Project Summary:

Coupling of Models for Energy and Environment (CMEE)

The goals of this Alpha project are twofold. The first is to improve our ability to model the flow of contaminants through the ecosystem accurately by coupling ground water and surface water simulations. This software could be used for a variety of purposes, including groundwater remediation, crisis management during oil and chemical spills, hurricanes and storms, and oil reservoir modeling. The second goal of this project is to integrate a robust, Grid-based computational and data handling infrastructure. This infrastructure will enable on-demand simulation, exploration and comparison of multiple scenarios that are of vital importance to energy and environmental modeling. This infrastructure will apply more generally to a wider range of applications that involve coupled simulations and data exploration. 

The participants of this project are

Current Status of Alpha Project

We provide plans for three projects whose details have not been fully worked out. This will be done when budgets have been formalized. 

Project 1: Heterogeneous Computing Using Parssim Simulator Transformation of Parssim into tool for interactive exploration of implications arising from various scenarios These scenarios include environmental remediation involving single phase flow with biogeochemistry reactive transport and implications of assumptions about aquifer geology. We will run the computationally intensive flow portion of the calculation on a high end compute server and run the data intensive portion of the computation (reactive transport, post-processing and data management) on a less computationally capable platform that has excellent connectivity to storage devices. Clients at various locations would be used to display output. We will also consider this approach for treating multiple realizations. 

UT will partition Parssim into coupled flow and reactive transport components. ISI will implement Globus supported coupling and demonstrate that the flow component and reactive transport component can be run on separate platforms. Maryland will develop ADR interface with reactive transport component so data can be stored and retrieved on multiprocessor storage system. UT, Maryland, ISI will define data analysis and visualization scenarios. Analysis will include selection of particular spatial regions, specific reaction products and calculation of derived quantities associated with concentrations and concentration gradients of products. Visualization details are unclear but will presumably include is contouring and possibly volume rendering. Maryland and ISI will define which visualization computations supported by ISI clients should be done in ADR and will adapt ADR to make use of data formats used by ISI visualization clients. Maryland, ISI, UT will then implement feedback between visualization clients so that Parssim problem definition can be interactively modified.

Project 2: Coupling of Surface Water Codes

The end result would be a 3-D version of the suite of programs used to carry out surface water pollution remediation using a processing chain consisting of flow codes, ADCIRC or UTBEST, projection code UTPROJ3D, and reactive transport codes, UTTRANS and/or CE-QUAL-ICM. The flow, projection, and reactive transport codes should be able to run on different platforms. Output from a single flow realization of flow will be stored in ADR and the data will be reused to carry out different remediation scenarios.

Plan: We are building on the successful 2D version of this project. A 3-D version of ADCIRC already exists. UTPROJ3D will be ported using KeLP by UT Austin with help from UCSD. The 3D codes described above will be coupled using MetaChaos by Maryland. This will provide basic functionality. We will support the use of multiple (more than four) MPI processes on a Teraflops node and support metacomputing on the teraflops platform by porting MPICH-G (G stands for Globus) to the Teraflops platform (to be carried out by SDSC). SDSC and UCSD will cooperate to optimize performance of the flow codes and UTPROJ3D on Teraflops platform. This will involve machine dependent optimizations as well as latency tolerance related Kelp optimizations. MetaChaos will be incorporated into Kelp by UCSD and Maryland (MetaChaos already has a Kelp interface, the idea is to optimize interface performance and reinterpret and reimplement MetaChaos in the context of abstract Kelp).

Project 3: History Matching with IPARS

Use of ADR to explore history matching scenarios. This involves collaboration between UT and Maryland to explore outputs from 100's of reservoir simulation realizations, each for 10^4-10^7 cells and hundreds of time steps and up to 10^3 wells. Computations will produce production histories for various well placements, production strategies and geostatistical scenarios. Additionally, the values of scalar and vector quantities of interest like pressures and concentrations will be stored by ADR. The production histories will be compared with experimental data to determine the closest reservoir description corresponding to the match. The experimental data is likely to be distributed over a different time stepping scale than the one used by the simulator and ADR interpolation operations will be used for comparison. In geostatistical setting and well optimization scenarios, the ADR reduction operations will deliver bounds for well performance and pressure and concentration values. x Finally, the details of reservoir simulation results will be compared for the realizations that were found to be of interest or which require further refinement. This includes querying ADR for spatial or temporal slices of data which can be postprocessed for some visualization output. For example, one would request a 2D slice of pressure and saturation data between two wells, which would offer  useful insight into determining coning problems and resolving poor well productivity.

Presentation:

Questions? Email us!