, 50]:

This matrix is obtained as a function of X matrix s × t which denotes the number of samples and model variables, respectively. Another major parameter of this method is the critical leverage value () which is shown as follows [51]:

As shown in Figure 7, the reliable area for experimental data is defined as the zone inbounded between red and green lines. The main number of normal boiling points is placed in a reliable zone, so the overall databank is dependable for preparing models.

4. Conclusions

In the present study, novel ELM and EDT boosted algorithms are used to predict the normal boiling point of refrigerants in terms of 16 molecular groups and the topology index EATII. The suggested outputs and actual T_b have been compared by different methods, including simultaneous visual comparison, cross plot, relative error depiction, and statistical analysis. These comparisons confirm the high ability of algorithms in the estimation of T_b. In addition, sensitivity analysis is applied to distinguish the effect of molecular descriptors on the determination of T_b for refrigerants. Finally, it is recommended to employ other available machine learning models trained on a wider databank for the utilization as software. These models should be compared with each other to choose the most accurate model.

Nomenclature

T_b:	Boiling point
ELM:	Extreme learning method
EDT:	Ensemble decision tree
CAMD:	Computer-aided molecular design
UNIFAC:	UNIQUAC functional-group activity coefficients
ORC:	Organic rankine cycle
SLFN:	Single hidden layer feedforward neural network
MSE:	Mean squared error
MRE:	Mean relative error
RMSE:	Root mean square error
STD:	Standard deviation
R²:	R-squared
r:	Relevancy factor.

Data Availability

The references of experimental data used to support the findings of this study are included within the article.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was supported by the 2021 Scientific Research Fund Project of Liaoning Provincial Education Department “Study on the development of compound juice of small berry and its quality change during storage” (LJKZ1127).

References

P. Giménez-Prades, J. Navarro-Esbrí, C. Arpagaus, A. Fernández-Moreno, and A. Mota-Babiloni, “Novel molecules as working fluids for refrigeration, heat pump and organic Rankine cycle systems,” Renewable and Sustainable Energy Reviews, vol. 167, Article ID 112549, 2022.
View at: Publisher Site | Google Scholar
M. Ghodbane, Z. Said, O. Ketfi et al., “Thermal performance assessment of an ejector air-conditioning system with parabolic trough collector using R718 as a refrigerant: a case study in Algerian desert region,” Sustainable Energy Technologies and Assessments, vol. 53, Article ID 102513, 2022.
View at: Publisher Site | Google Scholar
Z. Said, P. Sharma, A. K. Tiwari et al., “Application of novel framework based on ensemble boosted regression trees and Gaussian process regression in modelling thermal performance of small-scale Organic Rankine Cycle (ORC) using hybrid nanofluid,” Journal of Cleaner Production, vol. 360, Article ID 132194, 2022.
View at: Publisher Site | Google Scholar
P. Sharma, Z. Said, A. Kumar et al., “Recent advances in machine learning research for nanofluid-based heat transfer in renewable energy system,” Energy &amp; Fuels, vol. 36, no. 13, pp. 6626–6658, 2022.
View at: Publisher Site | Google Scholar
M. El Hassan, “Numerical characterization of the flow dynamics and COP estimation of a binary fluid ejector ground source heat pump cooling system,” Fluid, vol. 7, p. 250, 2022.
View at: Publisher Site | Google Scholar
Y. Maalem, Y. Tamene, and H. Madani, “Behavior of the thermo-physical properties and performance evaluation of the refrigerants blends of (Fluorocarbon/Hydrocarbon) for cooling cycle,” Recueil de mécanique, vol. 6, pp. 544–559, 2022.
View at: Google Scholar
H. M. Meacock, Refrigeration Processes: A Practical Handbook on the Physical Properties of Refrigerants and Their Applications, Elsevier, Amsterdam, The Netherlands, 2013.
A. R. C. Morais, L. D. Simoni, M. B. Shiflett, and A. M. Scurto, “Viscosity and density of a polyol ester lubricating oil saturated with compressed hydrofluoroolefin refrigerants,” Journal of Chemical & Engineering Data, vol. 65, no. 9, pp. 4335–4346, 2020.
View at: Publisher Site | Google Scholar
F. Yousefi, S. M. Hosseini, K. Hamidi, and M. Pierantozzi, “Viscosities of liquid refrigerants from a rough hard-sphere theory-based semi-empirical model,” International Journal of Thermophysics, vol. 40, no. 8, pp. 74–18, 2019.
View at: Publisher Site | Google Scholar
Y. Dong, M. Coleman, and S. A. Miller, “Greenhouse gas emissions from air conditioning and refrigeration service expansion in developing countries,” Annual Review of Environment and Resources, vol. 46, no. 1, pp. 59–83, 2021.
View at: Publisher Site | Google Scholar
Y. Khetib, A.-H. Meniai, and A. Lallemand, “Computer-aided design of CFC and HCFC substitutes using group contribution methods,” Desalination, vol. 239, no. 1-3, pp. 82–91, 2009.
View at: Publisher Site | Google Scholar
A. Samudra and N. V. Sahinidis, “Design of heat-transfer media components for retail food refrigeration,” Industrial & Engineering Chemistry Research, vol. 52, no. 25, pp. 8518–8526, 2013.
View at: Publisher Site | Google Scholar
L. Zhang, H. Mao, L. Liu, J. Du, and R. Gani, “A machine learning based computer-aided molecular design/screening methodology for fragrance molecules,” Computers & Chemical Engineering, vol. 115, pp. 295–308, 2018.
View at: Publisher Site | Google Scholar
A. Baghban, P. Abbasi, and P. Rostami, “Modeling of viscosity for mixtures of Athabasca bitumen and heavy n-alkane with LSSVM algorithm,” Petroleum Science and Technology, vol. 34, no. 20, pp. 1698–1704, 2016.
View at: Publisher Site | Google Scholar
A. Bahadori, A. Baghban, M. Bahadori et al., “Computational intelligent strategies to predict energy conservation benefits in excess air controlled gas-fired systems,” Applied Thermal Engineering, vol. 102, pp. 432–446, 2016.
View at: Publisher Site | Google Scholar
B. E. Poling, J. M. Prausnitz, and J. P. Oconnell, The Properties of Gases and Liquids, McGraw-Hill, New York, NY, USA, 2001.
K. G. Joback and R. C. Reid, “Estimation of pure-component properties from group-contributions,” Chemical Engineering Communications, vol. 57, no. 1-6, pp. 233–243, 1987.
View at: Publisher Site | Google Scholar
S. Devotta and V. R. Pendyala, “Modified Joback group contribution method for normal boiling point of aliphatic halogenated compounds,” Industrial & Engineering Chemistry Research, vol. 31, no. 8, pp. 2042–2046, 1992.
View at: Publisher Site | Google Scholar
L. Constantinou and R. Gani, “New group contribution method for estimating properties of pure compounds,” AIChE Journal, vol. 40, no. 10, pp. 1697–1710, 1994.
View at: Publisher Site | Google Scholar
J. Marrero‐Morejón and E. Pardillo‐Fontdevila, “Estimation of pure compound properties using group‐interaction contributions,” AIChE Journal, vol. 45, no. 3, pp. 615–621, 1999.
View at: Publisher Site | Google Scholar
Q. Wang, P. Ma, C. Wang, and S. Xia, “Position group contribution method for predicting the normal boiling point of organic compounds,” Chinese Journal of Chemical Engineering, vol. 17, no. 2, pp. 254–258, 2009.
View at: Publisher Site | Google Scholar
D. Abooali and M. A. Sobati, “Novel method for prediction of normal boiling point and enthalpy of vaporization at normal boiling point of pure refrigerants: a QSPR approach,” International Journal of Refrigeration, vol. 40, pp. 282–293, 2014.
View at: Publisher Site | Google Scholar
S. Deng, W. Su, and L. Zhao, “A neural network for predicting normal boiling point of pure refrigerants using molecular groups and a topological index,” International Journal of Refrigeration, vol. 63, pp. 63–71, 2016.
View at: Publisher Site | Google Scholar
I. Louaer, A.-H. Meniai, O. Larkeche, and M. Bencheikh-Lehocine, “Computer-aided design and test of new refrigerants for anabsorption cycle using group contribution methods,” Desalination, vol. 206, no. 1-3, pp. 620–632, 2007.
View at: Publisher Site | Google Scholar
J. R. Quinlan, “Machine learning, chap,” Induction of decision trees, Kluwer Academic Publishers, Boston, MA, USA, 1986.
View at: Google Scholar
A. Amirabadizadeh, H. Nezami, M. G. Vaughn, S. Nakhaee, and O. Mehrpour, “Identifying risk factors for drug use in an Iranian treatment sample: a prediction approach using decision trees,” Substance Use & Misuse, vol. 53, no. 6, pp. 1030–1040, 2018.
View at: Publisher Site | Google Scholar
K. Khosravi, B. T. Pham, K. Chapi et al., “A comparative assessment of decision trees algorithms for flash flood susceptibility modeling at Haraz watershed, northern Iran,” Science of the Total Environment, vol. 627, pp. 744–755, 2018.
View at: Publisher Site | Google Scholar
J. Wang, P. Li, R. Ran, Y. Che, and Y. Zhou, “A short-term photovoltaic power prediction model based on the gradient boost decision tree,” Applied Sciences, vol. 8, no. 5, p. 689, 2018.
View at: Publisher Site | Google Scholar
M. Akin, C. Hand, E. Eyduran, and B. M. Reed, “Predicting minor nutrient requirements of hazelnut shoot cultures using regression trees,” Plant Cell, Tissue and Organ Culture, vol. 132, no. 3, pp. 545–559, 2018.
View at: Publisher Site | Google Scholar
J. Franklin, R. Tibshirani, J. Friedman, and J. Franklin, “The elements of statistical learning: data mining, inference and prediction,” The Mathematical Intelligencer, vol. 27, no. 2, pp. 83–85, 2005.
View at: Publisher Site | Google Scholar
J. Elith, J. R. Leathwick, and T. Hastie, “A working guide to boosted regression trees,” Journal of Animal Ecology, vol. 77, no. 4, pp. 802–813, 2008.
View at: Publisher Site | Google Scholar
M. Zięba, S. K. Tomczak, and J. M. Tomczak, “Ensemble boosted trees with synthetic features generation in application to bankruptcy prediction,” Expert Systems with Applications, vol. 58, pp. 93–101, 2016.
View at: Publisher Site | Google Scholar
M. Abdar, N. Y. Yen, and J. C.-S. Hung, “Improving the diagnosis of liver disease using multilayer perceptron neural network and boosted decision trees,” Journal of Medical and Biological Engineering, vol. 38, pp. 953–965, 2018.
View at: Google Scholar
G. DeAth, “Boosted trees for ecological modeling and prediction,” Ecology, vol. 88, no. 1, pp. 243–251, 2007.
View at: Publisher Site | Google Scholar
Y.-S. Chung, “Factor complexity of crash occurrence: an empirical demonstration using boosted regression trees,” Accident Analysis & Prevention, vol. 61, pp. 107–118, 2013.
View at: Publisher Site | Google Scholar
E. Keybondorian, H. Zanbouri, B. Amin, and T. Hamule, “Application of MLP-ANN strategy to predict higher heating value of biomass in terms of proximate analysis, Energy Sources, Part A: recovery, Utilization,” And Environmental Effects, vol. 39, pp. 2105–2111, 2017.
View at: Google Scholar
M. Abdi-Khanghah, B. Amin, Z. Naserzadeh, and Z. Zhang, “Prediction of solubility of N-alkanes in supercritical CO2 using RBF-ANN and MLP-ANN,” Journal of CO2 Utilization, vol. 25, pp. 108–119, 2018.
View at: Google Scholar
R. Eghtedaei, M. Abdi-khanghah, B. S. Najar, and A. Baghban, “Viscosity estimation of mixed oil using RBF-ANN approach,” Petroleum Science and Technology, vol. 35, no. 17, pp. 1731–1736, 2017.
View at: Publisher Site | Google Scholar
F. Zarei and A. Baghban, “Phase behavior modelling of asphaltene precipitation utilizing MLP-ANN approach,” Petroleum Science and Technology, vol. 35, no. 20, pp. 2009–2015, 2017.
View at: Publisher Site | Google Scholar
Y. Bengio, “Learning deep architectures for AI,” Foundations and trends® in Machine Learning, vol. 2, pp. 1–127, 2009.
View at: Publisher Site | Google Scholar
C. R. Rao and S. K. Mitra, “Further contributions to the theory of generalized inverse of matrices and its applications,” Sankhyā: The Indian Journal of Statistics Series A, vol. 33, pp. 289–300, 1971.
View at: Google Scholar
G.-B. Huang, Q.-Y. Zhu, and C.-K. Siew, “Extreme learning machine: theory and applications,” Neurocomputing, vol. 70, no. 1-3, pp. 489–501, 2006.
View at: Publisher Site | Google Scholar
A. Baghban and A. Khoshkharam, “Application of LSSVM strategy to estimate asphaltene precipitation during different production processes,” Petroleum Science and Technology, vol. 34, no. 22, pp. 1855–1860, 2016.
View at: Publisher Site | Google Scholar
A. Bemani, A. Baghban, and A. H. Mohammadi, “An insight into the modeling of sulfur content of sour gases in supercritical region,” Journal of Petroleum Science and Engineering, vol. 184, Article ID 106459, 2020.
View at: Publisher Site | Google Scholar
R. Razavi, A. Sabaghmoghadam, A. Bemani, A. Baghban, K.-W. Chau, and E. Salwana, “Application of ANFIS and LSSVM strategies for estimating thermal conductivity enhancement of metal and metal oxide based nanofluids,” Engineering Applications of Computational Fluid Mechanics, vol. 13, no. 1, pp. 560–578, 2019.
View at: Publisher Site | Google Scholar
M. H. Ahmadi, A. Baghban, M. Ghazvini, M. Hadipoor, R. Ghasempour, and M. R. Nazemzadegan, “An insight into the prediction of TiO2/water nanofluid viscosity through intelligence schemes,” Journal of Thermal Analysis and Calorimetry, vol. 139, no. 3, pp. 2381–2394, 2020.
View at: Publisher Site | Google Scholar
D. D. Nguyen, R. Daneshfar, A. H. S. Dehaghani, and C.-H. Su, “The effect of shear rate on aggregation and breakage of asphaltenes flocs: experimental study and model-based analysis,” Journal of Molecular Liquids, vol. 325, Article ID 114861, 2021.
View at: Publisher Site | Google Scholar
E. Khamehchi and A. Bemani, “Prediction of pressure in different two-phase flow conditions: machine learning applications,” Measurement, vol. 173, Article ID 108665, 2021.
View at: Publisher Site | Google Scholar
A. Bemani, A. Kazemi, and M. Ahmadi, “An insight into the microorganism growth prediction by means of machine learning approaches,” Journal of Petroleum Science and Engineering, vol. 220, Article ID 111162, 2023.
View at: Publisher Site | Google Scholar
A. Bemani, A. Kazemi, M. Ahmadi, R. Yousefzadeh, and M. K. Moraveji, “Rigorous modeling of frictional pressure loss in inclined annuli using artificial intelligence methods,” Journal of Petroleum Science and Engineering, vol. 211, Article ID 110203, 2022.
View at: Publisher Site | Google Scholar
R. Razavi, A. Bemani, A. Baghban, A. H. Mohammadi, and S. Habibzadeh, “An insight into the estimation of fatty acid methyl ester based biodiesel properties using a LSSVM model,” Fuel, vol. 243, pp. 133–141, 2019.
View at: Publisher Site | Google Scholar

Copyright

Copyright © 2023 Bo Liu and Maryam Karimi Nouroddin. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

566

Downloads

458

Citations