Evaluating microgrid benefits requires mapping stakeholder relations and viewing outcomes mainly from the operator’s perspective while accounting for external influences. Four actors are involved: the operator, external grid, users, and government. The operator finances planning, construction, and technology, earns supply revenue and subsidies, and raises project value by lowering costs and risks through technological optimization. The external grid exchanges power with the microgrid, improving reliability and reducing costs for both sides. Users pay for reliable electricity and, through feedback and flexible demand, help improve service quality; they may also obtain indirect qualitative benefits (e.g., improved service experience and local economic vitality). However, in this paper the quantified evaluation focuses on energy-saving/emission-reduction benefits and technical performance indicators. Government guides planning and operations via policies and subsidies, promoting renewables, reducing fossil dependence, and stimulating investment and tax revenue. Benefits are evaluated across economic (investment and finance), social (energy saving and emissions), and technical (design, coordination, efficiency) dimensions. The SD model captures causal links—operator actions → costs and reliability, grid exchanges → reliability and revenues, user feedback → service quality—to assess dynamic outcomes.

The microgrid benefit-evaluation system is a dynamic, multi-factor system with nonlinearity, high order, and multiple feedback loops. Due to this complexity, system dynamics (SD) is used to link factors and track the system’s evolution. Following the parsimony rule, only elements that generate observable behavior are retained, while weakly related elements are removed. The model focuses on the benefit-realization process during microgrid operation, capturing operational schemes, internal structures, and their connections to benefit dimensions. Simulation is used to observe changes and obtain evaluation results. The indicator set is organized into three dimensions: economic benefits (electricity sales revenue and storage arbitrage), social benefits (energy savings and emission reductions from clean generation), and technical benefits (improvements in power reliability and reductions in line loss). This structure forms a compact, mechanism-oriented SD model, suitable for causal-loop, stock-flow, and equation development.

The causal network—bounded by the model scope and drawn in standard SD notation—uses signed arrows and closed reinforcing/balancing loops; variables may appear in multiple loops. It comprises three coupled subsystems: economic, social, and technical. In the economic subsystem, two reinforcing loops drive growth: E1 (storage capacity → arbitrage profit → economic benefit → comprehensive benefit → investment → microgrid scale → storage capacity) and E2 (DG capacity → power supply → electricity-sales revenue → economic benefit → investment → microgrid scale → DG capacity). Increased investment expands DG and storage, lifting profits and momentum. In the Environmental (energy-saving/emission-reduction) subsystem, cleaner energy yields three reinforcing loops: DG → supply → fuel saving → energy benefit → investment → scale → DG; DG → supply → GHG reduction → environmental benefit → investment → scale → DG; DG → supply → pollutant reduction → environmental benefit → investment → scale → DG. In the technical subsystem, local generation and storage improve reliability and cut line losses, raising technical and overall benefits via two reinforcing loops centered on reliability gains and loss reduction. These feedbacks inform the stock–flow state/rate variables and highlight sensitivity levers—microgrid scale, utilization hours, price/subsidy signals for economics, and carbon/pollutant prices for social benefits—each tied to explicit causal paths.

Guided by the causal network and feedback loops, we replace the flowchart with a compact, text-based process linking three interacting subsystems—economic, social, and technical—over 2015–2030. Each year begins with exogenous drivers (utilization hours, prices, subsidies, emission factors) and updated stocks of distributed generation and storage. The economic stream computes direct economic benefit by tracking electricity-sales revenue and storage arbitrage, then deducting investment and O&M costs. The environmental stream maps clean generation to fuel savings and CO₂/SO₂/NO_x reductions, monetized via carbon prices and pollutant fees to yield energy-saving and emission-reduction benefits. The technical stream evaluates how local generation and storage shorten delivery paths and buffer variability, improving reliability and lowering line-loss rates; saved energy and avoided outage losses translate into technical benefits. “In this paper, the reliability benefit is quantified using an outage-duration-based valuation. Specifically, the model converts the reliability improvement into avoided outage duration (hours) and then monetizes it using the unit outage cost (RMB 100,000 per hour). The detailed calculation formula of the technical subsystem (including reliability and line-loss benefit equations) is provided in the Appendix.” These three streams aggregate into a comprehensive benefit index that guides next-period capacity expansion and operational adjustments, closing the annual loop. Model execution follows “validity check → scenario setting → result analysis,” producing baseline and sensitivity trajectories for subsequent benefit-sharing and incentive-contract design.

Given the large number of variables in the SD model, we focus on the key variables that are most relevant to benefit evolution, in order to reduce the influence of secondary factors. Variables are organized by subsystem (economic, environmental/social, and technical). Detailed variable definitions and units are provided in the Supplementary Material.

In the constructed system dynamics model, the relevant variable equations and parameter settings are as follows:

Economic subsystem. The formula can be found in the attached material. In the SD model, INTEG (·) denotes the stock (accumulation) operator, i.e., a state variable obtained from its initial value and the integral of its rate of change; WITH LOOKUP(·) is a table function that maps a variable to an exogenous driver (e.g., time), allowing it to vary over the simulation horizon. In the economic-benefit subsystem, the equations use the following parameters: G ₀ , C _0,0 , I _D,0 , and I _E,0 denote the initial values of annual electricity supply, annual O&M cost, investment in distributed generation, and investment in storage, respectively. r _ID and r _IE are response coefficients that translate the comprehensive-benefit signal into additional investment in distributed generation and storage. c _D and c _E are the unit capital costs (per kW) of distributed generation and storage. r _D and r _E measure the effect of distributed generation and storage on changes in O&M cost.

With these settings, the SD equations fully encode inter-variable links (no flowchart needed). We then apply the model to a representative microgrid to validate it and trace the dynamic evolution of economic, environmental, and technical benefits, providing evidence for designing benefit-realization mechanisms.

Before conducting numerical simulations, the effectiveness of the system dynamics model needs verification. This section tests the model’s effectiveness in two ways: model structure verification and behavior reproduction verification. In the structure verification, the model aims to represent the interaction pathways that generate microgrid benefits and to evaluate these benefits dynamically, the model aims to show the interaction paths of factors contributing to microgrid benefits and dynamically evaluate these benefits. The model focuses on indicators strongly linked to microgrid benefits, such as power supply, electricity revenue, costs, pollutant reduction, and reliability. To ensure clarity, processes like energy coupling and device scheduling are simplified to avoid unnecessary exogenous variables. Thus, the model boundaries are reasonable.For behavior reproduction verification, microgrid project 1 is analyzed, and key variables reflecting the microgrid’s operational status and multidimensional benefits are used to assess how well the model fits the trends of these variables. Using historical data from 2015 to 2020, the key variables include power supply, revenue, costs, annual coal reduction, and line loss reduction. These variables represent economic, social, and technical impacts. The actual and fitted values of all key variables are shown in Figure 2. The closeness of these values indicates that the model accurately reflects the microgrid’s operation and benefits, passing the behavior reproduction test (Sonnenberg et al., 2018).

This section continues with Microgrid Project 1 and uses the SD model to trace how factors interact in forming multi-dimensional benefits. In the constructed causal network, variables are linked through functional relations, so a shock to one variable propagates through the system. We take power supply as the central driver and treat the utilization hours of distributed units as an exogenous lever: it directly affects supply and, through the transmission paths, shapes the economic, social, and technical benefits. Accordingly, we design three comparison scenarios that vary utilization hours (see Supplementary Table S1) and observe the resulting system responses.

Based on the dynamic evaluation system dynamics model of the multidimensional benefits of microgrids, relevant indicators are selected for analysis from two aspects: the operating status of microgrids and the multidimensional benefits of microgrids. Among them, the operational indicators of microgrids include six indicators: power supply, annual operation and maintenance costs, distributed power investment, energy storage investment, annual standard coal savings, and reduction of line losses. The multidimensional benefit indicators of microgrids include eight indicators: direct economic benefits, emission reduction benefits, energy conservation benefits, power reliability benefits, reduction of line losses benefits, and comprehensive benefits.

The simulation results of the operation status of microgrids in different scenarios are shown in Figure 3. Figure 3 shows that utilization hours substantially affect microgrid operation. Higher utilization increases electricity output, investment in distributed generation and storage, standard-coal saving, and line-loss reduction, while the impact on O&M cost is relatively limited. Over the long term, power supply and investment exhibit a steady upward trend across scenarios, indicating continuous scale expansion. Although O&M costs increase over time, the overall economic, environmental/social, and technical performance improves, supporting sustained microgrid development.

The simulation results of multidimensional benefits of microgrids in different scenarios are shown in Figure 4, which shows that: The direct economic benefits of microgrids continue to decline in different scenarios, due to the influence of electricity sales revenue, operation and maintenance costs, and investment costs on the direct economic benefits of microgrids. As the scale of microgrids expands, the power supply of the system continues to increase (see Figure 5), resulting in a continuous increase in electricity sales revenue. However, at the same time, the expansion of the system scale will inevitably lead to an increase in investment and operation costs. Simulation results show that the increase in investment and operation costs is higher than the increase in electricity sales revenue, resulting in a decrease in the direct economic benefits of the system. The emission reduction and energy-saving benefits of the system continue to increase in different scenarios, due to the expansion of the system scale and the increase in clean energy supply. On the one hand, the equivalent saved coal increases, resulting in an increase in the energy-saving benefits of the system. On the other hand, the carbon and pollutant emissions of the system decrease, resulting in an improvement in emission reduction benefits. With the increase of power generation utilization hours, the emission reduction and energy-saving benefits of the system also increase.

Across the three utilization-hour scenarios, the system’s line-loss-reduction capability shows a clear improving trend, whereas the reliability-related benefit is less evident. With microgrid expansion, more distributed resources are integrated and electricity is consumed closer to where it is generated, which shortens delivery paths and reduces losses. By contrast, the reliability evaluation in this study is expressed using the reduced outage duration (Tf) and is monetized by the loss per unit outage duration (Cf) to obtain the power reliability benefit (Bf) (see Supplementary Text for variable definitions). Therefore, the reliability-related benefit does not present a monotonic increase across scenarios, and Case 3 can be lower than Case 1 under the evaluated operating conditions and parameter settings. Nevertheless, from the perspective of overall system performance, the integrated benefits of microgrid development continue to grow across scenarios, which supports further expansion through the reinforcing feedback mechanism captured by the SD model.

Overall, the comprehensive benefits of microgrids are increasing year by year, mainly due to the continuous growth of their economic and social benefits. Based on the above analysis, the benefits generated by microgrids during operation are multidimensional and comprehensive, and their benefits are dynamically changing over time. The dynamic evaluation of microgrid benefits using system dynamics methods can effectively reflect the time-varying and sustainable characteristics of microgrid benefits, providing useful reference for future research on the implementation mechanism of multidimensional benefits of microgrids.

In the new power system, a microgrid operates as a cooperative alliance—integrating micro gas turbines, storage, and demand-side users—and participates in the market as a single entity. Fair, transparent sharing of alliance gains is therefore essential. Among cooperative-game rules (e.g., the kernel and the Shapley value), we adopt the Shapley value as the core rule for redistribution (Guo et al., 2021). It allocates by each member’s marginal contribution and follows a clear, operational procedure that supports perceived fairness (Ross et al., 2015). To reflect real-world heterogeneity, we extend the classical Shapley rule with weighted correction factors embedded in the marginal-contribution calculation (Yadav et al., 2018). In practice, participants contribute unevenly in capital at risk, operational flexibility, energy output, and stability support; a uniform Shapley split may miss these differences. Our improved scheme quantifies three core dimensions and converts them to stakeholder weights: (1) investment proportion (capital contribution), (2) energy supply or demand level (measured output/consumption supporting system operation), and (3) provision of ancillary services (e.g., verified demand response, spinning reserve). After normalization, these weights adjust each coalition member’s marginal contribution, yielding shares that reflect both theoretical coalition value and audited operational input. The weights can be recalibrated over time using metered indicators—such as dispatched energy, reserve-provision hours, or validated DR events—so the allocation adapts as configurations, load profiles, or market conditions evolve (Chen et al., 2021). This dynamic, weighted Shapley approach preserves transparency and coalition stability, enhances incentive alignment, and fits the decentralized, changing architecture of modern microgrids.

In new-type power systems, microgrids generate private economic value and positive externalities—emission reduction, peak shaving, and stronger grid resilience—that markets rarely price well. To monetize these benefits, we adopt a principal–agent framework in which regulators (principals) aim to maximize social welfare and microgrid operators (agents) maximize profit under information asymmetry. Two classic frictions follow: hidden information about the true cost of providing external services and hidden action leading to under-delivery. The solution is a compensation contract that satisfies incentive compatibility and individual rationality. Payments are tied to verifiable performance indicators—e.g., measured peak-shaving capacity delivered during critical hours, renewable absorption (curtailment avoided), or certified emissions abated—so operators truthfully reveal costs and supply the targeted services. Contract design can include auditing, baseline setting, and clawbacks to limit gaming, plus multi-tier rates that reflect service quality and temporal scarcity. Empirical work suggests such mechanisms increase operator participation in grid-supportive services (Guo et al., 2020). To balance efficiency and fairness, terms can be negotiated using game-theoretic tools such as Nash bargaining or contribution-based allocation rules that credit parties according to marginal impact. In sum, performance-based cost compensation aligns private incentives with social goals, internalizes microgrid externalities, and supports the financial sustainability of all participants.

To address the challenges of incentivizing microgrid operators to provide external benefits (e.g., emission reduction, grid stability, demand response) under the new power system, this section introduces a compensation mechanism based on principal-agent theory. The goal is to construct a contractual model that satisfies both incentive compatibility and individual rationality, accounting for asymmetric information and risk-sharing between the regulator (principal) and the microgrid operator (agent). (1) Basic Modeling of External Benefit Compensation.

In the principal-agent framework, the regulator acts as the principal aiming to maximize social welfare, while the microgrid operator (agent) pursues private profit. The regulator cannot fully observe the agent’s effort or cost, creating hidden action and hidden information problems. To construct a compensation model, we define the expected utility of the principal as: E u π = u E π (1) E π = 1 ‐ b E y ‐ s = 1 ‐ b k a + e + b y 0 ‐ s (2) max ⁡ E π = max 1 ‐ b k a + b y 0 ‐ s (3)

Intuition: Equations 1–3 define the regulator’s objective and the contract structure. In essence, the principal seeks to maximize the expected net social value generated by external services delivered by the microgrid (e.g., emission reduction or reliability support), minus the compensation paid to the operator. The contract contains a fixed payment and a performance-linked component, so that higher verified external-benefit output leads to higher compensation. This structure allows the principal to share risk with the agent while still providing incentives for service delivery.

Based on the above analysis, a basic principal-agent model is established to design an external benefit compensation scheme under asymmetric information. The goal is to maximize the expected utility of the regulator (principal), while ensuring both incentive compatibility (IC) and individual rationality (IR) for the microgrid operator (agent). In this paper, Equations 4–6 are used to characterize the dynamic interactions among the economic, social, and technical subsystems of the microgrid. Equations 4–6 constitute the core of the principal–agent model, where the regulator maximizes expected utility subject to incentive compatibility (IC) and individual rationality (IR) constraints. max   1   ‐   b ka + by 0   ‐   s (4) s . t .   max s + b k a ‐ b y 0 ‐ α ‐ β a 2 2 ‐ V a ‐ 1 2 r b 2 σ 2   IC (5) s + b k a ‐ b y 0 ‐ α ‐ β a 2 2 ‐ V a ‐ 1 2 r b 2 σ 2 ≥ ω ¯   IR (6)

Why IC and IR are needed: Under asymmetric information, the regulator cannot directly observe the operator’s true cost/effort when providing external-benefit services. The incentive compatibility (IC) constraint ensures that the operator’s optimal choice is to exert the intended effort and report truthfully under the designed payment rule, rather than deviating to a lower-effort strategy. The individual rationality (IR) constraint ensures participation: the operator will accept the contract only if the expected utility under the contract is no lower than its reservation utility. Together, IC and IR guarantee both feasibility (participation) and effectiveness (incentives) of the compensation mechanism.

To derive the optimal solution to the above principal-agent model, the regulator must simultaneously satisfy the incentive compatibility (IC) and individual rationality (IR) constraints, while maximizing their own expected utility. This forms a typical constrained nonlinear optimization problem. Under the assumptions of risk-averse agents and observable output, the first-order condition (FOC) can be used to solve for the optimal incentive coefficient b^∗, compensation s^∗, and effort level a^∗. From the Kuhn-Tucker conditions, it can be inferred that when the IR constraint is binding, i.e., the agent’s utility is exactly equal to the reservation utility ω ¯ , the optimal contract satisfies Equation 7: S ※ + b ※ k a ※ ‐ b y 0 ‐ α ‐ β a ※ 2 2 ‐ V a ※ ‐ 1 2 r b ※ 2 σ 2 = ω ¯ (7)

Solving this condition jointly with the first-order conditions with respect to the incentive coefficient and effort level yields the optimal contract parameters. To solve the above constrained optimization, we use first-order conditions together with the IC and IR constraints. The resulting optimal contract can be expressed in terms of three key outputs: the incentive coefficient (how strongly compensation responds to verified external-benefit output), the fixed payment (ensuring participation), and the agent’s effort level (which determines the delivered external benefit). When the IR constraint is binding, the fixed payment adjusts so that the agent’s expected utility exactly equals its reservation utility; the incentive coefficient then governs the trade-off between stronger incentives (higher effort and output) and greater risk borne by the agent.

The optimal expected return difference for power regulatory agencies under information symmetry and information asymmetry is Equation 8: Δ E π = E π ※ ‐ E π ※ ※ = k a ※ ‐ a ※ ※ ‐ C a ※ ‐ C a ※ ※ + R (8)

This section mainly studies how microgrid operators can reasonably allocate direct benefits to relevant entities such as energy hub operators, renewable energy operators, and energy users within the system during operation, in order to motivate all parties to actively participate in microgrid operation alliances and pursue “excess returns”. The calculation results show that the direct benefit generated by microgrids in the previous text is 534500 yuan, which is the benefit of cooperation among energy hub operators, renewable energy operators, and energy users. To achieve fair and efficient distribution of economic benefits generated through collaboration, this study adopts the classical Shapley value method, which allocates benefits according to the marginal contributions of each participant to the total coalition value (Zhu 2021).

Using the revenues in Supplementary Table S2, we apply the Shapley value to allocate benefits to the energy hub operator, renewable operator, and users. The formula is: ω 1 = 1 6 × 2 × 27.34 + 2.6 ‐ 14.08 + 30.61 ‐ ‐ 8.49 + 2 × 53.45 ‐ 23.64 = 3.22 (9) ω 2 = 1 6 × 2 × 14.08 + 2.6 ‐ 27.34 + 23.64 ‐ ‐ 8.49 + 2 × 53.45 ‐ 30.61 = 6.36 (10) ω 3 = 1 6 × 2 × ‐ 8.49 + 11.76 ‐ 27.34 + 23.64 ‐ 14.08 + 2 × 53.45 ‐ 44.02 = 10.94 (11)

According to Equations 9–11, under the traditional Shapley value calculation, energy hub operators, renewable energy operators, and energy users should receive revenue shares of 32200 yuan, 63600 yuan, and 109400 yuan, respectively. The final allocation result can be obtained: After the correction, the revenue that can be allocated to energy hub operators, renewable energy operators, and energy users in microgrids is different. Specifically, after revision, the revenue of energy hub operators, renewable energy operators, and energy users is 63500 yuan, 84800 yuan, and 56900 yuan, respectively. Compared with the original Shapley value calculation results, it can be found that energy hub operators and renewable energy operators have received more allocation amounts. This is because in actual grid operation, their actual role is greater, so their weight is also higher, resulting in more allocation.

On the basis of calculating the external benefits (emission reduction benefits, energy conservation benefits, power reliability benefits, and line loss reduction benefits) that microgrid operators (entity A) bring to stakeholders such as grid operators (entity B), power generation companies (entity C), and society (entity D) during their operation, these external benefits can be further redistributed to ensure that some of the external benefits obtained by grid operators, power generation companies, and other stakeholders from microgrid operators are returned, in order to accelerate the cost recovery of microgrid operators. Based on this, combined with the measurement results of external benefits (emission reduction benefits, energy conservation benefits, power reliability benefits, and line loss reduction benefits) mentioned earlier, the distribution of these benefits among microgrid operators and various stakeholders is carried out.

To effectively internalize the external benefits generated by microgrid participants (such as emission reduction, energy conservation, voltage support, and grid stability enhancement), and to align the interests of regulators and market entities, this section applies principal-agent theory to design incentive contracts between the grid supervision agency and various stakeholders (grid operators, power generation companies, and society). The goal is to determine optimal contract parameters that not only satisfy rational behavioral assumptions (such as risk aversion and incentive compatibility) but also promote the realization of external value within the microgrid ecosystem. (1) Basic Parameter Settings The external stakeholders are assumed to be risk-averse utility maximizers. The regulator acts as the principal, while each participant (grid operators, power generators, society) acts as an agent. According to the Supplementary Tables S3-1, parameters including marginal contribution coefficient k, risk coefficient σ 2 , risk aversion coefficient r, output cost sensitivity β , and target revenue ω ¯ are configured as follows: (2) Optimal Contract Design Using the basic incentive contract model, and incorporating the above parameters, the incentive coefficients a i * and fixed payments s i * for each stakeholder group are determined. For example, for the grid operator: ω i g y ˉ i = 9.7 + 0.9996 × y ˉ i ‐ 113.5

These results indicate that the grid operator receives the highest fixed incentive due to its higher target revenue and marginal contribution coefficient. In contrast, the societal agent receives lower compensation due to its limited quantifiable output despite its positive external impact. According to the Supplementary Tables S3-2, the contract design achieves an appropriate balance between risk-sharing and incentive compatibility. Compared to the power generation company and society, the grid operator receives stronger incentives due to its larger marginal impact and risk-bearing capability. This differentiation enhances the effectiveness of external benefit realization and improves the overall efficiency of microgrid operation.

The development of microgrids in China is gaining traction under the backdrop of the national “dual carbon” goal, aiming to peak carbon emissions and achieve carbon neutrality. Microgrids, as a new model of distributed energy supply and consumption, are increasingly seen as an effective supplement to traditional centralized energy systems, especially in enhancing energy efficiency, resilience, and environmental sustainability (Liu and Lo 2022; Xu and Lu 2019). However, large-scale adoption of microgrids still faces multiple challenges. Policy uncertainty, technical immaturity in some cases, and financing difficulties have hindered the progress of demonstration projects, despite initiatives launched since 2015. To advance microgrids, the central government should define their role in the energy transition and align planning with renewable integration and carbon goals, with clear rules and incentives for interconnection, operation, and trading (Haoming 2011). Deepen market reforms for distributed trading and tariffs—use pilot zones to test peak shaving, frequency support, and ancillary-service payments, plus competitive bidding and time-of-use pricing. Establish cross-ministerial coordination to cut red tape. Provide stronger finance, prioritizing rural/energy-poverty projects. Allow local flexibility in subsidies, access standards, and planning integration.