Geologic Sequestration Software Suite
From Carbon Sequestration Initiative
The Geologic Sequestration Software Suite, GS³, is a computational platform that will enable national and international collaborations. It provides collaborators dynamic access to evolving field and laboratory datasets, tools to facilitate the modeling process, and the ability to rapidly incorporate and validate new science into subsurface simulators. GS³ is designed to reduce modeling process burdens on scientists and engineers by encouraging interdisciplinary teams, making computational tools more accessible, and automatically tracking data provenance.
An extensible, dynamic and integrated computing environment, GS³ stewards users through the modeling process, from data mining, to creating the site geologic model, to launching scientific simulators, to analyzing simulated data. The GS³ framework creates rigor in this modeling process while providing scientists and engineers the flexibility to use their own favorite tools and software. Through the use of GS³, direct comparisons of best modeling practices can be made to accelerate the permitting of new geologic sequestration sites.
GS³ is web-based so team members can collaborate regardless of their geographic location. In addition, GS³:
- streamlines geologic model development and provides access to a growing subsurface database
- stores all the site data in one location where it can be updated and protected if proprietary; it also stores a history of the geologic models and previous simulations.
- keeps track of data input from different team members working on the same site and tracks models through the simulations process.
- allows application users access to the most advanced science-based simulators for modeling a site
- provides simulator developers a platform to quickly validate changes and improvements in the scientific simulators, and regularly benchmark the different simulators
- supplies a development site for a new community subsurface model.
GS³ will connect two user communities: application engineers, who are modeling a specific site for geologic sequestration; and development engineers, who want to advance the scientific simulators. As the geologic sequestration technology matures through laboratory experiments and field demonstrations, GS³ allows the testing of new theoretical ideas against recognized standards and field observations. With widespread use, GS³ will answer important questions such as the impact of multiple injection sites in the same reservoir or regional basin at resolutions not currently available.
Geological Sequestration Software Suite Core Architecture and Simulation Framework
PIs: Ian Gorton, Gary Black, Karen Schuchardt, Signe White
Advanced multi-phase flow and transport simulators play a crucial role in understanding and evaluating the feasibility and long-term effects of sequestering CO2 in deep geologic reservoirs. However, the process of managing and interpreting raw data, building the geologic model, and finally transforming/scaling this information into simulation input files is currently a tremendous challenge even for experienced modelers. The number of data sets to prepare and the number of tools that can be applied under different conditions is extensive and difficult to manage. Additionally the process may involve the development and evaluation of several geologic models leading to tens to thousands of simulations being run for just a single site. Organizing, managing, and tracking the data used during all of these stages is critical to the integrity, verifiability, and repeatability of the analysis, yet is currently ad hoc, time-consuming, and error prone.
The Geological Sequestration Software Suite (GS³) framework incorporates off-the-shelf tools that many modelers are already familiar with; facilitates the creation/integration of custom tools, utilities, and scripts; and integrates seamlessly with data management capabilities including provenance capture and user annotation. In addition to the component integration framework, this project will provide the data management services, standardized formats, data translations, and automated computational job management. Ultimately, the GS³ framework will vastly reduce the burden on modelers for manually organizing and tracking the volumes of data and applying the variety of tools needed throughout the sequestration modeling lifecycle.
Data Assimilation Tools for CO2 Reservoir Model Development
PIs: Mark Rockhold, Chris Murray, E. Charlotte Sullivan, Gary Black
This project is developing focused computational tools that allow scientists and engineers to efficiently assemble, assimilate, and track large volumes of disparate data for building geologic models of CO2 sequestration. Open-source and PNNL-developed reservoir characterization and visualization software tools will be available through GS³ for data assimilation and geologic model building in support of carbon sequestration modeling. The GS³ is flexible so that people can use their own commercial tools for developing their reservoir model. Resulting geologic models will form the basis for numerical modeling of reservoir processes at multiple scales.
The project is building a software environment composed of programs, utilities, and scripts for:
- evaluating, preparing, and pre-processing subsurface characterization data
- facilitating import, export, and linkage to external geologic sequestration-specific modules
- assimilating large, diverse data for building and parameterizing conceptual-mathematical models, emphasizing data and processes that are specific to geologic sequestration
- tracking authorship, data transformations, and analysis decisions used in calibrating and integrating diverse hard and soft data to ensure efficient updating of auditable, fully traceable, and defensible models.
The outputs of the resulting geologic models and parameterization tools will be appropriate for simulation of subsurface multi-fluid flow and reactive transport using a variety of simulators.
Geologic Sequestration Software Suite: Numerical Model Development
PIs:Mark Williams, Signe White, Paul Thorne, Andrew Kuprat, Diana Bacon
This project addresses the need for numerical model development tools and innovative methods to support the generation of simulator input files as part of the Geologic Sequestration Software Suite (GS³). New algorithms and tools will improve translation of conceptual models to the numerical framework required by simulator(s) and provide the capability of generating multi-scale (spatial and temporal) numerical models to support the operational and long-term performance simulation needs for geologic sequestration technology.
An important characteristic of the subsurface modeling process is the distinction between our conceptual understanding of the natural system, and the numerical implementation of that understanding for simulation. To preserve this distinction, it is critical to be able to manage data associated with each aspect independently. This allows for: 1) building multi-scale numerical models from a common conceptual model; 2) building numerical models from multiple conceptual models; 3) building numerical models and input files for different simulators; 4) ease in re-generating numerical models in response to revisions of the conceptual model; and 5) revising the numerical model specification during the development process (e.g., grid modifications and resulting re-assignment of material property values and distributions).
Eventually, the developed suite of tools will support input file generation for multiple simulators (i.e. STOMP, TOUGH2, PFLOTRAN, FEHM, and others, such as a community model), parallel computing platforms, stochastic simulations, and inverse modeling and parameter estimation. The ability to rapidly develop and run multi-scale simulations based on one or more conceptual models using various simulators in a parallel computing environment will result in better comprehension of the efficacy of the geologic sequestration approach given the uncertainties of the natural environment.
Visualizing Uncertainty in Conceptual and Numerical Models for Geologic Sequestration
PIs: Luke Gosink, Mark White, Diana Bacon, Bruce Palmer, Landon Sego
Scientific visualization and visual analytics are two classes of fundamental strategies that can provide insight into the complex events and processes modeled by numerical simulations. Such strategies are integral to the process of verifying the feasibility and identifying the impact of long term green house gas storage in deep underground reservoirs.
To address the challenges associated with the increasing complexity and expanding scale of numerical simulations, researchers at PNNL are building a suite of visualization software tools that will: 1) support cross-model comparisons and analysis based on a collection of simulated data ; 2) visually identify salient trends in numerical model uncertainty and sensitivity; 3) provide distributed visualization support for large-scale data; and 4) help track and maintain provenance data for numerical models and simulations. Components from this software will integrate into the GS3 framework and visually create an information feedback loop to accelerate the process of developing, analyzing, and verifying the integrity of a given numerical model, or a suite of models.
Geological Sequestration Software Suite Framework
PIs: Ian Gorton, Gary Black, Chandrika Sivaramakrishnan, Signe White
Advanced multiphase flow and transport modeling codes coupling physical, mechanical, chemical, and biological processes are expected to play a crucial role in understanding/evaluating the feasibility and long-term effects of sequestering CO2 in large-scale deep geologic reservoirs. However, the process of managing and interpreting raw data, building the conceptual model of the subsurface domain from that data, and finally transforming/scaling the conceptual model into the numerical model such that it can be evaluated by these types of simulation codes is currently a tremendous challenge even for experienced modelers. The number of data sets to prepare and the number of tools that can be applied under different conditions is extensive and difficult for modelers to manage. Additionally the process may involve the development and evaluation of several conceptual and numerical models leading to tens to thousands of simulations being run for just a single site. Organizing, managing, and tracking the data used during all of these stages is critical to the integrity, verifiability, and repeatability of the analysis, yet is currently ad hoc, time-consuming, and error prone.
In this project we will design and create novel provenance capture and analysis, and tool integration frameworks for carbon sequestration modeling. These capabilities will advance the state of modeling practices by providing the ability to track the data and assumptions in a given model, and make available a broad toolset for modelers to contribute to and exploit. To deploy these capabilities for modelers, we will integrate them with the Geological Sequestration Software Suite (GS3), and demonstrate their capabilities in support of uncertainty quantification and visualization. Together, these will create a transformational capability for both carbon sequestration modelers working on site models, and the creators of the simulators that are at the core of these modeling activities.
Uncertainty Quantification and Risk Assessment Pipeline for Carbon Sequestration
PIs: Guang Lin, Paul Eslinger, Yilin Fang, Jason Hou, Jian Yin
Predictive modeling of multiscale and multiphysics subsurface systems for CO2 geological sequestration requires accurate data-driven characterization of the input uncertainties and understanding how they propagate across scales and alter the final solution. In this project, we will leverage existing state-of-the-art tools to provide uncertainty quantification capabilities for the suite of simulation tools. We will also develop a uncertainty quantification workflow management capability to launch large ensemble of calculations and quantify the uncertainty, evaluate the risk and obtain the confidence interval for decision making. The uncertainty quantification toolkit and workflow management software involves individual uncertainty quantification tools such as sampling methods, statistical analysis methods, numerical integration routines, response surface evaluation and workflow management capabilities. In cases where existing tools do not provide appropriate capabilities, we will develop new approaches for characterizing uncertainty. The result will be an end-to-end uncertainty quantification and risk assessment pipeline that allows full uncertainty quantification studies of ensemble sets with a variety of adaptive sampling methodologies populating high dimensional uncertainty spaces, the capability to launch and monitor a large ensemble of calculations, and the functionality to collect and analyze the output data. This uncertainty quantification toolkit and workflow management software will provide a reliable means of quantitatively predicting the uncertainty and assessing the risk of CO2 geological sequestration. The uncertainty quantification pipeline toolkit will be open-source software and available to the GS community.