MODELING OF FLOW IN LAMINA FREE ON NATURAL CHANNELS. ANALYSIS INTEGRATED WITH BLUEPRINTS FINITE VOLUME IN ONE TWO DIMENSIONS.Author:
BLADE CASTELLET ERNEST.
Year:
2005.
University:
POLITÉCNICA DE CATALUÑA [
www.upc.edu].
Place of defense: EDIFICI C-1, AULA 002-CAMPUS NORD.
Place of preparation: ETSECCPB, EDIFICI C2 Campus NORD.
Summary: Understanding the hydraulic behaviour of rivers during floods is crucial for the resolution of a variety of problems of hydraulic engineering and river dynamics as flood areas mapping, embankments and hydraulic structures design, streambank stabilization, dam break studies, river rehabilitation, or risk assessment in extraordinary precipitation events. That is the reason for studying unsteady open channel flow in irregular geometries through the development of numerical simulation tools. The main objective of this work is generating mathematical modelling tools for unsteady open channel flow in irregular geometries, as natural rivers are. The developed numerical schemes are aimed to be able to properly simulate discontinuous flows (front waves, hydraulic jumps, transcritical flows) as occurs during a real flood in most rivers, especially those in Mediterranean areas. Explicit numerical schemes, based on the finite volumes technique, for the resolution of the Saint Venant equations in conservative form, are developed. This shock capturing schemes are most suitable for the simulation of flows with discontinuities. The developed schemes are high resolution schemes: second order precision away from flow discontinuities, no spurious oscillations and no extra dissipation (as with first order schemes) around them. Flow patterns in rivers depend on their geometry. When there exists a predominant flow direction a one dimensional approach can be used, but other times (river confluences, flow around structures, compound channels, river channel overflow) a two dimensional approach is needed. This last one is more expensive as needs more topographic information, model development is complex, and computational time is greater. New methodologies for one and two dimensional modelling are developed, but also both approaches have been integrated in order to be able to model big areas using a one dimensional approach when it is enough, and a two dimensional one when it is required by flow or geometry characteristics. In that way the efficiency of existing modelling methodologies is improved. Due to the special characteristics of Saint Venant equations, modelling methods that work for other hyperbolic equations can lead to important errors. In one dimension and irregular geometries, the flux vector of the equations has a spatial dependency on the geometry variations. A methodology that takes into account that dependency is developed. That, together with a correct treatment of the equations source term, allows a correct balance with the discretised term of the rest of the equations, leading to one dimensional high resolution schemes for irregular geometries. Similar schemes in known previous works were not able to converge to steady state solutions or, if they did, they did not converge to the correct one. A correct balance of the discrtetised source term is also achieved in two dimensions. Also, wetting and drying of the domain and precipitation inputs are implemented. In such way, the developed model can also be seen as a hydrological distributed rainfall-runoff transformation model fully integrated in a hydraulic model. The domain discretisation can be done using triangles or quadrilaterals, and the whole system has been integrated in a user friendly pre-process and post-process interface. High resolution schemes are based in a mathematical theory which is only valid for hyperbolic equations much simpler than Saint Venant equations. For that 8 reason 3b5 an exhaustive verification of the methodology is carried out. Verification is done with comparison against problems with analytical solution, other numerical models and laboratory experiments. Finally, some real applications of the methodology to engineering and river dynamics problems are presented.
