Modeling and validation of flows in a low speed centrifugal separator

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Helsinki University of Technology | Diplomityö
Checking the digitized thesis and permission for publishing
Instructions for the author
Mekaaninen prosessi- ja kierrätystekniikka
Degree programme
ix + 111 s. + liitt. 17
The aim of this Master's thesis is to assess the possibilities of Computational Fluid Dynamics (CFD) modelling in FLUENTr to predict fluid flow behaviour in a low speed centrifugal separator. From literature it is known that model limitations of the basic k-e turbulence model lay predominantly in the field of prediction of confined swirling flow behaviour. In this thesis CFD modelling of pipe flow and swirling annular flow is done in FLUENTr and validated by flow visualization experiments. The study includes a discussion of the fundamental physical theory behind the behaviour of fluid flows and particles in them and a low speed integrated centrifugal separator is introduced. With the theoretical insights, the classification mechanism of the device is explained. Subsequently the physical theory is used to explain the working of CFD flow modelling and what can be achieved. Also, a concise overview of the available techniques of flow visualization is given. In the experimental part of the thesis first a well established flow phenomenon, Reynolds' pipe flow experiment, is CFD modelled in FLUENTr and then validated by a flow visualization experiment. FLUENTr's flow predictions for pipe flow fit experimental data especially well for more turbulent flow conditions; a good result for modelling the turbulent swirling annular flows in the main experiment. With this practical experience gained the proposed centrifugal separation device was modelled in FLUENTr and parallel to that a flow visualization experiment run. Physical assumptions of the standard k-c turbulence modelling algorithms suggest model limitation especially in the type of flows encountered in the device studied. However, the model validation experiment indicates that CFD flow modelling of this type can in fact predict confined swirling flows fairly accurate. The model fit to the validation experiment was found to have a fractional bias of maximum 0.257 with common values well below 0.2. The average hit rate of CFD prediction to experimental data varied between 0.77 to 0.85 for low flow rate and between 0.75 and 0.92 for higher flow rates; unity being a perfect match. It can be concluded that strongly turbulent, confined swirling flows can be predicted well by the standard k-c turbulence model in CFD modelling and that accuracy even increases for increasing turbulence. Further development lays mainly in the investigation of this type of fluid flows in different geometric configurations and up-scaling.
Heiskanen, Kari
Thesis advisor
Turunen, Janne
Niitti, Timo
Computational Fluid Dynamics modelling, CFD, turbulence, swirling flow, centrifugation, flow visualization, Computational Fluid Dynamics model validation
Other note