Optimal planning of grid-connected energy storage and renewable energy sources integration using an improved whale algorithm Authors: Pranda Prasanta Gupta, Deepak Gupta, Kavita Tukaram Patil, Kishor S Wagh, Chanakya Kumar Jha, Pallavi Jha, Akhilesh Kumar Sharma Published: 2026 ABSTRACT AbstractThe integration of renewable energy sources (RESs) into energy systems poses considerable operational challenges, due to their intermittent and stochastic nature. Grid-connected energy storage systems (ESSs) present a compelling alternative for reliably accommodating various RESs. This paper presents an optimal scheduling for allocating wind-storage system capacity in high mountain regions with abundant wind resources and irrigation pumping demand. To capture uncertainty in wind and solar generation, a scenario-free stochastic formulation is adopted, enabling tractable and scalable uncertainty modelling. To reduce the overall lifecycle comprehensive cost (COC), a synergistic optimization model is developed by integrating features of wind energy, pumping unit parameters, and storage configurations. The model considers startup/shutdown losses, optimal flow distribution, and operating expenses of the pumping station while optimizing energy storage capacity. A single-battery system is compared with a hybrid battery-hydrogen storage strategy. Results show the hybrid solution reduces COC by 5.4%, alleviates battery operational stress, and maintains pumping station efficiency. The framework incorporates a reliability model for wind power generation, ensuring robust lifecycle cost optimization. This approach demonstrates the financial and technological benefits of hybrid storage in renewable energy-driven irrigation systems. Simulations in the case study validate the flexibility and effectiveness of the proposed approach across various practical settings.Keywords: Pumping station, Optimization, Uncertainty, Renewable energy sources, Hybrid energy storage systemSubject terms: Energy science and technology, Engineering Introduction Background and motivation Growing environmental concerns have accelerated the adoption of renewable energy sources (RESs) owing to their sustainability and widespread availability; however, their inherent intermittency and uncertainty introduce considerable challenges to the reliability, stability, and efficient operation of modern energy systems1,2. In addition to demand-side uncertainties, RESs such as wind and photovoltaic (PV) systems contribute further variability due to their strong dependence on weather conditions, thereby increasing system complexity and making optimization problems more challenging to solve3. By 2050, increasing energy efficiency and developing renewable energy technologies will be essential components of the energy transition, according to a report by the International Renewable Energy Agency. Technologies for renewable energy provide substantial socioeconomic advantages, scalability, broad resource availability, and economic feasibility. Renewable energy is expected to contribute significantly to reducing greenhouse gas emissions and to meet two-thirds of the world’s total energy consumption by 20504. Thus, it is crucial to conduct relevant research on power systems capable of efficiently integrating large amounts of renewable energy. Energy security is further complicated by the depletion of fossil fuels and price volatility. In contrast, renewable energy sources are plentiful and widely available, helping reduce dependence on external energy sources and improve energy supply stability5. Energy storage system (ESSs) offer a promising solution for the reliable integration of RESs into energy systems6. Although the precise definition of ESSs is still under discussion in technical forums, they are generally described as groups of interconnected loads and distributed energy resources within clearly defined electrical boundaries that act as single, controllable entities relative to the main grid. Thus, ESSs can import energy from the main grid when local generation is insufficient to meet demand requirements7. Consequently, explicitly incorporating pumped station aging into ESS operation has become critical for sustainable and economically efficient power system operation. Literature review Many researchers are currently working on economic dispatching with large-scale integration of renewable electricity8,9. More challenges could arise from the unpredictability and variability of renewable electricity output. As a result, many studies have been conducted on forecasting wind and solar PV power10. Accurate forecasting remains difficult to achieve, and wind power forecasting errors are expected to persist for some time. To address uncertainty in renewable power integration, a multi-timescale optimal scheduling model incorporating demand response and renewable power uncertainty was proposed in11. To account for wind power variation and uncertainty, an adjustable interval optimization model for energy and reserve clearance was introduced in12. Subsets with variable intervals were suggested as more prudent options for cost-effectively managing renewable uncertainties, replacing traditional projected intervals. By using reserve auxiliary variables to connect the reserve under the projected scenario and uncertain scenarios, a wind power reserve model incorporating uncertainty was developed in13. This robust scheduling model quantified the contributions of wind farms to power system frequency support and reserves. Considering the uncertainty in the probability distribution of wind power forecast errors, a distributionally robust coordinated reserve scheduling model was proposed in14 to minimise the total procurement cost of conventional generation and reserves while satisfying security requirements under all possible probability distributions of wind power forecast errors. Additionally, a distributionally robust optimization model was presented in15 for reserve scheduling decision-making with partial information on wind power, seeking a robust solution to wind power probability distribution uncertainty. In the meantime, high-altitude areas have produced superior, nutrient-rich, and environmentally adaptive speciality crops due to their distinct geographic setting and climate. Nevertheless, these areas face significant problems with their water supply, which largely depends on erratic sources such as seasonal snowfall, rainfall, and glacial meltwater. This causes severe water shortages during dry seasons and directly affects agricultural output. Furthermore, high-altitude areas face major power supply issues due to their distinct geography and climate. In this regard, building high-altitude irrigation pump-station systems has emerged as a viable means to alleviate the unequal distribution of water resources in space and time, thereby ensuring consistent and substantial crop yields while harnessing the full ecological and economic potential of high-altitude farming. A suitable model that accurately describes the optimisation problem, along with an efficient computational technique, is required to enhance the operational efficiency and energy-saving potential of pump stations and to address economic operating challenges. Numerous studies have employed optimisation techniques, including particle swarm optimization (PSO) algorithms16,17, to develop models that minimize the energy consumption and operational costs of pump station systems18,19. Storage technologies such as hydropower and compressed air (long-term storage), batteries (medium-term storage), flow batteries, flywheels, SMES, and supercapacitors (short-term storage) are frequently used to provide a steady, continuous electrical supply to end consumers. RES are limited to producing energy under specific operating conditions, such as sunny or windy periods, in contrast to conventional fossil-fuel-based generation, which can be dispatched on demand and stored with relative ease20. ESSs improve the system’s overall reliability by allowing excess energy to be stored during periods of high generation or low demand and to be provided when needed21. ESSs also contribute to power quality improvement, frequency control, voltage quality improvement, power fluctuation reduction, and ancillary service providing22. The security and economy of power system operation appear to be compromised by economic dispatch with large-scale wind power integration, which relies solely on ESSs to handle wind power fluctuations and unpredictability. Shortfalls in wind power forecasts can be compensated for by an ESS, which can dynamically absorb and release energy quickly23. For a wind-pumped storage-thermal integrated energy system, an economical and ecological scheduling approach was proposed in24, and the units were logically ordered according to the energy-saving and emission-reduction principles. In order to improve pumped storage units’ regulation performance, the scheduling method optimizes operational decision-making. In25, it was shown that a hybrid ESS with a battery and a supercapacitor could efficiently dispatch utility-scale wind power at one-hour increments for a full day. In26, the coordinated optimal dispatching model for a hydropower group system comprising traditional hydropower and pumped-storage power plants was examined. Wind-photovoltaic-pumped storage mitigates the impact of renewable energy’s unpredictability and intermittency on the power system’s stability. However, to prevent unfavourable effects on the power grid, such as overvoltage, increased power losses, reduced power quality, and reduced network dependability, integrating renewable sources into the current distribution grid as distributed generation requires careful planning and strategy. Hence, determining the optimal penetration levels and deployment strategies for distributed renewable energy sources within the distribution network is of paramount importance. Accordingly, the development and application of efficient optimization algorithms are essential for the optimal planning of distributed generation in power systems, as they enable mathematical modeling of network components and facilitate the identification of cost-effective and technically feasible solutions that satisfy multiple constraints in complex optimization scenarios. Researchers have explored a wide range of optimization techniques to address the optimal integration of distributed energy sources, combining conventional methods such as Genetic Algorithms (GA)27, Artificial Bee Colony (ABC)28, Biogeography-based Optimization (BBO)29, Whale optimization algorithm30, and Cuckoo Search Algorithms (CSA)31 with more advanced metaheuristic approaches like Harris Hawks Optimization (HHO)32, and Elephant Herding Optimization (EHO)33, where the former provide well-established solution frameworks while the latter offer improved exploration-exploitation balance, faster convergence, and enhanced capability to solve complex, nonlinear, and multi-objective power system problems. The Improved Mayfly Optimization Algorithm (IMO) was created in34 to address the unified economic emission dispatch problem in a microgrid aspect. Additionally, in reference35, Chaotic Fast Convergence Evolutionary Programming was offered as a solution to the combined power and heat economic scheduling problem. This system included pumped hydro storage, renewable energy, and demand-side control. In36,37, the Social Group Entropy Optimisation (SGEO) approach was proposed to address the fuel-constrained economic dispatch problem. This idea incorporates renewable energy sources and a pumped hydro storage facility. Reference38 employs several nature-inspired swarm intelligence (metaheuristic) algorithms for optimal scheduling, and the results are reported based on multiple runs to ensure solution reliability and accuracy. Although traditional methods have simple algorithmic structures and are widely used across many fields, their biggest drawback is that they tend to converge to local optima, leading to poor performance and stability. Meanwhile, more modern algorithms outperform traditional methods. However, they still have disadvantages that need to be addressed, such as slow data processing, heavy dependence on initial setup parameters, low stability, and no guarantee of convergence. The higher the performance and stability of the algorithm, the better the quality of the solution to the optimization problem. Therefore, developing a powerful and efficient algorithm for use in power network problems is very important. Wind-solar-storage hybrid systems play a pivotal role in next-generation power systems by leveraging the complementary nature of renewable resources and the buffering capability of energy storage to achieve a balance between cost, reliability, and sustainability; this integrated approach is particularly effective in remote or weak-grid regions, where optimized resource allocation, cost management, and scenario-based planning enable dependable power supply while maintaining economic efficiency under varying operating conditions39,40. However, these studies are limited by the absence of comprehensive lifecycle assessments, as they overlook the impact of extreme weather on system durability, depend heavily on static economic assumptions, and fail to incorporate dynamic cost factors such as battery degradation, equipment aging, and maintenance expenses, thereby reducing the accuracy and practicality of long-term economic and reliability evaluations. Several studies41,42 have concentrated on increasing system efficiency through intelligent control and dynamic optimization of wind-solar-storage systems using various algorithms. By using experimental simulations from various dimensions, research43 has improved the practical application of wind-solar-storage systems. They considered the cost-benefit index and the probability of power supply loss as key operational constraints. However, although approaches like the IWA have been utilized to determine optimal ESS sizing and evaluate system performance under different RES scenarios, these investigations predominantly focus on short-term or real-time conditions, thereby lacking a comprehensive assessment of long-term operational stability and the sustained effectiveness of control strategies. Research gap and paper contribution Many existing studies do not adequately account for uncertainty in key parameters, such as fluctuations in renewable generation and variations in electrical load demand. Overall, most existing studies focus on the optimal scheduling and energy dispatch of RES and ESSs, with an emphasis on short-term operational stability. These analyses were limited to a long-term planning schedule, without careful spatio-temporal considerations, single-objective operations, or intelligent, algorithm-driven fine-tuning. In addition, limited efforts have been made to assess grid-connected ESSs and the impacts of the decrease in economic benefits of wind-solar PV power and the rise in renewable energy costs if a substantial portion of pumped storage is under large-scale system control to address charging-induced grid operational security. Based on the literature review and the limitations in existing research, two key research gaps are identified: Most current research techniques rely on thermal unit reserves and ESSs for regulation to address wind power forecast errors. Thermal unit reserves, however, only mitigate variations caused by inaccurate wind power predictions. Wind curtailment may occur if the system does not establish adequate energy storage capacity when actual wind power exceeds forecasts, potentially hindering the large-scale development of wind-solar PV power. In addition, the economic benefits of wind-solar PV power decrease, and the costs of renewable energy rise if a substantial portion of pumped storage is allocated to renewable farms. Most current research techniques rely on thermal unit reserves and ESSs for regulation to address wind power forecast errors. Thermal unit reserves, however, only mitigate variations caused by inaccurate wind power predictions. Wind curtailment may occur if the system does not establish adequate energy storage capacity when actual wind power exceeds forecasts, potentially hindering the large-scale development of wind-solar PV power. In addition, the economic benefits of wind-solar PV power decrease, and the costs of renewable energy rise if a substantial portion of pumped storage is allocated to renewable farms. Research on wind-solar-storage systems has advanced significantly in areas like multi-energy coordination, intelligent algorithms, and energy storage technology, but it still faces technical bottlenecks, economic constraints, and limited engineering validation. Currently, no researchers have integrated the findings of wind-solar-storage system energy configuration optimization with pump station economic operation considerations. Research on wind-solar-storage systems has advanced significantly in areas like multi-energy coordination, intelligent algorithms, and energy storage technology, but it still faces technical bottlenecks, economic constraints, and limited engineering validation. Currently, no researchers have integrated the findings of wind-solar-storage system energy configuration optimization with pump station economic operation considerations. To fill this previous research gap, this paper integrates RES and energy storage into grid security operations. The proposed framework coordinates optimal planning and scheduling across hybrid ESSs while ensuring cost-effective pump station operation under grid security constraints. An economic scheduling model was presented to evaluate the impact of renewable costs on balancing factors, including pump station startup/shutdown losses, operational depreciation losses, and the variability of wind and solar power generation, with the aim of increasing the economic benefits of renewable power. Thus, by creatively proposing a wind-solar-storage pump-station system, this work addresses the energy-supply challenges faced by irrigation pump stations in mountainous areas. The schematic diagram of the integrated wind-solar-storage pump station system solution is shown in Fig. 1. By constructing a wind-solar-storage system capacity configuration optimization model based on the pump station economic operation model. The distinctive features of the proposed strategy, in comparison with conventional grid-connected ESS and RES, are summarized in Table 1. According to the above analysis, the main contributions of this paper are summarized as follows: Schematic diagram of wind and solar with a pumping station system. Table 1The comparison of the proposed model and related studies.ReferencesAdopted techniquesESSUncertaintyconsiderationGlobaloptimalityModelingsimplicityComputational scalability3,6SensitivityanalysisowindΔoo14,15Mixed integer programmingoElectrical load and windxoΔ16,17PSOxWind, PVΔoo24,25Mixed integer linear programmingoSolar, windΔΔo30,34,36WOA, IMO, SEGOoWind, PVΔΔo39Optimum power flowoWindxΔΔ41,42PSO, GA,BBOoElectric load and solarxΔΔProposedIWAoooooo : strong output, Δ : partial output, x : not supported. The comparison of the proposed model and related studies. Sensitivity analysis PSO, GA, BBO o : strong output, Δ : partial output, x : not supported. A scheduling model is presented for operational strategies for pump stations to manage a grid-connected ESS, coordinating PV generation and wind while considering uncertainties associated with renewable energy generation. A scheduling model is presented for operational strategies for pump stations to manage a grid-connected ESS, coordinating PV generation and wind while considering uncertainties associated with renewable energy generation. Effectively utilizing the abundant wind and solar resources in high-altitude regions to eliminate electricity purchase costs while mitigating fluctuations in wind and solar PV power generation through the pump station economic operation model, thereby reducing pressure on the energy storage system. Effectively utilizing the abundant wind and solar resources in high-altitude regions to eliminate electricity purchase costs while mitigating fluctuations in wind and solar PV power generation through the pump station economic operation model, thereby reducing pressure on the energy storage system. The optimal scheduling model is solved using the enhanced IWA. By leveraging straightforward implementation, fast convergence, and high search accuracy, this work enhances the IWA by balancing local and global searches and improving the algorithm’s search speed and convergence accuracy. The optimal scheduling model is solved using the enhanced IWA. By leveraging straightforward implementation, fast convergence, and high search accuracy, this work enhances the IWA by balancing local and global searches and improving the algorithm’s search speed and convergence accuracy. Outline The remainder of this paper is organized as follows: Sect. 2 introduces the power economic operation of pumped station model. Section 3 introduces the renewable uncertainty model. Section 4 provides the solution methodology of our model. In Sect. 5, cases are simulated, and the numerical results are discussed. Section 6 concludes this study. Economic operation models of different types of pump station units Realizing the multi-objective synergistic optimization of unit operation efficiency, energy consumption economy, and scheduling reliability while adhering to engineering constraints through system analysis and optimization techniques is the foundation of a pumping station’s economical operation. To determine the optimal operating parameters under various working conditions, it is essential to develop a quantitative model that relates energy consumption characteristics to the pump station’s operating state. This model is based on optimal choice theory. The economic operation model of the pumping station is formulated to account for start-stop losses and operational depreciation costs, and a corresponding solution strategy is proposed. This model is grounded in the pumping station’s economic operation principles and accounts for three different types of units operating in parallel. Pump station economic operation of the mathematical model Objective function The scheduling problem for the multi-model pumping station unit considers the depreciation cost of the unit during operation, the additional cost of starting and stopping the unit, and the minimum total daily operation cost as the optimization objective44. Three main factors are involved in the constraints: the amount of water that must be consumed within the allotted time, the maximum number of times the unit may start and stop, and the limitations on power operation during various time periods. To create the entire objective function, which looks like this: Constraints The sum of the water lifting capacity of each unit should be equal to the total water lifting capacity of the pumping station: The sum of the water lifting capacity of each unit should be equal to the total water lifting capacity of the pumping station: The pump unit states include the shutdown and running states. In the running state, to ensure the unit operates in the high-efficiency work area, it must meet the specific head conditions of the flow constraints. The pump unit states include the shutdown and running states. In the running state, to ensure the unit operates in the high-efficiency work area, it must meet the specific head conditions of the flow constraints. To avoid excessive start-stop cycles, it is necessary to constrain the number of start-stop cycles for the unit: To avoid excessive start-stop cycles, it is necessary to constrain the number of start-stop cycles for the unit: During operation, the power required in each time slot should be less than or equal to the maximum power that the wind energy storage system can provide in that time slot: During operation, the power required in each time slot should be less than or equal to the maximum power that the wind energy storage system can provide in that time slot: The health degree of each unit during operation should meet the requirements: The health degree of each unit during operation should meet the requirements: Curve fitting of pump data Three different pump models are prepared; each model has three pumps, and there is no difference between units of the same model. Thus, three fitting results are obtained, and the results for each pump are shown in Fig. 2. Data fitting diagram of three different types of pumps. Calculation of start/stop loss c Many researchers have examined the negative consequences of the start-stop operation of major pumping stations. Pumping station starting and shutdown consumption is influenced by a variety of factors, such as mechanical equipment wear and tear and control equipment failure during startup. The maintenance and staffing expenses related to wear and tear brought on by equipment failure account for the majority of the costs. The mathematical model of pump start/stop loss is shown in the following equation : Calculation of operating depreciation cost of pumping station units The depreciation cost for a pump unit running is usually calculated using the annual method. The formula is as follows: Capacity allocation of the wind solar PV energy storage system based on the economic operation of the pumping station Based on the constructed economic operation model of the pumping station, the optimal allocation model of the capacity of the wind and light storage system is established, and the two models are optimized to solve the problem. Output calculation of wind and solar PV energy storage system The output model of wind power generation The dynamic relationship between the output power of the wind power generation system and wind speed can be precisely expressed by a segment function, and its mathematical expression is as follows: Output model of photovoltaic power generation The intensity of local solar radiation and the ambient temperature have the greatest effect on the output power of the solar PV power generation system. The following mathematical statement describes the relationship between this output power and the PV array’s performance: Capacity configuration model In the capacity configuration study of wind and solar PV storage systems, although an appropriate increase in hybrid energy storage capacity can improve power supply reliability, it will lead to a significant increase in system investment costs, creating a trade-off between economy and reliability. Therefore, how to achieve the economic optimization of energy storage capacity configuration under the premise of ensuring system performance has become a key issue in current research. This study systematically explores the cost-effectiveness of two energy storage solutions: a hybrid energy storage device combining batteries and hydrogen storage, and batteries alone. Problem formulation It seeks to increase the wind solar energy storage system’s economy and, after verifying that the power supply reliability index satisfies the necessary standards, develops an optimization model to reduce the combined cost over the life cycle (COC) and increase the system’s overall cost-effectiveness. The operating life of wind and solar storage systems in use is influenced by several factors and typically ranges from 15 to 20 years. Wind turbines typically have a 15-year lifespan, while solar panel components can last 20 years. The lifespan of hydrogen storage systems is 15–20 years, and that of converters is 10–15 years. As a result, these four equipment types only require replacement once over the system’s lifetime, which lowers the cost of system replacement. On the other hand, the battery has a much shorter lifespan-only three to five years-implying it must be replaced up to four times over the power generation system’s 20-year lifespan, resulting in comparatively high replacement costs. The type of equipment, technical features, market conditions, and industry norms must all be considered when determining the operating and maintenance expenses of a renewable energy system. Based on current market data and research, the operating and maintenance costs are set at 2.5% annually for wind power generation facilities, 2% annually for photovoltaic arrays, 1.5% annually for battery systems, and 1.8% annually for energy storage converters. Equality and inequality constraints In the capacity allocation study of the wind energy storage system, in addition to considering the comprehensive cost of the whole life cycle, the following key operational constraints need to be met: Load power shortage rate LPSP. Load power shortage rate LPSP. This study adopts Loss of Power Supply Probability (LPSP) as the key index for assessing the power supply reliability of the wind energy storage system, which is mathematically defined as the ratio of the system power shortage to the total load demand. In engineering practice, to ensure that the system power supply reliability meets the design requirements, the LPSP must be strictly controlled within a predetermined threshold (usually ≤ 5%): Wind turbine output power constraints. Wind turbine output power constraints. Power constraints of PV power generation output. Power constraints of PV power generation output. Power balance constraints. Power balance constraints. In the equation, and represent the number of wind power, thermal energy and solar PV units, as well as pumped storage units. Power output from the ith thermal unit, wind turbine, solar PV array, and pumped storage unit is denoted as and according to their respective sources. Power constraints of thermal units. Power constraints of thermal units. Each generator’s active output must adhere to certain minimum and maximum limits. Constraints on power generation are the rules and regulations that limit how power-generating units can function. The safe and dependable functioning of the electricity system depends on these constraints Thermal generator ramp rate constraints. Thermal generator ramp rate constraints. Critical operating and control parameters for power systems include thermal generator ramp rate constraints. The ramp rate is the maximum pace at which a generator’s power output may fluctuate over a certain period. Solution methodology The capacity planning for the wind energy storage system should account for the total load demand of the pumping station. At the same time, the economic operation model developed in this study incorporates the real-time variability in the output power of the wind energy storage system. The overall modelling approach is illustrated in the flowchart in Fig. 3. Constraints (14–21) are applied to evaluate the pumping station’s performance under both equality and inequality conditions, using the initial grid data as defined in the proposed framework. Finally, thermal constraints are managed through a grid-connected renewable energy system that ensures optimal power dispatch to the irrigation pumping station over a 24-hour period. The optimal planning of RES and ESSs in a power system involves a systematic approach. The objective is to reduce the overall operating cost while satisfying all applicable constraints. The figure further presents the proposed optimization framework, highlighting how wind energy, solar PV generation, and network constraints are coordinated to ensure efficient and economical operation of the energy system. Ultimately, the integrated model simultaneously optimizes both RES and ESSs, and through an iterative computational process, it converges to a globally optimal solution. Flow chart of economic optimization calculation of wind power and storage system. Step 1: Collection of data. Step 1: Collection of data. This step involves collecting both current and historical data related to factors influencing load demand, RES, and ESSs. Key datasets also include wind speed data (as shown in Fig. 3) and solar irradiance levels, which directly affect the performance of photovoltaic systems. Furthermore, load data represents electricity demand over various time scales, including hourly, daily, and seasonal variations. Step 2: Wind-solar output power forecasting. Step 2: Wind-solar output power forecasting. An iterative approach is applied over defined time intervals to generate and refine scenarios until the required number is reached. During this process, the time counter is initialized. Subsequently, the forecasted data for wind and solar power generation is analyzed over a 24-hour time horizon. Step 3: Optimization algorithms and modelling framework. Step 3: Optimization algorithms and modelling framework. The significant aspect involves conducting research and developing optimization models and algorithms. These models are designed to determine the optimal scheduling of renewable energy generation and consumption based on forecasted data. Techniques such as pumped station optimization model, and IWA are employed to formulate and solve the scheduling problems effectively. Step 4: Development of constraints. Step 4: Development of constraints. The implementation of constraints is a crucial step in scheduling to ensure that the solutions comply with both operational and technical limitations. Figure 3 illustrates key operating constraints, including ramp-rate limits, thermal constraints, power-flow constraints, and pumped-station constraints. Identifying and integrating these constraints into the model is essential, as it enables the development of feasible, practical schedules. Step 5: Optimal capacity configuration. Step 5: Optimal capacity configuration. Optimal scheduling represents the fifth key stage, as it aligns with existing scheduling practices within an organization, such as lean manufacturing approaches. It follows the development of optimization models and the definition of system constraints. In this phase, optimal schedules for the generation and utilization of wind-solar energy storage are determined based on the solutions derived from the optimization process. Improved whale algorithm In this study, the model is developed using the Improved Whale Algorithm (IWA), which builds upon the Whale Optimization Algorithm (WOA)35 proposed by Mirjalili et al. in 2016 at Griffith University, Australia, recognized as a population-based optimization technique characterized by a simple structure, minimal parameter requirements, fast convergence, and ease of implementation. The key difference between the Improved Whale Optimization Algorithm (IWA)36 and the conventional WOA lies in the incorporation of the Grey Wolf Optimization hunting mechanism during the development phase. Specifically, the Grey Wolf Algorithm’s position update equation is integrated to enhance the bubble-net attacking strategy. The standard WOA identifies optimal solutions by mimicking three hunting behaviors of humpback whales: encircling prey, searching for prey, and the bubble-net feeding mechanism. It effectively combines global and local search capabilities, with its strength rooted in the adaptive adjustment of parameters to balance exploration and exploitation throughout the iterations. In the global search phase, the algorithm utilizes both encircling and random search strategies, enabling flexible changes in search direction. In contrast, the local search phase focuses on the bubble-net attacking mechanism to refine solutions around the current best candidate. Second, to enhance solution acceptance while maintaining strong global exploration, a conditional evaluation mechanism is introduced into the stochastic search process, enabling the algorithm to dynamically identify and retain promising solutions. The subsequent subsections present the workflow, solution representation, parameter update strategies, and execution details of the proposed IWA. Workflow of the algorithm Based on pertinent initialization rules, the IWA first produces initial solutions with a hybrid structure. The algorithm chooses the search procedure for the current iteration if the iteration termination criteria are not satisfied. To find better answers, the global search phase uses the surrounding and search methods. During the local search phase, the bubble-net encircling mechanism refines the current optimal solution by exploring its surrounding search space. Two features are shared by all search techniques in the IWA: the improved greedy search (IGS) algorithm chooses continuous observation targets based on the revolution sequence, while whale-inspired actions define the orbital revolution sequence. The IWA achieves efficiency and stability by avoiding local optima in the early stages and by concentrating on refining high-quality solutions in the latter stages, dynamically regulating the selection of search algorithms across iterations. Proposed solution structure design As illustrated in Fig. 4, the IWA solution incorporates three key constraints connected through a single revolution: the position parameter , revolution index and the target selection outcome . The spatial evolution of is characterised by , where m and n denote the initial and updated positions, respectively; when no positional update occurs, or the solution is excluded from the search process, m = n. In addition, the IWA effectively integrates bubble-net encircling for local exploitation with global search strategies to enhance exploration capability. As shown in Fig. 5, the IGS algorithm is used to dynamically segment candidate targets in order to minimize computing complexity and accomplish effective optimization under time and energy constraints, which presents the solution structure of the proposed IWA; furthermore, the solution must satisfy specific conditions, including fully organized revolutions, inclusion of all targets within the solution, and allowance for partially unfilled target positions within each revolution. Parameters connected through one revolution. Solution structure. The rules for updating parameters WOA execution is guided by continuously updating its parameters. The IWA adaptively adjusts these parameters to balance global and local search processes. The coefficient A, which influences global exploration, decides whether the algorithm searches new areas or moves toward the best-known solution. Its mathematical formulation is presented below: where a is a linearly decreasing variable from 2 to 0 over the iterations, and is an arbitrary number in the range [0, 1]. In addition, D represents the distance control parameter, which is defined as follows:: where t denotes the current iteration, represents the solution being approached or avoided, and. X(t)denotes the result to be encircled or located. The value of D is continuously rounded down to the near positive integer. The coefficient C is defined as follows: where represents a random value in the interval [0, 0.5]. The distance to be approached in the subsequent iteration is given by: where represents the final distance to be encircled. The function ensures that the computed distance is rounded down to the nearest integer. Encircling prey During the IWA search process, the distance between individuals is computed as the sum of spiral turns based on their particular positions. The reserve increases by 1 whenever the corresponding positions differ. Here, denotes the individual being updated, while signifies the control individual. If , the distance is incremented by 1. The following is the maximum separation between two people during the encircling phase: where n denotes the maximum number of revolutions and represents the Kronecker delta. Search for prey The formula for the greatest separation between two people during a search phase is provided by: where represents whether the ith position corresponding to X(t) and Xbest(t) is elaborate in a group at a distance. Also, indicates whether the ith position corresponding to Xrand(t) and Xbest(t) is a effective distance. The above-mentioned search process attempts to keep the IWA from approaching a random solution while converging to the ideal solution. IGS Although greedy search can efficiently and fast determine optimal solutions among candidate sets, it requires significant computational effort because it examines all possible combinations at every iteration. To mitigate this complexity, fixed-point constraints are applied to candidate targets, thereby restricting the search space. The IGS-based target selection in revolution is expressed as: The starting and ending points of imaging in a revolution are denoted by and the initial and final positions, while the intermediate targets form a set of candidate points divided into n segments. From separately segment, one target is arbitrarily chosen as the observation point . The selection range of these targets is defined within specified positional limits Uniform(1,k), where k is a constant, L represents the total number of candidate targets, and the complete set of candidates is considered for each revolution . This strategy effectively narrows the search space, thereby reducing computational effort while still satisfying all constraints. Optimization results of different energy storage schemes Parameter configuration In this study, an alpine region in Tengchong City, Yunnan Province, China, was selected as a case study, and the region’s wind speed and light intensity data for the entire year were extracted from the Meteonorm database. These data were scientifically referenced45 As shown in Fig. 6, the discrete representation of annual solar radiation intensity and wind speed data over time highlights their variability and stochastic nature. Discrete map of annual scenery data. This study proposes a wind-solar-storage hybrid power system for irrigation pump stations in alpine regions, comprising a 20 kW wind turbine, 0.8 kW of PV modules, a 3 kW·h lead-acid battery, and a 95% efficient converter. The system ensures stable power through wind-solar complementarity, with the battery mitigating fluctuations and converter losses limited to 5%. It also integrates a 5 kW electrolyser (producing 0.01798 kg/h of hydrogen per kWh), a 1 kg hydrogen tank, and a 1 kW fuel cell. The design optimizes cost-effectiveness by incorporating operation/maintenance (coefficient: 0.1) and disposal costs (coefficient: 0.06), balancing reliability and economic efficiency. Detailed economic analysis confirms its feasibility for renewable energy supply in Alpine areas. Table 2 presents a comparative analysis of equipment cost parameters, including purchase cost, replacement cost, and operation and maintenance cost for various energy system components. The lifetime of the wind and solar storage system is set at 20 years, during which the specific lifetimes of each subsystem are shown in Table 3. The table shows the lifespan of each component in the system, highlighting the variation in operational years across different pieces of equipment. Table 2Equipment cost parameter comparison table.EquipmentPurchase costReplacement costOperation and maintenance costPhotovoltaic cell (million yuan/kW)1.20.960.024Wind Turbine (million yuan/kW)1.61.280.032Battery (million yuan/kW-h)0.120.10.002Converter (million yuan/kW)0.670.670.006Electrolyzer (million yuan/kW)0.50.40.03Hydrogen storage tank (million yen/kg)110Fuel cell (million yuan/kW)0.60.50.438 Equipment cost parameter comparison table. Table 3A comparative table of equipment cost parameters.EquipmentWind turbinePhotovoltaic cellBatteryConverterFuel cellsElectrolysis tanksHydrogen storage tankLifespan/year152041552025 A comparative table of equipment cost parameters. Optimization model of hybrid energy storage with battery Assuming that there are 30,000 mu of land to be irrigated in this alpine area, 1 mu of land needs about 20 ~ 60 tons of water for a single irrigation, so it can be assumed that the water needed in this area is about 1.3 million tons of water per day. The comprehensive model is optimized by improving the whale algorithm to get the results of the pump station via operation and the capacity allocation of the wind energy storage system for this typical day. The flow allocation of different pumps for Scenario 1 is shown in Table 4. Table 4Each pump different time flow distribution table (m³/s).Time periodPump 1Pump 2Pump 3Pump 4Pump 5Pump 6Pump 7Pump 8Pump 9Pw P pv 12.762.762.7600000014.26022.642.642.6400000017.94032.522.522.5200000016.45042.262.262.2600000017.50051.741.741.7400000017.50062.312.312.3100000017.50072.572.572.5700000017.50080000003.253.253.2517.502.4690000003.643.643.6417.5029.3100000004.124.124.1217.5032.2111.561.561.560004.374.374.3717.5054.3121.391.391.392.782.782.784.624.624.6217.5067.5 Each pump different time flow distribution table (m³/s). Table 4 illustrates the detailed time-dependent flow distribution for each pump, showing that the operation of Pumps 1–9 varies across different time periods, while also incorporating the corresponding wind power and photovoltaic power outputs to reflect the coordinated utilization of available renewable energy resources. Figure 7 illustrates the hourly variation of power generation and consumption, integrating wind power, photovoltaic output, battery charging/discharging behavior, and load demand over a 24-hour period. The wind power contribution (blue bars) remains relatively stable throughout the day, with slight fluctuations, while photovoltaic generation (orange bars) increases significantly during daylight hours, peaking around midday and dropping to near zero at night. The load generated by the pumping station on a typical day for Scenario 1, the power from the PV and wind, and the battery input and output are shown in the figure. Overall, the figure demonstrates the coordinated operation of renewable energy sources and energy storage, where surplus energy is stored during high generation periods and later utilized to balance the load, ensuring a stable and efficient energy supply throughout the day. Photovoltaic wind power and battery input and output. Table 5 presents the optimized capacity configuration of the integrated energy system, summarizing the selected sizes of key components based on the optimization results. The system includes 56 wind turbines, indicating a significant reliance on wind energy as a primary renewable source. In addition, photovoltaic panels with a total capacity of 2400 kW are installed, complementing wind generation and enhancing energy supply during daylight hours. The number of capacity configurations of different devices for Scenario 1 is shown in Table 5. To achieve dependable, effective, and sustainable energy system performance, the table presents a balanced, optimal configuration that integrates generation, storage, and conversion components. Table 5Capacity configuration result.Equipment nameWind turbinePhotovoltaic panelsBatteryConverterOptimization results56 units2400 kW1760 units400 kW Capacity configuration result. Figure 8 presents the comparative overview of the system COC for different device types within the integrated energy system. It is evident that the photovoltaic system is the most capital-intensive component, accounting for the largest portion of the overall cost, at about 4300 (10,000 yuan). Figure 8 displays the system COC for Scenario 1. The wind power system, on the other hand, has a reasonable cost contribution of about 1400 (10,000 yuan), indicating a balanced investment in relation to its generation capacity. The battery system’s reduced cost, roughly 900 (10,000 yuan), highlights its role as a supporting element for energy storage and system stability rather than primary generation. In a similar vein, the present converter’s cost level is comparable, little under 1000 (10,000 yuan), highlighting its auxiliary role in enabling energy conversion and system component integration. Overall, the picture shows that supporting technologies like batteries and converters play crucial but less capital-intensive roles in maintaining effective and dependable system operation, even when renewable energy sources, especially solar systems, dominate investment costs. Overview of system COC. Table 6 presents the lifecycle cost distribution of the system’s components, including acquisition, replacement, and operation and maintenance (O&M) costs, all expressed in yuan. Due to the high cost of solar panels and energy storage, the photovoltaic-battery configuration is the most expensive subsystem, with an acquisition cost of 28,800,000 yuan and additional replacement and O&M costs of 7,183,987 and 7,363,219 yuan, respectively, for a total of 43,347,206 yuan. With a relatively modest initial investment of 8,900,000 yuan, replacement costs of 2,990,957 yuan, and maintenance costs of 2,290,779 yuan, the overall cost of wind turbines is 14,181,736 yuan, showing a moderate and balanced cost structure. The acquisition cost of the battery component alone is 6,336,000 yuan; replacement and O&M costs are 1,646,331 yuan and 1,349,923 yuan, respectively, for a total of 9,332,254 yuan, which represents the ongoing expenditures related to battery deterioration and replacement. With an initial cost of 7,500,000 yuan, a replacement cost of 2,086,325 yuan, and an O&M cost of 95,875 yuan, the converter has a total cost of 9,682,200 yuan, indicating strong dependability and minimal maintenance needs. With replacement and O&M costs of 13,907,600 yuan and 11,099,796 yuan, respectively, the total acquisition cost of the entire system is 51,536,000 yuan. This results in a total lifecycle cost of 76,543,396 yuan, indicating that although the initial investment makes up the largest portion of the total cost, ongoing replacement and maintenance expenses also contribute significantly. Table 6System COC specification.Configuration typeAcquisition cost (yuan)Replacement cost (yuan)Operation and maintenance cost (yuan)Total cost (yuan)Photovoltaic battery28,800,0007,183,9877,363,21943,347,206Wind turbines8,900,0002,990,9572,290,77914,181,736Batteries6,336,0001,646,3311,349,9239,332,254Converter7,500,0002,086,32595,8759,682,200Systems51,536,00013,907,60011,099,79676,543,396 System COC specification. Optimization model of hybrid energy storage with battery and hydrogen storage The hourly power output distribution of RES, such as wind and solar power, as well as hybrid energy storage contributions (battery and hydrogen systems), is shown in Fig. 9 in relation to load demand over a 24-hour period. While photovoltaic generation is low in the morning, it gradually increases after sunrise, peaks around midday (roughly 12:00–14:00), and then declines toward the evening as solar irradiance decreases. Wind power output stays relatively constant throughout the day, providing a continuous baseline supply. The load generated by the pumping station on a typical day for Scenario 2, the respective output powers of wind and PV, and the input and output of the battery and hydrogen storage system are shown in Fig. 9. On the other hand, both battery discharge and hydrogen energy discharge help balance the system by providing stored energy during early morning and evening hours when renewable generation is not enough to meet the load. The load profile, shown by the black line with red markers, gradually increases in the morning, peaks at midday, then declines in the evening before slightly increasing again at night. The efficient integration of multiple energy sources and storage technologies to maintain system reliability and stability under varying load and generation conditions is demonstrated by the coordinated operation of wind power, solar PV, and hybrid energy storage, ensuring that supply consistently meets demand. Output power of RES and hybrid energy storage system. Table 7 illustrates the ideal configuration of the hybrid renewable energy system, comprising 56 wind turbines, 2400 kW of photovoltaic capacity, 820 battery units, a 3000 kW converter, and a hydrogen subsystem with a 500 kW fuel cell, a 500 kW electrolysis unit, and 240 storage tanks. This balanced design integrates wind, solar, battery, and hydrogen storage to ensure a dependable and effective energy supply. Table 7 displays the quantity of capacity configurations for the various devices in Scenario 2. Table 7Simulation result of the capacity configuration.Equipment nameWind turbinePhotovoltaic panelBatteryConverterFuel cellElectrolysis tanksHydrogen storage tankOptimization results56 units2400 kW820 units3000 kW500 kW500 kW240 Simulation result of the capacity configuration. An overview of the COC for various device types in the energy system, expressed in units of 10,000 yuan, is shown in Fig. 10. At a cost of almost 4,300 (×10,000 yuan), the photovoltaic subsystem is by far the most expensive component, signifying that solar generation accounts for the majority of the total investment. With a cost of about 1,400, the wind power component is the second most expensive, suggesting a substantial but significantly lesser contribution than photovoltaics. The electrolytic bath (electrolyzer), which plays a part in the creation of hydrogen, is somewhat less expensive at roughly 550, while the current charger comes in second at nearly 1,000. The fuel cell, battery, and hydrogen storage tank, on the other hand, individually account for a comparatively small amount of the overall cost, less than $200. Overall, the chart makes it abundantly clear that the majority of system expenses are attributable to power production technologies, particularly photovoltaic systems, while storage and hydrogen-related components account for only a small portion. System COC overview. Acquisition, replacement, operation and maintenance (O&M), and total costs for each configuration type are broken down in Table 8 to summarize the system COC. At 43,347,206 yuan, the photovoltaic system has the highest overall cost. This is mostly due to its extremely high acquisition cost of 28,800,000 yuan, as well as significant replacement and maintenance costs. At 14,181,736 yuan, wind turbines are the second-largest expense, with significant contributions from both replacement (2,990,957 yuan) and acquisition (8,900,000 yuan). Despite having relatively low O&M expenditures, the converter’s total cost of 9,682,200 yuan is unusually high due to acquisition and replacement expenses. Fuel cells, batteries, electrolyzers, and hydrogen storage tanks, on the other hand, contribute significantly less to the total system cost. The overall cost of batteries is 1,691,182 yuan, and the need for frequent replacements is indicated by the replacement cost exceeding the acquisition cost. While hydrogen storage tanks cost 1,115,050 yuan and have no replacement costs but substantial O&M expenses, the electrolyzer has one of the lowest overall costs (566,227 yuan) with little O&M expenses. The replacement cost (1,085,187 yuan) accounts for the majority of the fuel cells’ overall cost (1,831,887 yuan). Table 8System COC specification.Configuration typeAcquisition cost ($)Replacement cost (yuan)Operation and maintenance cost (yuan)Total cost (yuan)Photovoltaic battery28,800,0007,183,9877,363,21943,347,206Wind turbines8,900,0002,990,9572,290,77914,181,736Batteries295,200767,041628,9411,691,182Converter7,500,0002,086,32595,8759,682,200Electrolyzer500,00062,3623865566,227Hydrogen storage tanks1,000,0000115,0501,115,050Fuel cells670,0001,085,18776,7001,831,887Systems47,665,20014,175,85910,574,42972,415,488 System COC specification. Results and discussion The impact of a single energy storage device and a hybrid energy storage device on the battery is primarily reflected in the battery’s State of Charge (SOC). In this paper, the SOC change graph of the battery in the two energy storage schemes is obtained through simulation and serves as the basis for assessing the impact on battery performance. The battery’s state-of-charge diagram for single-battery energy storage is shown in Fig. 11, while that for hybrid energy storage is shown in Fig. 12. Graph shows the variation of the battery SOC for Scheme 1. Graph shows the variation of the battery SOC for Scheme 2. The hourly variation of battery SOC for two different schemes is shown in Figs. 11 and 12. Figure 10 (Scheme 1) shows highly fluctuating SOC values over a wider range (roughly 30% to 100%), indicating less stable battery management, while Fig. 11 (Scheme 2) shows relatively smoother and more uniformly distributed SOC levels with fewer extreme highs and lows, suggesting improved control, better energy balancing, and enhanced battery utilization throughout daily and seasonal cycles. By comparing and analyzing Figs. 11 and 12, it is easy to find that the battery in single energy storage is in the peak state for a long time, and such a high load use state not only affects its performance, but also shortens its service life; while the battery in hybrid energy storage is in a relatively reasonable interval, which indicates that the hybrid energy storage system proposed in this paper can effectively alleviate the state of the battery being in the peak state for a long time, and it can effectively alleviate the pressure of the battery and prolong its service life. Therefore, in terms of battery SOC analysis, Scheme 2 is superior to Scheme 1. Re-evaluating the economic advantages of the two schemes. After performing the optimization operation for each of the two sets of schemes, the results are shown in Table 9. While both Scheme 1 and Scheme 2 use the same number of wind turbines (56 units), photovoltaic capacity (2400 kW), and fuel cell capacity (500 kW), Scheme 1 heavily relies on a larger battery storage system (1670 units) with no hydrogen-related components. In contrast, Scheme 2 adopts a hybrid approach by reducing battery capacity (820 units) and incorporating additional technologies like a 5000 kW electrolysis tank and 120 hydrogen storage tanks, which results in a lower total cost (72,415,488 Yuan) than Scheme 1 (76,543,396 Yuan). Table 9A comparison table of different energy storage schemes.EquipmentWind turbinePhotovoltaic panelBatteryElectrolysis tankHydrogen storage tanksFuel cellTotal amount(Yuan)Scheme 156 units2400 kW1670 units00500 kW76,543,396Scheme 256 units2400 kW820 units5000 kW120 units500 kW72,415,488 A comparison table of different energy storage schemes. Secondly, through the analysis of the results in Table 10, it can be obtained that under the condition of the same installed capacity of the wind and light storage system, the impact of Scheme 1 and Scheme 2 on the economic operation of the pumping station is very weak, but presents a significant difference in the economy. The COC of scheme 1 is 76,543,396 yuan, and the COC of scheme 2 is 72,415,488 yuan, and this data shows that the economy of scheme 2 is significantly better than that of scheme 1 for a certain amount of water lifting, and the cost difference between them is 4,127,908 yuan, which reaches 5.4%. Although Scheme 2 adds a hydrogen energy storage system and increases system complexity, it is more economical than Scheme 1 because it reduces battery use and losses, extends battery life, and is particularly well-suited to improving the cost-effectiveness and sustainability of power supply systems in remote or alpine areas. Table 10Comparison of different algorithms with solar-PV energy storage.AlgorthimMinimumoperating cost($)Meanoperating cost($)Maximumoperating cost($)Standarddeviation cost($)Executiontime (s)WOA3078,765,42178,765,41978,765,415---47.532IMO3477,325,43277,325,42677,325,422---41.543SGEO3677,324,32177,324,31877,324,3150.24334.654IWA (proposed)76,543,39676,543,39276,543,3870.35432.765 Comparison of different algorithms with solar-PV energy storage. Table 10 summarizes the statistical performance of different algorithms under solar–PV energy storage conditions, including minimum, mean, and maximum operating costs, standard deviation, and execution time. As observed, the proposed IWA algorithm achieves the lowest operating cost among all methods, with minimum, mean, and maximum costs of $7,654,396, $7,654,392, and $7,654,387, respectively, outperforming WOA, IMO, and SGEO. In addition, IWA records the shortest execution time of 32.765 s, compared to 34.654 s for SGEO, 41.543 s for IMO, and 47.532 s for WOA, indicating faster computational efficiency. Although SGEO exhibits slightly lower variability, the overall results indicate that IWA offers a better trade-off between solution quality and computational speed. This improved efficiency enables faster decision-making in applications such as demand response, load shifting, and load shedding, thereby enhancing demand-side management, grid reliability, and the effective integration of renewable energy sources under uncertain conditions. These have shown that the proposed method exhibits superior performance and convergence properties compared to other methods in the same domain. As illustrated in Fig. 13, all considered algorithms exhibit a typical three-stage convergence behaviour, characterised by a rapid decrease in fitness during the initial iterations, followed by a gradual decline, and finally reaching a stable plateau. The IWA algorithm achieves the lowest final fitness value of 0.1, compared with 0.2 for WOA, 0.4 for SGEO, and 0.8 for IMO. Relative to SGEO, IWA reduces the final fitness by 0.7 (87.5% improvement); relative to IMO, the reduction is 0.3 (75% improvement). From the figure, the proposed IWA algorithm converges faster and achieves a lower fitness value than WOA, SGEO, and IMO. Specifically, IWA reaches its convergence region earlier, indicating faster convergence and greater stability. In contrast, IMO and SGEO show slower convergence and higher final fitness values, while WOA demonstrates moderate performance. These results highlight that the proposed IWA effectively balances exploration and exploitation, benefits from parallel search capability, and efficiently avoids local minima, thereby significantly enhancing convergence performance for the considered optimization problem. Comparison of different optimization algorithms. Conclusion In this paper, an energy allocation problem for a wind-solar PV energy storage system to solve the uncertainty problem is proposed to support the economic operation of irrigation pumping stations in alpine areas, and the system’s optimisation method is studied to minimise total cost over the whole life cycle. The main conclusions are as follows: Curve fitting is performed for the power–flow, head–flow, and efficiency–flow characteristics of three distinct pump types. Using the investment data of the pumping station, parameters for start–stop losses and operational depreciation costs are determined, and an economic operation model is developed for the station with multiple pump types operating in parallel, incorporating these cost factors. Curve fitting is performed for the power–flow, head–flow, and efficiency–flow characteristics of three distinct pump types. Using the investment data of the pumping station, parameters for start–stop losses and operational depreciation costs are determined, and an economic operation model is developed for the station with multiple pump types operating in parallel, incorporating these cost factors. An output model of the wind energy storage system is developed to establish a capacity allocation model that minimizes the cost of operation. Two scenarios-battery storage and battery–hydrogen storage—are analyzed to determine the optimal wind–solar storage configuration, along with pump station flow distribution and unit operation for a typical day. Results show that incorporating hydrogen storage enhances both economic and sustainability performance. In terms of cost reduction, IWA achieves a 3.3% decrease in annual cost, outperforming WOA (2.9%), IMO (1.02%), and SGEO (1.1%). Additionally, IWA converges faster, reaching stability around iteration 700 with a final fitness value of 0.1, compared to 0.2 (WOA), 0.4 (IMO), and 0.8 (SGEO). An output model of the wind energy storage system is developed to establish a capacity allocation model that minimizes the cost of operation. Two scenarios-battery storage and battery–hydrogen storage—are analyzed to determine the optimal wind–solar storage configuration, along with pump station flow distribution and unit operation for a typical day. Results show that incorporating hydrogen storage enhances both economic and sustainability performance. In terms of cost reduction, IWA achieves a 3.3% decrease in annual cost, outperforming WOA (2.9%), IMO (1.02%), and SGEO (1.1%). Additionally, IWA converges faster, reaching stability around iteration 700 with a final fitness value of 0.1, compared to 0.2 (WOA), 0.4 (IMO), and 0.8 (SGEO). These results confirm that the IWA algorithm achieves superior performance in optimization accuracy, convergence speed, and grid stability maintenance compared with traditional and recent swarm intelligence methods, effectively balancing local and global searches, and improving the search speed and convergence accuracy of the algorithm. These results confirm that the IWA algorithm achieves superior performance in optimization accuracy, convergence speed, and grid stability maintenance compared with traditional and recent swarm intelligence methods, effectively balancing local and global searches, and improving the search speed and convergence accuracy of the algorithm. In future work, research will be extended to the user side to improve the algorithm’s adaptability and solving speed under more complex demand response scenarios, including uncertainty quantification for renewable generation and EV user behavior. Also, future work will focus on applying and validating the proposed model and algorithm in large-scale industrial scenarios, incorporating more realistic operational constraints. Abbreviations Renewable energy sources Photovoltaic Energy storage system Cost over the life cycle State of charge Whale optimization algorithm Improved mayfly optimization algorithm Social group entropy optimization Improved whale algorithm Index, sets, parameter and variables Minimum total cost of the pumping station Number of pump Number of time slots divided Time period length of the time period and depreciation cost of the unit during operation Starting and stopping status of the unit in the time period Loss cost of starting and stopping the unit once during the operation period Total amount of water lifted by the pumping station The flow of the unit in the time period The minimum and maximum flow rates of the unit in the time period The power required for time period in the water lifting process The volumetric weight of water, generally + taken as 9800kN/m3 The head of the unit at the flow rate of the time period , m The efficiency corresponding to the flow rate of the unit in the time period The maximum power that can be supplied by the wind and solar storage system during the water lifting process in the time period The health status of the unit The health degree of the generator, pump, valve and pipeline of the unit, respectively Economic life of the project The investment cost of the project Social discount rate The number of years that the pump investment Equipment purchase cost Residual value generally takes 3–5% Equipment life The minimum cost of water lifting for a single unit in one day Water lifting volume of the unit The total amount of water lifted by the system Output power of the wind turbine Rated power of the wind turbine Cut in wind speed and rated wind speed Cut out wind speed Output power of PV generation system, kW Light intensity, kW/m2 Photovoltaic module area, m2 Photovoltaic module nominal efficiency Temperature coefficient Cell operating and standard test temperature Combined cost over the life cycle Acquisition and replacement cost Operation and maintenance costs The load demand power at time of , kW Actual output power of wind turbine at the time of , kW Actual output power of the photovoltaic generation system at the time of , kW The maximum discharge power of the battery storage system at the time of , kW The maximum permissible power shortage rate of the power generation system, 5%, is selected in this study Author contributions All authors have contributed to the paper. Following are the contribution details: Pranda Prasanta Gupta was involved in Conceptualization, writing-original draft and visualization. Deepak Gupta and Kavita Tukaram Patil participated in problem formulation and simulations. Kishor S. Wagh and Chanakya Kumar Jha were responsible for result validation and the proposed methodology. Pallavi Jha and Akhilesh Kumar Sharma took part in the writing – review & editing and supervision. Funding Open access funding provided by Manipal University Jaipur. Data availability No datasets were generated or analysed during the current study. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Pranda Prasanta Gupta, Email: prandaprasantagupta@gmail.com. Akhilesh Kumar Sharma, Email: akhileshkumar.sharma@jaipur.manipal.edu. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.