«OPTIMIZATION AND MONITORING OF GEOLOGICAL CARBON STORAGE OPERATIONS A DISSERTATION SUBMITTED TO THE PROGRAM IN EARTH, ENERGY AND ENVIRONMENTAL ...»
OPTIMIZATION AND MONITORING OF GEOLOGICAL
CARBON STORAGE OPERATIONS
SUBMITTED TO THE PROGRAM IN
EARTH, ENERGY AND ENVIRONMENTAL SCIENCE
AND THE COMMITTEE ON GRADUATE STUDIES
OF STANFORD UNIVERSITY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
In this work we develop and apply computational optimization procedures to minimize the risk of CO2 leakage and to perform data assimilation in order to identify the location of the CO2 plume and detect CO2 leakage through the cap rock. The risk of leakage is quantiﬁed both in terms of the mobile CO2 in the formation and in terms of the total mobility of free CO2 at the top of the storage aquifer. Risk minimization is accomplished by determining optimum locations and time-varying injection rates for a set of horizontal CO2 injection wells. Both Hooke-Jeeves Direct Search and Particle Swarm Optimization algorithms are used for this purpose. A brine cycling procedure, in which brine is periodically produced at the bottom of the aquifer and reinjected at the top of the aquifer, is also considered, and the parameters associated with this operation are optimized. For data assimilation (or history matching), aquifer geology is represented in terms of a relatively small number of parameters using a KarhunenLo`ve (K-L) expansion. Sensor and CO2 injection-well data provide the measurements e to be matched. A procedure for optimizing the placement of monitoring wells and v the weights of the various types of measured data, with the goal of maximizing the eﬃcacy of the history matching procedure, is also presented.
Optimization results for both deterministic and uncertain aquifer models (in the latter case, the aquifer is represented using multiple realizations) are presented for a variety of cases, and reduction in the risk of leakage is consistently achieved. Specifically, by optimizing well placement and control (with known geology), the mobile CO2 fraction is reduced from around 0.32 to 0.22, and the total mobility is decreased by around 39%. For cases with uncertain geology, the reduction in mobile CO2 from optimization is only 7%, highlighting the need for a-priori geologic characterization.
Optimizing brine cycling processes leads to further risk reduction, and a plot of risk of leakage versus pore volume of brine injected (which is related to cost) provides a Pareto front for a bi-objective optimization involving these two variables as objectives. The data assimilation procedure is shown to improve predictions for the CO2 plume location relative to results from prior geological models. Speciﬁcally, in a series of tests, this procedure reduces the average error in the predicted CO2 mobility in the top layer of the model (which is the quantity of interest) by 46% relative to the error using the prior model.
Finally, we investigate the early detection of leaks in the cap rock using pressure data. We introduce a three-region model to quantify the amount of leakage for a large number of leakage cases (some including multiple leaks). A data assimilation method is applied to determine leakage locations and permeabilities for a number of cases, with pressures at sensor wells and injection wells providing the measured data.
Particle Swarm Optimization is used for the minimizations associated with this data assimilation problem. A data-rich scenario with nine sensor wells (completed in the overlying aquifer and storage formation) and a data-scarce scenario with four sensor wells (completed only in the overlying aquifer) are considered. Results indicate that the history matching procedure eﬀectively locates leakage positions in cases with a single leak, for both the data-rich and data-scarce scenarios. For cases with multiple leaks, however, the procedure is less reliable, though the data-rich scenario is shown to provide better matches than the data-scarce scenario.
Firstly, a huge thank you to my adviser, Prof. Louis Durlofsky, who not only conceived the idea for this research, but also helped guide it toward success. Lou’s expertise, professionalism, ideas and writing skills were appreciated and extensively utilized in this work. I’d also like to acknowledge the members of my defense committee, Profs. Sally Benson and Roland Horne, and Dr. Anshul Agarwal, for their insightful comments on this dissertation. Sally Benson, in particular, has provided helpful advice on numerous issues related to carbon storage modeling. Several others have also contributed to this work by kindly sharing their knowledge and sometimes also their codes. These include Profs. Peter Kitanidis, Tapan Mukerji, Andre Journel and Michael Saunders, as well as Drs. Obi Isebor, David Echeverr` Ciaurri, Samuel ıa Krevor and Jermome Onwunalu.
The Department of Energy Resources Engineering faculty and staﬀ deserve credit for their hard work in maintaining an outstanding learning environment. I am also extremely grateful to the Smart Fields Consortium and Stanford Carbon Capture and Storage (SCCS) program for providing ﬁnancial support during my studies.
In my time at Stanford, I’ve made some amazing friendships that will endure for a long time to come. Dan, Mike, Zach, Clement, the Maddux guys, and the Cowper bunch have helped to make my life here extremely fun. Special thanks to my parents, Robin and John Cameron, and sisters, Helen and Kirsty Cameron, whose love and support have meant so much over these years and who did such a great job bringing me to where I am today. Finally, a massive thanks to my wife Jessica Reeves. Jess made my time here the time of my life, and I know she will continue to do so wherever we travel to next. Cheers!
4.1 Results for default and optimized cases for TR1 - TR5........ 81
4.2 Scaled optimal data weight parameters, where wP weights the sensor pressure data, wBTT weights the CO2 breakthrough time data, and wBHP weights the BHP data at the injection wells........... 86
4.3 Errors in prior and history matched models for ﬁve geological realizations 89
4.6 Optimal injection rates for the four wells during the 30-year injection phase for geology TR1.......................... 74
4.7 Progression with time of mobile, immobile (residually trapped) and dissolved CO2 in the default and optimized cases for geology TR1.. 75
4.8 Mobile CO2 in the top layer of the model for default and optimized cases for geology TR1........................... 75
4.9 Pareto front for optimizations with brine cycling for geology TR1.. 77
4.10 CO2 injection well placements with time-averaged (over 500 years) CO2 mobility in the top layer for the case with 0.03 PV brine cycling and geology TR1................................ 77
4.11 Optimized CO2 injection rate and brine cycling parameters for the case with 0.03 PV brine cycling and geology TR1(B1 - B4 refer to brine injection wells corresponding to CO2 injectors W1 - W4).... 78
4.12 Progression with time of mobile, immobile (residually trapped) and dissolved CO2 for default and optimal conﬁgurations for the case with
0.03 PV brine cycling and geology TR1................. 79
4.13 CO2 injection well placements with expected value of time-averaged (over 500 years) CO2 mobility in the top layer (optimization based on 10 prior geological realizations)..................... 80
4.14 Optimal sensor placements for the two scenarios considered for TR1.
Injection wells (placed according to the a-priori optimization) provide bottomhole pressure data. Vertical sensor wells provide pressure and CO2 breakthrough time in each block they intersect.......... 84
4.15 Optimal sensor placements for the two scenarios considered for TR1 to TR5.................................... 85
4.16 Progression of the GPS minimization of data misﬁt.......... 87
4.17 Comparison of history matched solution with the true model (TR1) and with expected values from the prior model............. 88
4.18 Error in time-averaged mobility in top layer for history matched and prior models, for true model TR1.................... 90
Introduction In 2007, The Intergovernmental Panel on Climate Change (IPCC) concluded, with very high conﬁdence, that anthropogenic greenhouse gases are responsible for observed global warming (Solomon, 2007). Moreover, warming is expected to increase with possibly catastrophic consequences, unless emissions of greenhouse gas to the atmosphere are signiﬁcantly reduced. An obvious target for reductions are fossil fuel burning power plants, which account for around 30% of global carbon dioxide emissions. Ideally these power plants would be replaced with cost-eﬀective renewable sources of energy, but this transition may not be fast enough to mitigate large scale climate change. A potential interim solution is to capture the carbon dioxide produced at power plants and store it in underground geological formations. This overall process is known as carbon capture and storage (CCS).
CCS is a developing ﬁeld and there are many proposed approaches for carbon storage. One of the most promising solutions is to inject supercritical carbon dioxide into underground saline aquifers. Several projects using this approach are currently in operation. For example, the Sleipner project in Norway has been injecting 1 megatonne of CO2 per annum into a subsea saline aquifer since 1996. The CO2 in this case is a byproduct from oﬀshore natural gas production. Similar projects in In Salah, Algeria and Weyburn, Canada are also underway.
A major concern associated with CCS in saline aquifers is the potential for leakage. The consequences of leakage may include groundwater contamination, ecosystem
2 CHAPTER 1. INTRODUCTIONdamage, and emissions of greenhouse gases into the atmosphere. The majority of previous leakage-related research has focused on detection and remediation. While this is very important, CCS risk management should also include optimization of injection processes such that the CO2 is stored as safely as possible.
The task of optimizing CO2 storage operations to maximize ‘safe’ trapping mechanisms (for example, dissolution and residual trapping, discussed below) is complicated due to the physics of the problem, the need to account for geological uncertainty, and the lack of data to test model predictions and assertions. The goal of this thesis is to introduce and test computational procedures that can be used to predict and optimize the performance of CO2 storage operations.
We propose a closed-loop modeling and optimization framework to systematically predict and optimize CO2 storage operations. The goals of our procedure are to facilitate safe modes of trapping, minimize the risk of leakage, and reduce uncertainty in our knowledge of the CO2 plume location. The framework for a-priori optimization and closed-loop modeling and optimization is illustrated schematically in Figure 1.1.
We propose and test methodologies for the key components of this framework, namely optimization of well placement and control and data assimilation to update the geologic model.
Figure 1.1: Flow chart for a-priori optimization and closed-loop aquifer management
1.1. LITERATURE REVIEW
1.1 Literature review We begin our literature review with a discussion of previous studies relating to carbon storage modeling. We then provide descriptions of literature relevant to carbon storage optimization. In particular, we focus on a-priori optimization and closedloop aquifer management. We review research pertaining to data assimilation (also referred to as model updating or history matching) and leak detection methods. Because there is extensive literature in many of these areas, we restrict our discussion to the papers that are most relevant to the work in this thesis.
1.1.1 Carbon storage modeling Numerical procedures for simulating carbon storage in subsurface formations are fairly well developed, though there are still modeling and computational issues that require further study. A number of simulators have been developed or adapted for CO2 storage, and most can model the key physical eﬀects relevant to CO2 trapping and migration (see, e.g., the discussion and code comparisons in Pruess et al., 2004; Class et al., 2009). Achieving accurate modeling of CO2 storage in reasonable computational time is essential for optimization and data assimilation, because many simulations (hundreds or thousands) must often be performed. Thus, it is important to identify the key physical eﬀects to be included in the model, and to identify secondary eﬀects that can be neglected (particularly those that lead to excessive computational demands).