Explore chapters and articles related to this topic
Analyses and Numerical Simulations of Basic Two-Phase Flows
Published in Clement Kleinstreuer, Theory and Applications, 2017
The DNS approach does not require any modeling but is extremely taxing on computer resources. For example, Joseph (2001) reported meshed gap sizes between (solid spherical) particles as small as 10−5dp. Specific DNS references include Hu et al. (2001), Patankar et al. (2000), and Glowniski et al. (1999), among others. Computational aspects of the Monte Carlo method are reviewed in Oran et al. (1998). With respect to the separated flow approach, the Eulerian-Lagrangian formulation is most powerful for the simulation of particle suspension flows as indicated in Fig. 4-2 (cf. Sect. 4.1.2), especially in light of the advancements made with DNS, i.e., resolved-volume (or volume-of-fluid) techniques (cf. Scardovelli & Zaleski, 1999). In contrast, the Eulerian formulation, when simulating dispersed phase behavior, cannot easily account for the effects of particle size and shape, interparticle collision, and trajectory crossing (cf. Pozrikidis, 1999).
Encapsulation Methods
Published in Magdalini K. Krokida, Thermal and Nonthermal Encapsulation Methods, 2017
Maciej Jaskulski, Abdolreza Kharaghani, Evangelos Tsotsas, Magdalini K. Krokida
Although the origin of research on fluid flow turbulence dates back to a century ago, there is still no solution to this problem owing to its multiscale nature. Several methods have already been developed to solve this problem. The first method is Direct Numerical Simulation (DNS), though it requires high computing power due to the necessity of using dense computational meshes. This method is applied to solve the N-S equations for turbulent fluid flow without using additional models of turbulence (Ferziger and Perit 2002). As a result, a full instantaneous flow velocity field can be obtained. In this method, mesh elements cannot be bigger than the so-called Kolmogorov length scale, which decreases with an increase in Reynolds number. In addition, for calculations in transient state, the time step must be chosen such that fluid particles move by one mesh element in each time step, satisfying Courant–Friedrichs–Lewy (CFL) conditions (Courant number <1). The DNS is widely used to simulate the shape of turbulent eddies, the dispersion of the so-called aerodynamic noise, or calculations of hydrodynamic resistance during flow around solid bodies.
Domain Naming System
Published in Giovanni Bartolomeo, Tatiana Kováčiková, Identification and Management of Distributed Data: NGN, Content-Centric Networks and the Web, 2016
Giovanni Bartolomeo, Tatiana Kováčiková
The Domain Name System (DNS) is an abstraction layer decoupling the identifier of a host from its IP address, which allows these, and other, example problems to be solved efficiently. The primary advantages of the DNS are that it is not necessary to update the clients directly and that it is possible to use more easily remembered mnemonic names instead of numbers. The DNS defines a protocol that is essential to the effective functioning of the open Internet; it would be impossible to notify Internet clients when an IP address changes since the clients would frequently be completely unknown to servers.*
Direct Numerical Simulation of Low and Unitary Prandtl Number Fluids in Reactor Downcomer Geometry
Published in Nuclear Technology, 2023
Cheng-Kai Tai, Tri Nguyen, Arsen S. Iskhakov, Elia Merzari, Nam T. Dinh, Igor A. Bolotnov
Recent advances in high-performance computing has made DNS an affordable option for investigating complicated mixed convection phenomena. DNS allows direct access to detailed flow behavior, such as distribution of higher-order statistics, that facilitates the development of modeling approaches. Existing DNS efforts have largely been devoted to the prototypical Poiseuille-Rayleigh-Bénard (PRB) flow (horizontal turbulent channel with unsteady temperature stratification created by isothermal wall of different temperature). For the widely studied PRB flow of air, the influence of buoyancy facilitates the eruption of plumes from the heated wall. Consequently, these plumes facilitate the emergence of a large-scale roller in the velocity field and thermal structures.[11,12] The buoyancy effect also significantly promotes turbulence intensity on the wall-normal direction, which reportedly has an influence on the Nusselt number. With decreasing Prandtl number, the buoyancy promotion of the wall-normal turbulence fluctuation intensity and heat flux becomes more evident due to the strong thermal diffusivity of liquid metal. Also, large spatial scale eruption of hot plumes that reaches the opposite end can be clearly identified, signifying a stronger buoyancy effect.[4,13]
A three-step reduced mechanism for MILD combustion
Published in Combustion Science and Technology, 2022
Reiley T. Weekes, Keiko K. Nomura, Antonio L. Sánchez, Forman A. Williams
There have been a number of numerical studies of MILD combustion by DNS (Direct Numerical Simulation) with detailed reaction mechanisms (Swaminathan 2019). The mechanisms employed had no less than nine chemical species for the fuel H2 (Aspden, Day, Bell 2011), while that number increases rapidly for more complex fuels, reaching the 35 species of the GRI 3.0 Mechanism without nitrogen chemistry for CH4, which was the fuel addressed by Aspden, Day, and Bell (2016), and it would become even larger for larger fuel molecules. Introduction of these detailed mechanisms into DNS algorithms generates exceptional computational costs that impose limits on the geometrical complexity of simulations that can be performed. Consequently, most of the existing DNS investigations are limited to simple box-turbulence (Minamoto et al. 2013) and/or mixing-layer-type (van Oijen 2013) analyses, with even the simplest box analyzed by Minamoto et al. (2013) requiring 120 hours of simulation time on 4096 computational cores. More complicated and practical flows require even more time; a reacting swirling hydrogen jet studied by Minamoto et al. (2015) needed 135 days of simulation time. These high computational costs impose limits on the geometrical complexity and the Reynolds numbers that can be considered in direct numerical simulations (DNS) employing detailed chemistry.
Development and Optimisation of a DNS Solver Using Open-source Library for High-performance Computing
Published in International Journal of Computational Fluid Dynamics, 2021
Hamid Hassan Khan, Syed Fahad Anwer, Nadeem Hasan, Sanjeev Sanghi
The laminar, transition and turbulent fluid flow are physical phenomena found in nature and daily life. Computational fluid dynamics (CFD) is utilised for numerical solution of a wide range of natural, industrial, and scientific problems. The most popular computational tools to study turbulence are Reynolds-averaged Navier–Stokes (RANS), Large-eddy simulation (LES), and Direct Numerical Simulation (DNS). These computational models are classified based on accuracy and the resolved length scale. DNS is the most popular and accurate technique to study fluid flow's turbulence and chaotic nature. DNS solves the governing equations numerically so that the entire range of length scales, starting from the smallest dissipative scales to the largest scales up to the order of the physical domain are resolved. The DNS accuracy (Moin and Mahesh 1998) is acquired by using the smallest length scale ; this leads to a small-time step and an increase in computational burden. The demand for DNS of turbulence (Alfonsi 2011) advents high-performance computing, which is now an essential area in scientific computing.