Screening CO2 storage options in The Netherlands

International Journal of Greenhouse Gas Control 4 (2010) 367–380

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc

Screening CO2 storage options in The Netherlands Andrea Ramı´rez a,*, Saskia Hagedoorn b, Leslie Kramers c, Ton Wildenborg c, Chris Hendriks b a

Utrecht University, Copernicus Institute, Group Science, Technology and Society, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands Ecofys Netherlands, B.V., P.O. Box 8408, 3503 RK Utrecht, The Netherlands c TNO Built Environment and Geosciences, P.O. Box 80015, 3508 TA Utrecht, The Netherlands b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 17 April 2009 Received in revised form 13 October 2009 Accepted 15 October 2009 Available online 27 November 2009

This paper describes the development and application of a methodology to screen and rank Dutch reservoirs suitable for long-term large scale CO2 storage. The screening focuses on off- and on-shore individual aquifers, gas and oil fields. In total 176 storage reservoirs have been taken into consideration: 138 gas fields, 4 oil fields and 34 aquifers, with a total theoretical storage potential of about 3200 Mt CO2. The reservoirs are screened according to three criteria: potential storage capacity, storage costs and effort needed to manage risk. Due to the large number of reservoirs, which limits the possibility to use any pair-wise comparison method (e.g. Multi-Criteria programs such as Bosda or Naiade), a spreadsheet tool was designed to provide an assessment of each of the criteria through an evaluation of the fields present in the database and a set of scores provided by a (inter)national panel of experts. The assessment is sufficiently simple and allows others to review it, re-do it or expand it. The results of the methodology show that plausible comparisons of prospective sites with limited characterization data are possible. ß 2009 Elsevier Ltd. All rights reserved.

Keywords: CO2 storage Screening Costs Risks

1. Introduction Carbon dioxide capture, transport and geological storage (CCS) is increasingly being considered as a significant greenhouse gas (GHG) mitigation option that could allow the continued use of fossil fuels while providing time to develop and deploy renewable energy at large scale. The 2008 Energy Technology Perspectives report of the International Energy Agency estimated a CCS potential of about 20% of global CO2 emissions in 2050 (IEA, 2008), with other studies showing similar ranges (e.g. IPCC, 2005). The significant role forecasted for CCS is based on three main assumptions:  CO2 capture technology will be available;  The option will be competitive (i.e., cost reductions will be achieved via learning and market conditions will be present) and,  Sufficient and suitable underground storage capacity will be available and accessible. Work on storage capacity evaluation has focused on two main areas: (i) assessments of total capacities per country or region; (ii) risk assessment and uncertainty analysis of CO2 storage. The first approach has resulted in inventories and global atlases, with an

* Corresponding author. Tel.: +31 30 2537639; fax: +31 30 2537601. E-mail address: [email protected] (A. Ramı´rez). 1750-5836/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijggc.2009.10.015

emphasis on storage capacities in developed countries. Such capacities are being used by modelers and policy makers to set the physical boundaries for national deployment of CCS. The second approach has resulted in the on-going development of methodological frameworks for risk assessments. Almost all frameworks are based on quantitative mathematical modeling (e.g., Wildenborg et al., 2003; Walton et al., 2004) and only a few semiquantitative assessments have been developed (e.g., Oldenburg, 2008; Bachu, 2002). The fully quantitative methods comprise major involvement of many experts, detailed knowledge of local geology and extensive modeling work. Their applicability for a large set of potential storage sites is therefore limited unless significant time and effort is committed to the outcome. Most information used by national policy makers on the possible role of CCS in a portfolio of mitigation measures is based on national estimations of CO2 storage capacities that, in most cases, assume all fields to be available and suitable. Risk factors associated with CO2 storage also influence the suitability of a reservoir and consequently have an effect on regional and national capacity estimates. It is important to take risk aspects into account as it will result in more realistic valuations of total CO2 storage potential. A simplified screening method for comparison of the portfolio of CO2 storage options that uses (publicly) available data would be helpful. However, until now few efforts have been made to develop such methods (e.g. Van Egmond, 2006; Meyer et al., 2006). The objective of this study is to develop a methodological framework to screen and rank CO2 reservoirs in The Netherlands.

368

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

Fig. 1. Schematic diagram of the methodology used in this research.

Such a methodological framework should enable integration of theoretical knowledge, expert knowledge and data of the storage locations in The Netherlands in a consistent systematic and transparent way.

uncertainty in their knowledge base. Results for the latter point are presented in Hagedoorn et al. (2008).

2. Methodology

In The Netherlands there are over 500 potential oil and gas fields and structural traps in aquifers that are potentially useful for storing CO2. The total number of options has been reduced by applying threshold values (Table 1). This pre-screening resulted in 176 storage locations which will be considered in this study. Of the

This project aims to develop a methodology that allows the screening of CO2 storage fields in The Netherlands. The screening of potential locations should deal with qualitative and quantitative data and parameters. An initial literature survey pointed out Multi-Criteria Analysis (MCA) as a methodological framework that could integrate these two aspects (e.g., Figueira et al., 2005; Scholz and Tietje, 2002; Gough and Shackley, 2006). However, the number of storage options to be compared in this research is large (>170) and limits the possibility to use any formal pair-wise comparison method (e.g. Multi-Criteria programs such as Bosda (Janssen et al., 2000) or Naiade (Munda, 1995)). We have therefore used the main aspects of MCA in a linear aggregation tool. The results are further assessed with a sensitivity analysis on weights and a detailed analysis of uncertainty, especially of the knowledge base. In this research, the screening of storage options is based on a set of three criteria: potential storage capacity, storage costs and effort needed to manage risk. Furthermore, the results will be analyzed taking into account availability (before and after 2020). Fig. 1 shows a schematic diagram of the methodology used and of the flows of information among the different steps. Note that inputs obtained from consultation rounds with an expert panel play a central role in the methodology. In the first expert consultation an (inter)national panel of experts was asked their opinions on the method, criteria, and indicators selected. The feedback of the expert evaluation resulted in changes in the original indicators. The second consultation focused on gathering the information needed to aggregate and score the values for the criteria ‘‘effort needed to manage risk’’. In this consultation round each expert was asked to provide scores for categories defined within each indicator, to define weights among the indicators so they could be aggregated in the screening tool and to quantify the

2.1. Identification of fields

Table 1 Thresholds used for the pre-screening of CO2 storage options in The Netherlands. Parameter

Threshold

Capacity Thickness reservoir Depth top reservoir Porosity reservoir Permeability reservoir

4 Mt for gas/oil and 2 Mt for aquifersa >10 m 800 m Aquifers: >10%b Aquifers: an expected permeability of 200 mD or moreb 10 m. Both simple seals as well as complex caprock have been taken into accountc Salt, anhydrite, shale or claystones Aquifers: sandstones, hydrocarbon fields: limestone, sandstone, siltstone, carbonates Overpressure areas excluded Relevant for aquifers. Traps located alongside/ near salt domes/walls have been excluded because there is a high risk of salt cementation

Thickness seal Seal composition Reservoir composition Initial pressure Salt domes

a In an early study (TNO, 2007) it is considered that these fields are not economically feasible for CO2 storage. b Previous studies indicate that aquifer properties for CO2 injection should have a porosity greater than 10% and a permeability in the order of 200 mD to prevent extreme overpressures near the injection well. As permeability measurements are scarcely entered in the national database especially for aquifers, the threshold of 200 mD is not strictly used here. Instead, we have studied lithology descriptions to find out which aquifers are most likely candidates for CO2 storage. In this sense, the selected aquifers are expected to have moderate to good permeability and porosity. c Simple caprocks are located directly on top of the aquifer, while sequences of claystones intercalated with carbonates, thin (<10 m) sandy layers, etc. are considered complex seals.

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

369

Table 2 Main equations used to calculate potential CO2 storage capacities. Type

Equation

Nomenclature

Gas fields

SCCO2 ¼ ðV URðstpÞ =Eg ÞrCO2

SCCO2 ¼ CO2 storage capacity (Mt) VUR(stp) = volume of ultimate gas recovered at standard conditions (15 8C, 1 bar) (109 m3) Eg = gas expansion factor rCO2 ¼ density of CO2 at reservoir conditions (kg/m3)

Oil fields

SCCO2 ¼ ðV oilðstpÞ =Boil ÞrCO2

Voil(stp) = volume Boil = formation volume factor rCO2 ¼ density of CO2 at reservoir conditions (kg/m3)

Aquifers

SCCO2 ¼ Vr N=G Ef FrCO2

Vr = bulk aquifer volume (m3) N/G = net to gross ratio Ef = efficiency factor (constant = 0.02) F = porosity rCO2 ¼ density of CO2 at reservoir conditions (kg/m3)

176 cases 138 are gas fields, 4 are oil fields and 34 are deep saline aquifers.1 2.2. Potential storage capacities The CO2 storage capacities used in this project have been estimated based on data and results of previous studies performed by TNO (e.g. TNO, 2002). In those studies, three different approaches were followed depending of the type of field (equations used to calculate theoretical storage capacities are shown in Table 2). In the case of depleted gas fields the main parameters used to estimate the CO2 storage capacity were the ultimate produced gas volume, the gas expansion factor and CO2 density. Values for ultimate gas recovery (UR) and historic cumulative production per individual field from before 2003 are confidential in The Netherlands, making it impossible to estimate UR figures based on historic production data, unless the operator provides the data. The volume of recovered gas per field was estimated from the total future production estimate and the ratio between areal extent per field and the total area of gas fields. The gas expansion factor and CO2 density are based on depth dependent relations (TNO, 2007). The main parameters to determine the CO2 storage capacity of a depleted oil field are the ultimate oil recovery volume, the formation volume factor2 and the CO2 density at reservoir conditions. For the calculations, it has been assumed that the amount of CO2 stored in the reservoir after enhanced oil recovery is approximately 30% of the oil initially in place. Since the ultimate recovery of most oil fields is in the order of 30–35%, the storage capacity can be approximated by the ultimate reserves. Furthermore, given the limited amount of oil fields meeting our criteria on minimum storage capacity (at least 4 Mt) and the relatively small storage capacity represented by those oil fields, CO2 density at reservoir conditions is set at 0.70 t/m3 and the formation volume factor at 1.2. The possible effects of water invasion are not accounted for in the storage capacity calculations of hydrocarbon fields.3 The assessment of storage capacity in aquifers focused on sandstones of Permian, Triassic, Upper Jurassic and Lower 1 The gas field ‘‘Slocheteren’’ with an estimated potential capacity of 7 Gt is excluded because according to its operators it will not be available for storage before 2050. 2 The formation volume factor brings the oil volume from standard conditions to in situ conditions. 3 The subsurface of the Netherlands is heavily faulted and production from gas and oil fields show that depletion is mainly the result of the reduction of pressure. Assuming that no water (or aquifer) support occurs seems valid as 85–90% of the Dutch fields are not water (or aquifer) supported.

Cretaceous age. Sandstone members, with small lateral extensions and limited local deposition, were not included in the assessment. In this study only structural traps are accounted for. This includes 4-way dip closures (anticline), compartments like horst and grabens, or dip-closures comparable to structural gas/oil traps.4 The areal extent of a potential structure is determined from the location of spill points from cross sections and elevation contour lines. Aquifer thickness is obtained from well data when present or, alternatively, from extrapolated thicknesses from a 3D model.5 Volumes are calculated by multiplying the areal extent with the aquifer thickness, assuming an average thickness based on well data or extrapolated thicknesses from the model. For the efficiency factor it was chosen to use a value of 2%.6 This is an assumed average value for structures in the faulted subsurface of The Netherlands where the CO2 injection, besides the size of the connected aquifer, depends on the water and rock compressibility. Average porosities and permeabilities of the aquifer were determined from well data on the specific stratigraphic unit within a 20–40 km radius from the structure midpoint. Note that this approach strongly depends on the availability of plug data within the radius and on which level a plug has been taken in the lithology.

4 The database contains a collection of all kinds of structural traps, like fault trapping, anticlinal traps, combinations of faults and anticlinal traps, pinchout, stratigraphic trapping, etc. These are the most common hydrocarbon traps in the Netherlands. Note that screening potential traps in aquifers regional maps were used. On this scale only fault bounded traps and anticlinal traps could be identified as potential storage locations. 5 A static geological 3D model in the Netherlands was constructed in Petrel (software which allows 3D static model building) from depth maps of Super Group and Group base levels, thickness maps of the sandstone and caprock of interest, and major or regional faults. 6 The value for the storage efficiency factor is still in a subject of (worldwide) debate. Pore space between the grains of an aquifer is filled with formation water. When injecting matter into an aquifer, water must be pushed aside to create a volume for, in this case, CO2. However, this enforced migration of water strongly depends on the properties of the formation from sub-pore to reservoir scale. The 2% factor used to compute the storage potential of a structural trap is based on reservoir simulations performed by TNO taking into account the Dutch geology and reservoir characteristics. Collected data from Dutch gas and oil fields indicate that many faults in the subsurface are sealing. Exploration maps from producing fields show that the production from most gas and oil fields ceases as a result of pressure depletion instead of water invasion. This holds for 85–90% of the existing fields. Assuming that this is applicable for the injection of CO2 in aquifer traps, the injection of CO2 into a structural trap can be compared with injecting into a closed container, resulting in lower efficiency factors (e.g., 0.5–1%). It is likely however that the trap is hydraulically connected to some (limited) part of the aquifer and therefore pore water can ‘‘escape’’ allowing the storage efficiency to increase to 2%, according to reservoir calculations. The 2% is an average value for a trap that is in pressure communication with a large aquifer.

370

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

Table 3 Cost figures used in this study.

Drilling costs per new well (euro/m) Cost of reusing an existing well (Meuro) Site development costs (Meuro)a Surface facilities (Meuro)b Monitoring costs (Meuro) O & M costs (% investment, cost per year)

Hydrocarbon fields onshore

Hydrocarbon fields offshore

Aquifers onshore

Aquifers offshore

2750 (depth < 3000 m) 3980 (depth > 3000 m) 1 3 1.5 1 5

3830 (depth < 3000 m) 5890 (depth > 3000 m) 2 3 15 1 5

2750 (depth < 3000 m) 3980 (depth > 3000 m) – 24 1.5 1.5 5

3830 (depth < 3000 m) 5890 (depth > 3000 m) – 24 61 1.5 5

a Aquifer site development costs are higher than those of hydrocarbon fields due to the exploration costs. A main reason is that data on the geological structure and reservoir properties of hydrocarbon fields are available, but are scarce for aquifers. b The surface facility costs for offshore aquifers are four times higher than those for offshore hydrocarbon fields because there are no platforms that can be reused.

Table 4 Average CO2 injection rate of hydrocarbon fields and saline aquifers per stratigraphic unit, in Mt/y per well.

Lower Cretaceous Group Lower Germanic Trias Group Niedersachsen Group Upper Rotliegend Group

Formation

Hydrocarbon field

Saline aquifer

Vlieland Sandstone Fm Lower Buntsandstein Fm Friese front Fm (sandstone members) Zechstein Fm (carbonate members) Slochteren Fm (sandstone members)

1 0.4 0.4 0.2 1 0.4

0.5 0.2 0.2 0.1 0.5 0.2

Limburg Group

2.3. Storage costs Table 3 shows an overview of the cost values used in this study. Site characterization and development include investigation costs (geo-characterization), preparation of drilling site and equipment, environmental impact assessments, engineering, licensing, and lease costs. The drilling costs depend on the number of wells needed, hence on the injection rates per well, possibility to use existing wells and project lifetime. Injection rates are sitedependent and as data are not available for each site, average injection rates were used that depend on the stratigraphic unit (Table 4). Drilling costs also depend on the depth and thickness of the reservoir. The costs of a well increase linearly with length of drilling up to a depth of about 3 km.7 After 3 km, costs increase exponentially with depth. Therefore, drilling costs are represented by two segments (<3000 and >3000 m). For simplicity the relation between costs and depth in each segment is taken as linear. This is partly justified because most reservoirs in The Netherlands are shallower than 4 km. Furthermore, it is assumed that a part of the current production wells can be used for CO2 injection. The actual percentage of reusable wells depends very much on site-specific factors. Based on a recent study (Van der Velde et al., 2008), it is assumed that two-thirds of the existing wells that are still in use today could be reused for CO2 injection and that wells that have already been plugged and abandoned already cannot be reused. To account for price changes since the publication of the studies used as sources of information, prices have been adapted using the Upstream Capital Cost Index (UCCI). Clearly, the costs as displayed in this study are rough estimates. They should mainly be regarded as relative numbers, distinguishing between various storage locations instead of absolute numbers. 2.4. Effort needed to manage risk This third indicator is a semi-quantitative proxy of the effort needed (material and personnel resources) to develop a safe and effective storage site. This means that in this research, we aim to 7

Directional drilling is not taken into account, because this parameter is very site-specific and because it would make no distinction between different storage options.

quantify the effort needed to manage risk, but not quantify the risk as such. Quantifying risk requires that both the cause and the effects of a possible event are assessed (risk = probability of an event  impact). The effort needed to manage risk, as defined here, focuses on reducing the likelihood of the hazardous event occurring and does not include measures needed to avoid or minimize escalation of the event into a larger consequence. For example, the presence of a large number of relatively old wells implies that more management effort will need to be done to increase the certainty of safe storage of CO2, since old wells represent a potential leakage path. Selecting reservoirs with the less possible number of old wells therefore reduces the effort needed to manage risk. Furthermore, in this study CO2 leakage is considered the main risk of CO2 storage in terms of likelihood of occurrence, magnitude, and impact. Therefore the effort needed to manage risk in this study focuses on managing/reducing CO2 leakage. CO2 leakage can occur in various ways. CO2 can migrate through the reservoir to areas were caprock is absent or weak. At places where the cap rock is not sufficiently sealing, CO2 can leak through the seal, either through the upper or lower bounding seal. When entering geological formations outside the storage reservoir, it is possible that the CO2 will be trapped by secondary seals, or that the rate of migration through the outside geological formations is so slow that CO2 storage is effectively permanent. A larger effort is needed to manage risk of leakage when there are pathways through which CO2 can reach the surface. Earlier studies have shown that those pathways are related to faults and fractures, wells and insufficiently sealing cap rock (e.g., Benson et al., 2002; Damen et al., 2006; Holloway, 1997; Tsang et al., 2007). A literature survey and consultation with the expert panel resulted in a selection of five leakage paths and twelve indicators that will be used here to rank potential storage sites (Table 5). The rationale behind the selection of each indicator and main methodological aspects for their determination are highlighted in the rest of this section. 2.4.1. Category: faults. Indicators: fault displacement, number of faults Many gas and oil reservoirs and potential aquifer traps in The Netherlands are bound or transected by one or more faults with a

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

371

Fig. 2. Example of the spreadsheet used in the consultation to experts to gather scoring values.

throw of 50 m or more (defined here as major faults). These faults may extend upwards to younger formations at shallow depth or rock towards the surface or to permeable layers in the overburden. Faults transecting geologically young formations are expected to pose a higher risk than faults transecting only older formations. The Netherlands is almost completely covered by Miocene and younger sedimentary rock, making the presence of a fault in these young sediments an important risk factor. TNO has mapped the main geological formations and faults of The Netherlands on a regional scale based on 2D and 3D seismic data. For each geological formation (Permian, Zechstein, Triassic, Jurassic, Lower and Upper Cretaceous, Tertiary and Base

Table 5 Indicators and paths used to screen storage options according to the effort needed to manage risk of leakage. Leakage path

Indicators used

Leakage through faults

Fault displacement in or above reservoir Number of major faults in or above reservoir

Leakage because of natural (or induced) seismicity

Seismicity zone

Leakage through wells

Wells abandoned before 1967 Wells abandoned between 1967 and 1976 Wells abandoned after 1976 Wells not abandoned Accessibility of wells

Leakage through the seal

Thickness of primary seal Seal composition Proven sealing capability

Leakage through the overburden

Reservoir depth

Miocene) group fault maps have been compiled.8 By plotting all fault maps and outlines of hydrocarbon reservoirs and aquifers on top of each other, fault displacement can be visualized.9 With this approach it is possible to quantify the number of (major) faults that confine a potential storage site and determine to which formation a fault extends. The number of major faults per storage field is considered therefore an indicator of the effort needed for risk management e.g., characterization, pressure tests. It is important to note that offshore faults are currently less well mapped and therefore an underestimation of the number of offshore faults is possible. 2.4.2. Category: natural seismicity. Indicator: seismicity type Together with fault maps, well data and exploration maps, the onshore Netherlands has been subdivided in ‘‘stable,’’ ‘‘slightly unstable,’’ and ‘‘unstable’’ tectonic zones. Tectonically unstable regions increase the effort needed to manage risk. In the case that there is evidence of several faults transecting Pleistocene rocks, a zone is defined as slightly unstable. A zone with more recent seismic activity is classified as unstable. A zone is stable when seismic activity has not occurred in the Pleistocene. A similar, but somewhat less detailed approach was 8 Fault maps show faults with a throw of 50 m or more. Faults with a throw of less than 50 m are not mapped and for this reason not taken into account in our database. When the throw of a fault is large enough (e.g. larger than the total thickness of the cap rock) the storage reservoir could be truncated against one or more porous layer(s) in the overburden, posing a relatively larger risk of leakage. It should be noted, however, that faults with less throw could still pose a hazard as they could act as pathways for gas/fluids to migrate, for instance through the cap rock. It strongly depends on whether a fault is impermeable. 9 Faults with a displacement of at least 50 m (visible on seismic reflections lines) were mapped. When displacement decreases (at the tip of a fault) it may not be visible any more on seismics. In these cases, we extrapolated the interpret fault lines to the next horizon.

372

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

performed for this study to take offshore seismicity into account, since this data were not included in the primary database. In order to do so we took into account mapped faults, literature review, seismicity measures and knowledge from experts from the Dutch National Center for Weather, Climate and Seismology (KNMI) and measured seismicity. The results showed that the Dutch North Sea area is a fairly stable seismic area with the exception of an area known as the Dogger Bank, which is marked as unstable (in 1930 an earthquake occurred here with a magnitude of 6.0 on the Richter scale).

2.4.3. Category: wells. Indicators: wells drilled before 1967, between 1967 and 1976, and after 1976, well status, well accessibility Several wells may have been drilled in gas and oil fields for exploration appraisal, water injection, or production. Some of these wells are abandoned. Degradation of cement plugs or inadequate plugging of abandoned wells may pose a risk of CO2 leakage. In The Netherlands legislation on cementing and plugging of wells was not in place before 1967; thus the degradation risk is assumed to be higher for wells drilled and abandoned before 1967. The legislation enforced in 1967 was amended in 1976, which

Fig. 3. (A) Map of potential storage locations by CO2 capacity of storage, type of field. (B) Frequency distribution of potential storage capacities.

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

Fig. 3. (Continued ).

again led to higher quality plugs and cementation. In this study, it is assumed that a well drilled for exploration or production should fulfill the regulations defined at that moment. Assuming that the well has not been worked over since, the completion date is used to group the wells according to the periods when legislation changed (1967, 1967–1976 and after 1976). To obtain the number of wells by period that are abandoned and the number of wells that are still in use (wells not abandoned) a compilation had to be made of three different types of data namely:  End depth. To select the wells that pierce a reservoir or aquifer, field outlines from GIS maps compiled by TNO in the past and coordinates of the end point of a well from the existing (DINO) database10 were used. It was decided to use the end point of a well as subsurface coordinates may be located outside a field outline, which does not necessarily mean that a well does not pierce a reservoir (deviate drilling).  Age of the well. The project was faced with a practical problem as specific information on the technical status of the well is not available in a useful form. Gathering data of the exact abandonment date, possible work over of a well and total completion (casing and cemented) would take significant effort as it concerns about 5000 well reports, which is beyond the scope of the current project. However, as the completion date of a well is available from the DINO database, this date is used instead as an indication for the abandonment date of a well. The main drawback of this assumption is that it does not take into account possible degradation of the cement after completion, i.e., even though a well could have been completed in a time when high standards were required, a combination of chemical and geomechanical effects can degrade the cement so that, from the perspective of leakage potential, the well would pose a risk similar to a well completed decades earlier.  Well status. A well can consist of several boreholes. Based on the end depth of the primary well or sidetrack and the top depth of the reservoir, each part of the well system is evaluated. Wells that have the status closed, suspended, 10 DINO stands for Data and Information on the Dutch Subsurface. For more information see: http://www.dinoloket.nl/en/DINOLoket.html. 11 The NLOG website (http://www.nlog.nl/nl/home/NLOGPortal.html) provides information about oil and gas exploration and production in the Netherlands and the Dutch sector of the North Sea continental shelf. 12 In the Netherlands, when data is delivered by an oil or gas company the end result of a well must be clearly stated. In case a well is drilled into a reservoir that contains oil or gas, but is not economically valuable, a ‘gas shows’ or ‘oil shows’ status is assigned to it. These ones are not included in the category ‘dry’.

373

producing, or water injection are considered to be accessible and can be worked over before CO2 is injected into a reservoir. Wells with status Plugged and Abandoned (PAA) and FAS (failure and sidetracks) have been plugged with cement. These wells can potentially function as a bypass to the surface in case cement plugs and casings have been disintegrated or corroded over time. When a technical failure occurs part of a well will be plugged and a sidetrack can be drilled. It is possible that the sidetrack also fails and a new sidetrack will be drilled. When a well with a status PAA or FAS ends in the overburden (i.e., depth less than cap rock) it is not considered a risk for leakage in this study. When there are indications that a well or one of its sidetracks or both, are drilled in the caprock (but not necessarily in the reservoir itself) it is considered a potential risk. Hence, the number of wells considered equals the number of well sections drilled through the caprock and reservoir. Data from wells with construction year more recent than 2003 are still confidential. These wells have the status ‘‘unknown’’. It is assumed that these wells are used for production and are still accessible. In case well status is classified as unknown for a well older than 2003 the NLOG-website11 and production plans were consulted. In the event that no data were available for a well, which was completed before 2003 and could not be traced by any other means, the well was assumed to be abandoned. Wells near a hydrocarbon field that have a status ‘‘Dry’’ were not considered.12 It is assumed that these well sections missed the gas or oil reservoir. The accessibility of wells was determined by plotting the outlines of hydrocarbon fields, well data, and a topographic map of The Netherlands. Each field is quantitatively evaluated based on the number of wells that are located in an urban or rural area. In general, offshore abandoned wells are harder to locate and therefore more difficult to access, as they are often covered by sediments. Onshore abandoned wells may also be hard to access due to, for instance, buildings in urban areas. For onshore fields, wells in a populated area, such as a city, were classified as urban. Wells that are located in open areas (national parks, agricultural land) were classified as rural. 2.4.4. Category: caprock. Indicators: caprock thickness, caprock composition, proven sealing capability Gas and oil reservoirs in The Netherlands are mostly trapped by a caprock at least 10 m thick. CO2 can leak through the caprock, for example through fractures. The effort needed to manage risk is reduced when the cap rock is thicker. For most fields, production plans were used to find out which geological members act as seals. When a production plan was not available the first geological member on top of the reservoir rock was considered as caprock. Gas fields are predominantly fault-confined, i.e., gas is trapped by faults or lateral seals. We considered only top seals, because lateral seals of a reservoir are not reported in production plans. However, lateral seals have commonly the same composition as the top seal. Besides thickness, sealing performance depends on permeability of the rock, which is related to its composition. Caprock composition is determined from production plans and stratigraphy descriptions. In our database a caprock is classified as salt, shale,13 claystone, anhydrite, or marl. As explained before, the caprock can 13 It is necessary to elaborate on the definition of shale used in our database. In the petroleum industry, shale is defined as a rock member that is composed of sandy silt and clay (sandy or silty clay). In geology, shale is a fine-grained sedimentary rock whose original constituents were clay minerals or muds and has hardened due to (heavy) compaction. By this latter definition shale is considered to have better sealing properties than claystone, which is less compacted. In this study we used the geology definition.

374

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

also be a mixture of these components. Due to the regional scale inventory in this study, homogeneity of the rocks cannot be proven explicitly. This means that when the caprock is a mixture of components, the composition in the database refers to the component most present. The risk management effort decreases when the composition of the seal contains on average rocks which are more impermeable. Sandstone, siltstone, carbonate/limestone and dolomite are not regarded as a seal at all. Finally, the indicator proven sealing capability is based on the type of storage location and

the evidence of having trapped a gas or fluid before. The following classification has been used: field evidence of gas; field evidence of oil; no field evidence. Former gas fields have proven that these structures can sufficiently trap gas. Oil fields on the other hand have proven to trap a viscous fluid, but not necessarily gas, which increases the risk management effort. Structures that do not contain any hydrocarbon have no proof of the ability to trap CO2. This is also the case for aquifers.

Fig. 4. (A) Map of potential storage locations by CO2 costs of storage and type of field. (B) Frequency distribution of potential storage cots.

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

375

Table 6 Average weights (over the experts) used in this study. Leakage path

Indicators

Average weight (%)

Leakage through faults

Fault displacement in or above reservoir Number of major faults

12

Leakage because of natural (or induced) seismicity

Seismicity zone

Leakage through wells

Wells abandoned before 1967 Wells abandoned between 1967 and 1976 Wells abandoned after 1976 Wells not abandoned Accessibility of wells

Fig. 4. (Continued ).

2.4.5. Category: reservoir depth. Indicator: reservoir depth Reservoir depth, from the top of the reservoir to the surface (overburden thickness), is considered an indicator of the presence of secondary seals above the storage reservoir. A typical overburden in The Netherlands consists of an alteration of aquifers and aquitards. Aquitards could serve as a secondary seal. The overburden may reduce the risk when unexpected leakage through the top seal or a fault occurs. Impermeable rock in the overburden may block or slow down the upward migration of CO2. When CO2 is sufficiently slowed down by impermeable layers in the overburden, mineral or solubility trapping may lower the likelihood of CO2 leakage to the surface. From the stratigraphy data that were collected to determine the reservoir and caprock thickness, the total average overburden thickness was determined. 2.4.6. Additional data  Reservoir thickness. Reservoir thickness was not classified as an indicator, but is used for the computation of costs as it determines the length of a well. The reservoir thickness entered into the database states the gross thickness because the net thickness is unknown for most hydrocarbon fields. Based on the stratigraphy from well interpretations the true vertical thickness of a reservoir rock can be determined. When several wells are drilled in a reservoir the average thickness was computed.  Estimated year of depletion. These data are collected from the production plans. For this exercise the most recent plans were used (dating from 2003 to 2005). The estimated depletion year of a hydrocarbon field depends on, for instance, the oil or gas price, unexpected production of more or less gas or oil, and geological effects that may force production to stop (e.g., production of undesirable formation water) and may result in a revision of the depletion year in future reports. 2.5. Development of the screening tool To develop a relative ranking of potential storage options, it is necessary to aggregate and score the values provided by the different criteria and indicators. The output for the criteria storage capacity and costs per storage field will consist of two values expressed in Mt and euro/t CO2, respectively. Ranking the options according to storage capacity or costs is therefore straightforward. The criterion effort needed to manage risk, on the other hand, is made up of qualitative and quantitative indicators. Each indicator has been divided into categories. The categories are set in quantitative (e.g. categories for tectonic seismic risk are >7, 6–7, 4–6, 2–4, <2) or in qualitative terms (e.g. categories for the primary seal composition are Carbonate, Anhydrite, Clay, Shale, Halite). To obtain a ranking of the

7 6

12 10 9 7 10

Leakage through the seal

Seal thickness Seal composition Proven sealing capability

7 8 7

Leakage through the overburden

Reservoir depth

5

reservoirs it is necessary to translate the categories into scores. These scores reflect the relative importance of the categories within a given indicator. Then, in a second step, the scores of the different indicators for each field are summed up into a unique value. Since all indicators may not be equally important, it is necessary to assign weights that reflect the relative importance of the different indicators in terms of risk management effort. A second consultation with a (inter)national panel of eight experts was performed to obtain the scoring and weighting values. The consultation made use of a spreadsheet tool containing three types of entries: scoring categories, evaluation of the knowledge base and weights for the indicators. As an example Fig. 2 shows how data were gathered in the first entry type. The input generated in the tool is then used to calculate an average score per site. The average weights used to obtain the average score for the criterion are shown in Table 6. Note that the experts considered the relative importance of the indicators to be at the same level. The basic calculation is a simple linear aggregation using the scores and weights between categories and indicators. The resulting scores per site from the assessment are representative for the relative scoring and do not indicate an absolute site performance. 3. Results Potential storage capacity. The total potential capacity of CO2 storage in The Netherlands is calculated at 3.2 Gt (the Slochteren gas field which has a potential of around 7 Gt is not included since it is not expected to be available for CO2 storage before 2050). Fig. 3 shows the screening of the fields according to capacity. The results show that about 52% of the storage locations (accounting for roughly 20% of the total capacity) is composed of relatively small fields (<10 Mt). Fields with larger capacities are gas reservoirs, while oil fields and aquifers have lower capacities. However, the methodology used to calculate storage capacities of deep aquifers is considered preliminary. Since there was no commercial interest in these fields until a few years ago, the data available are limited. Several projects in The Netherlands are being carried out to address storage capacity estimations, and it is expected that better data and approaches for calculation of storage capacities will be made available in the next 2 years. In terms of time availability, 70% of the storage locations will become available before 2020, representing 74% of the total

376

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

storage capacity. Roughly half of those storage locations are already available (before 2010); these fields represent roughly 25% of the total storage capacity. Two-thirds of the storage locations that are available before 2010 are fields smaller than 10 Mt, which explains the difference between the amount of storage locations and storage capacity of the ‘‘early fields’’. CO2 storage costs. Fig. 4 shows the screening by location and a frequency distribution according to the costs of CO2 storage. Around 39% of the fields (representing 68% of the total potential storage capacity) have estimated storage costs lower than 3 s

per tonne of CO2. These fields are mainly onshore gas fields (about 70%). Most offshore gas fields show prices in the ranges of 3–14 s per tonne of CO2. Aquifers on the other hand show the highest prices (17–107 s per tonne of CO2), being located at the right tail of the histogram in Fig. 4B. This is not surprising since storing CO2 in these fields will imply not only that complete seismic studies have to be done from scratch but also that there is not any infrastructure that can be reused (e.g., wells or platforms). It is important to point out that less than 10% of the fields that become available before 2010 have estimated costs of

Fig. 5. (A) Screening of potential storage locations by effort needed to manage risk. (B) Frequency distribution of scores.

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

377

will be available before 2020. It is important to highlight that most aquifers are located in this category (about 83%), the rest is located in category B.

Fig. 5. (Continued ).

lower than 3 s per tonne of CO2. The storage locations that become available before 2020 have a higher price range because these include most of the aquifers. The percentage of storage locations with estimated costs of less than 3 s per tonne of CO2 increases to 25% when storage locations are considered that become available before 2020. Effort needed to manage risk. For this criterion the results are presented according to categories, instead of scores, as they are more appropriate given the character of the data and the scores.14 The three categories used are (the average highest score was 94 and the lowest 26):  Category A: composed of fields with the highest scores (82). These fields relatively require the least effort to manage risk. It corresponds to the third interquartile of the sample.  Category B: intermediate.  Category C: composed of fields with the lowest scores (<70). It corresponds to the first interquartile of the sample.

The results (see Figs. 5 and 6) indicate that:  43 fields score in Category A. These are gas fields representing a potential storage capacity of 541 Mt (about 17% of the total potential capacity). 54% of the total capacity in this category is made up of fields with potential storage capacities smaller than 15 Mt (30% are smaller than 10 Mt, though no fields have a capacity lower than 5 Mt). Larger fields (e.g., >25 Mt) account for only 16% of the total capacity. Of the total potential capacity, about 60% is available before 2020. In terms of location (onshore vs. offshore), 60% of the fields located in this category are onshore.  Most fields (83) are found in category B. This category is composed by a mix of gas fields, oil and some aquifers. The potential storage capacity amounts to 1837 Mt (58% of the total potential capacity). In this category larger fields (>40 Mt) account for about 40% of the total capacity. 68% of the capacity in this category will be available before 2020.  The rest of the fields (50) are found in category C. Their potential storage capacity amounts to 801 Mt. 45% of the fields in this category have categories larger than 40 Mt. Of the 801 Mt, 96% 14 Given the current assumptions used, the uncertainty in the data and in the knowledge base, it cannot be stated that a field located in the ranking position as 24 is indeed better than field located at position 25 or worse than field located at number 23. What can be assured is that fields in positions 23, 24 or 25 require fewer efforts to manage risks than those in positions 100, 160 or 173.

The results can also be analyzed by looking at the categories and the costs of storage. Frequency distributions of the costs by category are depicted in Fig. 7. Note that the width of the histograms increases from category A to C. Although the costs, as calculated in this study, do not account for the costs that would have to be made in order to minimize risk of leakage (which are larger for category C than A), they do reflect (some of) the characteristics of the fields in the categories. For instance, as explained before, the large majority of the aquifers fall under category C. Storing in aquifers require new platforms and wells (which is not always the case with hydrocarbon fields) as well as complete seismic studies (which have been done for hydrocarbon fields) and are therefore on average more expensive. Also, a relative larger number of smaller fields are located in category C (compared to A and B). These fields are more expensive per tonne of CO2 stored. 4. Discussion This research aimed to develop a methodology to screen storage fields that used publicly available data. The results of such an exercise provide insights into the potential capacity of The Netherlands once considerations on geological characteristics that could limit the potential leakage of CO2 are taken into account. This kind of information could allow a better understanding of the feasibility of CO2 deployment at large scale under different scenarios (e.g., what would be the impact of a given CO2 implementation path if all fields with lower scores for risk management were not considered?). The results are however not to be used as a selection guide for specific storage places. Before a potential storage location can be marked as sufficiently safe, local variations in lithology of both the reservoir (aquifer) and caprock should be examined by site-specific feasibility studies. Such a feasibility study should address, among other things, the local stress at faults (geo-mechanical state), small-scale faulting and heterogeneity of both the cap and reservoir rock, reservoir simulations, and variations in caprock thickness. Geological variance will become clear during such a study. An additional result of this project was the development of a detailed database where geological information at a field level was gathered in a systematic and consistent way. Such a database can be further improved, especially regarding aquifers and offshore fields. Furthermore, while in The Netherlands, oil and gas companies are obliged to supply data of their hydrocarbon fields15 (e.g., permeability, porosity, stratigraphy, pressure measurements, and temperature), availability of this data depends on confidentiality, the number and quality of measurements performed during drilling, and processing of the data by the DINO database. It is therefore important to develop ways in which the data can be used for, for instance, potential CO2 storage calculations or screening exercises of the type made in this research. Further, it is recognized that the results of the project may be biased in the sense that the confidence level is going to be biased towards the sites at high risk than the ones that show low risk (i.e., absence of proof is not proof of absence). Unfortunately this kind of issue can only be solved by gathering extra data that would allow fields to be assessed under ‘‘equal’’ conditions. 15 Also data from exploration wells that did not encounter gas or oil have to be provided.

378

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

Fig. 6. Locations of fields according to effort needed to manage risk, storage costs and time availability.

Another area for further development is the (possible) dependence among variables. In this study the indicators used were taken as independent. Therefore it is possible that some of the effects are under or over estimated. Also it should be noted that the methodology for the development of the indicators, scores and weights for the criteria ‘‘effort needed to manage risk’’ were based on expert elicitation. CO2 storage is a relatively new field with relatively large uncertainties due to lack of data and scientific controversy. The consultation rounds allowed us to make explicit the knowledge of experts, based on accumulated experience and expertise, including their insights into the

limitations, strengths, and weakness of the published knowledge and available data. The relations between categories within the indicators and their scores need therefore to be corroborated. This is only possible by carrying out experimental work. Finally, there is a relation between the effort needed to manage risk and costs (a high risk level usually results in higher costs to bring down the risk again to acceptable levels). In this research we considered the dependency between both criteria to be low and therefore they are treated as independent. This point should be explored in further research.

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

379

Fig. 7. Frequency distributions of fields according to effort needed to manage risk (category A, B or C), and storage costs.

5. Conclusion This project aimed to develop a methodology that allowed the screening of CO2 storage options in The Netherlands. In this study three criteria have been used: potential storage capacity, costs of storage, and effort needed to manage risk. The methodology provides quantitative results for the first two criteria and is qualitative with respect to risk. It uses geological proxies for understanding the level of effort that would have to be made in order to manage a safe storage place. For the development of the methodology a detailed database for 176 storage locations has been developed. This allowed us to assure consistency in data compilation and interpretation, to the extent that this is possible given the variability in the extent and quality of geological and geophysical data. A spreadsheet was designed to provide an assessment of each of the criteria through an evaluation of the fields present in the database and a set of scores provided by a (inter)national panel of experts. The assessment is sufficiently simple and allows others to review it, re-do it or expand it. The results of the methodology show that plausible comparisons of prospective sites with limited characterization data are possible. Several extensions to the methodology are possible. First, more detailed data could be used, especially concerning offshore fields and aquifers. Second, further knowledge on the influence of a parameter on the risk of CO2 leakage and on how it interacts with other parameters/indicators could be gathered. This knowledge could then be reflected in new data, indicators, or criteria being added into the tool. Third, by experimental verification of the relations between geological characteristics and scores for risk management effort (best, worst and intermediate categories) within a given indicator could be attempted. In this paper, these relations are based on the knowledge of the experts consulted which in many cases are based on analogous situations. Fourth, the methodology could be extended by including multiple dependencies among the data and indicators and by including additional risk drivers (e.g., displacement of brine into potable water). Finally, data uncertainty could be explicitly included in the methodology.

Though the results of this study are not to replace detailed risk assessments that would need to be carried out for specific site selection, they can help policy and decision makers to understand the influence of several criteria in the suitability and availability of potential storage locations. For instance, our results indicate that 25% of the theoretical potential storage capacity in The Netherlands falls under the category with the lowest scores regarding the criterion effort needed to manage risk. Since 96% of these fields will be available before 2020, the effects of not using these fields for the (short term) national strategy on CO2 storage should be examined in detail. Acknowledgements For their feedback and input into this project, the authors are deeply grateful to John Bradshaw (Greenhouse Gas Storage Solutions), Marjeta Car (Geoinzeniring), John Gale (IEA-GHG), Sam Holloway (British Geological Survey), Jerome Le Gouvec (OXAND), Stefan Bachu (Alberta Energy and Utilities Board), Margriet Kuijper (NAM), Isabel Czernicchowski-Lauriol (BRGM), Bill Gunter (Alberta Research Council), C.R. Geel, T. Benedictus, A.N.M. Obdam and E. Simmelink (TNO). This project is part of the National Programme on Carbon Capture and Storage in The Netherlands (CATO). This programme is financed by The Netherlands Ministry of Economic Affairs. For more information see: http://www.CO2-CATO.nl. References Bachu, S., 2002. Sequestration of CO2 in geological media in response to climate change: roadmap for site selection using the transform of the geological space into the CO2-phase space. Energy Conversion and Management 43 (1), 87–102. Benson, S.M., Hepple, R., Apps, J., Tsang, C.F., Lippmann, M., 2002. Lessons Learned from Natural and Industrial Analogues for Storage of Carbon Dioxide in Deep Geological Formations. Earth Sciences Division, E.O. Lawrence Berkeley National Laboratory, Berkeley, p. 135. Damen, K., Faaij, A., Turkenburg, W., 2006. Health, safety and environmental risks of underground CO2 storage—overview of mechanisms and current knowledge. Journal Climatic Change 74 (1–3), 289–318. Figueira, J., Salvatore, G., Ehrgott, M. (Eds.), 2005. Multiple Criteria Decision Analysis: State of the Art Surveys. Springer, Berlin, Heidelberg, New York, p. 1045.

380

A. Ramı´rez et al. / International Journal of Greenhouse Gas Control 4 (2010) 367–380

Gough C., Shackley S., 2006. Towards a Multi-Criteria Methodology for Assessment of Geological Carbon Storage Options, vol. 74, issue: 1–3 (January 1, 2006), pp. 141–174. Hagedoorn S., Ramirez A., Kramers L., Hendriks C., Wildenborg T., 2008. Pre-screening and ranking Dutch CO2 storage options, Ecofys EEP03017-WP1.3, The Netherlands. Holloway, S., 1997. Safety of underground disposal of carbon dioxide. Energy Conversion and Management 38, S241–S245. International Energy Agency, 2008. Energy Technology Analysis. CO2 capture and storage. A key abatement option. OECD/IEA, Paris. Intergovernmental Panel on Climate Change IPPC, 2005. In: Benson, S., Cook, P. (Eds.), Special Report on Carbon Dioxide Capture and Storage, Chapter 5 Underground Geological Storage. Janssen, R., van Herwijnen, M., Beinat, E., 2000. Bosda voor Windows. Beslissingsondersteuning systeem voor discreet alternativen. SDU Uitgevers, Den haag, The Netherlands, 431 pp. (in Dutch). Meyer, R., May, F., Muller, C., Geel, K., Bernstone, C., 2006. Regional research, selection and geological characterization of a large anticlinat structure, as a candidate site for CO2 storage in northern Germany. Environmental Geology 2006. Munda, G., 1995. Multicriteria Evaluation in a Fuzzy Environment—Theory and Applications in Ecological Economics, Contributions to Economics Series. Physica-Verlag, Heidelberg. Oldenburg, C.M., 2008. Screening and ranking framework for geologic CO2 storage site selection on the basis of health, safety, and environmental risk. Environmental Geology 54, 1687–1694.

Scholz, R.W., Tietje, O., 2002. Embedded Case Study Methods: Integrating Quantitative and Qualitative Knowledge. Sage, Thousand Oaks. TNO, 2002. GESTCO: storage capacity and quality of hydrocarbon structures in the North Sea and the Aegean region. TNO Report NITG 02-020-B, The Netherlands. TNO, 2007. Options for CO2 Storage in the Netherlands—Time Dependent Storage Capacity, Hazard Aspects and Regulations. TNO, The Netherlands. Tsang, C., Lewicki, J.L., Birkholzer, J., 2007. Natural and industrial analogues for leakage of CO2 from storage reservoirs: identification of features, events, and processes and lessons learned. Environmental Geology 52 (3), 457–467. Van der Velde, R., Mieog, J., Breunese, J., Remmelts, G., 2008. Potential for CO2 storage in depleted gas fields on the Dutch continental shelf. NOGEPA Report, 2008. Available online at: http://www.nogepa.nl. Van Egmond, B., 2006. Developing a method to screen and rank geological CO2 storage sites on the risk of leakage. NWS-E-2006-108, Utrecht University and TNO. Walton, F.B., Tait, J.C., LeNeveu, D., Sheppard, M.I., 2004. Geological storage of CO2: a statistical approach to assessing performance and risk. In: Rubin, E.S., Keith, D.W., Gilboy, C.F. (Eds.), Seventh International Conference on Greenhouse Gas Control Technologies, Cheltenham, UK. Wildenborg, A.F.B., Leijnse, A.L., Kreft, E., et al., 2003. CO2 Capture Project—An Integrated, Collaborative Technology Development Project for Next Generation CO2 Separation, Capture and Geologic Sequestration Safety Assessment Methodology for CO2 Sequestration (SAMCARDS) TNO-NITG, Report nr. 2.1.1, Utrecht.