Cost-efficient CO2 monitoring technology (Task 12)

The main ambition of Task 12 is to develop and demonstrate monitoring technology that will enable safe operation in compliance with laws and regulations in the most cost-effi cient manner.

This is achieved by research on five topics:

methods for assessing cost vs value-of-information for geophysical data acquisition
ways of extracting all available information from acquired geophysical data
strategies for comparing/combining prior reservoir modelling to geophysical observations for containment and conformance verifi cation
unconventional data acquisition technologies and survey layouts
use of machine learning for geophysical monitoring

To ensure optimal industry relevance and the highest possible relevance for the Northern Lights project (the carbon storage project of Equinor, Shell, and Total) on the Norwegian Continental Shelf, the industry will be closely involved through the lifetime of the Task. Involvement of vendors with expertise from oil and gas monitoring will also be important, and we will seek to have a dialogue with regulators.

Most of the monitoring activities in this Task are related to both Deployment Cases, since similar geophysical monitoring technologies are expected to be used in both cases.

Results 2021

The focus on methods for the optimal use of geophysical information continued, with further development of the method for rock physics inversion that this year focused on a framework for inversion based on selected machine learning algorithms. Four different algorithms were tested with respect to their efficiency at determining porosity, bulk and shear modulus from a velocity model.

Initial studies were also carried out to better understand the potential use of PS-converted seismic waves for the discrimination of pressure and saturation changes in the subsurface. The results were encouraging but obtained using a highly simplified synthetic model. The plan for 2022 is to carry out similar tests with a much more realistic model and potentially later with real data, adding significant complexity.

While much of the focus in Task 12 is on monitoring Norwegian CO₂ storage projects, a major study has also been initiated relateing to the characterization and modelling of the Bunter Sandstone Formation saline aquifer in the southern North Sea. In 2021, a model of the area was constructed and large-scale injection modelling carried out. Additional geomechanical modelling indicated a level of subsurface displacement that may be observed using InSAR if a suitable set of point sources overlay the region. It was pointed out that planned windfarms in the region may provide a suitable monitoring reflector to assess seabed movement. In addition, a set of seabed tiltmeters may provide the necessary resolution for the regional geomechanical monitoring.

Within the EM4CO2 project, some of the most important results include the demonstration of the feasibility of using CSEM for monitoring at Smeaheia, how CSEM survey layouts can be minimised without a loss of information, and how to account efficiently for infrastructure effects during EM modelling. The feasibility study showed promising results, with the CO₂ plume after 25 years of simulated injection being clearly visible on inversion results. Using a recently developed survey optimisation approach, it was also verified that a significantly reduced, but a wisely designed survey gave a nearly equally well imaged plume (see Figure 1). Another highlight in the EM4CO2 project was the recruitment of a postdoc researcher who will work on a new approach for time-lapse CSEM inversion.

Figure 1: Resistivity models obtained using CSEM inversion for a synthetic Smeaheia model. The top row shows the inversion of baseline data before injection, while the middle row shows the inversion of data after injection. The bottom row models show the difference before and after injection. [left column] Results using all data (approximately 17000 data points) from a dense survey. [right column] Results using 200 carefully selected data points. Results are similar despite the significant reduction in acquired data.

For the TOPHOLE project, the year meant further work on the monitoring concept integrating the screening of legacy wells, and the localisation, characterisation and monitoring of these wells with non-invasive methods. The screening approach has been tested on a large number of wells in the Troll area (see Figure 2), and more recently as part of a detailed study for the Smeaheia structure, where the full concept has been applied. The interest in and need for this type of tool has resulted in plans for preparing a professional version of the tool during 2022. Experimental work has been started and is focusing on acoustic measurements on a well replica and testing the leakage detection abilities of accelerometers and fiberoptics. Ehsan Hosseini (NTNU PhD student) has been working on electromagnetic modelling of well signatures and effects of cement and corrosion on the recorded signals.

Figure 2: (A) Map of the North Sea, with well locations and a white square indicating the studied area. (B) Map of the Troll area offshore SW Norway, with well integrity screening results for 93 P&A wells indicated by red dots (Emmel and Dupuy, 2021). Major faults are shown and the location of well 32/4-1 T2 within the Smeaheia alpha structure is highlighted and marked with a green star. From (Romdhane et al., 2022).

Results 2020

In 2020, effort was made to disseminate important recent results from the CO₂ monitoring task of the NCCS centre.

The Bayesian Rock Physics Inversion technique (completed and demonstrated in 2019) and its application to Sleipner data was thoroughly described in a paper (“Combined geophysical and rock physics workflow for quantitative CO₂ monitoring”) for the International Journal of Greenhouse Gas Control (IJGGC) that was accepted for publication at the end of the year. Similarly, another IJGGC paper, “Assessing the value of seismic monitoring of CO₂ storage using simulations and statistical analysis”, summarizing previous work on a value-of-information assessment workflow was accepted for publication at the very end of the year.

At the beginning of the year, four abstracts were also sent to the GHGT conference, which was unfortunately delayed until 2021. All abstracts were accepted and draft papers have been written related to each of these topics.

One of the papers describes the current status of work done on combining monitoring data and reservoir simulations of CO₂ storage sites for improved understanding of a storage site. The method, based on a flexible modelling framework in the open source Matlab Reservoir Simulation Toolbox (MRST), allows optimization of CO₂ storage modelling to any combination of observed monitoring data. It has been applied to the new, layered, Sleipner benchmark model, which was optimised to fit plume outlines, gravity monitoring data and CO₂ saturations inferred from joint full waveform and rock physics inversion with promising results.

Another GHGT draft paper describes how SINTEF's optAVO method can be used to map the extent of the CO₂ plume and provide an estimate for seismic parameters for a synthetic Sleipner case. Using the results (a P-wave model) of the optAVO as an initial model for Full Waveform Inversion (FWI) was seen to significantly improve the final FWI results.

The two other GHGT draft papers focused on monitoring at Smeaheia; one about application of a workflow for baseline quantitative characterisation using synthetic and real data from Smeaheia and one about a feasibility study on marine CSEM monitoring of CO₂ flow along the Vette Fault. This latter study is the first of its kind as the only previous evidence of the efficiency of CSEM for detection of CO₂ flow through faults/fractures was made through a laboratory test. The work has resulted in knowledge about how the CSEM data may interact with CO₂ flow through the Vette Fault. Issues for which the CSEM technique should be improved were also identified.

Aerial view of the Svelvik CO2 Field Lab with injection (#2) and monitoring wells (M1-M4) marked. The loops with fibre optic cables (several different commercial and research types) are shown yellow with tongues indicating how the cables also go down into the wells. — Aerial view of the Svelvik CO₂ Field Lab with injection (#2) and monitoring wells (M1-M4) marked. The loops with fibre optic cables (several different commercial and research types) are shown yellow with "tongues" indicating how the cables also go down into the wells.

Two successful webinars were carried out in 2020, both with a focus on more novel and unconventional technologies for CO₂ monitoring. The spring webinar summarized work done within the task related to "A machine learning based monitoring framework for CO₂ storage". The approach is based on integration of reservoir modelling, geophysical monitoring, and decision-making theory, and it was shown that a neural network can be trained to optimize geophysical data acquisition in terms of its value for verification of site performance. This is a first step towards a novel AI-based technique to support the decision-making process related to cost-efficient MMV. The autumn webinar, with nearly 230 registered, had a more general look at "Safe and cost-efficient CO₂ storage: Emerging monitoring technologies" with an overview of what we see as promising future techniques. An informal survey on the view of the participants was also carried out.

In addition to this survey of emerging technologies, two smaller studies were carried out in 2020 to investigate the potential of using ECCSEL infrastructure in future work within the monitoring task.

The first investigated the possibility and usefulness of performing measurements of elastic and electric parameters at controlled laboratory and test site conditions in order to explore the physics and the interrelationships of these parameters as functions of saturation, pressure and temperature. Such measurements will contribute to calibrating rock physics models and would consequently result in more reliable interpretation of monitoring data. Several potential research tasks were suggested. Similarly, a study of potential research efforts on development and testing of fibre-optic (DAS/DTS/DSS) technology was carried out. This emerging monitoring technology is still not properly explored for CO₂ monitoring purposes.

The recently upgraded Svelvik CO₂ Field Lab has several monitoring wells instrumented with various types of optical fibres and offers great possibilities for such studies. A number of potential research tasks and spin-off projects have been suggested for the next few years in NCCS.

Word cloud summarizing webinar particpants answers to What are the most promising emerging technologies for CO2 monitoring. (right) Boston square matrix showing webinar participants answers to Estimate potential impact of technology when deployed and the research effort needed to get there. — Word cloud summarizing webinar particpants answers to "What are the most promising emerging technologies for CO₂ monitoring". (right) Boston square matrix showing webinar participants answers to "Estimate potential impact of technology when deployed and the research effort needed to get there".

Main results 2019

Reliable monitoring of a CO₂storage site is essential for safe and efficient operation, as well as for public acceptance. By carefully monitoring the site before, during, and after CO₂injection, the risk for very costly intervention, remediation, or site closure is significantly reduced. Such surveillance can potentially be very expensive. The main ambition of Task 12 is to develop and demonstrate monitoring technology which will enable safe operation in compliance with laws and regulations in the most cost-efficient manner.

During the year, we applied a newly developed approach for reservoir parameter estimation with uncertainty quantification (Bayesian Rock Physics Inversion) to Sleipner 2008 seismic data. We demonstrated how important reservoir parameters, such as CO₂saturation and reservoir pressure, can be estimated with a simultaneous assessment of uncertainty, providing essential operational information to the storage site owner. Initial studies also showed how the estimated reservoir parameters can be used to constrain and calibrate reservoir simulations of the Sleipner injection. This calibration enables improved prediction of the future behaviour of the storage site.

The development and testing of a compressive sensing technique for enhanced geophysical data acquisition and interpretation continued in 2019. This technique, which can help to reduce the need for dense (and expensive) seismic surveys, was succesfully verified for sparse subsets of Sleipner data.

Reliable and cost-efficient monitoring will be essential for the Northern Lights project. Task 12 developments like the ones described above will support the design of an optimal monitoring scheme. Another such example is how the research and development of Controlled Source Electro-Magnetics (CSEM), as a complement to seismic, could provide a more cost-efficient and accurate approach for quantitative CO₂monitoring. Based on synthetic Smeaheia data, a quantitative CSEM inversion study was successfully concluded in 2019 (see the figure). Results show that CSEM can be used to give accurate volume estimates.

For any future storage project, there is also a great value in the research efforts on value-of-information, which offer new ways for an operator to analyse and select optimal geophysical monitoring strategies. During 2019, a conceptual Smeaheia case was used to demonstrate how a novel method for value-of- information analysis, based on machine learning, can be used to determine the optimal way of detecting potential leakage from CO₂storage.

In 2019, we also initialised two spin-off projects (EM4CO2 and Tophole) for more detailed studies of two important topics. EM4CO2 investigates the use of electro-magnetic methods as a complement to seismic for more quantitative reservoir monitoring information. Tophole studies how the integrity of plugged and abandoned wells can be cost-effectively monitored to enable CO₂storage in regions with existing wells.

CO₂ volume estimations from CSEM inversion results. Left: True model, Middle: Result with unconstrained inversion, Right: Result with constrained inversion.

Results 2018

Main Results

Smeaheia baseline geophysical models and rock physics models built using Gassnova seismic data (Fig. 1)
Sensitivity test of CO₂ injection on seismic observables at Smeaheia
Initial sensitivity studies for use of CSEM at Smeaheia (Fig. 1)
Demonstration of joint rock physics inversion approach at Sleipner using CSEM and seismic 2008 datasets
Validation of compressive sensing strategy for improved cost-efficient imaging
New survey optimization strategy tested
Work on combined modelling-monitoring and "history matching" initiated
Evaluation of cost-saving potential of NCCS CO₂ monitoring developments

Impact and innovations

First application of FWI to get most out of Gassnova's 3D seismic data at Smeaheia could become useful for Northern Lights project for reservoir seal characterization and monitoring planning
The compressive sensing approach can help to reduce the need for dense (and expensive) seismic surveys
Survey optimization technique and combined modelling-monitoring workflow will help to find cost-efficient ways of confirming site conformance during injection

Selected results from Task 12 in 2018. [top left] P-wave interval velocity model in depth for a subset of the Inline 1024 from Gassnova's seismic cube GN1101. [top right] P-wave velocity model derived from FWI for the subset of Inline 1024 (from Dupuy et al. 2018, to be published). [middle left] 2.5D true resistivity models for simulated injection at Smeaheia until 2045. [middle right] CSEM inversion results. [bottom] A survey optimization strategy was developed and tested for synthetic Sleipner CSEM case. Circles representing optimal lateral source position and sensitivity (area). Each source is used together with all five receivers (represented by red triangles).

Results 2017

The task focused on setting up synthetic Smeaheia geophysical models, on developing new approaches for efficient use of available data for quantitative CO₂ monitoring, and on using a statistical value-of-information concept for cost-minimization of CO₂ monitoring.

For Smeaheia, Statoil's CO₂ injection simulations were used to build synthetic models of the subsurface acoustic velocity, resistivity, and density at different times during and after the injection. These models together with Smeaheia data provided by Gassnova will serve as very important input for targeted Smeaheia monitoring studies in the years to come in NCCS.

Work on quantitative CO₂ monitoring at Sleipner led to promising results that was presented at several workshops and conferences and published in several journals. In total, six publications were produced.

Industry partners Statoil, Shell and Total, as well as vendor Quad Geometrics have contributed to the task and participated in a workshop in September. Late in 2017, EMGS confirmed that they want to join the task as a vendor.

Publications

Task leader

Jon Peder Eliasson

Senior Research Scientist

47 36 97 32

Name: Jon Peder Eliasson
Title: Senior Research Scientist
Phone: 47 36 97 32
Department: Petroleum
Office: Trondheim
Company: SINTEF AS

Norwegian CCS Research Centre

Cost-efficient CO₂ monitoring technology (Task 12)