Shared Power–Hydrogen Energy Storage Capacity Planning and Economic Assessment for Renewable Energy Bases

Abstract

Large-scale renewable energy bases in desert regions face challenges of unstable output and inefficient utilization due to the fluctuating nature of wind and solar power. To address these issues, this study proposes an optimization model for shared hybrid electricity–hydrogen energy storage across multiple micro-energy systems. The model minimizes the total investment and operation cost under electricity–hydrogen coupling and system balance constraints, and an improved Shapley value method is introduced to ensure fair cost allocation among participants. A case study based on a desert renewable base shows that the proposed shared configuration reduces the total annualized cost by 10.36% and increases renewable energy utilization by 12.19% compared with independent electrical storage systems. These results demonstrate that shared hybrid storage can effectively enhance energy utilization and cost efficiency in large-scale renewable energy bases, providing a feasible approach for integrated power–hydrogen energy management.

1. Introduction

Against the backdrop of establishing large-scale renewable energy bases in desert regions abundant in wind and solar resources, energy storage systems have emerged as low-carbon flexibility resources to address the challenges posed by the concentrated development of high proportions of renewable energy sources and their local consumption [1]. Especially in desert areas rich in wind and solar resources, the contradiction between the development and consumption of new energy is more prominent. Energy storage systems smooth the fluctuations of new energy on the source side and ensure high-quality energy use on the load side [2], becoming an important technical support for the safe and stable operation of the new power system, the deep decarbonization of the energy structure and the realization of the “dual carbon” goals. At the same time, energy storage systems are giving rise to new forms of energy in China. The Implementation Plan for the Development of New Energy Storage during the 14th Five-Year Plan period clearly requires that new energy storage should move from the initial commercial stage to the large-scale development stage by 2025 [3], and encourages the layout of energy storage projects in areas rich in new energy, such as deserts. Many places are accelerating the deployment of energy storage equipment, and the industrial layout momentum is strong. It is evident that in the context of the accelerated development and scenario-based implementation of energy storage systems, there is an urgent need to build energy storage solutions that are economically viable, have superior performance and high reliability.

The microgrid (MG) with clean energy generation, due to its “self-generated and self-consumed” power feature, enables energy storage configuration to have dual functions on both the power generation side and the user side, and is a solution [4] for load parties and power generation enterprises to jointly build and share energy storage in the implementation of the “green power direct connection” policy. The multi-micro energy grid (MMEG) as an MG cluster has significant complementary characteristics [5] in terms of energy generation and consumption and energy storage charging and discharging behaviours among its members. In desert areas, different microgrids may face different temporal and spatial distributions of wind and solar resources and load demands. The development constraints, such as high cost and low utilization rate of energy storage in the independent configuration mode of MG, can be broken to a certain extent by co-building energy storage and sharing costs [6]. The Action Plan for Accelerating the Construction of a New Power System (2024–2027) sets out requirements for the layout of shared energy storage power stations, strengthening the capacity for new energy consumption and power supply security. Shared energy storage (SES) is a new business model based on the concept of “sharing economy”, which achieves unified energy dispatch through the mechanism of “crowdfunding and co-construction, cluster sharing”, and plays a role in enhancing power interaction among MG [7].

The current research on SES mainly focuses on rational capacity allocation and economic analysis: Yu et al. [8] consider the source-load uncertainty and MG power interaction and verifies that MMEG co-construction of SES can significantly reduce the configured capacity; Jin et al. [9] confirmed the economic benefits and risk control gains through two-stage robust optimization; Shen [10] constructed SES energy optimization scheduling mechanism based on demand response mechanism; Song et al. [11] proposed a method for optimizing the configuration of SES under multiple uncertainties, verifying that equipment utilization efficiency was significantly improved compared to independent configuration. All of the above studies demonstrated the superiority of SES in terms of utilization efficiency and economic benefits. However, the current dominant force in this field is still electrochemical shared storage (ESS) [12], which has limitations in terms of long-term regulation capacity [13], cycle life and environmental friendliness. Therefore, integrating diversified energy storage technologies to further break through storage duration and economic bottlenecks, and rationally allocating hybrid energy storage capacity, is the key path to deepening the value of shared energy storage.

Hydrogen energy storage, with its long-term regulation capacity, high energy density and clean characteristics, has significant advantages in achieving cross-temporal and spatial energy transfer [14], and has become a key adjustable resource for the new power system. The abundant wind resources in desert regions provide ideal conditions for electrolytic hydrogen production. However, hydrogen’s characteristics of requiring large-scale, long-term storage often led to periods of equipment idleness, resulting in suboptimal resource utilisation and economic efficiency [15]. At the same time, the reduction in its unit investment cost depends on economies of scale [16]. Therefore, co-building and sharing hydrogen storage in MMEG with power mutual aid characteristics can improve resource utilization efficiency.

Existing studies have verified the value of shared hydrogen energy storage: Xu et al. [17] verified that shared hydrogen energy storage in multiple parks can reduce system operating costs and increase operating benefits compared to self-distribution; Ren et al. [18] indicate that the introduction of shared hydrogen storage in distributed wind power user groups can reduce initial investment; Shi et al. [19] suggest that although shared hydrogen storage increases the investment cost, cooperative and mutual dispatching among microgrids can still enhance the net operating income. Other studies have shown that the optimized configuration model of electric–hydrogen hybrid energy storage can effectively avoid excessive investment in lithium battery energy storage [20], and it performs better in terms of economy and new energy consumption capacity compared with single electric–hydrogen energy storage [21]. In addition, the existing literature mostly uses game theory to analyze the benefits of MMEG members and uses Shapley values to allocate cooperative costs and benefit distribution in SES configuration problems [22]. Therefore, to address the requirements for establishing renewable energy bases in desert regions, constructing a cross-timescale dispatch framework tailored to the characteristics of integrated wind–solar–storage operations, establishing hydrogen electric shared storage (HESS), and employing game theory for benefit analysis represent effective approaches to overcoming the limitations of energy storage development while enhancing system operational efficiency and fairness.

To sum up, existing research on SES primarily focuses on ESS, with preliminary explorations into standalone hydrogen energy storage configurations. However, studies examining how HESS supports coordinated energy scheduling across temporal scales remain limited, particularly lacking application validation in the specific environmental conditions of desert regions. At the same time, although existing results have verified the benefits of the shared model over the independent model, research on quantifying the multi-dimensional contributions of cooperative alliances and the mechanism of fair cost-sharing is still insufficient.

This paper constructs the MMEG and SES system architecture, establishing an energy storage system capacity configuration model tailored to the characteristics of typical wind and solar resources in desert regions. Unlike conventional hybrid electricity–hydrogen storage models that mainly focus on single microgrids or separate optimization stages, the proposed model integrates multi-microgrid coupling and shared operation into a unified optimization framework. The model aims to minimize the combined costs of SES equipment and MMEG operational expenses, yielding optimal energy storage configurations and MMEG operational outcomes across multiple scenarios. Further, on the basis of the optimized operation of MMEG, benefit analysis was conducted, and the improved Shapley value method based on interaction contribution was used to fairly allocate the cost savings of each MG, providing a basis for cost allocation. Finally, simulations of typical cases in desert areas show that compared with independent planning of energy storage or co-construction of ESS, MMEG co-configuration of HESS has significant superiority in terms of system economy and new energy consumption rate.

2. Micro-Energy Network Architecture and Shared Energy Storage System Within Renewable Energy Bases

2.1. System Architecture

The renewable energy base incorporates MMEG, which comprises multiple MGs integrating four energy flows: electricity, natural gas, cooling and heating. Its structure and energy flow are shown in Figure 1. It specifically consists of a renewable energy power generation (REPG) system, a combined cooling heating and power (CCHP) system and auxiliary equipment. The REPG consists of photovoltaic arrays (PVs) and wind turbines (WTs); The CCHP system includes gas turbines (GTs), absorption chillers (ACs) and waste heat boilers (HRSs); auxiliary equipment includes electric refrigerators (ECs) and gas boilers (GBs).

MMEG members jointly deploy centralized SES with minimal energy cost and no commercial profit. SES uses an electric–hydrogen energy storage hybrid architecture where electrochemical energy storage is responsible for power regulation within the day; hydrogen storage consists of an electrolyzer (EL), a hydrogen storage tank (HST) and a fuel cell (FC) for seasonal energy transfer. Members share costs and waive internal transaction fees. The MMEG co-construction SES structure is shown in Figure 2.

MG shares wind and solar output projections and load information, and based on the complementarity of energy storage demand time series, uniformly plans SES and participates in optimal scheduling. The interaction logic between MMEG and SES is as follows: when the net charge–discharge demand of the alliance is zero, internal energy self-balancing is achieved through virtual buffer channels; when there is a net load deviation, the energy storage device is invoked in response. MMEG provides support for capacity optimization by coordinating the production and sales behavior of multiple MG of energy to enhance energy storage utilization and new energy consumption capacity.

2.2. MMEG Equipment Model

The mathematical model of the output of the MG equipment within the MMEG is presented in Appendix A.

2.3. SES Equipment Model

The mathematical model of the output of the internal equipment of SES is presented in Appendix B.

3. A Shared HESS Capacity Configuration Model

To ensure that MMEG is not an isolated island and avoid exerting excessive pressure on the external power grid, MG maximizes the utilization of self-generated electricity and relies on SES to achieve energy mutual assistance. Therefore, the capacity configuration of SES should be planned in combination with the operational optimization problems of each MG.

3.1. Objective Function

The objective of the optimization model is to minimize the sum of the SES investment cost and MMEG operating cost. The objective function is as follows:

$$\min F = C_{\mathrm{i},\mathrm{ess}}+C_{\mathrm{i},\mathrm{hss}}+C_{\mathrm{p},\mathrm{e}}+C_{\mathrm{p},\mathrm{f}}+C_{\mathrm{p},\mathrm{p}}+C_{\mathrm{om}}$$

(1)

where F is the total cost; $C_{\mathrm{i},\mathrm{ess}},C_{\mathrm{i},\mathrm{hss}}$ share the investment costs of electricity and hydrogen storage, respectively, and the specific calculations are detailed in Equations (2)–(6); $C_{\mathrm{p},\mathrm{e}}$, $C_{\mathrm{p},\mathrm{f}},C_{\mathrm{p},\mathrm{p}}$ are, respectively, the cost of purchasing electricity, the cost of gas and the cost of abandoning electricity, and the specific calculations are detailed in Equations (7)–(9); $C_{\mathrm{om}}$ MMEG and SES operation and maintenance costs, and the specific calculations are detailed in Equation (10).

$$C_{\mathrm{i},\mathrm{ess}}=({\delta }_{\mathrm{ess},\mathrm{p}}{P}_{\mathrm{ess}}+{\delta }_{\mathrm{ess},\mathrm{v}}{V}_{\mathrm{ess}}){\displaystyle \frac{r{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{ess}}}}{{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{ess}}}-1}}+Z_{\mathrm{ess}}$$

(2)

$$C_{\mathrm{i},\mathrm{hss}}=C_{\mathrm{i},\mathrm{el}}+C_{\mathrm{i},\mathrm{fc}}+C_{\mathrm{i},\mathrm{hst}}+Z_{\mathrm{h}}$$

(3)

$$C_{\mathrm{i},\mathrm{el}}={\displaystyle \frac{r{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{el}}}}{{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{el}}}-1}}{\delta }_{\mathrm{el},\mathrm{p}}{P}_{\mathrm{el}}$$

(4)

$$C_{\mathrm{i},\mathrm{fc}}={\displaystyle \frac{r{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{fc}}}}{{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{fc}}}-1}}{\delta }_{\mathrm{fc},\mathrm{p}}{P}_{\mathrm{fc}}$$

(5)

$$C_{\mathrm{i},\mathrm{hst}}={\displaystyle \frac{r{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{hst}}}}{{\left(1+r\right)}^{{\mathrm{t}}_{\mathrm{hst}}}-1}}{\delta }_{\mathrm{hst},\mathrm{v}}{V}_{\mathrm{HST}}$$

(6)

In the formula, $C_{\mathrm{i},\mathrm{el}}$,$C_{\mathrm{i},\mathrm{fc}}$ and $C_{\mathrm{i},\mathrm{hst}}$ are EL, FC and HST investment costs, respectively; ${\delta }_{\mathrm{ess},\mathrm{p}}$, ${\delta }_{\mathrm{el},\mathrm{p}}$ and ${\delta }_{\mathrm{fc},\mathrm{p}}$ are the investment costs per unit power for electrical energy storage, EL and FC, respectively; ${\delta }_{\mathrm{ess},\mathrm{v}}$ and ${\delta }_{\mathrm{hst},\mathrm{v}}$ are the investment costs per unit capacity for electrical energy storage and HST, respectively; $t_{\mathrm{ess}}$, ${ t}_{\mathrm{el}}$, $t_{\mathrm{fc}}$, and $t_{\mathrm{hst}}$ represent the life cycle of electric energy storage, EL, F and HST, respectively; $Z_{\mathrm{ess}}$ and $Z_{\mathrm{h}}$ are the operation and maintenance costs of electric and hydrogen energy storage, respectively, and are the benchmark discount rates.

$$C_{\mathrm{p},\mathrm{e}}={\displaystyle \sum _{i=1}^{\mathrm{N}}{\displaystyle \sum _{t=1}^{\mathrm{T}}{\phi }_{\mathrm{p},\mathrm{e}}(t)P_{\mathrm{p},\mathrm{e}}^i(t)}\Delta t}$$

(7)

$$C_{\mathrm{p},\mathrm{f}}={\displaystyle \sum _{i=1}^{\mathrm{N}}{\displaystyle \sum _{t=1}^{\mathrm{T}}{\phi }_{\mathrm{p},\mathrm{f}}\left(\frac{P_{\mathrm{GT}}^i(t)}{{\eta }_{\mathrm{GT}}{L}_{\mathrm{NG}}}+\frac{Q_{\mathrm{GB}}^i(t)}{{\eta }_{\mathrm{GB}}{L}_{\mathrm{NG}}}\right)}\Delta t}$$

(8)

In the formula, ${\phi }_{\mathrm{p},\mathrm{e}}(t)$ and ${\phi }_{\mathrm{p},\mathrm{f}}$ are electricity prices and gas prices, respectively; $P_{\mathrm{p},\mathrm{e}}^i(t)$ is the power purchased for $\mathrm{M}{\mathrm{G}}_i$; N represents MG count and T represents the annual operating hours.

$$C_{\mathrm{p},\mathrm{p}}={\displaystyle \sum _{i=1}^N{\displaystyle \sum _{t=1}^{\mathrm{T}}{c}_{\mathrm{p},\mathrm{p}}{P}_{\mathrm{p},\mathrm{p}}^i(t)}\Delta t}$$

(9)

In the formula, $P_{p,p}^i(t)$ is $\mathrm{M}{\mathrm{G}}_i$ the abandoned power for the t period.

$$C_{\mathrm{om}}={\displaystyle \sum _{i=1}^N{\displaystyle \sum _{m=1}^M{\displaystyle \sum _{t=1}^{\mathrm{T}}{c}_{\mathrm{m},\mathrm{om}}{P}_m^i(t)}}}$$

(10)

In the formula, $m=\mathrm{WT}, \mathrm{PV}, \mathrm{GT}, \mathrm{GB}, \mathrm{HRS}, \mathrm{EC}, \mathrm{AC}$, $c_{\mathrm{m},\mathrm{om}}$ is the operation and maintenance cost of each device, and $P_m^i(t)$ is the output power of the device m.

3.2. Constraints

The constraints are in Appendix C. Specific constraints include power balance, equipment output, maximum power purchase, power interaction constraints, energy storage constraints, etc.

4. Cost Allocation Mechanism Based on Game Theory

Against the backdrop of renewable energy-based development in desert regions, one key challenge facing the co-construction and shared use of SES within multi-microgrid systems lies in designing a scientifically sound cost-sharing mechanism to achieve equitable distribution of cooperative surplus benefits. This paper constructs an SES cost-sharing mechanism tailored to renewable energy-based scenarios, based on the Shapley value framework within cooperative game theory. Addressing the limitation of traditional Shapley value models in adequately assessing users’ actual marginal contributions, it innovatively introduces an interactive value contribution factor. This quantifies users’ differentiated value during joint operation from multiple dimensions, including energy interaction intensity and regulation support capabilities, thereby effectively enhancing the fairness of cost sharing and user acceptance.

4.1. Traditional Shapley Value Cost Allocation

Suppose there is an MG willing to form an alliance, then S represents the alliances that might be formed without participation; $\left|S\right|$ is the number of MG participating in the alliance; $\left|S\right|!$ represents the possible permutations of MG in MMEG; and $\left(n-\left|S\right|-1\right)!$ represents the remaining possible permutations of MG divided by $\mathrm{M}{\mathrm{G}}_i$ and participates in the alliance S. Further, ${\displaystyle \frac{1}{n!}}\left|S\right|!\left(n-\left|S\right|-1\right)!$ indicates the frequency at which all possible permutations $\mathrm{M}{\mathrm{G}}_i$ occur at a specific position in the alliance S; $\mathrm{M}{\mathrm{G}}_i$ after joining the alliance S, a marginal contribution is made $\nu \left(S\cup \left\{i\right\}\right)-\nu \left(S\right)$. The marginal contribution is multiplied by the corresponding weight and summed up to obtain the cost savings that $\mathrm{M}{\mathrm{G}}_i$ should be allocated. Here is the Formula (11):

$${\phi }_i(\nu )={\displaystyle \sum _{S\subset N}\frac{\left(\left|S\right|\right)!\left(n-\left|S\right|-1\right)!}{n!}\left[\nu \left(S\cup \left\{i\right\}\right)-\nu \left(S\right)\right]}$$

(11)

In the formula, to save costs for the division, ${\phi }_i\left(\nu \right)$ is the cost savings allocated to $\mathrm{M}{\mathrm{G}}_i$;$N=\left\{\mathrm{1,2},\dots ,n\right\}$; $\nu \left(S\right)$ is the benefit generated by the entire coalition when $\mathrm{M}{\mathrm{G}}_i$ participates in coalition $S$; $\nu \left(S\cup \left\{i\right\}\right)$ is the benefit generated after $\mathrm{M}{\mathrm{G}}_i$ joins coalition $S$.

4.2. An Improved Shapley Value Method Based on Interaction Value Contribution

Within the SES of the renewable energy base, both the charging and discharging activities of the MG towards the SES may be regarded as its overall contribution to the alliance. However, existing extensions of the Shapley value still have limitations in shared energy-storage applications. Weighted Shapley methods rely on fixed importance factors that do not reflect the time-varying charging and discharging behaviours of microgrids. Interaction-index-based methods can capture synergies within coalitions, but they do not incorporate the temporal economic value of energy transactions under time-of-use pricing. Weighted Shapley formulations rely on exogenous and fixed weights, which do not reflect the time-varying energy exchange behaviours of microgrids. Interaction indices, although capable of capturing coalition synergies, do not explicitly incorporate the temporal economic value of charging or discharging actions, particularly under time-of-use electricity pricing.

To address these gaps, the proposed method introduces a dynamic interaction-value-based weighting mechanism that directly links each microgrid’s marginal contribution to its actual charging/discharging behaviour and the associated time-dependent economic value. This enables the allocation weights to reflect not only the magnitude of energy exchanged with the shared storage but also the temporal value of that energy. In this way, the improved method differs from existing weighted or interaction-index Shapley extensions by integrating temporal energy value and operational interaction intensity into the cost-allocation logic. This provides a more behaviour-consistent and economically grounded representation of contribution within shared-storage cooperation. Based on this, this paper calculates the contribution of MG using an exponential function $R_i$:

$$R_i={\displaystyle \frac{v_i}{V}}$$

(12)

$$v_i=e^{{\displaystyle \frac{E_{\mathrm{ES},\mathrm{su}}^i}{E_{\mathrm{ES},\mathrm{su}}}}}-e^{-({\displaystyle \frac{E_{\mathrm{ES},\mathrm{ob}}^i}{E_{\mathrm{ES},\mathrm{ob}}}})}$$

(13)

$$E_{\mathrm{ES},\mathrm{S}}^i={\displaystyle \sum _{w=1}^W{\displaystyle \sum _{t=1}^{N_T}\left[{\phi }_{\mathrm{p},\mathrm{e}}(t)\left(P_{\mathrm{ESS},\mathrm{S}}^i(t)+P_{\mathrm{HSS},\mathrm{S}}^i(t)\right)\right]\Delta t}}$$

(14)

$$E_{\mathrm{ES},\mathrm{B}}^i={\displaystyle \sum _{w=1}^W{\displaystyle \sum _{t=1}^{N_T}\left[{\phi }_{\mathrm{p},\mathrm{e}}(t)\left(P_{\mathrm{ESS},\mathrm{B}}^i(t)+P_{\mathrm{HSS},\mathrm{B}}^i(t)\right)\right]\Delta t}}$$

(15)

$$E_{\mathrm{ES},\mathrm{S}}={\displaystyle \sum _{i=1}^N{E}_{\mathrm{ES},\mathrm{S}}^i}$$

(16)

$$E_{\mathrm{ES},\mathrm{B}}={\displaystyle \sum _{i=1}^N{E}_{\mathrm{ES},\mathrm{B}}^i}$$

(17)

$$V={\displaystyle \sum _{i=1}^N{v}_i}$$

(18)

In the formula, $v_i$ is the contribution value of ${\mathrm{MG}}_i$; V is the overall contribution value of the MMEG; $E_{\mathrm{ES},\mathrm{S}}^i$ and $E_{\mathrm{ES},\mathrm{B}}^i$ represent the value of energy provided to and obtained from the SES by ${\mathrm{MG}}_i$, respectively;$E_{\mathrm{ES},\mathrm{S}}$ and $E_{\mathrm{ES},\mathrm{B}}$ are the total value of energy provided to and obtained from the SES by the MMEG, respectively.

The difference between the new and old weights of MG can be expressed as follows (19):

$$\Delta R_i=R_i-{\displaystyle \frac{1}{N}}$$

(19)

For the sum of the differences of all MG weights, there is ${\displaystyle \sum _{i=1}^N\Delta R_i=0}$. Using the differences in weights to appropriately adjust the allocation results based on the traditional Shapley value method, the fine-tuning amount and the new allocation results are as follows:

$$\Delta {\phi }_i(\nu )=\Delta R_{i}v(s)$$

(20)

$${\phi }^{\ast }{}_i(\nu )={\phi }_i(\nu )-\Delta {\phi }_i(\nu )$$

(21)

In the formula, ${{\phi }^{*}}_i\left(\nu \right)$ represents costs of $\mathrm{M}{\mathrm{G}}_i$ saved after improving the allocation model.

5. Case Analysis

For three representative microgrids (MGA, MGB, MGC) situated within the same desert renewable energy base, this study establishes a planning and operational analysis framework for their jointly constructed and shared energy storage system. The region possesses abundant and highly complementary wind and solar resources, yet exhibits clear differences in load characteristics across the microgrids. To simplify the calculation, the whole year is divided into four seasonal scenarios, each covering 91 days. A typical day is selected as the representative day with a 24-hour scheduling horizon. The representative days were obtained using a K-means clustering algorithm applied to the full-year wind, solar and load data, ensuring adequate representation of seasonal variability.

The capacity configuration model contains both continuous and binary variables. The original formulation is a mixed-integer nonlinear programming (MINLP) problem due to nonlinear operating relationships and logical constraints. To enable efficient computation, the nonlinear expressions are linearized using the Big-M technique. After linearization, the problem becomes a mixed-integer linear programming (MILP) model that can be solved by commercial solvers. The Big-M constants are selected based on the maximum feasible ranges of equipment power, rated capacities and operational states. In this study, a unified Big-M value of 10,000,000 is used. Although this value is large, it remains sufficiently above all physical upper bounds in the model and therefore does not distort feasible operating regions. We verified that it does not introduce excessive relaxation or numerical instability within the MILP solution process.

The simulation is implemented on the MATLAB R2021a platform, where the model is constructed using YALMIP 2021 and solved with CPLEX 12.10. The convergence tolerance is set to 0.001. The model converges within approximately 3–5 s on a standard workstation, demonstrating both high computational efficiency and stable solution performance.

5.1. Basic Data

The three MGs showed significant differences in resource and load characteristics. MGA is configured with a large-capacity wind turbine photovoltaic, multi-power type; MGB is configured with small capacity photovoltaics and is a low-power type; MGC solar and wind capacity is basically matched to the load and is a flat power type. Given the pronounced diurnal and seasonal variations in wind and solar resources within desert regions, MG provides typical daily wind and solar output, base load curves, equipment technical parameters and energy pricing in Figure 3 and Table 1, Table 2, Table 3 and Table 4. Reference [23] contains information for time-of-use electricity tariffs, natural gas prices, electricity storage investment and operational costs, and equipment operating parameters; Reference [24] for hydrogen storage equipment investment and operational costs, and operating parameters; Reference [25] for microgrid equipment parameter settings.

It should be noted that this study does not model uncertainties in renewable generation, load demand or electricity prices, and does not incorporate transmission or distribution network power-flow constraints. The analysis, therefore, reflects system behaviour under idealized operating conditions. Although idealized, these assumptions do not change the main findings of the study. The derived results still reflect general operational patterns of multi-microgrid coordination and shared energy-storage interaction, and the comparative relationships observed across different scenarios remain valid under a wide range of practical conditions. All conclusions should be interpreted within the scope of these assumptions.

• Typical daily wind and solar output and base load curves

Figure 3. The wind and solar output and load curves of MMEG.

Figure 3. The wind and solar output and load curves of MMEG.

2.

Energy price data

The purchase price of natural gas is set at 2.3 CNY/m³ and the electricity price is set according to the time-of-use pricing system, as shown in Table 1.

Table 1. Energy price data table for each period.

Time Period	Valley Section	Flat Section	Peak Section
	0:00–8:00	12:00–17:00	8:00–11:00
		21:00–23:00	18:00–20:00
Time-of-use electricity price (CNY/kWh)	0.37	0.82	1.36

Table 1. Energy price data table for each period.

Time Period	Valley Section	Flat Section	Peak Section
	0:00–8:00	12:00–17:00	8:00–11:00
		21:00–23:00	18:00–20:00
Time-of-use electricity price (CNY/kWh)	0.37	0.82	1.36

3.

Economic and technical parameters of the equipment

Technical parameters of integrated energy microgrid equipment. The parameters of each MG equipment are shown in Table 2.

Table 2. Technical parameters of integrated energy microgrid equipment.

Equipment	Parameter	Values
GT	Rated power/kW	1500
	Power generation efficiency	0.35
	Unit maintenance cost/(CNY/kW)	0.06
	Thermoelectric ratio	0.9
HRS	Rated power/kW	1200
	Operational efficiency	0.8
	Unit maintenance cost/(CNY/kW)	0.02
EC	Coefficient of refrigeration	3
	Rated power/kW	800
	Unit maintenance cost/(CNY/kW)	1.1 × 10−2
AC	Coefficient of refrigeration	0.8
	Rated power/kW	400
	Unit maintenance cost/(CNY/kW)	1.6 × 10−2
GB	Rated power/kW	800
	Thermal efficiency	0.85
	Unit maintenance cost/(CNY/kW)	1.5 × 10−5
PV	Unit maintenance cost/(CNY/kW)	2.4 × 10−2
WT	Unit maintenance cost/(CNY/kW)	1.9 × 10−2

Table 2. Technical parameters of integrated energy microgrid equipment.

Equipment	Parameter	Values
GT	Rated power/kW	1500
	Power generation efficiency	0.35
	Unit maintenance cost/(CNY/kW)	0.06
	Thermoelectric ratio	0.9
HRS	Rated power/kW	1200
	Operational efficiency	0.8
	Unit maintenance cost/(CNY/kW)	0.02
EC	Coefficient of refrigeration	3
	Rated power/kW	800
	Unit maintenance cost/(CNY/kW)	1.1 × 10−2
AC	Coefficient of refrigeration	0.8
	Rated power/kW	400
	Unit maintenance cost/(CNY/kW)	1.6 × 10−2
GB	Rated power/kW	800
	Thermal efficiency	0.85
	Unit maintenance cost/(CNY/kW)	1.5 × 10−5
PV	Unit maintenance cost/(CNY/kW)	2.4 × 10−2
WT	Unit maintenance cost/(CNY/kW)	1.9 × 10−2

(1)

SES device technical parameters

The parameters of each SES device are shown in Table 3.

Table 3. Technical parameters of integrated energy microgrid equipment.

Equipment	Parameter	Values
ESS	Power cost/(CNY/kW)	1000
	Capacity cost/(CNY/kWh)	1200
	Unit maintenance cost/(CNY/kW)	1.8 × 10−4
	Service life/year	10
EL	Investment cost/(CNY/kW)	3200
	Unit maintenance cost/(CNY/kW)	1.4 × 10−2
	Service life/year	20
FC	Investment cost/(CNY/kW)	2200
	Unit maintenance cost/(CNY/kW)	1.0 × 10−2
	Service life/year	10
HST	Investment cost/(CNY/kW)	600
	Unit maintenance cost/(CNY/kW)	1.0 × 10−2
	Service life/year	10

Table 3. Technical parameters of integrated energy microgrid equipment.

Equipment	Parameter	Values
ESS	Power cost/(CNY/kW)	1000
	Capacity cost/(CNY/kWh)	1200
	Unit maintenance cost/(CNY/kW)	1.8 × 10−4
	Service life/year	10
EL	Investment cost/(CNY/kW)	3200
	Unit maintenance cost/(CNY/kW)	1.4 × 10−2
	Service life/year	20
FC	Investment cost/(CNY/kW)	2200
	Unit maintenance cost/(CNY/kW)	1.0 × 10−2
	Service life/year	10
HST	Investment cost/(CNY/kW)	600
	Unit maintenance cost/(CNY/kW)	1.0 × 10−2
	Service life/year	10

(2)

Conversion efficiency of energy storage equipment

The conversion efficiency of the energy storage equipment is shown in Table 4.

Table 4. The conversion efficiency of the energy storage equipment.

Parameter	Values
Hydrogen produced by the electrolytic cell per kWh/(g/kWh)	25.2
Electricity generated per unit mass of fuel cell/(kWh/g)	0.03175
The charging and discharging efficiency of electrical energy storage	0.95
The hydrogen inlet and outlet efficiency of the hydrogen storage tank	0.98
Electrolytic cell electrical efficiency	0.6
Fuel cell electrical efficiency	0.6
Upper and lower limits of hydrogen storage ratio in hydrogen storage tanks	0.1/0.9

Table 4. The conversion efficiency of the energy storage equipment.

Parameter	Values
Hydrogen produced by the electrolytic cell per kWh/(g/kWh)	25.2
Electricity generated per unit mass of fuel cell/(kWh/g)	0.03175
The charging and discharging efficiency of electrical energy storage	0.95
The hydrogen inlet and outlet efficiency of the hydrogen storage tank	0.98
Electrolytic cell electrical efficiency	0.6
Fuel cell electrical efficiency	0.6
Upper and lower limits of hydrogen storage ratio in hydrogen storage tanks	0.1/0.9

5.2. Analysis of Multi-Scenario Configuration Results

5.2.1. Base Case

To verify the applicability and economy of SES in the MMEG scenario, especially in the context of combined operation of wind, solar and energy storage in desert areas, four comparative scenarios were designed: Scenarios 1 and 2 were MG independently configured with electricity and electric–hydrogen energy storage, respectively. Scenarios 3 and 4 are MMEG combined with electricity and electric–hydrogen energy storage, as shown in

Table 5. Scene setting.

Scene	Whether to Configure Independent ESS	Whether to Configure Independent HESS	Whether to Configure Shared ESS	Whether to Configure Shared HESS
Scene 1	YES	NO	NO	NO
Scene 2	YES	YES	NO	NO
Scene 3	NO	NO	YES	NO
Scene 4	NO	NO	YES	YES

.

As shown in

Table 6. Configuration results and user costs.

Variable	Scene 1	Scene 2	Scene 3	Scene 4
ESS power/kW	8559.10	8013.54	4410.10	3925.56
ESS capacity/kWh	23,965.47	22,437.92	12,348.27	10,991.57
EL power/kW	—	1688.15	—	622.84
FC power/kW	—	635.55	—	369.16
HST power/kW	—	14,340.78	—	5493.42
Abandoned wind and solar power rate/%	12.24	0.37	0.60	0.05
ESS investment cost/thousand CNY	2488	2259	1282	1141
HESS investment cost/thousand CNY	—	1371	—	511
Operation and maintenance cost/thousand CNY	2515	2581	2521	2553
Electricity purchase cost/thousand CNY	5379	4741	4843	4586
Gas purchase cost/thousand CY	19,790	19,572	19,832	19,911
Power abandonment cost/thousand CNY	1820	56	368	8
Total cost/thousand NY	31,992	30,581	28,845	28,710

, Scenario 4 achieved overall optimization of cost and wind and solar consumption, with a total cost of CNY 2.87 million and an abandoned wind and solar power rate of 0.05%, representing a reduction of 10.26% and 12.19% compared to the baseline Scenario 1. The system performance before and after SES configuration was verified by comparing Scenarios 1 with 3 and 2 with 4. The cost of energy storage investment was reduced by 48.47% and 54.34%, respectively, and the rate of abandoned wind and solar power was reduced by 11.64% and 0.32%, respectively. By comparing Scenarios 1 and 2 and 3 and 4, the effectiveness of HESS over standalone energy storage can be verified, with total system costs reduced by CNY 1.410 million and CNY 0.134 million, respectively, and wind and solar power abandonment rates reduced by 11.87% and 0.55%, respectively. The total cost reduction trend and sub-item cost composition of each scenario are shown in Figure 4.

The introduction of hydrogen storage increases the initial investment cost, but complements electrical storage effectively, enhancing the overall economy by reducing the cost of power curtailment. In desert areas where wind and solar resources are abundant but prone to wind and solar power curtailment, specifically, compared with Scenario 1, the power and capacity of electric energy storage decreased by 6.4%, while the cost of curtailment was reduced from CNY 1.8198 million to CNY 55,600. In Scenario 4, the cost of power curtailment was reduced by CNY 360,000 compared to Scenario 3, and the rate of new energy consumption was significantly improved, verifying that electric–hydrogen hybrid energy storage effectively enhances the system’s regulation capacity and the rate of new energy consumption, especially suitable for solving the problem of new energy consumption in desert areas. In the co-construction and sharing mode, the power and capacity of electricity storage in Scenario 3 decreased by 48.5% compared to Scenario 1, corresponding to a reduction of CNY 1.2819 million in investment cost, and the cost of power curtailment dropped to CNY 367,600, demonstrating the economic benefits of ESS. Scenario 4 has the smallest configuration of electric energy storage, which is 54.1% lower than Scenario 1, while the investment cost of hydrogen energy storage is 62.7% lower than Scenario 2. This fully validates the technical synergy and economic advantages of HESS and provides an effective solution for the integrated development of wind, solar and energy storage in desert areas.

5.2.2. Sensitivity Analysis

The advancement of future energy storage technology can increase the lifespan of energy storage. To reflect the impact of this situation on the planning results, we have constructed four scenarios with a lifespan increase of 10–50%. We have calculated the economic benefits of shared HESS relative to the independent mode, and the results are shown in Figure 5:

As shown in Figure 5, the economic benefits of shared HESS have remained around 10.26% as shown in the base case, indicating that the improvement of energy storage lifespan does not affect the changes in economic benefits of shared HESS.

At the same time, in order to verify the impact of energy prices on energy storage planning and operation results, a price fluctuation scheme of −20% to 20% was also set up, and the economic benefits of shared HESS were calculated. The results are shown in Figure 6:

As shown in Figure 6, when energy prices increase, the economic benefits of shared HESS show a linear downward trend. This is because as energy prices increase, the advantages of shared HESS in purchasing electricity and gas decrease, resulting in a corresponding decrease in its economic benefits.

5.3. Analysis of Operation Results

5.3.1. Subsubsection Analysis of Operation Results

In co-built HESS mode, both MMEG energy management and operational strategies change compared to independent planning. Based on the coordinated scheduling and multi-energy complementarity of multiple types of energy storage, the economic efficiency and operational stability of MMEG are effectively enhanced. Taking into account the output characteristics of desert regions, the MMEG power balance is illustrated in Figure 7.

As can be seen from Figure 7, during the typical spring days of wind and solar power output surplus period (0:00–4:00) in desert areas, after the system meets the load demand, the remaining electricity is prioritized for storage in the electrical energy storage system to minimize power curtailment. In summer and winter, the typical daily load demand is high. The system selects to purchase electricity for storage during the low electricity price period and release it during the peak electricity price period to optimize the overall economic benefits. During the typical spring days when photovoltaic output is abundant (13:00–18:00), electricity storage cannot fully absorb the surplus electricity, and hydrogen storage systems are activated. Excess electricity is stored for hydrogen through electrolysis, providing a guarantee for energy distribution across seasons and enhancing the flexibility of the system’s energy supply. During the typical summer days when new energy output is insufficient (8:00–11:00, 18:00–21:00), hydrogen is converted and released through fuel cells to make up for the shortage of wind and solar output. The aforementioned operational mechanism reduces reliance on purchased electricity and energy procurement costs, enhances the MMEG’s autonomous regulation capabilities and renewable energy absorption rate, improves system stability and effectively addresses the spatio-temporal mismatch between energy supply and demand in desert regions.

5.3.2. Analysis of Shared Energy Storage Charging and Discharging Conditions

Within the integrated wind–solar–storage operational framework, electrical energy storage primarily undertakes short-term power allocation tasks within the MMEG system. Its charging and discharging behaviour exhibits regular cyclical patterns across typical daily cycles, effectively mitigating intraday localized supply–demand imbalances caused by fluctuating wind and solar output. Hydrogen storage, conversely, focuses on facilitating energy transfer across seasonal scales. It capitalizes on the pronounced seasonal variations in wind and solar resources within desert regions. During typical days in spring and autumn when wind resources are relatively abundant, hydrogen is concentrated for storage, converting surplus electricity into hydrogen energy. Conversely, during typical days in summer and winter when energy demand peaks, hydrogen is concentrated for release. This enables power generation via fuel cells or hydrogen gas turbines, thereby balancing system supply and demand. The charging and discharging power of the HESS and the SOC of the energy storage are shown in Figure 8.

As shown in Figure 8, the seasonal dispatch of hydrogen energy storage and the intraday regulation of electric energy storage complement each other, jointly constructing a “short-term–seasonal” two-level energy storage system to enhance the energy utilization efficiency of the system. Specifically, on typical days in spring and autumn when wind and solar power outputs are relatively abundant, hydrogen energy storage converts surplus electricity into hydrogen energy storage and releases it on typical days in summer and winter. This strategy, through the bidirectional conversion of electricity to hydrogen, enables two cross-seasonal storage and utilization of energy, demonstrating the ability of hydrogen in long-term storage and cross-seasonal dispatching, effectively reducing system energy costs and external dependence, and meeting the long-cycle energy management requirements of combined operation of wind, solar and storage in desert areas.

5.4. Cost-Saving Apportion Analysis

Although co-building SES can effectively reduce the overall cost of energy storage investment, a fair cost-sharing mechanism needs to be established, as it involves multiple participants. The mechanism should be based on the cost savings achieved by each entity through participation and sharing. This paper uses the improved Shapley value method for cost allocation and compares the results with those obtained by the conventional Shapley value method to verify the effectiveness of the proposed improved method. The cost savings were zero when MG was independently configured for energy storage, and in co-built HESS, all possible MG combinations achieved cost savings relative to the independent configuration, as shown in

Table 7. Cost savings resulting from MG combinations.

Serial Number	Configuration Form	Cost Savings/Thousand CNY
1	MGA independent configuration	0
2	MGB independent configuration	0
3	MGC independent configuration	0
4	MGA, B combined configuration	1074.7
5	MGA, C combined configuration	1662.5
6	MGB, C combined configuration	2647.3
7	MGA, B, C combined configuration	3282.5

.

As shown in Table 3, due to significant differences in resource endowment and load demand among MG, especially in desert areas, microgrids at different locations may have different access conditions for wind and solar resources and load characteristics, and different combinations may produce different power mutual aid effects. Under the combination of two MGS, the synergy between MGB and MGC was the most significant, saving up to CNY 2.6473 million. The maximum cost savings reached CNY 3.2825 million when all three types of MGS with strong complementarity—more, less and equal—were combined. The cost-sharing results based on the conventional Shapley value method are shown in

Table 8. Cost allocation results based on the conventional Shapley value method.

Allocate Costs	MGA	MGB	MGC
Independent operating cost/thousand CNY	8258.1	13,891.1	9842.8
Share the cost savings/thousand CNY	667.9	1160.4	1454.2
Allocate operating costs/thousand CNY	7590.2	12,730.7	8388.6

.

As shown in Table 8, MGC shares the most cost savings, while MGA shares the least. However, the conventional Shapley value method does not fully reflect the differences in the actual value contributions of each subject to the alliance. In this paper, the interaction value contribution degree is introduced on the basis of the conventional Shapley value method to calibrate the allocation weights. By calculating the power provided and obtained by MG to shared energy storage and the time-of-use electricity price at the corresponding time, the value of its electricity provided and the value of its electricity obtained were obtained, respectively. The energy interaction of each MG and the contribution of the interaction value calculated accordingly are shown in

Table 9. Energy interaction of each MG and contribution of interaction value.

Parameter	MGA	MGB	MGC
Supply electrical power to HESS/kW	40,835	15,724	68,331
Obtain the electrical power from HESS/kW	17,997	77,186	19,203
Electrical energy value provided/CNY	32,402	11,510	50,543
Value of electrical energy acquisition/CNY	20,057	80,730	22,971
$\Delta R_i$	0.01	0.04	−0.05
New contribution degree	0.34	0.38	0.28

.

MGC provides the highest electrical power and corresponding electrical value to HESS, while MGB acquires the highest electrical power and electrical value from HESS, highlighting the limitations of the conventional Shapley value method, which only considers marginal contribution and ignores the actual energy interaction value. Based on the interaction value contribution calculated by the improved Shapley value, the weights of cost savings allocated by each MG in the conventional Shapley value method were adjusted, with the weights of MGA and MGB increased by 0.01 and 0.04, respectively, and the weight of MGC decreased by 0.05. This weight adjustment aims to more accurately reflect the value creation of each MG in the interaction with HESS energy, so that MGA and MGB obtain a relatively larger share when sharing the cost savings. The cost-sharing results based on the improved Shapley value method are shown in Table 6. Figure 6 compares the operating costs of each MG under three modes: independent operation, allocation based on the conventional Shapley method and allocation based on the improved Shapley method.

As shown in

Table 10. The cost allocation result based on the improved Shapley value method.

Allocate Costs	MGA	MGB	MGC
Independent operating cost/thousand CNY	8258.1	13,891.1	9842.8
Improving the allocation saves costs/thousand CNY	706.1	1306.3	1270.1
Improve the allocation of operating costs/thousand CNY	7552.0	12,584.8	8572.7

and Figure 9, each MG has the highest operating costs when independently configured with energy storage, and participation in co-construction and co-sharing significantly reduces their respective operating costs. In addition, compared with the conventional Shapley value method for traditional allocation, the allocation results of the improved Shapley value method are more reasonable. Specifically, after the contribution recalibration, MGC savings were reduced by USD 184,100, while MGA and MGB savings were increased by USD 38,100 and USD 146,000, respectively. This adjustment direction corresponds to the assessment results of interaction value contribution, verifying that the proposed improved Shapley value method can more reasonably quantify the actual contribution differences of each MG in the alliance, thereby reasonably correcting the cost allocation results and promoting the fairness of benefit distribution.

6. Conclusions

This paper addresses the issue of joint planning and shared energy storage for multi-microgrid systems operating under coordinated wind–solar–storage conditions in desert regions. A microgrid model incorporating renewable energy generation systems, combined heat and power systems, and auxiliary equipment was constructed, proposing an optimised configuration method for hybrid electricity–hydrogen storage capacity. Four representative scenarios were established: standalone electrical energy storage, standalone electrical–hydrogen storage, electrochemical shared storage and hybrid electrical–hydrogen shared storage. Storage configuration schemes and system operational outcomes were obtained for each scenario. Subsequently, a cooperative benefit analysis employing an improved Shapley value method based on interactive value contribution was conducted to achieve equitable cost-saving allocation. Simulation validation was performed using the MATLAB R2021a platform, and the conclusions were as follows:

(1)

The proposed model differs from conventional hybrid storage approaches by integrating cross-microgrid coupling and an improved game-theoretic cost allocation mechanism, forming a unified optimization–allocation framework. The optimal configuration scheme of HESS in the co-construction and sharing mode was obtained. In the collaborative scenario, the capacity configuration scheme for the energy storage system that achieves global optimization is obtained: electrical energy storage power of 3925.56 kW and capacity of 10,991.57 kWh. The hydrogen energy storage system consists of 622.84 kW electrolyzers, 369.16 kW fuel cells and 5493.42 kW hydrogen storage tanks. This mode offers the best economic performance and the highest rate of new energy consumption.

(2)

The addition of hydrogen energy storage enables the energy storage system to have “short-term–seasonal” flexible dispatching capabilities. In the renewable energy bases of desert regions, the total system cost of electric–hydrogen hybrid energy storage is reduced by CNY 1.4103 million and CNY 135,500, respectively, in independent mode and co-construction and sharing mode, and the curtailment rate of wind and solar power is reduced by 96.98% and 91.67%, respectively. Verify that the long-term storage characteristics of hydrogen energy enhance the system’s economic efficiency, capacity regulation and green power direct connection effect.

(3)

Establishing SES can optimize the energy storage configuration and enhance the operational performance of MMEG. Within the integrated operation framework of wind, solar and storage, the co-construction and sharing model of energy storage configuration reduces the cost of energy storage equipment by 2.34% and 32.31%, respectively, compared with the independent model, and reduces the rate of abandoned wind and solar power by 95.12% and 86.39%, respectively, verifying the optimization effect of internal resource aggregation of MMEG on the efficiency of energy storage investment.

(4)

The improved Shapley method based on interaction value contribution effectively quantifies the energy interaction differences among MGS. The operating cost of MGC increases by CNY 184,100, while the operating costs of MGA and MGB decrease, achieving reasonable cost allocation and ensuring the fairness and stability of the cooperative alliance.

In addition, the proposed shared electricity–hydrogen storage framework is compatible with China’s current policy direction, which promotes shared energy storage, renewable-powered hydrogen production, and market-oriented mechanisms such as time-of-use pricing and emerging hydrogen trading pilots. By adjusting market parameters, the model can be readily applied under existing or evolving regulatory environments.

Although the proposed shared hybrid electricity–hydrogen energy storage optimization model demonstrates significant economic and operational advantages, several challenges remain in practical applications. These challenges include hydrogen leakage risk, electrolyzer and fuel-cell degradation, and the difficulty of maintaining real-time coordination between electricity and hydrogen subsystems under variable renewable conditions. Recent studies have shown that accounting for degradation mechanisms can significantly improve the realism and robustness of storage-planning models. For example, degradation-aware modelling of non-ideal battery behaviour, such as that presented in [26], provides a methodological reference for future work on integrating aging effects into coordinated ESS/HESS optimization. Future research will incorporate component degradation models into the planning framework, develop adaptive control and predictive-maintenance strategies, and further extend the model to regional multi-microgrid systems coupled with thermal and gas networks.