Published in Sarov (Arzamas-16), Nizhegorodskaya oblast
NUCLEAR CENTER -
ALL-RUSSIAN RESEARCH INSTITUTE
OF EXPERIMENTAL PHYSICS
Русский | English
Issue No 2, 2019
CONSERVATIVE REMAPPING IN GAS-DYNAMIC DIFFERENCE SCHEMES ON LAGRANGIAN-EULERIAN MESHES
A. M. SteninThe paper describes a conservative-in-total-energy global remapping algorithm for the Eulerian stage of the time step in gas-dynamic difference schemes on regular tetragonal Lagrangian-Eulerian spatial meshes. The values of the gas-dynamic quantities are remapped with conservation of the respective integral values. The values of specific internal energy are remapped subject to the conservation of the integral internal energy with corrections that provide total energy conservation. If a fully conservative difference scheme is used at the Lagrangian time-step stage, then the conservative remapping renders the whole scheme fully conservative at both Lagrangian and Eulerian stages. The results of benchmark computations of a strong point explosion are presented.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 3-16.
Key words: gas dynamics, Lagrangian-Eulerian meshes, remapping, a fully conservative scheme, the Sedov blast.
|TVDR-SCHEMES TO SOLVE A SYSTEM OF RADIATIVE HEAT TRANSFER EQUATIONS
A. A. ShestakovA great number of works are devoted to the problem of designing numerical methods to solve the radiative transfer equation. A separate group among them consists of the methods for solving the system of radiative heat transfer equations with an additional equation of internal energy of matter to account for the exchange processes of radiation/matter interaction. Due to the considerable non-linearity of these processes, high requirements are set for the quality of the selected numerical techniques when solving the radiative heat transfer problems. Firstly, these techniques should incorporate an efficient method to resolve the non-linearity of this problem. Secondly, because of the 7D space of all variables, the approximation system should be solved by a cost-effective method, in which the number of arithmetic operations is proportional to the number of nodes of the difference mesh. Thirdly, along with natural requirements (approximation, absolute stability and convergence), the difference discretization should satisfy additional requirements of conservatism, unconditional monotonicity and positivity for positive functions. The paper presents the research on new implicit unconditionally monotone difference schemes of a higher order of approximation (higher than the first one) for the problems of radiative heat transfer.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 17-36.
Key words: radiative transfer equation, difference schemes.
|RESULTS OF NUMERICAL SIMULATIONS OF VARIATIONS IN THE SHOCK SENSITIVITY OF TATB-TYPE HIGH EXPLOSIVES IN RESPONSE TO MULTIPLE SHOCK WAVES IN THE LEGAK CODE
N. A. Volodina, V. B. Titova, M. O. ShirshovaThe paper presents the results of modification of the MK detonation kinetics model in the LEGAK code to enable universal accounting for the shock sensitivity of TATB-based HE depending on its state before the front of the initiating shock wave. The changes in the model apply to the formula responsible for the density of hot-spot formation near the shockwave front, and to the algorithm for shock wave front tracking and analysis of the state of matter before the shock-wave front. The realized modification of the MK model was tested using experimental data for TATB-based HE with initial densities of 1,3 г/cm3 ≤ ρ0 ≤ 1,91 г/cm3. The proposed algorithm enables automatic accounting for the state of HE before the shock front and uniform description of experimental data for both sensitizing and desensitizing of HE in response to multiple shock waves.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 37-49.
Key words: detonation kinetics, shock sensitivity, TATB.
|MAJORANT SELECTION CONDITION IN THE MAXIMUM CROSS-SECTION METHOD
N. V. IvanovThe maximum cross-section method is widely used in the Monte Carlo simulations of the particle transport equations to simplify the pathlength modeling procedure. Using a numerical example we show that this method delivers a correct pathlength distribution if only one condition is observed: the pathlength distribution density corresponding to the maximum cross-section should be normalized.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 50-54.
Key words: particle transport equation, Monte Carlo method, pathlength modeling, choice of the maximum cross-section.
|NUMERICAL SIMULATION OF AERODYNAMIC PROBLEMS WITH STATIC MESH ADAPTATION FOR SOLUTION FEATURES
A. V. Struchkov, A. S. Kozelkov, R. N. Zhuchkov, A. A. Utkina, A. V. SarazovThe algorithm for numerical solution of aerodynamic problems based on the identification of shock-wave regions and subsequent adaptive mesh refinement in such regions is described. The method is based on the cell partitioning by adding new nodes to cell faces. The shock-wave regions, where the mesh will be refined, are identified using a pressure and density gradient criterion. The proposed algorithm is implemented on arbitrary unstructured meshes and designed for the finite-volume discretization of the system of Navier-Stokes equations. Application of the algorithm is demonstrated on the problems of transonic airflow of the NACA0012 airfoil and supersonic airflow of a wedge. The proposed adaptive mesh refinement method is shown to improve considerably the quality of the numerical solution produced on coarse meshes.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 55-67.
Key words: aerodynamics, numerical simulation, adaptive mesh, transonic and supersonic flow, shock waves.
|EFFICIENT JOB TIMING DATA PREDICTION BASED ON EXTENDED METADATA
S. A. PetuninMost of today’s high-performance computing systems use the Slurm workload manager and job scheduler in their system software. The scheduler’s statistics database is the main source of workload analysis, and its understanding makes it possible to improve the efficiency of workload management in computing systems. Application of the Random Forest statistical method from the machine learning group for predicting the job timing data (queuing time and runtime) was studied. All the experiments were done with a large body of data logged by Slurm during different calendar periods. The data processing programs were implemented as scripts within the R statistical system. The key idea of the work was to incorporate new unique job metadata, which have never been employed before, into the structure of the predictive models. The results of the studies demonstrated that the proposed predictive models were twice as efficient as the estimates of users and the job scheduler. This work is the first step in the implementation of automatic job management services at the level of the metadata-based job scheduler to provide automatic processing of resource requests and more efficient notification of users.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 68-80.
Key words: prediction, job metadata, runtime, latency, Random Forest method.
|PROBABILISTIC RISK ASSESSMENT OF TSUNAMI ON THE COAST OF EGYPT NEAR THE EL DABAA NPP CONSTRUCTION SITE
A. I. Zaytsev, S. M. Dmitriev, A. A. Kurkin, E. N. PelinovskyThe need to assess the risk of tsunami on the Egyptian coast is associated with the El Dabaa NPP construction project (coordinates 28,5270° E, 31,1032° N), the tender for which has been won by the Russian corporation Rosatom. To carry out the long-term tsunami risk analysis, we used the popular PTHA method based on the statistical analysis of a great number of real and predicted earthquakes and subsequent simulations of tsunami waves triggered by hypothesized earthquakes. It is shown that the predicted water rise level in the region of the NPP makes 6 ± 2 m for the period of 10 thousand years.
VANT. Ser.: Mat. Mod. Fiz. Proc. 2019. No 2. P. 81-89.
Key words: tsunami, long-term prediction, NPP, Egypt.
|[ Back ]