In Argentina, over the last decade mature field management has been of paramount importance in ensuring economic sustainability for many oil field operators. Success requires solving a specific suite of problems comprising mixed generations of technology (in particular logging tools), long and complex well histories and often the sheer size of the dataset. The field in this case study has been in production since 1964 reaching its maximum production capacity of 32000 bpd during 1969, after which decline began. Consequent depletion from 140kg/cm2 to 20kg/cm2 drove the need for pressure support that was achieved through waterflooding which was implemented in two major campaigns during the 70's and the 80's. A total of 354 wells comprise the historical dataset with recent re-drills, extensions and infills bringing the total well count to 422. Re-evaluation of the remaining target via a series of studies carried out between 2006 and 2009 indicated an attractive opportunity for 7 spot waterflooding and saw the commencement of a massive re-development of the field. Behaviour of recent wells has been worse than predicted. This deviation from expectation initiated a series of studies to better characterize the reservoir with the objective of re-defining targets for incremental development. Associated with these studies new geological and dynamic models were built using re-evaluated historical data integrated with information from 68 new wells and 8 new cores. In particular, the impact of textural variation and thin bed architecture on the meso-scale oil distribution was assessed allied with a range of different techniques to identify macro-scale compartmentalization. The result was an integrated model that enabled comprehensive re-evaluation of the remaining targets. The approach used in this study to identify and characterize thin beds in this type of setting, define the impact on OOIP and determine the remaining oil in place in order to evaluate opportunities can assist many operators who experience various challenges associated with developing mature acreage.

A polymer injection pilot started in 2015 in the Rayoso Formation, Desfiladero Bayo field (Neuquen Basin, Argentina) A detailed monitoring of the pilot was planned and the amount of data anticipated led to the construction of a reservoir model with a high level of detail in order to compare the results to prognosis and to reduce the uncertainty to the business case forecast. A specific complication was that the Rayoso Fm. had been produced to date in a commingled way with no exclusive production data before the pilot. Subsequently other data was needed to constrain the simulation model. Clearly, when the pilot started, the Rayoso Fm. was not in a virgin state (after ten years of waterflooding), so RFT data, saturation logs of infill wells, injection logs, tracer and geochemical data were used instead, adding production and injection exclusively from the Rayoso Fm. in the new wells and workovers subsequent to the pilot starting. The history match of the water injection stage of the pilot showed things that were not initially obvious; one was that the production of one of the pilot producer wells was not consistent with the log response and core data. It was subsequently established that errors were occurring in the production tests of this well in the field that were then corrected. Breakthrough times for water were consequently history matched and it was established that even after polymer injection started, water cut was expected to initially rise, this being key for expectation management. As the pilot has progressed, the tracer dosed with the polymer was seen before polymer, this suggesting either the presence of a thief zone or as a methodology for estimating the polymer retention. Additionally the mismatch of tracers between the simulation and actual data was used to estimate the real injection rates into the Rayoso Fm. of commingled injectors lying near the pilot area. In this paper we discuss how a detailed pilot simulation model can be constructed and constrained when production data specific to the formation is absent. Subsequently, when field results are different to expectation, simulation can be used as tool to establish the probable cause of the deviation from prognosis and hence identify remedial action if required. Finally, the simulation model was key to manage expectations; unrealistic expectations are one of the leading causes of pilots failing.

A semi-regional 3D model of the Rayoso Formation in the Neuquén basin, Argentina, has been used to predict that several millions of barrels of additional oil can be recovered by polymer flooding in six fields of YPF S.A. As a result of static and dynamic modelling, two polymer flood pilots in separate fields have been approved. The first pilot started polymer injection in July 2016, while the second one is expected to commence in late 2017. This paper describes the approaches currently used and proposed for surveillance to ensure that the pilot objectives are met and to demonstrate the feasibility of field-scale application of polymer flooding across the Rayoso formation. A multidisciplinary integrated surveillance plan has been implemented to prevent the pilot from failing due to avoidable causes. Data on polymer quality monitoring, tracer surveys, interwell tracers, fall-off tests, and production and injection review are regularly fed into simulation models. We describe how this is achieved while limiting additional staffing and equipment costs. The pilot test discussed here lies in a multi-reservoir field where oil production is commingled with two deeper reservoirs. In order to make the oil production response to polymer injection directly detectable, the central producing wells in the pilot were recompleted to produce only from the Rayoso formation. Since there was no practical possibility to do the same for the second-line (offset) production wells, geochemistry techniques were used to identify the oil produced from each formation and thus estimate incremental produced outside of the pilot area. Although the Rayoso formation has been exploited for more than 30 years, only recently has it been given priority as a production target. At the beginning of the study, several major uncertainties exisited that could adveesely affect the potential success of polymer flooding in this reservoir, especially those regarding the initial and remaining oil saturation in the formation. As a result of the appraisal and surveillance plan applied, key project uncertainties were addressed or assessed by monitoring potential deviations of a set of variables from the project prognosis in real time. Interwell tracers and pressure fall off tests during the baseline waterflood provided data to reduce the uncertainty in the presence of possible flow barriers in the pilot area. Frequent production tests and continuous well bottomhole pressure data enabled calibration of the simulation model and to predict the behavior of the declining phase of the baseline waterflood with resasonable confidence. A single well polymer injectivity test with back flow, performed in one of the pilot injectors, showed that the selected polymer at the target concentration was not significantly degraded in the reservoir.

No abstract

Improved oil recovery is more sensitive to sub-log scale heterogeneity than primary recovery processes. Determination of porosity, net-to-gross, saturation and permeability, both magnitude and direction, at scales below log resolution is often required in secondary and tertiary processes and can have an impact on the estimation of original hydrocarbon in place, by-passed oil and recovery factors. Conventional log interpretation and conventional static modeling workflows may not capture sub-scale log heterogeneities, often due to the averaging involved in the upscaling process. Even though log values have been generally calibrated to plug scale (cm3) measurements, they are averaged over volumes much larger than their scale(s) of key heterogeneity. This paper compares the simulated response derived from conventional and from effective property and lamina-scale modeling in three different oil bearing formations in the Argentinian Neuquen Basin, namely Quintuco, Sierras Blancas and Lotena Formations. All three Formations are candidates for waterflood developments and enhanced recovery processes options are being examined. Permeability, porosity, saturation and net-to-gross ratio were determined by constructing high resolution models that enabled capturing sub-log scale heterogeneities. These models were subsequently upscaled at representative elementary volumes and the derived effective properties were used in building simulation models through different upscaling methodologies. Results of the two scale models for each reservoir are compared. Strengths and weakness of the applied methodology, as applied to these specific cases, are evaluated and discussed.