Vazquez_1_2013_Produced Water Chemistry History Matching in the Janice Field SPE-164903-MS-P

Produced Water Chemistry History Matching in the Janice Field
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
    SPE 164903 Produced Water Chemistry History Matching in the Janice Field O. Vazquez 1 , C. Young 2 , V. Demyanov 1 , D. Arnold 1 , A. Fisher 2 , A. MacMillan 2 , Mike Christie 1 ; 1 Heriot-Watt University, 2 Maersk Oil Copyright 2013, Society of Petroleum Engineers This paper was prepared for presentation at the EAGE Annual Conference & Exhibition incorporating SPE Europec held in London, United Kingdom, 10–13 June 2013. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright. Abstract Produced Water Chemistry data (PWC) is the main source of information to monitor scale precipitation in oil field operations. Chloride concentration is used in order to evaluate the seawater fraction of the total produced water per producing well and is included as an extra history matching constraint to revaluate a good conventionally history matched reservoir model for the Janice field. Generally PWC is not included in conventional history matching and this approach shows the value of considering the nature of the seawater injection front and the associated brine mixing between the distinctive formation water and injected seawater. Adding the extra constraint resulted in the re-conceptualization of the reservoir geology between a key injector and two producers. The transmissibility of a shale layer is locally modified within a range of geologically consistent values. Also, a major lineament is identified which is interpreted as a NW-SE trending fault, whereby the zero transmissibility of a secondary shale in the Middle Fulmar is locally adjusted to allow cross-flow. Both uncertainties are consistent with the complex faulting known to exist in the region of the targeted wells. Other uncertainties that were carried forward to the assisted history matching phase included: water allocation to the major seawater injectors; thermal fracture orientation of injectors and the vertical and horizontal permeability ratio (kv/kh) of the Fulmar formation. Finally, a Stochastic Particle Swarm Optimization (PSO) algorithm is used to generate an ensemble of history matched (HM) models using seawater fraction as an extra constraint in the misfit definition. Use of addition data in history matching has improved the srcinal good history matched solution. Field Oil Production Rate is interpreted as improved over a key period and although no obvious improvement was observed in Field Water Production Rate, Seawater fraction in a number of wells was improved. Introduction Scale precipitation is a major flow assurance problem where minerals precipitate and further nucleate on surfaces such as production tubing, reservoir pore or pore throats, perforation intervals and surface facilities. These deposits can inhibit well inflow and outflow performance which may result in costly well interventions, downtime or ultimately abandonment. The sampling of produced water chemistry and wellbore monitoring surveys can however aid oilfield scale detection and its management (Carbone et al., 1999). One of the most common occurring oilfield scales is sulphate minerals, which form due to the mixing of formation water (rich in cations such as Ba, Sr, Mg, Ca) and injected seawater rich in sulphate ions. Predicting the location of the front and hence sulphate mineral deposition is an intricate process; its prediction requires accurate modelling of the seawater and formation water mixing front and associated breakthrough time. The use of produced water chemistry, due to their clear distinctive chemistries, has been used for seawater fraction determination in a number of techniques, such as reacting ions method (Ishkov et al., 2009) and Multivariate Analysis (Scheck and Ross, 2008). In this particular study, Chloride ion concentration is considered as it is one of the most common methods used in the oil industry. Barium sulphate (barite) is relatively acid insoluble and is considered as one of the most challenging and expensive scales to remove, where its precipitation is the result of fluid-fluid incompatibility. Pressure maintenance and secondary oil recovery by the injection of seawater in to the reservoir is common in field development strategies and the interaction of equilibrated, Barium (Ba 2+ ) rich formation water, with the injected seawater rich in sulphate (SO 42- ), can form barium sulphate scale (Puntervold and Austad, 2007). It is therefore favourable to predict and prevent sulphate scale by sulphate reduction prior to injection or regular scale inhibition to prevent its occurrence rather than costly removal by well intervention or at worst,  2 SPE 164903 abandonment. The same is also true for the Janice field, where the formation water is rich in Calcium (Ca 2+ ), forming an insoluble scale with the sulphate in the injected seawater. Conventional history matching is a standard industry practice, whereby the adjustment of physical parameters, such as the permeability and porosity of the geological model, is made in order to replicate the production field observations. A reasonably well matched model is a necessity for proficient reservoir management, as it is intuitive to draw confidence from the ability of the model to replicate the past, and, therefore, use it as an aid for decisions regarding the reservoir future performance, where associated facilities can be optimised. Bypassed or stranded oil can be targeted, and also water and gas breakthrough time can be anticipated. This procedure of adjusting the model is generally carried out through the lifetime of a reservoir as further data is gathered and a model continually updated to retain a match. It is well understood that due to the limited and spatially restricted confidence in data pertaining to geologically complex reservoirs, significant uncertainty exists in any reservoir model. It is also well understood that there is no unique solution, due to the fact that a number of different configurations of geological parameters can yield multiple well matched models, each with different forecasts of reservoir performance. It is therefore essential to quantify the uncertainty related to multiple well history matched realisations (multiple local minima) of relatively geologically consistent models (Hajizadeh et al., 2010). For the purpose of this study, particle swarm optimization (PSO) is utilised where it has been successfully shown to find well history matched models of synthetic and real-life case-studies quickly (Mohamed et al., 2010a; Mohamed et al., 2010b), while retaining model diversity for the purpose of uncertainty quantification of forecasts. PSO is a stochastic sampling algorithm which is not limited to integers and therefore has the advantage over classical genetic algorithms where it samples the complete range of variability (Arnold et al., 2012). A set of N particles, initially randomly generated and described by laws of motion, solve the optimization problem by convergence towards the best solution an individual particle (pbest) has seen from a population, and also the best solution from the best generation of particles (gbest). This avoids trapping in local minima such as in conventional gradient based algorithms (Kelley, 1999). PWC is not conventionally included in history matching, although the amount of data used is crucial to improve conditioning of an ill-posed inverse problem. Therefore, integration of PWC data is seen as a unique opportunity to increase the justification and confidence of the predictions based on HM models on a synthetic modified PUNQ-S3 case study. A pilot study in (Arnold et al., 2012) showed improvement in HM with PWC. The present paper extends further the methodology applied to a real field case. Produced Water Chemistry (PWC) To include the PWC in the reservoir history matching exercise, the seawater fraction was calculated as a function of the Chloride concentration. This is a common technique, where considering Cl -  is a non-reacting ion, it is expected that the concentration of Chloride follows a linear behaviour with respect to seawater fraction (Braden et al., 1993). The linear relationship for the mixture of seawater and formation water was determined for Chloride concentrations between the 116,950 mg/l and 19,700 mg/l endpoints, Janice formation water and seawater concentration, respectively. A schematic of the relationship between chloride concentration and therefore historical and simulated PWC is shown below in Figure 1. History Match Assessment of the Original Reservoir Model Generally, conventional reservoir history matching considers parameters such as well gas rate, oil rate and bottom-hole pressure. Other authors have proposed the inclusion of other observed data as extra constraints, such as time lapse seismic data (Kazemi et al., 2011; Stephen et al., 2009) and PWC (Arnold et al., 2012; Huseby et al., 2005). Water is present in sedimentary deposits where oil is found, which may dissolve minerals present in the formation, so formation waters have distinguishable chemistry (Ishkov et al., 2009), which may be traceable. In this study, two main types of water are considered, formation (including connate and aquifer water, as normally aquifer and formation water is assumed to have the same chemistry) and injection water (present in water-flooding as a secondary recovery mechanism). PWC is the main source of information for the detection of scale precipitation in oil field operations, where it is common practice to analyse the composition of formation waters present in the reservoir and injected waters. One of the most common scales occurring and one of the most difficult to treat is BaSO 4 , which is formed when seawater rich is SO 4  ions mixes with formation water, rich in Ba 2+  ions. Below, an assessment of the history match exercise will be carried out. First, considering solely observed produced water, oil and gas. Then the history matched model will be assessed using PWC, which is used to calculate seawater fraction. Assessment without Produced Water Chemistry - Conventional The base case reservoir model is considered to conventionally history match against observed oil, water and gas production rates. The resulting history matched model shows a field wide good match with respect to observed field oil production rate (Figure 2), field water production rate (Figure 3) and field gas production rate (Figure 4. A detailed assessment of this history matched reservoir model was reviewed by the Operator. Assessment with Produced Water Chemistry - Seawater Fraction To assess the value of adding PWC as an extra constraint, the observed and calculated seawater fraction is compared using the srcinal reservoir model, which was conventionally history matched. Three producers in particular, Wells A, B and C,  SPE 164903 3 showed good matches to the total water cut, however they provided poor matches to the observed seawater fraction, shown in Figure 5, 6 and 7, respectively. The poor seawater fraction match in Well A has important implications, as the simulated results suggest that Well A is producing ~50% seawater, but based on the observed PWC it should mainly be formation water. Although well B showed a reasonable match to the water cut, producing solely formation water, the observed seawater fraction is above 80% (Figure 6). Finally, Well C did not capture the seawater breakthrough time. In conclusion, the srcinal reservoir model did not represent adequately the water (injected and formation water) flow paths to these producers. Therefore, a re-evaluation was required that formed the basis of geological uncertainty identification and subsequent new parameterisation. Reservoir Uncertainty Identification, Quantification and Parameterisation After further investigation, two different kinds of uncertainties were identified, namely geological and water allocation. Geological uncertainties consist of the sealing potential of two shale layers in the Middle Fulmar, thermal fracture orientation in some of the injectors and finally the vertical and horizontal permeability ratio. Water allocation uncertainty is largely related to the water split between injectors. A uniform distribution was chosen between the ranges of possible values for each uncertain parameter, since each value was interpreted as equally likely and finally, to allow model diversity. The ranges were constrained by geological or engineering evidence when available. Geological Uncertainties  Middle Fulmar Shale Layer 1 There is uncertainty with regards to the sealing potential of a shale layer in the eastern regions of the field (Figure 8). Considering Well B is completed above the sealing shale and the closest injector is structurally and stratigraphically lower, there exists no significant pressure differential to encourage vertical movement of injected seawater. In order to initiate communication the transmissibility of this sealing layer had to be adjusted. Although it did not significantly improve the contribution of injected seawater to Well B, it had a positive effect on Well C, where an increased contribution from injected sea water was observed during initial sensitivity runs.  Middle Fulmar Shale Layer 2 This second shale layer is located close to Well B, and prevented vertical communication with its closest injector. Adjusting the transmissibility provided a good communication between the surrounding injectors of Well B and C. After this adjustment, there was no significant effect on the water and oil production rates. In addition, the pressure field also remained relatively unaltered away from this region, which was a necessity to retain good well matches at field scale. Thermal Fracture Orientation, Length and Permeability Due to the fact that injection is above fracture pressure, and the temperature contrast between the reservoir rock and injected seawater, it is interpreted that induced fracture wings are present. Fracture length has been limited to data from analogue literature and pressure fall-off data provided by the Operator. Fracture morphology may have a significant impact on the injected water flow paths and associated well water production rates; therefore it is reasonable to consider the uncertainty in the fracture orientation where the present day maximum stress direction is known to be variable in the region. Fractures are assumed to only propagate in the X and Y directions to retain consistency with the srcinal reservoir grid.  Absolute Vertical permeability, Kv/Kh Based on the facies dependent Kv/Kh values obtained from the Reservoir Field Development Review, Kv/Kh ranges were applied to the Upper Fulmar, Upper-Middle Fulmar, Lower-Middle Fulmar and Lower Fulmar, in order to keep a reasonable number of parameters. Water Allocation Uncertainties and Parameterisation  Injector A/B water split Injected water is pumped to the seabed via a riser to a manifold. Water distribution after the manifold to individual wells however is uncertain, which is largely due to a lack of direct injection well testing. Recent information gathered by the Operator suggested that Injector A may take a larger fraction of the total Injector A/B water split than is currently assumed. First considering that Injectors A and B are a significant distance apart, and second, that they provide the majority of pressure support through the field’s history and finally that these injectors are completed through the majority of the geological formations, Injector A/B water split uncertainty could have a significant impact on the history matching results.  Injection Well Uptime and Downtime For completeness, the injection wells uptime, downtime and plug failure records were cross-checked with the srcinal reservoir model schedule, which was updated accordingly.  4 SPE 164903 PWC History Matching and Uncertainty Quantification Misfit definition Choice of the misfit definition is one of the crucial tasks in history matching. A conventional misfit definition commonly used is the least squared norm:  󰀽  󰀨 −󰀩  󰀲 󰀬󰀬󰀬󰀬  Where W   is the number of wells, V   is the number of production variables and T   is the number of time steps for each. This misfit definition is then used in the likelihood model for the posterior inference, see below,  L=exp(-M) Use of such a likelihood model implies that the model errors are independently normally distributed. However, this may not be the case and the choice of σ  ijk   in the denominator becomes vital. Generally speaking the σ   corresponds to the level of confidence attached to every observation, in other words it describes how close the match is expected to approach the observation. In this case the σ  ijk    does not correspond just to the measurement device error, but reflects the overall uncertainty associated with the observation (e.g. due to averaging over a period of time, accumulation, allocation, calibration, etc.). A common practice is to choose a constant σ  ijk   throughout assuming they are independent and identically distributed. This is a statistically sound assumption. However, this is not always practical as it does not reflect larger uncertainty of higher observed values. Another common approach is to set the value of σ  ijk    to be a fraction of the observed value (or in some proportion to it), which inevitably leads to propagation of the correlation from the reservoir response into the match errors. Therefore, a sensible guidance is to assess the value of σ  ijk   with the desirable match error, which can be obtained based on engineering judgment. Also, it is important to take into account the correlation between the errors, which is generally done through a full covariance matrix. Autocorrelation or the error through the time steps occurs due to the imposed influence of the numerical simulation and the periods of stationary model behaviour. Thus, correlation between the errors from one time step to another means that the σ  ijk    are no longer independent and the impact of each single error should be mitigated by a weighting factor. The weighting factor mitigates the impact of the mismatch from multiple history observations with correlated errors. Statistical correlation analysis with an autocorrelations or variogram function provides a way to measure the correlation in the errors. Figure 9 shows an absence of correlation in the errors for observations from one of the wells with the corresponding variogram nugget behaviour. The errors for the observations from another well (see Figure 10) demonstrate periodic correlations and its period can be measured on the variogram. Based on this information the weight for this series of data is chosen inversely proportionate to the number of points within the correlation range. This procedure of computing the weights is performed for all matched variables across all the production wells. It appeared that water production error is highly correlated in all the wells, while the tracer (i.e. seawater fraction) error is correlated only in some of the wells, leaving the error of the sparse pressure data uncorrelated. Factoring this information into the misfit definition leads to mitigation of the influence of relatively large amounts of water production data in favour of the tracer water production, so the value of the latter becomes more important in history matching. This technique also allows decreasing the range of the operating misfit values from tens of thousands (for thousands of observed history data) to hundreds. PWC History matching With the new updated parameterisation based on the information provided by the PWC and the misfit definition described above, a new history matching exercise was performed including the PWC as a constraint. The same three producers, Wells A, B and C which showed a good match to the total water cut, provided poor matches to the observed seawater fraction. The overall match for Well A (Figure 11), is significantly improved as the observed seawater fraction and watercut is extremely well matched. Well B is slightly better matched (Figure 12 and finally, Well C is slightly better matched, where the seawater breakthrough time is accurately captured (Figure 13). Uncertainty Quantification In this section the uncertainty in oil and water production, including seawater, was predicted for the following three producers, Wells B, D and E. These three producers would be potentially actively producing in future years in the Janice field. The uncertainty quantification is based on multiple model realizations, which provides an ensemble of good history matched models, which then determine the uncertainty of the well fluid production. These calculations provide a Bayesian confidence interval (P10-P50-P90) in time for oil production, which can be used to evaluate the value of the well. Then, this information is combined with the water production and the seawater fraction predictions to evaluate the scale risk associated
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks