Review of Virtual Power Plant Response Capability Assessment and Optimization Dispatch

Abstract

Functioning as a smart aggregation entity that combines distributed energy resources, energy storage systems, and flexible loads, virtual power plants (VPPs) serve as a pivotal technology in advancing the decarbonization and flexibility enhancement of modern power systems. Initially, we summarize the developmental context, evolutionary trajectory, and conceptual framework of VPPs. The architecture is functionally partitioned into three tiers: the aggregation layer, communication layer, and dispatch optimization layer (central layer). The dispatch optimization layer of VPPs serves as the “intelligent brain” connecting physical resources with electricity markets, whose core lies in achieving “controllable, adjustable, and optimizable” distributed resources through algorithmic and data-driven approaches, driving the energy system transition towards low-carbon, flexible, and efficient directions. Next, we critically examine core technologies in the dispatch optimization layer, particularly the response capacity assessment and optimal resource scheduling. Its content mainly focuses on the latest research on the aggregated resource response capability evaluation, virtual power plant dispatching optimization models, and dispatching strategies. Conclusively, we analyze prevailing technical bottlenecks and summarize significant advancements, concluding with prospective insights into future research frontiers and developmental priorities for VPPs. In the future energy system transition, VPPs will play an increasingly important role. It is foreseeable that the utilization efficiency of renewable energy will be significantly enhanced, and the energy market will become more diverse and vibrant. We look forward to VPPs integrating more quickly and effectively into daily life, transforming lifestyles and helping people collectively step into a low-carbon, green future.

1. Introduction

The essence of the global energy transformation is rooted in diminishing the reliance on fossil fuels and advancing sustainable low-carbon development. Since the onset of industrialization, the utilization of energy resources has been fundamentally coupled with societal productivity growth, with fossil fuels sustaining contemporary industrial infrastructure at the expense of escalating climatic risks. As documented by the Intergovernmental Panel on Climate Change (IPCC), the 2011–2020 global mean surface temperature exhibited a 1.09 °C rise relative to preindustrial baselines, of which 1.07 °C stemmed from anthropogenic drivers, predominantly fossil fuel utilization [1]. Combating climate change has evolved into a global imperative. The UN 2030 Agenda for Sustainable Development formally designates “affordable and clean energy” as critical targets, thereby positioning renewable energy technologies as pivotal enablers of the fourth energy revolution.

Technological advancements and declining costs in renewable energy have catalyzed the transformation of energy infrastructure. Renewable energy technologies, notably solar PV and wind power, have achieved a remarkable cost competitiveness. Between 2009 and 2023, the global LCOE for solar PV experienced an 83.29% reduction, with China contributing 40% of worldwide renewable installations, attaining a cumulative capacity of 1516 GW by 2023. Moreover, increasing computational demands have increased energy consumption, with the projected global computing device electricity usage reaching 44.8–67.2 trillion kWh annually by 2030, necessitating accelerated decarbonization and efficiency enhancements in energy systems [2]. This evolving landscape has fostered novel energy paradigms through proliferating distributed energy resources (solar PV, energy storage, and EVs), exposing conventional centralized grids to operational constraints such as flexibility deficits and inadequate peak regulation capabilities, thereby driving system resilience improvements via technological convergence.

VPPs are gaining critical importance in modern energy systems. The grid stability requirements are escalating due to the decentralized characteristics and inherent volatility of proliferating distributed energy deployments. By intelligently aggregating distributed generation, storage systems, flexible loads, and EV clusters, VPPs achieve optimized resource orchestration and synergistic operations. Under dual policy–market dynamics, renewable energy emerges as a strategic enabler of next-generation productivity paradigms. Through initiatives such as the Integrated Energy Development Acceleration Program, China achieved 9.495 million NEV sales (65% global market share) and 2.525 million charging stations in 2023, establishing comprehensive industrial ecosystems spanning from raw materials to service networks. According to IEA projections, the renewable generation capacity is expected to exceed coal-based power in early 2025, signifying a paradigm shift from centralized transmission architectures toward a hybrid distributed autonomy with grid synergies [3]. By functioning as critical grid–DER interfaces, VPPs optimize energy distribution efficiency, increase renewable penetration levels, and enable multistakeholder participation in evolving power markets.

While China maintains global leadership in renewable energy deployment, critical challenges remain regarding spatial resource disparity and systemic flexibility deficits in the power infrastructure. By aggregating geographically dispersed resources, VPPs significantly improve grid regulation capacities and renewable integration efficiency, establishing themselves as pivotal technologies for next-generation energy architectures. Data from the China VPP Industry Development White Paper (2023) indicate that the nation’s aggregated VPP capacity reached 3.7 GW in 2022 (17.5% global proportion), with forecasts suggesting a 39 GW deployment by 2025. This technological progression not only embodies innovation breakthroughs, but also, more critically, constitutes a strategic mechanism for realizing carbon peaking/neutrality targets and fostering advanced productivity paradigms.

The convergence of IoT (Internet of Things), artificial intelligence, and blockchain technologies positions VPPs as pivotal enablers in the smart energy system evolution and energy market transformation. Blockchain’s inherent features of decentralization, immutability, and transparency offer groundbreaking solutions for ensuring data integrity and enabling cross-stakeholder coordination in VPP ecosystems. The operational paradigm of VPPs necessitates real-time data interoperability among distributed energy resources, storage systems, and demand-side loads, whereas conventional centralized architectures remain vulnerable to single-point failures and malicious data manipulation. By implementing distributed ledger technology (DLT), blockchain guarantees the cryptographic verification of data provenance and maintains immutable audit trails throughout data lifecycle management. Advanced cryptographic mechanisms in the blockchain architecture, such as zero-knowledge proofs, enable privacy-preserving data transparency through selective information disclosure, significantly improving consumer engagement in demand response programs. The Enerchain initiative in Germany has demonstrated blockchain’s capacity to facilitate cross-border power transactions among heterogeneous stakeholders, establishing a scalable technical blueprint for VPP-based energy markets. Artificial intelligence empowers VPP operations through sophisticated algorithmic frameworks encompassing machine learning, deep neural networks, and reinforcement learning, revolutionizing resource forecasting and dispatch optimization capabilities [4,5]. For renewable energy integration, AI-driven predictive analytics leverage multisource data (meteorological patterns, historical generation profiles, and equipment telemetry) to establish high-fidelity forecasting models that capture the stochastic nature of solar/wind generation, thereby increasing the prediction accuracy for day-ahead market operations [6]. At the demand-side management layer, AI-enabled load forecasting synthesizes consumption behavior analytics, seasonal variability patterns, and dynamic pricing signals to generate temporally granular load predictions, enabling an optimized demand-response alignment. AI-driven optimization transcends conventional mathematical programming constraints through self-adaptive learning architectures that continuously refine dispatch strategies in response to renewable generation volatility, stochastic load patterns, and dynamic market conditions. Reinforcement learning paradigms exemplify this capability by autonomously coordinating storage system operations, a flexible load management, and a distributed generation dispatch through the real-time assimilation of grid-state measurements and market signals, thereby achieving Pareto-optimal solutions that balance economic efficiency, system reliability, and carbon reduction. The integration of digital twins with AI-powered simulation engines enables the creation of physics-informed virtual testbeds for VPPs, facilitating a scenario-based contingency analysis and operational strategy validation under complex grid conditions, thereby substantially improving the system adaptability and operational resilience. Emerging technologies are transforming VPP operations by enhancing security, efficiency, and responsiveness [7]. Blockchain ensures tamper-proof, decentralized energy transactions with smart contracts and privacy-preserving tools like zero-knowledge proofs, as seen in Germany’s Enerchain and California’s LO3 Energy. AI-driven optimization, using techniques like reinforcement and federated learning, improves forecasting accuracy and cybersecurity through real-time anomaly detection and digital twins, demonstrated by Google DeepMind and Japan’s JERA VPP. IoT–edge computing enables sub-second control, with 5G-enabled devices and local processing achieving a 50 ms latency in Spain’s Naturgy and securing data via encryption and TPM chips. Synergistically, platforms like Australia’s Power Ledger combine blockchain for settlements, AI for dynamic dispatch, and IoT for real-time monitoring within a 200 ms latency. While scalability and interoperability challenges persist, hybrid models—such as Ethereum PoS with DAG-based IoT layers—offer promising solutions. The ongoing advancement of foundation models in AI promises to streamline dispatch decision-making workflows and human-machine interactions, reducing operational overheads and accelerating the maturation of VPP technologies from pilot demonstrations to grid-scale deployments [7].

Recent reviews on VPPs, such as Gao et al. (2024) [8] and Naval et al. (2021) [9], have centered on market mechanisms or technical structures. However, few studies comprehensively explore the interaction between the response capability assessment and dispatch optimization under dynamic market conditions. This paper addresses this gap by presenting a “capability cognition–action decision” closed-loop framework in Section 4, which clarifies the connection between the two and highlights the latest research findings in both areas. Our analysis of several global projects, such as the EU FENIX, Tesla–SCE VPP, and China’s policy-driven pilot projects, offers practical insights for both academia and industry, particularly in balancing economic incentives with carbon neutrality goals. We think VPPs may not perfectly solve all issues from the new energy grid connection, but their development will be an indispensable technical supplement for the new energy transition.

2. Development of Virtual Power Plants

2.1. Concept and Development of VPPs

2.1.1. Definition and Origin

Originally conceptualized by Dr. Shimon Awerbuch in his seminal 1997 work The Virtual Utility: Description, Technology and Competitive Considerations, the VPP paradigm establishes a market-oriented coordination framework that virtually aggregates geographically dispersed energy assets into a unified dispatchable entity, delivering grid services without requiring physical infrastructure ownership [10]. This revolutionary approach transcends the conventional generation-consumption dichotomy in power systems, leveraging an advanced ICT infrastructure and cyber–physical coordination protocols to holistically manage heterogeneous resources, including DERs, storage systems, demand-side flexibility, and EV fleets, thereby enabling coordinated participation in energy markets and grid operations for enhanced resource efficiency and socioeconomic value creation [11].

The FENIX initiative [12] in Europe established formal VPP specifications requiring the following:

(1)

Capacity-agnostic DER aggregation;

(2)

Operational parameter characterization with a network impact assessment;

(3)

The formation of market-qualified flexible portfolios capable of delivering ancillary services to transmission system operators.

The US approach emphasizes demand-side resource orchestration, where VPP architectures predominantly aggregate flexible loads through dynamic pricing signals and automated control systems to provide peak load management and frequency regulation services [13]. While the implementation paradigms differ, all VPP frameworks fundamentally rely on cyber–physical coordination frameworks that enable (1) secure data exchange through advanced communication protocols and (2) algorithmic resource orchestration via distributed control architectures for the optimized coordination of heterogeneous assets.

2.1.2. Development Process

As a key technology for integrating distributed energy resources and enhancing power system flexibility, the development of VPPs is closely intertwined with the global energy transition, electricity market reforms, and digital innovation. From their conceptual emergence in the late 20th century to current large-scale commercialization, domestic and international landmark projects have continuously broken through technical barriers and explored business models, forming distinct development pathways.

• Global implementation pathways: pilot to commercial scaling

Europe: VPP research began in the early 21st century. The EU FENIX project (2005–2009) was an early prototype aiming to achieve a reliable DG grid integration and electricity market operation. The project defined VPPs as “aggregations of multitype DERs”, focusing on addressing challenges of a small DG capacity, dispersed distribution, and strong volatility. Germany’s ProViPP pilot [14] (2008) demonstrated VPP effectiveness in terms of grid frequency and voltage control by coordinating combined heat and power (CHP) units. The Dutch “PowerMatcher” project [15] utilized Universal Mobile Telecommunications System (UMTS) technology for a real-time DER dispatch, becoming a classic demand response case [16].

North American context: The U.S. VPP development centers on demand response strategies. The 2009 Auto-DR initiative [17] in California aggregated commercial/industrial loads to create a 1GW dispatchable VPP, demonstrating the economic viability of the load control in peak demand management. The 2016 Tesla–SCE partnership deployed the “Powerwall+Solar” VPP, integrating residential PV systems with battery storage to provide frequency regulation services, signifying the expansion of VPP applications to residential end-users.

2.

Chinese implementation: policy incentives and technical innovation

While China’s VPP development began relatively late, its recent progress has been rapid. The 2010 “Energy Efficiency Power Plant” initiative by the State Grid Corporation established demand-side management as a functional equivalent to conventional plants, representing the conceptual foundation of Chinese VPPs [18]. The 2017 Jibei pilot marked China’s first integration of wind/solar generation, energy storage, and industrial loads (650 MW total capacity). Implementing time-of-use pricing with capacity compensation mechanisms reduced the energy curtailment from 12% to 5% through production process optimization, while revealing industrial load response limitations that were later addressed through phased start-stop control strategies. Following the 2021 dual carbon policy, Shenzhen deployed China’s inaugural city-scale VPP platform, which integrates 290 MW of 5G base stations, EV chargers, and commercial cooling systems. While achieving a 180 MW peak shaving during summer, operational challenges emerged from the 5G backup battery management, which was resolved through SOC thresholds (>40%) prioritizing nonessential loads. The initial residential participation remained below 30% due to incentive gaps and awareness deficiencies. The implementation of a VPP credit system—offering bill discounts and carbon credits for demand response, coupled with real-time emission tracking via mobile apps—increases engagement, generating over CNY 20 million in annual operator revenue. The 2023 International Digital Energy Expo showcased Shenzhen’s VPP 2.0, which was built on a city-wide autonomous spatiotemporal modelling platform. Capable of managing 10 million device connections with a millisecond-level responsiveness, it currently integrates 310,000 charging points, 5100 5G storage nodes, 6000 e-bike charging units, 1200 solar installations, and multiple innovative stations (49 battery swaps, 144 V2G nodes, and 487 superchargers), constituting China’s most extensive distributed energy management system in its scale, diversity, and geographical coverage. Chinese VPPs currently remain in early development phases, constrained by a limited user engagement and homogeneous business models, with most implementations still at the pilot demonstration stage and lacking operational maturity [19].

The specific scales of the aforementioned virtual power plant projects are presented in the table (

Table 1. Global VPP applications: technical specifications and outcomes.

Project	Region	Capacity	Key Tech	Main Outcome
FENIX	Spain and UK	1.2 GVA DER	Layered control, IEC 104/OPC	Validated large-scale DER aggregation
ProViPP	Germany	8.6 MW hydro	Siemens DEMS, GPRS link	Provided frequency and peak services
Auto-DR	California, USA	~7.7 GW potential	Automated DR, AMI	Saved USD 94 M/yr; cut peak by 2.3%
Powerwall + Solar	Australia and USA	~500 MW dispatchable	Integrated PV + storage, Tesla VPP	Enabled household peak shaving
Shenzhen VPP 2.0	Shenzhen, China	310,000+ devices	5G-IoT, digital twin, blockchain	180 MW peak shaving, CNY 20M annual revenue

) below.

3.

Comparison and integration

Table 2. Comparative analysis of international and Chinese VPPs.

Dimensions	Global Implementations	Chinese Practice
Resource Aggregation Priorities	Residential energy storage and transnational flexibility resources (EU) Storage–demand response integration (North America)	Dominated by industrial adjustable loads, accelerated integration of new loads (5G, EV)
Technical Challenges	DER communication protocol harmonization (EU) End-user device response latency (North America)	Heavy industry load flexibility retrofitting Cross-stakeholder data interoperability frameworks
Policy Frameworks	Transnational regulatory alignment (EU) Competitive market mechanisms (North America)	National strategic planning with provincial pilots Hybrid incentive–mandate storage policies
Business Models	Market-driven revenue models (EU/NA)	Security-oriented service models transitioning towards market-based mechanisms

provides a systematic multidimensional comparison delineating the contextual environments and developmental distinctions between international and Chinese VPP ecosystems.

The global development trajectories exhibit distinct paradigms: market-driven mechanisms dominate Europe (along with Kraftwerke) and North America (Tesla Autobidder) through electricity spot markets, whereas China’s Jibei/Zhejiang pilots emphasize industrial load flexibility and grid coordination under the “national blueprint + regional experimentation” policy framework. At the technological level, Europe has made breakthroughs in transnational protocol standardization, whereas China addresses multiple stakeholder data interoperability challenges. Commercially, market-based bidding mechanisms prevail in Western systems, whereas China’s Guangdong pilot demonstrates a shift from subsidy-driven models to hybrid capacity leasing and energy splitting schemes. Shared technical challenges stem from the resource heterogeneity-induced dispatch complexity, with global implementations adopting hierarchical control architectures and China pioneering digital twin simulations integrated with blockchain-based trust mechanisms. The EU’s Energy Data Space initiative and China’s National Unified Electricity Market are poised to spearhead cross-border coordination and domestic market integration, respectively. Concurrently, the convergence of vehicle-to-grid (V2G) technologies with AI-driven dispatch algorithms is anticipated to enable over 500 GW of global VPP capacity by 2030, positioning virtual power plants as critical infrastructures in low-carbon energy transitions.

4.

Functional evolution and technological advancement

VPPs exhibit two distinct functional archetypes:

(1)

Commercial VPP (CVPP): market-driven architectures that optimize DER portfolios for electricity market participation, as demonstrated by the European FENIX initiative’s market contracting mechanisms;

(2)

Technical VPP (TVPP): grid service-oriented systems delivering ancillary services including frequency regulation and voltage stabilization, with its technical validity proven by the German ProViPP project’s successful aggregation of DERs for reserve capacity provisions [11,20].

The evolution of core technologies accelerates the VPP intelligence through three key dimensions:

(1)

Control architectures: paradigm shift from centralized EMS frameworks to decentralized multiagent systems (MASs), achieving scalability enhancements and system resilience improvements [21];

(2)

Communication infrastructure: integration of the 5G NR (New Radio) and IoT enables a <100 ms latency, as demonstrated by the Dutch PowerMatcher initiative’s UMTS-based load modulation (1.2 s response cycles);

(3)

Predictive analytics: hybrid architectures that combine big data analytics with AI algorithms (e.g., the CEEMDAN-DBN framework) achieve a 20% prediction error reduction in day-ahead markets, as evidenced by [22].

2.2. Composition and Functions of VPPs

2.2.1. Core Constituent Elements

The fundamental principle of VPPs is to aggregate distributed energy resources into a unified dispatchable entity via advanced communication technologies and optimal control strategies. The primary constituents consist of four resource types:

• Distributed energy resources (DERs)

DERs constitute the fundamental building blocks of VPPs, including renewable generation systems (photovoltaic arrays and wind turbines) and low-emission power sources (micro gas turbines and fuel cells) [23]. Geographically dispersed across distribution grid nodes, these resources demonstrate a pronounced output intermittency and volatility. Residential DER deployments (e.g., rooftop solar) prioritize locals load serving with grid exports of excess generation, whereas utility-oriented DER installations (e.g., clustered wind farms) are designed for bulk energy production and grid integration [24].

2.

Energy storage systems (ESSs)

ESSs serve as pivotal solutions for renewable output smoothing and grid flexibility enhancement. Predominant technologies include electrochemical storage (lithium-ion batteries), kinetic energy storage (flywheels), and capacitive storage (supercapacitors) [23]. By executing charge-discharge cycles, ESSs enable a temporal-spatial energy redistribution, delivering ancillary services (frequency regulation and peak load management) and participating in capacity markets. An optimal ESS performance depends on sophisticated SoC management systems that continuously monitor and adjust operations on the basis of environmental conditions (temperature variations) and degradation patterns (capacity fading) [25].

3.

Controllable loads (CLs)

CLs comprise demand-side resources with operational adjustability, including industrial processes, commercial cooling/heating systems, and residential IoT-enabled appliances. By implementing DR programs, VPPs actively reshape load curves for a peak demand reduction and contingency reserve provision [23]. Electric vehicles exemplify advanced CL applications, where intelligent charging coordination enables a bidirectional energy exchange (vehicle-to-grid) to support the grid frequency stability during supply-demand imbalances.

2.2.2. Fundamental Operational Capabilities

VPPs fulfil three primary operational roles in Chinese modern power systems via the coordinated management of heterogeneous energy assets:

• Resource aggregation and coordinated control

The essential capability of VPPs resides in creating unified control domains from spatially distributed assets with diverse operational characteristics. By utilizing centralized or distributed control frameworks, VPPs synchronize three operational dimensions: the DER generation dispatch, energy storage system (ESS) operational strategies, and controllable load (CL) demand modulation. During frequency regulation events, VPPs execute sub-second adjustments to storage systems and electric vehicle (EV) charging profiles to compensate for grid frequency deviations [26].

2.

Market operations and economic optimization

Functioning as competitive market entities, VPPs operate across three market layers: energy commodity trading, ancillary service provisioning (including frequency regulation and operating reserves), and capacity commitment mechanisms. Advanced bidding algorithms enable VPPs to optimize either profit margins through revenue maximization or system-level efficiency via cost minimization. Day-ahead scheduling integrates price forecasting models with resource availability predictions, whereas real-time operations employ adaptive control strategies to respond to the market price volatility within 5 min intervals [27].

3.

Grid resilience and operational flexibility

VPPs mitigate renewable integration challenges through three stability services: the synthetic inertia emulation for frequency stabilization, reactive power compensation for voltage regulation, and spinning reserve provision for contingency responses. In systems with a reduced rotational inertia, VPPs implement virtual synchronous machine (VSM) control algorithms to mimic conventional generator dynamics. During voltage excursion events, the coordinated control of DER inverters and ESS power converters ensures nodal voltage compliance within ±5% of the nominal values [28].

Internationally, VPPs provide critical services to DSOs/TSOs [29], including the following:

• Frequency Regulation: German ProViPP pilots demonstrate a 100 MW capacity for ±0.1 Hz adjustments via CHP units [14].
• Voltage Support: Dutch PowerMatcher projects use DER inverters to maintain the nodal voltage within ±5% [15].
• Black Start Capability: Tesla’s South Australia VPP leverages battery clusters for a 150 MW emergency restoration [30].

2.2.3. Typical Architecture

The typical VPP architecture comprises three layers–an aggregation layer, a communication layer, and a dispatch optimization layer (see Figure 1), where the dispatch optimization layer serves as the operational nucleus for the advanced functionality realization.

• Aggregation layer

The aggregation layer manages the physical resource integration and local control, including terminal devices of DERs, ESSs, and CLs [31]. This layer acquires the real-time resource status (e.g., PV output, battery state of charge (SoC), and load demand) through sensors and controllers and executes low-level control commands (e.g., inverter power settings and load switching operations).

2.

Communication layer

The communication layer establishes information channels between resources and upper management systems, enabling a bidirectional data exchange. Key technologies include low-latency communication protocols (e.g., MQTT, DDS), edge computing nodes, and cybersecurity protection mechanisms [32].

3.

Dispatch optimization layer

As the operational core of the VPPs, this layer performs market decision-making, resource dispatch, and risk management, specifically including the following:

(1)

Multitime-scale optimization: VPPs require coordinated scheduling across day-ahead, intraday, and real-time horizons [33]. Day-ahead scheduling develops generation plans on the basis of electricity price forecasts and resource availability predictions, whereas the real-time dispatch dynamically adjusts power outputs through model predictive control (MPC) or reinforcement learning (RL) to address forecasting errors and contingencies [34].

(2)

Uncertainty management: The stochastic nature of renewable energy generation and market price volatility constitute major optimization challenges [35]. Stochastic optimization (SO) and robust optimization (RO) are predominant methodologies [36]. For example, a scenario analysis generates multiple wind power output scenarios, with scenario reduction techniques selecting representative cases to decrease the computational complexity.

(3)

Market gaming strategies: VPPs must account for strategic interactions with other market participants. In bi-level programming models, the upper level simulates market clearing processes, whereas the lower level optimizes VPP bidding strategies. Cooperative game frameworks enable a joint optimization with distribution utilities to achieve an equitable benefit distribution.

(4)

Cross-market coordination: VPPs participate concurrently in energy markets, ancillary service markets, and carbon markets. Through carbon quota trading and Green Certificate mechanisms, VPPs can monetize low-carbon energy generation as additional revenue streams [37].

The dispatch optimization layer critically determines the VPP’s economic viability and operational reliability. Its complexity stems from multiobjective conflicts (e.g., profit maximization vs. carbon emission minimization), high-dimensional decision variables (e.g., thousands of DER output combinations), and real-time requirements. Mixed-integer linear programming (MILP) and deep reinforcement learning (DRL) are widely employed for large-scale optimization, whereas edge-cloud collaborative computing architectures enhance the solution efficiency through distributed processing [38]. Furthermore, cybersecurity risks in cyber–physical systems (CPSs) (e.g., cyberattacks and data tampering) necessitate incorporating resilience constraints into optimization models to ensure the VPP’s robustness under abnormal conditions.

Synthesis: VPPs enable a transformative grid-edge flexibility through cyber–physical integration. The dispatch optimization tier serves as the intelligent nexus translating physical resource capabilities into market-responsive strategies.

2.3. Significance of VPPs

Driven by “dual carbon” goals, VPPs have emerged as critical technological solutions for the transition to green energy through the aggregation of distributed renewable energy, energy storage, and flexible loads. The new power system dominated by renewable energy faces an inherent stochasticity and volatility in generation outputs, rendering the traditional “source-follows-load” balancing paradigm unsustainable. VPPs integrate massive distributed resources into flexible controllable entities via advanced control technologies, significantly enhancing the renewable energy accommodation capacity and mitigating supply-demand imbalances. In enhancing the power system flexibility, VPPs achieve a dynamic supply-demand balance across multiple temporal and spatial scales through coordinated control and optimized dispatch technologies. Their core significance manifests in three dimensions:

• Strengthening system regulation capability: aggregated energy storage and flexible loads in VPPs enable rapid responses to grid peak shaving and frequency regulation requirements;
• Improving supply-demand matching efficiency: VPPs dynamically optimize generation schedules and load responses via AI algorithms and real-time communication technologies;
• Expanding system operational boundaries: VPPs transcend traditional grid physical constraints through cross-regional resource coordination.

Furthermore, VPPs drive electricity market mechanism refinements through innovative business models. Their participation in energy markets, ancillary service markets, and demand response programs generates economic benefits for distributed resource owners, stimulating user engagement. The integration of economic incentives with low-carbon objectives accelerates energy transition processes.

In conclusion, VPPs not only ensure the secure operation of new power systems but also serve as strategic instruments for achieving dual-carbon goals. Through multidimensional innovations in technology, mechanisms, and markets, VPPs promote greener energy structures and more flexible power systems, providing replicable paradigms for the global energy transition.

3. Measurement and Assessment of Responsiveness

Responsiveness in VPPs is defined as the system’s ability to coordinate DER portfolios (energy storage units, photovoltaic generation, and flexible loads) for an agile adaptation to grid operational requirements and market price fluctuations. This operational capacity forms the fundamental basis for the critical roles of VPPs in grid ancillary services, energy trading, and system resilience assurance. The responsiveness measurement acts as the perceptual foundation for optimization frameworks, delivering accurate operational constraints and evaluation benchmarks. Dispatch optimization algorithms operate as the cognitive core, converting the theoretical response capacity into operational value through intelligent control strategies. The responsiveness assessment forms the critical pathway from concept validation to market deployment, enabling technical refinement, risk-constrained operations, and a systemic transition to sustainable power systems. Systematic evaluation frameworks allow VPPs to optimally match grid demands, maximize DER aggregation benefits, and achieve a balanced optimization of societal welfare and economic returns.

3.1. Current Status of Response Capacity Evaluation Systems

In the field of demand response potential assessments, the existing research has focused primarily on two directions:

• Analyzing the demand response capabilities of individual load resources;
• Evaluating the comprehensive demand response potential of all users within a region.

The former investigates the response characteristics of specific load types under different electricity price incentives through modelling and simulation, enabling the precise quantification of the response potential for individual load categories. However, this approach often fails to comprehensively reflect the aggregate potential when applied to large-scale regional integrated demand response assessments, exhibiting inherent limitations. The current research predominantly focuses on modelling the independent response capabilities of individual distributed resources while neglecting the analysis of aggregated system-wide response characteristics, particularly with respect to dynamic variations in response times, regulation directions, and metric weighting allocations. Furthermore, conventional methods struggle to effectively manage coordination and instability among aggregated resources during grid contingencies or complex operational scenarios, resulting in deviations between evaluation results and actual regulation capacities. Therefore, Reference [39] proposes establishing a comprehensive response capacity evaluation system that integrates dynamic metrics, including the regulation direction, magnitude, timing, duration, and ramping rates, while quantifying aggregated feasible regulation domains through optimization models to support grid dispatch decisions.

3.2. Metrics for Assessing Aggregated Response Capacity

To evaluate the aggregated response capacity of the VPPs accurately, multidimensional quantitative metrics must be established on the basis of the individual characteristics of the distributed resources. The core metrics proposed in Reference [39] comprise five categories:

• Adjustment direction (D)

The regulation direction is defined as the variation trend of the resource output relative to the operational baselines, including upwards regulation (output increase) and downwards regulation (output decrease). It refers to a VPP’s ability to perform “peak shaving” (reducing load) or “valley filling” (increasing load) within the power system.

2.

Adjustment amplitude (A)

The adjustment amplitude indicates the maximum output deviation relative to the current operating point, which is divided into the upwards regulation magnitude ($A^{+}$) and the downwards regulation magnitude ($A^{-}$). It measures the maximum power variation range of adjustable resources in a virtual power plant and is a core indicator for assessing its regulation potential. Taking hydropower units as an example, their regulation magnitudes can be calculated via the differences between the current output ($P_h\left(t\right))$ and output limits ($P_{h,\max }$), ($P_{h,\min }$):

$$A_h^{+}=P_{h,\max }-P_h(t),A_h^{-}=P_h(t)-P_{h,\min },$$

(1)

3.

Response time (T^R)

The response time refers to the shortest time required from receiving a regulation command to fully achieving the target output, determining the VPP’s ability to respond rapidly to a sudden demand.

4.

Duration (T^D)

Duration indicates the maximum time that a resource can maintain the target output state, reflecting its regulation sustainability. For energy storage systems, the duration is constrained by capacity limits, with the discharge duration calculated as follows:

$$T^D={\displaystyle \frac{E_b\left(t\right)-E_{b,\min }}{P_b\left(t\right){\eta }_b}},$$

(2)

where $E_b\left(t\right)$ represents the current stored energy and where ${\eta }_b$ denotes the charge/discharge efficiency.

5.

Ramping rate (R)

The ramp rate is defined as the rate of change in the resource output per unit time, reflecting the dynamic performance of the resource regulation. For distributed resources, the ramping rate can be directly calculated through output differences:

$$R={\displaystyle \frac{P\left(t\right)-P\left(t-1\right)}{\mathsf{\Delta }t}},$$

(3)

For aggregated VPPs, ramping rates require dynamic adjustments while considering individual resource output constraints and power balance conditions.

3.3. Responsiveness Evaluation Framework

• On the basis of the aforementioned metrics, Reference [18] evaluated the aggregated VPP response capacity in four aspects: the tie-line power baseline, upper/lower tie-line power limits, VPP ramping rate limits, and VPP response time.

(1) The tie-line power baseline, defined as the power profile without response requirements, is calculated through the day-ahead optimal dispatch with the minimum total cost. The baseline serves as the reference benchmark for measuring a VPP’s actual regulation capability; meanwhile, the grid must establish the VPP’s “original state” to accurately allocate regulation tasks (e.g., peak shaving and frequency regulation). In electricity markets (such as ancillary services markets), the baseline is also used as the benchmark value for revenue settlement, preventing resource misreporting or inflation. The objective function is as follows:

$$\min {\sum }_{t\in \psi }\left({\sum }_{i\in S} c_i\left(p_i\left(t\right)\right)+c_{tie}\left(t\right)p_{tie}\left(t\right)\right),$$

(4)

(2) Upper/lower tie-line power limits: The maximum and minimum achievable tie-line power at time $t$. The grid must evaluate, based on the upper and lower power limits of the interconnection lines, whether the VPP satisfies the transmission corridor’s capacity constraints to avoid overload risks. Additionally, the diversity of aggregated resources (e.g., energy storage, controllable loads, and distributed generation) must be uniformly characterized—via these upper and lower bounds—to represent the VPP’s overall regulation capability.

(3) VPP ramping rate limits: The maximum allowable power increases/decreases from the current time $t$ to $t+1$ when operating at the power baseline. The grid must rapidly balance the variability of new energy sources (e.g., sudden drops in the PV output or spikes in wind generation). The ramp rate upper and lower limits determine whether a VPP can match the power change rate required by the grid. Differences in ramp rates among resources (such as fast-responding storage versus slow-starting gas turbines) must be uniformly represented by aggregate bounds. Furthermore, both the spot market and the frequency regulation market impose explicit ramp rate requirements (for example, PJM in the U.S. mandates that frequency regulation resources have a ramp rate ≥ 1 MW/min). The objective function is as follows:

$$\mathrm{Max}\left({\sum }_{i\in S} p_i\left(t+1\right)-{\sum }_{i\in S} p_i\left(t\right)\right) t\in \psi ,$$

(5)

$$\mathit{min}\left(\sum _{i\in S} p_i\left(t+1\right)-\sum _{i\in S} p_i\left(t\right)\right) t\in \psi ,$$

(6)

The objective function is subject to the following constraints:

$$X(t)=X^{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}}(t),$$

(7)

This equation indicates that all device states at time $t$ remain identical to baseline states, meaning that the VPP operates along the baseline at time $t$.

(4) VPP response time ($T_{agg}^R$): When operating at the power baseline, the minimum time required for aggregated resources to achieve the regulation command $\mathsf{\Delta }P$ after receiving the instruction at time $t$, reaching the target at $t$ + $T_{agg}^R$.

The baseline provides the starting point for regulation; the upper and lower limits define the regulation range; the ramp rate determines the regulation speed; and the response time constrains the regulation timeliness—together, these four elements form the spatiotemporal framework of a VPP’s regulation capability.

2.

The established distributed resource responsiveness model enables the formulation of VPP operational constraints including the power balance, tie-line power limits, and equipment operational constraints. Combined with the objective function and partial constraints, this forms the optimization model. Solving this model yields the aggregated response capability.

3.

Weighting analysis and evaluation

Indicator weighting coefficients are determined through an entropy-based objective weighting methodology:

$$w_{m,n}={\displaystyle \frac{1-e_{m,n}}{\sum _{n=1}^M 1-e_{m,n}}},$$

(8)

The final aggregated responsiveness score $F_m$ is obtained through a weighted summation of the normalized indicators $x_{m,n,t}^{\prime }$:

$$F_m={\displaystyle \frac{1}{24}}{\sum }_{t=1}^{24} {\sum }_{n=1}^M w_{m,n}{x}_{m,n,t}^{\prime },$$

(9)

where the score $F_m$ ranges between [0, 1], with higher values indicating a better responsiveness.

The table (

Table 3. Symbol legend.

Symbol	Signification	Symbol	Signification
$\psi$	operating time set (1 day, i.e., 24 time periods in this study)	$c_{tie}\left(t\right)$	electricity purchase price at tie-line during time t
$S$	set of all devices (including thermostatically controlled loads)	$X\left(t\right)$	the decision variable matrix at time t, including device outputs, workstation reserves, ESS stored energy, etc.
$i$	device index	$X^{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}}\left(t\right)$	matrix of baseline decision variable values at time t
$p_i\left(t\right)$	power output of device i at time t	$M$	the maximum index number
$p_{tie}\left(t\right)$	tie-line power at time t (positive direction defined as power inflow from upper grid to VPP)	$e_{m,n}$	represents the entropy value of the nth indicator for the mth VPP, reflecting its information content.
$c_i\left(p_i\left(t\right)\right)$	cost function of device i at time t

) below delineates the semantic significance conveyed by each symbol.

4.

Implementation process

The tie-line power baseline calculated and reported by the VPP serves as the settlement reference. Tie-line power limits define the operational feasible domain as grid dispatch constraints. When dispatching regulation commands and calculating next-period scheduling plans, the VPP’s ramping rate determines the feasible output domain for the next period as scheduling constraints. The VPP response time required to achieve specified power adjustments serves as a scheduling computation constraint. Additionally, the calculated aggregated responsiveness score provides reference metrics for the VPP optimization configuration to ensure a sufficient response capacity [39].

3.4. Summary

The evaluation system proposed in Reference [39] quantifies the feasible regions and dynamic characteristics of VPP aggregated response capabilities through multidimensional metric modelling, optimization solving, and a weighting analysis, providing a refined decision-making support for the grid dispatch. Future research could further consider multitemporal coupling and market interaction mechanisms to enhance the applicability of the evaluation model.

4. Resource Optimization Dispatch

The responsiveness assessment and resource optimization dispatch form the “capability cognition–action decision” closed loop in VPPs. The former addresses “what can be achieved”, whereas the latter resolves “how to implement”. A dispatch without an assessment resembles “map-free navigation”, risking resource conflicts or market exposure; an assessment decoupled from the dispatch becomes a speculative discourse, failing to unlock the commercial and technical value of VPPs. The direct integration of both is essential for realizing the “measurable, controllable, optimizable” characteristics of VPPs. VPP development requires establishing demand-aligned dispatch models and strategies on the basis of resource responsiveness evaluations.

4.1. Dispatch Objectives

• Economic objective: maximize aggregated resource value

Overall economic efficiency can be achieved through optimized energy production, distribution, and consumption. For example, dynamically adjusting renewable–thermal generation ratios in electricity markets, utilizing energy storage charging during low-price periods, and discharging during high-price periods can reduce system costs and increase revenues. A comprehensive consideration of generation costs, market prices, and subsidy policies is needed.

2.

Technical objective: ensure power system security

Maintain the grid frequency and voltage within safe thresholds to prevent overloads or failures. Examples include optimizing the reserve capacity allocation and adjusting the power flow distribution to handle contingencies. N-1 criterion compliance and dynamic stability analyses are needed.

3.

Flexibility objective: enhancing uncertainty resilience

Improving the system adaptability to renewable generation volatility and load demand randomness [40]. The demand response for interruptible loads can be implemented, or storage fast charging/discharging can be utilized to mitigate power fluctuations. The flexibility resource regulation potential (e.g., storage capacity and adjustable load ratios) must be quantified.

4.

Coordination objective: multistakeholder benefit equilibrium

Coordinate interests among generators, consumers, and storage operators. Equitable revenue allocation mechanisms should be designed to incentivize distributed PV participation in peak shaving or to achieve cross-regional resource complementarity through bilateral contracts. Game theory or cooperative optimization models are needed.

4.2. Dispatch Optimization Model

A dispatch optimization model is a mathematical or algorithmic framework that generates optimal dispatch strategies under given constraints (e.g., resource capacity, grid demand, and market regulations) [41].

4.2.1. Model Classification and Composition

Optimization models for dispatch can be categorized into the following three types on the basis of their optimization objectives:

• Economic models

Economic models are the most common type of optimization model for VPP dispatch. Their core objective is to maximize the economic benefits of the VPP. By integrating DERs, such as solar power, wind power, energy storage systems, and controllable loads, VPPs participate in electricity market transactions to obtain economic benefits.

Guo Hongxia et al. [42] proposed an optimization dispatch model for VPPs under a unified electricity trading market. This model aggregates resources such as distributed power sources (e.g., wind and photovoltaic power), energy storage systems, and controllable loads. The constraints include power balance constraints, equipment operation constraints, and market trading constraints. The model combines day-ahead and real-time markets to optimize the electricity trading volume of the VPP and its internal energy flow. It considers the operation constraints of distributed power sources, energy storage systems, and interruptible loads. Through case studies, the model effectively coordinates the internal energy allocation of the VPP and optimizes its electricity trading with external markets. Compared with independent market models, the unified market model significantly increases the total revenue of the VPP, achieving the lowest overall purchase costs and the highest sales revenue, thereby verifying the effectiveness and superiority of the model.

Niu Dongxiao et al. [43] developed a bi-level optimization dispatch model for VPPs and energy-efficient power plants (EEPPs). This model aggregates resources such as wind power, photovoltaic power, gas turbines, energy storage systems, and EEPs. The constraints cover electricity supply and demand balance constraints, thermal power unit output constraints, and system reserve constraints. The upper level aims to minimize the system’s generation cost, whereas the lower level aims to maximize the revenue of the EEP. The model considers the optimization effects of energy storage systems, demand response, and energy-saving projects on the operation of the VPP. Through simulation systems, the model is shown to significantly enhance the system’s ability to accommodate VPPs and reduce generation costs. The introduction of EEPs leads to a significant increase in grid-connected electricity from wind and solar power, a reduction in the net load demand, and the smoothing of the load curve, providing more space for the grid connection of VPPs.

2.

Reliability models

The optimization objective of reliability models is to ensure the operational safety and power supply reliability of VPPs. VPPs integrate various types of DERs, the outputs of which may be affected by factors such as weather conditions and equipment failure.

Xu Hui et al. [44] developed a stochastic dispatch optimization model for VPPs that integrates wind power, photovoltaic (PV) power, energy storage, and the demand response. This model aggregates resources such as wind and PV power generation, gas turbines, energy storage systems, and the incentive-based demand response. The constraints include the uncertainty of the wind and PV power outputs and load demand fluctuations. The model incorporates the Conditional Value at Risk (CVaR) theory and confidence level methods to quantify operational risk and optimize dispatch strategies. It also considers the impact of the price-based and incentive-based demand response on VPP operations. Through a case study on an improved IEEE 30-bus system, the model effectively smooths the load curve and reduces the operational risk of the VPP. Although the operational revenue of the VPP slightly decreases after uncertainty is considered, the risk is significantly reduced. The model provides decision-makers with a flexible risk control tool, allowing them to adjust dispatch strategies on the basis of their risk preferences.

3.

Multiobjective Models

Multiobjective models consider multiple objectives of VPP dispatch optimization, such as economic efficiency, reliability, and environmental benefits. In practical applications, single-objective optimization often fails to meet the complex needs of VPPs.

Liu Yujia et al. [45] proposed a blockchain-based VPP optimization dispatch model that aggregates resources such as distributed power sources, energy storage systems, controllable loads, and blockchain technology. The constraints include distributed characteristic constraints, information security constraints, fast transactions, and coordinated dispatch constraints. By combining a blockchain-based distributed particle swarm optimization (BD-PSO) algorithm and a proof of computation work (POCW) consensus algorithm, an optimization computation blockchain (OCB) system was constructed. This model enhances the information security and dispatch efficiency of the VPP through blockchain technology while considering the uncertainty of distributed resources. Verified through actual data from a certain region, the OCB-VPP model outperforms traditional algorithms in terms of its computational speed and accuracy. It reduces system operating costs by 13.29%, decreases carbon emissions by 14.55%, and increases the consumption capacity of new energy by 9.95%. The model shows a significant performance in reducing operating costs, improving energy utilization, and reducing carbon emissions.

Zhou Yizhou et al. [46] established a comprehensive energy coordinated dispatch optimization model for multiregional VPPs, aggregating resources such as gas turbines, boilers, fans, PV panels, electric energy storage, thermal energy storage, electrical loads, thermal loads, and cooling loads. The constraints include power balance constraints for electricity, heat, and cooling; interregional energy transmission constraints; and reserve capacity constraints. The model aims to maximize economic benefits in the energy market and spinning reserve market while considering environmental costs and start-up/shutdown costs. Verified through a case study on a multiregional combined cooling, heat, and power (CCHP) system in Changsha, the model effectively reduces the operating costs of the VPP and achieves an optimized coordinated dispatch of cooling, heat, and electricity among different regions. Compared with the scheme that does not consider CCHP and interregional energy interactions, the model significantly improves the system economic efficiency and reduces environmental pollution.

4.2.2. Current Status Analysis

A comparative analysis of the dispatch models mentioned above is presented in

Table 4. Comparison of dispatch models.

Dispatch Model	Literature 19	Literature 20	Literature 21	Literature 22	Literature 23
Objective Type	Economic		Reliability	MultiObjective
Time Scale	Day-ahead (24 h) + real-time dynamic adjustment	Long-term planning (annual/seasonal) + real-time dispatch	Day-ahead + intraday rolling (15 min-1 h) + real-time control	Multitime–scale coordination (day-ahead–real-time dynamic adjustment)	Day-ahead (24 h)
Uncertainty Handling	Not explicitly mentioned, implied deterministic model	Stochastic optimization (Weibull distribution for wind speed, Beta distribution for PV radiation)	CVaR + confidence level method	Improved particle swarm optimization for distributed uncertainties	Based on forecast data (electricity prices, wind/solar output, load), deterministic model used
Mathematical Tools	Benders decomposition algorithm, MILP	Bi-level stochastic programming, MILP	Stochastic programming, CVaR model, reinforcement learning (RL)	BD-PSO, POCW consensus algorithm	MILP, Boolean variables to describe equipment start-up/shutdown states
Resource Types	Distributed generation (DG), ESS, interruptible loads (ILs)	Wind, PV, gas turbines, energy storage, energy saving/efficient power plants (ESEPP/DREPP)	Wind, PV, gas turbines, energy storage, incentive-based demand response (IBDR)	Wind, PV, energy storage, gas turbines, controllable loads (blockchain nodes)	gas turbines, boilers, fans, PV, electric/thermal energy storage, cooling/heating/electricity loads, electric chillers, absorption chillers, etc.
Market Mechanism Adaptability	Unified electricity market (day-ahead/real-time market arbitrage)	Energy market + Ancillary service market, coupled with carbon trading	Dynamic pricing mechanism (PBDR+IBDR), participation in frequency regulation market	Blockchain-supported decentralized trading, optimizing price arbitrage and capacity market	Compatible with EM and SRM, dynamic response to electricity prices (selling electricity/interrupting loads during peak times), additional revenue from reserve capacity services

.

As shown in Table 4, optimization dispatch models for VPPs have been widely applied in various scenarios. The most common scenarios include multiregional coordinated dispatch, uncertainty handling, and participation in electricity market transactions. In these scenarios, commonly used models include dispatch models based on MILP, stochastic optimization models, and multiobjective optimization models. However, these conventional models are typically only suitable for relatively simple scenarios and often fail to achieve the desired outcomes in more complex situations. As a result, numerous specialized models have been developed to address these complexities. For example, in multiregional coordination, the integration of distributed energy resources and loads from different regions enables interregional energy interactions and optimized dispatch, significantly enhancing resource utilization efficiency and system flexibility. In terms of uncertainty handling, the introduction of the CVaR theory and multiparameter programming methods has effectively quantified the uncertainties in the distributed energy output and load demand, optimized dispatch strategies, and reduced operational risks. The application of blockchain technology has provided VPPs with more efficient and secure distributed computing and information management tools, thereby improving dispatch efficiency and information security levels. Although dispatch models vary in their objective types, their practical application effects commonly focus on economic benefits, with some even emphasizing this aspect. Moreover, interactions with the market have become increasingly sophisticated. Given that these studies are recent research findings, they reflect that the VPP model in China is actively seeking to transition towards marketization and commercialization.

Despite the significant progress made, several challenges remain to be addressed:

• Adaptability to complex scenarios: Existing models still have limitations when dealing with multiregional, multitype distributed energy resources and complex market mechanisms. Particularly under dynamic and multi-scenario operating conditions, the adaptability and solution efficiency of the models need to be further improved.
• Refined modeling: There are significant differences in the operational characteristics of equipment within VPPs. The existing models still need to be refined, especially in terms of modelling the charging and discharging losses of energy storage systems and the output fluctuations of distributed energy resources.
• Market mechanism and VPP interaction: With the increasing penetration rate of distributed energy, how to better coordinate the interaction between VPPs and electricity markets and how to maximize economic benefits while ensuring stable system operations remain pressing issues [47].
• Data quality and forecasting accuracy: Optimization dispatch models rely on high-quality historical data and accurate forecasting results. However, existing forecasting technologies still fall short, particularly in terms of forecasting accuracy for the distributed energy output and load demand.

In summary, current VPP optimization dispatch models have made significant progress in their multiregional coordination, uncertainty handling, and integration with market mechanisms. However, they still face challenges such as insufficient adaptability to complex scenarios, inadequate refined modelling, and imperfect market interaction mechanisms. In the future, the development of multi-scenario analysis, refined modelling, the integration of blockchain technology, and improvements in forecasting accuracy are expected to further increase the optimization dispatch level of VPPs and promote their widespread application in new types of power systems.

4.3. Dispatch Strategies

4.3.1. Strategy Classification

Dispatch strategies refer to a set of rules or control logic applied in practice to dynamically adjust the operating state of a VPP in response to real-time demands or unexpected events. On the basis of the optimization solutions provided by models and combined with real-time data (such as electricity price fluctuations, load changes, and renewable energy output variations), dispatch strategies enable rapid decision-making. Depending on the focus of different VPPs, dispatch strategies can be categorized into the following three types on the basis of their orientation:

• Economic-oriented strategies

(1) Time-of-use (TOU) tariff response

By integrating various types of DERs, VPPs can flexibly respond to TOU tariff mechanisms to achieve peak-valley arbitrage [48]. During off-peak periods, VPPs can schedule energy storage systems, V2G electric vehicles, and controllable loads for charging or energy storage operations to take advantage of low electricity prices and reduce operating costs. In contrast, during peak periods, VPPs can release stored energy from energy storage systems or adjust the output of distributed generation units to supply electricity to the grid and earn a higher revenue from high electricity prices. In this way, VPPs not only optimize their own economic performance but also help users reduce electricity costs while providing peak shaving and valley filling services to the grid, thereby improving the grid operating efficiency.

Gao Jinrui et al. [49] proposed a two-level economic dispatch strategy. The upper level uses information gap decision theory (IGDT) to robustly model the intermittency of the DG output, whereas the lower level optimizes the charging and discharging plans of electric vehicles (EVs) by combining dynamic time-of-use pricing (DTOUP) and “vehicle–road” information to minimize the total cost of EV users and the total cost of the VPP. Taking a VPP that includes distributed generation, energy storage, charging stations, and load units as an example, the effectiveness of the strategy was verified through simulation. The results showed that the DTOUP model is more flexible in guiding EV charging and discharging than the traditional TOU pricing (TOUP) model is. This reduces the interaction cost between the VPP and the grid and improves the overall economic benefits.

(2) Dynamic adjustment of market bidding

By utilizing advanced monitoring and forecasting technologies, VPPs can track real-time fluctuations in electricity market prices and dynamically adjust their bidding strategies on the basis of real-time prices [50]. By integrating the flexibility of distributed energy resources, VPPs can respond to market signals in a short time and flexibly adjust the output plans of generation or energy storage equipment. When real-time electricity prices rise, VPPs can prioritize scheduling low-cost distributed energy resources (such as wind and solar power) and energy storage system discharges to earn greater economic benefits. Conversely, when prices drop, they can increase the charging of energy storage devices [51].

Gao Yuanhao et al. [30] proposed a dynamic aggregation and multi time-scale bidding strategy for VPPs on the basis of reinforcement learning. By constructing a “cloud-edge-end” hierarchical and zonal control architecture, this strategy enables the dynamic aggregation of flexible resources and the identification of external characteristics. It uses a constraint-aware deep reinforcement learning method (MIP-DQP) to solve the day-ahead and real-time two-stage bidding model. Taking the example from the literature, this strategy dynamically aggregates fourteen flexible resources into five VPPs, significantly reducing communication costs and total operating costs, with a cost savings rate of 35.86%. In terms of the bidding strategy, the VPPs formulate preliminary output plans in the day-ahead market and correct them in the real-time market on the basis of more accurate forecast data. This effectively reduces the risks associated with inaccurate forecasts and increases the overall revenue. Compared with participating only in the day-ahead market, the two-stage bidding strategy increases the total market revenue by 22.05% and reduces deviation assessment costs by 462.41%. Moreover, the strategy yields results close to the theoretical optimum under the constraint conditions, verifying its effectiveness and practicality.

2.

Reliability-oriented strategies

(1) Dynamic allocation of reserve capacity

VPPs can dynamically allocate the reserve capacity by integrating distributed energy resources to cope with the volatility of renewable energy and equipment failure [52]. VPPs can use energy storage systems, fast-start backup power sources (such as small gas turbines), and controllable loads to flexibly adjust the allocation of the reserve capacity on the basis of real-time wind and solar power generation forecasts and load demands. When fluctuations in the wind or solar power generation occur, VPPs can quickly schedule energy storage systems to release power and compensate for the missing generation. In the event of equipment failure, backup power sources can be activated to ensure stable grid operations. This dynamic allocation strategy not only enhances the system reliability but also reduces the dependence on traditional backup power sources.

Li Dongdong et al. [53] proposed a scenario generation and reduction method based on Monte Carlo and Manhattan probability distances, modelling wind and solar power outputs and electricity prices. They also used interval methods to model carbon prices and the demand response, establishing an optimization dispatch model based on interval linear programming. The model aims to minimize the operating cost of the VPP while considering electricity and carbon trading and demand response mechanisms. Taking a small VPP system in East China as an example, which includes wind power, photovoltaic power, gas turbines, energy storage, and flexible loads on the user side, the simulation analyses the impact of different uncertainty handling methods on dispatch results. The results showed that considering multiple uncertainties allows the model to more accurately reflect dispatch results and improve the system’s economic performance.

(2) Emergency control for power balance

In emergency situations of the power grid, VPPs can quickly respond to power balance requirements. By integrating intelligent control technologies, VPPs can rapidly curtail controllable loads (such as the noncritical equipment of industrial users and EV charging loads) when the grid frequency decreases. At the same time, they can urgently start energy storage systems or backup generation equipment to restore the power supply quickly. VPPs can establish real-time communication with grid operators and immediately initiate emergency control procedures upon detecting grid faults to restore the power balance within a short time and prevent large-scale power outages [54].

Liu Dongqi et al. [55] proposed a dispatch strategy based on Nash-Q reinforcement learning, combining game theory and reinforcement learning to establish a hybrid objective dispatch model for VPPs. They introduced the Nash equilibrium model and optimized the VPP’s output strategy through multiagent learning. Taking a VPP that includes gas turbines, wind and solar power units, and energy storage batteries as an example, the simulation results showed that compared with traditional dispatch methods, this strategy increased revenue, reduced carbon emissions, and enhanced the system’s inertial support capability by utilizing the virtual inertia of energy storage batteries.

3.

Flexibility-oriented strategies

(1) Multienergy synergy

Breaking through traditional power system regulation boundaries through energy form conversion chains and combining market game mechanisms to achieve the multiobjective optimization of the economy, the environment, and safety represent important directions for the evolution of VPPs toward integrated energy system operators. Its core value lies in transforming volatile renewable energy into high-value stable energy products while creating new carbon asset management models [56].

Li Zhaoze et al. [57] proposed a flexibility adjustment strategy based on a green electricity-to-hydrogen-to-methanol system, establishing a flexible supply and demand model and introducing a Stackelberg game model. They optimized the flexibility regulation of the VPP through dynamic pricing and demand response mechanisms. Taking an improved IEEE 30-bus system as an example, which includes wind power, PV power, and methanol synthesis systems as flexible resources, the simulation results revealed that the strategy effectively reduced the volatility of the VPP’s external power purchase, increased the flexibility redundancy, and decreased carbon emissions.

(2) EV cluster V2G power smoothing

Incorporating EV clusters into their resource management systems, VPPs can achieve the power smoothing of EVs through V2G technology. VPPs can monitor the charging status and battery capacity of EVs in real time and flexibly schedule their charging and discharging power according to grid demands. During peak grid load periods, VPPs can schedule EVs to discharge power to the grid, whereas during off-peak periods, EVs can charge. This strategy not only enhances grid flexibility but also provides additional economic benefits for EV users and promotes the deep integration of transportation and energy systems [58].

Lu Zhigang et al. [59] established a two-level inverse robust optimization dispatch model for VPPs with V2G technology. The inner layer aims to maximize the total profit of the power generation, whereas the outer layer introduces the optimal inverse robustness indicator (OIRI) to quantify the uncertainty of the wind power and V2G charging and discharging power, optimizing the VPP’s output strategy. Taking a VPP that includes wind turbines and V2G users as an example, the simulation results showed that the strategy can optimize the V2G charging and discharging plan while ensuring wind power consumption, increasing the system’s total profit and robustness.

4.3.2. Current Status Analysis

Current optimization dispatch strategies for VPPs focus mainly on addressing the uncertainty of distributed energy resources, the complexity of the demand response, and the variability of market environments. In practical applications, optimization dispatch strategies based on deterministic models are relatively common, such as dispatch models that consider single uncertainty factors (e.g., fluctuations in wind or PV power output).

To address the aforementioned challenges, the new strategies introduced (as shown in

Table 5. Comparison of optimization dispatch strategies.

Strategy	Literature 24	Literature 25	Literature 26	Literature 27	Literature 28	Literature 29
Orientation	Economic-Oriented		Reliability-Oriented		Flexibility-Oriented
Uncertainty Handling	IGDT for robust modelling of DG output intermittency	Multi-scenario stochastic optimization with rolling time window in real time	Scenario generation and reduction for wind/solar output and electricity prices; interval method for carbon prices and demand response	Nash-Q reinforcement learning for high stochasticity and uncertainty	Copula theory for source–load scenarios + historical CDF for flexibility demand	Copula theory for source–load scenarios + historical CDF for flexibility demand
Optimization Algorithm	Two-level economic dispatch model: IGDT robust model (upper level), DTOUP and “vehicle–road” information for EV scheduling (lower level)	Constraint-aware deep reinforcement learning (MIP-DQP) with MDP, solved in training and online execution phases	Two-stage decomposition algorithm (YALMIP + CPLEX)	Nash-Q learning algorithm	Stackelberg game model (VPP operator vs. load aggregator) + CPLEX solver	GMOBCC algorithm + topological mapping/bisection method
Application Scenario	Coordinated optimization dispatch of VPP and EV aggregators	VPP participation in combined energy and reserve markets	VPP with sources, loads, and storage	Multitime-scale coordination of distributed energy resources	VPP with green hydrogen-to-methanol synthesis, integrating EV/HVAC flexibility resources	VPP with V2G, coordinating wind power accommodation and EV charging/discharging

) incorporate a variety of innovative methods. For example, a scenario generation and reduction method based on Monte Carlo simulations and the Manhattan probability distance has been proposed. This method, combined with interval methods to handle the uncertainties of carbon prices and the demand response, establishes an optimization dispatch model based on interval linear programming. This approach can comprehensively consider multiple uncertainties and provide decision-makers with a more flexible choice space in interval form, while improving the economic performance of the system on the basis of ensuring a safe operation.

Another example is the integration of game theory and reinforcement learning, which has led to the development of a Nash-Q reinforcement learning dispatch strategy with virtual inertia. By utilizing a multiagent game coordination dispatch model, this strategy effectively enhances the low-carbon operation capability and economic performance of VPPs while also strengthening the system’s inertial support capability.

However, these new strategies still face several pressing issues in practical applications. First, the increased model complexity leads to significant increases in the computational difficulty and solution time, especially when considering multiple time scales and multiobjective optimization. How to improve the solution efficiency is a key issue. Second, the new strategies have high requirements for data accuracy and real-time performance, and the difficulty of data acquisition and updating in practical applications is considerable. Moreover, how to further optimize the economic dispatch strategies of VPPs in a complex market environment to adapt to the dynamic changes in electricity markets, carbon markets, and ancillary service markets is also an important direction for in-depth research. Finally, with the expansion of the VPP scale and the diversification of technologies, ensuring system safety and reliability while reducing operating costs is another significant challenge for future research.

4.4. Case Studies

Table 6. Comparison of optimization scheduling models and dispatch strategies in virtual power plants.

Case Name	Dispatch Optimization Model	Dispatch Strategies	Country/Region
State Grid Gansu VPP	Carbon–Green Certificate Joint Trading Mechanism + Integrated Demand Response Model	Link electricity, carbon and green certificate markets to cut carbon emissions and energy purchase costs; introduce price- and substitution-based demand response models to explore demand-side flexibility.	China (Gansu)
State Grid Shanghai VPP [19]	Energy Blockchain Distributed Scheduling Strategy	Use PBFT consensus algorithm for autonomous load allocation and on-chain consensus; optimize power dispatch with equal incremental fuel consumption criteria.	China (Shanghai)
Guangdong Power Grid Multiregional VPP	Improved Grey Wolf Optimization (GWO) Algorithm	Jointly schedule wind, solar, and carbon capture units to coordinate multiregional distributed resources; enhance global optimization and reduce carbon emissions and net costs.	China (Guangdong)
Next Kraftwerke [60]	MILP optimization scheduling – Based on 15 min spot market prices – Includes minimum up/down times, storage and generator constraints	Periodic scheduling: switches between weekly, daily, and peak-load schedules, with parameterized asset flexibility; cloud platform re-optimizes in real time.	Germany
Tesla SA VPP [61]	Proprietary “Opticaster” algorithm – Real-time “Price-to-Use” optimization – Maximizes charge/discharge economic benefit	Real-time market response: discharges in high-price periods, charges in low-price periods; automatically bids into FCAS (frequency control ancillary services) market.	South Australia
Green Mountain Power VPP [61]	Simplified price–signal control – Residential storage responds to high-price forecasts – Cost-sensitive start/stop logic	Rolling periodic control: customers cede partial control; operator remotely adjusts in real time based on demand and price signals.	USA (Vermont)

below summarizes the optimization and scheduling models and strategies used in typical VPP cases both domestically and internationally. Overall, domestic research and pilots mainly center on mathematical optimization methods, such as stochastic programming, IGDT, chance-constrained models, and hierarchical coordination, to handle the uncertainty of electricity prices and the renewable energy output. In contrast, international practices more often adopt MILP, proprietary scheduling algorithms, and real-time market response strategies, leveraging market price signals and cloud platforms for precise scheduling in spot and ancillary service markets.

4.5. Summary

Dispatch optimization models are based on mathematical and algorithmic frameworks that generate optimal dispatch plans under constraints, including economic, reliability, and multiobjective models. These models cover fundamental units such as DERs, adjustable loads, energy storage strategies, and energy efficiency forecasting. Dispatch strategies, on the other hand, are dynamic adjustment rules that are based on real-time data and are categorized into economic-, reliability-, and flexibility-oriented strategies, such as time-of-use tariff responses, reserve capacity allocations, and energy storage frequency regulations. The two are closely linked: models provide optimization plans for strategies, whereas strategies offer real-time feedback to models, working together to achieve the efficient operation of VPPs. This enhances economic benefits, ensures power supply reliability, increases system flexibility, and promotes multiobjective optimization, driving the sustainable development of VPPs.

5. Challenges and Difficulties

5.1. Technical Challenges

The following will provide a summary focusing on four key aspects: the resource integration and modeling, response capability assessment, optimal scheduling, and standardization of VPPs.

• The integration and modelling of massive heterogeneous resources

VPPs need to aggregate a variety of resources, including DERs, ESSs, controllable loads, and electric vehicles. These resources are characterized by a large scale, wide distribution, and heterogeneous model parameters, making it difficult to aggregate and model them effectively. For example, the chemical properties of energy storage systems (e.g., differences between lithium-ion batteries and flow batteries) and the response characteristics of controllable loads (e.g., delay effects of thermostatically controlled loads) add complexity to the modelling process. While VPPs currently have various methods for modelling single-type common energy sources, they lack effective methods for quickly modelling complex load sides or demand responses. This requires the integration of large amounts of data and the design of specialized modelling solutions. In this context, modelling resources can be time-consuming and expose the lack of a template for scalability, which is not conducive to the large-scale commercialization of VPPs [62,63,64,65,66].

2.

The assessment of aggregated resource response capabilities

The technical challenges and difficulties in assessing the response capabilities of aggregated resources in VPPs mainly involve three aspects. First, regarding resource heterogeneity and model compatibility, significant differences in the dynamic characteristics of distributed energy, energy storage, and load devices (e.g., response delay, regulation accuracy) necessitate the development of a unified modelling framework across protocols and interfaces to address the challenges of multisource data integration and standardized representation. Second, dynamic coupling and uncertainty quantification—fluctuations in the wind and solar power output, random user behavior, and market signal disturbances are intertwined—require the development of spatiotemporally correlated joint probability models to overcome the bottlenecks of high-dimensional nonlinear optimization and multitime-scale coupling solutions. Third, in regards to balancing real-time performance and security, the second-level state perception of massive resources relies on low-latency communication and edge computing capabilities, but large-scale distributed architectures are vulnerable to network attacks and data tampering threats. It is necessary to integrate federated learning and lightweight encryption technologies to achieve reliable and traceable assessment results under privacy protection [67,68,69].

3.

Large-scale optimization dispatch

Optimization dispatch is a key aspect of VPP operation management, but the complexity of dispatch algorithms is one of the main technical difficulties currently faced. VPPs need to integrate various types of DERs, such as wind power, photovoltaic power, energy storage systems, and controllable loads, and achieve optimal dispatch under system constraints. However, owing to the intermittent and uncertain nature of DER outputs and the complex interactions between resources, dispatch algorithms have to handle many variables and constraints, leading to a significant increase in computational complexity. Moreover, as the scale of the VPPs expands and the number of DERs increases, the computational efficiency and real-time performance of dispatch algorithms are also challenged. Existing dispatch algorithms often require longer computation times when dealing with large-scale VPPs, making it difficult to meet real-time dispatch requirements.

4.

Energy communities and net billing strategies

Energy communities and net billing strategies are crucial concepts for the market operations of VPPs [70]. The rise of energy communities (ECs) introduces decentralized decision-making paradigms, where prosumers collaboratively manage DERs under net billing schemes. ECs, defined as locally organized entities that jointly invest in and operate renewable generation and storage assets, are reshaping VPP architectures by embedding social governance into technical dispatch models. For instance, Spain’s Recast Renewable Energy Directive (2023) incentivizes ECs to aggregate rooftop PV and EV clusters into VPPs, enabling peer-to-peer (P2P) trading via blockchain. The directive mandates a 30% tax rebate for ECs participating in VPPs, with Barcelona’s pilot project demonstrating a 25% increase in solar self-consumption through real-time blockchain settlements [71].

Net billing’s dynamic pricing mechanisms—ranging from feed-in tariffs (FiTs) to time-of-use (TOU) rates—fundamentally reshape VPP revenue models. In California, the Net Billing 3.0 policy (2024) replaces fixed FiTs with hourly locational marginal pricing (LMP), forcing VPPs to align storage charging with solar surplus periods. Tesla’s Los Angeles VPP, integrating 50,000 residential Powerwalls, achieves a 35% cost savings by discharging batteries during peak price intervals (18:00–21:00) and leveraging AI-driven price forecasting [72]. However, TOU-based net billing amplifies uncertainty. A 2023 EU study found that 1 h prediction errors in solar generation reduce VPP profits by 12% under dynamic pricing [73].

Technical integration challenges persist. ECs require interoperable communication protocols to harmonize heterogeneous DERs. The Dutch Energyshare platform, for example, uses OpenADR 3.0 and IEC 61,850 standards to connect 200+ ECs into a national VPP network, enabling 500 ms response times for grid frequency regulation. Blockchain further enhances transparency: Germany’s Enerchain project employs zero-knowledge proofs (ZKPs) to anonymize prosumer transactions while ensuring auditability, reducing settlement costs by 18% [71].

5.

The standardization of virtual power plants

The above difficulties are partly due to the lack of standardization in VPPs. The main challenges in the standardization of VPPs are as follows: First, the standardization process lags behind technological development, and the existing standard system is not yet comprehensive or coordinated, making it difficult to fully support the diverse business needs of VPPs. Second, the standardization work of VPPs in China started relatively late, with most existing standards being local or enterprise standards. There is a lack of unified national standards, leading to issues such as nonuniform standards and insufficient interoperability in the actual application of VPPs. Additionally, the existing standard system has significant gaps in the dynamic aggregation of multitype resources, cross-platform collaboration, communication security encryption, and market transaction rules, making it difficult to meet the complex technological and market-oriented needs of VPPs. Finally, the business models and management rules of VPPs are becoming increasingly complex, and the differences in policies and market rules across regions further complicate standardization efforts. These challenges restrict the large-scale and standardized development of VPPs and need to be addressed through the improvement of standard systems and strengthened policy guidance [74,75].

5.2. Market and Policy Challenges

VPPs also face numerous challenges in the market and policy arenas. First, the market mechanism design of the electricity market poses high requirements for VPP participation. VPPs need to compete with other power generation entities in the electricity market, but their uncertain output and complex scheduling of flexible resources cause VPPs to face greater risks in market transactions. As mentioned in the literature, when participating in the electricity market, VPPs need to consider market price fluctuations, incentive mechanisms for demand response, and coordination with grid operators [76,77,78].

Second, policy support is crucial for the development of VPPs, but the current policy environment is still not perfect [79]. Existing policies lack clear positioning for VPPs, especially in terms of their roles in carbon emission trading and green certificate markets. For example, although China’s “electricity demand response management measures” encourage VPPs to participate in peak shaving, the compensation mechanism for ancillary service markets is not detailed, creating many uncertainties for VPPs during investment, construction, and operation. Policy factors such as the entry threshold for VPPs, subsidy policies, and carbon emission trading mechanisms directly affect the economic performance and market competitiveness of VPPs.

The global development of VPPs is hindered by regional disparities in regulatory policies. The European Union promotes ECs’ participation in ancillary service markets through its Clean Energy Package, but inconsistent communication protocols among member states (e.g., OpenADR 3.0 in The Netherlands vs. localized standards in France) reduce the cross-border VPP coordination efficiency. China adopts a “national blueprint + regional pilot” model, where Shenzhen’s VPP achieves a 1000 MW regulation capacity through dedicated policies, yet marketization gaps between Guangdong (focused on spot markets) and Yunnan (reliant on administrative directives) limit interprovincial resource sharing. In the U.S., California’s Net Billing 3.0 enhances VPP economics via dynamic pricing, while Texas restricts participation in ancillary markets, stifling functional expansion. Emerging markets like Vietnam face stalled investments due to the absence of technical standards for VPP grid integration [80].

Current regulatory initiatives for VPPs include national and regional efforts such as China’s release of four industry standards—like the Technical Specifications for VPP Resource Configuration and Assessment—and pilot projects in Changzhou and the Yangtze River Delta enabling cross-regional interoperability via unified protocols (IEC 61850). Guangdong supports full VPP participation in spot markets using edge cluster technology, while the EU’s Energy Data Space promotes transnational data sharing despite privacy challenges. To harmonize standards, international communication protocols (e.g., OpenADR 3.0), global grid testing norms, and market mechanisms like the EU’s layered auction model are recommended. Legal frameworks should recognize VPPs as independent entities (as in Germany) and encourage cross-border cooperation (e.g., APEC). Future development includes expanding belt and road pilots, promoting ISO technical guidelines on VPPs, and forming a Global VPP Alliance (GVPPA) that uses blockchain for decentralized, transnational resource coordination.

6. Conclusions

As key technological carriers of new types of power systems, VPPs have made systematic breakthroughs in recent years in terms of technology research and development, resource integration, and market mechanism explorations. On the technical front, the core progress has focused on the diverse innovations and the intelligent upgrading of optimized dispatch models. Traditional dispatch frameworks are gradually being replaced by multitime-scale coordinated optimization models, which effectively address the output volatility of DERs and the uncertainty of load demands through stochastic and robust optimization methods. The deep integration of artificial intelligence technologies has significantly improved prediction accuracy and decision-making efficiency. Reinforcement learning frameworks and adaptive algorithms have provided new paradigms for multiobjective optimization in dynamic scenarios, making dispatch strategies more robust in complex market environments and grid demands. Moreover, the introduction of blockchain technology has solved the challenges of multiparty data sharing and trust collaboration, laying a technical foundation for VPPs to participate in cross-regional energy transactions. Another important achievement is the improvement of resource modelling and evaluation systems. The proposal of multidimensional dynamic indicators (such as the regulation direction, response time, and ramp rate) and the construction of quantitative models for aggregated response capabilities have provided a fine-grained decision-making basis for grid dispatch, promoting the transition of VPPs from theoretical validation to commercial implementation. In terms of the market and policy, differentiated development paths have gradually emerged both domestically and internationally. Europe and America focus on market-oriented mechanisms, increasing the economic value of distributed resources through electricity spot and ancillary service markets. China, on the other hand, relies on top-level policy design and local pilot demonstrations to explore synergistic models of industrial load transformation and city-level resource aggregation. The iterative upgrading of policy tools (such as carbon market integration and demand response compensation mechanisms) has provided institutional guarantees for VPP business model innovations, but there is still a need to address issues such as the low user participation and uneven cross-party benefit distribution. Overall, VPPs have shown significant potential in enhancing the integration of renewable energy, increasing grid flexibility, and promoting the energy structure transformation. However, their technological maturity and large-scale application still need to break through core bottlenecks, such as heterogeneous resource integration, algorithm complexity, and market rule adaptation [81].

The development of VPPs will be driven by new technologies such as 5G, big data, and artificial intelligence (AI) [82]. The future development of VPPs will be deeply integrated into the overall framework of the energy system intelligence and low-carbon transformation. Technical evolution: The focus will be on strengthening multienergy coordination and cross-domain interaction capabilities. At the technical level, new digital technologies (such as 5G, edge computing, and quantum communication) will drive communication architectures toward low-latency and high-reliability upgrades, supporting the real-time perception and millisecond-level response of massive heterogeneous resources. The future of VPP response capability assessments and optimization dispatch will revolve around dynamism, intelligence, and collaboration [83]. A response capability assessment will achieve real-time perception and dynamic correction through edge computing, digital twins, and graph neural networks (GNNs), quantifying the generalized flexibility of multienergy flow resources [84]. Optimization dispatch relies on multiagent collaboration, reinforcement learning, and carbon-electricity coordinated models to break through the centralized decision-making problem.