Introduction and objective
Despite the problem of insufficient power being a real condition of the SIN (National Interconnected System), the LRCAP 2026 (Brazilian Electricity Regulatory Agency Contract 2026) ended up being criticized by several associations, in addition to raising doubts about the necessity and the amount tendered.
This article presents relevant points that should be observed for sound public policy. The evolution of the power adequacy problem, along with the arguments of the ONS/EPE (Brazilian National System Operator/Energy Research Company), will be addressed, incorporating pertinent observations, mainly regarding the path adopted.
The government's decision to hold an auction that burdens consumers and indirectly society should be preceded by an analysis considering four points:
- Power requirements and sizing;
- Auction drawing
- Technological and systemic alternatives
- Socioeconomic impacts.
This article discusses these points, highlighting shortcomings that could harm electricity consumers and society in general.
Sizing and power
Since electrical energy is a commodity that is consumed at the same time it is produced, it has distinct characteristics compared to other energy commodities such as coal, oil, etc. This characteristic necessitates a more complex operation, as it requires a network of wires to connect the source to the point of consumption.
Monitoring demand and the resulting dispatch of power plants is done second by second and is part of the real-time operation of the system operator, which in Brazil is the ONS. The imbalance in the energy balance that flows through the grid can be measured through the frequency variation of the system.
There are times when demand is high, and any shortage in the generation system, as well as an unexpected increase in load, can cause problems for the system. At these times, it is necessary to have an operational reserve to avoid load shedding, both in terms of power and energy.
Therefore, the need for a reserve exists, but its size depends on the risk the consumer is willing to take to avoid being without electricity during this period and how much they are willing to pay for this "insurance".
Since this question is not asked of consumers in Brazil, the models for sizing this reserve were developed centrally by the system operator.
History of the models
The models for calculating the reserve have gone through several phases since Eletrobrás coordinated the operation.
First phase – classic deterministic criterion (until the 2000s)
For decades, the planning of the SIN (National Interconnected System) under the coordination of the GCOI (General Coordination of Internal Operations) used a relatively simple criterion:
RPO = 5% of total load + largest machine
Since the reserve power (RPO) had to cover the randomness of the load, the system had to support an additional 5% increase at peak demand. This value represented a statistic observed over the years for the four regions of Brazil. Furthermore, it was necessary to cover the loss of the largest machine in the SIN (National Interconnected System), which was a unit of Itaipu.
This criterion stemmed from the tradition of classic hydroelectric systems where the electricity matrix was dominated by hydroelectric plants with low generation variability and little intermittency. An improvement was made to attempt a correlation between load increase events and generator failure using generator failure rates, giving rise to the concept of the largest probabilistic machine.
Second phase – classic semi-deterministic criterion (2000 – 2020)
With the massive influx of wind power in the Northeast region, the ONS (National System Operator), which replaced the GCOI (Intermunicipal Consortium Group), began introducing additional reserves. This was necessary because, unlike hydroelectric plants with reservoirs and thermal plants with reliable available capacity, the new wind farms might not have wind available during peak hours. A statistical analysis was conducted, and as a result, it was determined that the reserve should be increased to address this uncertainty.
In the case of the Northeast region, the added value would be 6% of the total wind power generation capacity of this region. For the South region, the value would be 15%. This difference is due to the variability of the wind in the South, which is greater than in the Northeast region where the winds are more constant.
Although it was a statistically based model, a probabilistic assessment of the combined effects was not performed.
Third phase – probabilistic criterion (current model)
Today, Brazil has migrated to a fully probabilistic method, similar to that used in systems like PJM or Europe. Power reserve is now calculated considering the probability of failures and variability of the sources.
This model includes:
- Probability of power plant failure
Similar to previous models that introduced the highest probability rate, this model includes downtime rates and mean repair time, but these must coincide with the peak load period.
- Variability of renewables
Wind and solar energy will be represented by monthly probability density functions differentiated by region (Northeast coast, Northeast interior, South).
- Operating scenarios
Using probability and density functions, combinations of load, renewable generation, unit failures, and outages are simulated. These combinations are simulated using the Monte Carlo method, which is widely employed in reliability studies.
Laughter Criterion
Today, power reserve is no longer determined as a percentage of load or generation, but based on risk criteria for service failure. The main indicator is the LOLP (Loss of Load Probability), which represents the probability of not meeting the load, and in this particular case, during peak hours of the National Interconnected System (SIN).
Another important point was abandoning the logic of using installed capacity as a proxy for capacity contribution and instead using installed capacity as a proxy for capacity contribution (ELCC – Effective Load Carrying Capacity). In reality, the notion of how much a given technology actually contributes to meeting peak demand is incorporated. For example, thermal and hydroelectric plants with reservoirs have a high contribution, while wind power contributes very little (5 to 20%) and solar power almost zero if the peak occurs at night.
Power flow
For this LRCAP, there were new updates to the model that were presented by ONS at the end of last year through a technical note.[1] Defining the need for flow from the transmission when power is required at the system's endpoint:
- methodology used to assess power requirements
- SIN operating premises
- network topology considered
- reliability criteria
- treatment of renewables
- assumptions of power plant unavailability
- load and expansion assumptions.
The key assumptions are the explicit considerations of forced and scheduled outages, as well as regional electrical constraints, which will define the effective available capacity.
There was concern about the capacity to handle power generation in several regions, as the response to a power shortage at a given time depends not only on the conditions of the generation system, but also on the transmission system.
The need for coupling generation with transmission to assess remaining capacity and ensure power availability for events that may utilize it is not in question. The problem is that the ONS (National System Operator) established, based on the structural nature of the reserve, the worst-case scenario for the PAR/PEL (Planned Energy Reduction/Power Generation Plan) horizon at each busbar, sub-region, and region, resulting in a rather conservative approach.
___________________________________
[1] Methodology, Premises and Criteria for Capacity Reserve Auctions in the form of Power – LRCAP 2026, NT-ONS DPL 0114/2025/ EPE-DEE-RE-093/2025-r0
The problem is that when thermal generation is defined to provide power in a specific bus, sub-region, or even region, it is necessary to verify the local availability of fuel.
In the case of natural gas power plants, access becomes more critical, requiring pipeline infrastructure or even gasification systems that must be located on the coast.
CNPE Criteria
Resolution 29/19, established in 2019, defined the explicit risk of insufficient power (LOLP) and the expected value conditioned on a given confidence level (CVaR). It also establishes that the Ministry of Mines and Energy (MME) will periodically determine these parameters, which represent the degree of risk aversion imposed on electricity consumers.
These criteria are related to the methodology that has already been used to calculate the physical guarantee of hydroelectric plants that use NEWAVE and SUSHIO. Therefore, it is a structural solution that basically depends on the scenarios used in planning, that is, the vision of EPE and ONS regarding the development of the electricity sector is reflected in what is currently dimensioned as power reserve.
For this auction, an annual LOLP of 5% and a CVaR of 5% of the monthly PNS were used, with 5% of the maximum instantaneous demand. This is the crucial point of the methodology where the sensitivity to the adopted assumptions makes the result strongly dependent on the modeling. In reality, the LOLP does not directly affect the result, and what increases the most is the monthly CVaR. In simplified terms, the criterion states that the system can fail in up to 5% of scenarios, but when it fails, the average power loss in the worst cases cannot exceed 5% of the maximum load.
The problem is which criteria to adopt, given that the consumer is the one who pays the bill. If the criteria are too conservative, the cost ends up being too high. Recent minutes from the CNPE (National Energy Policy Council) express concern, based on PEN (National Energy Policy) reports, that the LOLP (Lower Current Percentage Rate) and CVaR (Cost Per Receipt) values related to peak demand are increasing, suggesting a possible violation starting in 2026, which motivated the structural need for the LRCAP (Law on Energy Consumption and Preservation).
Another important point is CVaR itself, which originated from VaR (Value at Risk) in the financial market. With reliable statistics, these parameters are interesting within certain limits. However, they depend heavily on assumptions and fall towards the end of the probability density curve where the extremes become evident.
Supply and demand scenario
Based on the historical load growth observed throughout 2024 and early 2025, an average growth of 3,4% was obtained, which was projected for each region, reaching a value of 94,6 GW in 2029. The power requirements assessment used the coincident peak demands obtained from typical load curve profiles.
These were the criteria adopted in the PEN (National Energy Plan) that were preserved for calculating the power requirement of the LRCAP (Local Regulatory Capacity of Public-Price Consumers). It was also observed that the entry of Group A free consumers with loads greater than 500 kW or less through a retail agent from 2024 onwards would alter the structural peak of the SIN (National Interconnected System).
If energy forecasting is already a huge problem in the current context of technological changes and economic scenarios in the country, even greater uncertainty arises in power forecasting. What is observed is that any assumption adopted can significantly change the result, even considering the probabilistic effects in the reliability studies used for LRCAP. New loads with different characteristics, such as data centers, and the alteration of traditional load curves due to technological advancements completely change the framework used.
Offer
Brazil's energy and electricity matrix has changed significantly over the years. The introduction of wind and solar sources has provided significant support for energy generation, but comes with intermittency problems due to the reliance on primary sources.
The power calculation programs used wind and solar generation density functions taken from historical data that have been changing due to climatic variables. This already highlights uncertainty in the medium and long term regarding the best solutions for peak demand requirements in the Brazilian Interconnected System (SIN).
Another dimension of the problem is water supply, which is also suffering from climate change. Inflow scenarios can significantly alter both the energy supplied and the available power.
Climate variables end up dominating energy and power supply scenarios, even in combined modeling of precipitation, wind, and sun, which cannot be independent as in current models.
Predicting climate variables is a crucial point where the use of current stationary models fails to reflect the reality of weather and climate dynamics. In addition to the combination of these variables, we can add the increase in temperature as a determinant on the demand side. These combined variables act to define the peak of the system that current models cannot represent. The analysis of independent maxima is not a good practice and can yield results far from future reality.
Auction as a solution
Because the Brazilian model still lacks price signals, inefficiencies arise as it links power supply decisions to models that have a number of deficiencies, as presented in the previous section. Other solutions found in other markets, such as the energy-only model in Texas or the regulation market (ancillary service) in European countries, end up proving to be more efficient in dealing with uncertainties, with greater participation from generation and consumption agents in solving operational problems.
Given that the Brazilian model still features centralized governance, auctions end up being an important instrument for providing for the reserve. However, we can list some problems observed in these recent auctions.
Technological direction
Specifying thermal power plants as an almost exclusive solution implies the elimination of competition with other technologies. BESS (Built-in-Solid Wastewater Treatment), reversible power plants, and hybrid systems present themselves as viable solutions not listed. This leads to a loss of economic efficiency and the risk of technological lock-in incompatible with the energy transition.
One point that could be raised is that the problem is not only one of power but also the need for a stable, sustained energy supply or even assistance with stability issues. BESS effectively handles ramps, frequency, and ancillary services, but still has economic limitations for longer durations such as multi-day events. Regarding stability, the need for grid-forming technologies may still be under testing.
This mix of auction purposes shows a lack of explicit definition of attributes such as firm power (MW) and duration (h) to allow other technologies to compete. In this specific auction, specific attributes for thermal and hydroelectric plants were designed.
Improving the characterization of this auction would allow: BESS to compete in short blocks, hybrids (solar + BESS, wind + BESS) to enter, and thermal plants to remain where they are actually most efficient.
Reservation price and implicit subsidies
During the auction planning process, there were discussions about the reserve price, which doubled slightly before the auction took place. A high reserve price can lead to unnecessary contracting, guarantee returns on specific technologies, and transfer costs to the consumer without transparency. It is known that a reserve price outside the market can create a risk of non-contracting, but the way it was defined raises questions.
The risk lies not in the reserve price itself, but in the combination of conservative reliability criteria, a lack of transparency in defining capacity demand, and a high reserve price—which, together, can lead to contracting that is structurally above the economic optimum.
Technological and systemic alternatives
To solve the capacity problem, there are now technologies that are more aligned with the energy transition. Gravitational storage and electrochemical storage have gained ground in various countries facing the same difficulties with curtailment, ramp-up, and peak-up. In this context, we can separate storage into centralized and distributed.
Centralized storage
Centralized storage today forms a "modern equivalent" of thermal power plants from a systemic operation standpoint. Despite the Ministry of Mines and Energy (MME) stating that the battery auction is on the agenda, obstacles remain, such as the specific regulations governing the process. ANEEL which does not move forward with CP 39/23. Given that power requirements are urgent according to the ONS (National System Operator), the lack of regulation of the BESS (Balanced Subsystem for Energy Storage) ends up favoring the thermal solution. We can highlight some important characteristics of centralized storage:
- BESS centralized
- Excellent for ancillary services, ramp and hourly arbitration;
- A still relevant limitation in duration for prolonged events;
- Highly modular and with rapid deployment — a key advantage over thermal power plants.
- Flexible location within the National Interconnected System (SIN) adapting to drainage margins.
- Reversible plants
- Structural solution for long-term storage;
- High systemic efficiency and long service life;
- But there are significant obstacles: high CAPEX, licensing costs, and implementation time.
- Location dependent on geography.
- Hybrids (solar/wind + storage)
- They reduce curtailment and increase predictability;
- They introduce a renewable "semi-firmness";
- They can compete directly with thermal power plants if well-structured in the auction.
The current LRCAP design does not explicitly state the reliability attributes (power, duration, availability), which implicitly favors thermal power plants even when technological alternatives already exist.
In addition to the aspects mentioned above, it is important to highlight another relevant characteristic of energy storage: the need for energy for charging. This characteristic, far from being a limitation, allows BESS to act as an active resource in absorbing surplus energy, shifting this energy to periods of higher systemic value, such as peak load.
In this way, storage not only contributes to meeting power demand, but also acts to reduce curtailment, increasing the overall efficiency of the system. It is, therefore, a solution that simultaneously addresses two structural problems experienced in the Brazilian case: excess generation in certain periods and insufficient power in others.
On the other hand, a significant portion of thermal power plants, especially those based on steam cycles or combined cycles, present important operational restrictions associated with start-up time (T-on). These times can exceed several hours, depending on the need to heat boilers, turbines, and other thermal components.
When these start-up times are high, the operation of these plants loses adherence to the intraday dynamics of the system, characterized by steep ramps and rapid variations in renewable generation. In particular, to keep up with the load ramp-up at the end of the day, it may be necessary to anticipate the dispatch of these thermal plants even during the period of high solar generation.
This early dispatch implies, in practice, maintaining thermal generation at times when there is already a surplus of energy in the system, which could worsen the curtailment problem by further displacing renewable generation.
While storage tends to reduce curtailment by absorbing surpluses, certain thermal configurations, due to their operational constraints, can contribute to its expansion by requiring dispatch outside the optimal period of power demand.
An additional relevant aspect refers to the increasing loss of controllability of the SIN (National Interconnected System) during certain operating periods, a phenomenon already identified by the ONS (National System Operator) in its most recent planning studies (PAR/PEL 2025).
This phenomenon has translated into an increase in the number of days with loss of controllability, as indicated in the ONS's prospective studies for the horizon up to 2029. During these periods, the operator has less room to maneuver on the balance between generation and load, which can compromise operational security, especially in scenarios of high variability in renewable sources.
In this context, the curtailment problem becomes just one manifestation of a broader issue: the insufficiency of resources with the capacity for fast and bidirectional modulation, capable of absorbing surpluses and, at the same time, supplying power when needed.
Given that BESS has the capacity to operate both as load and generation, it can absorb excess energy, contributing to reducing curtailment in the maintenance of controllable units in operation.
The evolution of the system suggests a transition from a problem that is predominantly one of power adequacy to an increasingly important problem of flexibility and controllability adequacy.
Distributed storage
The expansion of distributed generation (DG) has brought benefits to the system in terms of energy supplied near load centers, but with its high degree of penetration experienced in recent years, problems in the load curve, such as flow reversal and a ramp-up to peak consumption, have also been causing problems for the operation of the National Interconnected System (SIN).
Just as distributed generation is already a reality, "behind the meter" storage located at or near the point of consumption is a transformation in the logic of the electrical system that is already driving many countries.
Instead of solving the problem in a traditional centralized way, that is, "top-down" (generation → transmission → consumption), there is an approach that:
- directly in the load profile;
- reducing local peaks;
- avoiding or postponing network investments;
- and, most importantly, transforming demand into an active resource.
We can cite the emblematic case of Australia where:
- State programs encourage residential battery batteries;
- Aggregators form VPPs (Virtual Power Plants);
- Battery prices are determined in the market.
- Batteries provide ancillary services as a direct response to demand by shifting the peaks that cause LRCAP.
The distributed solution tackles the problem at its source — the peak load — while the centralized model reacts to it.
The discussion about LRCAP cannot be limited to the choice between thermal and centralized renewable energy sources. There is a third way—distributed storage—that directly impacts the load profile and can reduce the systemic need for power contracting.
By not explicitly incorporating distributed resources and aggregators into the mechanism's design, the current model risks oversizing the need for centralized capacity. Since the auction establishes a 10-year contract, we will inevitably have a "sunk cost" to be paid by generators and consumers, or by the taxpayer.
The ONS (National System Operator) still sees little use of these resources due to the lack of full integration with aggregators and the absence of models for coordinated dispatch.
By not explicitly incorporating distributed energy resources and aggregators into the mechanism's design, the current model risks overestimating the need for centralized capacity.
Socioeconomic impacts
The previous items addressed the intrasectoral impact of the LRCAP (Long-Term Auction for Thermal Power Plants). Since the decision to define the source rested with the Ministry of Mines and Energy (MME), one of the externalities affecting the adopted public policy must be incorporated. This perspective also discusses economic and social efficiency in a broader sense. An auction restricted to thermal power plants not only produces a technological outcome but also determines the distribution of costs and benefits, as well as risks to society.
The first effect is on the energy consumer. Even when the auction presents a discount, the economic problem does not disappear, because the discount occurs within a previously defined universe. The 3rd LRCAP, for example, was focused on fuel oil, diesel oil, and biodiesel; therefore, the competition occurred between thermal power plants within this group, not between all possible reliability solutions. This means that the final price may even fall below the ceiling, but still remain above the social cost of alternatives excluded from the bidding process. This ends up affecting the tariff moderation so strongly advocated by the Ministry of Mines and Energy.
The second effect is the environmental and health externality. When the solution is concentrated in thermal power plants, especially those fueled by liquid or solid fossil fuels, the sector assumes indirect costs linked to emissions, local pollution, intensive logistical use of fuel, and greater exposure to supply chain volatility. Even if some of these costs do not fully appear in the winning bid, they reappear in society in the form of environmental deterioration, pressure on public health, and the future need for more expensive decarbonization.
The third effect is technological stagnation. By contracting power with requirements implicitly designed to favor thermal power plants, the sector postpones the learning curve, industrial scale-up, and regulatory maturity of solutions such as centralized BESS, renewable hybrids with storage, and aggregated distributed resources. This is particularly relevant because the Ministry of Mines and Energy itself has already recognized battery storage as an object of future contracting, which weakens the argument that it is a "premature" technology or outside the institutional radar.
The fourth effect is locational and distributive. Centralized thermal power plants solve a systemic problem from the operator's point of view, but they don't always solve the problem at the electrical source. In many cases, what puts pressure on the system is peak demand, ramp-up, or local grid constraints. In these cases, distributed or semi-distributed solutions can generate greater value per contracted MW, as they simultaneously address load, distribution networks, and reliability.
The problem, therefore, is not in recognizing the role of thermal power plants in the reliability of the National Interconnected System (SIN), but in transforming this role into an almost exclusive solution through regulatory design. When this occurs, the auction ceases to be an instrument for efficient cost discovery and begins to function as a mechanism for pre-selecting winners, with impacts that extend beyond the electricity sector and reach the economy, the territory, and society.
Conclusion
The LRCAP 2026 highlights the need for mechanisms that complement the current pricing model to ensure adequate power capacity for the National Interconnected System (SIN). However, the adopted design—by combining conservative risk criteria, implicit technological direction, and a lack of integration of new alternatives—may lead to a structurally more costly solution, compromising the development of the electricity sector.
The analysis suggests that the challenge lies not in capacity acquisition itself, but in how that capacity is defined, valued, and allocated within the system. The lack of explicit reliability attributes and mechanisms that allow competition between different technologies limits the efficiency of the process.
For example, in determining the power requirement to be contracted, market solutions such as distributed resources, demand response, or even importing energy from other countries to increase the contracted thermoelectric demand were ignored.
Energy storage systems (BESS) are the most suitable and efficient technology to address the challenge of variability in uncontrollable renewable sources and consumption, by providing flexibility and immediate response for instantaneous balance between supply and demand for electricity in the National Interconnected System (SIN), and to meet the requirement for systemic reserve capacity, by shifting wasted renewable energy to the most critical supply periods, and therefore could be widely (or even entirely) used for the purpose set out in the LRCAP.
An aggravating factor in not considering storage technologies, such as BESS, is that, from the perspective of the overall system optimum, these resources could significantly reduce the generation cuts currently observed in renewable plants (curtailment). In addition to meeting ramp and power needs, storage allows for the absorption of energy surpluses and restores the flexibility and controllability of the SIN (Brazilian Interconnected System) — aspects that are becoming increasingly critical in a system with a high share of non-dispatchable generation.
In summary, while the Brazilian National Interconnected System (SIN) is discarding renewable energy (curtailment), it has opted for expensive, inflexible, polluting thermal power plants that are dependent on global geopolitics for the supply of fossil fuels, which will intensify the waste of renewables due to lengthy commissioning ramp-ups.
The opinions and information expressed are the sole responsibility of the author and do not necessarily represent the official position of the author. Canal Solar.
