H12D-01 INVITED
Adequacy of Monitoring Methods and Strategies for Detecting Carbon Dioxide Leakage from Geological Storage Reservoirs
Significant progress has been made over the past decade resulting in the development of many approaches for monitoring carbon dioxide leakage from geological storage reservoirs. Some of these methods, such as 3-D seismic, cross-well seismic and vertical seismic profiling have been demonstrated to be successful. Others such as above-zone pressure monitoring show theoretical potential. Surface monitoring methods such as eddy- covariance flux towers and flux accumulation chambers have been used to monitor surface releases from natural seeps and controlled releases with good success. New open-path systems have also been successfully used to detect elevated concentrations of carbon dioxide in the air above controlled release experiments. Sophisticated algorithms for distinguishing leakage from natural variations in ecosystem carbon fluxes have also been proposed. As a result of these studies, it is becoming apparent that, if we know that leakage is occurring, existing methods are quite adequate for locating and quantifying them. More challenging is the design cost-effective approaches leakage detection—and providing confidence that leakage is below a prescribed detection threshold. The comparatively small footprint of the monitoring methods that are useful for the location and quantification of leakage may require an uneconomical number of monitoring stations if they are used for leakage detection. This paper examines this issue and proposes a number of options for designing effective monitoring networks for leak detection.
H12D-02
Monitoring Surface CO2 Fluxes Associated with Shallow Subsurface CO2 Release Experiments
A new facility designed by the ZERT Project to release CO2 into the shallow subsurface provides the opportunity to test field methods to detect and quantify potential CO2 leakage from geologic storage sites. CO2 release experiments were conducted in an agricultural field at Montana State University, in Bozeman, MT, where a ~100-m long horizontal well was installed at ~2.5 m depth, sub-water table, within a sandy gravel. The well was divided into zones separated by inflatable packers, from which 0.1 t CO2 d- 1 was released from 07/09/2007 to 07/19/2007 (Release 1), and 0.3 t CO2 d-1 was released from 08/05/2007 to 08/10/2007 (Release 2). We measured soil CO2 fluxes using the chamber method on grids repeatedly on a daily basis and net CO2 fluxes continuously using the eddy covariance technique. Based on chamber measurements near the well, CO2 breakthrough at the surface occurred on day two of Release 1. The spatial distribution and magnitude of leakage fluxes reached quasi-steady state by day six. Fluxes returned to near those measured at background locations two days following the end of Release 2. Spatial patterns in chamber CO2 fluxes were strongly related to well design. Estimates of total, background (soil respiration), and leakage CO2 discharges (t d-1) based on chamber measurements showed that during Release 1, background CO2 discharge was relatively high but declined at nearly the same rate as leakage discharge increased, leading to little change in total discharge. Conversely, during Release 2, background soil respiration remained low relative to leakage discharge. Because the comparatively large spatial footprint of the eddy covariance measurements averages over both background and well locations, CO2 leakage signals were difficult to detect by eddy covariance during Release 1, while during Release 2, signals were clearly detectable. Results emphasize the influence of background CO2 flux variations on the ability to detect leakage signals.
H12D-03
The influence of seal thickness and rate of pressure buildup on CO2 migration and sequestration
This research investigates the nature of hydrodynamic seal capacity in potential CO2 sequestration reservoirs and the conditions that lead to initial migration from an injection horizon resulting from pressure buildup and vertical brine displacement. We developed numerical codes adapted from the natural gas storage industry to investigate the influence that seal thickness and the rate of pressure buildup have on CO2 migration rates and net volume retained. In methane storage, computed vertical migration rates (and therefore the volume retained) depend on the shape of the injection profile (pressure increase versus time), suggesting that optimization of CO2 storage volumes (net capacity) should consider how the CO2 is injected (dynamic capacity) in addition to simply the petrophysical properties of the host reservoir and primary seal (threshold capillary entry pressure; static capacity). This paper investigates whether the methodology developed by the natural gas industry could be relevant to CO2 sequestration. Such analyses provide practical guidance regarding the injection profile advisable for a range of seal characteristics. Economic implications of various injection profiles that are compatible with the host rocks will need to be addressed to achieve optimal storage.
H12D-04
Evaluation Of CO2 Injection Into Brine Formations Below Oil Reservoir
Two models evaluating CO2 trapping mechanisms in the SACROC northern platform were developed using an upscaled geocellular model. The first model was designed for simulating CO2 trapping mechanisms in a reservoir saturated only with brine. The other model was designed for simulating CO2 trapping mechanisms in a reservoir saturated with both brine and oil. CO2 trapping mechanisms in the brine-only model showed distinctive stages of trapping; the major CO2 trapping mechanisms were hydrostratigrapic (mobile), residual, and solubility trapping during the 200 year simulation period. However, in the brine-plus-oil model, the trapping mechanisms did not vary distinctly over time. Both oil solubility trapping and hydrostratigraphic (mobile) trapping were dominant mechanisms for the entire simulation period. The contrast in CO2 trapping mechanisms in the two models is attributed to differences in thermophysical properties between oil and brine. Based on the two simulation models, injecting CO2 into the brine-plus-oil model reduces the amount of mobile CO2, which would otherwise likely migrate vertically to the top seal. Several advantages of CO2 injection into reservoirs saturated with both brine and oil include: (1) CO2 solubility in oil is significantly greater than CO2 solubility in brine. (2) CO2 does not tend to migrate vertically because oil density is similar to CO2 density, (3) CO2 mobility is reduced when more than two phases coexist, (4) although mobile CO2 may reach the top of the target formation, an oil reservoir is typically capped by a sealing layer, and (5) infrastructure for CO2 injection is already built. As suggested, injecting CO2 into oil reservoirs provides several advantages for minimizing potential CO2 vertical migration. However, oil reservoirs are typically limited with respect to CO2 storage capacity. Therefore, to obtain advantages in terms of CO2 storage capacity and protection from potential leakage, we propose CO2 injection into brine formations below oil reservoirs.
H12D-05
A Hybrid Numerical-Analytical Model for Wellbore Leakage in Reservoir-Scale Simulations
Large-scale implementation of geological CO2 sequestration requires quantification of risk and leakage potential. One potentially important leakage pathway for the injected CO2 involves existing oil and gas wells. Wells are particularly important in North America, where more than a century of drilling has resulted in millions of oil and gas wells. There is significant uncertainty surrounding the integrity of existing wells, and very little data to support quantification of material or hydraulic properties associated with these wells. Models of CO2 injection and leakage will involve large uncertainties in parameters associated with wells, and therefore a probabilistic framework is required. These models must also be able to capture both the large-scale CO2 plume associated with the injection and the small-scale leakage problem associated with localized flow along wells. This is made even more difficult by the fact that within a typical simulation domain, many hundreds of wells may exist. Traditional numerical methods are not suitable because grid refinement is needed to capture wellbore flow, which soon becomes computationally prohibitive because of the large number of wells and the need for multiple realizations in a probabilistic framework. Recent developments in analytical models are promising, however, these methods are limited to idealized geological systems. In this paper, we present a new model that combines both numerical and analytical models into a single hybrid model. The governing equations are vertically-averaged and solved numerically on a relatively coarse grid, thereby capturing the large- scale injection problem. Within this coarse-grid simulation, an analytical model is embedded to solve for wellbore flow occurring at the sub-gridblock scale. This hybrid numerical-analytical method is a powerful tool because it combines the advantages of both numerical and analytical methods. We obtain greater flexibility with the numerical model, which can be used solve heterogeneous and geologically complex systems, while the analytical method provides quick and accurate solutions to the wellbore flow problem, thereby eliminating expensive grid refinement. We show that this method compares well with traditional numerical simulations. It also compares well to the fully analytical model, which applies for appropriately simple systems. We also model a more complex system to demonstrate the flexibility of this model for handling CO2 injection into a dome-shaped aquifer with leakage through an existing well. We believe that the hybrid model provides a simple but powerful tool to evaluate leakage risk in geological CO2 sequestration.
H12D-06
The Role of Optimality in Characterizing CO2 Seepage from Geological Carbon Sequestration Sites
Storage of large amounts of carbon dioxide (CO2) in deep geological formations for greenhouse-gas mitigation is gaining momentum and moving from its conceptual and testing stages towards widespread application. In this talk we explore various optimization strategies for characterizing surface leakage (seepage) using near-surface measurement approaches such as accumulation chambers and eddy covariance towers. Seepage characterization objectives and limitations need to be defined carefully from the outset especially in light of large natural background variations that can mask seepage. The cost and sensitivity of seepage detection are related to four critical length scales pertaining to the size of the: (1) region that needs to be monitored; (2) footprint of the measurement approach; (3) main seepage zone; and (4) region in which concentrations or fluxes are influenced by seepage. Seepage characterization objectives may include one or all of the tasks of detecting, locating, and quantifying seepage. Each of these tasks has its own optimal strategy. Detecting and locating seepage in a region in which there is no expected or preferred location for seepage nor existing evidence for seepage requires monitoring on a fixed grid, e.g., using eddy covariance towers. The fixed-grid approaches needed to detect seepage are expected to require large numbers of eddy covariance towers for large-scale geologic CO2 storage. Once seepage has been detected and roughly located, seepage zones and features can be optimally pinpointed through a dynamic search strategy, e.g., employing accumulation chambers and/or soil-gas sampling. Quantification of seepage rates can be done through measurements on a localized fixed grid once the seepage is pinpointed. Background measurements are essential for seepage detection in natural ecosystems. Artificial neural networks are considered as regression models useful for distinguishing natural system behavior from anomalous behavior suggestive of CO2 seepage without need for detailed understanding of natural system processes. Because of the local extrema in CO2 fluxes and concentrations in natural systems, simple steepest-descent algorithms are not effective and evolutionary computation algorithms are proposed as a paradigm for dynamic monitoring networks to pinpoint CO2 seepage areas. This work was carried out within the ZERT project, funded by the Assistant Secretary for Fossil Energy, Office of Sequestration, Hydrogen, and Clean Coal Fuels, National Energy Technology Laboratory, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
H12D-07
Monotoring of CO2 Sequestration at Sleipner Using Full Waveform Inversion in Time-lapse Mode.
It is now widely admitted that recent increase of CO2 in the atmosphere is due to human activities. The consecutive greenhouse effect is a major ecological concern. Geological storage is one proposed way to reduce atmosphere CO2 emissions. The Sleipner methane field, North Sea, is the very first site where CO2 has been injected back into a deep saline aquifer. In 1996, the Norwegian company Statoil and its partners began the production of the methane. The extracted methane contains a relatively high ratio of CO2, between 4% and 9%, that has to be reduced below 2.5% before delivering into the pipeline. An environmental tax introduced in Norway as early as 1991 prompted the company to store the separated CO2 instead of releasing it into the atmosphere as usually done. The CO2 is injected at the base of the Utsira sands. This water bearing formation lies at a depth between 800 and 1000m and is sealed by a thick shale layer. Seismic monitoring is a key tool in this strategy from a security standpoint and for sequestration optimization itself. Consequently, 3D seismic data were acquired before injection in 1994 and after injection in 1999, 2001, 2002, 2004 and 2006. Well-log revealed that the reservoir is crossed by thin shale layers that are 1 to 10m thick. CO2 rises up and is confined vertically by the shale layers, favouring horizontal gas migration and creating gas bearing thin beds. Seismic imaging of the gas pockets is therefore a challenging problem because large velocity variations occur on very short distance. Classical processing of time-lapse data consists in subtracting repeated survey seismic traces from the pre- injection baseline traces to exhibit changes within the reservoir. This approach remains qualitative, providing only the shape and extent of the gas cloud. Instead, we propose to compare elastic models of the subsurface computed through 2D full wave form inversion, an advanced seismic imaging technique. This method is based on the wave equation numerical simulation and can account for complex propagation effects as encountered in the Sleipner time-lapse data. This makes possible quantitative estimation of P and S-wave velocities on the meter scale. We applied the technique to 2D lines from the 1994, 1999 and 2006 vintages. The resulting post- injection models were subtracted to the pre-injection model to determine both the geometry and the velocity structure of the gas bearing areas which will be used to quantify the amount of CO2 in different forms (free versus dissolved).
H12D-08
Simulations of dry-out and halite precipitation due to CO2 injection
Although H2O is not very soluble in supercritical CO2, a continuous stream of CO2 injected into a formation, will cause a region around the injection well to dry out. As the water of the formation brine is continuously evaporated into the CO2, the irreducible water saturation may attain practically zero. Enhanced injectivity is the result of this process in a low salinity brine environment. In formations saturated with highly saline brine (e.g. Northern German Basin) the outcome is opposite: injectivity is impaired. In this case, the brine becomes supersaturated as continuously H2O evaporates into the CO2 phase and salt (halite) precipitates in the pores. The porosity and permeability diminish, which can lead to the loss of a well. We present simulations of these processes as an example of pre-injection study for a CO2 injection and storage site. The simulation tool consists of a commercial compositional code used extensively in the oil and gas industry to simulate the flow of multiple phases (oil, water, gas) in porous or fractured media. The mutual solubility of CO2 and H2O with a correction for salinity is implemented as described in Spycher and Pruess (2005). The brine salinity is adjusted accordingly until the saturation threshold is reached and halite is precipitated. The distribution of precipitation in the reservoir depends not only of the relationship between permeability change as a function of porosity change, but also on the relative permeability curves for the CO2-brine system. Therefore it is essential to establish relative permeability curves in the laboratory, as well as to obtain a relationship between porosity change and permeability variation with precipitation by laboratory experiments on cores to obtain meaningful results from numerical simulations. References Spycher N. and Pruess, K. (2005), CO2-H2O mixtures in the geological sequestration of CO2, II Partitioning in chloride brines at 12-100o and up to 600 bar, Geochim. Cosmochim. Acta 69, 13, 3309-3320.