Renewable-only Virtual Power Plant (RVPP)

• Renewable-only Virtual Power Plants are defined as aggregators composed solely of renewable resources and clean flexibility, eliminating conventional fossil backup.
• They integrate non-dispatchable renewables like wind and PV with dispatchable sources such as hydro, CSP with storage, and demand response for enhanced operational stability.
• Robust optimization techniques, sequential market bidding, and marginal contribution methods ensure optimal profit allocation amid price, production, and demand uncertainties.

Searching arXiv for recent and related papers on renewable-only/renewable-based VPPs, robust bidding, multi-market participation, and control. A Renewable-only Virtual Power Plant (RVPP) is a Virtual Power Plant whose portfolio is entirely or predominantly composed of renewable energy sources, complemented by “clean” flexibility on both the supply and demand side. In the formulation centered on southern Spain, the RVPP aggregates non-dispatchable RES such as wind farms and solar PV, dispatchable RES such as hydro and biomass plants, concentrated solar power with thermal storage, and flexible demand, all coordinated by an operator that collects technical and forecast data, optimizes schedules and bids in several markets, and allocates profits among technologies via a marginal contribution method (Nemati et al., 14 Oct 2025). Related work uses closely aligned notions such as “RES-only Virtual Power Plant” and “renewable-only virtual power plants” for portfolios that exclude fossil units and rely on renewable generation, storage, and flexible demand for energy balancing, reserve provision, and market participation (Nemati et al., 2024). This suggests that the defining feature of the RVPP is not merely renewable generation, but the replacement of conventional backup by renewable and demand-side flexibility across planning, control, and settlement layers.

1. Definition and scope

In the southern Spain formulation, an RVPP is operated as a price-taker whose generation units are all RES: hydro, biomass, concentrated solar power, wind, and PV, while flexibility comes from dispatchable RES, CSP thermal storage, and demand response (Nemati et al., 14 Oct 2025). The architecture is organized into three levels: the asset level, where units $u \in \mathscr{U}$ include dispatchable RES, non-dispatchable RES, CSP with solar field and thermal storage, and flexible demand; the RVPP operator level, where optimization and profit allocation are performed; and the market level, where the day-ahead market, secondary reserve market, and intra-day markets are cleared (Nemati et al., 14 Oct 2025). In related robust bidding models, the same concept is rendered more narrowly as an aggregation of multiple non-dispatchable renewable energy sources and flexible demand, all connected at a single node, with no conventional thermal units (Nemati et al., 2024).

The distinction from conventional VPPs is explicit. A conventional VPP may include thermal or fossil units and electrical storage systems charged from the grid, whereas the renewable-based VPP considered here uses no fossil units and must achieve robustness with renewable and demand-side flexibility alone (Nemati et al., 14 Oct 2025). Earlier network-constrained formulations of RES-based VPPs are already close to this definition: they aggregate hydro, biomass, wind, PV, solar thermal with thermal storage, and flexible demand, with no fossil-fuel units included (Oladimeji et al., 2021). A stricter RVPP interpretation may exclude even dispatchable renewable combustion analogues, but the literature repeatedly treats hydro, biomass, solar thermal with storage, and pumped-storage hydro as compatible with renewable-only aggregation when they provide flexibility without fossil backup (Oladimeji et al., 2021).

A recurring misconception is that “renewable-only” implies only wind and PV. The cited formulations do not support that restriction. In the principal market-participation model, renewable-only operation explicitly includes dispatchable RES, CSP with thermal storage, and flexible demand (Nemati et al., 14 Oct 2025). In scenario-based high-RES system studies, PV and wind are identified as the renewable backbone, but solar thermal and pumped hydro become important to cover the last range of integration because they provide a high flexibility, which is crucial for high share (Collados-Rodríguez et al., 2021). This suggests that renewable-only portfolios are operationally defined by non-fossil energy and flexibility, not by the absence of dispatchability.

2. Resource portfolio and internal operating roles

The RVPP resource portfolio in the southern Spain case comprises hydro, biomass, wind, PV, CSP with thermal storage, and flexible demand (Nemati et al., 14 Oct 2025). Non-dispatchable RES supply low-cost energy but are weather-driven and have limited reserve capability. Their operating envelope is represented through bounds such as

$$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$

where $P_{r,t}$ is uncertain and forecast-dependent (Nemati et al., 14 Oct 2025). In the robust bidding literature focused on RES-only VPPs without dispatchable generation, wind and PV likewise appear as uncertain upper-bounded sources whose economic value depends on forecast credibility and reserve coupling (Nemati et al., 2024).

Dispatchable RES play the stabilizing role. Hydro and biomass are modeled with commitment binaries and bounds of the form

$$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$

with hydro contributing fast ramping and daily energy-constrained flexibility, and biomass behaving like a thermal unit but with renewable fuel (Nemati et al., 14 Oct 2025). In the marginal contribution analysis, dispatchable RES are the most stable and highest normalized contributors, with $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$ and allocated profit $\sim 43\text{–}45\ \mathrm{k€}$ across optimistic, balanced, and pessimistic strategies (Nemati et al., 14 Oct 2025). This indicates that renewable-only portfolios are often economically anchored by flexible renewable capacity rather than by the largest variable-energy block.

CSP with thermal storage is modeled as a semi-dispatchable renewable. The solar field thermal output satisfies

$0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$

the turbine energy balance is written as

$$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$

and storage dynamics follow

$$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$

Its role is to shift solar energy over time, provide upward reserve during hours with high solar input, and smooth ND-RES fluctuations (Nemati et al., 14 Oct 2025). Earlier network-constrained RES-based VPP work makes a closely related point, emphasizing that realistic solar thermal modeling with piece-wise linear efficiency and thermal storage materially affects optimal timing, efficiency, and profitability (Oladimeji et al., 2021).

Flexible demand completes the renewable-only flexibility stack. In the main formulation, each demand $d$ chooses exactly one profile $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$0 through

$$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$1

must remain above the chosen profile-dependent minimum,

$$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$2

and can provide upward and downward reserve within

$$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$3

Its role is to adjust consumption upward or downward to balance supply and provide reserve, transforming surplus ND-RES into useful consumption or reducing load when RES is scarce (Nemati et al., 14 Oct 2025). In the southern Spain study, increasing FD flexibility from $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$4 to $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$5 improves profit from $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$6 to $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$7 under the optimistic strategy, from $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$8 to $$p_{r,t} + r_{r,t}^{\uparrow} \le P_{r,t}, \quad P_{r} \le p_{r,t} - r_{r,t}^{\downarrow},$$9 under the balanced strategy, and from $P_{r,t}$0 to $P_{r,t}$1 under the pessimistic strategy (Nemati et al., 14 Oct 2025). A plausible implication is that demand-side elasticity is one of the strongest levers for keeping renewable-only portfolios viable under conservative uncertainty treatment.

3. Multi-market participation and bidding structure

The RVPP in the principal formulation participates in the day-ahead market for energy, the secondary reserve market for upward and downward reserves, and three intra-day markets for energy re-trading (Nemati et al., 14 Oct 2025). The deterministic objective in the joint DAM+SRM problem is

$P_{r,t}$2

where $P_{r,t}$3 is net traded energy, $P_{r,t}$4 and $P_{r,t}$5 are offered reserve capacities, and $P_{r,t}$6 are operating costs (Nemati et al., 14 Oct 2025). For SRM-only stages, energy revenues are fixed by DAM outcomes; for IDM sessions, reserve is omitted and only energy re-trading remains active (Nemati et al., 14 Oct 2025).

A defining modeling feature is the coupling between scheduled energy and reserve activation. The supply-demand balance is enforced across activation scenarios through the compact expression

$P_{r,t}$7

with $P_{r,t}$8 and $P_{r,t}$9 (Nemati et al., 14 Oct 2025). This enforces feasibility under no activation, upward activation, and downward activation using the same schedule. Similar coupling of energy and reserve bids appears in renewable-only robust bidding models for simultaneous day-ahead and secondary reserve market participation (Nemati et al., 2024).

Sequential market participation has also been developed explicitly. In the sequential framework covering day-ahead, secondary reserve, and intra-day markets, the RVPP solves a sequence of robust MILPs aligned with gate closures, using accepted bids from earlier markets as fixed parameters for later ones (Nemati et al., 2024). That formulation emphasizes asymmetric uncertainty in prices, production, and demand, and reports that robust scheduling shifts energy from earlier commitments toward later intra-day positions when forecast accuracy improves (Nemati et al., 2024). This is consistent with the southern Spain finding that DAM+SRM provide the main revenues while IDM sessions primarily adjust energy to updated forecasts and price signals (Nemati et al., 14 Oct 2025).

A misconception sometimes encountered is that renewable-only VPPs can profit mainly from reserve markets while treating energy as secondary. The cited results do not support that as a general rule. In the southern Spain case, DAM+SRM are the main revenues, and as strategy becomes more conservative, DAM energy bids and reserve bids are reduced while IDM revenue increases significantly, for example by $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$0 in IDM#1 pessimistic versus optimistic, yet total profit decreases and can become negative (Nemati et al., 14 Oct 2025). This suggests that reserve value is contingent on retaining enough controllable headroom after covering energy commitments.

4. Uncertainty modeling and robust optimization

The main market-participation model treats multiple uncertainties simultaneously: day-ahead and intra-day energy prices, reserve prices, wind production, PV production, CSP solar field thermal output, and flexible demand profile consumption (Nemati et al., 14 Oct 2025). Forecasts are represented as bounds derived from historical $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$1 percentiles, and budgets $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$2 control how many time periods can deviate to worst-case values (Nemati et al., 14 Oct 2025). The compact two-stage robust problem is

$$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$3

with $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$4 corresponding to the deterministic case and higher $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$5 producing more conservative schedules (Nemati et al., 14 Oct 2025). The min–max problem is reformulated as a single-level MILP via strong duality (Nemati et al., 14 Oct 2025).

The observed economic trade-off is explicit. In the multi-market southern Spain case, total profit falls from $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$6 under the optimistic strategy to $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$7 under the balanced strategy and to $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$8 under the pessimistic strategy, while revenues drop by $$p_{c,t} + r_{c,t}^{\uparrow} \le \bar P_{c} u_{c,t}, \quad P_{c} u_{c,t} \le p_{c,t} - r_{c,t}^{\downarrow},$$9 and $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$0 versus optimistic and costs rise by $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$1 and $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$2 (Nemati et al., 14 Oct 2025). Thus, robustification is not merely a constraint-tightening device; it reconfigures the resource mix actually used in bids, pushing the RVPP toward hydro, CSP, and flexible demand when uncertainty budgets grow.

Several related formulations deepen this theme. A single-level robust bidding model for a RES-only VPP in simultaneous day-ahead and secondary reserve markets explicitly maximizes worst-case profit under uncertainty in prices, ND-RES production, and flexible demand, with uncertainty budgets $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$3, $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$4, $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$5, $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$6, and $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$7 selecting the hours exposed to worst-case deviations (Nemati et al., 2024). In out-of-sample comparison, that profit-robust model attains net profits close to a much more complex multi-level adaptive robust model while solving in under $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$8 seconds instead of up to $$\rho_u \sim 0.65\text{–}0.68\ \mathrm{k€}/\mathrm{MW}$$9 minutes (Nemati et al., 2024). Another sequential-markets formulation uses asymmetric intervals around medians for prices and production, with global budgets over the horizon rather than hourly budgets, and reports net profit improvements of $\sim 43\text{–}45\ \mathrm{k€}$0, $\sim 43\text{–}45\ \mathrm{k€}$1, $\sim 43\text{–}45\ \mathrm{k€}$2, $\sim 43\text{–}45\ \mathrm{k€}$3, and $\sim 43\text{–}45\ \mathrm{k€}$4 for $\sim 43\text{–}45\ \mathrm{k€}$5 relative to an earlier robust approach (Nemati et al., 2024). These results support the broader interpretation that renewable-only bidding benefits from explicitly asymmetric uncertainty sets rather than symmetric dispersion models.

At finer temporal resolution, a multi-bound robust optimization framework extends this logic to quarter-hourly scheduling. It differentiates frequent moderate deviations from rare extreme ones through several deviation bounds and several corresponding budgets, and reports that normalized absolute differences between hourly and $\sim 43\text{–}45\ \mathrm{k€}$6-minute schedules are $\sim 43\text{–}45\ \mathrm{k€}$7 for day-ahead traded energy, $\sim 43\text{–}45\ \mathrm{k€}$8 for upward reserve, and $\sim 43\text{–}45\ \mathrm{k€}$9 for downward reserve (Nemati et al., 16 Feb 2026). Relative to classic robust optimization, the proposed multi-bound approach increases profit by $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$0 across the considered strategies (Nemati et al., 16 Feb 2026). A plausible implication is that time resolution and uncertainty geometry are not secondary implementation details for RVPPs; they alter both traded volumes and the profitability frontier.

A different branch of work replaces budgeted robust optimization with distributionally robust model predictive control under price uncertainty. In a renewables-batteries-buildings VPP setting aligned with renewable-only operation, a Wasserstein-ball DR-MPC adds a penalty $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$1 to the stage cost, producing revenue gains of up to $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$2 for small radii and revenue losses for larger, overly conservative radii (Recke et al., 14 May 2026). That framework is price-only in its robustification, whereas the multi-market RVPP formulations robustify generation and demand as well. This suggests a methodological distinction between market-centric robust bidding and receding-horizon operational control.

5. Flexibility, feasibility, and control architectures

The literature surrounding RVPPs treats flexibility as more than reserve headroom. One line of work distinguishes the Feasible Operating Region (FOR), the set of all feasible dispatch power points of a VPP, from the Flexibility Operating Region (FXOR), the set of achievable ancillary power points within a given time $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$3 from a dispatch point (Riaz et al., 2019). In that formulation, fast resources such as PV inverters, batteries, and flexible loads expand short-horizon FXOR, while slower resources constrain it for small $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$4 even if they belong to the static FOR (Riaz et al., 2019). Network topology and flexible load placement materially affect both regions (Riaz et al., 2019). Although this work is not specific to renewable-only portfolios, it offers a precise vocabulary for interpreting why dispatchable RES, thermal storage, and flexible demand are disproportionately important in renewable-only operation: they enlarge the dynamic region the aggregator can actually reach.

Dynamic control perspectives sharpen the same point. The Dynamic Virtual Power Plant concept treats a RES-based VPP as a set of renewable units plus control and operation procedures for local regulation, ancillary services, and interaction with neighboring grid elements (Marinescu et al., 2021). In that framing, the DVPP is composed of already installed RESs and demands that can provide some level of flexibility, and can emulate aggregate frequency and voltage response through coordinated control (Marinescu et al., 2021). This suggests that an RVPP need not be understood purely as a market aggregator; it can also be a dynamically coordinated renewable block whose aggregate behavior is engineered across local, global, and economic layers.

Building-centric and decentralized implementations show how this can be operationalized at lower voltage levels. A building VPP architecture aggregates rooftop PV, building loads, and feeder-level battery energy storage through interactions between autonomous BEMSs and a BVPP EMS, with local controllers optimizing appliance schedules while the BVPP EMS aggregates net loads, schedules storage, and interfaces with utility and markets (Luo et al., 2020). The case study reports a commercial BVPP with $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$5 buildings, rooftop PV, and a feeder BESS earning $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$6 goes to the BVPP operator and $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$7 is shared among buildings (Luo et al., 2020). While not formulated as a strict renewable-only market-bidding RVPP, it illustrates a decomposed control structure directly compatible with renewable-only aggregation.

Decentralized transactive control offers another design line. A blockchain-based decentralized VPP platform with distributed renewables, energy storage, and flexible loads formulates local battery dynamics

$0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$8

and user-level power balance including peer-to-peer trades (Yang et al., 2021). In a $0 \le p_{\theta,t}^{SF} \le P_{\theta,t}^{SF},$9-user study, individual cost reductions range from $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$0 to $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$1, while aggregate system cost falls by $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$2 (Yang et al., 2021). The data block explicitly states that a renewable-only VPP is essentially the special case where all generation is from distributed renewables and firm capacity comes from storage and flexible demand rather than fossil or other non-renewable sources (Yang et al., 2021). This suggests that control decentralization and privacy-preserving coordination are not peripheral to RVPPs, especially when portfolios are made of many independent prosumers.

6. Profit allocation, valuation of technologies, and recurring controversies

The southern Spain study allocates profit using a marginal contribution method simpler than the Shapley value but still contribution-based (Nemati et al., 14 Oct 2025). Defining the normalized marginal contribution

$$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$3

the incremental aggregation profit

$$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$4

and the allocation rule

$$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$5

the method is budget-balanced and capacity-weighted, and reflects the actual role of each technology in energy and reserve provision under a given uncertainty strategy (Nemati et al., 14 Oct 2025).

The resulting allocations clarify the internal economics of renewable-only aggregation. Dispatchable RES show the highest normalized marginal contributions and relatively stable allocated profits across strategies; ND-RES contributions are more volatile and decrease by $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$6 to $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$7 versus optimistic under more conservative strategies; CSP has moderate contribution buffered by thermal storage; and flexible demand can have negative solo profit but high normalized marginal contribution, making its allocation less negative than standalone operation (Nemati et al., 14 Oct 2025). This directly challenges a common intuition that technologies should be rewarded in proportion to their average energy output. The cited method instead rewards coalition value creation under uncertainty and reserve co-provision.

Another recurring controversy concerns whether renewable-only portfolios must include batteries to be credible. The principal market study does not include electrical storage systems; flexibility relies on CSP thermal storage, hydro, biomass, and flexible demand (Nemati et al., 14 Oct 2025). Other works explicitly define renewable-only VPPs as renewables plus batteries plus buildings, with no conventional dispatchable generation, and show that distributionally robust MPC can improve economic performance modestly under price uncertainty (Recke et al., 14 May 2026). A third line models solar, stationary batteries, EV charging stations, and controllable loads, reporting reductions of about $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$8 in energy purchased from the grid, about $$\frac{p_{\theta,t}}{\eta_{\theta}} = p_{\theta,t}^{SF} + p_{\theta,t}^{TS,-} - p_{\theta,t}^{TS,+} - K_{\theta} v_{\theta,t} \bar P_{\theta},$$9 in distribution losses, about $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$0 in grid purchase cost, and about $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$1 in EV charging cost, with the VPP supplying $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$2 to the distribution system and earning $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$3 profit (Harsh et al., 2024). The literature therefore does not support a single canonical storage choice. Rather, renewable-only feasibility can be achieved through multiple flexibility mixes, with batteries as one option, not a definitional requirement.

Forecast provision and information structure create another underappreciated issue. In a community-based VPP where uncertain renewable outputs $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$4 are predicted and sold to consumers, the decentralized prediction provision algorithm yields a demand gap relative to centralized prediction with zero expectation and bounded variance, while preserving privacy and reducing communication (Chen et al., 2020). Though this framework is not a market-bidding RVPP per se, it points to a deeper controversy: whether centralized forecasting is necessary for renewable-only coordination. The cited result suggests that decentralized prediction architectures can approximate centralized performance in expectation (Chen et al., 2020).

7. Design implications and future directions

Several practical design principles recur across the cited RVPP work. First, the portfolio mix matters. The southern Spain study recommends combining large ND-RES capacity with hydro and biomass as dispatchable RES, CSP with thermal storage, and substantial flexible demand, with $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$5 flexible share where possible (Nemati et al., 14 Oct 2025). Scenario studies of high-RES systems reinforce that PV and wind provide the renewable backbone but lack the flexibility needed to achieve a very high share, while solar thermal and pumped hydro become important for the last range of integration (Collados-Rodríguez et al., 2021). A plausible implication is that “renewable-only” does not obviate the need for dispatchability; it relocates dispatchability into non-fossil technologies and responsive demand.

Second, market strategy should be sequential and forecast-aware. The southern Spain study uses DAM for baseline commitments, SRM for reserve revenues, and IDMs aggressively under higher uncertainty to correct positions and reduce imbalance risk (Nemati et al., 14 Oct 2025). The sequential robust bidding framework formalizes the same idea across DAM, SRM, and multiple IDMs, showing that later markets are valuable because they exploit improved forecasts without enabling pure speculation (Nemati et al., 2024).

Third, uncertainty treatment should be tuned rather than maximized. Budgeted robust optimization exposes a direct conservatism-profit trade-off (Nemati et al., 14 Oct 2025). Multi-bound uncertainty sets at $$e_{\theta,t}^{TS} = e_{\theta,t-1}^{TS} + p_{\theta,t}^{TS,+} \eta_{\theta}^{TS,+} \Delta t - \frac{p_{\theta,t}^{TS,-}\Delta t}{\eta_{\theta}^{TS,-}}.$$6-minute resolution improve profitability relative to classic robust optimization (Nemati et al., 16 Feb 2026). Distributionally robust MPC improves performance only for small ambiguity radii, with larger radii becoming overly conservative (Recke et al., 14 May 2026). Across methodologies, overly pessimistic protection can destroy profitability.

Finally, multiple extensions remain open. The southern Spain paper notes or implies the omission of network constraints, batteries, full reserve activation dynamics, and price impact (Nemati et al., 14 Oct 2025). Related work proposes incorporating electrical storage, distribution network constraints, DSO coordination, distributionally robust optimization, scenario-based stochastic models, and dynamic participation incentives in profit sharing (Nemati et al., 14 Oct 2025). Dynamic and network-aware perspectives further suggest integrating AC or DC power-flow constraints, TSO/DSO interface modeling, and ancillary service dynamics into RVPP design (Riaz et al., 2019, Marinescu et al., 2021).

Taken together, the literature portrays the RVPP as a technically heterogeneous but conceptually coherent entity: an aggregation of renewable generation and clean flexibility operated as a single market-facing and potentially grid-supporting unit. Its central problems are the valuation and orchestration of renewable flexibility, the consistent coupling of energy and reserve across markets, and the management of asymmetric, multi-source uncertainty without recourse to fossil backup (Nemati et al., 14 Oct 2025, Nemati et al., 2024, Nemati et al., 2024).