University of Twente Student Theses

Login

Numerical Simulation over a backward facing step with OpenFOAM

Slaper, A.R. (2015) Numerical Simulation over a backward facing step with OpenFOAM.

[img] PDF
4MB
Abstract:A literary study was performed in order to understand the theory behind numerical ow simulations. There were also several test cases performed to see the behaviour of different turbulence models as well as different methods. The circular cylinder cases where used to get a feeling for the setup of cases as well as to get familiar with OpenFOAM. The results found were, except in the LES case, quite good in comparison to the reference papers. In the uRANS case, a little too much artificial dissipation reduced the sharp edges in the profiles. By applying a finer mesh in the wake region, this problem could be solved. Also the application of higher order schemes could improve the results. The LES case was a different story because the results seem totally off. The flow remains attached to the cylinder for too long because too little turbulence was created. It would be useful to try a different type of function in order to improve the results because the van Driest damping function damps some of the turbulence. Also by setting the initial condition for the turbulent viscosity to a very small value could trigger the simulation to produce more turbulence and improve the results. It is also interesting to try some different LES methods in order to see if the result will change. The backward facing step was the test case to see the influence of different LES and DES models implemented in OpenFOAM. The original fine grid setup took an exceptionally long time to finish. This could be caused by a mistake in setup as well as with the fact that the simulation would take a long time anyway. Since the cells at the step are very small (to get enough resolution at the wall), the time step has to remain very small in order to keep the maximum Courant number very low. To get some results for the backward facing step, the decision was made to change to a coarser grid. Two different types of simulations were performed on the coarse grid. The first one is the IDDES simulation while the second one was an one-equation-eddy LES model. This latter case was left out of the report, because the velocity profiles were very wrong. Although in both cases the initial conditions were the same as presented before in this report, the velocity profile before the step changed in the one equation eddy simulation. The boundary layers became very thick and made no sense at all. The reason for this could be that the coarse mesh was too coarse for this type of simulation. With a much finer mesh, the result could significantly improve. It should also be kept in mind that the application of wall models could also mean a significant improvement in the results. The IDDES case looked promising but it turned out that it also over predicted the length of the bubble. The mesh used was very coarse so the bubble looks a little like a Lego creation which is far from reality. By applying a finer mesh, the result could become much better. Because of the coarse mesh, the simulated eddy size also increases (since the smallest solvable eddy size depends on the size of the grid). Because these eddies are approximated and not solved exactly, errors are introduced as well. Although the differences with the experiments are clearly visible, it can also be seen that the averaging time influences the fluctuations as well. The differences in stations further down stream were visible especially in the fluctuations. The averaging time chosen should be sufficiently large to `guarantee' a stationary solution of the mean flow. Setting the average time to a much larger value is one way to make sure that the averages will be correct. Another way would be to make estimations of the frequency of the vortex shedding and choosing the time in such a way that at least a couple of vortexes have been shed. This will give a good solution without increasing the calculation too much.
Item Type:Internship Report (Master)
Clients:
CIRA, Italy
Faculty:ET: Engineering Technology
Subject:52 mechanical engineering
Programme:Mechanical Engineering MSc (60439)
Keywords:Numerical solutions, openFOAM
Link to this item:http://purl.utwente.nl/essays/69283
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page