Parallel Supercomputers Research Articles

Simulating Nonlinear Radiation Diffusion Through Quantum Computing

The recent success of the fusion ignition experiment at Lawrence-Livermore National Laboratory has renewed excitement for inertial confinement fusion. Designing such experiments relies on computational modeling using radiation hydrodynamic codes run on massively parallel supercomputers which simulate hydrodynamic flows, radiation diffusion, and thermonuclear burn. Constructing a quantum algorithm for radiation hydrodynamics that provides a quantum speedup is of great interest. A recent quantum algorithm that solves the Navier-Stokes equations of hydrodynamics with a quadratic speedup marks a first step towards this goal. Here we take the next step and present a quantum algorithm for nonlinear radiation diffusion that also has a quadratic speedup. To verify the algorithm we consider radiation striking a cold, optically thick target that generates a Marshak wave in the target. Results of numerical simulation of the quantum algorithm are compared with those of a standard partial differential equation solver and excellent agreement is found.

A parallel compositional reservoir simulator for large-scale CO2 geological storage modeling and assessment

Storing CO2 in deep aquifers and depleted gas reservoirs is an effective way to achieve carbon neutrality. However, the numerical simulation of CO2 storage in these formations is challenging due to the complexity of gases-brine systems. The number of gas species included in the gases-brine fluid models of existing simulators cannot meet the rapidly evolving CO2 sequestration scenarios. To address this intricate issue, we developed a three-dimensional fully implicit parallel CO2 geological storage simulator (PRSI-CGCS) on distributed-memory computers based on our in-house parallel platform. This simulator uses a compositional fluid model with a diverse range of gas species, including CO2, C1 ~ C3, N2, H2S, as well as newly added gases H2 and O2, which may be encountered in geological CO2 storages. Besides, we provide more suitable scaling factors for different gases in the stability analysis bypassing (SAB) method to accelerate the gases-brine phase equilibrium calculations. PRSI-CGCS does not incorporate energy conservation equations, and salt precipitation or dissolution is also not considered. Numerical experiments show that our simulator is scalable, robust and validated to simulate large-scale CO2 storage problems with hundreds of millions of grid blocks on a parallel supercomputer cluster. Besides, after our modification on scaling factors, the SAB method can reduce the number of stability analyses by 61.39 % to 88.71 %, thereby reducing simulation time. Furthermore, case studies indicate that injecting O2 and H2 along with CO2 reduces the stability or capacity of CO2 storage and increases the pressure required for injection. However, this impact is not significant when the impurity content is less than 10 %.

GENESIS 2.1: High-Performance Molecular Dynamics Software for Enhanced Sampling and Free-Energy Calculations for Atomistic, Coarse-Grained, and Quantum Mechanics/Molecular Mechanics Models.

GENeralized-Ensemble SImulation System (GENESIS) is a molecular dynamics (MD) software developed to simulate the conformational dynamics of a single biomolecule, as well as molecular interactions in large biomolecular assemblies and between multiple biomolecules in cellular environments. To achieve the latter purpose, the earlier versions of GENESIS emphasized high performance in atomistic MD simulations on massively parallel supercomputers, with or without graphics processing units (GPUs). Here, we implemented multiscale MD simulations that include atomistic, coarse-grained, and hybrid quantum mechanics/molecular mechanics (QM/MM) calculations. They demonstrate high performance and are integrated with enhanced conformational sampling algorithms and free-energy calculations without using external programs except for the QM programs. In this article, we review new functions, molecular models, and other essential features in GENESIS version 2.1 and discuss ongoing developments for future releases.

The Effective Fragment Molecular Orbital Method: Achieving High Scalability and Accuracy for Large Systems.

The effective fragment molecular orbital (EFMO) method has been developed to predict the total energy of a very large molecular system accurately (with respect to the underlying quantum mechanical method) and efficiently by taking advantage of the locality of strong chemical interactions and employing a two-level hierarchical parallelism. The accuracy of the EFMO method is partly attributed to the accurate and robust intermolecular interaction prediction between distant fragments, in particular, the many-body polarization and dispersion effects, which require the generation of static and dynamic polarizability tensors by solving the coupled perturbed Hartree-Fock (CPHF) and time-dependent HF (TDHF) equations, respectively. Solving the CPHF and TDHF equations is the main EFMO computational bottleneck due to the inefficient (serial) and I/O-intensive implementation of the CPHF and TDHF solvers. In this work, the efficiency and scalability of the EFMO method are significantly improved with a new CPU memory-based implementation for solving the CPHF and TDHF equations that are parallelized by either message passing interface (MPI) or hybrid MPI/OpenMP. The accuracy of the EFMO method is demonstrated for both covalently bonded systems and noncovalently bound molecular clusters by systematically examining the effects of basis sets and a key distance-related cutoff parameter, Rcut. Rcut determines whether a fragment pair (dimer) is treated by the chosen ab initio method or calculated using the effective fragment potential (EFP) method (separated dimers). Decreasing the value of Rcut increases the number of separated (EFP) dimers, thereby decreasing the computational effort. It is demonstrated that excellent accuracy (<1 kcal/mol error per fragment) can be achieved when using a sufficiently large basis set with diffuse functions coupled with a small Rcut value. With the new parallel implementation, the total EFMO wall time is substantially reduced, especially with a high number of MPI ranks. Given a sufficient workload, nearly ideal strong scaling is achieved for the CPHF and TDHF parts of the calculation. For the first time, EFMO calculations with the inclusion of long-range polarization and dispersion interactions on a hydrated mesoporous silica nanoparticle with explicit water solvent molecules (more than 15k atoms) are achieved on a massively parallel supercomputer using nearly 1000 physical nodes. In addition, EFMO calculations on the carbinolamine formation step of an amine-catalyzed aldol reaction at the nanoscale with explicit solvent effects are presented.

Molecular generation strategy and optimization based on A2C reinforcement learning in de novo drug design.

In the field of pharmacochemistry, it is a time-consuming and expensive process for the new drug development. The existing drug design methods face a significant challenge in terms of generation efficiency and quality. In this paper, we proposed a novel molecular generation strategy and optimization based on A2C reinforcement learning. In molecular generation strategy, we adopted transformer-DNN to retain the scaffolds advantages, while accounting for the generated molecules' similarity and internal diversity by dynamic parameter adjustment, further improving the overall quality of molecule generation. In molecular optimization, we introduced heterogeneous parallel supercomputing for large-scale molecular docking based on message passing interface communication technology to rapidly obtain bioactive information, thereby enhancing the efficiency of drug design. Experiments show that our model can generate high-quality molecules with multi-objective properties at a high generation efficiency, with effectiveness and novelty close to 100%. Moreover, we used our method to assist shandong university school of pharmacy to find several candidate drugs molecules of anti-PEDV. The datasets involved in this method and the source code are freely available to academic users at https://github.com/wq-sunshine/MomdTDSRL.git.

Task graph-based performance analysis of parallel-in-time methods

In this paper, we present a performance model based on task graphs for various iterative parallel-in-time (PinT) methods. PinT methods have been developed to speed up the simulation time of time-dependent problems using modern parallel supercomputers. The performance model is based on a data-driven notation of the methods, from which a task graph is generated. Based on this task graph and a distribution of time points across processes typical for PinT methods, a theoretical lower runtime bound for the method can be obtained, as well as a prediction of the runtime for a given number of processes. In particular, the model is able to cover the large parameter space of PinT methods and make predictions for arbitrary parameter settings. Here, we describe a general procedure for generating task graphs based on three iterative PinT methods, namely, Parareal, multigrid-reduction-in-time (MGRIT), and the parallel full approximation scheme in space and time (PFASST). Furthermore, we discuss how these task graphs can be used to analyze the performance of the methods. In addition, we compare the predictions of the model with parallel simulation times using five different PinT libraries.

The effects of incident light wavelength difference on the collective stimulated Brillouin scattering in plasmas

The first laser–plasma interaction experiment using lasers of eight beams grouped into one octad has been conducted on the Shenguang Octopus facility. Although each beam intensity is below its individual threshold for stimulated Brillouin backscattering (SBS), collective behaviors are excited to enhance the octad SBS. In particular, when two-color/cone lasers with wavelength separation 0.3 nm are used, the backward SBS reflectivities show novel behavior in which beams of longer wavelength achieve higher SBS gain. This property of SBS can be attributed to the rotation of the wave vectors of common ion acoustic waves due to the competition of detunings between geometrical angle and wavelength separation. This mechanism is confirmed using massively parallel supercomputer simulations with the three-dimensional laser–plasma interaction code LAP3D.

Acceleration of generalized replica exchange with solute tempering simulations of large biological systems on massively parallel supercomputer.

Generalized replica exchange with solute tempering (gREST) is one of the enhanced sampling algorithms for proteins or other systems with rugged energy landscapes. Unlike the replica-exchange molecular dynamics (REMD) method, solvent temperatures are the same in all replicas, while solute temperatures are different and are exchanged frequently between replicas for exploring various solute structures. Here, we apply the gREST scheme to large biological systems containing over one million atoms using a large number of processors in a supercomputer. First, communication time on a multi-dimensional torus network is reduced by matching each replica to MPI processors optimally. This is applicable not only to gREST but also to other multi-copy algorithms. Second, energy evaluations, which are necessary for the multistate bennet acceptance ratio (MBAR) method for free energy estimations, are performed on-the-fly during the gREST simulations. Using these two advanced schemes, we observed 57.72 ns/day performance in 128-replica gREST calculations with 1.5 million atoms system using 16,384 nodes in Fugaku. These schemes implemented in the latest version of GENESIS software could open new possibilities to answer unresolved questions on large biomolecular complex systems with slow conformational dynamics.

Topology optimization using the lattice Boltzmann method for unsteady natural convection problems

This paper proposes a density-based topology optimization method for natural convection problems using the lattice Boltzmann method (LBM). As the LBM can be developed as a completely explicit scheme, its attractive features over the traditional ones, such as the finite element method, are (1) suitability for solving unsteady flow problems and (2) scalability for large-scale parallel computing. We develop an LBM code for solving unsteady natural convection problems and provide its sensitivity analysis based on the so-called adjoint lattice Boltzmann method. Notably, the adjoint equation is derived from the discrete particle velocity Boltzmann equation and can be solved similarly to the original LBM concerning unsteady natural convection problems. We first show that the proposed method can produce similar results to the previous work in a steady-state natural convection problem. We then demonstrate the efficacy of the proposed method through 2D numerical examples concerning unsteady natural convection. As a large-scale problem, we tackle a 3D unsteady natural convection problem on a parallel supercomputer. All the developed codes written in C++ are available at https://github.com/PANFACTORY/PANSLBM2.git.

Electronic Excitation Response of DNA to High-Energy Proton Radiation in Water.

The lack of molecular-level understanding for the electronic excitation response of DNA to charged particle radiation, such as high-energy protons, remains a fundamental scientific bottleneck in advancing proton and other ion beam cancer therapies. In particular, the dependence of different types of DNA damage on high-energy protons represents a significant knowledge void. Here we employ first-principles real-time time-dependent density functional theory simulation, using a massively parallel supercomputer, to unravel the quantum-mechanical details of the energy transfer from high-energy protons to DNA in water. The calculations reveal that protons deposit significantly more energy onto the DNA sugar-phosphate side chains than onto the nucleobases, and greater energy transfer is expected onto the DNA side chains than onto water. As a result of this electronic stopping process, highly energetic holes are generated on the DNA side chains as a source of oxidative damage.

Modeling of radiation heat transfer in dense-bed flows of solids in indirectly heated rotary kilns

• modification and implementation of a DOM-radiation model which combines all required physical submodels to simulate heat radiation to dense beds of particles. • Full time and space resolved simulation of thermal conversion of wet moving bulk in a large-scale rotary kiln reactor. • Analytical validation of model and justification with laboratory-scale experiments. • Efficiency optimization of the numerical solver for industrial-scale and long-time simulations on parallel supercomputers. • Parameter studies for large-scale reactors to investigate conversion rate and product quality. This work presents further development and validation of the Discrete Ordinates Model for thermal radiation implemented in OpenFOAM® for application to packed beds of particles. The radiation model is an important part of a more comprehensive model for simulating the thermochemical conversion of the discrete phase (here for instance wet biomass particles). The comprehensive Eulerian-Lagrangian model is part of a three-dimensional, time-resolved simulation of the essential chemical and physical processes occurring within and in-between particles in a moving bed. For the thermal treatment of solid particles, convective and radiative heat transfer couple the energy exchange between the reactor wall, gas- and disperse phase. The original implementation of the finite volume Discrete Ordinate Model (fvDOM) present in OpenFOAM® is valid for a dilute particulate phase and neglects the effect of local opacity due to the existence of individual particles. In the present application, a dense-packed bed of the particulate phase exists in the reactor and, therefore, this direction-based radiation model is adjusted for a computational cell with arbitrary particle volume fractions. To validate the results with the present thermal radiation model, a simple test case is first presented, where a bed of particles is heated from the top of the computational domain. A second test case relates to an experimentally investigated laboratory-scale reactor. The results of the improved fvDOM are compared to the original implementation in OpenFOAM® and the more simple and computationally cheap P-1 radiation model. In the presence of a packed bed, the P-1 model largely overpredicts the radiative heat transfer while the original fvDOM underpredicts the heat flux by about 15 % for the first test case. The new improved model delivers results within 1% deviation at the expense of maximum of 10 % increase in the computational time. Large-scale parallel simulations of real setups are crucial tools to predict the efficiency of a process and improve the operational parameters. To achieve the pre-defined product quality at minimum cost, an example of the implemented radiation model in thermochemical conversion of wet biomass in a rotary drum is been given and the importance of the radiation heat transport to the bulk is highlighted.

Hybrid interconnection networks for reducing hardware cost and improving path diversity based on fat‐trees and hypercubes

AbstractFat‐tree topologies are widely used in interconnect network designs for parallel supercomputers. In the classic fat‐tree, compute nodes are connected to leaf stage switches by links. Given a large number of compute nodes, many switches and links are required, resulting in high hardware costs. To solve this problem, this paper proposes two hybrid topologies, ‐Cube ‐Ary ‐Tree (CAT) and Mirrored ‐Cube ‐Ary ‐Tree (MiCAT), based on fat‐tree and hypercube. Instead of connecting compute nodes directly to a leaf switch, we connect a ‐cube to the switch, and each switch in the ‐cube part connects compute nodes. That is, this ‐cube consists of switches and compute nodes. We give the shortest path routing algorithms and evaluate the path diversity, cost, performance, and average packet latency of CAT and MiCAT. The results show that CAT and MiCAT can save up to switches and links in a large‐scale parallel system, for example, compared to fat‐trees, and meanwhile, both CAT and MiCAT have higher path diversities than fat‐trees.

(Invited) Polaronic Transport in Iron Oxides from Density Functional Theory

Iron oxides such as hematite (α-Fe2O3) play an important role in diverse fields ranging from biogeochemistry to photocatalysis. In the majority of these cases their electrical properties are key because they facilitate electron transfer reactions occurring at their interfaces. The iron oxides tend to be wide band gap semiconductors with a narrow d band. Consequently, electron and hole charge carriers localize by self-trapping to form polarons, carriers whose mobilities are tied to the lattice distortions they create. Site-to-site polaron transport is thermally activated and the rate depends on the energy to reorganize the local structure into a suitable transition state, as well as the strength of the electronic coupling in that transient configuration. In turn, these depend on factors that determine the atomic and electronic structure, including crystallographic direction.Because the structure of polarons, the reorganization energy, and the electronic coupling matrix element are not readily experimentally accessible, quantum mechanical calculations have been extraordinarily useful. However, this also means that the accuracy of predicting polaronic charge carrier mobilities depends strongly on the computational method used. In the ~20 years since the first application of ab initio cluster model computations to this topic for hematite, both the methodological rigor and the supporting computational power have greatly advanced. After a brief historical overview, this talk will highlight current findings obtained with gap-optimized hybrid density functional theory with periodic boundary conditions performed on massively parallel supercomputers. Comparisons will be made to the original calculations for hematite published in 2003. The DFT-based predictions for other iron oxide phases will also be discussed.Specifically, we will present calculations of both the electron and hole polaron structures and associated reorganization energies for the bulk of hematite, lepidocrocite (γ-FeOOH), goethite (α-FeOOH) and white rust (Fe(OH)2).1 Through the use of gap-optimized hybrid functionals and large supercells under periodic boundary conditions, we remove some of the complications and uncertainties present in earlier cluster model calculations. It is found that while the hole polaron in these materials generally localizes onto a single iron site, the electron polaron delocalizes across two iron sites of the same spin layer as a consequence of the lower reorganization energy for electrons compared to holes. An exception to these trends is the hole of goethite, which according to our calculations does not form a localized polaron.For hematite,2 we find that upon ionization the hole relaxes from a delocalized band state to a polaron localized on a single iron atom with localization induced by tetragonal distortion of the six surrounding iron-oxygen bonds. This distortion is responsible for sluggish hopping transport in the Fe-bilayer, characterized by an activation energy of 70 meV and a hole mobility of 0.031 cm2/(V s). By contrast, the excess electron induces a smaller distortion of the iron-oxygen bonds resulting in delocalization over two neighboring Fe units. We find that 2-site delocalization is advantageous for charge transport due to the larger spatial displacements per transfer step. As a result, the electron mobility is predicted to be a factor of three higher than the hole mobility, 0.098 cm2/(V s), in qualitative agreement with experimental observations.The advances made on the theory and simulation front have provided increasing insight into the nature of polaronic transport in the iron oxides, and the methods have been widely duplicated successfully to a variety of materials. However, there is tremendous room for expansion of the work to treat the more challenging aspects governing iron oxide behavior in natural systems and in device application, including the role of interfaces, surface potential, and defects to name a few. Theoretical computations seem poised now more than ever to help unlock the fundamental insight into charge carrier transport in polaronic materials needed to understand and adapt their electrical transport properties. References Ahart C.S., Blumberger J., Rosso K.M. (2020) Polaronic structure of excess electrons and holes for a series of bulk iron oxides. Physical Chemistry Chemical Physics, 22, 10699-10709.Ahart C.S., Rosso K.M., Blumberger J. (2022) Electron and hole mobilities in bulk hematite from spin-constrained density functional theory. Journal of the American Chemical Society, 144, 4623-4632.

CitcomSVE: A Three‐Dimensional Finite Element Software Package for Modeling Planetary Mantle’s Viscoelastic Deformation in Response to Surface and Tidal Loads

AbstractThis article presents a comprehensive benchmark study for the newly updated and publicly available finite element code CitcomSVE for modeling dynamic deformation of a viscoelastic and incompressible planetary mantle in response to surface and tidal loading. A complete description of CitcomSVE’s finite element formulation including calculations of the sea‐level change, polar wander, apparent center of mass motion, and removal of mantle net rotation is presented. The 3‐D displacements and displacement rates and the gravitational potential anomalies are solved with CitcomSVE for three benchmark problems using different spatial and temporal resolutions: (a) surface loading of single harmonics, (b) degree‐2 tidal loading, and (c) the ICE‐6G GIA model. The solutions are compared with semi‐analytical solutions for error analyses. The benchmark calculations demonstrate the accuracy and efficiency of CitcomSVE. For example, for a typical ICE‐6G GIA calculation with a 122‐ky glaciation‐deglaciation history, time increment of 100 years, and ∼50 km (or ∼0.5°) surface horizontal resolution, it takes ∼4.5 hr on 96 CPU cores to complete with about 1% and 5% errors for displacements and displacement rates, respectively. Error analyses shows that CitcomSVE achieves a second order accuracy, but the errors are insensitive to temporal resolution. CitcomSVE achieves the parallel computational efficiency &gt;75% for using up to 6,144 CPU cores on a parallel supercomputer. With its accuracy, computational efficiency and its open‐source public availability, CitcomSVE provides a powerful tool for modeling viscoelastic deformation of a planetary mantle with 3‐D mantle viscous and elastic structures in response to surface and tidal loading problems.

Scalability in Computing and Robotics

Efficient engineered systems require scalability. A scalable system has increasing performance with increasing system size. In an ideal case, the increase in performance (e.g., speedup) corresponds to the number of units that are added to the system. However, if multiple units work on the same task, then coordination among these units is required. This coordination can introduce overheads with an impact on system performance. The coordination costs can lead to sublinear improvement or even diminishing performance with increasing system size. However, there are also systems that implement efficient coordination and exploit collaboration of units to attain superlinear improvement. Modeling the scalability dynamics is key to understanding efficient systems. Known laws of scalability, such as Amdahl's law, Gustafson's law, and Gunther's Universal Scalability Law, are minimalistic phenomenological models that explain a rich variety of system behaviors through concise equations. While useful to gain general insights, the phenomenological nature of these models may limit the understanding of the underlying dynamics, as they are detached from first principles that could explain coordination overheads among units. Through a decentralized system approach, we propose a general model based on generic interactions between units that is able to describe, as specific cases, any general pattern of scalability included by previously reported laws. The proposed general model of scalability is built on first principles, or at least on a microscopic description of interaction between units, and therefore has the potential to contribute to a better understanding of system behavior and scalability. We show that this model can be applied to a diverse set of systems, such as parallel supercomputers, robot swarms, or wireless sensor networks, creating a unified view on interdisciplinary design for scalability.

GronOR: Scalable and Accelerated Nonorthogonal Configuration Interaction for Molecular Fragment Wave Functions.

GronOR is a program package for nonorthogonal configuration interaction calculations. Electronic wave functions are constructed in terms of antisymmetrized products of multiconfiguration molecular fragment wave functions. The computational complexity of the nonorthogonal methodologies implemented in GronOR applied to large molecular assemblies requires a design that takes full advantage of massively parallel supercomputer architectures and accelerator technologies. This work describes the implementation strategy and resulting performance characteristics. In addition to parallelization and acceleration, the software development strategy includes aspects of fault resiliency and heterogeneous computing. The program was designed for large-scale supercomputers but also runs effectively on small clusters and workstations for small molecular systems. GronOR is available as open source to the scientific community.

Performance Evaluation of Massively Parallel Systems Using SPEC OMP Suite

Performance analysis plays an essential role in achieving a scalable performance of applications on massively parallel supercomputers equipped with thousands of processors. This paper is an empirical investigation to study, in depth, the performance of two of the most common High-Performance Computing architectures in the world. IBM has developed three generations of Blue Gene supercomputers—Blue Gene/L, P, and Q—that use, at a large scale, low-power processors to achieve high performance. Better CPU core efficiency has been empowered by a higher level of integration to gain more parallelism per processing element. On the other hand, the Intel Xeon Phi coprocessor armed with 61 on-chip x86 cores, provides high theoretical peak performance, as well as software development flexibility with existing high-level programming tools. We present an extensive evaluation study of the performance peaks and scalability of these two modern architectures using SPEC OMP benchmarks.

The modeling of two-dimensional vortex flows in a cylindrical channel using parallel calculations on a supercomputer

Objectives . The study aimed to examine vortex structures formed during the interaction of incident and reflected shock waves in a cylindrical channel. The shock wave was described by the Hugoniot relations, which make it possible to determine the parameters of the gas behind the shock front by a given Mach number and the values of the gasdynamic parameters ahead of the pressure jump. The propagation of a strong shock wave (Mach number was 20) in argon was simulated. Methods . The methods of mathematical modeling were used. A parallel algorithm for solving two-dimensional equations of gas dynamics in cylindrical coordinates (r, z, t) was developed and a new version of the NUTCY_ps program created. The calculations were performed on an MVS-100K supercomputer. Results. Two methods of parallelization when solving a system of equations were considered. Using a specific task as an example, a comparison of the effectiveness of these methods was conducted. A parallel algorithm was developed and a program was upgraded for solving two-dimensional equations of gas dynamics in cylindrical coordinates (r, z are spatial coordinates, t is time). Numerical calculations were performed to simulate: 1) the shock wave incidence to and reflection from a metal screen; 2) the propagation of the shock wave through a hole in the screen; 3) the propagation of the shock wave through a cylindrical channel and its reflection from the bottom of the channel and interaction with the incident wave. The results obtained by the parallel supercomputer with different numbers of processors are presented. It is shown that using 16 processors, it is possible to reduce the computation time for getting a solution for the test problem by approximately 12 times. Conclusions . It is shown that the interaction of incident shock wave and the one reflected at an angle leads to the formation of regions with low and high gas densities, as well as vortex flows. The vortex interaction area (turbulence zone) gets a complex shape. The article discusses the possibility of carrying out full-scale experiments in shock tubes or using a laser shock tube. Such studies would make it possible to compare experimental data with the results of numerical simulation and, on their basis, to develop more advanced models of turbulent motions.

Modeling Transport of SARS-CoV-2 Inside a Charlotte Area Transit System (CATS) Bus

We present in this paper a model of the transport of human respiratory particles on a Charlotte Area Transit System (CATS) bus to examine the efficacy of interventions to limit exposure to SARS-CoV-2, the virus that causes COVID-19. The methods discussed here utilize a commercial Navier–Stokes flow solver, RavenCFD, using a massively parallel supercomputer to model the flow of air through the bus under varying conditions, such as windows being open or the HVAC flow settings. Lagrangian particles are injected into the RavenCFD predicted flow fields to simulate the respiratory droplets from speaking, coughing, or sneezing. These particles are then traced over time and space until they interact with a surface or are removed via the HVAC system. Finally, a volumetric Viral Mean Exposure Time (VMET) is computed to quantify the risk of exposure to the SARS-CoV-2 under various environmental and occupancy scenarios. Comparing the VMET under varying conditions should help identify viable methods to reduce the risk of viral exposure of CATS bus passengers during the COVID-19 pandemic.

An experimental data-driven mass-spring model of flexible Calliphora wings

Insect wings can undergo significant deformation during flapping motion owing to inertial, elastic and aerodynamic forces. Changes in shape then alter aerodynamic forces, resulting in a fully coupled fluid–structure interaction (FSI) problem. Here, we present detailed three-dimensional FSI simulations of deformable blowfly (Calliphora vomitoria) wings in flapping flight. A wing model is proposed using a multi-parameter mass-spring approach, chosen for its implementation simplicity and computational efficiency. We train the model to reproduce static elasticity measurements by optimizing its parameters using a genetic algorithm with covariance matrix adaptation (CMA-ES). Wing models trained with experimental data are then coupled to a high-performance flow solver run on massively parallel supercomputers. Different features of the modeling approach and the intra-species variability of elastic properties are discussed. We found that individuals with different wing stiffness exhibit similar aerodynamic properties characterized by dimensionless forces and power at the same Reynolds number. We further study the influence of wing flexibility by comparing between the flexible wings and their rigid counterparts. Under equal prescribed kinematic conditions for rigid and flexible wings, wing flexibility improves lift-to-drag ratio as well as lift-to-power ratio and reduces peak force observed during wing rotation.