Supercomputing Frontiers and Innovations

Recent Technology Trends in High-Performance Computing: Quantifying the Divergence of Top and the Rest

Sergey S. Konyukhov — 2025-05-16

Exploratory analysis methods were used to study basic characteristics of computing systems from TOP500_Lists. One of the peculiarities of the distribution of computing systems by performance is that it sufficiently well obeys an analog of the empirical Zipf’s law, in which logarithm of performance is reciprocal to the rank of computing system. Based on this observation we can divide all systems from the lists into several performance classes: top, high, base, and entry levels. Our analysis also revealed differences between these classes in other characteristics besides, the computational performance, e. g., such as power consumption. For all performance classes, trends in evolution of the basic characteristics of TOP500 computing systems were described and, where possible, comments were provided to explain their behavior. Performance and energy efficiency of the TOP500_List computing systems in the next 5–10 years were estimated using simple linear models obtained by the least-square method. We have found that energy consumption needed for entry-level supercomputers to surpass the threshold value of performance and to enter into TOP500_List will decrease during this period.

3D Numerical Simulation of Micro-Jet Excitation

Luka S. Volkov — 2025-05-16

Excitation of round laminar air micro-jet by volumetric force and by pulse-periodic heat source was simulated using the FlowVision software package in 3D formulation at normal conditions. Heat source and volumetric force imitate an influence of electrical discharge. Air jet was formed by a circular cross-section channel with inner size of 1 mm with the Poiseuille velocity profile at inlet boundary, the maximum profile velocity was 5 m/s. The conditions corresponded to the formulation of the problem considered earlier in the experiment. Dependence of large-scale vortex formation from volumetric force frequency and amplitude was obtained. Amplitude of force corresponding to the effect of the discharge on the air jet was determined and was set to 5 μN. For round laminar jet the exited oscillations of the jet was obtained at frequency range 500–2500 Hz, further increase in the frequency of oscillations left the shape of the jet close to the initial. Lack of influence of pulse-periodic heat source on flow structure was discussed. The results obtained demonstrate that the main contribution from the corona discharge to the jet in the experiment is provided by the volumetric force (ionic wind), and not by the heating of the gas.

Numerical Simulation of the Growth of Localized Disturbances in a Supersonic Boundary Layer over a Plate with Longitudinal Slots

Aleksey A. Yatskikh — 2025-05-16

This paper presents the results of numerical simulations of the growth of small amplitude localized disturbances in the boundary layer on a flat plate and on a plate with rectangular longitudinal slots at Mach number 2. The simulations were performed with the FlowVision software under flow conditions corresponding to those in the T-325 wind tunnel of the Khristianovich Institute of Theoretical and Applied Mechanics SB RAS. Cases with slots no deeper than the boundary layer thickness were considered. The localized disturbances in the boundary layer were generated by a time-pulsed, spatially localized heat supply from the plate surface. For a smooth plate, the growth of the disturbances was compared with the results of the linear stability theory. Data agreement was achieved for the grid resolution used. For the cases of the smooth surface and the plate with longitudinal slots, the evolution of disturbances in the boundary layer was analyzed in physical and wave space. In case of the smooth plate, the disturbance introduced into the boundary layer increases monotonically downstream. However, in the presence of longitudinal slots, the growth of the disturbance growth is different and depends on the depth of the slots. It has been shown that longitudinal slots can affect the stability of the supersonic boundary layer. A frequency-wavenumber analysis of the disturbance evolution revealed that longitudinal slots shift the range of the most unstable waves towards a region of higher frequencies compared to the smooth plate. This effect becomes more pronounced with increasing depth of the slots.

Direct Numerical Simulation and Rank Analysis of Two-Dimensional Kolmogorov-type Vortex Flows

Mikhail A. Guzev — 2025-05-16

The article is devoted to direct numerical modeling of viscous weakly compressible Kolmogorov-type flows in a square calculation cell. Several different conditions are observed. One of them is dominated by a large vortex with a well-defined average profile. In another state, strong chaotic large-scale fluctuations prevail. In the third state, laminar flow is observed. The nature of the realized state depends on the coefficient of kinematic viscosity of the liquid, the amplitude of the external pumping force, and the bottom friction coefficient. At constant values of the kinematic viscosity and the wave vector, a small value of the friction coefficient leads to the appearance of the first state. As the bottom friction coefficient increases, there is a transition from a flow with one large vortex to a laminar flow through a series of states with several unstable vortices, which we call chaotic flow. A rank analysis of the values of vorticity, energy, and pressure, as well as the frequency of their occurrence, is proposed. It is shown that for chaotic, vortex, laminar and transitional regimes of fluid motion, the inflection point in the rank frequency distributions of the above fields is a universal characteristic for classifying various types of flow.

Application-Specific Parallel Linear Solver for Nonlinear Harmonics Method with Implicit Time Integration

Alexey P. Duben — 2025-05-16

The present paper covers specific parallel implementation details of the nonlinear harmonics (NLH) method within an implicit time integration framework. The NLH method plays an important role in industrial turbomachinery applications as it accounts for unsteady effects in modelling of compressors and turbines on a base of low-cost stationary approaches: the flow is modelled using the Reynolds-Averaged Navier–Stokes approach, the mixing plane method is used for the rotorstator interface, and only one periodic sector of a blade passage per row is considered. The main focus is on the adaptation of the linear solver used in the Newtonian process of the implicit scheme. The goal of this work is to significantly reduce memory consumption and improve performance. This goal is achieved by using a specialized block sparse matrix storage format, adapted linear solver preconditioners with approximate inverse diagonal blocks, and a combination of single- and double-precision real number formats.

Modelling and Supercomputer Simulation of Hinged Rotor

Ilya V. Abalakin — 2025-05-16

The paper presents a computational technology of numerical simulation of turbulent flow over a hinged rotor on high-performance heterogeneous computer systems. A key part of the technology is the developed mathematical model describing the complex motions of triple-hinged rigid blades of a helicopter under the action of external and aerodynamic forces and its implementation using an original unstructured mesh-deformation algorithm. The mesh-deformation method exploits an auxiliary web-structured mesh with its elastic compression-expansion controlled by low-cost quasi-one-dimensional strand-based algorithms. The mechanics model is verified by solving the pendulum problems. To demonstrate the correctness of the developed techniques, the problems on taper stabilization and blade motion under cyclic control for model helicopter rotors are considered. All the presented computations are carried out using the code NOISEtte for solving aerodynamics and aeroacoustics problem. The code implements higher-accuracy methods of computational fluid dynamics on unstructured mixed-element meshes and operates with a high efficiency on modern supercomputers with arbitrary architectures including CPU cores and GPU accelerators.

Methodology for Scale-Resolving Simulation of Unsteady Effects in Turbomachines

Alexey P. Duben — 2025-05-16

Methodology for studying effects associated with periodic unsteady impact of neighbouring rows in turbomachines is presented. The two-stage procedure of an investigation is as follows: simulation using an approach based on solving the Reynolds-averaged Navier–Stokes equations (RANS) of an entire turbomachine at the first stage and scale-resolving simulation (SRS) of a particular row at the second. The methodology exploits the following methods and technologies, which are implemented in the NOISEtte computational algorithm: the nonlinear harmonics method as a RANS approach to obtain unsteady inflow parameters for SRS; the hybrid Improved Delayed Detached Eddy Simulation approach for SRS of the row under detailed study. SRS considers using the dynamic synthetic turbulence generator in a form of volumetric source terms (VSTG) to reproduce unsteady periodic turbulent perturbations. A dynamic version of the VSTG, the parameters of which depend on the flow upstream the source region, is formulated. Details of the parallel heterogeneous implementation of the dynamic VSTG are discussed. To demonstrate the applicability of the presented methodology, a simulation of non-stationary effects in a cascade of T106 low-pressure turbine blades was performed.

Numerical Study of Noise Sources Generated by Wing of Supersonic Business Jet in Landing Mode

Tatiana K. Kozubskaya — 2025-05-16

The paper is devoted to the numerical study of wing noise for the prototype of supersonic business jet in landing mode. Near the wing, the acoustics is simulated using the CFD/CAA methods within the Delayed Detached Eddy Simulation approach. The Ffowcs Williams–Hawkings method is used for calculation of noise radiation in the far field. To localize the near-field acoustic sources, the advanced postprocessing including the numerical beamforming method is applied. The numerical beamforming formulated for monopole- and dipole-type sources allowed for detecting the main sources of the wing noise in the vicinity of leading and trailing edges. Analysis of the sound pressure level calculated for signals recorded on the Ffowcs Williams–Hawkings surface during the CFD simulations generally confirmed these results. Direct comparison of the noise spectra calculated by the Ffowcs Williams–Hawkings and numerical beamforming methods is provided for selected mid-field points. According to the presented noise radiation pattern, the far-field noise generated by the considered wing in landing mode has dominating dipole-type component for frequencies lower than 250 Hz and dominating monopole-type component for frequencies higher than 1 kHz.