Navegar por los elementos (1 total)
- Resumen es exacto "Laboratory-scale physical models continue to be the ultimate tool for the design, optimization, and verification of complex hydraulic structures. To achieve perfect similarity between physical models and prototypes a single, constant and uniform scale for all types of forces is required. In practice, this condition is virtually impossible to achieve, so the usual procedure is to identify and properly scale only the dominant type of force for the specific case under study. The distortive effects produced by out-of-scale forces are known as scale effect.
In principle, as long as the fundamental physics is well known, the effect of all forces actuating on both the prototype and the physical model can be represented accurately into an appropriate mathematical model, and then solved numerically. Given that this model would represent all forces with their correct scale, it could be used to determine the differences that appear when changing from one scale to the other, thus allowing to quantify and eventually correct the scale effects of the physical model. In practice, however, Computational Fluid Dynamic (CFD) models require the usage of simplified schematization to account for subgrid phenomena, especially turbulence. This introduces uncertainty about their capability of predicting scale effects related to these simplified processes.
To take into account these uncertainties on the process of numerical evaluation of scale effects, an original methodology is proposed, based on the definition of a Scale Effect Coefficient, which works as a conversion factor between any similar variable of the prototype and the physical model. It is shown that this coefficient can be computed from numerical model results as long as three Error Factors are maintained close to unity. These factors are associated with the mathematical model, the discretization and the input parameters respectively. Strategies to neutralize the Error Factors associated with input parameters and discretization are proposed. However, the Error Factor related to the model cannot be evaluated directly for a given problem of interest. Hence, to ensure that the selected mathematical model is capable of resolving the desired scale effects, the solution of one or more subrogate Test Problems is proposed. These must contain the same physical processes leading to scale effects on the main problem, and have experimental data available at different scales.
In this thesis, a number of test problems are solved with Reynolds-Averaged Navier Stokes (RANS) models with different turbulence models and wall treatments to evaluate their capability for solving viscous scale effects. Several applications to hydraulic structures using RANS models are then presented, including ship-locks and weir crests."
Ordenar por: