Fourteen International Conference on Thermal Engineering: Theory and Applications

May 25-27, 2023 Yalova, Republic of Türkiye

Machine Learning Applications on Membrane Distillation

E. Aytaç1, M. Khayet2,3
1
Department of Environmental Engineering, Zonguldak Bülent Ecevit University, 67100,
Zonguldak, Türkiye.
2
Department of Structure of Matter, Thermal Physics and Electronics, Faculty of Physics,
University Complutense of Madrid, Avda. Complutense s/n, 28040, Madrid, Spain.
3
Madrid Institute for Advanced Studies of Water (IMDEA Water Institute), Calle Punto Net Nº 4,
28805, Alcalá de Henares, Madrid, Spain.

Abstract
Membrane distillation (MD), a thermally driven phase separation method, has a number of benefits, including the
capacity to process a variety of input solutions, high purity and yield of the purified liquid, and minimal energy usage.
Applications such as, water treatment, chemical separation, and fuel purification are benefiting from its versatility and
scalability. Machine learning (ML) – in demand topic of the last years – is the use of statistical models and algorithms to
allow machines to learn and make decisions quickly for humans. In this study, we investigated the use of machine learning
in membrane distillation. It was understood that the number of articles in international peer reviewed journals/books was
very few (10 documents). Besides most of the documents (6) were published in 2022. We briefly presented what topics the
published articles focus on, which ML algorithms they use, and their results. The results showed that the integration of
machine learning methods in the field of membrane distillation is open and offers important opportunities to researchers
who want to work on this subject.
Keywords: Membrane distillation, Machine learning, Scopus, Web of Science.
disease diagnosis, weed detection, and yield prediction [7,
8]. Machine Learning is to perform particular actions
1. Introduction automatically utilizing computer algorithms through
Membrane distillation is the process that includes experience and the usage of data [9, 10]. The basic flow
semipermeable membranes to separate a mixture of of a machine learning task can be seen in Fig. 1.
compounds according to their molecular weight and size. Machine Learning have mainly 3 categories:
The vapor pressure difference between the two sides of supervised, unsupervised and reinforcement learning.
the membrane serves as the driving force behind the Labeled data are used in supervised learning techniques
thermal separation process. MD is a popular method for to build multivariate regression or classification models.
cleaning water and other liquid mixtures, and it works Unsupervised learning creates clusters from unlabeled
especially well to remove dissolved salts and other data to reveal previously undiscovered associations.
inorganic impurities from water [1]. Other liquids, such Models that use reinforcement learning are self-improving
organic solvents and fuels, can also be employed to and continually improve. Semi-supervised learning that act
separate and purify. Since MD can work at relatively low as a bridge the between supervised and unsupervised
temperatures and is typically more energy-efficient than machine learning is counted as the 4th category [11].
other separation techniques like distillation or reverse In direct proportion to the interest in the use of
osmosis, it is a good choice for separating heat-sensitive machine learning methods in many areas, the integration
compounds [2]. of these methods in membrane distillation processes has
Machine learning is an area of artificial intelligence also started. According to Scopus database results there
(AI) that focuses on the creation of algorithms and models are 9 documents in the peer-reviewed international
that allow computers to gain experience-based knowledge journals/books [12]. Web of Science (WoS), the other
and enhance their performance on a given task [3-5]. By global database include 8 documents [13]. The combined
automating operations that would take too much time or be results are given in Table 1.
too challenging for humans to complete manually, it has As can be seen in Table 1. there are limited research
the potential to change a wide range of areas [6]. (10 documents) on machine applications of membrane
In this paper we uncover the published literature and distillation. Another striking situation in the table is that
future opportunities of machine learning in membrane most of the published articles is very recent (belong to the
distillation. year 2022). These results are an important indication that
the use of machine learning methods in membrane
distillation is limited and that new study topics are open for
2. Applications of Machine Learning on the coming years.
Membrane Distillation Domain By going into the details of some of the studies carried
Since 1959, Arthur Samuel first coined the phrase out, we can reveal which types of applications researchers
"Machine Learning" the increasing popularity ML focus on. Derbali et. al (2016) [14] worked on faulty valves
applications are making decisions in the fields such as detection in water desalination systems. They utilized the
crop quality, water management, soil management, datasets from supplementary materials of different
research papers and they process the data with many

1
machine learning algorithms. They figured out that when performance of the process, including permeate flux and
processing speed and accuracy are balanced against one specific heat energy consumption under various feed inlet
another, the decision trees method performs better than temperature, feed flow rate, and membrane length
others. conditions. They found out that the behavior of the hollow
In 2020, Yang C. and his friends studied based on 36 membrane module could be accurately predicted using the
separate actual vacuum membrane distillation tests, an ANN model for the full range of input variables [15].
artificial neural network (ANN) was used to estimate the

Fig. 1. Basic workflow of Machine Learning

Table 1. Machine learning application in membrane distillation in the literature (Scopus + WoS)
No Authors Title Journal Year
Journal of Theoretical
Derbali, M., Fattouh, A., Improved fault detection in water desalination systems
1 and Applied Information 2016
Jerbi, H., Abdelkrim, M.N. using machine learning techniques
Technology
Improving the distillate prediction of a membrane
Acevedo, L., Uche, J., Del-
2 distillation unit in a trigeneration scheme by using artificial Water (Switzerland) 2018
Amo, A.
neural networks
Yang, C.H., Peng, X.,
Zhao, Y.J., Wang, Fu, Prediction model to analyze the performance of VMD Computers and
3 2020
J.X., Liu, K., Li, Y.D., Li, desalination process Chemical Engineering
P.L.
Dudchenko, A.V., Mauter, Neural networks for estimating physical parameters in Journal of Membrane
4 2020
M.S. membrane distillation Science
Jones, M.N., Stevnsborg,
Pilot Plant 4.0: A Review of Digitalization Efforts of the
M., Nielsen, R.F., (...), Computer Aided
5 Chemical and Biochemical Engineering Department at the 2022
Huusom, J.K., Dam- Chemical Engineering
Technical University of Denmark (DTU)
Johansen, K.
Development of machine learning and stepwise
Chemical Engineering
Behnam, P., Shafieian, A., mechanistic models for performance prediction of direct
6 and Processing - 2022
Zargar, M., Khiadani, M. contact membrane distillation module- A comparative
Process Intensification
study
Ray, S.S., Verma, R.K., Exploration of time series model for predictive evaluation Process Safety and
7 Singh, A., (...), Lee, H.K., of long-term performance of membrane distillation Environmental 2022
Kwon, Y.-N. desalination Protection
Hybrid desalination technologies for sustainable water-
Son, H.S., Nawaz, M.S., Desalination and Water
8 energy nexus: innovation in integrated membrane module 2022
Soukane, S., Ghaffour, N. Treatment
development
Qamar, A., Kerdi, S.,
A deep neural networks framework for in-situ biofilm
Amin, N., (...), Separation and
9 thickness detection and hydrodynamics tracing for filtration 2022
Vrouwenvelder, J., Purification Technology
systems
Ghaffour, N.
Behnam, P., Faegh, M., A review on state-of-the-art applications of data-driven
10 Desalination 2022
Khiadani, M. methods in desalination systems

Behman et. al (2022) [16] used machine learning ML approaches perform better than the mechanistic
(artificial neural network, support vector regression, and model. They showed some advantages of machine
random forest) to predict the performance of direct contact learning models such as their simplicity, increased
membrane distillation. The findings demonstrate two of the accuracy, and much shorter computation times.

2
Qamar and his colleagues worked on using ML for in-situ 3. Conclusion
biofilm thickness detection and hydrodynamics tracing for The use of machine learning approaches in industry,
filtration systems. The filtering channels module's biofilm academy, science, engineering and daily life is increasing
growth is difficult to see and has a significant propensity to rapidly every day. Researchers working on membrane
degrade process efficiency. They created a database of distillation have to keep up with this trend to expand their
biofilm images and they trained a Convolution Neural productivity and scope. In this study, we took a look at the
Network (CNN) to determine the biofilm thickness. The machine learning applications of membrane distillation in
results showed that the created framework might be used the global databases Scopus and Web of Science. The
in desalination facilities for proactive controls and early results revealed that the integration of machine learning
decision-making [17]. methods into the membrane distillation domain has started
We can say that 2022 is the golden year of in recent years and the number of published articles is
applications of machine learning to membrane distillation. limited. While this situation offers opportunities to young
Saikat Sinha et. al (2022) [18] worked with time series researchers who are starting to work on membrane
machine learning models to overcome the membrane distillation, it also shows the importance of data-driven
wetting problems in MD. Their ARIMA model was capable methods to membranologists.
of forecasting this membrane wetting problem.

[11] A. Carter, S. Imtiaz, G.F. Naterer, Review of

References interpretable machine learning for process industries,
[1] M. Essalhi, M. Khayet, S. Tesfalidet, M. Alsultan, N. Process Safety and Environmental Protection, 170 (2023)
Tavajohi, Desalination by direct contact membrane 647-659.
distillation using mixed matrix electrospun nanofibrous [12] Scopus, Scopus basic search, (2022).
membranes with carbon-based nanofillers: A strategic https://www.scopus.com/results/results.uri?sort=plf-
improvement, Chemical Engineering Journal, 426 (2021) f&src=s&st1="membrane+distillation"+and+"machine+l
131316. earning"&sid=f03760d20675961a0865521630f86ea4&s
[2] M. Al-Salmi, M. Laqbaqbi, S. Al-Obaidani, R.S. Al- ot=b&sdt=b&sl=61&s=TITLE-ABS-
Maamari, M. Khayet, M. Al-Abri, Application of KEY%28"membrane+distillation"+and+"machine+learn
membrane distillation for the treatment of oil field ing"%29&origin=searchbasic&editSaveSearch=&yearFr
produced water, Desalination, 494 (2020) 114678. om=Before+1960&yearTo=Present (accessed December
[3] E. Aytaç, Modeling Future Impacts on Land Cover of 26, 2022)
Rapid Expansion of Hazelnut Orchards: A Case Study on [13] WoS, Web of Science basic search, (2022).
Samsun, Turkey, European Journal of Sustainable https://www.webofscience.com/wos/woscc/summary/81
Development Research, 6 (2022) em0193. 2c7787-d9d8-4339-8ac9-2c1685b070dd-
[4] L. Velagapudi, F.A. Saiegh, S. Swaminathan, N. 6745bad5/relevance/1 (accessed December 26, 2022)
Mouchtouris, O. Khanna, V. Sabourin, M.R. Gooch, N. [14] M. Derbali, S.M. Buhari, G. Tsaramirsis, M.
Herial, S. Tjoumakaris, R.H. Rosenwasser, P. Jabbour, Stojmenovic, H. Jerbi, M.N. Abdelkrim, M.H. Al-
Machine learning for outcome prediction of neurosurgical Beirutty, Water Desalination Fault Detection Using
aneurysm treatment: Current methods and future Machine Learning Approaches: A Comparative Study,
directions, Clinical Neurology and Neurosurgery, 224 IEEE Access, 5 (2016) 23266-23275.
(2023) 107547. [15] C. Yang, X. Peng, Y. Zhao, X. Wang, J. Fu, K. Liu,
[5] E. Aytac, Exploring Electrocoagulation Through Data Y. Li, P. Li, Prediction model to analyze the performance
Analysis and Text Mining Perspectives, Environmental of VMD desalination process, Computers & Chemical
Engineering and Management Journal, 21 (2022) 671- Engineering, 132 (2020) 106619.
685. [16] P. Behnam, A. Shafieian, M. Zargar, M. Khiadani,
[6] E. Aytaç, Havzaların Benzerliklerini Tanımlamada Development of machine learning and stepwise
Alternatif Bir Yaklaşım: Hiyerarşik Kümeleme Yöntemi mechanistic models for performance prediction of direct
Uygulaması, Afyon Kocatepe Üniversitesi Fen Ve contact membrane distillation module- A comparative
Mühendislik Bilimleri Dergisi, 21 (2021) 958-970. study, Chemical Engineering and Processing - Process
[7] M. Doshi, A. Varghese, Chapter 12 - Smart agriculture Intensification, 173 (2022) 108857.
using renewable energy and AI-powered IoT, in: A. [17] A. Qamar, S. Kerdi, N. Amin, X. Zhang, J.
Abraham, S. Dash, J.J.P.C. Rodrigues, B. Acharya, S.K. Vrouwenvelder, N. Ghaffour, A deep neural networks
Pani (Eds.) AI, Edge and IoT-based Smart Agriculture, framework for in-situ biofilm thickness detection and
Academic Press, 2022, pp. 205-225. hydrodynamics tracing for filtration systems, Separation
[8] A.L. Samuel, Some Studies in Machine Learning and Purification Technology, 301 (2022) 121959.
Using the Game of Checkers, IBM Journal of Research [18] S.S. Ray, R.K. Verma, A. Singh, S. Myung, Y.-I.
and Development, 3 (1959) 210-229. Park, I.-C. Kim, H.K. Lee, Y.-N. Kwon, Exploration of
[9] R. Zhong, C. Salehi, R. Johnson, Machine learning for time series model for predictive evaluation of long-term
drilling applications: A review, Journal of Natural Gas performance of membrane distillation desalination,
Science and Engineering, 108 (2022) 104807. Process Safety and Environmental Protection, 160 (2022)
[10] M. Khayet, E. Aytaç, T. Matsuura, Bibliometric and 1-12.
sentiment analysis with machine learning on the scientific
contribution of Professor Srinivasa Sourirajan,
Desalination, 543 (2022) 116095.