We present a new closed-loop feedback inflow control strategy, based on downhole measurements of self-potential (SP), and quantify the benefit of this approach in preventing or minimizing water production. Previous studies have demonstrated that downhole measurements of SP in production wells, using permanently installed electrodes, can be used to monitor and image encroaching waterfronts that are tens to hundreds of meters away from an instrumented well; field trials have demonstrated that SP data acquisition is technically feasible. However, no previous studies have investigated the use of SP measurements for closed-loop inflow control of smart wells. SP signals arise naturally during production to preserve electrical neutrality when charge separation occurs due to gradients in pressure, temperature and composition (salinity) of the formation or injected brine. We use numerical modelling as a cheap alternative to field experiments, to develop a closed-loop feedback control strategy triggered by downhole measurements of SP at a production well and compare the performance with feedback inflow control strategies based on downhole watercut measurements. Inflow control is facilitated by ‘variable’ ICVs. Our results show that closed-loop, feedback control using downhole SP measurements can yield gains in well NPV when compared to inflow control based on conventional in-well data in a single smart horizontal well. Moreover, SP-based feedback control can be used to improve well NPV over uncontrolled well production when the reservoir does not behave as predicted. Gains in NPV are close to optimal. These results are significant because they suggest a new reservoir technology that can provide a simple but effective approach to production optimization using smart wells.
Numerous studies have demonstrated that closed-loop feedback control, facilitated by smart (or intelligent/advanced) wells equipped with downhole monitoring technology and inflow control valves (ICV), can be used for production optimization during waterflooding (e.g. Brouwer et al., 2004; Naus et al., 2004; Yeten et al., 2004; Aitokhuehi and Durlofsky, 2005; Elmsallati et al., 2005; Sarma et al., 2006; Meum et al., 2008; Chen et al., 2010). However, identifying the optimal ICV settings that maximize an objective function remains challenging, especially when the reservoir behavior is uncertain. Model-based control strategies allow control actions to be taken at any time during production and can, in principle, identify the optimal solution. However, they require predictive reservoir models which are rarely available at the spatial and temporal resolution necessary to make inflow control decisions; moreover, model-based techniques may suggest control actions that are counter-intuitive and are therefore unlikely to be implemented in practice (e.g. Dilib and Jackson, 2012). Direct feedback strategies, in which control actions are taken based on (often heuristic or ad-hoc) rules relating monitoring data to ICV settings avoid the problems associated with uncertain reservoir model predictions, but control actions are taken only after some adverse change in flow or related reservoir properties (such as temperature or pressure) is measured at the well; moreover, control actions may be far from optimal (e.g. Kharghoria et al., 2002; Elmsallati et al., 2005; Ebadi and Davies, 2006; Dilib and Jackson, 2012). As yet, no study has investigated the use of reservoir monitoring technology, in which adverse changes in flow are detected or monitored at some distance from the well, for inflow control in smart wells. Yet the potential benefits of such an approach are clear: control actions can be taken before adverse changes in flow are detected at the well, but do not depend upon the predictions of an uncertain reservoir model. For example, if unwanted fluids such as water encroaching on a production well can be monitored before they reach the well, control actions can be taken in order to minimize or prevent water production (Jackson et al., 2012).
Dilib, Fahad Ahmed (Imperial College) | Jackson, Matthew David (Imperial College) | Mojaddam Zadeh, Ali (Statoil Norway) | Aasheim, Robert (Statoil ASA) | Årland, Kristine (Statoil ASA) | Gyllensten, Atle J. (Statoil ASA) | Erlandsen, Sigurd Myge (Statoil ASA)
Important challenges remain in the development of optimized control strategies for intelligent wells, particularly with respect to incorporating the impact of reservoir uncertainty. Most optimization methods are model-based and are effective only if the model or ensemble of models used in the optimization capture all possible reservoir behaviors at the individual well and completion level. This is rarely the case. Moreover, reservoir models are rarely predictive at the spatial and temporal scales required to identify control actions. We evaluate the benefit of using closed-loop control strategies, based on direct feedback between reservoir monitoring and inflow valve settings, within a geologically heterogeneous, thin oil-rim reservoir. This approach does not omit model predictions completely; rather, model predictions are used to optimise a number of adjustable parameters within a general direct feedback relationship between measured data and inflow control settings. A high-resolution sector model is used to capture reservoir heterogeneity, which incorporates a locally refined horizontal grid in the oil zone, to accurately represent the horizontal well geometry and fluid contacts, and capture water and gas flow. Two inflow control strategies are tested. The first is an open-loop approach, using fixed inflow control devices to balance the pressure drawdown along the well, sized prior to installation. The second is a closed-loop, feedback control strategy, employing variable inflow control valves that can be controlled from the surface in response to multiphase flow data obtained downhole. The closed-loop strategy is optimized using a base case model, and then tested against unexpected reservoir behavior by adjusting a number of uncertain parameters in the model but not re-optimising. We find that closed-loop feedback control yields positive gains in NPV for the majority of reservoir behaviours investigated, and higher gains than the open-loop strategy. Closed-loop control can also yield positive gains in NPV even when the reservoir does not behave as expected. However, inflow control can be risky, because unpredicted reservoir behavior also leads to negative returns. Moreover, assessing the benefits of inflow control over an arbitrarily fixed well life can be misleading, as observed gains depend on when the calculation is made.
Significant challenges remain in the development of optimized control techniques for intelligent wells, particularly with respect to properly incorporating the impact of reservoir uncertainty. Most optimization methods are model-based and are effective only if the model or ensemble of models used in the optimization capture all possible reservoir behaviors at the individual well and completion level. This is rarely the case. Moreover, reservoir models are rarely predictive at the spatial
and temporal scales required to identify control actions. We suggest that closed-loop feedback control strategies, triggered by monitoring at the surface or downhole, can increase NPV and mitigate reservoir uncertainty. We do not neglect reservoir model predictions entirely; rather, we use a model-based approach to optimize adjustable parameters in the feedback control strategies. We assess the implementation of an intelligent horizontal well in a thin oil reservoir in the presence of reservoir uncertainty, and evaluate the benefit of using different inflow control actuators in conjunction with surface and/or downhole monitoring equipment.
Four inflow control strategies are tested. The first is an open-loop approach, using a fixed control device to balance inflow along the well, sized prior to installation. The second and third are closed-loop feedback control strategies, employing intelligent completions that can be controlled from the surface. The second strategy uses on/off inflow control devices, operated according to surface flow rates and phase measurements obtained during zonal well tests. The third strategy uses
variable control devices, operated according to downhole multiphase flow meters. The fourth strategy employs a gradientbased optimization algorithm to find the dynamic optimal inflow control behavior. This strategy assumes perfect reservoir knowledge and is implemented only for benchmarking of the feedback control strategies.
Our result suggests that closed-loop control based on direct feedback between reservoir monitoring and inflow valve settings can yield close-to-optimal gains in NPV compared to uncontrolled production, even if the reservoir does not behave as predicted. Open-loop control yields significantly lower gains in NPV and is also a riskier strategy, because unpredicted reservoir behavior can lead to negative returns. Closed-loop feedback control can mitigate uncertain reservoir behavior, even when this lies outside the range of model predictions; moreover, the uncertainties which have the most significant impact on production may not be the most difficult to mitigate.
Spontaneous potential (SP) is routinely measured using wireline tools during reservoir characterization. However, SP signals are also generated during hydrocarbon production, because of gradients in the water phase pressure (relative to hydrostatic), chemical composition and temperature. We suggest that measurements of SP during production, using electrodes permanently installed downhole, could be used to detect water encroaching on a well while it is several tens to hundreds of meters away. We simulate numerically the SP generated during production from a single vertical well, with pressure support provided by water injection. We vary the production rate, and the temperature and salinity of the injected water, to vary the contribution of the different components of the SP signal. We also vary the values of the so-called ‘coupling coefficients' which relate gradients in fluid potential, salinity and temperature, to gradients in electrical potential. The values of these coupling coefficients at reservoir conditions are poorly constrained.
We demonstrate that the SP signal peaks at the location of the moving waterfront, where there are steep gradients in water saturation and salinity. The signal decays with distance from the front, typically over several tens to hundreds of meters; hence the encroaching water can be detected before it arrives at the well. The SP signal at the well is dominated by the electrokinetic and electrochemical components arising from gradients in fluid potential and salinity. Larger signals will be obtained in low permeability reservoirs produced at high rate, saturated with formation brine of low salinity, or with brine of a very different salinity from that injected. Inversion of the measured signals in conjunction with normally available reservoir data could be used to determine the water saturation in the vicinity of the well, and to regulate flow into the well using control valves in order to maintain or increase oil production and delay or prevent water production.
Downhole monitoring of streaming potential, using electrodes mounted on the outside of insulated casing, is a promising new technology for monitoring water encroachment towards an intelligent well. However, there are still significant uncertainties associated with the interpretation of the measurements, particularly concerning the streaming potential coupling coefficient. This is a key petrophysical property which dictates the magnitude of the streaming potential for a given fluid potential. The coupling coefficient can be measured experimentally, but previous studies have obtained data for sandstone cores saturated with relatively low salinity brine (less than seawater). Formation and injected brine in hydrocarbon reservoirs is typically more saline than this. Extrapolating data obtained at low salinity into the high salinity domain suggests that the coupling coefficient falls to zero at approximately seawater salinity. If this is the case, then streaming potential signals will be very small in most hydrocarbon reservoirs.
We present the first measured values of streaming potential coupling coefficient in sandstone cores saturated with brine at higher than seawater salinity. We find that the coupling coefficient is small, but still measurable, even when the brine salinity approaches the saturated concentration limit. Consistent results are obtained from two independent experimental set-ups, using specially designed electrodes and paired pumping experiments to eliminate spurious electrical potentials. We apply the new experimental data in a numerical model to predict the streaming potential signal which would be measured at a well during production. The results suggest that measured signals should be resolvable above background noise in most hydrocarbon reservoirs, and that water encroaching on a well could be monitored while it is several tens to hundreds of metres away.
Significant challenges remain in the development of optimized control techniques for intelligent wells, particularly with respect to properly incorporating the impact of reservoir uncertainty. Most optimization methods are model-based and are effective only if the model can be used to predict future reservoir behavior with no uncertainty. Recently developed schemes, which update models with data acquired during the optimization process, are computationally very expensive.
We suggest that simple reactive control techniques, triggered by permanently installed downhole sensors, can enhance production and mitigate reservoir uncertainty across a range of production scenarios. We assess the implementation of an intelligent horizontal well in a thin oil rim reservoir in the presence of reservoir uncertainty, and evaluate the benefit of using two completions in conjunction with surface and downhole monitoring. Three control strategies are tested. The first is a simple, passive approach using a fixed control device to balance inflow along the well, sized prior to installation. The second and third control strategies are reactive, employing intelligent completions that can be controlled from the surface. The second strategy opens or closes the completions according to well water cut and flow rate and individual downhole rate and phase measurements obtained from a surface multiphase flowmeter and alternating zonal well tests. The third strategy proportionally chokes the completions as increased completion water cut is measured using downhole multiphase flowmeters.
A cost-benefit analysis demonstrates that reactive control strategies always yield a neutral or positive return, whereas a passive, model-based strategy can yield negative returns if the reservoir behavior is poorly understood. While reactive control strategies enhance production and mitigate reservoir uncertainty, they may not deliver the optimum possible solution. Proactive control techniques, which additionally incorporate data from downhole reservoir-imaging sensors, may yield near-optimal gains.
The ACG Oilfield caps an elongate anticline with three culminations - Azeri, Chirag and Gunashli - and is located in the offshore Azerbaijan sector of the south Caspian Basin. This study focuses on Azeri in the south-east of the structure, which has over 8 billion barrels of oil in place. The major reservoir interval, the Pliocene Pereriv Suite, is characterized by laterally continuous layers of variable net-to-gross (NTG) deposited in a fluvial-deltaic environment. Azeri is being developed by down-dip water injection, with up-dip gas injection on the more steeply dipping central north flank. At the planned offtake rates both recovery mechanisms are expected to be stable. However, these predictions are based on reservoir models which do not explicitly capture the full range of geologic heterogeneity present in the Pereriv Suite reservoirs.
We report the first detailed assessment of the impact of large- and intermediate-scale heterogeneities on flow. Experimental design techniques have been used to rank the impact of different heterogeneities. A key finding is that communication between adjacent high and low NTG reservoir layers significantly improves recovery, providing pressure support and a route for oil production from sandbodies within the low NTG layers which would otherwise be isolated. Heterogeneity within high NTG layers has only a small impact on recovery, but heterogeneity within low NTG layers is much more significant. In most cases, the same significant heterogeneities impact both water and gas displacements, because both displacements are stable at the planned production rates.
The results are applicable to Azeri, and to similar reservoirs in the Caspian Basin. They also represent the first comparison of water-oil and gas-oil displacements in fluvial-deltaic reservoirs using 3D geologic/simulation models derived from outcrop and subsurface data.
The giant Azeri-Chirag-Gunashli (ACG) Field occurs in a large, elongate anticlinal structure located in the offshore Azerbaijan sector of the south Caspian Basin (Fig. 1A). The structure has steeply dipping limbs and contains three culminations (Azeri, Chirag and Gunashli). This paper focuses on the Azeri accumulation in the south-east of the ACG structure, which contains an estimated 8 billion barrels of oil in place. The ACG field development project is one of the largest current energy projects (US$20 billion aggregate) in the world. The Azerbaijan International Operating Company, operated by BP, has been awarded a 30 year production license for ACG which expires in 2025, and plans to bring total production in ACG to 1 million bbls oil/day by 2008.
The oil fields of the South Caspian Basin (Fig. 1A) have reservoirs in the thick (up to 7000 m) latest Miocene to early Pliocene strata of the Productive Series (Fig. 2A). These strata record multiple, high-frequency cycles of deltaic shoreline advance and retreat in response to fluctuating lake levels in the isolated South Caspian Basin1-4 (Fig. 1B). The resulting Productive Series stratigraphy is strongly layered (Fig. 2A), and sandstone-bearing intervals are bounded by laterally extensive mudstones1-4. Productive Series sandstones in the northern part of the South Caspian Basin (including the ACG Field) are quartz-rich, well rounded and well sorted, indicating deposition by the paleo-Volga River and Delta5 (Fig. 1B).
It is widely agreed that gas reservoirs with a component of water drive should be produced at high rates to minimize the volume of gas which is trapped at high pressure by the advancing water (often termed ‘outrunning the aquifer'). Yet high production rates are also associated with coning (in vertical wells) or cresting (in horizontal wells) of the encroaching water, leading to early water breakthrough. In vertical wells, the formation of an inverse gas cone means that high gas rates can be maintained post-breakthrough until almost the whole perforated interval is flowing water. However, in horizontal wells, water breakthrough is a serious threat to gas deliverability, because the inverse coning mechanism does not apply and the well rapidly loads with water. Consequently, it is not clear whether producing at high rates is the best strategy to maximize recovery in gas reservoirs developed using horizontal wells.
We investigate the risk associated with producing horizontal wells at high rates by simulating gas recovery and aquifer response over a broad range of reservoir properties and production scenarios. We find that high rates always result in lower gas recovery unless the ratio of vertical to horizontal permeability is very low, in which case water cresting is suppressed. However, there are many instances where accelerating production recovers only slightly less gas over much shorter timescales, so may be economically favorable. Rate sensitivity increases in low permeability reservoirs with thin gas columns, because these conditions increase the tendency for water cresting, and decreases in reservoirs with strong aquifer support, since water breakthrough occurs regardless of the rate at which the well is produced. Our results can be used as a reference framework to rapidly assess gas production behavior and aquifer response within a wide range of field development scenarios.
Horizontal and highly deviated wells are increasingly being used in gas field developments worldwide.1-5 Large-bore horizontal wells can deliver significantly higher gas production rates than conventional completions,2 reducing field development costs by allowing reserves to be targeted with fewer wells.4-6 However, realizing the potential of high-productivity gas wells requires an understanding of the subsurface risks to deliverability, to ensure sustained gas production, maximize profitability, and establish large-bore completions as an economically viable development option. A key subsurface risk in gas reservoirs with a component of water drive is early water breakthrough.3-5,7,8 In large-bore horizontal wells, early water breakthrough is a particularly serious threat to deliverability, because of the significant reduction in gas flow capacity associated with flowing entrained water to the surface.4,9-11 At best, substantial water production will require expensive processing facilities; at worst, it will effectively ‘kill' the well.3-5,12
Based on material balance considerations, it is widely agreed that gas reservoirs with a component of water drive should be produced at high rates. This approach (often described as ‘outrunning the aquifer') maximizes gas recovery by reducing the volume of gas which is trapped at high pressure by the advancing water.13-18 In this context, high productivity horizontal wells might be expected to make a positive contribution to gas recovery, because they can generally produce at much higher rates than vertical wells. However, material balance approaches assume that the gas-water contact (GWC) remains flat during production.13,14,16,18,20-22 Yet bottom-water drive gas reservoirs are associated with coning (in vertical wells) or cresting (in horizontal wells) of the GWC towards the well.7,23-28 Cresting occurs when viscous forces associated with pressure drawdown overcome gravity forces resulting from the density contrast between gas and water, causing a crest or cone of water to be drawn upwards towards the producing well29 (Fig. 1). Water crest behavior has been described using analytical approaches to predict a ‘critical rate' above which water breakthrough is expected,30,31 and time to water breakthrough.32,33 These approaches imply a sensitivity of gas recovery to production rate that conflicts with material balance techniques, as the severity of water cresting is increased with accelerated production, and therefore water breakthrough is expected earlier at higher rates. Water crest development also becomes more significant as the separation between the well and GWC is reduced, the horizontal reservoir permeability decreases, and the vertical permeability increases.29
Steady-state upscaling techniques are attractive because they are quick and simple to implement; unlike dynamic methods, there is no need for fine grid simulation and the upscaled properties are not case dependant. They are based on the assumption that either capillary forces (capillary equilibrium limit, CL) or viscous forces (viscous limit, VL) dominate flow. However, the reservoir conditions for which these assumptions are valid have not been clearly defined. It is generally supposed that the CL method is valid at ‘low' flow rates over ‘small' lengthscales, while the VL method is valid at ‘high' flow rates over ‘large' lengthscales. These qualitative criteria are difficult to properly apply and can be easily violated, yielding significant errors in predicted reservoir performance.
We have identified a comprehensive suite of dimensionless groups which can be used to define the validity of steady-state methods. The groups account for the effect of heterogeneity, as well as the other parameters which control the balance between capillary and viscous forces. Numerical simulations have been used to identify the range of values for these groups over which steady-state methods are valid. Our results yield a practical set of quantitative criteria which can be used to determine the validity of steady-state upscaling methods for a wide range of geological models. They capture the effects of capillary trapping and are valid regardless of fluid mobility, wettability or end-point saturation.
We test our criteria against three realistic models of small-to intermediate-scale geological heterogeneity. We find that the criteria do a good job of predicting the range of validity for each method, and are conservative in all cases, suggesting that if they are met then steady-state upscaling techniques can be applied with confidence, and may still be valid for slightly less restrictive conditions. However, in the models investigated, we find that the validity of the CL method is restricted to very low flow rates which are unlikely to be encountered in most production scenarios. This is because the CL method overestimates the amount of capillary trapping. In general, VL upscaling is valid over a much more reasonable range of reservoir flow rates.
Copyright 2005, Society of Petroleum Engineers This paper was prepared for presentation at the 2005 SPE Annual Technical Conference and Exhibition held in Dallas, Texas, U.S.A., 9 - 12 October 2005. This paper was selected for presentation by an SPE Program Committee following review of information contained in a proposal submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300 words; illustrations may not be copied.