Numerical Investigations of the Unsteady Flow in the Stuttgart Swirl Generator with OpenFOAM

Detta är en Master-uppsats från Chalmers tekniska högskola/Institutionen för tillämpad mekanik

Sammanfattning: As a consequence of the current change in the electric energy supply structure and present available electricenergy storage capabilities, hydro power plants are tending to be operated on a wide spread of off-designconditions. The operation of turbines not running at their best operating point could lead to a physicalphenomenon called helical vortex or vortex rope. Although the helical vortex has been investigated for severaldecades already, it is still far from being completely understood. Moreover, it could have undesirable impactson the hydraulic system of the power plant, namely, an increase of the risk of fatigue failure, the developmentof resonance vibration and a decrease of the efficiency of the power plant. In order to analyze the vortex ropeand to assess its effects on the hydraulic system, it is therefore in the interest of both science and industry tosimulate the physical phenomenon of the helical vortex accurately. In response to this, the Institute of FluidMechanics and Hydraulic Machinery at the University of Stuttgart, Germany, constructed a Swirl Generator toinvestigate a swirling flow, which is similar to the downstream of the turbine runners. With the help of eightnon-rotating blades and a conical diffuser, this test rig is capable of generating a helical vortex.In this work, cases for the simulation of the unsteady flow in the Stuttgart Swirl Generator with OpenFOAMare presented. To do so, this work firstly explains the background theory of the helical vortex phenomenonand the method of Computational Fluid Dynamics. Furthermore, the underlying test rig, geometry and mesh,discretisation schemes, algorithm for interequation coupling, solvers for the systems of linear algebraic equations and boundary and initial conditions are elucidated. As the flow in the Swirl Generator is characterized by aReynolds number in a range of approximately 1.1'105 to 1.8'105 and unsteady flow features caused by thehelical vortex, the treatment of turbulence plays a vital role for an appropriate setup of the cases. Here, common Reynolds-Averaged Navier-Stokes (RANS) turbulence models are assumed to be limited in their capability of dealing with unsteady flows. The method of Large Eddy Simulation (LES) on the other hand, would be capable of addressing these drawbacks, but are still unfavorable in terms of their demands on the computational effort.This work therefore, investigates hybrid strategies, which combine the advantages of both the RANS and LESapproach. For this, cases based on two standard high-Reynolds number RANS models: k-ε and k-ω SST, andthree hybrid RANS-LES models:k-ω SST SAS, Spalart-Allmaras DDES and Spalart-Allmaras IDDES are setup. In order to assess the results of the simulations, the present work shows a general evaluation of the resultsand compares experimental and simulated measurement data.