Search

Seismic time-lapse (4D) imaging has been considered as a key solution to monitor CO2 reservoirs. However, traditionally this technology requires dense data acquisition to produce high-resolution images. It is anticipated that monitoring will be required for more than 50 years after CCS operations cease and the monitoring phase is profit-negative. Developing cheaper 4D seismic imaging techniques is necessary. Historical knowledge of the subsurface structure in and near abandoned hydrocarbon fields, could reduce the dense data requirement of 4D imaging.Here we present preliminary results of 4D seismic (image-domain) wavefield tomography (IDWT) using pre-stack gathers from a sparse monitoring acquisition. IDWT uses short-offset data to exploit primarily kinematic changes rather than amplitude changes. IDWT minimises the shift between baseline and monitor migrations by optimising the monitor velocity model. Pre-stack IDWT, unlike post-stack methods, can use individual shot gathers to calculate the migration images. This property is beneficial when using sparse data acquisition permitting reliable measurement of shifts between monitor and baseline. Knowing the structure of the subsurface, we can design sparse acquisition surveys for seismic deployment, to minimize uncertainty in target areas. We create synthetic models based on Tyra gas field, a prospective future repository of CO2 in the Danish sector of North Sea and simulate CCS and subsequent leakage scenarios. The presence of CO2 in the reservoir, as well as the effect of reservoir pressure on the overburden stress-state, changes the seismic velocity structure of the reservoir and the overburden. These velocity changes cause an apparent depth (or time) shift when migrating the data.

In this paper we present a new code to build geologically realistic models of natural fracture networks in geological formations, by simulating the processes of fracture nucleation, growth and interaction, based on geomechanical principles and the geological history of the formation. This code implements the fracture modelling algorithm described in Welch et al. (2020), developed to generate more accurate, better constrained models of large fracture networks than current stochastic techniques. It can efficiently build either implicit fracture models, explicit DFNs, or both, across large (km-scale) geological structures such as folds, major faults or salt diapirs. It will thus have applications in engineering and fluid flow modelling, including CO2 sequestration and geothermal energy, as well as in understanding the controls on the evolution of fracture networks. The code is written in C Sharp and is provided with two interfaces: a standalone interface with text file input and output, that can be compiled in standard C Sharp and can run simple models, and a plug-in interface for the Petrel geomodelling package from Schlumberger, that can run more complex models of real geological structures. The standalone version has been used to run extensive sensitivity analyses, which studied the influence of various mechanical and physical parameters (e.g. layer thickness, applied strain, Young’s Modulus, etc.) on the fracture evolution and geometry, by varying the parameters individually in simple models. The Petrel plug-in has been used to evaluate the code applicability by running simulations of actual fractured layers in outcrops and in the subsurface, and comparing the results with observed fracture patterns.

We combine a power-law microfracture size distribution function with an expression for fracture propagation rate derived from subcritical fracture propagation theory and linear elastic fracture mechanics, to derive a geomechanically based deterministic model for the growth of a network of layer-bound fractures. This model also simulates fracture termination due to intersection with perpendicular fractures or stress-shadow interaction. We use this model to examine key controls on the emergent geometry of the fracture network. First, we examine the effect of fracture propagation rates. We show that at subcritical fracture propagation rates, the fracture nucleation rate increases with time; this generates a very dense network of very small fractures, similar to the deformation bands generated by compaction in unconsolidated sediments. By contrast, at critical propagation rates, the fracture nucleation rate decreases with time; this generates fewer but much larger fractures, similar to the brittle open fractures generated by tectonic deformation in lithified sediments. We then examine the controls on the rate of growth of the fracture network. A fracture set will start to grow when the stress acting on it reaches a threshold value, and it will continue to grow until all the fractures have stopped propagating and no new fractures can nucleate. The relative timing and rate of growth of the different fracture sets will control the anisotropy of the resulting fracture network: if the sets start to grow at the same time and rate, the result is a fully isotropic fracture network; if the primary fracture set stops growing before the secondary set starts growing, the result is a fully anisotropic fracture network; and if there is some overlap but the secondary set grows more slowly than the primary set, the result is a partially anisotropic fracture network. Although the applied horizontal strain rates are the key control on the relative growth rates of the two fracture sets, we show that the vertical effective stress, the initial horizontal stress, the elastic properties of the rock and the inelastic deformation processes, such as creep, grain sliding and pressure solution, all exert a control on the fracture growth rates, and that more isotropic fracture networks will tend to develop if the vertical effective stress is low or if the fractures are critically stressed prior to the onset of deformation. Thematic collection: This article is part of the Naturally Fractured Reservoirs collection available at: https://www.lyellcollection.org/cc/naturally-fractured-reservoirs

The past two decades of research on Carbon Capture and Storage (CCS) seem to have finally become fruitful as global leaders and energy-intensive industries are cooperating to materialize CCS projects and reach the promised reduction in CO2 emissions. Traditionally, CCS projects targeted mostly high permeability sandstone formations, despite the numerous carbonate fields undergoing CO2 injection for Enhanced Oil Recovery (EOR) in the United States or Canada. Because of the reactivity between calcite minerals and CO2 saturated water, chalk formations, characterized by high porosity and low permeability, have been previously portrayed as infeasible CO2 storage sites. Although previous laboratory investigations were carried out to assess the performance of CO2-EOR in North Sea chalk fields, these studies did not result in any field-scale demonstration projects; this may soon change since a positive movement towards CO2 storage in depleted oil fields has been recently initiated. In this work, we reviewed existing studies on CO2 injection in chalk to address the suitability of this type of formation for CCS. Although the evidence on the thermo-hydro-mechanical-chemical behaviour of chalk in the presence of CO2-saturated aqueous solutions is mixed, the majority of flooding tests performed on reservoir core samples do not support further weakening relative to water injection conditions nor significant changes in the petrophysical properties. Along with the weakening effect and using the Danish North Sea chalk fields as a case study, we addressed events that impact the storage site safety such as fault reactivation, and caprock and well integrity. Furthermore, monitoring techniques relevant to offshore locations are also discussed. Based on studies on other types of carbonates, and considering the characteristics of chalk (e.g., permeability, wettability, and reactivity) we analysed the relevance of different trapping mechanisms (i.e., solution, capillary, and mineral) but also several effects (i.e., chemical, biological, mechanical) that can lead to loss of injectivity. The main observations and conclusions in this work can be easily extrapolated to other chalk formations worldwide.

Large scale modelling of fractured reservoirs is a persistent problem in representing fluid flow in the subsurface. Considering a geothermal energy prospect beneath the Drenthe Aa area, we demonstrate application of a recently developed approach to efficiently predict fracture network geometry across an area of several square kilometres.Using a strain based method to mechanically model fracture nucleation and propagation, we generate a discretely modelled fracture network consisting of individual failure planes, opening parallel and perpendicular to the orientation of maximum and minimum strain. Fracture orientation, length and interactions vary following expected trends, forming a connected fracture network featuring population statistics and size distributions comparable to outcrop examples.Modelled fracture networks appear visually similar to natural fracture networks with spatial variation in fracture clustering and the dominance of major and minor fracture trends.Using a network topology approach, we demonstrate that the predicted fracture network shares greater geometric similarity with natural networks. Considering fluid flow through the model, we demonstrate that hydraulic conductivity and flow anisotropy are strongly dependent on the geometric connection of fracture sets.Modelling fracture evolution mechanically allows improved representation of geometric aspects of fracture networks to which fluid flow is particularly sensitive. This method enables rapid generation of discretely modelled fractures over large areas and extraction of suitable summary statistics for reservoir simulation. Visual similarity of the output models improves our ability to compare between our model and natural analogues to consider model validation.

In this chapter, we examine the interactions between fractures. We examine three types of fracture interaction: microfractures which become enveloped by the stress shadows of propagating macrofractures; macrofractures which propagate into the stress shadow of another parallel macrofracture and macrofractures which intersect a perpendicular or oblique macrofracture. The likelihood of a specific fracture interacting in one of these ways will increase as the overall fracture population grows, and we derive quantitative expressions for the probability of the different types of fracture interaction as a function of overall fracture density. In each of the interactions above, we assume that the fracture in question will cease propagating. We can, therefore, subdivide the fracture populations into active fractures, that are still propagating, and static fractures, that have ceased propagating. We show how to combine the expressions for the probability of fracture interaction with the cumulative density distribution functions derived in previous chapters, to obtain cumulative density distribution functions describing the volumetric density and mean linear density of active and static fractures.

In this chapter, we summarise the key insights we gain from the new method and discuss its potential applications in geomechanical and fluid flow modelling. We consider the value that it could add in various industrial contexts, including tunnel excavation and mining, groundwater management, hydrocarbon exploration and production, geothermal energy extraction and CO2 sequestration. Finally, we discuss how the technique could be developed further, to model fractures that have been reactivated in multiple deformation episodes and to model fractures that propagate across multiple mechanical layers.

Summary We apply the mechanically driven approach to fracture modelling of Welch et al. (2019) to two fractured outcrops, Nash Point in South Wales and Robin Hood’s Bay in NE England. We show that the local stress field developed around larger faults is a key control on the geometry of the fractures, along with layer thickness, fracture propagation rate and layer stiffness. We can use this information to predict fracture anisotropy and connectivity.

In this chapter, we derive equations to describe the growth of layer-bound macrofractures in response to an applied horizontal strain. These form when microfractures grow large enough to span the entire layer, so we can calculate the rate of macrofracture nucleation using the equations for microfracture growth. The rate of macrofracture propagation is controlled by the layer thickness. We can, therefore, derive expressions for the total macrofracture area at any time based on the macrofracture nucleation and propagation rates. Using this information we can derive cumulative density distribution functions that describe the volumetric macrofracture density (MFP30) and mean linear macrofracture density (MFP32) through time. Finally we show how to use these results to calculate important parameters such as the fracture porosity, the total strain accommodated by fracture displacement and the total volume of the stress shadows surrounding the fractures.

In this chapter, we examine the impact of the growing fracture populations on the in situ stress in the fractured layer. We describe two potential scenarios. The first is the stress shadow scenario, in which displacement on a fracture accommodates all the horizontal elastic strain in a zone surrounding the fracture, known as the stress shadow. The second is the evenly distributed stress scenario, in which some of the horizontal strain is accommodated by displacement on the fractures, while the remainder is accommodated by elastic strain throughout the host rock, and there are no stress shadows. We show that in systems with stress shadows, the in situ stress outside them is the same as in an equivalent unfractured layer, while in the absence of stress shadows, we can derive modifiers for the bulk rock elastic moduli to take into account the effects of the fractures, based on the fracture densities calculated in previous chapters.

In this chapter, we use observations from three fractured outcrops in the UK (Nash Point, Robin Hood’s Bay and Pegwell Bay) to generate discrete fracture network models (DFNs). We show that these DFNs can replicate the complexity observed in the real fracture networks. We use only basic observations of the geometry of large-scale faults (>100 m in size) controlling their formation and simple assumptions regarding the mechanical properties and stress state at the time of deformation. This is sufficient for us to generate DFNs that replicate the fracture spacing (i.e. mean linear microfracture density, P32), the fracture length distributions, the spatial distribution of the fractures and the fracture anisotropy observed in the different outcrops. Furthermore, our application to these field analogues begins to reveal further questions as to the role of fluid pressure and stress shadows in generating fracture corridors.

In this chapter, we apply our method to two subsurface examples: the Kraka oilfield in the Danish North Sea and an onshore geothermal prospect in the Drenthe province in the Netherlands. Production from the Kraka field is highly dependent on the natural fracture network of the chalk reservoir. We test the method using two sets of strain input data, one based on a regional extension and one based on detailed backstripping and show how we can use the geometry of the salt pillow underlying the Kraka field, and the large-scale faults observed in seismic data, to create large-scale, detailed fracture models covering the entire field. We demonstrate a good match to the fractures observed on calibration data such as core, borehole images and high-quality three-dimensional seismic volumes. The Drenthe prospect has limited calibration data, so we generate a set of multiple geologically realistic fracture network models encompassing the uncertainty that can be used in scenario-based risk analysis. We then examine lateral variability in the resulting fracture networks and to identify areas that are more likely to develop more or less anisotropic fracture networks, and areas that are likely to have longer or shorter fractures. Finally, we consider the implications for fluid flow within the geothermal prospect.

Geoscience data often have to rely on strong priors in the face of uncertainty. Additionally, we often try to detect or model anomalous sparse data that can appear as an outlier in machine learning models. These are classic examples of imbalanced learning. Approaching these problems can benefit from including prior information from physics models or transforming data to a beneficial domain. We show an example of including physical information in the architecture of a neural network as prior information. We go on to present noise injection at training time to successfully transfer the network from synthetic data to field data., 5 pages, 5 figures, workshop, extended abstract, EAGE 2019 Workshop Programme, European Association of Geoscientists and Engineers

Summary Recent advances in machine learning relies on convolutional deep neural networks. These are often trained on cropped image patches. Pertaining to non-stationary seismic signals this may introduce low frequency noise and non-generalizability.

Summary We have developed a new technique to generate more realistic discrete fracture network (DFN) models by dynamically simulating the propagation of fractures with models that use the equations of motion and conserve mass and momentum, and include the interactions between different fractures. We use this to simulate the propagation of layer-bound fractures in a thin, homogeneous reservoir, and compare the resulting cumulative fracture density functions (P30 and P32) generated under different conditions. A critical factor in the models is the subcritical fracture propagation index b. Subcritical fracture propagation, which can occur at low stress but is characterised by low propagation rates controlled by chemical and thermal processes at the fracture tip, results in a large number of relatively short fractures. By contrast critical fracture propagation, characterised by rapid rupture at high stress, generates a smaller number of fractures with a greater range of lengths, including some very long fractures, although the total fracture area is similar to that generated by subcritical propagation. Intersection with orthogonal fractures under biaxial strain conditions, and stress shadow interactions between parallel fractures, can also affect the resulting fracture area and lengths.

Scanning-Electron images from North Sea Chalk are studied for important rock properties. To relieve this manual labor, we investigated several standard image processing methods that underperformed on complicated chalk. Due to the lack of manually labeled data, deep neural networks could not be adequately applied. Gaussian Mixture Models learnt a two-fold representation that separated the background well from the rock. Subsequent morphological filtering cleans up the prediction and enables automatic analysis.

Kilometre-scale geobodies of diagenetic origin have been documented for the first time in a high-resolution 3D seismic survey of the Upper Cretaceous chalks of the Danish Central Graben, North Sea Basin. Based on detailed geochemical, petrographic and petrophysical analyses it is demonstrated that the geobodies are of an open-system diagenetic origin caused by ascending basin fluids guided by faults and stratigraphic heterogeneities. Increased amounts of porosity-occluding cementation, contact cement and/or high-density/-velocity minerals caused an impedance contrast that can be mapped in seismic data, and represent a hitherto unrecognized, third type of heterogeneity in the chalk deposits in addition to the well-known sedimentological and structural features. The distribution of the diagenetic geobodies is controlled by porosity/permeability contrasts of stratigraphic origin, such as hardgrounds associated with formation tops, and the feeder fault systems. One of these, the Top Campanian Unconformity at the top of the Gorm Formation, is particularly effective, and created a basin-wide barrier separating low-porosity chalk below from high-porosity chalk above (a regional porosity marker, RPM). It is in particular in this upper high-porosity unit (Tor and Ekofisk formations) that the diagenetic geobodies occur, delineated by ‘Stratigraphy Cross-cutting Reflectors’ (SCRs) of which 8 different types have been distinguished. The geobodies have been interpreted as the result of: 1) escaping pore-fluids due to top seal failure, followed by local mechanical compaction of high-porous chalks, paired with 2) ascension of basinal diagenetic fluids along fault systems that locally triggered cementation of calcite and dolomite within the chalk, causing increased contact cements and/or reducing porosity. The migration pathway of the fluids is marked by the SCRs, which are the outlines of high-density bodies of chalk nested in highly porous chalks. This study thus provides new insights into the 3D relationship between fault systems, fluid migration and diagenesis in chalks, and has important applications for basin modeling and reservoir characterization.

Summary This work proposes a new workflow to characterize the 3-D fracture intensity along wells based on a mechanial approach to predict fracture distribution in 2-D along wells, an extrapolation of the 2-D fracture set in 3-D as discreete objects and an explicit calculation of the 3-D fracture intensity (i.e.: P32). Such an approach takes into account the mechanical stratigraphy, the stratabound fractures, cutting through faults, and shifts several issues inherent to traditional methods that involves a correction factor (such as terzaghi correction) to weigh the fracture intensity as function of the fracture orientation with respect to the well trajectory, the quality of the images logs, the interpreter for fracture picks.

Ground-penetrating radar (GPR) and satellite ERS-2 synthetic aperture radar (SAR) are used to map the thickness and extent of the superimposed ice (SI) zone on the surge-type glacier Kongsvegen, Svalbard. GPR imagery shows sub-horizontal SI layers lying unconformably above a discrete boundary. Below this boundary, the ice has a GPR signature similar to that of ice further down-glacier in the ablation zone. This boundary is posited to represent the closing of crevasses that were created during the last surge of Kongsvegen in ∼1948. Open crevasses would have interrupted the formation of sheet layers of SI due to efficient vertical drainage of the snowpack. Aerial photographs suggest that the crevasses closed sometime in the period 1956–66. A classified SAR image from 2003 is used to delineate the extent of the SI zone. The SI extent in the SAR image agrees well with the SI zone mapped by GPR. Using the SI spatial depth distribution, we estimate the mean annual accumulation of superimposed ice to be 0.16 ± 0.06 m w.e. a−1 (locally up to 0.43 m a−1 w.e.). This corresponds to ∼15–33% of the local winter balance and ∼5–10% of the total winter balance measured since 1987.

Chalk fields are in general of high porosity and low permeability. However, many chalk fields are heavily fractured which provides the effective permeability needed to produce from them. Understanding the geometry, density and distribution of the fractures is vital to predict fluid flow through the reservoir. An energy balance model that predicts the nucleation and propagation of fractures is tested on a chalk field in the Danish North Sea where hydrocarbons are produced from the Paleocene Ekofisk Formation and Late Cretaceous Tor Formation. Fracture densities are available from both core measurements and FMI data. The model predicts a fracture distribution similar to what is observed and with a close match to the fracture density from FMI data. Similar to 4D seismic results it is predicted that only a few fractures connects the two reservoirs. This study shows that the use of an energy balance model is an important step forward to improve the knowledge about the fracture distribution and vertical connection between different flow units in a reservoir.

We present a model study investigating the summer evolution of supraglacial lakes on the Greenland ice margin. Using a one-dimensional (1-D) model we calculate the surface ablation for a bare ice surface and beneath supraglacial lakes for 30 days in the summers of 1999 and 2001. The surface ablation beneath the lake was enhanced by 110% in 1999 and 170% in 2001 compared with the ablation for bare ice. We then use the results from the 1-D model to further model the vertical and horizontal evolution of the supraglacial lakes, the results of which are compared with satellite images. Within the region of the ice sheet where supraglacial lakes presently occur, the area covered by supraglacial lakes is found to be more or less independent of the summer melt rate but controlled by topography. We therefore predict that, inside this region, the area covered by supraglacial lakes will remain constant even in a warmer climate. However, in a warmer climate, surface melting will occur higher on the ice sheet where small surface slopes favour formation of large supraglacial lakes. Enhanced surface melting beneath such lakes is a hitherto overlooked feedback mechanism related to climate warming.

[1] We report measurements of ablation rates of the bottom of two supraglacial lakes and of temperatures at different depths collected during the summers of 2010 and 2011 in west Greenland. To our knowledge, this is the first time that such data sets are reported and discussed in the literature. The measured ablation rates at the bottom of the two lakes are of the order of ∼6 cm/day, versus a rate of ∼2.5–3 cm/day in the case of bare ice of surrounding areas. Though our measurements suggest the presence of a vertical temperature gradient, it is not possible to draw final conclusions as the measured gradient is smaller than the accuracy of our temperature sensors. In-situ measurements are compared with the results of a thermodynamic model forced with the outputs of a regional climate model. In general, the model is able to satisfactorily reproduce the measured quantities with RMSE of the order of 3–4 cm for the ablation and ∼1.5°C in the case of water temperature. Our results confirm that the ablation at the bottom of supraglacial lakes plays an important role on the overall lake volume with the ablation in the case of ice covered by a lake being 110–135% of that over bare ice at nearby locations. Beside ice sheet hydrological implications, melting at the bottom of a supraglacial lake might affect estimates of lake volume from spaceborne visible and near-infrared measurements.

The objective of this study is to develop improved methods for predicting pore pressure from seismic data. Basin modelling techniques have been combined with Rock Physics velocity modelling to obtain an accurate and robust prediction of pore pressures prior to drilling. The seismic wave velocities depend on porosity, pore pressure and the burial history of the sediments. Knowledge of velocity-depth trends is important when predicting pore pressure from seismic data. Information provided by basin modelling can help establish the appropriate velocity-depth relation. Shales make up over 80% of sediments and rocks in sedimentary basins. Hence, a proper description of velocities in the shaly intervals is therefore important for seismic data interpretation and for accurate estimation of the overburden properties. We have taken a model-based approach in order to make pore pressure predictions away from well-control. Rock Physics based models are used for calculating velocities at a given state of stress, mineralogy and diagenesis. The burial and diagenetic histories are used as input to the velocity model. The pressure predictions from seismic velocities have been coupled with pore pressure predictions from basin modelling. The model has been successfully applied to case studies from the North Sea.

Today we experience an accelerated melting of sea ice in the Arctic which global circulation models are inadequate to predict. We believe one of the reasons is the shortcomings in the sea ice albedo schemes for these models. This paper investigates a physically based sea ice albedo scheme for ECHAM5 GCM, which separates between snow-covered sea ice, bare sea ice, melt ponds, and open water (separately for the albedos and albedo fractions). The new albedo scheme includes important components such as albedo decay due to snow aging, bare sea ice albedo dependent on the ice thickness, and a melt pond albedo dependent on the melt pond depth. The explicit treatment of melt pond albedos has, to our knowledge, not been included in general circulation models before and represents a substantial improvement when simulating the annual cycle of sea ice albedo. The new albedo scheme overall reduces the sea ice albedo both in winter, because of snow aging, and in summer, because of melt ponds. The reduced sea ice albedo leads to overall reduced sea ice thickness, concentration, and volume, with large temporal and spatial variations. In the Northern Hemisphere in March, some areas experience increased albedo, resulting in thicker sea ice and higher ice concentration, but in August the pattern is spatially homogeneous, with reduced albedo, thickness, and concentrations for all areas where the new scheme has a significant effect. [References: 57]

[1] We present a mathematical model describing the summer melting of sea ice. We simulate the evolution of melt ponds and determine area coverage and total surface ablation. The model predictions are tested for sensitivity to the melt rate of unponded ice, enhanced melt rate beneath the melt ponds, vertical seepage, and horizontal permeability. The model is initialized with surface topographies derived from laser altimetry corresponding to first-year sea ice and multiyear sea ice. We predict that there are large differences in the depth of melt ponds and the area of coverage between the two types of ice. We also find that the vertical seepage rate and the melt rate of unponded ice are important in determining the total surface ablation and area covered by melt ponds.

This paper presents a new integrated approach for updating pore pressure predictions during drilling of a well. Pore pressures in sedimentary rocks can vary from hydrostatic (normal) pressures to very high (abnormal) overpressures. Sediment pore pressures are important constraints that need to be carefully considered when planning a new well and will influence drilling strategy, casing program, and mud-weights.A pre-drill prediction of pore pressures is used as input in the well planning phase and is based on an assumption of the local geology in the prospect, typically extrapolated from near-by wells. However, the pre-drill prediction may be inaccurate or erroneous due to incorrect assumptions.To help solving this problem we have developed an integrated approach for updating the pore pressure prognosis during drilling. The methods for pore pressure predictions are integrated with an eDrilling system for real-time simulation of the drilling process. Data from the drilling operation are processed real-time and includes drilling parameters, logging while drilling data, and other measurements performed during drilling. The base for a real-time update is pre-drill pore pressure predictions from basin modeling using Monte Carlo analysis. As drilling and log data becomes available during the drilling operation, fast calculations of pore pressure along the well are performed. The calculated pressure along the well is used to weight the pre-drill Monte Carlo results in order to update the pore pressure prediction ahead of the bit.Real-time updated pore pressure predictions are of vital importance for advanced computer based systems for online supervision and decision support during drilling. Early warnings can be supplied if abnormal overpressures are to be expected ahead of the bit, or if measures should be taken to avoid dropping below the pore pressure. This paper presents the methodology and simulation results using log data from the Norwegian North Sea. Simulations show that pore pressure predictions ahead of the bit are improved compared with the pre-drill predictions, when updating predictions using the drilling logs.The novel combination of automatically updated pore pressure data during drilling with an advanced real-time integrated drilling simulator provides a powerful tool for next generation drilling software.

Summary We present an integrated fracture study in the Ekofisk chalk reservoir of the Kraka Field, offshore Denmark, based on core, borehole images and seismic data. The core contains numerous fractures ranging from short (cm-scale) fractures, mostly associated with chert or stylolites, to large (m-scale) open, slickensided fractures likely related to halokinesis. On borehole images, especially larger fractures are identified, coinciding in dip and dip-azimuth. Seismic data at an approximate resolution of 40m would not resolve these local features around the well-bore. We show that chromatic analysis combined with an ant-tracking algorithm extracts several lineaments (> m-scale) from the seismic data. These correlate closely in orientation and distribution with the fractures logged in the well data. It is likely that these represent fracture corridors, small faults or damage zones in the chalk. The seismic data therefore provides a valuable method for mapping the size, orientation and connectivity of fracture zones away from the well. This gives insights into the scalability of local stress fields, and fracture distributions.

As the world’s largest emitter of greenhouse gasses (GHGs), China plays a central role in the suite of options for climate change mitigation. To analyze the importance of biomass and carbon capture and storage (CCS) availability in China, varying levels of these parameters are created and then global climate scenarios are simulated using TIAM (TIMES Integrated Assessment Model). TIAM is a 16-region global energy system optimization model that includes a climate module that calculates the global concentrations of GHGs in the atmosphere. We analyze the potential for using biomass, CCS, and bioenergy CCS (BECCS) in China under the constraint of meeting a climate stabilization target such that dangerous climate change (as defined by the Copenhagen Accord) is avoided. When considering hypothetical scenarios where GHG emissions are constrained, China consumes all available domestic biomass as a relatively inexpensive fuel source. However, while BECCS does have a small role to play, in general it is cheaper to use biomass for the transportation sector and CCS with fossil fuel in order to meet both the energy demand and emissions reduction goals in the cheapest way possible. Therefore, we find that while both utilization of biomass and CCS are essential options for reducing emissions in China, BECCS is not the most cost effective option in China. CCS is nevertheless an important option for China; in the climate mitigation scenarios modeled, by 2050, China is projected to employ CCS on at least 70% of fossil energy electricity generation. When CCS is excluded, the cost of mitigation is more than doubled compared to the scenarios where CCS is included as a mitigation option.

Summary We have developed a method of created geomechanically-based Discrete Fracture Network models by simulating the nucleation and propagation of natural fractures over geological time, based on linear elastic fracture mechanics and subcritical fracture propagation theory. In this presentation we apply the method to the Kraka field offshore Denmark, which comprises a fractured chalk reservoir developed over a salt pillow. We calculate the magnitude and orientation of the horizontal strain experienced during development of the salt structure by backstripping, and use this as input to the fracture propagation model. We compare the results with fractures interpreted on borehole images from 5 horizontal wells, as well as lineations observed on ant-tracked seismic data, and find a good match in both orientation and fracture density.

The South Arne chalk field is located in the Danish sector of the North Sea. It is a double-dipping anticline extending 12½ km by 3½ km. Production started in the summer of 1999 using horizontal wells and water flooding with a cumulative production to date of 141 MMbbl from the two oil bearing intervals: Maastrichtian (Late Cretaceous) Tor Fm and Danian (Paleocene) Ekofisk Fm. 3D seismic was shot in 1995 (pre-production), in 2005 and in 2011. The matrix permeability of the field is low but greatly enhanced due to natural fractures. Based on the 2005 seismic interpretations a geomodel was built with emphasis on capturing the permeability enhancement due to the occurrence of fractures. The modelling was done using discrete fracture network (DFN) modelling. Fracture density was calculated based on various attributes and the outcome calibrated to well test data. Calibration was performed using full field adjustments on fracture properties enhancing predictability in less data dense areas. The model has been history matched to production data with a good overall result. The seismic shot in 2011 was compared to pre-production seismic data giving valuable information about unswept areas and water cutting as well as differences between the two reservoir intervals. The 4D data also shows that the faults have significant influence on fluid flow as modelled in the geomodel. Using fluid substitution modelling, the strongest signal is expected to come from changes in Sw. Comparing the observed changes from 4D seismic to the history-matched modelled changes in Sw, from pre-production to 2011, a very good match is found, thus validating the geomodel.

Fracture prediction and modelling is one area where there is scope for significant technical advance. At present fractures are modelled either by modifying the bulk rock properties to take account of the fracture porosity and permeability, or using stochastically generated Discrete Fracture Network (DFN) models. Both methods tend to give a poor history match because the distribution, orientation, length and connectivity of fractures in the subsurface is not well constrained. We propose to improve the prediction of fluid flow in a fractured reservoir, by developing an algorithm to accurately model the key parameters of a fracture population (fracture density, fracture size distribution, and connectivity) based on the geomechanics of fracture nucleation and propagation, and using this to generate a mechanically-based DFN that more accurately represents the subsurface fracture geometry. We will use a preliminary version of this model to demonstrate some of the key geomechanical controls on fracture geometry, such as fracture propagation rate and strain anisotropy, and illustrate the different types of fracture network that can develop under different conditions.