Water values and marginal costs for power generation planning under more restrictive constraints and a more detailed representation of hydro plants

•Assessment of the effect of different types of hydraulic constraints in water values and system marginal costs for long term power generation planning.•Comparison of an individual vs aggregated representation of hydro plants for long term power generation planning.•Assessment of future cost functions obtained when solving the long term power generation problem with the stochastic dual dynamic programming approach (SDDP).•Results for realistic cases of the large-scale Brazilian system.•Discussion on the concepts of water values, incremental water values, and marginal cost of hydro generation. Power generation planning for real large-scale systems is a complex multistage stochastic problem, which requires modeling approximations for the representations of system components and operating constraints in order to be efficiently solved by the stochastic dual dynamic programming (SDDP) approach. This paper presents an analysis of the impacts of a more detailed representation of the hydro plants and the inclusion of more restrictive hydraulic constraints in the recourse functions that are built by the SDDP strategy, as well as in the simulation outputs of this problem. While some results are rather obvious - as the increase in systems costs - others are counterintuitive, such as the decrease in thermal generation levels and water values in some situations. Results are presented for the individualized modeling of the hydro plants in the official models applied for long-term hydrothermal planning in Brazil.