Marginal cost of energy volatility in the Brazilian Interconnected Power System

From 2018 to 2022, on average, 70% of the Brazilian effective electric generation
was produced by hydropower, 10% by wind power and 20% by thermal power plants.
Over the last five years, Brazil suffered from a series of severe droughts. As a result,
hydropower generation was reduced, but demand growth was also declined as results
of the COVID-19 pandemic and economic recession. From 2012 to 2022, the
Brazilian reservoir system operated with, on average, only 40% of the active storage,
but storage recovered to normal levels in the first three months of 2022. Despite large
capacity of storage reservoirs, high volatility of the marginal cost of energy was
observed in recent years. In this paper we used two optimization models, NEWAVE
and HIDROTERM. These two models were previously developed for mid-range
planning of the operation of the Brazilian interconnected power system. We used
these two models to optimize the operation and compared the results with observed
operational records for the period of 2018-2022. NEWAVE is a stochastic dual
dynamic programming model which aggregates the system into 4 subsystems and 12
equivalent reservoirs. HIDROTERM is a nonlinear programming model that
considers each of the 167 individual hydropower plants of the system. The main
purposes of the comparison are to assess cooperation opportunities with the use of
both models, and better understand the impacts of increasing uncertainties,
seasonality of inflows and winds, demand forecasts, decisions about storage in
reservoirs and thermal production on energy prices.