Energy and hydrology modeling for the Paraná basin
Hydropower in Brazil stands for 64.5% of all
energy sources in the country. The largest
number of hydropower plants is located in the
basins of the Paraná River, which together
have an installed capacity of approximately 40
GW including the world’s largest, in terms of
electricity production, hydropower plant, Itaipu
(ANEEL, 2013).
The wet season during Brazilian summer
(January-March) is the most important for
hydropower. Occasionally extreme events
called El Niño and La Niña connected to sea
surface temperature variability in the Pacific
Ocean affect climate patterns (ATMOS, 1998).
These events could lead to droughts and result
in dramatic consequences for hydropower
production. Rain patterns within the basin
reflect the flows during different seasons and
the largest flows in the Paraná River basin
occur between January and March. The largest
energy inflows from the main tributaries of
Paraná are in Paranaíba and Grande.
This project objective includes the adaption of
the Scania-HBV model for the Paraná basin
and the evaluation of the model. The model
was provided by Thomson Reuters Point
Carbon. Input data was precipitation and
temperature. Through simplified hydrological
processes, the model simulated inflow to the
basin in energy units. Limitations were mainly
connected to measurements and calculations of
input data. Thomson Reuters Point Carbon
uses information about energy inflow for
prediction of future hydropower production
and prices on the energy market. Severe peaks
in prices seem to be connected to long-term
droughts in the wet season and technical
problems in the hydropower system
(Söderberg, 2012).
Natural energy inflow was the target data used
for calibration of the model. The Paraná basin
was divided into five sub-basins: Paraná,
Grande, Paranaíba, Tietê and Paranapanema,
see observed energy inflow for the sub-basins
in Figure 1. Scania-HBV model was adapted
for each sub-basin, calibration period was
chosen from 2005 to 2012. Objective functions
were used to find the best fit between observed
and calculated energy inflow. Evaluation was
done by simulating energy inflow for a
validation period from 2000 to 2005. Finally
the results from the five models were put
together to receive information of energy
inflow for the whole basin.
Inflow per sub-basin (GWh)
13000
11000
9000
7000
5000
3000
1000
-1000
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
-3000
Tiete Qobs
Paranapanema Qobs
Parana
Paranaiba Qobs
Figure 1 Monthly energy natural inflow to rivers of the Paraná basin during period 2000-2012
Grande Qobs
Model results were generally satisfying.
Common characteristic patterns of energy
inflow for each sub-basin and peak events
were captured by the model. Hydrographs for
all the sub-basins showed an abnormal pattern
of natural energy inflow in May 2001. During
this period blackout crisis took place in Brazil.
Reasons
for
the
blackouts
were
mismanagement of the energy sector and lack
of precipitation due to an extreme event, La
Niña which led to low levels in hydroelectric
reservoirs.
Paranáiba, Tietê, Grande, the northernmost
sub-basins, overestimate observed inflow for
2000-2003. Stations selected captured mostly
the highland climate which means generally
lower
temperatures
and
less
evapotranspiration. Another reason could be
that wetter stations than are representative for
the northern sub-basins were chosen. The exact
pattern in the Paraná sub-basin cannot be
captured by the model since it is the result of
subtracting the whole basin with the other subbasins, this leads to odd values (e.g. negative
ones) in the target series as well as comparable
large negative Accumulated difference in
validation period. However the addition of all
sub-basins into one final minimizes the
influence of odd values.
All sub-basins, except one, showed high values
of the Coefficient of determination (r2) which
means good fit between modeled and observed
natural energy inflow. Weekly r2 values for the
sub-basins were of approximately 0.8 (monthly
0.9) in calibration and 0.7 (monthly 0.8) in
validation. The final result was satisfying;
validation showed weekly r2 0.87 (monthly
0.92) in calibration and 0.75 (monthly 0.82) in
Karin Olsson
Agnieszka Duma
Supervisors:
Prof. Cíntia Bertacchi Uvo
Stefan Söderberg
23/6-2013, Lund
Division of Water Resources Engineering
Lund Technical University
Thomson Reuters Point Carbon AS
validation and a relatively low Accumulated
difference of -2106 GWh.
Choice of stations for data collection is a
crucial part when building a trustworthy
model. General the chosen stations were
evenly spread except in centre of the Paraná
basin where no stations that reached the set
demand were found, this might affect a bit the
picture of the weather condition. Quality of
data was carefully checked and both
precipitation and temperature stations had a
high fillness rate of at least 95%. Focusing on
daily target data a relative large part (30-40%)
was estimated. If simulation on a daily basis
should be presented, further improvements
might be needed by collecting a complete data
series of real time daily natural energy inflow.
Other suggestions for improvement include
using other models to verify results and using
automatic calibration tools to receive optimal
parameter combinations.
All steps including validation, evaluation of
meteorological stations and target data, that is
comparison data with other sources, and
quality checks ensure a trustworthy model. The
evaluation of source data, together with good
final values for different objective functions,
and the fact that the calculated energy inflows
capture observed extreme weather conditions
verify that the model accurately reflects reality.
Overall, the Scania-HBV model was
successfully adapted to the Paraná basin and
together with satisfying final results give the
conclusion that the model is reliable and
probably useful to predict future hydropower
production at the Paraná River basin.
Bibliography
ANEEL, 2013. Matriz de Energia Elétrica. [Online]
Available at: http://www.aneel.gov.br/aplicacoes/capacidadebrasil/operacaocapacidadebrasil.asp
[Collected 24 April 2013].
ATMOS, 1998. Washington University, College of Environment. [Online]
Available at: http://www.atmos.washington.edu/1998Q4/211/kana.htm
[Collected 2013].
Söderberg, S., 2012. Hydrological energy modeling, u.o.: Thomson Reuters Point Carbon.