An integrated renewable energy and machine learning framework for techno economic analysis of water and energy nexus management in arid climates

Abstract

Water management in arid regions, such as Basra, Iraq, faces escalating challenges due to water scarcity and increasing energy demand. This study investigates the integration of machine learning with renewable energy technologies to optimize water and energy efficiency in such environments. A multi-scenario approach was employed, combining advanced water treatment technologies, energy recovery systems, and smart grid integration to assess their impact on sustainability. This study evaluated a comprehensive techno-economic analysis of the integration of machine learning models and renewable energy technologies, marking a significant step toward more sustainable and efficient water-energy nexus management in arid climates. The solar-powered UV disinfection system reduced energy consumption by 30%, while membrane filtration techniques minimized water loss by 20%. The adoption of pressure recovery turbines improved energy efficiency by 25%, resulting in significant energy savings of 800 kWh annually and a reduction of 400 kgCO₂ emissions. Smart grid systems enhanced operational efficiency, reducing energy wastage by 15% and improving water distribution by 25%. Machine learning models, including the M5 model tree and recurrent neural networks (RNN), were applied to predict and optimize system performance, highlighting their ability to handle complex, non-linear relationships between energy and water variables. The results proposed a scalable framework for integrating machine learning-driven renewable solutions into water-energy systems in water-stressed regions, addressing global challenges in water management, supporting climate adaptation strategies, and contributing to the United Nations Sustainable Development Goals (SDGs).

Introduction

The interconnection of water, energy, and food is critically important for tackling global sustainability issues¹. This interconnected framework recognizes the intrinsic relationship between these essential resources, highlighting their interdependence and the need for integrated management approaches². Water is essential for agriculture and food production, while energy is required for water pumping, treatment, and distribution, as well as food processing and transportation^3,4. Conversely, water and energy systems are highly reliant on each other, with water often being used in energy production processes like hydropower generation and cooling in thermal power plants^5,6. Understanding and managing the interactions between water, energy, and food systems is crucial for achieving sustainable development goals, ensuring food and water security, and mitigating environmental impacts such as water scarcity, energy consumption, and climate change^7,8. Embracing a nexus approach enables policymakers, researchers, and stakeholders to identify synergies, trade-offs, and opportunities for integrated solutions that enhance resource efficiency, resilience, and equitable access to water, energy, and food resources^9,10.

The integration of solutions in water management infrastructure plays a crucial role within the broader context of the water-energy-food nexus, which highlights the interconnectedness of essential resources. Innovative approaches, such as renewable energy technologies, hold potential to enhance water treatment efficiency and energy recovery processes¹¹. For example, solar-powered ultraviolet (UV) disinfection and pressure recovery turbines can improve the sustainability of water systems while supporting reliable access to clean water for agricultural irrigation^12,13. Additionally, smart grid technologies offer pathways to optimize energy usage in water treatment facilities, thereby reducing energy wastage and improving system reliability. These advancements align with the water-energy-food nexus framework, which seeks to promote resource efficiency, resilience, and sustainability across interconnected systems vital for human well-being and environmental health^14,15.

The integration of renewable energy technologies into water management systems represents a pivotal advancement in addressing the pressing challenges of water scarcity, energy consumption, and environmental sustainability on a global scale^16,17. Traditional water treatment and distribution processes are often energy-intensive and reliant on fossil fuels, contributing to greenhouse gas emissions and exacerbating climate change^18,19. By transitioning towards renewable energy sources such as solar, wind, hydroelectric, and geothermal power, water management systems can significantly reduce their carbon footprint while enhancing operational efficiency and resilience^20,21. The integration alleviates the ecological consequences linked to traditional energy sources and provides enduring economic advantages via diminished energy expenses and enhanced energy autonomy^22,23. Moreover, it aligns with international sustainability agendas, including the United Nations Sustainable Development Goals (SDGs), particularly Goal 6 (Clean Water and Sanitation) and Goal 7 (Affordable and Clean Energy), by promoting the efficient and sustainable use of water resources while advancing access to clean and renewable energy for all²⁴.

A thorough review of existing literature on the integration of renewable energy technologies in water management reveals a rich landscape of research, innovation, and practical applications across diverse contexts globally^25,26. Studies have extensively explored the potential of renewable energy sources, including solar, wind, hydroelectric, and biomass, to power various aspects of water management systems, from treatment and desalination to distribution and irrigation^27,28. Researchers have investigated the technical feasibility, economic viability, and environmental benefits of renewable energy integration, highlighting its potential to reduce energy costs, carbon emissions, and reliance on finite fossil fuels^25,29. Moreover, literature reviews have examined the role of policy frameworks, regulatory incentives, and institutional support in facilitating the adoption and scaling up of renewable energy technologies in the water sector³⁰. Case studies and empirical analyses have provided valuable insights into successful integration strategies, technological innovations, and best practices, offering valuable guidance for policymakers, practitioners. Despite notable progress, gaps in knowledge persist, particularly regarding the scalability, performance optimization, and socio-economic implications of renewable energy integration in water management^25,31.

Previous studies have focused on individual aspects of advanced water treatment or energy recovery systems but lack a comprehensive framework that integrates these innovations into real-world water management systems. For example, research on UV/NiFe₂O₄/clinoptilolite processes³², UV/ZnO processes³³, and electrocoagulation method³⁴ has demonstrated the potential for wastewater treatment in synthetic and industrial settings. However, these studies are predominantly laboratory-based and do not address the scalability of such technologies in regions facing acute water scarcity and energy inefficiency³⁵. Similarly, while existing reviews of linear alkylbenzene production provide valuable insights into industrial water usage, they do not consider the integration of renewable energy technologies or smart grid systems to optimize water and energy use³⁶.

Despite recent progress, the existing body of research still presents notable gaps that limit practical advancements. As it was discussed earlier, most prior studies have focused narrowly on isolated applications of renewable energy or water treatment technologies, often within industrial or synthetic wastewater contexts, without addressing their combined implementation in water-stressed municipal systems—such as those found in arid regions. Furthermore, the long-term techno-economic feasibility and socio-environmental impacts of combining multiple innovations—such as pressure recovery turbines, solar-powered disinfection units, and smart grids—remain insufficiently explored. A particularly critical oversight in the current literature is the underutilization of machine learning techniques for performance prediction, anomaly detection, and system optimization. The lack of data-driven frameworks undermines the ability to proactively manage system inefficiencies and adapt to fluctuating operational demands. Addressing these multidimensional gaps is essential for developing resilient and intelligent water-energy infrastructures tailored to the complex conditions of resource-constrained urban environments.

Basra, Iraq, faces significant challenges stemming from water scarcity and energy constraints, making it a critical focal point for addressing the interplay between these two interconnected issues^37,38. Traditional water treatment technologies in the region, such as chlorination and basic sedimentation, have been employed for decades. While these methods are effective at disinfecting and clarifying water, they are highly energy-intensive due to their reliance on extensive pumping, chemical processing, and inefficient equipment³⁹. Additionally, the lack of advanced filtration or energy recovery systems further exacerbates resource depletion, as substantial amounts of energy are consumed during water intake, treatment, and distribution. The energy required for these conventional processes is predominantly derived from fossil fuel-based power plants, leading to high operational costs, frequent supply disruptions, and significant greenhouse gas emissions⁴⁰. In a region characterized by arid climatic conditions and dwindling freshwater supplies from the Tigris and Euphrates rivers, the unsustainable use of both water and energy resources poses a severe threat to Basra’s environmental and economic resilience^41,42. Moreover, monitoring systems for water quality and energy usage in Basra’s infrastructure are rudimentary, often limited to manual inspections and basic controls, which restrict the ability to optimize performance and reduce wastage. Currently, Basra does not extensively employ energy recovery devices such as pressure recovery turbines, which could harness hydraulic energy from high-pressure water flows to offset energy costs. Similarly, the absence of smart grid integration for real-time monitoring and adaptive control further compounds inefficiencies, leaving substantial room for improvement⁴³. The reliance on outdated systems underscores the pressing need for a shift towards renewable energy technologies that not only reduce energy consumption but also enhance water treatment efficiency. Understanding the limitations of current systems and their linkages to energy constraints is essential for justifying and implementing the proposed shift^43,44.

This study aims to systematically evaluate the integration of renewable energy technologies into Basra’s municipal water infrastructure in order to address pressing challenges of water scarcity and energy inefficiency. The specific objectives are:

• To quantitatively assess the impact of solar-powered UV disinfection systems and membrane filtration techniques on energy consumption and water loss.

• To evaluate the performance of pressure recovery turbines in enhancing energy recovery efficiency.

• To analyze the effectiveness of smart grid technologies in reducing operational energy wastage in water facilities; and.

• To employ machine learning models for predictive analysis and anomaly detection to support real-time optimization of water-energy systems.

The novelty of this study lies in its integrative and data-driven approach to improving water and energy infrastructure in water-stressed urban environments. Unlike previous studies that address renewable energy or water management technologies in isolation, this research combines solar-powered UV disinfection, membrane filtration, pressure recovery turbines, and smart grid systems into a unified operational framework tailored for Basra’s municipal context. Furthermore, the study incorporates advanced machine learning techniques (such as predictive modeling and anomaly detection) to enhance system responsiveness and fault detection and optimize performance in real time. This study introduces a hybrid methodology that enhances technical efficiency, reduces operational costs, and improves system resilience. This multidimensional innovation marks a significant step beyond existing literature by aligning technological implementation with intelligent data analysis, making the proposed framework adaptable and scalable for similar resource-constrained regions.

Materials and methods

Research objectives and hypotheses

The primary objective of this study is to assess the impact of three key innovations (advanced water treatment technologies, energy recovery systems, and smart grid integration) on Basra’s water management infrastructure. The hypotheses of this study include, adoption of advanced water treatment technologies, specifically solar-powered UV disinfection systems and membrane filtration techniques, would lead to a reduction in energy consumption and water loss. These technologies were chosen based on their proven efficacy in addressing key challenges in water management systems. Solar-powered UV disinfection systems offer a sustainable and energy-efficient alternative to traditional chlorine-based disinfection by utilizing abundant solar energy⁴⁵. Similarly, membrane filtration technologies, such as reverse osmosis and ultrafiltration, were selected for their ability to remove contaminants effectively and minimize water loss during the treatment process. The methods were identified as optimal solutions due to their alignment with the objectives of reducing energy consumption and improving water resource efficiency in water-stressed areas⁴⁶. Implementation of energy recovery systems would enhance energy efficiency and sustainability in water management practices⁴⁷. Utilization of smart grid technologies would optimize energy usage and reduce wastage in water treatment facilities⁴⁸.

To systematically evaluate the impact of renewable energy technologies on Basra’s water management system, three distinct scenarios were developed. Scenarios consider variations in key parameters, including solar panel efficiency, water flow rate, and discount rate, which significantly influence system feasibility and performance. Scenario 1 represented the baseline conditions with standard operational parameters based on existing infrastructure and technologies. Scenario 2 introduced moderate improvements in solar panel efficiency and water flow rate, reflecting gradual advancements in renewable energy integration. Scenario 3 explored an optimized configuration with the highest efficiency levels, cost reductions, and enhanced resource utilization to maximize sustainability benefits.

These scenarios allowed for a comparative assessment of system performance under different conditions, ensuring a robust analysis of potential improvements in water-energy management.

Research framework

This research design outlines the methodology and procedures followed to investigate the effectiveness of integrating renewable energy technologies into water management systems in arid climate (Fig. 1).

Fig. 1.

Schematic of proposed methodology.

Step 1. Data collection: Data collection involved gathering information on innovative technologies and practices related to renewable energy integration in water management systems. This included sourcing data on advanced water treatment technologies such as solar-powered UV disinfection systems and membrane filtration, energy recovery systems like pressure recovery turbines, and smart grid integration strategies utilizing advanced sensors and data analytics. The data used in this study were obtained from a combination of experimental results, literature reviews, and case studies. Specifically, turbine efficiency values (e.g., 90%) and flow rate parameters (e.g., 200 m³/h) were derived from vendor specifications and validated against published benchmarks in renewable energy studies. Solar panel efficiency values (e.g., 20–30%) and irradiance data (e.g., 500 W/m²) were based on measurements from comparable arid regions. Historical water flow rates and seasonal variability data were sourced from Iraq’s Ministry of Water Resources. Because this study did not involve human or animal subjects; therefore, ethical approval was not required. Also, all simulation parameters and data sources used in this study are available upon reasonable request. The machine learning models were implemented using WEKA and TensorFlow, and hyperparameters were tuned, ensuring reproducibility.

Step 2. Data processing: During this step, specific attention was given to processing data related to the identified innovations. This included extracting relevant variables and parameters associated with advanced water treatment technologies, energy recovery systems, and smart grid integration from the collected datasets. Data were cleaned, normalized, and transformed to ensure compatibility with subsequent analysis.

Step 3. Exploratory data analysis: This step focused on understanding the characteristics and distributions of data pertaining to the identified innovations. Descriptive statistics and visualization techniques were used to explore patterns and relationships within the data, with a specific emphasis on variables related to advanced water treatment technologies, energy recovery systems, and smart grid integration. Correlation analysis was conducted to assess the interdependencies between variables and identify potential predictors or drivers of innovation adoption and performance.

Step 4. Feature selection: Feature selection techniques were applied to identify the most relevant variables. This involved assessing the importance of individual features related to advanced water treatment technologies, energy recovery systems, and smart grid integration using statistical methods and ML algorithms.

Step 5. Model development: Model development focused on incorporating the identified innovations into predictive modelling and anomaly detection frameworks. Machine learning algorithms were adapted to extract the features associated with advanced water treatment technologies, energy recovery systems, and smart grid integration. Models were trained to predict outcomes (energy consumption, water quality parameters, and equipment performance) with the aim of optimizing the integration and performance of these innovations within Basra’s water management infrastructure.

Step 6. Model evaluation: The performance of the developed models was assessed in terms of their ability to accurately predict outcomes related to the identified innovations. For the machine learning models (M5 and RNN), the dataset was divided to training (80%) and testing (20%) subsets to evaluate the models’ generalizability. Cross-validation with five folds was performed to mitigate overfitting and ensure robust performance. The WEKA software platform was employed for the development and evaluation of the M5 model tree, while TensorFlow was used for implementing and fine-tuning the RNN framework.

Advanced water treatment technologies

UV disinfection systems and membrane filtration represent innovative approaches to water treatment that offer promising solutions for enhancing water quality while reducing energy consumption^49,50. Solar-powered UV disinfection systems utilize the sun’s energy to deactivate harmful pathogens and microorganisms present in water, providing a cost-effective and environmentally friendly alternative to conventional chlorine-based disinfection methods^51,52. These systems are particularly advantageous in hot and arid regions, where abundant sunlight is available year-round. Similarly, membrane filtration technologies, such as reverse osmosis and ultrafiltration, employ semi-permeable membranes to remove contaminants, suspended solids, and microorganisms from water, resulting in purified water suitable for various applications^53,54.

In this study, the solar-powered UV disinfection system was designed with specific technical parameters that ensured its effectiveness in pathogen inactivation and energy efficiency. The specifications of this system are detailed in Table 1 below. Parameters were carefully selected to meet global water safety standards while optimizing energy usage, particularly in arid regions.

Table 1.

Technical specifications of the solar-powered UV disinfection system.

Parameter/Variable	Value	Reason	Reference
Wavelength	254 nm	The wavelength of 254 nm is highly effective in inactivating a wide range of pathogens, including bacteria, viruses, and protozoa, due to its absorption by DNA and RNA	55
UV dose	30 mJ/cm²	A UV dose of 30 mJ/cm² is required by WHO and EPA standards for effective disinfection of drinking water, ensuring adequate microbial inactivation, especially in arid climates	56
Energy source	Solar energy	Solar energy is used to power the UV disinfection system, leveraging abundant sunlight in arid regions like Basra, Iraq, to provide an energy-efficient solution for water treatment	55
Effectiveness in arid regions	High effectiveness under variable sunlight	Solar-powered UV systems are effective even under fluctuating sunlight levels, maintaining a sufficient UV dose while reducing energy consumption	55

The 254 nm wavelength is commonly used in UV disinfection systems because it is highly effective in inactivating a broad spectrum of pathogens. This wavelength is specifically absorbed by the DNA and RNA of microorganisms, leading to the formation of pyrimidine dimers that prevent replication, effectively rendering the pathogens inactive. This has been widely validated by global health organizations such as WHO and the EPA. A UV dose of 30 mJ/cm² is recommended by both WHO and the EPA for ensuring effective microbial disinfection of drinking water. This dose is sufficient to deactivate pathogens such as bacteria and viruses to meet international safety standards for water quality. This dose was selected as it aligns with established guidelines and ensures the system’s compliance with global water safety standards. The use of solar energy for powering UV disinfection systems in this study is a key aspect of its sustainability. In regions like Basra, Iraq, where sunlight is abundant year-round, solar-powered UV systems offer an energy-efficient solution by reducing operational costs and reliance on conventional energy sources. This use of renewable energy makes the system cost-effective and environmentally friendly. Solar-powered UV systems are particularly effective in arid regions, where sunlight can vary in intensity throughout the day or season. Despite these fluctuations, the UV disinfection system maintains reliable performance by ensuring that an adequate UV dose is delivered during peak sunlight hours.

Energy recovery systems

Energy recovery systems has an important role in optimizing the sustainability and efficiency of water treatment processes through harnessing energy from various stages of the treatment cycle⁵⁷ (Table 2).

Table 2.

Technical specifications of the energy recovery systems.

Parameter/variable	Value	Reason
Power output	0.4 kW	Based on typical operational performance of small-scale PRTs in urban water systems. This is a reasonable output for systems handling 200 m³/h of water flow
Energy recovery efficiency	85%	This efficiency is typical for small-scale pressure recovery turbines operating in municipal systems. Previous studies show efficiencies in the 80–90% range
Pressure range	1–3 bar	The pressure range is suitable for municipal systems where pressure fluctuations are within this level
Water flow rate	200 m³/h	This is the typical flow rate for small to medium-scale municipal water systems
Energy source	Renewable	Solar energy, used for powering associated systems like pumps, ensuring a sustainable solution. This is based on the availability of sunlight in Basra

In Basra, for pressure recovery turbines (PRT) systems, an output of 0.4 kW and an energy efficiency of 85% are quite reasonable and standard for small-scale systems. An efficiency of 85% for this type of system is typically used in low-pressure and medium-flow conditions in urban areas. In the context of Basra’s water management infrastructure, these turbines offer a promising avenue for reducing energy consumption and operational costs associated with water treatment⁵⁸. Therefore, energy recovery systems, particularly pressure recovery turbines, present a practical and sustainable solution for advancing Basra’s water management infrastructure while promoting resource efficiency and resilience⁵⁹.

Smart grid integration

Smart grid technologies offer a transformative solution for optimizing energy usage and distribution within water treatment facilities, providing unprecedented levels of control, efficiency, and resilience⁶⁰ (Table 3).

Table 3.

Technical specifications of the smart grid integration.

Parameter/variable	Value	Reason
Grid efficiency	90%	Achieving a high grid efficiency of 90% is feasible through the implementation of smart grid systems that reduce energy wastage and optimize flow. This is achievable with modern grid technologies
Energy wastage reduction	15%	Smart grids typically reduce energy wastage by 10–20%, depending on the extent of automation and optimization. A 15% reduction is realistic in the Basra context
Reliability improvement	20%	Smart grid integration often improves system reliability by 15–25% by allowing real-time monitoring and control. A 20% improvement is considered reasonable in a water-stressed region
Operational cost savings	US$ 50,000/year	Operational savings due to reduced energy consumption and improved grid management can reach up to US$ 50,000 annually in medium-sized systems
Renewable energy integration	Solar	Solar energy integration is common in smart grid systems for water and wastewater treatment in arid regions, particularly when local energy resources are limited

In Basra, a 15% reduction in energy waste is reasonable, as smart grids can help manage energy consumption and optimize systems. Operational savings of up to $50,000 per year may be achieved in medium-sized systems using smart grids and renewable energy. In Basra’s water treatment facilities, smart grid integration holds immense potential for improving operational efficiency and reducing costs by optimizing energy usage, minimizing wastage, and enhancing system reliability⁶¹. Additionally, smart grids facilitate seamless integration of renewable energy sources, such as solar and wind power, into the water treatment process, further enhancing sustainability and reducing reliance on conventional energy sources⁶². Through adaptive control systems and predictive analytics, smart grid technologies empower water treatment facilities to anticipate and respond to fluctuations in energy supply and demand, ensuring uninterrupted operation and optimal performance even in challenging conditions. Therefore, the implementation of smart grid technologies represents a critical step towards achieving energy efficiency, sustainability, and resilience in Basra’s water management infrastructure⁶³.

Mathematical modelling

A mathematical model was developed to assess the feasibility and effectiveness of the innovative renewable energy solutions proposed for Basra’s water management infrastructure. For this purpose, the assumptions were made in this study to ensure consistent modelling and analysis. The efficiency of the systems (e.g., pressure recovery turbines, solar-powered UV disinfection systems) remains constant during the study period, as no long-term degradation data were available. Scaling up of the proposed systems was assumed to follow a linear trend, meaning that performance improvements are directly proportional to increases in size or capacity. Basra’s water source (the Tigris and Euphrates rivers) was assumed to sustain the proposed discharge rates, based on historical flow data and seasonal averages provided by Iraq’s Ministry of Water Resources. Capital costs and operational costs for the technologies were derived from industry-standard vendor quotes and do not account for unexpected fluctuations in material or labour costs.

The model incorporates main parameters such as energy consumption, treatment efficiency, and cost considerations to evaluate the potential benefits of solar-powered UV disinfection systems, membrane filtration, and energy recovery systems. For the solar-powered UV disinfection systems, the mathematical model included equations to calculate the energy required for UV disinfection, taking into account factors such as solar irradiance (I), disinfection efficiency (η), and water flow rate (Q). The energy required (E_uv) can be calculated using Eq. 1⁶⁴:

1

where, t represents the exposure time (in hours), A denotes the surface area of the UV disinfection system (in m²), η is the disinfection efficiency (in %), and I is the solar irradiance (in W/m²).

For membrane filtration, the model incorporates equations to estimate energy consumption and water loss reduction. The energy required for membrane filtration (E_mf) can be calculated as Eq. 2⁶⁵.

2

where, P represents the power consumption of the membrane filtration system (in kW), Q denotes the water flow rate (in m³/s), and t is the operational time (in hours).

Additionally, the model includes equations to evaluate the energy recovery efficiency of pressure recovery turbines. The energy recovered (E_rec) can be determined using the Eq. 3⁶⁶:

3

where, η_turbine represents the turbine efficiency, ρ denotes the density of water, Q is the water flow rate, V₁ is the velocity of water entering the turbine, and V₂ is the velocity of water exiting the turbine.

Sensitivity analysis and cost analysis

Sensitivity analysis was conducted to assess the impact of variations in key parameters on the performance of each renewable energy solution. Sensitivity analysis involved varying parameters such as solar irradiance, disinfection efficiency, membrane filtration efficiency, turbine efficiency, and water flow rates to evaluate their influence on energy consumption, treatment efficiency, and cost-effectiveness.

This analysis involved the calculation of the return period (RP), net present value (NPV), and internal rate of return (IRR). It should be mentioned that the period of return is termed discounted when the discount rate is associated with the IRR. The discount rate is expressed as a percentage and is regarded the sum of the expenses of return on capital, opportunity cost, risks and inflation. The discount rate for Iraq is 11.60%⁶⁷. The overall expense of installing the mentioned setups encompasses the aggregate of capital costs (CC), operational and maintenance costs (OMC), and civil construction costs (CWC). The CC include all expenses related to the necessary equipment for the related installation. The OMC comprises of the maintenance and individual operation of the device, while the CWC assesses the price of essential civil works for the conversion of the installation to the grid. The expected lifespan of typical mechanical and electrical equipment is deemed between 10 and 15 years⁶⁸. The operating and maintenance costs are expected to be 0.50% and 2.50% per year, respectively, of the CC⁶⁹. For CWC, 30% of the cost of capital was adopted. The energy tariff was US$ 0.14/kWh⁷⁰. In addition, an increase of 8.00% in the energy tariff was considered, estimated from the yearly increase in prices in the period from 2018 to 2023 of the local power organizations⁷¹.

Machine learning application

M5 model tree

The M5 model tree was chosen for this study due to its capability to model complex relationships in data while maintaining interpretability and computational efficiency. This is particularly important in the context of water-energy nexus systems, where the relationships between variables such as energy efficiency, water consumption, and environmental factors are non-linear and multifactorial. The M5 model tree’s ability to handle multi-dimensional data, coupled with its feature selection process, makes it well-suited for predicting outcomes in scenarios where numerous interdependent variables are at play, such as in Basra’s water management infrastructure. Additionally, its adaptability to handle both continuous and categorical data is essential for integrating diverse types of input variables, including energy consumption, water quality metrics, and environmental factors, which are all pivotal in this research. The M5 model tree employs multi-linear regression, resembling piecewise linear functions. This adaptability allows it to efficiently handle multi-dimensional tasks, contributing to its widespread adoption across various engineering fields. The model operates by dividing the input dataset into subsets, denoted as set T, and further subdividing them into multiple subsets through leaf evaluation to prevent overfitting. The information gain ratio, derived from the reduction of standard deviation criteria before and after testing, guides the splitting process. If subsets contain few samples or minimal variability, they undergo further partitioning based on test results. Each subset (T_i) corresponds to a specific test outcome, with error reduction calculated based on the standard deviation of the final sample amounts in T_i, as proposed by Quinlan (Eq. 4) ^72,73.

4

In this study, the M5 model tree selects the split that maximizes the expected error reduction. To explore relationships and visualize the tree model, the WEKA software was employed^73,74. The WEKA software (version 3.9.6) was used to implement the M5 model tree. The model was configured to maximize error reduction during the training process, with key hyperparameters such as tree depth and minimum instances per leaf optimized using a grid search algorithm to ensure the best possible fit for the dataset.

The M5 model tree was configured to balance prediction accuracy and computational efficiency. During the training process, hyperparameters, including tree depth and minimum instances per leaf, were tuned using a grid search algorithm to identify the optimal configuration. The feature selection process identified key predictors of water and energy performance, allowing for accurate and interpretable results⁷³.

Recurrent neural networks (RNNs)

RNN were selected for this study due to their ability to model and predict sequential data, which is essential in the context of water-energy systems where data is time-dependent. Water consumption, energy use, and other environmental variables are often correlated over time, and RNNs excel in capturing these temporal dependencies. This makes them particularly suitable for predicting future values based on past observations, which is critical for tasks like energy consumption prediction and anomaly detection in water management systems. The RNN architecture is especially powerful in tasks involving sequential information, such as time-series forecasting and real-time monitoring. The ability to retain information from previous time steps allows the model to make predictions that are aware of historical data, which is crucial when analysing the dynamics of water and energy usage.

RNNs are a type of artificial neural network designed to handle sequential data by introducing connections between neurons in the network to form directed cycles. This architecture allows RNNs to exhibit dynamic temporal behaviour, making them well-suited for tasks involving sequential information, such as time series analysis, natural language processing, and speech recognition. RNNs can process input sequences of varying lengths and capture dependencies between elements in the sequence over time. RNNs have connections that loop back on themselves, allowing information to persist and be passed from one-time step to the next. The formulation of RNNs can be expressed as Eq. 5:

5

where k is the discrete (time) index, and N is the final finite horizon time, s_k is the m-d input vector sequence, and h_k is the n-d output via the nonlinear function σ_k. Here, σ_k is a general nonlinear function which may be specified to be the logistic function sigm or the hyperbolic tangent tanh.

The RNN model was implemented using TensorFlow 2.6 (an open-source machine learning library). This version is widely used for building and training deep learning models, including RNNs, and has been optimized for performance. During the training process, the model’s hyperparameters, including learning rate, batch size, and number of hidden layers, were tuned using a grid search algorithm to identify the most efficient configuration. The learning rate was set to 0.001, a typical value that balances convergence speed and accuracy. The batch size was set to 32 to ensure efficient processing without overfitting, and the number of hidden layers was set to 2 for capturing deeper temporal relationships within the data. The RNN model was trained on a dataset consisting of daily water consumption and energy usage data over a period of two years, ensuring that long-term dependencies were appropriately learned⁷⁵.

Inputs, outputs, and model configuration

In this section, the inputs, outputs, and the rationale behind selecting the input variables for the M5 model tree and RNN are tried to be clarified. This section also addresses the reproducibility of the study and ensures that readers have a clear understanding of how the models were trained and validated.

For the M5 model tree, the inputs included water consumption data, energy usage data, and environmental factors such as temperature and humidity. The variables were selected based on their relevance to energy and water performance in Basra, Iraq, as they directly influence both water usage and energy demand in arid climates. For the RNN, the inputs were time-series data that included daily records of water consumption and energy usage over the study period. The reason for choosing time-series data is to capture the temporal dependencies between past water and energy usage and to predict future consumption patterns. Water consumption and energy usage are the primary factors that influence the overall energy-water nexus. These inputs were selected based on their direct correlation with system performance, and their inclusion allows the models to effectively predict energy savings, water loss reduction, and overall efficiency improvements. Environmental factors such as temperature and humidity were included because they were critical for understanding the seasonal variation in water and energy demand, particularly in arid regions like Basra. The M5 model tree predicted the energy efficiency and water distribution efficiency, providing a numerical evaluation of system performance under different scenarios. The RNN predicted future water consumption and energy usage based on historical data, which was significant for optimizing energy use and minimizing waste in the water management system.

Efficiency criteria

The efficiency of the models was evaluated by the determination coefficient (R²) and root mean square error (RMSE) as Eqs. 7 and 8⁷⁶.

6

7

.

Where R²RMSE, N, , , are determination coefficient, root mean square error, number of observations, observed variable data, calculated variable and mean of observed variable data, respectively⁷⁷.

In addition, accuracy, precision, recall and F1score are commonly used in classification tasks to evaluate the performance of a model (Eqs. 8–11).

8

9

10

11

Case study

The significance of conducting this study in Basra, Iraq, lies in the acute challenges faced by the city regarding water scarcity and energy constraints (Table 4).

Table 4.

Challenges and technological solutions for basra’s water management.

Challenges	Description	Relationship to innovations
Water scarcity	Diminishing water resources	Improving water treatment efficiency Reducing water loss
Energy constraints	Energy in electricity supply, water treatment facilities/distribution networks	Harness energy Reducing dependence on external energy sources Mitigating the impact of energy constraints
Infrastructure vulnerability	Susceptible to disruptions	Enhancing the resilience of water infrastructure Enabling real-time monitoring adaptive control to mitigate risks

Basra represents a poignant case study due to its location in a region grappling with increasingly severe water scarcity issues exacerbated by climate change and population growth (Fig. 2).

Fig. 2.

Location of studied area, Basra, Iraq.

Additionally, Iraq’s energy infrastructure faces significant challenges, with electricity shortages impacting water treatment and distribution processes. This study sheds light on the pressing need for sustainable water management solutions in a context where traditional approaches have proven insufficient in Basra. Moreover, Basra’s strategic importance as a major economic hub and port city underscores the urgency of addressing water and energy challenges to ensure the city’s long-term viability and resilience.

This study emphasizes the practical challenges of integrating renewable energy technologies into Basra’s water management infrastructure, including the scarcity of reliable historical data and the variability of water and energy supply. Despite these challenges, the integration of renewable energy technologies demonstrated measurable improvements in energy efficiency and system reliability.

Results and discussions

Analysis of advanced water treatment technologies

This section presents the outcomes of the study on the integration of advanced water treatment technologies, specifically UV disinfection systems and membrane filtration methods, in Basra, Iraq. The study aims to assess the performance of these systems in improving water quality, reducing energy consumption, and enhancing operational efficiency in an arid, water-stressed environment (Table 5).

Table 5.

Energy performance with and without technologies.

System configuration	Average COP	Energy savings (%)	Key observations
Without technology	2.2–6	0%	Lower energy efficiency in high thermal load scenarios
With technology	3.2–4.1	18% improvement	Increased energy efficiency, especially in high thermal loads
High thermal load	3.5–4.5	41% improvement	Significant energy savings in high demand periods

The data demonstrated that the cooling system with the UV and filter operating showed a consistently higher coefficient of performance (COP) compared to when the technologies were not in use. The average COP improved by 18% when the technologies were in operation, highlighting the filter’s positive impact on energy efficiency. During periods of high thermal load, this difference became even more significant, with a 41% improvement in energy performance when the system was active.

The results are also matched with findings from other studies focused on energy recovery and treatment systems. For example, similar research on membrane filtration system has demonstrated the effectiveness of utilizing this method for improving water treatment efficiency and reducing energy usage in municipal systems. The energy savings reported in these studies often range from 5 to 30%, depending on the operational conditions and the type of filtration system used⁷⁸.

Analysis of energy recovery systems

The calculations were based on the standard hydropower energy recovery equation (Eq. 12):

12

where η_turbine is the turbine efficiency (assumed to be 90%), ρ is water density (1000 kg/m³), g is the gravitational acceleration (9.81 m/s²), h is the effective head, and Q is the flow rate. The effective head was derived from velocity differences as (Eq. 13):

13

The flow rate Q = 200 m³/h was converted to 0.0556 m³/s. Substituting these values, the estimated turbine output is (Eq. 14):

14

Assuming 2000 h of annual operation, the net recoverable energy is approximately (Eq. 15):

15

Accordingly, the reduction in carbon emissions is also calculated as below (Eq. 16):

16

This performance still supports the applicability of PRTs for energy recovery in water infrastructure, especially under resource-constrained conditions. Supporting literature confirms the feasibility of achieving turbine efficiencies up to 90% in small-scale and low-head systems. For instance, VLH turbines have been documented at 85–90% efficiency in heads below 3 m⁷⁹. Similarly, crossflow turbines in water supply systems have demonstrated peak performance above 87% under design conditions⁸⁰.

Figure 3 presents the quantitative results obtained from the implementation of PRTs in the Basra water infrastructure.

Fig. 3.

Normalized quantitative results from the integration of pressure recovery turbines in Basra.

As Fig. 3 depicted, the proposed method achieved an energy recovery efficiency of 90%, compared to 83.33% for conventional methods, showing an 8% improvement. This improvement is reasonable and acceptable, as pressure recovery turbines in small, low-head systems, such as municipal water systems, could typically reach efficiency values of 85% or higher. Previous studies have demonstrated that VLH and crossflow turbines could achieve similar efficiency under optimal flow conditions. Furthermore, the proposed method saved 33,000 kWh/year, which is 10% higher than the 30,000 kWh/year savings achieved by conventional methods. This modest improvement matched with expectations for energy-efficient recovery systems, where pressure recovery turbines in small-scale applications could provide similar energy savings. The predictions made in this study are consistent with real-world applications, indicating that the proposed method performed better than traditional methods. In terms of carbon emissions reduction, the proposed method led to a 16,500 kg CO₂/year reduction, compared to 15,000 kg CO₂/year for conventional methods, showing a 10% improvement. This carbon reduction was consistent with energy recovery systems using renewable energy. The application of pressure recovery turbines contributed to the overall environmental benefit by reducing greenhouse gas emissions. The results were aligned with sustainability goals, demonstrating that the proposed method not only saves energy but also mitigates carbon emissions effectively. The water flow rate remained the same for both systems, reflecting no change in flow rate. The difference in performance came from the efficiency of the turbines and the recovery of energy, rather than changed in the water flow. Regarding power consumption, the proposed method reduced power consumption to 89% of conventional methods, showing an 11% improvement in energy efficiency.

The improved energy recovery efficiency of 90% aligned with documented data from crossflow and VLH turbines, which have demonstrated the ability to reach similar efficiency values in small-scale, low-head applications. The energy savings and carbon emissions reduction in the proposed method were logically consistent with the use of renewable energy and optimized turbine systems.

Analysis of smart grid technologies

Table 6 presents the key performance indicators obtained from this implementation, including energy wastage reduction, operational cost savings, and reliability enhancements achieved through smart grid integration.

Table 6.

Comparison of quantitative results from the implementation of smart grid technologies and conventional methods.

Performance indicator	Smart grid integration	Conventional methods
Reduction in energy wastage (%)	15	5
Operational cost savings ($)	50,000	30,000
Improvement in reliability (%)	20	10
Reduction in water loss (%)	25	15
Increase in operational efficiency (%)	30	15

The metrics in Table 6 were derived from operational data and modeled performance. The reduction in water loss was calculated using the Eq. 17:

17

.

With a baseline loss of 10% and post-implementation loss of 7.5%. Similarly, the increase in operational efficiency (30%) was computed using (Eq. 18):

18

.

Table 6 provides a comprehensive overview of the quantitative results from the implementation of smart grid technologies in Basra’s water management infrastructure, highlighting notable improvements across multiple performance indicators compared to conventional methods. The data reveals that smart grid integration leads to a 15% reduction in energy wastage, 25% reduction in water loss, and 30% increase in operational efficiency compared to conventional methods. Moreover, smart grid integration results in significant operational cost savings, with an estimated $50,000 saved annually. Additionally, there is a 20% improvement in system reliability, underscoring the enhanced resilience and efficiency achieved through smart grid technologies. These findings collectively demonstrate the transformative impact of smart grid integration on energy optimization, resource efficiency, and system reliability in Basra’s water management infrastructure, providing valuable insights for stakeholders and policymakers involved in sustainable water management initiatives. The reduction in water loss and energy wastage achieved through smart grid integration is particularly significant in addressing the challenges posed by Basra’s aging infrastructure and fluctuating energy supply. Similar improvements have been observed in water-stressed urban areas where real-time monitoring systems and adaptive control technologies were deployed, as highlighted by Singh et al.⁸¹. The implementation of smart grid technologies demonstrated operational cost savings of $50,000 annually for Basra’s water management facilities. However, scaling these systems to larger infrastructures requires an evaluation of additional costs associated with sensor installations, data storage, and real-time monitoring systems. For instance, the upfront cost of deploying advanced smart grid infrastructure across larger regions may exceed current operational savings, necessitating careful cost-benefit analysis. Furthermore, the integration of smart grids with renewable energy systems, such as solar power, would require additional investment in compatible inverters and energy storage systems to maintain reliability during periods of intermittent power supply.

Energy recovery efficiency and cost savings

The comparative analysis of energy recovery efficiency and cost savings between applied hybrid and traditional methods in Basra’s water management infrastructure reveals compelling advantages. Figure 4 presents the quantitative results obtained from this comparative analysis, displaying energy recovery efficiency, energy savings, and capital investment costs for both pressure recovery turbines and traditional methods.

Fig. 4.

Comparative analysis of energy recovery efficiency and cost savings.

The comparative analysis shows differences between proposed and traditional methods. New methodology achieved annual energy savings of 100,000 kWh, compared to 70,000 kWh for traditional methods. The observed energy savings with PRTs is confirmed with prior experimental findings by Lehman and Worrell⁸²which emphasize the efficiency gains achievable through energy recovery systems in high-pressure pipeline environments. Moreover, the study supports the economic feasibility of transitioning to pressure recovery systems, as demonstrated by their ability to reduce capital costs and enhance lifecycle cost-efficiency compared to traditional methods. Capital investment costs were derived from vendor quotes, with turbines costing $50,000 versus $60,000 for traditional methods. The results of the comparative analysis demonstrate that the proposed hybrid methods outperformed the traditional ones in Basra’s water management infrastructure, with significantly higher energy recovery efficiency (85% vs. 60%) leading to substantial energy savings (100,000 kWh/year vs. 70,000 kWh/year) and lower capital investment costs ($50,000 vs. $60,000).

Sensitivity analysis

The sensitivity analysis conducted in this study aimed to assess the robustness of the mathematical model used to evaluate the performance of renewable energy technologies in Basra’s water management infrastructure. Figure 5 presents the results of the sensitivity analysis, showing the variation in key performance indicators under different parameter scenarios.

Fig. 5.

Sensitivity analysis via applied different scenario types.

The sensitivity analysis evaluates the performance of the proposed system under the predefined three scenarios to assess the influence of critical variables on system efficiency and economic feasibility. Scenario 1 reflects the current operational setup, maintaining existing infrastructure efficiency levels. Scenario 2 considers moderate technological improvements, leading to enhanced solar panel efficiency and water flow rates. Scenario 3 represents an optimal scenario, implementing advanced solutions to maximize system efficiency and cost-effectiveness. The comparative analysis of these scenarios provides insight into how improvements in technology and policy adjustments can influence the sustainability of Basra’s water management system. The sensitivity analysis shows how variations in solar panel efficiency (20–30%), water flow rates (100–300 m³/h), and discount rates (5–15%) impacted system performance. Increasing solar panel efficiency from 20 to 30% increased daily energy output from 18 kWh to 27 kWh.

An increase in solar panel efficiency from 20 to 30% resulted in a notable improvement in energy generation and cost savings, with Scenario 3 demonstrating the highest energy output and cost-effectiveness. Similarly, an increase in water flow rate led to higher energy production and greater cost savings, highlighting the importance of optimizing water supply for maximizing the efficiency of renewable energy systems. Moreover, reducing capital investment costs and discount rates resulted in enhanced economic feasibility and shorter payback periods for renewable energy projects, underscoring the importance of favorable financial conditions and incentives for promoting investment in sustainable infrastructure. An optimization analysis was performed by systematically adjusting solar panel efficiency, water flow rates, and discount rates to identify configurations that maximize energy output while minimizing costs. Scenario 3, with 30% solar panel efficiency and a 10% increase in water flow rates, yielded the most favorable results, with a 25% reduction in payback periods and improved net present value. The findings emphasize the diminishing returns on energy output when solar panel efficiency exceeds 30%, suggesting that further investments in optimizing water flow and reducing capital investment costs could yield greater financial benefits. The sensitivity analysis also highlights the importance of policy interventions, such as providing subsidies to lower capital investment costs and improving water infrastructure to support increased flow rates, as effective measures to enhance system scalability and feasibility. The sensitivity analysis further reinforces the potential for scalability of these renewable energy technologies. For instance, increasing solar panel efficiency aligns with advancements in photovoltaic materials reported Ji et al.⁸³, suggesting that ongoing technological innovation could further amplify the cost-effectiveness of such systems. Similarly, optimizing water flow rates offers the dual benefit of maximizing energy recovery and reducing system stress, highlighting the critical role of operational adjustments in achieving sustainability targets.

Economic evaluation and cost-benefit analysis

According to the economic analysis, the PR was determined only for the variable speed operation, whose operational advantages were technically better. The results exhibited a payback period of 2.08 years or 25 months, with a net present value of US$ 64,480.20 and an internal rate of return of 65% (Table 7).

Table 7.

PRT implementation cost.

Item cost of capital	Cost [US$]
Tools	1,930
Pressure meter	120
Frequency inverter	900
Block valve	180
Operation and maintenance cost	1,500
Cost of civil works	850
Total	5,480

The tools represent the specialized equipment needed for the installation and maintenance of the PRT system. These costs are considered essential for setting up the system and ensuring its proper functioning. The high upfront cost indicates the complexity of installing and maintaining such a renewable energy solution. The pressure meter is a necessary component for monitoring and ensuring the proper functioning of the system. It measures the pressure within the water distribution network, ensuring that the PRT system operates optimally. The relatively low cost compared to other components is typical for measurement devices, yet crucial for operational efficiency. The frequency inverter plays a key role in adjusting the speed of the turbine based on varying water flow rates. It converts electrical input into the right frequency to control the turbine, making it an important part of the energy recovery efficiency. The block valve is used for regulating the water flow, allowing for controlled shutdowns or maintenance. This component ensures that the system remains flexible in terms of operation and maintenance, making it a crucial part of managing the flow dynamics and ensuring efficient performance. The OMC represent the annual costs to keep the system running efficiently, including labor, spare parts, and routine checks. This recurring expense must be factored into the long-term economic feasibility of the system. The OMC is typically a percentage of the initial capital cost and is essential for calculating the NPV and Payback Period. The civil works costs are associated with the physical infrastructure needed to integrate the PRT system into the existing water grid. This includes the installation of pipes, supports, and other structures that allow the system to operate smoothly. While a smaller portion of the total cost, civil works are necessary for adapting the site for PRT installation.

In the context of Basra’s water management system, these costs provide a detailed breakdown of the investment needed to implement the PRT technology. The US$ 5,480 total cost is an important benchmark for assessing the initial investment required to set up the system. The OMC are a critical factor in calculating the long-term savings and determining the payback period for Basra. The investment in tools and specialized equipment, such as the frequency inverter, reflects the complexity of maintaining and optimizing the system’s energy recovery capabilities. Figure 6 depicts the behavior of PR during 6 years of operation.

Fig. 6.

Payback period.

This value aligns with Stefanizzi et al.⁶⁸ results and was significantly shorter in duration compared to the utilization of conventional turbines. When the system run within a power range of 1–500 kW, RPs of 2 years or fewer are recorded⁸⁴. The operation of the system at variable speed would be more advantageous for water supply companies⁷¹.

Mapping findings to SDGS

The integration of advanced water treatment technologies, such as UV disinfection and membrane filtration, along with energy recovery systems and smart grid technologies, contributes significantly to various SDGs, including SDG 6 (clean water and sanitation), SDG 7 (affordable and clean energy), SDG 9 (industry, innovation, and infrastructure), and SDG 12 (responsible consumption and production).

The adoption of solar-powered UV disinfection systems and membrane filtration technologies enhances the accessibility and safety of drinking water. These technologies reduce the reliance on energy-intensive conventional methods, making them especially suitable for water-stressed regions, where freshwater access is limited (Table 8).

Table 8.

Impact of UV disinfection and membrane filtration on water quality and efficiency.

Variable	UV disinfection	Membrane filtration	Improvement compared to traditional methods (%)
Water quality	98% microbial reduction	95% microbial reduction	20
Energy savings	15% energy reduction	10% energy reduction	25
Operational cost reduction	12% cost reduction	10% cost reduction	20
Treatment time	20% faster than traditional methods	15% faster than traditional methods	18

Both UV disinfection and membrane filtration systems effectively reduce microbial contaminants in water. The UV disinfection system achieves a higher microbial reduction rate of 98%, indicating its superior performance in improving water quality compared to membrane filtration, which achieves 95% reduction. Both systems contribute to energy savings, with UV disinfection achieving a 15% reduction in energy consumption, and membrane filtration saving 10%. The UV disinfection system’s energy efficiency benefits are attributed to its use of solar power in regions with abundant sunlight, making it more sustainable compared to traditional methods. The integration of these advanced treatment technologies leads to a notable reduction in operational costs. UV disinfection results in a 12% reduction, while membrane filtration shows a 10% decrease in costs. These savings are primarily due to the lower energy demand and reduced need for chemicals, making the systems more cost-effective. Both systems offer faster treatment times compared to conventional methods, with UV disinfection showing a 20% improvement in treatment speed, and membrane filtration achieving a 15% faster process.

The results of this study matched with findings from the other researches in the current field. For example, Bayoumi et al.⁸⁵ and Moharram et al.⁸⁶ studies have confirmed the effectiveness of UV disinfection and membrane filtration systems in reducing both energy consumption and operational costs, while also improving water quality. The results of these studies showed that these systems can be particularly beneficial in arid and water-scarce regions, such as Basra, where efficient water treatment and energy conservation are critical. Furthermore, the combination of renewable energy sources, such as solar power for UV disinfection, with membrane filtration, has proven to be an energy-efficient and cost-effective approach in similar environmental conditions, supporting the findings of this study.

The solar power utilization for UV disinfection systems, the proposed methods reduce the energy burden of water treatment processes, contributing to clean energy (SDG 6). Membrane filtration systems also benefit from energy optimization strategies, leading to a reduction in energy consumption for water purification (SDG 7). The use of predictive modeling and machine learning in optimizing these systems supports industry innovation, enabling smarter and more efficient water management (SDG 9). Energy recovery systems, such as PRTs, improve energy efficiency and contribute to responsible consumption and production. Systems allow for the recovery and reuse of energy within water management infrastructures, reducing waste and promoting circular economy principles (SDG 12).

Performance evaluation of machine learning algorithms

The application of the M5 model tree for predictive modeling and anomaly detection yielded promising results, as detailed in Table 9 below.

Table 9.

Performance metrics of M5 model tree for predictive modeling and anomaly detection.

Metric	Predictive modelling (%)	Anomaly detection (%)
Accuracy	85	90
Precision	88	91
Recall	82	89
F1 Score	85	90

The predictive modeling aspect of the M5 model tree achieved an accuracy of 85%, indicating its effectiveness in accurately predicting numerical variables. The precision, recall, and F1 score further support the model’s performance, with values of 88%, 82%, and 85%, respectively. These metrics indicate the model’s ability to correctly identify true positive instances while minimizing false positives and false negatives. In the context of anomaly detection, the M5 model tree exhibited even higher performance, achieving an accuracy of 90%. The precision, recall, and F1 score for anomaly detection were 91%, 89%, and 90%, respectively. These results highlighted the model’s capability to accurately detect anomalous instances within the dataset, ensuring reliable identification of irregularities and deviations from expected patterns. The superior performance of the M5 model tree in both predictive modeling and anomaly detection underscores its versatility and effectiveness in handling diverse tasks.

The M5 model tree’s predictive modeling outputs have practical implications for Basra’s water management systems. For example, predictions of energy consumption can inform the scheduling of energy-intensive processes during off-peak hours, leading to reduced operational costs and improved grid stability. Meanwhile, water quality parameter forecasts enable preemptive maintenance and adjustments in treatment processes, minimizing risks of contamination. Anomaly detection by the M5 model tree has demonstrated its ability to identify potential issues such as equipment malfunctions or irregular energy spikes. These early warnings allow operators to address problems before they escalate, ensuring uninterrupted operation and optimizing resource use.

The RNNs for predictive modeling and anomaly detection yielded insightful results, as summarized in Table 10 below.

Table 10.

Performance metrics of RNN for predictive modeling and anomaly detection.

Metric	Predictive modelling (%)	Anomaly detection (%)
Accuracy	87	92
Precision	89	93
Recall	84	91
F1 Score	86	92

The RNN achieved an accuracy of 87%, indicating its efficacy in accurately forecasting numerical variables. The precision, recall, and F1 score further validate the model’s performance, with values of 89%, 84%, and 86%, respectively. These metrics demonstrate the model’s ability to make precise predictions while capturing a significant portion of true positive instances. Turning to anomaly detection, the RNN exhibited even higher performance, achieving an accuracy of 92%. The precision, recall, and F1 score for anomaly detection were 93%, 91%, and 92%, respectively. These results underscore the model’s robustness in identifying anomalous instances within the dataset, ensuring reliable detection of irregularities and deviations from expected patterns.

The RNN’s ability to model sequential data provides significant operational benefits. For instance, its forecasts of daily and seasonal energy consumption patterns allow for dynamic resource allocation, ensuring energy supply meets demand while minimizing wastage. By capturing temporal dependencies, RNN models can also predict potential failures in equipment based on subtle trends in sensor data, enabling predictive maintenance strategies. In the context of water quality, RNN outputs can identify trends that signal potential contamination events, allowing timely interventions that protect public health. These practical applications underscore the importance of RNN-based predictive analytics in enhancing the reliability and efficiency of Basra’s water management systems, particularly in adapting to fluctuating energy supplies and water demands.

The comparison between the M5 model tree and RNN for predictive modeling and anomaly detection yielded insightful findings (Table 11).

Table 11.

Comparison of M5 and RNN performances.

Efficiency criteria	R 2		RMSE
Model	Train	Verify	Train	Verify
M5	0.83	0.75	0.03	0.05
RNN	0.90	0.85	0.01	0.03

The RNN achieved an accuracy of 87%, compared to 85% for the M5 model tree. Similarly, the precision, recall, and F1 score for the RNN were marginally higher than those of the M5 model tree, indicating its ability to make more precise predictions while capturing a greater proportion of true positive instances (Fig. 7).

Fig. 7.

Comparison of RNN and M5 model tree: (a) Time series, (b,c) Scatter Plots.

In terms of predictive modeling, both the M5 model tree and RNN demonstrated strong performance, with the RNN slightly outperforming the M5 model tree across all metrics. Specifically, the RNN achieved an accuracy of 87%, compared to 85% for the M5 model tree. Similarly, the precision, recall, and F1 score for the RNN were marginally higher than those of the M5 model tree, indicating its ability to make more precise predictions while capturing a greater proportion of true positive instances.

When it comes to anomaly detection, both models exhibited excellent performance, with the RNN again showing a slight advantage over the M5 model tree. The RNN achieved an accuracy of 92%, compared to 90% for the M5 model tree. Additionally, the precision, recall, and F1 score for the RNN were higher than those of the M5 model tree, highlighting its superior ability to detect anomalies within the dataset. These results suggest that while both the M5 model tree and RNN are effective for predictive modeling and anomaly detection tasks, the RNN tends to offer slightly better performance across various metrics. The RNN’s ability to capture complex temporal dependencies and patterns may contribute to its enhanced performance compared to the M5 model tree. However, it is essential to consider practical factors such as model interpretability, computational efficiency, and ease of implementation when choosing between the two approaches. While the RNN may offer better predictive performance, the M5 model tree’s simplicity and transparency may be preferable in certain contexts, especially when interpretability is a priority.

The choice between a linear approach, such as the M5 model tree, and a non-linear approach, such as RNNs, depends on the underlying nature of the data and the complexity of the relationships being modeled. Linear approaches like the M5 model tree are well-suited for datasets where the relationships between variables can be adequately captured by straight lines or planes. They are effective when the underlying phenomena exhibit linear behavior or when the relationships are relatively simple and can be approximated by linear functions. However, in cases where the relationships are non-linear or the data exhibits complex interactions, linear models may not capture the nuances of the data effectively. On the other hand, non-linear approaches like RNNs are capable of capturing more complex patterns and relationships in the data. RNNs excel in modeling sequences and time-series data where there are dependencies between previous observations and future outcomes. They are particularly effective in scenarios where the relationships between variables are non-linear and may involve intricate interactions or temporal dependencies. In the context of this paper, which investigates the integration of renewable energy technologies into water management systems, the phenomena being studied likely exhibit non-linear behavior. The interactions between renewable energy technologies, water management systems, and environmental factors are complex and dynamic, making them better suited for modeling using non-linear approaches like RNNs. Therefore, the superior performance of RNNs in this paper may be attributed to their ability to capture the non-linear nature of the phenomena under investigation and effectively model the complex relationships within the data.

Challenges, practical applications, and future research directions

This subsection aims to provide a comprehensive discussion of the challenges addressed by this study, its practical applications, and directions for future research. By reflecting on the limitations of the current approach and proposing actionable solutions for addressing critical gaps, the subsection underscores the significance of the study’s contributions to renewable energy integration in water management systems. It also highlights the broader implications of the findings for policymakers, engineers, and researchers, while outlining potential advancements and areas for further exploration.

Challenges overcome in the current study

This study addressed critical challenges in integrating renewable energy technologies into water management systems in resource-constrained regions such as Basra, Iraq. The adoption of solar-powered UV disinfection, membrane filtration, pressure recovery turbines, and smart grid technologies demonstrated significant improvements in energy efficiency, water quality, and operational resilience. Despite the complexities of implementing these technologies in a region with aging infrastructure and fluctuating energy supplies, the study successfully showcased their feasibility and scalability. By leveraging predictive modeling and anomaly detection algorithms, the research provides a replicable framework for improving water and energy systems in water-stressed regions.

Limitations of the current study

A notable limitation of this study is the use of the OFAT approach for parameter optimization. While OFAT provided valuable insights into the individual effects of key parameters, it does not account for interaction effects between variables, which are critical for achieving optimal system performance. For example, in this study, the interaction between solar panel efficiency and water flow rates may significantly impact energy recovery and overall cost-effectiveness but was not fully explored due to the constraints of the OFAT approach.

DOE, specifically response surface methodology (RSM) with central composite design (CCD) or Box-Behnken design (BBD), has been widely recognized as a more efficient and reliable tool for parameter optimization. DOE allows for the evaluation of interaction effects and provides a more holistic understanding of system behavior under varying conditions^87,88. Studies such as “Photocatalytic degradation of nitrotoluene in synthetic wastewater using Box-Behnken design”⁸⁹ and “Investigation of spent caustic effluent treatment by electro-peroxone process”⁸⁸ demonstrate the advantages of DOE for optimizing operational parameters in water and energy systems. Future research will prioritize the application of DOE methodologies to improve the robustness and reliability of optimization results, particularly for evaluating complex interactions between key parameters such as solar irradiance, membrane efficiency, and turbine performance⁸⁷.

This study primarily focuses on demonstrating the feasibility of renewable energy technologies and predictive modeling frameworks for small- to medium-scale water management systems. However, scaling these technologies to large-scale operations introduces additional challenges, particularly in economic feasibility. Comprehensive economic analyses that consider all costing factors—capital investments, operational and maintenance costs, lifecycle costs, and material costs—are essential to assess the long-term viability of large-scale implementations. For instance, the payback period of renewable energy technologies may extend significantly for larger systems due to the higher initial investments required for equipment and infrastructure. Furthermore, the variability in hydraulic flow rates and energy demands across large-scale systems necessitates a more detailed evaluation of economic trade-offs. Future research should prioritize the development of scalable financial models that integrate these factors to provide a holistic understanding of the economic implications of wastewater treatment systems at different operational scales.

Practical applications of the study

The findings of this study have significant practical implications for policymakers, engineers, and stakeholders involved in water and energy management. The integration of renewable energy technologies, such as pressure recovery turbines and smart grids, provides actionable solutions for reducing energy consumption, greenhouse gas emissions, and operational costs. Additionally, the predictive modeling and anomaly detection frameworks developed in this study enable water treatment facilities to proactively address issues, enhancing reliability and efficiency. For instance, energy consumption predictions can inform decision-making on scheduling energy-intensive processes during off-peak hours, while anomaly detection can prevent equipment failures by identifying irregular trends in operational data. These results can guide infrastructure development and policy initiatives in other water-stressed regions facing similar challenges.

Future research directions

Building on the findings of this study, future research will focus on incorporating DOE methodologies, such as RSM with BBD, to evaluate the interaction effects of parameters and achieve more comprehensive optimization. For example, evaluating the combined effects of solar panel efficiency, water flow rates, and capital investment costs can provide a deeper understanding of the trade-offs involved in optimizing renewable energy systems. This approach will enable the development of more reliable and cost-effective solutions for water and energy management.

Additionally, exploring alternative renewable energy technologies, such as UV/ZnO processes for advanced oxidation or electro-peroxone treatments⁹⁰could further enhance the effectiveness of water treatment systems. Socio-economic studies are also needed to address barriers to the adoption and scaling of these technologies, including cost reduction strategies, regulatory support, and community engagement. Finally, expanding this research to other regions facing water scarcity and energy constraints would validate the broader applicability of the proposed solutions.

Conclusion

This study conducted a comprehensive assessment of integrating renewable energy technologies and intelligent control systems into Basra’s municipal water infrastructure, focusing on performance, sustainability, and feasibility in arid regions. The primary findings are summarized as follows:

1. Achieved 30% improvement in operational efficiency through combined use of advanced treatment and recovery systems.

2. Enabled annual energy savings of 100,000 kWh, reducing dependency on conventional sources.

3. Attained carbon emission reductions of 16,500 kg CO₂/year, contributing to environmental sustainability.

In the following, the secondary outcomes can be listed as:

• 4.Machine learning models (M5 and RNN) provided high-accuracy predictions, enabling real-time fault detection, system optimization, and demand forecasting.
• 5.Scenario-based simulation demonstrated the system’s flexibility and scalability under varying operational conditions.
• 6.Economic indicators such as NPV and payback period supported long-term financial feasibility.

The results were aligned strongly with SDGs 6, 7, and 13, promoting access to clean water, affordable clean energy, and climate resilience. While the current study relied on a OFAT approach for sensitivity analysis, future work will incorporate DOE to improve optimization accuracy. Moreover, enhanced financial modeling with dynamic pricing scenarios and broader economic indicators is recommended for large-scale implementations. The framework is transferable to other arid and resource-constrained regions, providing a replicable model for integrating renewable energy with smart water infrastructure. Its policy relevance lies in supporting municipalities to make data-driven, sustainable investments that address both environmental and operational challenges.

Author contributions

Azfarizal Mukhtar: Methodology, validation, data curation, formal analysis, investigation.Jawdat N. Gaaib: Software, data curation, formal analysis, investigation, resources.Ahmed Sabeeh Abed Abood: Conceptualization, methodology, validation, data curation, writing—original draft preparation.Hyder Hassan Abd Balla: Methodology, software, formal analysis, resources, writing—review and editing. Farruh Atamurotov: Validation, data curation, writing—original draft preparation, resources.Natei Ermias Benti: Conceptualization, Methodology, data curation, formal analysis, investigation, resources.

Data availability

The datasets used during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.