Inicio | Hidro | Rios | Puentes | Investigación | EnseñanzaCV | Links | English
2D river morphodynamics

My PhD research intents to modify the existing hydrodynamic code River2D in order to simulate riverbed dynamics as well. River2D is a fairly advanced depth-averaged finite elements model with notable capabilities for river flow simulation (e.g. supercritical regime, wet-dry elements, automatic mesh refinement, mesh extraction).

River2D can also solve Vertically Averaged and Moment (VAM) of momentum equations to provide more detail in the vertical component, such as surface velocity and near-bed velocity. In a sense, River2D-VAM is similar to a 3D model with only 3 or 4 horizontal layers (typical 3D applications require 10 or more horizontal layers to resolve for the vertical profile of velocity). That ability to become a quasi-3D should be beneficial to resolve the complex helical flow in bends, without the computational demands of a fully 3D model. On the right there is a vector plot of vertically averaged velocities in a strongly curved bend; velocity vectors seem to follow the shape of the bend. The secondary flow effect of water in the surface moving towards the outer bend, and near bed water moving towards the inner bend is lost in the averaging process. Therefore, 2D depth-averaged models require some sort of secondary flow "correction" to account for this loss of information
The VAM model allows computing the surface velocity and near-bed velocity. Notice the direction of surface (outwards) and near-bed (inwards) velocity vectors, preserving some of the features of helical flow. Near-bed velocity should provide a better estimation of bed shear stress magnitude and direction, which is important for bedload sediment transport, and therefore for riverbed dynamics.
VAM model vector plots in a bend: (a) Surface velocity (b) near-bed velocity.

My research is planned to proceed in four stages: (1) Bed changes in straight channels, (2) Bend curvature effects using the VAM model, (3) Bank stability algorithms to account for channel width changes, and (4) Riverbed dynamics such as meander migration and meandering-braiding transition. At this moment I am at the initial stage, computing bed aggradation and degradation. Below are the first quantitative results of a comparison between the model (solid line) and experimental measurements (dots) of bed aggradation due to sediment overloading at the entrance of a straight flume (sediment load is 4 times the equilibrium sediment transport for the flume).
I have also performed some "qualitative" test to assess the model (and for fun mostly!), such as scour in a long constriction and local scour in a spur dyke (groine). In a long constriction (e.g. Piura River in the urban reach), initially there is intense scour in the contracting zone (flow acceleration) and deposition in the expansion zone (flow deceleration). After a sufficiently long time, the deposited sediment near the expansion is washed away. That may help explain why Bolognesi Bridge, which is located at the end of a constricted reach in the Piura River, suddenly failed without warning 4 days after Viejo Bridge failed during the peak discharge..
Scour in a long constriction: (a) Scalar plot of velocity (b) scour at two different time intervals (roll over the mouse)
Previous Research...

Genetic Programming applied to runoff-rainfal modelling and velocity estimation in gravel-bed rivers. Genetic Algorithms for water quality systems optimization under uncertainty... More details will be posted soon.

Several people contributed to the numerical modeling shown in these pages. Alex Herrera and Edgar Lazarte performed the 2D numerical computations of both Piura River and Tumbes River. Georg Premstaller made the 3D computations of local scour in piers and abutments. Juan Carlos Atoche performed the 3D simulations of sand traps as part of his undergraduate research thesis.
Inicio | Hidro | Rios | Puentes | Investigación | EnseñanzaCV | Links | English
Actualizacion: 15 de noviembre de 2003
Autor: José Vásquez