Parallel Computer Cluster Research Articles

(Invited) Challenges and Opportunities for the Computational Analysis of Electrochemical Energy Systems

Electrochemical systems are challenging to analyze and design due to their multi-dimensional, time-dependent, and multi-physics nature. For this reason, many computational electrochemical system models have been developed in the past three decades, some of which can now be found in commercial software packages, e.g., COMSOL or ANSYS Fluent. These models however: i) are based on simplifying assumptions, e.g., reduced dimensionality, infinite dilution, Tafel kinetics; ii) contain uncertain input parameters, e.g., effective transport properties and kinetic parameters; and, iii) are only validated with a limited number of experiments, e.g., polarization curves. These limitations, understandable at the time due to limited computational resources, characterization techniques, and material understanding and stability, are difficult to justify today that parallel programming and computer clusters enable scientists to perform large multi-dimensional simulations; electrochemical testing has expanded to include segmented cell testing, impedance spectroscopy, water flux estimation, and in-operando visualization; and, characterization tools and techniques have been developed to, for example, easily measure adsorption isotherms, effective proton conductivity and, within a given resolution, visualize the three-dimensional microstructure of the electrodes.Scientists working on computational analysis of electrochemical systems must acknowledge that further model development is still needed and must take advantage of new resources to improve the robustness and accuracy of cell-level models. The path is long and crooked, but it is very likely that the opportunity for a truly useful computational model, one that can be used for design, is beyond simplistic implementations and polarization curves. The effort is great, but it can be minimized by concurrent software development, as well as, by careful experimental work dedicated to parameter estimation and model validation.This presentation aims at outlining the limitations of state-of-the-art models, the challenges and opportunities of concurrent development and maintenance of a multi-scale, transient electrochemical energy system software, such as the open-source fuel cell simulation toolbox (OpenFCST) [1], and the new opportunities provided by combining advanced characterization tools with computational analysis. As an example, the development and validation of the transient, two-phase fuel cell and electrolyzer cell-level models in OpenFCST will be discussed [2]. The use of a hydrogen-pump model to study the effect of an active catalyst in CL effective proton conductivity measurements will also be discussed as an example of concurrent experimental/model design [3].

Cooperative Job Scheduling and Data Allocation in Data-Intensive Parallel Computing Clusters

Cooperative Job Scheduling and Data Allocation in Data-Intensive Parallel Computing Clusters

Implementation and analysis of high-performance cloud computing for underwater acoustic modeling applications

In the last few years advances in compute speeds have plateaued and multi/parallel processing (MPP) has been the way forward. With the advent of multiple processor cores packaged within a single microchip the architectures that were restricted to large servers has crossed over into consumer workstations. Now, more than ever, an understanding of how to maximize computational work is needed. The design and implementation of a Cloud based parallel-compute cluster for very large-scale processing of underwater acoustic models is presented. Ray-based, parabolic equation and wave number integration models were evaluated. This type of system allows for the creation of real-world sized solution sets which were previously prohibitive for most researchers. This is mainly due to their excessively long runtimes or being restricted to gated hardware, such as super-computer systems. Amazon Web Services is used as the underlying technologies for the creation of the Cloud based parallel-computer cluster, but the principles discussed are universal and may be applied to any MPP system. [Work supported by the Office of Naval Research.]

Flexible Array Probe With In-Plane Differential Multichannels for Inspection of Microdefects on Curved Surface

It is critical for maintaining the safe operation of turbine blades to inspect the potential structural damages promptly. However, the industry still lacks efficient tools for inspection of the curved surface. In this respect, in this article, a new flexible probe with in-plane differential coils is proposed. With the sophisticated configuration, the probe has four channels, which complement each other in terms of sensitivity to defects of different orientations. The in-plane differential scheme nulls the background signal and reduces the noise resulting in high sensitivity to microdefects. The probe can be bent according to the curved surface due to its flexible configuration. A three-dimensional finite-element method model based on the cluster parallel computation is utilized to study the feasibility and performance of the probe, and a prototype probe is developed and tested by scanning defects on the curved surface of a turbine blade sample. Furthermore, a data fusion method is proposed to obtain a better signal-to-noise ratio so that defects with length*width*depth = 2.5 mm*0.1 mm*0.1 mm can be identified. Based on repeated experiments, it is observed that the probability of detection with the fusion of the multichannels is higher than that of every single channel demonstrating that the data fusion increases the reliability of the inspection.

The generality of the GUGA MRCI approach in COLUMBUS for treating complex quantum chemistry.

The core part of the program system COLUMBUS allows highly efficient calculations using variational multireference (MR) methods in the framework of configuration interaction with single and double excitations (MR-CISD) and averaged quadratic coupled-cluster calculations (MR-AQCC), based on uncontracted sets of configurations and the graphical unitary group approach (GUGA). The availability of analytic MR-CISD and MR-AQCC energy gradients and analytic nonadiabatic couplings for MR-CISD enables exciting applications including, e.g., investigations of π-conjugated biradicaloid compounds, calculations of multitudes of excited states, development of diabatization procedures, and furnishing the electronic structure information for on-the-fly surface nonadiabatic dynamics. With fully variational uncontracted spin-orbit MRCI, COLUMBUS provides a unique possibility of performing high-level calculations on compounds containing heavy atoms up to lanthanides and actinides. Crucial for carrying out all of these calculations effectively is the availability of an efficient parallel code for the CI step. Configuration spaces of several billion in size now can be treated quite routinely on standard parallel computer clusters. Emerging developments in COLUMBUS, including the all configuration mean energy multiconfiguration self-consistent field method and the graphically contracted function method, promise to allow practically unlimited configuration space dimensions. Spin density based on the GUGA approach, analytic spin-orbit energy gradients, possibilities for local electron correlation MR calculations, development of general interfaces for nonadiabatic dynamics, and MRCI linear vibronic coupling models conclude this overview.

Dredging for dilution: A simulation based case study in a Tidal River

A 2-D hydrodynamic finite element model with a Lagrangian particle module is used to investigate the effects of dredging on the hydrodynamics and the horizontal dilution of pollutant particles originating from a wastewater treatment facility (WWTF) in tidal Oyster River in New Hampshire, USA. The model is driven by the semi-diurnal (M2) tidal component and includes the effect of flooding and drying of riverine mud flats. The particle tracking method consists of tidal advection plus a horizontal random walk model of sub-grid scale turbulent processes. Our approach is to perform continuous pollutant particle releases from the outfall, simulating three different scenarios: a base-case representing the present conditions and two different dredged channel/outfall location configurations. Hydrodynamics are investigated in an Eulerian framework and Lagrangian particle dilution improvement ratios are calculated for all cases. Results show that the simulated hydrodynamics are consistent with observed conditions. Eulerian and Lagrangian residuals predict an outward path suggesting flushing of pollutants on longer (>M2) time scales. Simulated dilution maps show that, in addition to dredging, the relocation of the WWTF outfall into the dredged main channel is required for increased dilution performance. The methodology presented here can be applied to similar managerial problems in all similar systems worldwide with relatively little effort, with the combination of Lagrangian and Eulerian methods working together towards a better solution. The statistical significance brought into methodology, by using a large number of particles (16000 in this case), is to be emphasized, especially with the growing number of networked parallel computer clusters worldwide. This paper improves on the study presented in Bilgili et al., 2006b, by adding an Eulerian analysis.

Large-Scale Optimization of Polymer Electrolyte Membrane Fuel Cells

The design of low-temperature polymer electrolyte membrane fuel cells (PEMFCs) consists of two separate but related problems: (1) geometry optimization, which focuses on finding the optimum channel configuration and overall cell dimensions; and (2) microstructure optimization, which focuses on finding the optimum distributions of the porosity, catalyst, and electrolyte that maximize the power density of the cell [1]. The number of design parameters in each of the above problems is relatively large, which makes the design of PEMFCs a difficult mathematical problem. For instance, let us consider problem (2): if we assume that the fuel cell is discretized using a 3-D finite element discretization with 106 nodes, the design variables consist of the mesh-point values of the porosity, catalyst concentration, and electrolyte density at each mesh point. Hence, the number of design variables is 3 ×106, which make the design of PEMFC a large-scale optimization problem. Usually, such large optimization problems are solved on parallel computer clusters, with a large number of concurrent processors. In this presentation we focus on the large-scale design of the microstructure of the PEMFC by developing a gradient-based method for the computation of the optimum (mathematically exact) distributions of the porosity, catalyst, and electrolyte that maximize the power density of the cell. It should be noted that traditional heuristic techniques such as genetic algorithms, swarm-optimization or other evolutionary algorithms are inappropriate for the large-scale optimization of PEMFCs. Indeed, the number of optimization variables in non-heuristic techniques in of the order or 5-10 variables, which is a few orders of magnitude smaller than the total number of design variables in PEMFCs. Since the computation time of heuristic techniques increases approximately exponentially with the number of design variables, the total optimization time of heuristic techniques becomes unpractical in large-scale optimization problems. The technique presented in this work is based on evaluating the sensitivity functions of the cell voltage and discharge current and applying a gradient-based maximization method to optimize the PEMFC. The sensitivity functions are computed efficiently using an adjoint space approach suitable for large-scale systems that can be simulated using 2-D or 3-D finite element models. The computational cost required to compute the sensitivity functions is relatively small since our technique requires solving only one sparse system of linear equations instead of performing multiple device simulations. For this reason, the numerical algorithm presented in this work can be implemented on single processor computers even in the case of 3-D finite element problems. Using the proposed algorithm we are able to predict the optimum distribution of platinum particles and electrolyte inside the catalyst layers, as well as the optimum porosity profile that maximizes the power density of the cell. The optimum distributions and porosity profiles depends on the positions of landings and openings, and on the geometry and dimensions of the layers. In agreement with existing experimental data, we obtain that the catalyst density should be distributed non-uniformly inside the cell in order to increase the power density of the cell (see figure). In the case of the porosity profile it is more efficient to increase the porosity towards the gas diffusion layer side in order to increase the flow of the oxygen and water vapors. In the case of the electrolyte distribution in the catalyst layers it is more efficient to increase the fraction of the electrolyte towards the membrane in order to decrease the ohmic losses of the protons. Finally, it should be noted that our numerical algorithm can also show how the optimum distributions of the catalyst, porosity and electrolyte vary with the discharge conditions. The results describe the complete profiles of the optimal catalyst and porosity by providing the full distribution of optimum locations and concentration of platinum particles in the catalyst layer, as well as the “ideal” porosity distribution of the PEMFC under the specified operating conditions. More details about the technique, the numerical implementation, and the final optimized structures will be presented at the conference.

From Gas to Stars Over Cosmic Time

From the time the first stars formed over 13 billion years ago to the present, star formation has had an unexpectedly dynamic history. At first, the star-formation rate density increased dramatically, reaching a peak 10 billion years ago of more than 10 times the present-day value. Observations of the initial rise in star formation remain difficult, poorly constraining it. Theoretical modeling has trouble predicting this history because of the difficulty in following the feedback of energy from stellar radiation and supernova explosions into the gas from which further stars form. Observations from the ground and space with the next generation of instruments should reveal the full history of star formation in the universe, and simulations appear poised to accurately predict the observed history.

A physics-based hydro-geomorphologic simulation utilizing cluster parallel computing

To conduct a large-scale hydrologic-response and landform evolution simulation at high resolution, a complex physics-based numerical model, the Integrated Hydrology Model (InHM), was revised utilizing cluster parallel computing. The parallelized InHM (ParInHM) divides the simulated area into multiple catchments based on geomorphologic features, and generates boundary-value problems for each catchment to construct simulation tasks, which are then dispatched to different computers to start the simulation. Landform evolution is considered during simulating and implemention in one framework. The dynamical Longest-Processing-Time (LPT) first scheduling algorithm is applied to job management. In addition, a pause-integrate- divide-resume routine method is used to ensure the hydrologic validity during the simulation period. The routine repeats until the entire simulation period is finished. ParInHM has been tested in a computer cluster that uses 16 processors for the calculation, to simulate 100 years’ hydrologic-response and soil erosion for the 117-km2 Kaho’olawe Island in the Hawaiian Islands under two different mesh resolutions. The efficiency of ParInHM was evaluated by comparing the performance of the cluster system utilizing different numbers of processors, as well as the performance of non-parallelized system without domain decomposition. The results of this study show that it is feasible to conduct a regional-scale hydrologic-response and sediment transport simulation at high resolution without demanding significant computing resources.

A Parallel Block Predictor-Corrector Method by Python-Based Distributed Computing

This paper is to discuss how Python can be used in designing a cluster parallel computation environment in numerical solution of some block predictor-corrector method for ordinary differential equations. In the parallel process, MPI-2(message passing interface) is used as a standard of MPICH2 to communicate between CPUs. The operation of data receiving and sending are operated and controlled by mpi4py which is based on Python. Implementation of a block predictor-corrector numerical method with one and two CPUs respectively is used to test the performance of some initial value problem. Minor speed up is obtained due to small size problems and few CPUs used in the scheme, though the establishment of this scheme by Python is valuable due to very few research has been carried in this kind of parallel structure under Python.

Effects of computational grids and turbulence models on numerical simulation of centrifugal pump with CFD

In order to verify accurately the effects of computational grids and turbulence models on CFD numerical simulation of centrifugal pump, the calculation results of the different mesh numbers coupling with five kinds of turbulence models are compared. These parts of models (e.g. wear ring and tongue) are meshed with the local refinement technology. At design condition, with CFX software on the Dawning TC3600 parallel computer cluster, the calculation results of phase coupling six different mesh numbers from 1 million to 25 million with five different turbulence models (K-Epsilon, SSG Reynolds Stress, K epsilon EARSM, RNG K-Epsilon, K-Omega) are used to make performance prediction and to analyze the internal flow field. And the comparison between performance prediction results and experimental results are analyzed. It is confirmed that the calculation results are quite different with different turbulence models, and the result under K epsilon EARSM model is better than the others. And it shows that the internal flow in centrifugal pump is depicted more perfect with the increase of mesh numbers.

Energy-saving control of parallel computer cluster

In order to save the energy in using the parallel computer cluster,a scheme was presented in which Open Portable Batch System(OpenPBS) was used to control the parallel computers' boot and shutdown automatically.Homeostasis can be achieved between task computing and energy saving.

Design of Microstrip Reflectarray Antenna Using a Genetic Algorithm Based Optimization Method

The design of the microstrip reflectarray antenna is commonly based on the reflection phase curve, but this is not an accurate method, as many parameters have been neglected in the design procedure. This work explores the genetic algorithm in conjunction with full-wave simulation and the cluster parallel computation to design a microstrip reflectarray antenna. A microstrip reflectarray antenna, which consists of a 7 × 7 rectangular-patch/ring-combination reflection elements and is illuminated by a patch antenna, is designed for high gain. Results demonstrate that in comparison with the design method based on the reflection phase curve, the proposed optimization method is able to design a microstrip reflectarray antenna with better performances; e.g., the gain has been considerably improved (from 18.1 dBi to 19.5 dBi) at the design frequency of 5.8 GHz.

Structure Based Protein Multiple Sequence Alignment Algorithm on a Parallel System

To enhance the speed and efficiency of structure based algorithms for protein Multiple Sequence alignment we need parallel processing. Secondary and tertiary structure of proteins are more important and informative then its primary structure. We get more accurate information's about proteins when we conduct structure based matching. In this paper we have to focus on how to use parallel computing in terms of structure based algorithms. In case of protein structure based multiple sequence alignment several searches and structure matching are involved which require a lot of time and processing speed. If we use a parallel computer cluster or Grid then we can reduce processing time and get an optimal result. It will predict the function of protein from its structure.

A study of light scattering by wavelength-sized particles covered by much smaller grains using the superposition T-matrix method

By using the results of a direct, numerically exact solution of the macroscopic Maxwell equations we analyze the behavior of light scattering characteristics for polydisperse spherical particles covered with a large number of smaller grains. We show that the effect of the presence of microscopic dust on the surfaces of wavelength-sized particles depends on the particle absorption and the relative size of irregularities. In our computations, a new parallel superposion T -matrix code developed for use on parallel computer clusters is applied.

A multiple sphere T-matrix Fortran code for use on parallel computer clusters

A general-purpose Fortran-90 code for calculation of the electromagnetic scattering and absorption properties of multiple sphere clusters is described. The code can calculate the efficiency factors and scattering matrix elements of the cluster for either fixed or random orientation with respect to the incident beam and for plane wave or localized-approximation Gaussian incident fields. In addition, the code can calculate maps of the electric field both interior and exterior to the spheres. The code is written with message passing interface instructions to enable the use on distributed memory compute clusters, and for such platforms the code can make feasible the calculation of absorption, scattering, and general EM characteristics of systems containing several thousand spheres.

Parallel hyperbolic PDE simulation on clusters: Cell versus GPU

Increasingly, high-performance computing is looking towards data-parallel computational devices to enhance computational performance. Two technologies that have received significant attention are IBM's Cell Processor and NVIDIA's CUDA programming model for graphics processing unit (GPU) computing. In this paper we investigate the acceleration of parallel hyperbolic partial differential equation simulation on structured grids with explicit time integration on clusters with Cell and GPU backends. The message passing interface (MPI) is used for communication between nodes at the coarsest level of parallelism. Optimizations of the simulation code at the several finer levels of parallelism that the data-parallel devices provide are described in terms of data layout, data flow and data-parallel instructions. Optimized Cell and GPU performance are compared with reference code performance on a single x86 central processing unit (CPU) core in single and double precision. We further compare the CPU, Cell and GPU platforms on a chip-to-chip basis, and compare performance on single cluster nodes with two CPUs, two Cell processors or two GPUs in a shared memory configuration (without MPI). We finally compare performance on clusters with 32 CPUs, 32 Cell processors, and 32 GPUs using MPI. Our GPU cluster results use NVIDIA Tesla GPUs with GT200 architecture, but some preliminary results on recently introduced NVIDIA GPUs with the next-generation Fermi architecture are also included. This paper provides computational scientists and engineers who are considering porting their codes to accelerator environments with insight into how structured grid based explicit algorithms can be optimized for clusters with Cell and GPU accelerators. It also provides insight into the speed-up that may be gained on current and future accelerator architectures for this class of applications. Program summary Program title: SWsolver Catalogue identifier: AEGY_v1_0 Program summary URL: http://cpc.cs.qub.ac.uk/summaries/AEGY_v1_0.html Program obtainable from: CPC Program Library, Queen's University, Belfast, N. Ireland Licensing provisions: GPL v3 No. of lines in distributed program, including test data, etc.: 59 168 No. of bytes in distributed program, including test data, etc.: 453 409 Distribution format: tar.gz Programming language: C, CUDA Computer: Parallel Computing Clusters. Individual compute nodes may consist of x86 CPU, Cell processor, or x86 CPU with attached NVIDIA GPU accelerator. Operating system: Linux Has the code been vectorised or parallelized?: Yes. Tested on 1-128 x86 CPU cores, 1-32 Cell Processors, and 1-32 NVIDIA GPUs. RAM: Tested on Problems requiring up to 4 GB per compute node. Classification: 12 External routines: MPI, CUDA, IBM Cell SDK Nature of problem: MPI-parallel simulation of Shallow Water equations using high-resolution 2D hyperbolic equation solver on regular Cartesian grids for x86 CPU, Cell Processor, and NVIDIA GPU using CUDA. Solution method: SWsolver provides 3 implementations of a high-resolution 2D Shallow Water equation solver on regular Cartesian grids, for CPU, Cell Processor, and NVIDIA GPU. Each implementation uses MPI to divide work across a parallel computing cluster. Additional comments: Sub-program numdiff is used for the test run.

A Fully Implicit, Compact Finite Difference Method for the Numerical Solution of Unsteady Laminar Flames

Forced laminar diffusion flames form an important class of problems that can help to bridge the significant gap between steady laminar flames in simple burner configurations and the turbulent flames found in many practical combustors. Such flames offer a much wider range of interactions between convection, diffusion, and chemical reaction than can be examined under steady-state conditions, and yet detailed simulations of them should be feasible without having to resort to “modeling” any of the relevant physics, above all without having prematurely to reduce the large kinetic mechanisms typical of hydrocarbon fuels. Nevertheless, the computation of time-dependent laminar diffusion flames with conventional numerical methods is hindered by technical challenges that, while not new, are more troublesome to surmount than in the calculation of otherwise similar, unforced flames. First, the intricate spatiotemporal coupling between fluid dynamics and combustion thermochemistry ensures that spurious numerical diffusion or spatial under-resolution of the mixing process at any stage of the computation can lead to inaccurate prediction of flame characteristics for the remainder thereof. Second, relatively long simulated flow times and extremely short chemical time scales make many standard time integration algorithms impractical on all but the largest parallel computer clusters. This paper introduces a new numerical approach for time-varying laminar flames that addresses these challenges through the use of high order compact finite difference schemes within a robust, fully implicit solver based on a Jacobian-free Newton–Krylov method. The capabilities of this implicit-compact solver are demonstrated on a periodically forced axisymmetric laminar jet diffusion flame with one-step Arrhenius chemistry, and the results are compared to those of a conventional low order finite difference solver.

Unified formulation for URANS and LES in DxUNSp code

The aim of this work is to find a unified and efficient implementation of a LES turbulence model in an existing URANS CFD code, initially based on unsteady RANS equations with a k-e turbulence model. This code has the capability to be developed for nonreacting/reacting multifluid flows in research applications. The paper intends to present mainly three aspects of this implementation for unstructured mesh based solvers, for high Reynolds compressible flows: the influence of the numerical reconstruction scheme on the results for compressible LES, the influence of the compressible SGS modeling and the efficient implementation of a wall-law based approach for complex geometry. The results will be presented for a test case (3D flow over a square cylinder at Reynolds = 22.000 [0]) and compared with experimental data and other simulations. Some details for the computational efficiency and implementation on parallel computer cluster at INCAS will also be presented [0].

A Novel Wideband Microstrip Fractal Antenna Based on the Fractal Binary Tree

This article explores designing wideband microstrip antennas utilizing fractal geometries. A novel microstrip monopole antenna using the geometry of the fractal binary tree is presented. To achieve a wide impedance bandwidth, the proposed microstrip fractal antenna is optimized in an automated design, making use of the genetic algorithm with the full-wave electromagnetic simulation and cluster parallel computation. A design result is presented. A prototype of the designed antenna has been fabricated and tested. Good agreement is obtained between numerical calculated and measured results, and results show the proposed fractal antenna obtains a quite promising S11 < −10-dB impedance bandwidth, which surpasses 42.8% (from 2.46 GHz to 3.80 GHz).