Large eddy simulation of particle-laden flow [Elektronische Ressource] / Christian Gobert

Christian Gobert

Informations

• Publié par: technische_universitat_munchen
• Publié le: 01 janvier 2010
• Nombre de lectures: 32
• Langue: English
• Poids de l'ouvrage: 9 Mo

Extrait

TECHNISCHE UNIVERSITAT MUNCHEN
Fachgebiet Hydromechanik
Large Eddy Simulation of Particle-Laden Flow
Christian Gobert
Vollstandiger Abdruck der von der Fakultat fur Bauingenieur- und Vermessungswesen der
Technischen Universitat Munchen zur Erlangung des akademischen Grades eines
Doktor-Ingenieurs
genehmigten Dissertation.
Vorsitzender: Univ.-Prof. Dr. rer. nat. E. Rank
Prufer der Dissertation: Univ.-Prof. Dr.-Ing. habil. M. Manhart
Prof. Dr. J.G.M. Kuerten, TU Eindhoven, Niederlande
Univ.-Prof. Dr. rer. nat. B. Simeon
Die Dissertation wurde am 15.12.2009 bei der Technischen Universitat Munchen eingereicht und
durch die Fakultat fur Bauingenieur- und Vermessungswesen am 02.03.2010 angenommen.Abstract
This thesis is a contribution to research in the eld of Large Eddy Simulation (LES) of
particle-laden ow with a focus on the e ect of unresolved small scale turbulence on sus-
pended particles. The rst substantial contribution of this thesis is a detailed quantication
of small scale e ects. The thesis contains new results from Direct and Large Eddy Simula-
tion of particle-laden forced homogeneous isotropic turbulence. Reynolds numbers based on
the Taylor length scale are Re = 34, 52 and 99. Stokes numbers based on the Kolmogorov
time scale range fromSt = 0:1 toSt = 100. The main conclusions of these numerical exper-
iments are the following: if subgrid scales are neglected for particle transport in LES, then
particle kinetic energy is underpredicted, particle dispersion is overpredicted but preferential
concentration is predicted satisfactorily.
The second contribution of this thesis is an analytical and numerical analysis of three com-
monly used models that describe the e ect of subgrid scales on particles: the Approximate
Deconvolution Method (ADM) and two stochastic models. Kuerten (Phys. Fluids 18, 2006)
proposed ADM for particle-laden ow. The stochastic models were proposed by Shotorban &
Mashayek (J. Turbul. 7, 2006) and Simonin et al. (Appl. Sci. Res. 51, 1993). Analytical and
numerical results show that ADM improves the particle dynamics, but for coarse LES, the im-
provement is very small. On the other hand, at high Stokes numbers the predictions from the
stochastic models are less accurate than those obtained by neglecting subgrid scales for par-
ticle transport. Furthermore, the stochastic models were found to destroy preferential con-
centration, whereas ADM preserves preferential concentration. In conclusion, the stochastic
models were found to perform less reliably than ADM.
The third contribution of this thesis is a new model that can be regarded as an extension
of ADM. The new model consists of a specic interpolation method, which is designed such
that statistically the numerical interpolation error can be identied with the e ect of the
unresolved scales. The new model was assessed analytically and numerically. For the nu-
merical assessment, simulations of forced homogeneous isotropic turbulence atRe = 52, 99
and 265 were conducted. Analytical and numerical assessments show very promising results.
In particular, the overall accuracy of the model is higher than the accuracy of ADM. As the
coarseness of LES increases, the gain in terms of accuracy of the new model in comparison
to ADM also increases. This means that for high Reynolds number congurations, where
only coarse LES is possible, the new model can be expected to produce signicantly better
results than ADM.
i