A sloping depositional surface, known as clinoform, is commonly associated with prograding strata deep into water and this surface can be imaged with seismic (J.L.Rich, 1951). During a sea level drop carbonate sediment factories tend to shut down and results in periods of non-deposition. These clinoform surfaces can be cemented, resulting in a partially sealing effect. Therefore characterizing clinoforms is crucial to better understand the reservoir dynamics and hydraulic communication throughout the field. In the Karachaganak field, Wireline Formation Pressure acquisition along the sub-horizontal well section has played a major role in the identification of semi-transmissible flow restrictions in the field. In particular, plots of depth corrected pressure measurements against distance from wellhead showed clear discontinuities confirming the existence of pressure baffles. Once these clinoforms are identified, their properties have to be calibrated against the measured pressure data, process that historically has been implemented by manual trial-and-error approach which is time-consuming and often frustrating. This paper proposes a methodology based on assisted history matching solutions to constrain the properties and sealing degree of clinoform's regions to both Wireline Formation and Static Bottom-Hole Pressure data. The proposed approach allows engineers to integrate geological and reservoir engineering workflows into a single model driven by state-of-the-art history matching optimization techniques. This makes possible to systematically sensitize the properties of the clinoforms assuring consistency between static and dynamic models. As a result, observed pressure has been matched, and hence the characterization of the clinoforms properties was considerably improved in a short time compared with the timeframe required by the traditional manual approach. The presented workflow is a valuable tool to set methods and gain experience using assisted history matching techniques, and furthermore, it contributes to a change in history matching philosophy by semi-automating laborious tasks achieving faster and more physically coherent solutions.

Well testing in gas-condensate and volatile oil reservoirs with flowing bottomhole pressure below the saturation pressure creates multiphase flow in the reservoir, resulting in relative permeability reduction and a rate-dependent skin factor. In lean gas condensate reservoirs, the wellbore skin effect calculated using single-phase pseudo-pressures has often been found to be constant or even to decrease with increasing gas rates, instead of increasing as in dry gas reservoirs. This behaviour has been tentatively attributed to capillary number effects compensating for condensate blockage and inertia effects, but no detailed study of this behaviour has appeared in the literature to-date. This paper investigates wellbore skin behaviours in lean and rich gas condensate reservoirs and in volatile oil reservoirs by using compositional simulation with capillary numbers and non-Darcy flow to generate well test data. It is shown that below saturation pressure, gas condensate well test analyses with single-phase pseudo-pressures and volatile oil well test analyses with pressures do not correctly estimate the rate-independent wellbore skin effects and the non-Darcy flow coefficients, whereas analyses with two-phase pseudo-pressure do, provided that non-Darcy and capillary number effects are included in the two-phase pseudo-pressure calculations. In gas condensate reservoirs below the dew point pressure, the rate independent skin factor and the non-Darcy flow coefficient calculated with two-phase pseudo-pressures are identical to the corresponding values calculated above the dew point pressure with single-phase pseudo-pressures. These simulation results are applied to actual field data.

This paper presents a successful case study of building a robust “in-situ” Integrated Production Systems Model (IPSM) directly in Reservoir simulator (ECLIPSE) without adding complexity of third-party network simulators (e.g. PETEX) and scripts. Based on the case study, ECLIPSE based IPSM model has proved to add incremental value over standalone ECLIPSE model by providing more accurate production profiles and reducing development costs by avoiding “under-designing” or “over-designing” of project facilities. High level group tree had been replaced by the detailed architecture that is well aligned with the actual surface field network layout. Considering the large number of production nodes, HFP (Horizontal flow performance) tables for existing and future pipelines generated and updated using the in-house developed automatic workflow based on OPEN SERVER platform and linked to the nodal analysis software. Flow rate-pressure calibration of simulated data to observed data across all 400+ manifolds, pipes and well chokes are performed beforehand based on regularly updated data from internal real-time production data management system. The current reservoir pressure at Karachaganak gas condensate field (KGK) is below the saturation point. Producing Gas-Oil ratio and water cut at well and field level increases over production life that in turn creates bottlenecks in the surface gathering system. The production system includes around 200 production and injection wells connected to three inter-flowing processing facilities, which simultaneously constrained by gas and water processing capacities, gas compression for re-injection and contractual gas sale obligations. Recent practice for assessment of oil plateau extension projects in the field was to use either standalone ECLIPSE model option or full-scale Integrated Asset Model (IAM) based on integration of RESOLVE, GAP and ECLIPSE. ECLIPSE based IPSM model reproduced historical pressure losses of the surface pipelines at similar resolution of IAM model at much shorter computational runtimes. It also improved the rate-pressure transition from history to forecast in comparison with standalone results. In addition, new model setup helped to identify bottlenecks in the well flowlines and to propose solutions by properly rerouting the wells. Furthermore, significant CAPEX cost savings achieved by finding optimal size and number of trunklines from manifolds to Processing plants. The novelty of this production-forecasting tool is the generation and integration of detailed surface gathering system into dynamic model without using third-party network simulators and to bring its accuracy to the levels sufficient not only for the long-term forecasting but also for medium-term and short-term optimization work. Mainly, this study is vital for a big oil and gas fields, where high precision is the critical for decision making and production success.

This paper will present a case study of an Uncertainty Analysis recently performed for the Kashagan Field. This is the first full scale uncertainty work performed on the field since it came online in September 2016. The paper will describe unique challenges given Kashagan peculiarities and describe methods and approaches taken to address those. A multi-scenario modeling workflow has been utilized to fully explore subsurface uncertainties in the decision space. First, uncertainties that were believed to have an impact on field performance based on the available data and learnings from previous studies were identified and ranked. Then, so-called categorical parameters were combined by grouping geologically relatable parameters into several scenarios with differentiating behaviors which were checked by performing screening simulation runs. This was followed by Design of Experiment (DoE) runs (Placket-Burman and Latin Hypercube), proxy modeling and Monte-Carlo analysis to aid selection of high, mid, and low models. More than 400 simulation runs were generated resulting in a wide range of outcomes for key performance indicators, enabling the selection of high, mid, and low model realizations to evaluate future field development decisions. The approach of combining categorical parameters into multiple scenarios followed by DoE runs offered several advantages including full sampling of the uncertainty space and clearer link between geologic input and dynamic output. It also allowed to get maximum information using the lowest number of simulation runs. Communication techniques such as plumbing diagrams representing geologic scenarios were effective in discussion of study results amongst experts.

No abstract