Predicting Engineered Natural Systems (PENs):
Applying GoldSim to Geological C02 Sequestration
Phil Stauffer, Hari Viswanathan, Rajesh Pawar, and George Guthrie
Los Alamos National Laboratory
Mounting evidence of global problems associated with excess atmospheric CO2 requires a solution that can be implemented quickly, safely, and economically. Given these constraints, researchers worldwide have concluded that geologic sequestration is the best available option. With so much CO2 to deal with, how will we find enough underground sites that are economical and safe for hundreds of years?
At Los Alamos, we believe the answer to this question lies in the development of a robust performance assessment framework for evaluating the suitability of specific sites to ensure that they will perform up to required goals as well as to determine the risks associated with long-term storage of CO2. The framework needs to take into account the fundamental physical and chemical interactions resulting due to presence of CO2 at multiple length and time scales (Pawar et al., 2006). To inform the decision making process, the performance assessment framework must be based on a rigorous, science based system-level model that tracks CO2 from capture through injection to storage while predicting the likelihood of leakage to the surface. Additionally, the system-level model must calculate risk to humans and the environment as well as economical impacts in order to make informed decisions on where to locate CO2 sequestration sites
LANL has developed the first such system-level model for geologic sequestration that tracks the fate of CO2 and is designed to include all associated risks (Stauffer et al., 2005; Viswanathan, 2005; Stauffer et al., 2006). Our system-level CO2 sequestration model is called CO2-PENS, where PENS is an acronym that stands for Predicting Engineered Natural Systems. Los Alamos has chosen GoldSim as the upper level of control in this system-level model. CO2-PENS uses existing knowledge from the oil industry, ties to practical risk and economic theory, and innovative fundamental science. Fundamental science is required to help answer questions that are unique to the long-term CO2 sequestration problem. For example, although we know a lot about oil reservoirs and how they behave over 20-50 years, the CO2 sequestration problem will require us to understand behavior in a wide variety of geological settings over many centuries. Figure 1 shows the basic conceptual framework on which the CO2-PENS model is based with the primary pathways used to track CO2.
Figure 1
CO2-PENS integrates a host of CO2 related processes including CO2 transport pathways, economics, and risk analysis. Each of these processes is built up using both existing knowledge and fundamental science that is unique to the CO2 sequestration problem. Figure 2 shows the highest level of the CO2-PENS model, as seen from within GoldSim.
Figure 2
In this figure, the flow of CO2 moves from the left to right, starting in the CO2 capture container and ending at the atmospheric container on the right. Each process can include any level of complexity that a user may need. For example, the CO2 Reservoir container could be composed of either a simple analytical solution for CO2 flow or it could include links to an entire 3-D reservoir simulation like the oil industry uses to predict how much oil they can get out of the ground.
Another key aspect of CO2-PENS is that the adjustable parameters are not single values but are defined as probability distributions. Because natural systems are highly variable and complex, we are not able to measure parameters at every point in space and time. With GoldSim as the top level controller for CO2-PENS, we are able to easily explore uncertainty using probability distributions for both model inputs and model outputs.
Because of the inherent flexibility of the GoldSim modeling environment, we can quickly adapt CO2–PENS to changes in understanding by modifying or adding processes and interactions. Identification of additional processes and interactions can come through use of the system model itself, through expert elicitation, and through independent investigations undertaken to support the decision making process. The highest level of CO2-PENS (Figure 2) is used to manage global variables such as time, CO2 mass balance, total risk, well statistics, and costs. These high level variables are then passed into subprograms that can be modified or replaced by anyone working on the system-model. Our approach allows use of both intrinsic GoldSim functions, such as accumulators and pipe pathways, while at the same time allowing users to create more complex algorithms in separate programs and link to GoldSim using dynamic link libraries. By making CO2-PENS modular, we have coupled the powerful system-level bookkeeping and stochastic functions of GoldSim with state-of-the-art reservoir simulators, new analytical techniques, and GIS software.
Modularity and flexibility will allow CO2-PENS to be used throughout the DOE complex, universities, and possibly industry. We have used the GoldSim GUI capability to allow easier interaction with users (Figure 3 and 4).
Figure 3
The GoldSim Player combined with the GUI allows us to create a model with limits on variable manipulation. These example models can be given to potential collaborators or policy makers and will allow them to better understand the CO2 sequestration system model. Potential collaborators could use the GoldSim Player to see how their work would fit into the framework of our growing program while policy makers would be able to use the GoldSim Player examples to leverage more funding for the integrated modeling approach.
We envision that CO2-PENS will initially be used as a screening tool that can help to make quick decisions on the applicability of sequestration sites. As site selection proceeds, CO2-PENS can be modified to generate complete Performance Assessments for individual sites as detailed site specific information becomes available. Currently, CO2-PENS contains many of the modules required for a performance assessment of geological CO2 sequestration, while in the future we will add components of risk analysis that require collaboration with experts in the field of risk analysis. We believe that it is very important to have a consistent platform for calculations of performance and risk at different sites to provide meaningful comparisons, and by using the GoldSim modeling environment, CO2-PENS will provide the necessary transparency to ensure QA/QC requirements that will no doubt be mandated for geologic sequestration.
We next present an example of how process level models are include in CO2-PENS and some of the links that we have built between GoldSim and modules that have been built as dynamic link libraries.
We have two process level models that can be used to calculate injection of CO2 into the subsurface. The first module uses an analytical solution which is tuned to 3-D reservoir calculations to estimate the amount that each well can inject. The module also uses another analytical solution (Nordbotten et al., 2005) to calculate plume extent and also gives an estimate of the amount of the plume that has spilled over the edge of the reservoir. The module is intended for use during the initial site selection when estimates on the reservoir capacity and the likely number of injection wells are required. The second injection module, designed for assessing performance of specific sites, calls FEHM (Zyvoloski at al., 1997), a Los Alamos reservoir simulator that can perform complex calculations employing physics of 3-dimensional, multi-phase fluid flow. CO2-PENS can also be linked to TOUGH2 (Zhang et al., 2006) or other multiphase flow and transport simulators. Both injection modules allow the user to link GoldSim to reservoir databases that contain site specific permeability and porosity measurements. Output from the injector module is fed back into GoldSim and displayed on the reservoir dashboard (Figure 4).
Figure 4
In summary, we have described how GoldSim is used as a platform to implement our performance assessment framework in a comprehensive system-level simulator. The simulator, named CO2-PENS, uses processes level laboratory experiments, field experiments, modeling, economic data, and risk theory to support a system level model that can provide a robust tool in the decision making process for selection and implementation of CO2 sequestration sites. CO2-PENS, is already proving to be useful in showing complex interactions between the different components of the framework. The system model also provides a consistent platform to document decisions made during the site selection, implementation, and closure periods. Because of the modular design, GUI interface, and powerful underlying GoldSim modeling environment, we envision that CO2-PENS may become a tool used throughout the DOE complex, universities, and possibly industry to help analyze potential CO2 sequestration sites.
Nordbotten, JM; Celia, M.A.; Bachu, S (2005). Injection and storage of CO2 in deep saline aquifers: Analytical solution for CO plume evolution during injection, Transport in Porous Media; 58, p.339-360.
Pawar, R., J. C., S. Chipera, J. Fessenden, J. Kaszuba, G. Keating, P. Lichtner, S. Olsen, P.H. Stauffer, H. Viswanathan, H. Ziock, G. Guthrie, (2006) Development of a framework for long-term performance assessment of geologic CO2 sequestration sites, Proceedings of the 8th International Confernce on Greenhouse Gas Control Technologies, Trondheim, NO.
Stauffer PH, Viswanathan HS, Pawar RJ, Klasky ML, Guthrie GD. (2006) CO2-PENS a CO2 sequestration system model supporting risk-based decisions. Proceedings of the 16th International Conference on Computational Methods in Water Resources; Copenhagen, Denmark, June 19-22, 2006.
Stauffer, P.H. Viswanathan, H, et al., (2005) CO2-PENS: A CO2 Sequestration Systems Model Supporting Risk-Based Decisions, AGU Fall Meeting, 2005
Viswanathan, H, Stauffer, P.H. et al., (2005) The Development of a Performance Assessment Framework for Geologic CO2 Sequestration, AGU Fall Meeting, 2005
Zhang, Y., C. M. Oldenburg, S. Finsterle, and G. S. Bodvarsson (2006), System-Level Modeling for Geological Storage of CO2: TOUGH2 Users Conference, Earth Sciences Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, California
|
