Potential applications of time-lapse CSEM to reservoir monitoring
technical article first break volume 33, April 2015 Potential applications of time-lapse CSEM to reservoir monitoring O. Salako1, 2, C. MacBeth1* and L. MacGregor3 Abstract Monitoring of changes in brine chemistry (salinity and temperature), during water-flooding is important for injector optimisation, understanding efficiency, detecting early water breakthrough, locating bypassed hydrocarbons or detecting scaling in the heterogeneous reservoir. It is already known that water injection into the oil-leg of a hydrocarbon reservoir can be monitored by both seismic acquisition and also CSEM methods. The problem of an interfering pressure signal for seismic impacts quantitative evaluation for time-lapse analysis, while with EM there are the counterbalancing effects of salinity and temperature. CSEM becomes favoured when the pressure effects on seismic are dominant, or heavy oil is present with similar acoustic properties as the formation and injected waters. The quality of measurement for both methods is influenced by reservoir facies variations, acquisition repeatability and overburden heterogeneity. Time-lapse seismic is unlikely to detect brine distributions injected into the water-leg or aquifer, although it may detect associated pressure up effects. However, our calculations show that it may be possible for time-lapse CSEM to distinguish the inter-mixing of different brines in the subsurface hydrocarbon reservoir. Specifically, low salinity injection or injection into a highly saline formation can clearly be detected with this technique. Introduction Electromagnetic (EM) methods in industry geophysics are generally employed to measure the conductivity (or resistivity) of fluid-saturated rocks, which may aid in discriminating highly resistive hydrocarbon-bearing rocks from those with relatively more conductive saline formation water. Such resistivity measurement forms the basis for one of the oldest (1927) wireline logs to be run for the purposes of reservoir evaluation. In modern quantitative petrophysical analysis, direct resistivity from deep and shallow laterologs or induction logs are routinely interpreted together with porosity from density or neutron density logs and brine resistivity calibrated in the laboratory, to determine water saturation and hence recognise hydrocarbon-bearing zones. EM measurement techniques have also evolved to occupy not just the log domain but other wider aspects of inter-well imaging. A recent example is high resolution cross-well EM tomography methods that can track details of a water flood in a hydrocarbon reservoir between wells 500 m to 1 km apart (for example, Al Ali et al., 2009). Another example is surfacebased EM surveying, and in particular the marine controlled source EM (CSEM) method which has become popular over the past 15 years since first publication of the proof of concept for imaging hydrocarbon reservoirs (Ellingsrud et al., 2002). Here, a source in the form of an electrical dipole transmitter is towed close (approximately 25 m) to the seafloor and used to induce horizontal and vertical loop currents in the subsurface that are then measured on a receiver sensitive to (typically) the electrical and magnetic components. The received signal is a function of source-receiver geometry (offsets usually in the range of a few hundred metres to 10 or 15 km), seafloor resistivity and transmitted frequency, which is typically in the frequency range 0.1 to 10 Hz (MacGregor and Tomlinson, 2014). Although superficially similar to seismic methods because of the multi-component and frequency dependent aspect of the data, the governing physics is very different and care must be exercised when drawing parallels with seismic data. A natural and obvious drawback of CSEM, as a potential field method, when used for hydrocarbon exploration or as a pre-drill appraisal tool, is the poor structural constraint afforded by the method. As the signal evolves spatially according to a diffusion equation rather than the familiar propagative seismic wave equation, vertical detail is much more poorly resolved than in seismic data. These limitations require that CSEM data should be interpreted jointly with seismic data (e.g. MacGregor et al., 2012). Given the limited thickness of most hydrocarbon reservoirs, producing units or intervals (10 to 50 m), the most that Institute of Petroleum Engineering, Heriot Watt University, Edinburgh EH14 4AS, UK. Department of Geology, Osun State University, P.M.B. 4494, Osogbo, Nigeria. 3 RSI, 2600 South Gessner, Suite 650, Houston, TX 77063 USA. * Colin.MacBeth@pet.hw.ac.uk 1 2 © 2015 EAGE www.firstbreak.org 35 technical article can therefore be expected of CSEM is an estimate of bulk electrical resistivity for the reservoir at each location – but as we shall see, that particular measure may be very useful in reservoir management and evaluation. More recently, there has been the suggestion that timelapse EM imaging could provide information to supplement 4D seismic techniques when monitoring hydrocarbon reservoirs during production and recovery (Orange et al., 2009; Constable, 2010; Andreis and MacGregor, 2011). This new application to producing reservoirs at a later stage in their life cycle benefits in particular from pre-existing knowledge of the reservoir depth, structure and thickness, and hence well developed geological and simulation models that provide constraints to negate the accepted weaknesses of the EM approach. Indeed, one of the obvious benefits of the timelapse EM imaging most commonly cited in the literature, is the ability to sense changes in water saturation without the interference of the pressure effects to which the seismic data is variously susceptible. Thus, it is believed that EM may be used to help track injected water during the improved oil recovery stage of a reservoir’s life. Particular attention is paid here to lateral drive, with less attention on basal aquifer drive. Clearly, the method is of value in geographical locations where the injected and displaced fluids have similar seismic properties, the monitoring of production mechanisms in which pressure obfuscates water saturation, or quantifying water distribution in the calculation of remaining oil. There have been several modelling studies in the past that investigate this topic and assess the likely magnitude of the EM signal. These studies have considered representations of the waterflood as an idealised piston-like lateral displacement mechanism (Lien and Manneseth 2008; Orange et al., 2009) to modelling of the response with a more realistic fluid flow simulator (Shahin et al., 2010; Liang et al., 2011). These studies conclude that the absolute magnitude of the EM signal appears sufficient to be detectable with current acquisition and hardware. Given that one of the largest benefits of EM monitoring is surveillance of a waterflood, one important area for potential technology development is the introduction of more realism into the representation of the waterflood process and development of an understanding of the corrersponding reservoir recovery processes. This will enable us to better characterize the EM signal with a view to assessing its ultimate value for reservoir management. In this context, one important difference between seismic and EM is that the latter is strongly sensitive to the physical and chemical properties of the aqueous fluids. Thus, in this study we highlight the role of EM in tracking specific brine distributions. To take this further we must firstly describe the injection of water for improved oil recovery. Water injection Basic principles Water injection is the oldest and most commonly used secondary oil recovery method, whether onshore or offshore, 36 first break volume 33, April 2015 as it is relatively inexpensive and in most cases appropriate sources of water (for example, seawater) are readily available. Many fields are on water injection from close to the start of production, and this continues for most of the production lifetime (for example, the Girassol field – Bouchet et al., 2004). Such injection helps to maintain reservoir pressure near its initial level, and displaces (‘sweeps’) oil to the producing wells via voidage replacement. This process enhances the recovery factor beyond the forces of natural drive and maintains the production rate for a longer period, despite the need for additional water handling at the producers. Injection is preferably into the water column/aquifer to avoid bypassing down-dip oil accumulations. Understanding the aquifers and the bottom- or edge-drive that injectors provide is critical to such reservoir operations. However if permeability in the water leg is significantly reduced due to compaction or diagenesis, it may be necessary to supply higher energy by injecting directly into the oil column at the beginning or later in the field life. For this, injection is in low permeable zones at the flanks of the main reservoir to avoid early water breakthrough and promote a reasonable sweep (as, for example, in the Nelson field – MacBeth et al., 2005). Water injection is effective as the mobility contrast between oil and water is favourable for a continuous sweep, unlike gas injection. However, the efficiency of the displacement depends on many factors, such as viscosity and rock characteristics, pattern of injectors and producers, and hence this process has a high degree of uncertainty. Thus, over the several decades it takes to fully complete a water injection programme, effective reservoir management requires close monitoring of the evolution of the injected water. This is necessary in order to determine its efficiency at restoring pressure or control the relative volumes of injected water into different reservoir intervals, to ensure the process is efficient and that excessive water is not produced. Monitoring needs Production technology aims to monitor and control the movement of the injected water. In particular, steps to modify the original design of optimally placed injectors and producers, or injection intervals, are required if adverse conditions prevail. Adverse conditions include early water breakthrough, where water channels along highly permeable pathways to the producer, or a heterogeneous and inefficient displacement process with areas of bypassed oil. Assessment of the waterflood can be achieved by monitoring at the wells. Waterflood monitoring includes direct observation of the volume of produced oil and water, bottomhole pressures and gas-oil ratio, as a basis of evaluating the voidage replacement rates, water injection pattern efficiencies and water injection rates. As injected and formation (aquifer and connate water) water have different chemical signatures, measuring the brine chemistry of the waters produced can identify their individual origin (and even help history to match the simulation model – Arnold et al., 2012). Reservoir production www.firstbreak.org © 2015 EAGE first break volume 33, April 2015 engineers are particularly interested in identifying zones for infill drilling, and this is clearly where techniques that can image the inter-well space and monitor water sweep progression can add benefit – provided they are cost-effective. Any prospective monitoring technique therefore requires a basic fundamental ability to discriminate between oil-filled (plus connate water) zones of the reservoir rock and those that are predominantly water-filled (injected water plus connate water and residual oil). Also of fundamental importance is the ability to track the separate progress of injected and formation waters within the formation after injection into the water leg or aquifer. This is where 4D seismic or EM surveying can help, as in many cases injected and formation water have a favourable bulk seismic or resistivity property contrast with the ‘oil’ zone which has the smaller water content. For 4D seismic data, in particular, there are many good examples of direct imaging of waterflood direction, shape and distribution (for example, Røste et al., 2009). However, an important disadvantage of the seismic is the lack of acoustic contrast between some heavier oils and the injected/formation waters (see, for example, the measurements of Batzle et al., 2006). Additionally, there is probably insufficient seismic property contrast to distinguish injected and formation brines. Both of these disadvantages point to techniques such as EM imaging as being of value, as the electrical resistivity they sense is dependent not only on the amount of brine, but also the chemical state of the brine. We consider the latter point in further detail in the next section. technical article the injected water, followed by aquifer water. The amount of produced connate water depends on whether the water is injected into the aquifer or the oil leg. Direct injection into the oil leg tends to yield a small connate water production to be followed by the injected water. For increased reservoir heterogeneity (especially vertical), then the time sequence of the different breakthroughs and the appearance of the different brines will be more indistinct and complex. Aquifer water may be produced early by coning, whereas injected water breaks through at the bottom of the well. Of concern in this process of complex fluid interaction is the region or time period over which the injected and formation brines get a chance to interact. Of particular interest in oil-field chemistry is the mixing of incompatible brines with differing ionic composition. Specifically, seawater rich in sulphate ions can mix with formation water rich in barium (also strontium and calcium) ions to create a hard crystalline mineral scale deposit within the formation as well as at the wellbore. The exact location of the mixing zones is defined by the reservoir heterogeneity and injector placement. These scale deposits mainly have an effect at the well to restrict flow and limit performance at the production wellbore and production facilities (Sorbie and MacKay, 2000). Scale deposits in the deep reservoir do not have an adverse effect on porosity or permeability, and may actually prevent near wellbore scale build-up. Scale deposition is more likely to occur when seawater is injected into the aquifer. It is therefore important to monitor the location of the individual brines. Brine distribution tracking Brine mixing in the reservoir Whilst the physical displacement mechanisms associated with water injection are well known, less well known is how the injected water interacts with the subsurface aqueous fluids. Brine chemistry may impact souring, scaling and hence oil recovery. The flow of injected water through the reservoir and complex interaction with the reservoir fluids is a strong function of reservoir heterogeneity. Within the reservoir, the injected water interacts with the formation waters (connate and aquifer) both physically, to displace the mobile oil phase, but also chemically at distinct mixing zones. The composition of the formation brines depends on their source, history, and present-day thermodynamic conditions. Their distribution in turn depends on petrophysical conditions of the rocks, and physical and chemical properties of the fluids themselves. Injected water displaces the oil and mixes with the connate water in the oil zone, transition zone, or directly with the aquifer water – depending on where the injected well is completed. Numerical studies have shown that co-production of these different brine mixtures or individual brines is normally expected, and indeed complex temporal sequences of water production can occur (Sorbie and Mackay, 2000). Connate water is generally produced first at the producers by connate water banking ahead of © 2015 EAGE www.firstbreak.org Brine resistivity as a function of salinity and temperature It is well known from published literature that the acoustic density and bulk modulus of reservoir brines (for example, Batzle and Wang, 1992) or seawater (for example, Bertrand and MacBeth, 2005) is mainly a function of salinity (that is, the NaCl equivalent of the ionic composition) and temperature, but less dependent on the variety of other minor ionic components. The same dependence is true of the electric resistivity (Rw) of brines, a fact identified from the early petrophysical work by Archie. A convenient equation that captures the general dependency of brine resistivity on salinity and temperature is given by Crain (1986) (1) where Τ and Sal are the temperature in degrees Celsius and salinity in parts per million equivalent of sodium-chloride solution respectively. In addition, we know that the salinity and temperature of the brines and their mixtures vary for hydrocarbon reservoirs in different geological and hence geographical environments (see Table 1), and they also depend on the time and rate of production/injection. Thus, for a static reservoir prior to production, formation brine (i.e., aquifer or movable connate water in the transition zone of the reservoir, for which chemical differences are often 37 technical article first break volume 33, April 2015 small) varies from almost pure water (1000 ppm) to very saline (300,000 ppm) (Rider and Kennedy, 2013). As a rule, salinity generally increases with depth in the reservoir, and if near-bedded salt (such as in the Gulf of Mexico) can exceed 100,000 to 300,000 ppm. If there is no salt present, salinity will generally not exceed 30,000 ppm. The temperature of a hydrocarbon reservoir depends on the proximity to geothermal sources, but roughly varies between 75 and 100oC (100 and 150oF) – excluding high pressure – high temperature reservoirs. The conditions of the water injected into the hydrocarbon reservoir are determined mainly by its origin from a variety of available sources. The nature of this water depends on whether the reservoir is onshore or offshore, the reservoir type, and geographical location. For example, in offshore production (or near onshore) it is very common to use seawater, filtered and collected appropriately to avoid algae, corrosive oxygen content, and other undesirable chemicals. This typically has a salinity of around 30,000 ppm and temperature of 30oC (86oF) Fluid type/Cases Subsurface Sandstone aquifer water 2,3 Sea water Injected water 1,7,9 Example of oilfield/province Temperature (°C) Salinity (ppm of NaCl) Resistivity, Rw, (Ω-m) Saudi Arabia (Wasia and Biyadh) - 5,000 - 18,000 (av. of 10,000) 0.57 Niger Delta/Nigeria 24 30,000 - 35,000 0.19 - 0.22 Alaskian (Arctic) 5 - 17 30,000 - 35,000 0.22 - 0.36 UKCS 5 - 17 30,000 - 35,000 0.22 - 0.36 15 - 20 (summer) 5,000 - 14,000 0.47 - 1.31 500 - 1,500 (desalinized) 3.05 - 8.03 EOR in an oil-wet reservoir. Not < 5% of the salinity of formation water to avoid clay swelling. - Values are function of the in situ conditions and earlier injected water River water 10 Low salinity water 4 Endicott Produced water 18.7 (desalinized) Comment Rw calculated at 24°C Ave. summer; 15°C; 30,000ppm; 0.27Ω-m - - - Simpson sd, Oklahoma - 298,497 0.000 89 200,000 0.018 Temperature calculated Burgan, Kuwait - 154,388 0.053 Rw calculated at 24°C US average - 94,000 0.08 1 Saudi Arabia (Arab - D) 2,3 1 1 Formation water (drawn from the top layer of the seawater column). For onshore reservoirs, river water is utilized and after a similar treatment may be typically 10,000 ppm and 20oC (68oF). Interestingly, by contrast, in some onshore areas of Saudi Arabia, subsurface aquifer water with an average salinity of 10,000 ppm is injected into reservoirs with a very high salinity formation water 200,000 ppm on average (Rafle and Youngblood, 1987). Other options for injected water are to extract from the same or nearby water-bearing formations other than the oil reservoir. This water is purer and less saline than seawater, being around 3000 ppm and 15oC (59oF). It is also possible to intentionally inject low salinity water of between 500 ppm and 1500 ppm for enhanced oil recovery. A final option, aimed at reducing potential formation damage due to incompatible fluids, is to inject produced water. For this, work must be done to remove hydrocarbon and solid contaminants before reinjection, and this is often quite costly. This volume is never sufficient to replace all produced volumes and additional water from other sources is required to make up the volumes. Woodbine, E, Texas - 68,964 0.1 100 15,000 0.16 Niger Delta/Nigeria 54.4 20,000 0.19 UKCS 57.7 18,000 1 4,5 Endicott/Alaska 6 7 California petroleum basins - Daquing Field, China 45 Laugunillas, Venezuela - 8 6 1 0.2 30,000 - 35,000 0.19 - 0.22 5,000 - 7,000 0.55 - 0.74 Rw calculated at 24°C CASE X 7,548 0.77 CASE Y Rw calculated at 24°C CASE Z Table 1 Resistivity properties for reservoir formation or injected fluids used in secondary and tertiary recovery from a range of geographical locations. Superscripts refer to the following references: 1Rider & Kennedy 2013; 2Rafle and Youngblood, 1987; 3Youngblood, 1980; 4McGuire et al., 2005; 5State of Alaska, 2011; 6Shehata et al., 2012; 7Martin and MacDonald, 2010; 8Batzle and Wang, 1992; 9Constable, 2013; 10River and Lake Swimming Association, 2013. Examples given here are ranked according to Rw values that are calculated, when appropriate, using Crain (1986). 38 www.firstbreak.org © 2015 EAGE first break volume 33, April 2015 technical article Figure 1 (a) Brine resistivity as a function of salinity and temperature. This is compared against the acoustic properties of bulk density (b) and bulk modulus (c). Table 1 details a range of salinity and temperature for formation and injected waters, in a selection of hydrocarbon reservoirs from around the world. In some notable cases the contrast in salinity between formation and injected water is quite large. This is particularly true of the highly saline formation water of the Arab-D reservoir and much lower salinity injected water from massive underlying aquifers cited by Rafie and Youngblood (1987) (which we define as Case X in our study below), or the low salinity water injected into the Endicott field described by McGuire et al. (2005), for improved oil recovery (defined as Case Y). The latter is important, as it is now being currently considered as an effective tertiary oil recovery procedure. In the reservoir, large temperature contrasts between the brines may also be present close to the injector, particularly in environments such as the North Sea. Figure 1 shows how the seismic and resistivity properties respond to salinity and temperature variations. Both bulk density and bulk modulus vary in an approximately linear fashion with salinity and temperature. Temperature variations are predicted to have a small effect on the seismic data, with less than 5% change in bulk modulus occuring over a span of 100oC. Salinity variations generate between 10 and 14% change in bulk modulus for every 50,000 ppm over a similar temperature span, suggesting the possibility that in the extreme cases mentioned above, injected and formation waters may possibly be identified separately with high-quality data. Interestingly, although there is no literature available on the separate seismic detection of individual brines within the hydrocarbon reservoir itself, many seismic examples do exist from seismic acquisition in deep seawater columns. For example, seismic reflections have been observed from the mingling of flows with differing salinity and used to image the thermohaline structure of oceans (Holbrook et al., 2003), and have in the past been associated with difficulties © 2015 EAGE www.firstbreak.org in seismic imaging for reservoir exploration (MacKay et al., 2003). Resistivity on the other hand has an obvious stronger response to both salinity and temperature, and must be plotted on a logarithmic scale in Figure 1. Interestingly, however, for the North Sea example in Table 1 (which we refer to as Case Z), injected water of 30,000 ppm and (15oC) 59oF has a resistivity of 0.27 Ωm and formation water of 18,000 ppm and (58oC) 136oF has a resistivity of 0.20 Ωm (Figure 2). This difference is not as high as one might expect, and is due to the effect of salinity variation on resistivity opposing that of temperature. This interaction suggests that the situation best suited for CSEM monitoring application should have variations of either salinity or temperature, but not necessarily both. These resistivity changes should be compared with those for bulk modulus (2.42 GPa and 2.58 GPa) and bulk density (1.03 g/cm3 and 1.01 g/cm3), which are much weaker (reference reservoir pressure is 20 MPa). For comparison, the predicted properties for Cases X, Y and Z are compared in Figure 2. Finally, it should be noted that there is a lack of data examples in the seismic or CSEM literature desmonstrating the identification of individual brines in a hydrocarbon reservoir. This point is considered later, where modelling based on a UKCS field assesses identification under the subsurface conditions of a producing hydrocarbon reservoir. Example of expected variation during injection To understand the evolution of the brines over the production time-scale, their effect on the distribution of salinity and temperature through the reservoir, and their relative time constants and resultant spatial patterns, we take an example of a UKCS clastic field. For this purpose, a well history matched simulation model is used for a segment of the field. 3D numerical fluid-flow simulation is performed to obtain 39 technical article the water saturation, pressure, salinity and temperature distributions due to production of oil and water, and injection of seawater from vertical wells. In the well example we select, seawater is injected into the oil-leg, therefore formation brine is produced first, and over time the seawater fraction gradually increases. In this simulation we track the different contributions from the injected and formation water (however, no distinction is made between aquifer and connate water) by calculating the salinity and temperature values for the water component saturating each cell of the model. The modelling is performed for the pre-production (baseline) reservoir conditions and then for up to several years of production and recovery via a combination of aquifer support and a constant rate of water injection into the oil leg. The salinity and temperature of the brines are as given for the North Sea example (case Z) in the previous section, this yielding formation brine with a resistivity of 0.20 Ωm in the pre-production reservoir state and injected water with 0.27 Ωm. Figure 3 shows the evolution of the brine resistivity with production time for all cells in the simulation model, together with their corresponding salinity and temperature for reference. Initially, the cells have a common resistivity, and a single peak exists at the salinity and temperature of the formation water. When injection starts, another peak appears at the higher resistivity of the injected water. Over time, temperature first break volume 33, April 2015 and salinity gradients are established as these diffusive effects equilibrate. The injected water is heated up by the formation, and the higher salinity dilutes out into the larger formation volume. Temperature has the faster speed of equilibration, whilst salinity still remains fairly localised. As a consequence, the injected water heats up quickly and resistivity drops in the injected water – this can be observed in Figure 3 as a red/ orange/yellow tail emanating from the initial point for the injected water. Similarly, a smaller volume of formation water in contact with the injected water slightly cools, thus giving rise to the shorter purple upward trend from the formation water point. There is also a drop of resistivity in the formation water due to the impact of the salinity diffusion. The equilibrium condition for the temperature and salinity profiles is defined by the injection and production rates, rock and fluid properties, and the initial conditions of the reservoir. However, equilibration is complicated by the physical movement of the brines in the subsurface. Both equilibration effects are faster, however, than the physical progression of the water flood. Cells with a different combination of salinity and temperature from the initial state may be considered to be in mixed brine conditions. Figure 4 shows the changes in temperature and saturation predicted from one of the injectors in our UKCS field. Changes are a complicated interaction of both the temperature and Figure 2 (a) Brine resistivity as a function of salinity and temperature. This is compared against the bulk density (b) and bulk modulus (c) for cases X, Y and Z described in the text. The brine resitivity is calculated using Crain (1986) while the brine bulk density and bulk modulus are calculated using Han and Batzle (2000) at a reference pressure of 20 MPa. 40 www.firstbreak.org © 2015 EAGE first break volume 33, April 2015 technical article Figure 3 Cross-plots of brine resistivity for each cell of our North Sea simulation model against the corresponding temperature, colour-coded by salinity. These plots are drawn as a function of injection time. The initial temperature and salinity of the injected and formation water are labelled as IW and FW respectively. Production occurs after August 1998, and injection starts at the same time as production. Figure 4 The evolution of: (a) water saturation; (b) salinity; (c) temperature; and (d) resistivity outwards from a seawater injector over one year (between Aug 98 and Aug 99 in Figure 3) of production and injection in our example North Sea clastic reservoir (Case Z). The composite brine resistivity changes at each location around the well are shown in (d). Changes for each property are displayed as percentages. salinity variations. To understand these effects further when combined with the impact of the rock framework, the spatial distribution of the resistivity must be modelled. This is the subject of the next section. Modelled response for a producing reservoir To better understand how the fluid properties impact the observed CSEM and seismic responses recorded at the surface, © 2015 EAGE www.firstbreak.org full modelling and rock physics calculation is performed for our field example. The reservoir intervals of interest are buried at between 1990 m and 2050 m below the seafloor, and are between 25 and 50 m thick. Porosity ranges from 25 to 30%, and net-to-gross between 0 and 1 (see Figure 5). Before modelling can proceed, the resistivity of the fluidsaturated rock of variable porosity and shale content must be calculated. This will modify the fluid detectability, and thus 41 technical article first break volume 33, April 2015 Figure 5 Resistivity log for our North Sea field, from which lithology and fluid conditions are calibrated for our modelling exercise. The mean resistivity for the water-saturated sand is 2.5 Ωm, the shale is 2.8 Ωm, and an oil-saturated sand is 60 Ωm. The blue horizontal lines denote the reservoir interval considered in the fluid flow simulation in this work. lower the chance of a successful direct interpretation of the fluid response. The composite resistivity of rock and fluids In seismic data, to understand the impact of fluids and the porous rock matrix, the Gassmann (1951) equation is normally used. In the case of electrical resistivity, the equivalent equation originates from Archie (1942), where total formation resistivity Rt is determined from the fluid resistivity Rw, (2) where account is taken of the brine saturation Sw present in the pore space. The saturation exponent, n, is usually assumed to be 2. The formation resistivity factor F is a function of sediment texture and is independent of Rw (3) 42 where α is the formation tortuosity (generally regarded as unity), Φ is the total fractional rock porosity and m the cementation factor typically taken as 2 in clastic settings. The above equations were originally fit by a range of porosities for laboratory data for clean consolidated brine-saturated sandstones, but despite these restrictive origins they are still found to be approximately true for a wide range of cases in log analysis (Rider and Kennedy, 2013), with only a slight variation in both a, m and n from the ‘typical’ values above. With a fairly uniform formation rock, EM methods can monitor total resistivity and hence changes in water content (Sw). Although it is believed that resistivity is a poor indicator of lithology, some rocks, such as cemented sands, coals, anhydrites and limestones, have a high resistivity, and hence may affect the ability to interpret fluids. Whilst the basic exponents m and n in the above relation do not vary significantly with the formation, a strong contribution is played by porosity, and thus the Rw signature as discussed in the www.firstbreak.org © 2015 EAGE first break volume 33, April 2015 previous section will be amplified, or not, by the effects of the formation. Thus, CSEM does not escape the need for us to know porosity in the formation (in a similar way to seismic data). Indeed, high resistivity may indicate low porosity and not hydrocarbon. An important input to this interpretation is thus a calibration from the resistivity log data. For our reservoir example, we must also generalise this equation to the specific case of shaley sand for which the laminated shale model holds. Treatment of electrical resistivity in shaley sand occupies a considerable body of research (Worthington 1985 cites and critically discusses more than models) and is complicated by the ionic distribution in the bounding water layers. We take advantage of the fact that the horizontal electrical dipole source and receiver configuration we are considering is sensitive only to the vertical resistivity within the reservoir interval (see, for example, Key, 2009; Brown et al., 2012). Thus, in our case an arithmetic average which approximates vertical resistivity can be used, as the (thin) reservoir acts as a serial resistor network. Thus, the total resistivity in the reservoir is (Tsili and Sheng, 2001) (4) where Rsh is the resistivity of brine-saturated shale (usually lower than that of the oil-filled sands), and NTG is the netto-gross of the reservoir rocks. Modelled EM and seismic responses To simulate the EM response recorded at the surface due to changes in the reservoir conditions, we use the results of the fluid flow simulator to create a subsurface cellular model of resistivity distribution. The simulator provides pressure and saturation changes at the time of each seismic survey acqui- technical article sition, which we also assume to be the time at which the CSEM survey is acquired. These changes are then converted into either seismic impedances and velocities, or vertical resistivities. A homogeneous overburden resistivity model was generated from the well logs – this is used as a guide, despite the fact that resistivity well logs are horizontally directed while the CSEM-measured resistivity is vertically directed. We then apply a 1D dipole modelling approach to the in-line acquisition geometry of Key (2009) on a point-by-point basis across the reservoir. It should be noted that the 1D responses obtained using this approach represent the best case scenario in terms of signal strength since no 3D effects are included in the modelling. Such higher dimensional effects will reduce the effect of the reservoir on the CSEM response significantly. However, the time-lapse 1D approach used here provides an instructive starting point to examine potential applications of time-lapse CSEM in tracking different brines involved in secondary and enhanced oil recovery of hydrocarbon reservoirs. Responses are presented in the common offset domain, normalised against a baseline model, which in this case is the starting state of the reservoir. The normalised response for given source-receiver separation is plotted at the midpoint between the source and receiver to produce a map of anomalous response. Numerical computation determines that an offset of 9 km and frequency of 0.2 Hz gives an optimal response for the particular reservoir and overburden structure under consideration. For comparative purposes, seismic modelling is also performed on the same model using the work of Amini et al. (2012), which incorporates a calibrated petroelastic model for the reservoir and convolutional modelling to generate a seismic data volume. Four scenarios are modelled – two extreme cases (X and Y, as detailed previously), together with a UKCS Figure 6 Time lapse seismic amplitude changes for the four scenarios after 12 months of water injection: (a) reference case, with constant brine properties for both injected and formation water; (b) Case X, in which subsurface aquifer water (10,000 ppm, 24ºC) is injected into a highly saline formation water (200,000 ppm, 89ºC) in a Middle East reservoir; (c) Case Y, in which low salinity water (1500 ppm, 18.7°C) is injected into formation water (15,000 ppm, 100ºC) in Endicott field reservoir; and (d) Case Z, in which seawater (30,000 ppm, 15ºC) is injected into formation water (18,000 ppm, 58ºC) in a North Sea reservoir; (e) A bar chart displaying the percentage time-lapse seismic amplitude changes for the four scenarios. © 2015 EAGE www.firstbreak.org 43 technical article first break volume 33, April 2015 Figure 7 As in Figure 6, but for the CSEM response. case (Case Z). For reference, a model that holds all brine properties constant over time and does not differentiate between injected and formation water is also considered. Figure 6 shows the corresponding mapped seismic timelapse response for all four scenarios, six months after water injection. The change in water saturation generates a fairly substantial hardening signal in the seismic of about 30% for the reference model – easily sufficient to be detected in practice. When the brine chemistry effects are included, there is a difference of only a few fractions of a per cent between Case X (high saline formation water) and Case Z (UKCS), but 3% difference relative to the reference for Case Y (low salinity injected water) where the effect of salinity decrease can be clearly observed around the injector. Figure 7 shows the results of the CSEM calculation. There are now some distinct differences between the scenarios and the reference, and the effects of salinity compete against those of temperature to create signals that exhibit both positive and negative polarity. Time-lapse EM responses range from -13.5% to +6.5%, and at least when only these 1D models are considered, are likely to be detectable. For Case X, there is a weak 0.1% (positive) change as the resistivity of the injected water is much higher than that of the strongly saline formation water. The salinity effect completely counteracts the response due to water saturation alone. The largest positive change (12.1%) is for Case Y, and arises due to injection of the low salinity, high resistivity water. Relative permeabilities of water through oil are also adjusted as a result of the injection, and the water front progression is slower than that of the reference in Figure 7(a) – hence the strong low salinity anomaly in the vicinity of the injector (see also Figure 2). Diffusion of the higher salinity formation water into the displacing front ensures the water saturation response can be 44 detected, unlike case X, albeit reduced relative to the reference. Case Z displays a similar response to the reference. This is due to the injected and formation water having a similar resistivity as salinity and temperature counteract each other for both of these brines. Overall, the deviations of the EM response from the reference are more significant than we observed from seismic. In summary, both time-lapse seismic and CSEM can detect signals of water saturation (except perhaps case X), but the seismic response is, in general, stronger. However it is unlikely that the seismic will detect changes of brine chemistry, or distinguish between the scenarios we have considered. Conclusions and discussion It is already well known that both CSEM and seismic timelapse acquisition can be used to monitor the development of waterfloods in hydrocarbon reservoirs. The CSEM response is also strongly influenced by brine salinity and temperature variations in the water itself, and the seismic by a strong pressure up signal and to a lesser extent salinity and temperature. In practice, either imaging method may be used at a qualitative level to assess basic injector performance such as the likelihood of water breakthrough or inefficiencies due to channelling. In addition to this acknowledged benefit, our study shows that CSEM has the theoretical potential to track individual brines in the reservoir, provided they have distinctly different chemical signatures (salinity-temperature footprints). However, in the three cases we selected from producing fields worldwide, only the low salinity water flood case revealed a strong and detectable signal. It is probable that water injected into an aquifer will not be detected by seismic, but CSEM may stand a chance. If injection is into the oil leg, the flood will be visible to both techniques. Indeed, www.firstbreak.org © 2015 EAGE technical article first break volume 33, April 2015 any likely benefit will be strongly field and reservoir dependent. This finding is important for reservoir monitoring and management as brine mixing and rock-fluid interaction is critical to understanding the proper functioning of reservoir flow. Whilst monitoring of connectivity can to some extent be accomplished by time-lapse logging, and tracer studies using production chemistry, no current method exists to be able to directly detect the spatial distribution of the individual brines in the hydrocarbon reservoir. Three dimensional imaging of the formation, aquifer and injected brines stands as a compelling challenge to our current monitoring techniques. In this study we have focused on the detection of brines. However many other potential applications for CSEM have been cited in the literature. Another beneficial area includes the polymer flood, where for certain polymers the resistivity of the polymer to chase injected water is favourable (Jun et al., 2012). There is an opportunity, as the 4D signal for both polymer and chemical floods is thought to be negligible and unsuitable for 4D seismic monitoring due to the small acoustic contrasts (Johnston, 2013). Other possible applications include monitoring methane or CO2 injection. As more CSEM data are recorded over producing hydrocarbon reservoirs, then the detection limitations defined above will become clearer. Batzle, M. and Wang, Z. [1992] Seismic properties of pore fluids. Geophysics, 57 (11), 1396–1408. Batzle, M., Hofmann, R. and Han, D.-H. [2006] Heavy oils-seismic properties. The Leading Edge, 25 (6), 750–756. Bertrand, A. and MacBeth, C. [2005] Repeatability enhancement in deep-water permanent seismic installations: a dynamic correction for seawater velocity variations. Geophysical prospecting, 53 (2), 229–242. Bouchet, R., Levallois, B., Mfonfu, G. and Authier, J.-F. [2004] Girassol field optimized development. AAPG International Conference, Extended Abstracts. Brown, V., Hoversten, M., Key, K. and Chen, J. [2012] Resolution of reservoir scale electrical anisotropy from marine CSEM data. Geophysics, 77, E147–E158. Constable, S. [2010] Ten years of marine CSEM for hydrocarbon exploration. Geophysics, 75 (5), 75A67–75A81, doi: 10.1190/1.3483451. Constable, S. [2013] Review paper: Instrumentation for marine magnetotelluric and controlled source electromagnetic sounding. Geophysical Prospecting, 61 (sup 1), 505–532. Crain, E.R. [1986] The Log Analysis Handbook. Pennwell Publishing, Tulsa. Retrieved on 30 November, 2011, from: http://www.spec2000. net/00-resume.htm. Ellingsrud, S., Eidesmo, T., Jonansen, S., Sinha, M.C., MacGregor, L.M. and Constable, S. [2002] Remote sensing of hydrocarbon layers by Acknowledgements We would like thank the sponsors of the Edinburgh Time Lapse Project, Phase V (BG, BP, CGG, Chevron, ConocoPhillips, ENI, ExxonMobil, Hess, Ikon Science, Landmark, Maersk, Nexen, Norsar, Petoro, Petrobras, RSI, Shell, Statoil, Suncor, Taqa, TGS and Total) for supporting this research. We would also like to thank SchlumbergerGeoquest for the use of its Petrel and Eclipse software. Olarinre Salako acknowledges financial support from the Petroleum Technology Development Fund, Nigeria. Dennis Obidegwu, Veronica Omofoma and Erick Mackay are thanked for their technical advice and discussions. seabed logging (SBL): Results from a cruise offshore Angola. The Leading Edge, 21, 972–982, doi: 10.1190/1.1518433. Gassmann, F. [1951] Über die Elastizität poröser Medien. Vier. der Natur Gesellschaft, 96, l–23. Han, D. and Batzle, M. [2000] Velocity, density and modulus of hydrocarbon. SEG 70th Annual Meeting, Expanded Abstracts, 1892–1866. Holbrook, W.S., Páramo, P., Pearse, S. and Schmitt, R.W. [2003] Thermohaline fine structure in an oceanographic front from seismic reflection profiling. Science, 301, 821–824. Johnston, D.H. [2013] Making a Difference with 4D: Practical Applications of Time-Lapse Seismic Data. SEG Distinguished Instructor Short Course. Jun, Y., Bao-Zhi, P. and Qing-Shan, Y. [2012] Characteristics of resistiv- References Al Ali, M., Vahrenkamp, V., Elsembawy, S., Bhatti, Z., Reeder, S. L., ity log response of oil layers under polymer flooding. Applied Geophysics, 9 (2), 187–194. Clerc, N. and Wilt, M. [2009] Constraining interwell water flood Key, K. [2009] 1D inversion of multicomponent, multifrequency marine imaging with geology and petrophysics: An example from the CSEM data: Methodology and synthetic studies for resolving thin Middle East. SPE Middle East Oil and Gas Show and Conference, SPE 120558. resistive layers. Geophysics, 74 (2), F9–F20. Liang, A., Abubakar, A. and Habashy, T.M. [2011] Feasibility study of Amini, H., Alvarez, E., MacBeth, C. and Shams, A. [2012] Finding a marine CSEM for reservoir monitoring using joint 3D EM modelling petro-elastic model suitable for sim2seis calculation. 74th EAGE and fluid flow simulator. 73rd EAGE Conference & Exhibition, Conference & Exhibition, Expanded Abstracts. Andreis, D and MacGregor, L. [2011] Using CSEM to monitor production from a complex 3D gas reservoir – a synthetic case study. The Leading Edge, 30 (9), 1070–1079. Extended Abstracts. Lien, M. and Mannseth, T. [2008] Sensitivity study of marine CSEM data for reservoir production monitoring. Geophysics, 73 (4), F151–F163. MacBeth, C., Stephen, K.D. and McInally, A. [2005] The 4D seismic Archie, G.E. [1942] The electrical resistivity log as an aid in determining signature of oil–water contact movement due to natural production some reservoir characteristics. Petroleum Transactions of the AIME, in a stacked turbidite reservoir. Geophysical Prospecting, 53 (2), 146, 54–62. 183–203. Arnold, D., Vazquez, O., Demyanov, V. and Christie, M.A. [2012] Use of MacGregor, L. and Tomlinson, J. [2014] Marine controlled source water chemistry data in history matching of a reservoir model. 74th electromagnetic methods in the hydrocarbon industry: A tutorial on EAGE Conference & Exhibition, Expanded Abstracts. method and practice. Interpretation, 2, SH13–SH32. © 2015 EAGE www.firstbreak.org 45 technical article first break volume 33, April 2015 MacGregor, L., Bouchrara, S., Tomlinson, J., Strecker, U., Fan, J., Ran, Røste, T., Moen, A.S., Kolstø, E., Brekken, M., Thrana, C., Husby, O. and X. and Yu, G. [2012] Integrated analysis of CSEM, seismic and well Lescoffit, G. [2009] The Heidrun Field: monitoring fluid flow in the log data for prospect appraisal: a case study from West Africa. First Break, 30 (4), 77–82. complex Åre Formation. First Break, 27 (4), 41–49. Shahin, A., Key, K., Stoffa, P.L. and Tatham, R.H. [2010] Time-lapse MacKay, S., Fried, J. and Carvill, C. [2003] The impact of water-velocity CSEM analysis of shaley sandstone simulated by comprehensive variations on deepwater data. The Leading Edge, 22, 344–350. petro-electric modelling. SEG 80th Annual Meeting, Expanded Martin, K. and MacDonald, C. [2010] The Schiehallion Field: Applying a geobody modelling approach to piece together a complex turbidite field. Devex Conference, Abstracts. Abstracts. Shehata, A.M., Ghatas, A., Kamel, M., Aly, A. and Hassan, A. [2012] Overview of Polymer Flooding (EOR) in North Africa Fields – McGuire, P.L., Chatham, J.R., Paskvan, F.K., Sommer, D.M. and Carini, F.H. [2005] Low salinity oil recovery: An exciting new EOR opportunity for Alaska’s North Slope. SPE Western Regional Meeting, Expanded Abstracts. Elements of Designing a New Polymer/Surfactant Flood Offshore (Case Study). Society of Petroleum Engineers, 151952. Sorbie, K.S. and Mackay, E.J. [2000] Mixing of injected, connate and aquifer brines in waterflooding and its relevance to oilfield scaling. Orange, A., Key, K. and Constable, S. [2009] The feasibility of reservoir monitoring using time-lapse marine CSEM. Geophysics, 74 (2), F21–F29. Journal of Petroleum Science and Engineering, 27, 85–106. State of Alaska. [2011] Alaska Oil and Gas Conversation Commission. Retrieved: Rafle, M.Y. and Youngblood, W.E. [1987] Advances in Quantitative Reservoir Description and Monitoring in Saudi Arabia. 12th World Petroleum Congress. Abstracts. 2014, from http://doa.alaska.gov/ogc/annual/cur- rent/18_Oil_Pools/Endicott%20-%20Oil/Endicott,%20Ivishak%20 Oil/1_Oil_1.htm. Tsili, W. and Sheng, F. [2001] 3D electromagnetic modelling using finite Rider, M. and Kennedy, M. [2013] The Geological Interpretation of Well Logs (Third ed.). Rider-French Consulting Limited, Glasgow. differences. Geophysics, 66 (5), 1386–1398. Worthington, P.F. [1985] The evolution of shaly-sand concepts in reser- River and Lake Swimming Association. [2013] River and Lake Swinmming Association. Retrieved from RALSA: http://www.riverswimming.co.uk/safety.htm. voir evaluation. The log analyst, 23–40. Youngblood, W.E. [1980] The application of pulsed neutron decay time logs to monitor waterfloods with changing salinity. SPE, 957–963. Register now! 21st EUROPEAN MEETING OF ENVIRONMENTAL AND ENGINEERING GEOPHYSICS FIRST CONFERENCE ON PROXIMAL SENSING FIRST EUROPEAN SUPPORTING AIRBORNE AGRICULTURE CONFERENCE PRECISION ELECTROMAGNETICS NEAR SURFACE GEOSCIENCE CONFERENCE & EXHIBITION www.eage.org/event/environmental-engineering-2015 www.eage.org/event/proximal-sensing-2015 www.eage.org/event/airborne-em-2015 NS15-V3H.indd 2 46 6-10 September 2015, Turin, Italy 13-03-15 14:07 www.firstbreak.org © 2015 EAGE
© Copyright 2024