Abstract 

This paper deals with the prospects for the use of Artificial Intelligence (AI) in water desalination by applying it to the promising KSA experience. The paper included four themes; AI applications in desalination with renewable energy, AI applications in automating “adsorption” processes, Filterless Technology, and the recent trends in the localization of AI in water desalination in the Kingdom of Saudi Arabia. The paper sheds light on the developments of the efforts of the Kingdom of Saudi Arabia in the localization of AI techniques in desalinating water for agricultural purposes. It concludes that the applications of AI have become a major factor in all details of desalination of drinking water and agriculture in the Kingdom.

Introduction 

The Kingdom of Saudi Arabia is heading towards the fourth revolution in all development fields. The executive frameworks of development programmes adopted by the Kingdom within the framework of the National Transformation Program “Vision 2030”, especially at the level of facing the challenges related to the water situation and the ambitions of agricultural expansion and development projects associated with it. Despite the distinguished global position of KSA in the field of water desalination, (it is one of the largest producers of desalinated water in the world with 18% of global production), the enormous development ambitions of the Kingdom to meet the requirements of future development require more innovative solutions to all problems. This underlines the need to take advantage of the progress achieved in all fields, especially AI, and to employ its techniques in water treatment and desalination projects.

First: AI applications in desalination with renewable energy

AI uses a range of data, including satellite imagery, climate, temperature, humidity, and weather forecasts to make or suggest optimal decisions to manage crops and waste the least amount of water possible. Soil and light sensors provide data to the AI about the most appropriate times to provide the soil with water and fertilizers. The smart irrigation systems provide water with high accuracy. They have a solution to the problem of leakage. AI approaches provide decision makers with different investment options, each consisting of different desalination groups with respect to locations, capacities, and energy sources on many scales. Smart decisions determine the optimal location of the plants and the capacity of the desalination system for future expectations. It also provides decision makers with the possibility of constructing a network of pipelines and transporting water between planting sites. AI is being used more and more to control the best way to treat water and remove salt from seawater.

As shown in the figure, an AI desalination system can be mainly divided into four categories: decision making, parameter optimization, parameter prediction, and control. The features of artificial intelligence employed in the design of desalination systems not only realize the maximum efficiency and minimum cost, but also free human resources.

Firstly; Decision-making by AI in RE-desalination systems

Determining the location of a plant is a strategic decision, as decision makers need a variety of different criteria and factors. Environmental, social, and economic sustainability must also be considered when selecting a site (Dweiri and Almulla, 2018). AI plays an important role in making such decisions through the Multi-criteria Decision Support System (DSS) model to rank these criteria (Ishimatsu et al, 2017). This model is consistent with the environmental, technological, and financial constraints associated with common desalination technologies. This model is based on a decision-making approach based on an Analytical Hierarchal Process to determine the best RO plant technology to maintain strategic decision-making on the site selection of seawater desalination plants (Bick and Oron, 2011).

Secondly: Parameter optimization by AI in desalination systems

Operating parameters in desalination systems are divided into four categories: energy parameters (energy input), structure parameters (size), feed parameters (pressure, pH value, feed flow rate, total dissolved solids (TDS), seawater temperature) and surrounding parameters (ambient temperature, wind speed, solar radiation).

In addition, there are many variables involved in the seawater desalination process, so traditional mathematical methods cannot obtain the optimal design efficiently. Accordingly, artificial intelligence techniques such as Genetic Algorithm and Particle Swarm Optimization Algorithm solve this problem to improve the situation of the operating parameters, which is reflected in the operating performance of the systems through algorithms that represent different search strategies that in turn affect the calculations and solve complex engineering problems using a multivariate method to obtain the best solution.

Thirdly; Parameter prediction by AI in desalination systems

It is necessary to evaluate the scope of the relevant parameters to achieve the optimal configuration in a certain space. The most widely used prediction is energy prediction, which mainly includes the energy input of solar energy and hybrid energy, as well as the scale degree and the system efficiency. Artificial intelligence supports forecasting, as algorithms have unique advantages in the power and customization of the system. (Mashaly and Alazba, 2017) An Artificial Neural Network (ANN) is used to predict the efficiency of solar energy production, and there is also a Back-Propagation Artificial Network (BP-ANN). (Braga et al, 2002).

Fourthly; Automatic control by AI in desalination

The operation stability of seawater desalination processes is of great significance for stable fresh water output. However, there are many variables involved such as such as water temperature, flow rate and operating modes. Thus, timely response and quantitative control are needed for the best working conditions.

The stability of the operation of seawater desalination processes is of great significance for stable fresh water output. However, there are many variables involved, such as water temperature, flow rate, and operating modes. Thus, timely response and quantitative control are needed for the best working conditions.

AI support the power of the machine used and the existing operational management like the human brain (Kalogirou, 2003). While issues such as membrane contamination and concentration polarization of reverse osmosis technology are leading to increased operating costs, AI solves this problem through Model Predictive Control (MPC) that helps overcome the limitations of the control unit in desalination plants. ANN algorithms are the most widely used in AI control because they have the advantages of being simple algorithms and having appropriate accuracy (Zilouchian and Jafar, 2001).

AI applications in water efficiency in agriculture (electronic agriculture, or aquaponics)

Smart technologies, in parallel with AI and Machine Learning methods (ML), have gained great interest from researchers. Smart Tech associated with artificial intelligence methods, machine learning models, and other intelligent systems has proven to be effective in automating and monitoring the growth and safety of these water-based agricultural systems. All models were developed using the open source R programming language, known for its use in statistical computation. For training and testing the results, the data set was divided into two groups, with 80% used for training and the remaining 20% used for testing (Hajeeh and Al-Othman, 2005).

Hydroponics systems are becoming more important, as plants are grown in a nutrient solution that is often custom designed, so that the plants are supplied with all their nutritional and water needs. This is one of the technical forms of farming compared to conventional farming, hydroponics has the particular advantage of producing higher yields, with a higher vegetation density in a very small area and with a lower average water use. The crops are grown suspended in a separate nutrient solution that is kept at a pH suitable for growth. Growing rooms should also be maintained at an appropriate temperature and humidity. The nutrient solutions are usually stored in separate tanks and delivered to crops using a pump and piping. In aquaculture systems, the plants receive their nutrients from fish by-products (usually excreta) stored in adjacent (or close to) tanks and connected through a pump and pipe network. In contrast, crops often act as a water softener for fish by removing their by-products and excreta (AlZu’bi, et al, 2019).

AI and ML methods and other smart applications are used in water and environmental studies, known as “Smart Models Technology”, are adopted; It includes many applications of control and monitoring using smart devices based on the Internet of Things (IoT) which uses smart sensors for control and automation. These settings depend on the use of important inputs to monitor the information and procedures of the systems used (Subramani and Jacangelo, 2015).

In such cases, common water and water sensor data include pH, water temperature, air temperature, humidity, and nutrients that are measured using an ultrasonic sensor, and electrical conductivity. Other inputs to the aquaculture system include total dissolved solids (TDS) and ammonia concentration. With IoT or Smart Sensing for system health monitoring and automation, the outputs are usually associated with nutrient pump feeding, humidity control, temperature control, pH control, and light control. To achieve these levels of automation, IoT systems are coupled with artificial intelligence methods and machine learning models such as smart neural networks and FIS applications. Such systems are coupled with central processing or control units. While Arduino-based controllers are the most common, they have shown tremendous results in implementing machine learning models on harvests, higher yields, fish weights, and nutrient solution component concentrations. These results were achieved using artificial neural network models (He et al, 2022).

Accordingly, AI relates to all the details of modern trends in water treatment for all purposes, especially agriculture. By building intelligent models for algorithms, AI can solve problems sensibly and empirically. With the improvement of goals and the maximization of interests, AI is more suitable to achieve energy saving, environmental protection, and the requirements of effective engineering construction of contemporary society. So, the performance of AI mainly depends on the advanced algorithm. ANN has a significant advantage in regression models. Genetic algorithms seem to be getting better globally. In this context, the combination of reverse modelling methods and AI tools can generate greater potential for optimal operations, especially in complex operating environments. Hybrid AI systems have become increasingly popular in recent years, due to their success in solving many complex problems in the real world. By providing complementary thinking and research methods, components of computational intelligence such as machine learning, fuzzy logic, neural networks, and genetic algorithms generate the synergies that drive success. As such, in the future, AI will tend to use a composite algorithm for prediction and optimization. Through the full use of the advantages of different algorithms, smart control can be improved, which contributes to improving efficiency and increasing fresh water productivity by 10% (Zheng et al, 2022).

Second: AI applications in automating of Adsorption

Adsorption is a process in which attractive (intensive) forces bind to a solid surface (adsorbent). The solid used for adsorption consists of a porous medium with a high internal surface area. It involves the accumulation of atoms or molecules on the surface of a substance. Such a process creates a layer of molecules or atoms that have densely accumulated on the surface of the absorbers. It is different from absorption by diffusion of a substance in a liquid or solid to form a solution. Adsorption has two modes, which can be referred to as follows:

Chemical adsorption: It occurs using what is known as “Van der Waals Forces”, which means a mutual influence forces between the molecules of an electrically neutral substance with each other, where the heat released for this type of adsorption does not exceed the order of 5 free/mol. This adhesion increases with increasing gas pressure, and decreases with increasing temperature; At a given temperature and pressure, the gas is more likely to sorb. The easier it is to liquefy.

Physical adsorption: It is the formation of bonds between the surface molecules of a metal (or any other substance with high surface energy) and another substance (Gas or Liquid) in contact with it. These bonds formed are comparable in strength to ordinary chemical bonds, they are much stronger than the Van der Waals forces characteristic of physical desorption; the heat released by this sorption is in the order of 100 free/mol (Alam et. al, 2022).

Adsorption has received increasing attention from researchers, it focuses on the prediction of the removal of copper ions based on attapulgite clay as the adsorption feedstock; Through artificial intelligence techniques, models have been developed to determine the optimal form of forecasting. These models included an artificial neural network, a support vector machine, and a random field based on artificial neural network optimization. These models combine the smart choices of initial copper concentration, adsorbent dosage, water pH, contact time, and solution ionic strength when added (Al-Gobaisi, 1994).

 Third; Filterless Technology

Researchers at the Massachusetts Institute of Technology have developed a portable desalination device that weighs less than 10 kg. It removes particulates and salts to generate drinking water. This device requires little energy to operate and automatically generates drinking water that meets quality standards. It uses electrical energy to remove particles from drinking water, eliminating the need to replace filters, dramatically reducing long-term maintenance requirements. It may allow the unit to be deployed in remote areas with very limited resources, such as communities where water is not readily available. The device addresses the shortcomings of common desalination methods such as reverse osmosis, which use salt-filtration membranes. But, it requires powerful pumps to maintain the high pressure needed to push water through the membranes. However, it is prone to waste deposition and clogged pores in a membrane with salt and pollutants (Shirazian and Alibabaei, 2017).

The device relies on ion concentration polarization technology instead of filtering water, it applies electric fields to membranes placed above and below a channel of water, and the membranes repel positively or negatively charged particles, including salt particles, bacteria, and viruses, as they pass. The particles are directed charged to a second stream of water to be finally drained. It uses microfluidic device manufacturing methods similar to microchips using materials such as silicon (synthetic rubber). The dissolved and suspended solids process is removed, allowing clean water to pass through the channel itself. As it requires a low-pressure pump, it uses less energy than other technologies. The suspended salts are still at the beginning of the process, so, a second process known as electrodialysis to remove the remaining salt ions is incorporated. AI models such as ML have been used to find the perfect combination of ion concentration polarization and electrowash units. The optimal configuration includes this two-stage polarization process, where water flows through six units in the first stage through three units in the second stage, followed by one electrowashing process. To develop such a dynamic, mobile apps have been created to be controlled wirelessly and monitor real-time data on energy consumption and water salinity.

Fourth: KSA efforts to localize AI in water desalination 

In recent years, AI technologies have been rapidly localized in all development projects within the National Transformation Program (Vision 2030) to meet future challenges in KSA. Due to the relative scarcity of water as the main challenge, water desalination projects are at the heart of the development map in parallel with the boom in the Kingdom in relation to AI. On December 7, 2021, Saline Water Conversion Corporation (SWCC), (was established in 1974 as an independent governmental institution with a legal personality), concluded an MoU with Schneider Electric, a global company specializing in energy utilization and operation, with the aim of activating advanced digital solutions and advanced technologies in AI for the purposes of future projects in the water sector. It boosts exchange of experiences and knowledge to localize artificial intelligence solutions. On June 2, 2022, “SWCC” and the “International Desalination Association” launched the AI system, which includes a set of platforms and technologies related to the activation of artificial intelligence solutions as follows: the Integrated Platform for Security Management, Smart Safety SSS, the Smart Operation and Maintenance (SOM), the Smart Facilities platform (SC), the Smartlog Warehouse Management System (SWM), the Smart Projects Management platform (SPM), the integrated platform for 3D VXP Platform, and the Drones Platform (DP). The conference comes as a pioneering step in the international efforts to face the water desalination challenges, both at the cost level (reaching the cost of desalination per litre to $0.32), or at the level of reducing carbon emissions (one of the conference objectives is to reduce it to 50% ), or even at the level of non-financial revenues that the conference aims to increase to 10%.

It is worth mentioning the magnesium processing facilities, the first commercial system in the world ever to use nanotechnology to produce magnesium and pump it into the water through Mineral Extraction from Reverse Osmosis Water, developed by the General Organization for Water Conversion in Saudi Arabia. In addition, the Kingdom is organizing an international conference under the title “The Future of Desalination” in Riyadh (11 – 13 September 2022).

The conference represents a pioneering step in the international efforts to meet the water desalination challenges, both at the cost level (reaching the cost of desalination per litre to $0.32), or the level of reducing carbon emissions (one of the conference objectives is to reduce it to 50% ), or even at the level of non-financial revenues that the conference aims to increase to 10%. So, the rich agenda of the conference reflects an ambitious step within the National Transformation Program “Vision 2030” adopted by the Saudi leadership to meet the challenges of the future and open broad horizons for the localization of AI technologies in water desalination for agricultural purposes, in addition to supporting the ambitious and serious development programs to ensure the Kingdom’s regional and global leadership in this vital area.

Saudi’s Vision 2030 Studies Unit *