A quasi-multidimensional Eulerian-Lagrangian numerical model for turbulent subcooled boiling flows in single and branched channels

Himanshu Pokharna, Purdue University

Abstract

The Particle-Fluid Model (PFM) is a modeling approach for two-phase flow which has promise for overcoming some of the limitations of the existing two-fluid models and leads to enhanced predictive capability. The model incorporates a Lagrangian representation for the dispersed phases and thus reduces the numerical diffusion associated with Eulerian schemes. In addition, it permits the interfacial interactions as well as inter-particle interactions to be modeled in a natural way. This thesis concerns further improvements to a PFM code called DISCON. In particular, stochastic models for bubble interactions such as bubble coalescence, turbulence, breakup and wall nucleation have been incorporated. Capability to model heat structures has also been added. The boiling model produces good agreement when compared to experiment data. A primary emphasis of the development was the quasi-multidimensional lateral phase distribution model and its incorporation into the Lagrangian-Eulerian single channel model. This phase distribution model is based on the physics of eddy-bubble interaction. Eddies are simulated using simple algebraic models and the bubble lateral motion is established by tracking the bubbles through a succession of eddies in turbulence. Results from the model are compared with experimental data and good agreement is obtained. Furthermore, the single channel code was extended to a pilot systems code to model branched flow passages. Models for merging and dividing flows were incorporated into the PFM code. Modeling results for a few conceptual problems have been obtained using this enhanced capability and are presented herein. Finally the modeling studies are augmented with theoretical dispersion and characteristics analysis of the conservation equations used in both the two-fluid model and in the PFM. It is shown that the Lagrangian-Eulerian system of equations results in a well posed problem. This is in contrast to the usual one-dimensional two-fluid model which constitute an ill-posed problem. The dynamic character of the two-fluid models as formulated in the RELAP5 and CATHARE codes were investigated and their respective regularization procedure were analyzed. Conclusions from these studies are summarized along with recommendations for future research.

Degree

Ph.D.

Advisors

Ransom, Purdue University.

Subject Area

Nuclear physics|Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS