OR/18/013 Introduction

Quinn, M, Hannis, S, Williams, J, Kirby, G, and McCormac, M. 2018. Multiscale Whole Systems Modelling and Analysis Project — A description of the selection, building and characterisation of a set of 3D generic CO2 storage models. British Geological Survey Open Report, OR/18/013.

The aim of this Open Report is to provide a reference document detailing the selection and construction of three 3D geological models carried out as part of a multi-disciplinary NERC funded project entitled ‘Multiscale Whole Systems Modelling and Analysis for CO₂ capture, transport and storage (CCTS)’ (Grant Reference: NE/H013946/1 and henceforth referred to as the 'Whole Systems' project). The 3-year project (2010–2013) was led by Imperial College, London with Cranfield University (later Hull), Sussex University and the British Geological Survey (BGS) as partners.

The report outlines the methodology adopted in building these three very different models, along with a full understanding of their scope and limitations, and provides essential background information for model use. In addition, the report also briefly describes how the 3D models were used to assess CO₂ storage performance; results to date from these studies have been published in a number of peer-reviewed journals and presented at international conferences.

Implementation of large-scale carbon capture and storage requires substantial capital investment in CO₂ capture, transport systems and storage complex management. The Whole Systems project aimed to examine the performance of the different parts of the CCTS chain assessing these, where appropriate, at microscopic through to macroscopic scales. Understanding the performance and controlling factors of each technology enabled their interaction with other parts of the CCTS chain to be better understood. Results from this project should contribute to designing optimal systems for a range of potential CO₂ capture and storage scenarios. The technologies considered and partners responsible are shown in Table 1 below.

Table 1 Multiscale Whole Systems Modelling and Analysis project — summary of the different parts of the CCTS chain considered, shown with responsible project partners.

Technology	Responsible partner
Analysis of future energy systems	Sussex
Power plant model	Cranfield (later Hull)
Capture plant model	Cranfield
Transport model	Imperial College
Injection and storage model; Geological model Reservoir model	BGS & Imperial College
Network design	Imperial and Sussex
Life Cycle and Integrated assessment	All
Single chain dynamics, network stability and operability.	Imperial and Cranfield

Overview of utilisation of the 3D geological models in the project

For CO₂ storage sites, Imperial College and BGS focused significant effort on improving their understanding and ability to characterise and predict CO₂ storage site evolution syn- and post- the injection phase. Investigations focused on the performance of different types of geological settings (utilising different geological reservoirs) and the resultant consequences for the other parts of the CCTS chain. To be useful to the CCTS community, the stores chosen for this study were those most likely to be utilised in the near term.

Four potential sandstone reservoirs were identified (Figure 1):

• the Permian Leman Sandstone Formation in the UK Southern North Sea;
• the Triassic Bunter Sandstone Formation in the UK Southern North Sea;
• a submarine fan sandstone reservoir comprising the Palaeogene Forties and Cromarty sandstone members in the UK Central North Sea;
• and the Lower Cretaceous Captain Sandstone Member in the UK Northern North Sea.

A regional 3D model for part of the Captain Sandstone (Quinn et al., 2010) and a detailed model of the Captain Sandstone in and adjacent to the Blake hydrocarbon Field (Quinn et al., 2012) are available as confidential reports and this reservoir is not considered further in the report. The remaining three models were built from defined locations in the UKCS using published information and released well data. The three models are referred to throughout this report as follows:

1. The Rotliegend model is located within the Permian Leman Sandstone Formation (see Figure 1, Selection of the Rotliegend generic model and The 'Rotliegend' generic model);
2. The Bunter model is located within the Triassic aged Bunter Sandstone Formation (see Figure 1, Selection of the Bunter generic model and The 'Bunter' 3D generic model);
3. The Cenozoic model is located within the Palaeogene Forties and Cromarty Sandstone members (see Figure 1, Selection of the Cenozoic generic model and The 'Cenozoic' 3D generic model).

The distribution of sedimentary facies was defined within each model utilising published studies, BGS in-house knowledge and interpretation of well records. For two of the models (Rotliegend and Cenozoic), petrophysical data, including, permeability, porosity and Net-to-Gross, were interpreted by the BGS and supplied to Imperial College for attribution. The BGS fully attributed the Bunter model prior to submission to Imperial College.

The reservoirs characterized in the three models were deposited in very different geological environments. In addition, each reservoir has a large regional extent and the characteristics of each will vary in detail away from the site where each model was built. In order to assess how the performance of a CO₂ store might change with location, the regional extent of each reservoir was divided, where possible, into different sectors or ‘Area Types’ to represent the broad variations predicted for each reservoir. Each Area Type is defined by a unique set of petrophysical values as well as changes to the depth and thickness of the potential storage reservoir. Thus, each of the 3D models built for this project is generic, in that their reservoir divisions and structure can be transposed to any part of the basin in which that particular reservoir has the potential to be a CO₂ store.

Following handover of the models, Imperial College populated the models with petrophysical attributes and performed a series of numerical flow simulations to quantify the storage performance of the different reservoirs. For example, Korre et al., 2013 modelled an injection rate of 1 MtCO₂/yr of CO₂ from 1 injection well placed in the Permian Leman Sandstone Formation reservoir in the Rotliegend model. Injection from a second well was added when pressure in the first reached its maximum acceptable value. In this way a set of performance metrics, known as Key Performance Indicators (KPI’s), were derived to characterize the performance of the potential store in different parts of the studied reservoir (Korre et al., 2013). The Key Performance Parameters (KPI’s) defined were:

• Period of Sustained Injection (PSI); the duration wherein a pre-specified constant injection rate can be maintained;
• Fraction of Capacity Utilised (FCU); the fraction of available pore space within the reservoir occupied by CO₂ during the PSI.

In the case of the Rotliegend and Cenozoic models, where a substantial volume of the reservoir included hydrocarbon field/s, it was possible to carry out quality assurance (QA) on the built and attributed models by modelling production of hydrocarbons from the field and comparing the amount produced with actual production (Korre et al., 2013; Babaei et al., 2014a). Comparison of modelled production with actual published production thus provided a level of confidence in the attributed model used in assessment of a potential CO₂ store.

The Cenozoic model, capturing geology in and around the Forties and Nelson oilfields, has been utilized to investigate the possibilities of using upscaled models to speed up the identification of optimal solutions for injection and well placement and CO₂ storage potential in a potential CO₂ store (Babaei et al., 2014a; Babaei et al., 2014b). To date the Bunter model has not been used for simulation purposes.

Figure 1 Map showing extents of the saline aquifers utilised by the three generic models for this project. Hydrocarbon fields are shown in red. The saline aquifers shown are from Knox and Holloway (1992); Johnson and Lott (1993) and Johnson et al. (1994)

Regional context and selection of generic models

At the beginning of the project, a list of potential CO₂ stores was compiled (Table 2). From this review, three potential reservoir stores; the Leman Sandstone Formation (depleted gas fields), the Bunter Sandstone Formation (saline aquifer and depleted gasfields) and the Forties and Cromarty sandstone members (depleted oil field with surrounding saline aquifer) were identified for this study (Figure 1).

Selection of the Rotliegend generic model

Depleted Rotliegend gas fields in the Southern North Sea provide a significant portion of the UK’s CO₂ storage potential (Bentham, 2006; Holloway et al., 2006). Gas has been produced from the Leman Sandstone Formation in the UK sector since the discovery of the West Sole gas field in 1964 (Hardman, 2003). Many of the fields have either ceased production or are nearing their expected cessation dates, so will likely be available for storage in the near-future. In fact, one field, the Rough gas field is now used as a seasonal natural gas storage facility (Ellis, 1993). Reservoir quality is generally good, and valid structural traps have already been proven by the large number of natural gas discoveries.

The Rotliegend sandstone 3D geological model (see The 'Rotliegend' generic model) was built from published surface and well information from the depleted Ravenspurn North and South Gas fields (Ketter, 1991a). The BGS model comprised six reservoir layers with major internal and bounding faults.

Initially, the Rotliegend sandstone reservoir within the UK SNS was divided into 5 ‘Area Types’ based on variation in reservoir Depth and Thickness. A set of performance indicators were generated from these different Area Types that were then further subdivided on the basis of known regional heterogeneity and production performance values (including facies distribution, cementation of the reservoir, pressure and the economics of extraction for individual fields), increasing the number of Area Types to 10.

Selection of the Bunter generic model

The Lower Triassic Bunter Sandstone Formation is considered to have significant CO₂ storage potential within closed structural domes in the saline aquifer parts of the formation (Bentham, 2006; Holloway et al., 2006). Potential storage sites are currently being actively explored by industry with a view to utilising the Formation as a demonstration of industrial scale CO₂ storage in the UK. Additionally, the Formation has properties that meet best practice requirements (Chadwick et al., 2008) and the presence of several natural gas accumulations demonstrate locally the ability of the overburden to retain buoyant fluids (e.g. Williams et al., 2014).

The Bunter Sandstone 3D geological model (see The 'Bunter' 3D generic model) was built around an area within the so-called ‘Silverpit Basin’ of the Southern North Sea. In this area, the Bunter Formation has been gently folded by mobilisation of the Zechstein salt in the underlying Permian strata to form a series of anticlines and synclines. These folds, which typically form the culminations of NW–SE trending periclinal ridges, are characteristic of the structures actively being considered for CO₂ storage in the Bunter Sandstone.

The extent of the model itself was designed to encompass several different structures over which seismic and well data were available to the project, with a likely cumulative storage capacity of at least 5 Mt/year over a 30-year period. The selected area includes the producing gas fields; Caister B and Hunter, both of which produce natural gas from the Bunter Sandstone. An additional brine-saturated closure within the Bunter Sandstone aquifer, known as 3/44 (Bentham, 2006), is also included in the model area.

For the purposes of 'genericising' the model, the gas fields were not attributed with hydrocarbon fluids and were treated as part of the hydraulically connected saline aquifer. Note however, hydraulic connection may be compromised in some areas. For instance, in the Caister Gas Field halite cementation below the Gas Water Contact (GWC) may inhibit water influx.

The model surfaces were derived using a combination of 1 km x 1 km depth converted seismic surfaces, provided to the project by Petroleum GeoServices (PGS) (derived from their Southern North Sea (SNS) MegaSurvey data), augmented with available information from exploratory, appraisal and development wells.

Selection of the Cenozoic generic model

The Cenozoic submarine fan sandstone reservoir is considered to be a near term choice for CO₂ storage for the following reasons:

• Of the ten potential saline aquifer stores identified in the Scottish Joint Study (Scottish Centre for Carbon Storage, 2009), eight were of Paleocene/Eocene age and deposited in a submarine fan environment;
• Were the Goldeneye hydrocarbon field to become the UK’s first CO₂ storage demonstrator, this would increase the accessibility and attractiveness of the Cenozoic hydrocarbon fields and saline aquifers to the east and south-east in the Outer Moray Firth and Central North Sea;
• Reservoir parameters in parts of the Cenozoic submarine fan system meet best practice requirements (Chadwick et al., 2008) for CO₂ storage, for instance in the up-dip proximal part of the Forties Fan system, Net to Gross can be 65%, porosity 23–26% and permeabilities, 100’s mD to Darcies;
• Hydrocarbon exploitation of these reservoirs means good data availability.

The Cenozoic submarine fan sandstone 3D geological model (see The 'Cenozoic' 3D generic model) was built around two depleted oil fields; Forties and Nelson, and includes part of the adjacent water-filled reservoir. The modelled reservoir is primarily based upon the Paleocene/Eocene (i.e. Palaeogene) Forties Sandstone Member. Published surface and thickness information, constrained by released well data was used to build the different layers within the model. The model comprises 7 reservoir zones together with two major pressure discontinuities and top seal.

Attribution of the model was constrained by two broad facies distributions i.e. channel and interchannel. These were recognized in 6 of the reservoir zones and were built into the model in the form of limiting polygons on the appropriate layers.

Three ‘Area Types’, based on the different positions on the south-easterly advancing submarine fan, were defined. These Area Types governed depth, the number of reservoir layers and petrophysical values applied to the model. By changing these parameters, the model could be made to represent different parts of the submarine fan and enabled the storage potential of different parts of the fan to be quantified.

Table 2 Potential CO₂ stores located in the Central, Northern and Southern North Sea areas, United Kingdom Continental Shelf. Red text highlights the stores considered to be the most likely to be utilised first and have been selected for modelling.

	Location	Lithostratigraphic unit	Reservoir	Contained fluids	Depositional environment	Seal	Trap	Type of Store	Boundary conditions
Cenozoic	Southern North Sea					It is unlikely the Cenozoic will be utilised as a CO2 store in the SNS
Cenozoic – Miocene to Oligocene	Central and Northern North Sea					It is likely that these sandstones will be too shallow for CO2 storage
Cenozoic – Paleocene Ecocene	Central and Northern North Sea	e.g. Tay, Grid, Frigg sandstone members	Sandstone	Saline water or hydrocarbons	Submarine Fan System	Mudstones and Siltstones	Ultimately stratigraphic but size of reservoir means capillary trapping and dissolution would be the chief trapping mechanism	Saline aquifer/depleted hydrocarbon field	OPEN SYSTEM. Few faults means that the rate of pressure increase will be related to permeability of the reservoir and could be controlled by rate of injection
		e.g. Cromarty, Mey, Heimdal and Forties sandstone members
Upper Cretaceous	All areas		Chalk	Saline Water or hydrocarbons		It is unlikely that the Late Cretaceous chalk would be utilised as a long term CO2 store
Lower Cretaceous	Outer Moray Firth	e.g. Captain Sandstone Member	Sandstone	Saline water and hydrocarbons. In the Captain Field the oil is heavy (19-21 deg API)	Submarine Fan System	Mudstone and occasionally chalk	To the west, the Captain reservoir subcrops at the sea bed and, outside structural closure, the amount of CO2 injected will be governed by rate if migration and its storage by capillary and dissolution.	Saline aquifer/depleted hydrocarbon field	OPEN SYSTEM. Faults do not appear to compartmentalise this reservoir. It is likely that the Captain reservoir subcrops at sea bed.
Jurassic	Southern North Sea					It is unlikely that the Jurassic will be utilised as a CO2 store in the SNS
Upper Jurassic	Central and Northern North Sea	e.g. Brae and Fulmar formations, Magnus Member.	Sandstone	Saline water or hydrocarbons	Submarine Fan Systems (Brae, Claymore, Magnus). Shallow marine, low to moderately high energy storm influenced setting	Mudstone and Siltstones	Structural (crests of tilted fault blocks, anticlinal over salt induced highs or deeper fault blocks), structural stratigraphic or purely stratigraphic. CO2 stores are more likely to be depleted hydrocarbon fields as these are the proven traps. Note: Some Middle and Lower Jurassic reservoirs may include a Triassic component.	Depleted hydrocarbon field	Most Likely CLOSED SYSTEM
Middle Jurassic		e.g. Brent Group and Beatrice Fm.			Delta system (Brent), marine barrier bar (Beatrice)
Lower Jurassic		e.g. Dunlin Group e.g. Statfjord Fm.			Wave influenced lower shoreface to offshore
Upper Triassic Middle Triassic	All areas	It is unlikely that the Mid to Upper Triassic would be utilised as a CO2 store in the near term (but see note above).
Lower Triassic	Southern North Sea	e.g. Bunter Sandstone Fm.	Sandstone	Saline water or methane	Fluvial	Interbedded mudstones	4-way dip closure over salt diapirs	Saline aquifer/depleted gas field
Late Permian	Southern North Sea	e.g. Rotliegend Leman Sst. Fm.	Sandstone. Most likely a depleted gas reserve	Methane	Aeolian	Evaporite	Hydrocarbon fields tend to be uplifted or inverted fault blocks top sealed by thick layers of evaporites.	Saline aquifer/depleted gas field