Vitor Luiz de Matos 
Paulo Vitor Larroyd
Erlon Cristian Finardi


For real-life problems arising in the energy sector,
we give some insight on the trade-off that needs to be found between solving 
until optimality a simple representation of the problem 
and stopping prematurely a more realistic model that is too difficult to solve.

As this study is part of an ongoing project on Hydrothermal 
Scheduling for the Brazilian power system, we will focus 
on this type of problems. 

The goal of the Hydrothermal Scheduling problem 
is to define a policy for the use of water. As such, the model 
plays a fundamental role in hydro-dominated power systems, like Brazil's.

Since the Brazilian system has hundreds of hydro plants, various simplifications 
are made in their modeling in order to reduce the computational burden. 
Such simplifications result in inefficiencies that
increase not only operational costs but also the risk of load shedding. 
We discuss in particular the impact of a simplification made 
in the official models used by the Independent System Operator, 
which consists in aggregate Hydro Power Plants into a single generation
unit, the so-called Energy Equivalent Reservoir.