Computational fluid dynamics

Jyeshtharaj B. Joshi, Krishnaswamy Nandakumar, Ashwin W. Patwardhan, Arun K. Nayak, Vishnu Pareek, Monica Gumulya, Chunliang Wu, Nitin Minocha, Eshita Pal, Mukesh Kumar, Vishal Bhusare, Shashank Tiwari, Dhiraj Lote, Chaitanya Mali, Ameya Kulkarni, Sarang Tamhankar

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

6 Scopus citations


In nuclear reactors, heat is generated in the fission process which is harnessed by transferring it to fluid. The transfer also maintains the integrity of the fuel. The subject of nuclear reactor heat transfer and thermal hydraulic is crucial as it establishes the relation between the heat generated in fission and its transfer to the coolant. The heat accumulated in the coolant is used for various purposes viz. for power generation in Rankine cycle, for propulsion of an engine by providing thrust and for other heating applications. One of the major goal of the subject is to provide flow and temperature distribution in the components/equipment of the system so as to enable the safe operations. In addition, the knowledge of thermal hydraulics provides the inputs for the structure design of the nuclear reactor. In view of the above, thermal hydraulic analysis of nuclear reactor plays an important role in designing an efficient, economical and safe reactor. Such a knowledge base was not available during the initial period of inception of nuclear reactor design. This was because the governing equations (arising out of laws of conservation of mass, momentum and energy) could not be solved for any combination of laminar/turbulent single phase/multiphase flows in complex geometries of reactor components. Therefore, during the first 50 years of nuclear reactors, design calculations were usually carried out using preliminary knowledge base obtained by incorporating extreme assumptions in the governing equations. The major emphasis was on the experimental data from pilot plants and the scale-up in a gradual manner. During the period of sixties, seventies and eighties, one dimensional (1-D) code were in continuous development resulting into commonly employed 1-D thermal hydraulic codes like RELAP, CATHARE, ATHELETE, MARS, etc. These codes have been popular because they (1) utilize comparatively lesser computational resources and (2) facilitate system/reactor scale simulations. The development of these codes is completely based on extensive experimental database collected over the years which have been useful in generating empirical models that are plugged into these 1-D codes. However, the validity and reliability of these empirical models is sceptical, for designing new generation of advanced nuclear reactors for which extensive database (flow and temperature patterns) is not available in advance. In order to overcome the limitations of 1-D codes and designing new generation of nuclear reactors, high fidelity and reliable numerical tools which can provide detailed 3-D distribution of thermal hydraulic parameters is required. Due to recent advancements in computational power, the detailed 3-D velocity and temperature patterns (thermal hydraulic parameters) can be obtained using 3-D Computational Fluid Dynamics (CFD) codes where transient conservation equations of mass, momentum and energy are solved for different species with a possibility of phase change. The most accurate and detailed approach for simulating turbulent flows is Direct Numerical Simulation (DNS). However, in order to resolve all the eddy scales and ensure spatial as well as temporal accuracy, DNS requires very high computational power and non-dissipative numerical schemes. A compromise to DNS is Large Eddy Simulation (LES), in which large scales of turbulence are resolved and the small scales are modeled. For LES, the grid requirement and hence the computation cost increases with Reynold’s number [endif]-->, making the resolution of the wall layer infeasible at high Re. As a result, LES is also not suitable for performing CFD simulations for industrial scale. Recently, hybrid LES/RANS (Reynold’s Average Navier Stoke’s) models have become very popular in which the inner region (close to the wall) is solved using RANS model and outer layer using LES model but the problem is the selection of appropriate RANS model to study the inner layer. Hence, the present chapter focuses majorly on the development of advanced RANS models which are computationally less expensive and can be easily coupled with hybrid LES/RANS models. In the recent years, varieties of RANS models have been proposed. The performance of any RANS model depends entirely on the selection of appropriate closure modeling strategy which in turn totally depends entirely upon the applicability of the set of assumptions behind these closure models. The closure models are not derived from first principles and are completely empirical or data driven. Therefore, the present chapter describes the relative merits of large number of RANS, LES and DNS models. Besides, the chapter also incorporates the issue and challenges associated with two phase flow modeling. It includes developments in the formulation of conservation equations, interfacial forces and turbulence models for two phase flows. In a nut shell, the chapter describes the development of CFD right from its origin with its application to single phase as well as multiphase flows. In addition, the chapter also covers wide spectrum of dense two phase flow modeling (DPM and DEM) including various mathematical formulations viz. Euler-Euler, Euler-Lagrangian, etc., closures required for the mathematical models and turbulence modeling. Moreover, the chapter also presents a variety of numerical algorithms used for RANS, LES and DNS models for single and multiphase flows.

Original languageEnglish
Title of host publicationAdvances of Computational Fluid Dynamics in Nuclear Reactor Design and Safety Assessment
Number of pages218
ISBN (Electronic)9780081023372
ISBN (Print)9780081023389
StatePublished - 1 Jan 2019

Fingerprint Dive into the research topics of 'Computational fluid dynamics'. Together they form a unique fingerprint.

Cite this