February 2014 - Allocations-admins - mailman3.cels.anl.gov

Re: [allocations-admins] [LCRC Accounts] Project Request: NEXTGENOPT
by Bair, Raymond A. 27 Feb '14

27 Feb '14

This is Victor Zavala's postdoc in MCS. Since it is a 1M hour request, he should present at the allocations meeting for their decision about the 2nd half (or reduce his request). In the meantime, I suggest starting the project and scaling the request to the end of the quarter: Allocation 90K Priority 4 Physical science Applied Math Ray -----Original Message----- From: allocations-admins-bounces(a)lists.lcrc.anl.gov [mailto:[email protected]] On Behalf Of accounts(a)lcrc.anl.gov Sent: Thursday, February 27, 2014 1:10 PM To: allocations-admins(a)lcrc.anl.gov Subject: [allocations-admins] [LCRC Accounts] Project Request: NEXTGENOPT 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: Naiyuan Chiang Applicant's institution: ANL Applicant's division: MCS Project Name: NEXTGENOPT Project title: Next generation optimization Associated funding: Office of Science Other Systems: Fusion Science: Advances in algorithms for stochastic optimization (SO) and uncertainty quantification (UQ) are critical for the efficient design and real-time optimization of next-generation national infrastructures. Anticipating and mitigating uncertainty of weather, demands, and contingencies in a more integrated environment is necessary to mitigate market volatility and prevent cascading failures that can ultimately lead to catastrophic shortages of supply. The integrated monitoring and optimization of infrastructures systems will generate huge amounts of data that cannot be possibly processed without exploiting underlying probabilistic structures (e.g., sparsity in spatio-temporal correlation patterns) and without exploiting the properties of the particular decision-making objective at hand (e.g., stochastic, PDEs). It is thus at the heart of our current research the philosophy that a structure-oriented and integrated approach to UQ/SO is necessary. Project description: Sponsored by a Department of Energy (DOE) Early Career Award, we are currently developing new algorithms for UQ/SO that have a holistic view of the entire data-modeling-optimization process. A particular example of such algorithms is a new clustering-based interior-point strategy for stochastic programs in which scenarios are compressed based on their influence on the first-stage decision, not on data. This has shown to lead to drastic compression rates (and solution times) of above 80%, which cannot be achieved with data compression alone (see Figure 1). The compression is done adaptively, at the linear algebra level, by looking at the different contributions of the scenarios on the Schur complement. We are also currently developing new interior-point algorithms for nonconvex structured optimization capable of dealing with complex multi-scale physics (typically involving large networks of partial differential equations) and stochastic components. These problems arise, for instance, from natural gas inventory (line-pack) management (see Figure 2). This requires of a redesign of existing algorithms and modeling environments capable of conveying problem structure down to the linear algebra kernels. In addition, hybrid linear algebra kernels need to be designed to exploit different structural facets. Problems with billions of variables are being targeted. Project URL: http://www.mcs.anl.gov/~vzavala/earlycareer.html Requested allocation: 1000000 Q1: 250000 Q2: 250000 Q3: 250000 Q4: 250000 Justification: This is the scalability of our optimization problem with up to 50 MPI processes on FUSION: 1MPI, walltime: 2:29:58 = 8998 sec 2MPI, walltime: 1:24:48 = 5088 sec 5MPI, walltime: 0:36:08 = 2168 sec 10MPI, walltime: 0:18:17 = 1097 sec 25MPI, walltime: 0:07:40 = 460 sec 50MPI, walltime: 0:04:07 = 247 sec 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 _______________________________________________ allocations-admins mailing list allocations-admins(a)lists.lcrc.anl.gov https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins

1 0

[LCRC Accounts] Project Request: NEXTGENOPT
by accounts＠lcrc.anl.gov 27 Feb '14

27 Feb '14

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: Naiyuan Chiang Applicant's institution: ANL Applicant's division: MCS Project Name: NEXTGENOPT Project title: Next generation optimization Associated funding: Office of Science Other Systems: Fusion Science: Advances in algorithms for stochastic optimization (SO) and uncertainty quantification (UQ) are critical for the efficient design and real-time optimization of next-generation national infrastructures. Anticipating and mitigating uncertainty of weather, demands, and contingencies in a more integrated environment is necessary to mitigate market volatility and prevent cascading failures that can ultimately lead to catastrophic shortages of supply. The integrated monitoring and optimization of infrastructures systems will generate huge amounts of data that cannot be possibly processed without exploiting underlying probabilistic structures (e.g., sparsity in spatio-temporal correlation patterns) and without exploiting the properties of the particular decision-making objective at hand (e.g., stochastic, PDEs). It is thus at the heart of our current research the philosophy that a structure-oriented and integrated approach to UQ/SO is necessary. Project description: Sponsored by a Department of Energy (DOE) Early Career Award, we are currently developing new algorithms for UQ/SO that have a holistic view of the entire data-modeling-optimization process. A particular example of such algorithms is a new clustering-based interior-point strategy for stochastic programs in which scenarios are compressed based on their influence on the first-stage decision, not on data. This has shown to lead to drastic compression rates (and solution times) of above 80%, which cannot be achieved with data compression alone (see Figure 1). The compression is done adaptively, at the linear algebra level, by looking at the different contributions of the scenarios on the Schur complement. We are also currently developing new interior-point algorithms for nonconvex structured optimization capable of dealing with complex multi-scale physics (typically involving large networks of partial differential equations) and stochastic components. These problems arise, for instance, from natural gas inventory (line-pack) management (see Figure 2). This requires of a redesign of existing algorithms and modeling environments capable of conveying problem structure down to the linear algebra kernels. In addition, hybrid linear algebra kernels need to be designed to exploit different structural facets. Problems with billions of variables are being targeted. Project URL: http://www.mcs.anl.gov/~vzavala/earlycareer.html Requested allocation: 1000000 Q1: 250000 Q2: 250000 Q3: 250000 Q4: 250000 Justification: This is the scalability of our optimization problem with up to 50 MPI processes on FUSION: 1MPI, walltime: 2:29:58 = 8998 sec 2MPI, walltime: 1:24:48 = 5088 sec 5MPI, walltime: 0:36:08 = 2168 sec 10MPI, walltime: 0:18:17 = 1097 sec 25MPI, walltime: 0:07:40 = 460 sec 50MPI, walltime: 0:04:07 = 247 sec 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 23 Feb '14

23 Feb '14

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: spray-modeling Title: Development of a near nozzle spray modeling approach Description: The project is divided into 5 major tasks to be performed in 3 years in collaboration with Convergent Science Inc., and Cummins Engine Company: Task 1: Development and validation of a nozzle flow model to account of transient effects such as needle movement, needle off-axis motion, rail pressure fluctuations etc. Task 2: Further advancement of KH-ACT (Kelvin Helmholtz – Aerodynamic Cavitation Turbulence) primary breakup model to calculate more accurately relevant length and time-scales for the cavitation, turbulence, and aerodynamically induced breakup processes. Task 3: In the past nozzle flow and spray simulations have been statically coupled. A dynamic coupling approach will be developed so that the nozzle flow simulations can influence the spray development and vice versa. This is quite complicated since the length and time scales associated with nozzle flow processes are significantly different from spray processes. Task 4: An one equation non-viscosity dynamic structure model large eddy simulation (LES) based turbulence modeling approach will be implemented in CONVERGE. Grid-convergence on critical spray parameters will be evaluated with the implemented LES models. Task 5: Implementation of an Eulerian-Lagrangian spray model: While dynamic coupling of the nozzle flow and KH-ACT models will enhance the accuracy of spray simulations, a transition between the Eulerian and Lagrangian phases must take place at the nozzle exit with this approach. Moving the transition point downstream allows for a more natural bridge between the coupled simulations as the liquid fuel is modeled in an Eulerian framework on both sides of the nozzle exit. A specific challenge of this task is developing an algorithm to convert the Eulerian liquid to Lagrangian parcels outside of the dense spray region. 1) Finite volume schemes 2) LES (dynamic structure) and RANS turbulence models used 3) Detailed chemical kinetic models for diesel fuels 4) CONVERGE software is being used - ~50 licenses available 5) Scales well up-to 256 processors 6) Detailed spray models used for different spray processes 7) Coupled nozzle flow and spray simulations Current: undetermined amount Justification: Requested: 150000 A specific reason has been given: in the past couple of weeks, my team has been running massive optimization simulations on Fusion and Blues clusters. We have exceeded our requested core hour limits on all the projects. If there is still time available on Fusion, I would request 100K core hours on this project. We have a paper deadline on March 28th which has instigated these massive simulations. 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 23 Feb '14

23 Feb '14

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: cavitation-modeling Title: Development of cavitation models for compression ignition engines Description: This research activity will mainly focus on comparing the cavitation models existing in the state-of-the-art engine modeling codes CONVERGE, OpenFOAM, and AVL-FIRE. Most cavitation models assume that vapor formation is initiated by the presence of cavitation nuclei which grow to become bubbles and then to form the complex structures also observed experimentally. The transport of vapor can be modeled either as discrete vapor bubbles or as a continuum. The Lagrangian tracked approach can handle natively different velocities for bubbles and liquid by solving the Newton law expressing the balance of forces acting on them. Among Eulerian methods, the multi-fluid model (implemented in AVL-FIRE) is conceptually superior to the single-fluid model (implemented in CONVERGE), since the latter is a homogeneous method therefore only one velocity is shared by all phases; while in a multi-fluid model each phase has its own velocity since momentum is solved separately for each fluid. In this case proper source terms accounting for the exchange of momentum between the phases have to be accounted for. In addition to that, a number of other physical processes have to be simulated. The most important one is the mechanism leading to bubble growth/collapse, i.e. the mass exchange model in cavitating conditions. The so-called Rayleigh-Plesset equation forms the basis for this modeling. The complete non-linear form, even if it is numerically stiff to be integrated, can be implemented in a multi-fluid model. Cavitation threshold can be expressed in terms of saturation pressure if the flow is quasi-steady, or adding viscous shear stress effects for high viscous or high velocity flows. By introducing a plenum for the outlet chamber and using a three-fluid model consisting of liquid, vapor and air, the nozzle outlet region can be resolved. This avoids the setting of unphysical boundary conditions at the nozzle exit and it allows reproducing the flow field accurately even in the proximity of the outlet. In case of super-cavitation the vapor plume can extend slightly beyond the outlet ; in case of hydraulic flip air can re-enter the hole region. An advantage of this approach is that the nozzle flow analysis is coupled to the in-cylinder flow, which results in the possibility of consideration of in-cylinder pressure variations, recirculation in cavitating flows (hydraulic flip), etc. Specific deliverables: - Comparison between single-fluid model (CONVERGE) and multi-fluid (FIRE) models and validation against experimental x-ray radiography data from Argonne. - Perform transient simulations of ECN injectors and perform validation against ECN data - Compare CONVERGE and FIRE results against transient ECN data - Assessment of transient effects: Series of steady simulations at fixed lift vs. transient simulations. - True needle movement (by x-ray visualization, Argonne) vs. idealized needle movement with n-dodecane as diesel surrogate - Coupling with outlet chamber in steady and time varying pressures (combustion chamber conditions) Current: undetermined amount Justification: Requested: 150000 A specific reason has been given: in the past couple of weeks, my team has been running massive optimization simulations on Fusion and Blues clusters. We have exceeded our requested core hour limits on all the projects. If there is still time available on Fusion, I would request 100K core hours on this project. We have a paper deadline on March 28th which has instigated these massive simulations. 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 23 Feb '14

23 Feb '14

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: LES-sprays Title: LES modeling for diesel spray and engine simulations Description: The project is divided into four major tasks to be performed in 3 years in collaboration with Convergent Science Inc., and Cummins Engine Company: Task 1: An one equation non-viscosity dynamic structure model large eddy simulation (LES) based turbulence modeling approach will be implemented in CONVERGE. Grid-convergence on critical spray parameters will be evaluated with the implemented LES models for non-evaporating and evaporating diesel sprays. Different flavors of LES models such as dynamic structure and Smagorinsky will be compared against RANS approach and without any sub-grid scale (SGS) model. Task 2: This task will focus on extending the LES spray modeling approach for performing combustion simulations. Grid-convergence of LES combustion simulations will be assessed and recommendations will be provided minimum grid-sizes keeping in mind run-time vs. accuracy for LES modelers. Task 3: The above tasks will benchmark LES approach for full-cycle engine simulations. The recommended minimum cell sizes will be implemented for performing engine simulations. Cycle-to-cycle variations will be captured using the LES approach. Validation of the engine simulations will be performed against experimental data available in literature. Task 4: Since LES requires small resolutions the wall-clock times for the above tasks will be large. In general, such high wall-clock times are not affordable in the industry. The last task will be focus on further improving the SGS models to obtain flow structures even at coarser resolutions. The specific methods used for the multi-phase, reacting flow simulations with moving boundaries will include: 1) Finite volume schemes 2) LES (dynamic structure) and RANS turbulence models used 3) Detailed chemical kinetic models for diesel fuels 4) CONVERGE software is being used - ~128 licenses available 5) Scales well up-to 128 processors 6) Detailed spray models used for different spray processes Current: undetermined amount Justification: Requested: 150000 A specific reason has been given: in the past couple of weeks, my team has been running massive optimization simulations on Fusion and Blues clusters. We have exceeded our requested core hour limits on all the projects. If there is still time available on Fusion, I would request 100K core hours on this project. We have a paper deadline on March 28th which has instigated these massive simulations. 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 23 Feb '14

23 Feb '14

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: biodiesel-combustion Title: Computation Fluid Dynamics Modeling of Petrodiesel and Biodiesel Description: The project is dived in two parts: In the first part, engine processes will be modeled using petrodiesel. The spray data obtained from Advanced Photo Source at Argonne and single-cylinder engine performance and emission data from CAT cell will be used to further develop and validate spray and combustion models in CONVERGE and perform parametric studies. The major motivation is to develop more predictive and robust sub-models for capturing diesel engine processes. In the second part, fundamental studies will be performed to establish flame stabilization and lift-off characteristics of biodiesel using CONVERGE. A new surrogate (consisting of methyl decanoate, methyl 9 decenoate, n-heptane) fuel will be used for modeling the combustion characteristics of biodiesel. Comparison will be performed between inner nozzle flow, non-evaporating and evaporating spray, combustion characteristics, engine performance and emission characteristics of biodiesel and petrodiesel. Biodiesel fuel obtained from different feedstocks such as Soy, Cuphea, Rape-seed will be modeled. Blends of biodiesel with diesel fuel will also be simulated. FORD motor company has requested this surrogate mechanism for performing some internal studies. Based on the feedback from FORD, the mechanism will be further improved for the benefit of the industry. 1) Finite volume schemes 2) RANS turbulence models used 3) Detailed chemical kinetic models for diesel and biodiesel fuels 4) CONVERGE software is being used - ~50 licenses available 5) Scales well up-to 256 processors 6) Detailed spray models used for different spray processes 7) Full cycle engine simulations Current: undetermined amount Justification: Requested: 150000 A specific reason has been given: in the past couple of weeks, my team has been running massive optimization simulations on Fusion and Blues clusters. We have exceeded our requested core hour limits on all the projects. If there is still time available on Fusion, I would request 100K core hours on this project. We have a paper deadline on March 28th which has instigated these massive simulations. 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 23 Feb '14

23 Feb '14

Hello, A change in allocation has been requested: Requester: ssom (Sibendu Som) Project: cfd_enginemodeling Title: Computation Fluid Dynamics Modeling of Diesel Engine Processes Description: The project is divided into three major phases: 1) Cavitation modeling: A two-phase cavitation modeling approach using the homogeneous relaxation method has been developed and implemented in CONVERGE. Extensive validation of the new model will be performed against experimental data available in literature and new data from Advanced photon source at Argonne. 2) Spray Modeling: Traditional Eulerian-Lagrangian spray modeling approaches are hampered by the fact that the results are not grid-independent. Grid-convergent spray modeling approaches will be developed and validated against experimental data from the Engine Combustion Network. The simulations will be performed in a constant volume combustion vessel since it is a more effective crucible for model validation 3) Turbulence Modeling: Traditionally engine simulations have been performed using the Reynolds Average Navier Stokes Equations (RANS) approach. RANS employs filtering in time to derive the governing equations for the mean state. Turbulent interactions over the full range of dynamic scales are averaged to make the calculations affordable for engineering analysis. Only the largest energy containing features in a flow are resolved and no information exists to describe broadband small-scale interactions. Thus, computational requirements for RANS are relatively low and the most affordable. However, case-by-case calibration of models is required, and stochastic processes such as cycle-to-cycle variations cannot be captured in RANS. Large Eddy Simulation (LES) directly resolves the large scale unsteady motions that account for the bulk spatial transportation, while smaller scales of the flow are removed through a filtering operation and treated by using subgrid scale (SGS) models f or reduced computational cost. By resolving the large-scale turbulence, LES can significantly improve the accuracy of flow predictions compared to RANS. A dynamic structure based LES model will be developed and the results will be compared against RANS. The CONVERGE software has been observed to scale well up to 256 processors and full-cycle engine simulations. Current: undetermined amount Justification: Requested: 150000 A specific reason has been given: in the past couple of weeks, my team has been running massive optimization simulations on Fusion and Blues clusters. We have exceeded our requested core hour limits on all the projects. If there is still time available on Fusion, I would request 100K core hours on this project. We have a paper deadline on March 28th which has instigated these massive simulations. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System

1 0

Re: [allocations-admins] [LCRC Accounts] Project Allocation Request
by Bair, Raymond A. 20 Feb '14

20 Feb '14

John Low checked with him, and he has enough time to last until 4/1. Ray ---------- Ray Bair Argonne National Laboratory, and The University of Chicago rbair (at) anl.gov<http://anl.gov> Sent from my iPad On Feb 20, 2014, at 9:02 PM, "John Valdes" <valdes(a)anl.gov<mailto:[email protected]>> wrote: Any thoughts on this request? They originally requested 400K total for the year, 100K per quarter, and this request is for another 400K. They didn't say if that's for the current quarter or for the rest of the FY. Given their usage rate, I say we give them 100K unless someone knows more about this request. John On Tue, Feb 18, 2014 at 11:26:01AM -0600, accounts(a)lcrc.anl.gov<mailto:[email protected]> wrote: A change in allocation has been requested: Requester: heinonen (Olle Heinonen) Project: Meso Title: Mesoscale computational materials science [...] Requested: 400000 A specific reason has been given: I ran out of the current allocation. I have been running an OpenMP code that runs very efficiently on 16 cores - I need more runs of that code. Also, we are ramping up a project using a developer version of SIESTA/SMEAGOL that will need many more nodes/cores for pre-production runs before we move to BG/Q _______________________________________________ allocations-admins mailing list allocations-admins(a)lists.lcrc.anl.gov<mailto:[email protected]> https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins

1 0

Re: [allocations-admins] [LCRC Accounts] Project Allocation Request
by John Valdes 20 Feb '14

20 Feb '14

Any thoughts on this request? They originally requested 400K total for the year, 100K per quarter, and this request is for another 400K. They didn't say if that's for the current quarter or for the rest of the FY. Given their usage rate, I say we give them 100K unless someone knows more about this request. John On Tue, Feb 18, 2014 at 11:26:01AM -0600, accounts(a)lcrc.anl.gov wrote: > > A change in allocation has been requested: > > Requester: heinonen (Olle Heinonen) > Project: Meso > Title: Mesoscale computational materials science [...] > Requested: 400000 > > A specific reason has been given: > I ran out of the current allocation. I have been running an OpenMP code that runs very efficiently on 16 cores - I need more runs of that code. Also, we are ramping up a project using a developer version of SIESTA/SMEAGOL that will need many more nodes/cores for pre-production runs before we move to BG/Q >

1 0

Re: [allocations-admins] [LCRC Accounts] Project Allocation Request
by John Valdes 20 Feb '14

20 Feb '14

Granted. John On Mon, Feb 17, 2014 at 10:06:55PM +0000, Bair, Raymond A. wrote: > This request seems reasonable > > On 2/17/14 10:57 AM, "accounts(a)lcrc.anl.gov" <accounts(a)lcrc.anl.gov> wrote: > > >A change in allocation has been requested: > > > > Requester: sarich (Jason Sarich) > > Project: UNEDF_Optimization [...] > > Requested: 120000

1 0