PNNL

Abstract

The Designs for Risk Evaluation and Management (DREAM) tool was developed as part of the effort to quantify the risk of lack of containment for geologic storage of carbon dioxide (CO2) under the U.S. Department of Energy’s National Risk Assessment Partnership (NRAP). DREAM is an optimization tool created to identify optimal monitoring schemes that minimize the time to first detection of CO2 leakage from a subsurface storage formation. DREAM acts as a post-processor on user-provided output from subsurface leakage simulations. While DREAM was developed for CO2 leakage scenarios, it is applicable to any subsurface leakage simulation of the same output format. The DREAM tool is comprised of three main components: (1) a Java wizard used to configure and execute the simulations, (2) a visualization tool to view the domain space and optimization results, and (3) a plotting tool used to analyze the results. A secondary Java application is provided to aid users in converting common American Standard Code for Information Interchange (ASCII) output data to the standard DREAM hierarchical data format (HDF5). DREAM employs a simulated annealing approach that searches the solution space by iteratively mutating potential monitoring schemes built of various configurations of monitoring locations and leak detection parameters. This approach has proven to be orders of magnitude faster than an exhaustive search of the entire solution space. The user’s manual illustrates the program graphical user interface (GUI), describes the tool inputs, and includes an example application. The latest version of the software, DREAMv3.0, introduces the ability to optimize surface geophysical surveys taken over various areas of the field site at irregular intervals. This optimization can be run using surface geophysics only, wellbore instruments only, or using a combination of surface geophysical surveys and wellbore instruments. These two approaches generally have broadly overlapping sensitivities, as a surface geophysical survey detects over a wide area of the field but only detects signals that are detectable along the surface at the time the survey was conducted, whereas a wellbore sensor can generally detect signals in the subsurface continuously the entire time it is in place, but only at one point in space. Therefore a pairing of wellbore sensors limited in their spatial coverage with geophysical surveys limited in their temporal coverage can produce a much improved monitoring plan. This software assumes that a set of hypothetical simulated CO2 leaks are available, and that those leaks are representative of the underlying risk profile for a site. This allows the software to remain lightweight and functional on a typical laptop, and reduces the barrier to entry for users who lack the expertise of budget for high performance computing. This also improves compatibility across various application domains, allowing the tool to remain independent from any software dependencies related to forward modelling or model calibration. Leak scenarios can be provided in the HDF5 file format, as NRAP-Open-IAM output files, or as STOMP, NUFT, Tecplot or Tough2 simulation files.