Sediment flow in the Angostura reservoir in Costa Rica

The Angostura Reservoir is located in Costa Rica, east of San Jose. It has been in operation since October 2000. Before it was built, our department took part in a sedimentation study of the reservoir. This included a field trip to the Angostura river.

Picture from the river upstream of the dam site before the reservoir was built (photo: N. Olsen)

Note the stones on the original river bed, which will not be eroded during a flushing.

N. Olsen and L. Hoven visted the reservoir in February 2010 to investigate how well the SSIIM model could compute the sediment flow in the reservoir. Both sediment deposition and flushing of the reservoir are included in the study. Bed surveys of the reservoir has been carried out one or two times each year. The reservoir is flushed regularly.

Picture from the reservoir (photo: N. Olsen)

Looking at the inflow gauging station to the reservoir: Cola Embalse Angostura. (photo: N. Olsen)

Investigating the cohesive properties of the bank sediments (photo: L. Hoven)

Initial computations were done in the spring 2010. One of the main problems was to be able to get a stable free surface algorithm to model the drying of the reservoir during the flushing. In the flushing modelling, the grid initially covered the whole reservoir. As the downstream water level is drawn down, the reservoir partly dries up, leaving a channel with high velocity and sediment concentration.

Some of the results from the flushing modelling are given below. Note that many of the sediment processes in the reservoir have not yet been taken into account when producing these results, for example effects of cohesion, sediment movments on sloping beds, sand slides, hiding/exposure of different particle sizes, bed forms, variations in effective bed roughness etc.

Grid seen from above at the beginning of the flushing

Grid seen from above at the end of the flushing

Velocity vectors at the end of the flushing

Bed level changes after the flushing, when the water is raised back to normal level.

Four animations of the results during a flushing has been made. The animations view the reservoir from above. The first three animations were made with ParaView. The last animation was made with Tecplot. One animation shows the velocities in the reservoir during the whole flushing 34 MB. Another 5 MB shows the velocities only during a drawdown and a quick water elevation raise. The colours shows the water velocity at the surface. Red is high velocity and blue is zero velocity. The third anmiation 38 MB shows the bed elevation changes during the whole flushing. A fourth animation shows the bed elevations at the downstream part of the reservoir during the flushing, including the velocity vectors close to the bed: 94 MB. There are 10 minutes between each frame, and the whole flushing lasts for 4 days and 6 hours. The computational time for the simulation was 105 minutes on a dual-core 2.8 GHz PC.

June 2010 Lisa Hoven wrote her Master Thesis in hydraulic engineering on this project. Her thesis can be downloaded here (2.5 MB PDF file).

New field measurements were done in September 2011, using both the ADCP and the LISST-SL instruments. The LISST measured sediment concentrations and grain size distributions instantly. Some pictures from the field work are given below:

The LISST seems to be giving good results (Photo: S. Løvfall)

Measuring along a section in the reservoir (Photo: S. Løvfall)

December 2011 Sigurd Løvfall wrote his Master Thesis in hydraulic engineering on this project. His thesis can be downloaded here (116 MB PDF file).

June 2012 Halvor Kjærås wrote his Master Thesis in hydraulic engineering on this project. His thesis can be downloaded here (6 MB PDF file).

This project is carried out in cooperation with Roberto Rodríguez and Laura Lizano from the Instituto Costarricense de Electricidad in Costa Rica.

Back to CFD page

This page was last updated: 12. June 2012