The Fluids & Energy Group investigates challenges of storing CO2 below ground. The researchers use computer simulations, field studies, and laboratory experiments to study processes at molecular to basin scales. The PI’s of the Group are Michael Celia and Jean Prévost of Civil and Environmental Engineering, Pablo Debenedetti and Thanos Panagiotopoulos of Chemical and Biological Engineering, Howard Stone (Mechanical and Aerospace Engineering), and Jeroen Tromp (Geosciences and Applied and Computational Mathematics).
New initiative on natural gas issues
Initial measurements of methane leakage from abandoned wells in Pennsylvania suggest that this may be a significant source of unreported emissions, accounting for 5-7 percent of anthropogenic methane emissions from the state.
A study of the fate of fluids used for fracking and shale-gas production indicates that the simplest model for absorption of the fracking fluid by the rock matrix cannot account for the 75% of fluid that does not return to the well.
Simple models are proving useful for the study of injection of CO2 into depleted shale-gas formations for the purpose of enhanced methane recovery and CO2 storage.
Multi-scale models of CO2 injection, migration and leakage
CMI’s numerical and semi-analytical models are providing analyses of field sites - including the Basal Aquifer of Canada and the northern United States, the Utsira formation associated with the Sleipner injection, and the Gorgon Project in Australia – that are generally consistent with those from more complex simulators but require far fewer computing resources.
The Dynaflow model can now represent the effects of boiling CO2-brine mixtures that might occur through leakage from a storage reservoir, raising the risk of runaway discharge or blowouts at CO2 injection sites.
Bench-scale investigations of flows
Theory and laboratory experiments are providing insights into the behavior of CO2 in viscous gravity currents belowground.
Molecular models of ice nucleation can illuminate rates of ice formation relevant to the genesis of gas hydrates.
In 2013, Michael Celia and colleagues initiated a new line of inquiry focused on leakage of methane along old wells, and on development of models to predict fluid movement and potential leakage associated with hydraulic fracturing and shale-gas production. This work follows naturally from their earlier and ongoing research focused on CO2 injection and the possible leakage scenarios associated with CCS (see “Models for CO2 injection, migration and leakage” below).
A new field measurement program, led by senior graduate student Mary Kang, involves direct measurements of methane fluxes along old abandoned wells in northwestern Pennsylvania. Mary and others from Princeton made a total of four rounds of measurements in the last year, involving 19 abandoned wells. All 19 wells showed positive methane emissions to the atmosphere, independent of their plugging status (Figure 1). The mean methane flux for the 14 wells is 0.27 kg/day/well, and the chemical composition of the gas indicates a mixture of thermogenic and biogenic sources. These fluxes are much higher than soil fluxes measured adjacent to the wells and used as controls. If the mean flux for these 19 wells is used as an average for all abandoned wells in Pennsylvania, this would represent methane leakage that is between 5 and 7 percent of the total estimated anthropogenic methane emissions in the state of Pennsylvania. This number is also similar to the estimated leakage rate from operating fracking wells, although the temporal extent of leakage from abandoned wells is likely to be much longer than those associated with fracking and production operations. The team hopes to measure additional wells, from different oil and gas fields and with different attributes, to get a better measure of the overall leakage rates.
As a parallel activity, the researchers have begun to develop models to study fluid movements associated with hydraulic fracturing (“fracking”) and shale-gas production. The focus is the fate of the fracking fluids, where about 75% of the injected aqueous-based fracking fluid does not return to the well in the flow-back period. Recent literature has hypothesized this as an indication of leakage upward toward potable aquifers. The Celia group is looking at two-phase flow effects and the role of spontaneous imbibition, to determine whether this absorption into the rock matrix can reasonably account for the fluid losses. This work is ongoing, although early results indicate that simple imbibition away from the main fracture planes cannot account for this loss of mass.
In a related effort, Celia and colleagues have recently begun to consider whether CO2 might be injected into depleted shale-gas formations for the purpose of enhanced methane recovery with the co-benefit of CO2 storage driven by preferential sorption within the rock. The researchers are making initial calculations using simplified models at the large scale while also building new porescale models to represent the shale system at the small scale. The large-scale calculations are part of a collaboration with Eric Larson from CMI’s Low-Carbon Energy Group and Lynn Loo from Chemical Engineering, while the pore-scale modeling is being developed in collaboration with Vahid Joekar-Niasar who is at the Shell Research Laboratory in Rijswijk, The Netherlands.
Michael Celia and colleagues continue to develop a suite of models to simulate CO2 injection, migration, and potential leakage, with applications to a number of active or potential injection sites. These models predict pressure buildup in the formation, movement of both CO2 and brine, capillary trapping of CO2, and dissolution of CO2 into the brine phase (that is, solubility trapping).
The group’s work in the past year has focused on applications of models to real field sites, on explicit identification and evaluation of differences among a wide range of models with different complexity, and on new numerical algorithms and analytical solutions to enhance model capabilities. Applications include large-scale analysis of the Basal Aquifer of Canada and the northern United States, covering more than a million square kilometers (Figure 2); the top layer of the Utsira formation and CO2 migration associated with the Sleipner injection; and ongoing simulations focused on the Gorgon Project in Australia.
The researchers have written a manuscript (Bandilla et al., 2014) that provides details on the range of models that have been, and can be, used for CO2 injection problems, from fully coupled multicomponent multi-phase models to simplified dynamic models to very simple percolation-type models.
This year Celia and colleagues also developed new numerical algorithms for their multi-scale models. One advance maintains much of the computational advantages of vertical equilibrium models while allowing for a relaxation of the equilibrium assumption (see Guo et al. (2014)). They have also developed a new analytical solution for leakage into a leaky fault; this analytical solution can be used to provide local-scale information within coarse grid blocks of larger-scale simulations. Much of their ongoing modeling work is in collaboration with partners at Lawrence Berkeley National Laboratory and the University of Bergen.
The Celia group continues to study additional aspects of CO2 injection, migration, and leakage, including pore-scale models for geochemically reactive systems, and multi-phase models for trapping and hysteresis.
The geochemical modeling work includes explicit modeling of both precipitation and dissolution with the associated changes in porosity and permeability tracked as the reactions proceed. The researchers also identify conditions under which unique relationships between porosity and permeability should be expected.
Pore-scale models also provide new insights into how nonwetting fluids are trapped in porous media, including underlying mechanisms and the extent to which continuum-scale trapping functions are hysteretic (Figure 3). The hysteretic nature of all multi-phase constitutive functions can be carried from the small scale to the large scale; Doster et al. (2013) shows how this can be done for Vertical Equilibrium models and the conditions under which hysteresis may be important at the large scale.
The Prévost group develops simulation tools that capture the effects of coupling between fluid flow, thermal, and geomechanical effects. Since the inception of CMI, the group’s Dynaflow model has been adapted to simulate aquifer geochemistry and multiphase flow as well as to predict physical stresses induced by CO2 injection. The model has recently been extended with new features, making it capable of simulating the flow and heat transfer for fluids with changing number of phases. The current implementation can simulate leaks of CO2-brine mixture which can form one, two or three fluid phases.
Previous CMI research and particularly results of the Prévost group revealed possible risks of CO2 leakage from long-term storage in deep aquifers. At the depth of an aquifer, the temperature and pressure are high, so that CO2 is in a supercritical state and forms two almost immiscible phases with brine. However, if a leak causes upward migration of CO2 and brine to shallow depths, where the temperature and pressure are lower, CO2 becomes subcritical and starts boiling, forming an additional gas phase. Appearance of this third phase may change the characteristics of the flow dramatically.
The researchers have enhanced Dynaflow with a thermodynamic module based on an equation of state that is capable of predicting the number of phases and their composition based on temperature or energy of an element for every time step of the finite element simulation of the fluid flow. Modeling the flow with changing numbers of phases also required a new numerical procedure for solving the energy transport equation.
Basic test cases showed that, where the temperature substantially drops, there is the possibility of formation of a wide (about 100 meters) region of coexistence of three phases. Current and future work is focused on testing of the implemented algorithms for more complicated conditions representing realistic CO2 injection sites.
Predicting the fate of CO2 injected underground for carbon storage or enhanced oil recovery requires understanding physical processes associated with the underground rearrangement of a buoyant material. In addition, geological transport processes require understanding the flows of granular materials. These themes raise new questions regarding multiphase flows in porous media, which the Stone group is investigating with a combination of bench-scale laboratory experiments and theory.
The gravitationally driven spreading of viscous fluids, also known as viscous gravity currents, is a process common to a large number of industrial and geological phenomena, including carbon sequestration. It is always helpful to supplement common numerical simulations with analytical results, which can highlight spreading rates as a function of relevant input variables, and laboratory experiments, which can test and characterize assumptions made in the modeling efforts. In the Stone group, new experimental, theoretical and numerical results have been reported for the effects of horizontal heterogeneities on the propagation of viscous gravity currents.
In the analytical work, two types of self-similar solutions – so-called first- and second-kind similarity – were explored to predict the time-dependence of invasion processes. These theoretical predictions were compared to experimental results and numerical solutions of the governing partial differential equations and all three results were found to be in good agreement.
Modeling transport of particulate materials is important in geophysical flows such as snow avalanches, mud and landslides. For example, in a polydisperse avalanche, segregation among different particle types drives the large particles to the front. The resulting distribution of debris upon the cessation of flow can dictate the ecological impact of the event, hence it is important to know how the various constituent materials are dispersed during the landslide. Similar questions arise in the industrial handling of granular materials.
It is well known that rapidly flowing dense granular materials can behave similarly to fluids and can be approximated as a continuum, hence the dispersion can be studied using methods familiar from the similar problem of a chemical dispersing in a pressure-driven pipe flow. The Stone group has formulated and solved a model problem of dispersion of dense granular materials in rapid shear flow down an incline. The findings can be applied to transport of all manners of granular materials, such as coal, sand, and powders, and also to natural processes such as landslides.
Pablo Debenedetti, Athanassios Panagiotopoulos, and Jeroen Tromp use molecular simulations to provide new insights into processes relevant to CO2 storage and climate.
The Debenedetti Group is using advanced molecular simulation techniques to study the rates and mechanisms of formation of ordered phases of water, including ice and gas hydrates. Studying ice nucleation at atmospherically-relevant conditions is of interest to atmospheric and environmental sciences and can be pivotal in developing more accurate predictive tools for weather forecasting, as well as for improving fundamental understanding of climate change. Studying nucleation and stability of hydrates is relevant to CO2 capture and storage, as well as to climate change-induced release of CH4.
In 2013, Debenedetti and colleagues developed an advanced sampling technique for computing nucleation rates in slowly relaxing systems, and began using it to compute the rate of ice nucleation in molecular models of water. Due to the extremely slow structural relaxation in such systems, existing techniques are unsuccessful in computing the rate of ice nucleation. With a new algorithm, based on molecular dynamics and a coarse-grained version of the forward-flux sampling (FFS) technique, they have obtained a statistically representative picture of the nucleation process. Calculations are still in progress, but the results to date give the team confidence in the eventual success of this approach. Figure 6 depicts ice crystallites of different sizes obtained from the algorithm.
The researchers are also continuing work on molecular modeling of CO2-brine systems of relevance to CO2 storage. In the coming year they expect to publish the ice nucleation calculations, and to apply the method to heterogeneous ice nucleation and to nucleation of hydrates.
In 2010, a molecular modeling program was initiated within the Fluids & Energy Group. Pablo Debenedetti, Athanassios Panagiotopoulos, and Jeroen Tromp use molecular simulations to provide new insights into the behavior of CO2 and methane in subsurface environments.
Al-Housseiny, T.T., J. Hernandez and H.A. Stone, 2014. Critical conditions for instability in the invasion of identical parallel channels. Submitted.
Al-Housseiny, T.T., I.C. Christov and H.A. Stone. 2013. Fluid displacement under elastic membranes: dynamics and interfacial instabilities. Phys. Rev. Lett., 111: 034502.
Al-Housseiny, T.T., and H.A. Stone, 2013. Controlling viscous fingering in tapered Hele-Shaw cells. Phys. Fluids ,25: 092102.
Bandilla, K.W., M.A. Celia, A. Cihan, J.T. Birkholzer, and E.C. Leister, 2014. Overview of approaches for modeling of geologic carbon sequestration in saline aquifers. Ground Water, to be submitted.
Christov, I.C., and H.A. Stone, 2014. Shear dispersion in dense granular flows. Granular Matter, accepted.
Doster, F., J.M. Nordbotten, and M.A. Celia, 2013. Impact of capillary hysteresis and trapping on vertically integrated models for CO2 storage, Advances in Water Resources, 62: 465-474.
Doster, F., J.M. Nordbotten, and M.A. Celia, 2014. Hysteretic upscaled constitutive relationships for vertically integrated porous media, accepted for publication, Computing and Visualization in Science.
Elliot, T.R., T.A. Buscheck, and M.A. Celia, 2013. Active CO2 reservoir management for sustainable geothermal energy extraction and reduced leakage, Greenhouse Gases: Science and Technology, 3: 50-65, DOI: 10.1002/ghg.
Gor G. Y., H.A., J.H. Prévost J. H., 2013. Fracture propagation driven by fluid outflow from a low-permeability aquifer, Transport in Porous Media, 100: 69-82. doi: 10.1007/s11242-013-0205-3.
Gor G. Y., J.H. Prévost J. H., 2013. Effect of CO2 injection temperature on caprock stability, Energy Procedia, 37: 3727-3732. Proceedings of GHGT-11 Conference. Ed. by Tim Dixon and Kenji Yamaji. doi: 10.1016/j.egypro.2013.06.267.
Guo, B., K.W. Bandilla, F. Doster, E. Keilegavlen, and M.A. Celia, 2014. A vertically-integrated model with vertical dynamics for CO2 storage. Water Resources Research, under review.
Huang, X., K.W. Bandilla, M.A. Celia, and S. Bachu, 2014. Basin-scale modeling of CO2 storage using models of varying complexity,” International Journal for Greenhouse Gas Control, 20: 73-86.
Joekar-Niasar, V., F. Doster, R.T. Armstrong, D. Wildenschild, and M.A. Celia, 2013. Trapping and hysteresis in two-phase flow in porous media: A pore-network study”, Water Resources Research, 49: 4244-4256, doi:10.1002/wrcr.20131.
Kang, M., J.M. Nordbotten, F. Doster, and M.A. Celia, 2014. Analytical solutions for two-phase subsurface flow to a leaky fault considering vertical flow effects and fault properties, under revision, Water Resources Research.
Kang, M. C. Kano, M. Reid, X. Zhang, D. Mauzerall, M. Celia, Y. Chen, and T.C. Onstott, 2014. Measurement of methane emissions from old abandoned oil and gas wells in northwest Pennsylvania. Science, to be submitted.
Liu, Y., T. Lafitte, A.Z. Panagiotopoulos and P.G. Debenedetti, 2013. Simulations of vapor-liquid phase equilibrium and interfacial tension in the CO2-H2O-NaCl System, AIChE J., 59: 3514. doi: 10.1002/ aic.14042.
Nogues, J.P., J.P. Fitts, M.A. Celia, and C.A. Peters, 2013. Permeability evolution due to dissolution and precipitation of carbonates using reactive transport modeling in pore networks, Water Resources Research, 49: 6006-6021. doi: 10.1002/wrcr.20486.
Person, M., M. Schlegel, Y. Zhang, J. Rupp, K. Ellett, B. Bowen, C. Gable, J. McIntosh, K. Bandilla, M. Celia, 2013. Effects of Pleistocene glaciations on the distribution of freshwater and brines across the Illinois Basin, USA. Geofluids, under revision.
Tsai, P.A., K. Riesing and H.A. Stone, 2013. Density-driven convection enhanced by an inclined boundary: Implications for geological CO2 storage. Physical Review, E 87: 011003(R). Highlighted by Physics OnLine: http://physics.aps.org/synopsisfor/ 10.1103/PhysRevE.87.011003.
Zhang, Y., M. Person, J. Rupp, K. Ellett, M.A. Celia, C.W. Gable, B. Bowen, J. Evans, K. Bandilla, P. Mozley, T. Dewers, and T. Elliot, 2013. Hydrogeologic controls on induced seismicity in crystalline basement rocks due to fluid injection into basal reservoirs, Ground Water, 51(4): 525-538.
Zheng, Z., B. Soh, H. Huppert and H.A. Stone, 2013. Fluid drainage from the edge of a porous reservoir. J. Fluid Mech., 718: 558-568.
Zheng, Z., I.C. Christov and H.A. Stone, 2014. Influence of heterogeneity on second-kind self-similar solutions for viscous gravity currents. Journal of Fluid Mechanics, accepted.