January 2012 - Allocations-admins - mailman3.cels.anl.gov

[LCRC Accounts] Project Request: XRayFuelSpray
by accounts＠lcrc.anl.gov 31 Jan '12

31 Jan '12

Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Chris Powell Applicant's institution: ANL Applicant's division: ES Project Name: XRayFuelSpray Project title: Fuel Injection and Sprays Studied Using X-Ray Diagnostics Associated funding: DOE-EERE Other Systems: None Science: At the Advanced Photon Source Sector 7-BM at Argonne National Laboratory, we are presently engaged in an ongoing investigation of sprays in high pressure fuel injection systems using X-ray diagnostic techniques. An improved understanding spray formation physics is essential to the development of more efficient engine technology. As a part of this ongoing research, we are investigating the presence of cavitation phenomena in fuel injection systems and how these phenomena impact spray formation. Cavitation is a problem in fuel injection systems which operate under high pressure. At the Advanced Photon Source, we are undertaking a series of experiments to develop a novel methodology for measuring real-time cavitation inside fuel injection systems using advanced X-ray imaging techniques. A model nozzle and fuel injection system have been designed specifically for this purpose. In order to develop these novel experimental techniques, it is necessary to have an accurate and detailed prediction of the cavitation zones inside the model nozzle at precisely defined boundary conditions, so that the measurement may be compared against an accurate baseline. To achieve this, numerical simulations of the fluid flow are necessary. The prediction of cavitation in a turbulent flow at high pressure requires the use of a high-resolution three-dimensional mesh with high levels of local refinement. As such, the computation time for these highly resolved meshes is beyond the reasonable capability of a desktop computer. The Fusion cluster will allow us to make these computations. The availability of accurate and detailed prediction of cavitation zones in a model nozzle will permit the exposition of these new experimental techniques, which will enable the measurement of cavitation zones in fuel injection systems and lead to improved design for purpose to reduce or enhance local cavitation. Project description: A CFD model will be used to predict the cavitation zones in a custom-meshed model of a 0.5mm diameter nozzle. Two codes will be employed: 1. An incompressible large eddy simulation, to provide an accurate pressure and velocity distribution without cavitation effects. 2. A barotropic compressible large eddy simulation, to provide prediction of cavitation zones. Comparison of both solutions permits the observation of expected differences in velocity and pressure due to local cavitation effects, and also permits prediction of geometric zones where cavitation may be expected over a range of supply pressures and back pressures. Large Eddy Simulation (LES) is employed in preference to the typically used RANS codes, as the local Reynolds Number is relatively high (up to approx. Re = 30,000) and it is necessary to consider the effects of turbulent shear between the higher and lower velocity regions in the expansion and contraction zones on the local cavitation which also occurs in these high-shear regions. The use of LES adds some computation penalty over RANS but the trade-off is a more accurate prediction of the mean turbulent fluctuations in the high shear regions. The subgrid scaling properties of LES are advantageous for the prediction of bubble behaviour in nucleation zones which will be smaller than the mesh resolution. The mesh of the custom model nozzle consists of approximately 2.4 million cells, and is locally refined in the expansion/contraction region of the nozzle to capture the features of interest. A grid independence study has been carried out on a series of coarse mesh solutions with several stages of refinement. The mesh to be used in the final simulation is refined to be within 1% of the grid-independent centreline velocity and pressure solution, based on a Richardson extrapolation of the incompressible LES solution. Further adaptive local refinement in the cavitation zones is also undertaken in the multiphase model. The boundary conditions are supplied via prescribed supply pressure and back pressure. The solver iterates to find the velocity solution to the LES form of the Navier-Stokes Equations based on the desired pressure. The solution will be verified via mass flow measurements made during the experimental analysis at the Advanced Photon Source. The variables of the simulations will include different fluid properties (water and gasoline) as well as a range of supply pressures and back pressures. For this initial study, fluid temperature is maintained around 20 degrees C. The code to be used is the free, GNU Public License CFD package OpenFOAM. For single-phase modeling, the incompressible iterative solver "pisoFoam" will be used. We have adapted the source code for a self-adjusting time step to ensure stability of the solution. For cavitation modeling, the barotropic compressible "cavitatingFoam" solver will be used. The single-phase incompressible turbulence solver includes the following features: * Large eddy simulation for closure of turbulent fluctuation terms * Iterative corrector PISO algorithm to solve for velocity from pressure * Adaptive time step code for stable solution (local Courant number limiter) In addition to the above, the multiphase cavitation solver includes a barotropic multiphase model using tabulated, known physical properties. The present code is well-suited for the identification of cavitation zones, which is the primary aim of the project. If more detailed modeling of the transport and collapse of cavitation bubble clouds is necessary, the code may be adapted to include compressible acoustic effects, as the local sound speed inside the vapor bubbles is low, and the Mach number is high. The OpenFOAM solver permits a simple discretisation of the domain for parallel computation. This parallelisation scales well up to 16 cores. Scaling to larger numbers of cores results in diminishing returns due to I/O overhead at cell boundaries, and due to the requirement of implicitly solving the velocity field from the specified pressure gradient through the Navier-Stokes equations. As such, we plan to run longer wall times on fewer CPUs, to maximise the efficiency of the calculation. If further parallelisation efficiency is required, the source code may be adjusted to permit cross-node communication of velocity & pressure on a time step interval rather than every time step. This approach has been shown to be highly effective in comparable LES studies. The OpenFOAM code is well established and the stability of the solver has been verified for coarser mesh solutions on a desktop computer, and is ready for scale-up to cluster application. Project URL: Requested allocation: 150000 Q1: 0 Q2: 50000 Q3: 50000 Q4: 50000 Justification: The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Project Allocation Request
by accounts＠lcrc.anl.gov 24 Jan '12

24 Jan '12

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: foam-EngineProcesses Title: Assessment of OpenFOAM for different Engine-Process Applications Description: CFD problems in general are not linearly scalable because of the need for high inter-processor communication. This results in loss of computational efficiency. Performance monitoring tools will be used extensively to understand computation and communication patterns for various physical models (turbulence, transport properties), chemistry mechanisms (varying number of species and reactions) and grid-sizes. Based on these studies, different domain decomposition approaches will be assessed for optimum load-balancing among different processors. Strong and weak scalability for various problem-sizes will thus be assessed. Some key features of the simulations include: 1) Finite volume schemes 2) LES and RANS turbulence models used 3) Detailed chemical kinetic models for diesel and biodiesel fuels 4) Detailed spray models used for different spray processes 5) Full cycle engine simulations Current: undetermined amount Justification: 99000 Requested: 100000 A specific reason has been given: We quickly consumed our 1st half allocation with several 1024 core jobs. We would like an additional 100K for the 2nd quarter. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Project Allocation Request
by accounts＠lcrc.anl.gov 24 Jan '12

24 Jan '12

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: CAT-engine-modeling Title: Detailed Engine modeling using CONVERGE Description: The commercial CFD code, CONVERGE will be used to conduct 3-D diesel engine simulations. Currently, CONVERGE has been run on FUSION on 24 cores. Good scaling has been seen up to 24 cores. The current study aims to increase the computational load using greater resolution, bigger chemistry mechanism and possibly more complicated geometries. Scaling of CONVERGE on a larger number of cores will be studied. It is expected that 3 ANL researchers along with 3 CAT engineers will be using these core hours. Current: undetermined amount Justification: Requested: 250000 A specific reason has been given: A series of large core count (512 and 1024) cases were run to study the scaling issues of CONVERGE with large mechanisms. Several interesting observations were noted about the scalability factors and bottlenecks. A larger suite of runs on smaller number of cores will be conducted in the remaining 3 quarters. Some of these will short runs (1000-5000 core hours) to study scalability issues. Once the number of cores required for good scaling is determined, productions runs will be conducted. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System

1 0