This paper discusses the use of a novel data-driven method for automated facies classification and characterization of carbonate reservoirs. The approach makes an extensive use of wireline and while drilling electrical borehole image logs and provides a direct and fast recognition of the main geological features at multi-scale level, together with secondary porosity estimation. This embodies an unbiased and valuable key-driver for rock typing, dynamic behavior understanding and reservoir modeling purposes in these puzzling scenarios.
The implemented methodology takes advantage of a non-conventional approach to the analysis and interpretation of image logs, based upon image processing and automatic classification techniques applied in a structural and petrophysical framework. In particular, the Multi-Resolution Graph-based Clustering (MRGC) algorithm that is able to automatically shed light on the significant patterns hidden in a given image log dataset. This allows the system to perform an objective multi-well analysis within a time-efficient template. A further characterization of the facies can be established by means of the Watershed Transform (WT) approach, based on digital image segmentation processes and which is mainly aimed at quantitative porosity partition (primary and secondary).
The added value from this data-driven image log analysis is demonstrated through selected case studies coming from vertical and sub-horizontal wells in carbonate reservoirs characterized by high heterogeneity. First, the MRGC has been carried out in order to obtain an alternative log-facies classification with an inherent textural meaning. Next, the WT-based algorithm provided a robust quantification of the secondary porosity contribution to total porosity, in terms of connected vugs, isolated vugs, fractures and matrix contribution rates. Finally, image log-facies classification and quantitative porosity partition have been integrated with production logs and pressure transient analyses to reconcile the obtained carbonate rock types with the effective fluid flows and the associated dynamic behavior at well scale.
The presented novel methodology is deemed able to perform an automatic, objective and advanced interpretation of field-scale image log datasets, avoiding time-consuming conventional processes and inefficient standard analyses when the number of wells to be handled is large and/or in harsh circumstances. Moreover, secondary porosity can be proficiently identified, evaluated and also characterized from the dynamic standpoint, hence representing a valuable information for any 3D reservoir models.
In this work we discuss the successful application of our previously developed automated scenario reduction approach applied to life-cycle optimization of a real field case. The inherent uncertainty present in the description of reservoir properties motivates the use of an ensemble of model scenarios to achieve an optimized robust reservoir development strategy. In order to accurately span the range of uncertainties it is imperative to build a relatively large ensemble of model scenarios. The size of the ensemble is directly proportional to the computational effort required in robust optimization. For high-dimensional, complex field case models this implies that a large ensemble of model scenarios which albeit accurately captures the inherent uncertainties would be computationally infeasible to be utilized for robust optimization. One of the ways to circumvent this problem is to work with a reduced subset of model scenarios. Methods based on heuristics and ad-hoc rules exist to select this reduced subset. However, in most of the cases, the optimal number of model realizations must be known upfront. Excessively small number of realizations may result in a subset that does not always capture the span of uncertainties present, leading to sub-optimal optimization results. This raises the question on how to effectively select a subset that contains an optimal number of realizations which both is able to capture the uncertainties present and allow for a computationally efficient robust optimization. To answer this question we have developed an automated framework to select the reduced ensemble which has been applied to an original ensemble of 300 equiprobable model scenarios of a real field case. The methodology relies on the fact that, ideally, the distance between the cumulative distribution functions (CDF) of the objective function (OF) of the full and reduced ensembles should be minimal. This allows the method to determine the smallest subset of realizations that both spans the range of uncertainties and provides an OF CDF that is representative of the full ensemble based on a statistical metric. In this real field case application we optimize the injection rates throughout the assets life-cycle with expected cumulative oil production as the OF. The newly developed framework selected a small subset of 17 model scenarios out of the original ensemble which was used for robust optimization. The optimal injection strategy achieved an average increase of 6% in cumulative oil production with a significant reduction, approximately 90%, in the computational effort. Validation of this optimal strategy over the original ensemble lead to very similar improvements in cumulative oil production, highlighting the reliability and accuracy of our framework.
Identification of tidal channels fairways is key for predicting behavior of areas at higher risk to water breakthrough or otherwise have a significant impact on the development and monitoring of reservoir performance. However, tidal channels in carbonates are not often easily characterized using conventional seismic attributes. It is important to decipher the complexity of the carbonate tidal channel architecture with integrated multisource data and a variety of approaches.
In this paper, petrological characteristics and petrographic analysis is conducted on well logs and validated carefully using core data. Then, the second step is to compare the carbonate channel systems with modern analogue in Bahama tidal flat and outcrop scales in Wadi Mi'Aidin (Northern Oman). Thereafter, the supervised probabilistic neural network (PNN) and linear regression method were undertaken to detect an additional channel distribution.
The relationship of high porosity with low acoustic impedance appeared mostly in the channel facies which reflects good reservoir quality grainstone channels. Outside these channels, the rock is heavily mud filled by peritidal carbonates and characterized by a high acoustic impedance anomaly with low quality of porosity distribution. The new observation of PNN porosity volume revealed a lateral distribution of the Mishrif carbonate tidal channels in terms of paleocurrent direction and the connectivity. Additionally, the prior information from core data and the geological knowledge indicate a good consistency with classified lithology. These observations implied that Mishrif channels consist of a wide range of lithology and porotype fluctuations due to the impact of depositional environment.
The work enables us to provide a new insight into the distribution of channel bodies, and petrophysical properties with quantification of their influence on dynamic reservoir behavior of the main producing reservoir. This work will not only provide an important guidance to the development and production of this case study, however also deliver an integrated work path for the similar geological and sedimentary environment in the nearby oil fields of Southern Iraq.
With the advent of high-resolution methods to predict hydraulic fracture geometry and subsequent production forecasting, characterization of productive shale volume and evaluating completion design economics through science-based forward modeling becomes possible. However, operationalizing a simulation-based workflow to optimize design to keep up with the field operation schedule remains the biggest challenge owing to the slow model-to-design turnaround cycle. The objective of this project is to apply the ensemble learning-based model concept to this issue and, for the purpose of completion design, we summarize the numerical-model-centric unconventional workflow as a process that ultimately models production from a well pad (of multiple horizontal laterals) as a function of completion design parameters. After the development and validation and analysis of the surrogate model is completed, the model can be used in the predictive mode to respond to the "what if" questions that are raised by the reservoir/completion management team.
Saputelli, Luigi (ADNOC) | Celma, Rafael (ADNOC) | Boyd, Douglas (ADNOC) | Shebl, Hesham (ADNOC) | Gomes, Jorge (ADNOC) | Bahrini, Fahmi (Frontender Corporation) | Escorcia, Alvaro (Frontender Corporation) | Pandey, Yogendra (Prabuddha)
Permeability and rock typing are two of the main outputs generated from the petrophysical domain and are particularly contributors to the highest degree of uncertainty during the history matching process in reservoir modeling, with the subsequent high impact in field development decisions. Detailed core analysis is the preferred main source of information to estimate permeability and to assign rock types; however, since there are generally more un-cored than cored wells, logs are the most frequently applied source of information to predict permeability and rock types in each data point of the reservoir model.
The approach of this investigation is to apply data analytics and machine learning to move from the core domain to the log domain and to determine relationships to then generate properties for the three-dimensional reservoir model with proper simulation for history matching. All wells have a full set of logs (Gamma Ray, Resistivity, Density and Neutron) and few have routine core analysis (Permeability, Porosity and MICP). On a first pass, logs from selected wells are classified into Self Organizing Maps (SOM) without analytical supervision. Then, core data is used to define petrophysical groups (PG), followed by linking the PG's to NMR pore-size distribution analysis results into pre-determined standard pore geometry groups, in this step supervised PGs are generated from the log response constrained by the relationship between pore-throat geometry (MICP) and pose-size distribution (NMR). Permeability-porosity core relationships are reviewed by sorting and eliminating the outliers or inconsistent samples (damaged or chipped, fractures or with local features). After that, the supervised PGs are used to train and calibrate a supervised neural network (NN) and permeability and rock type's relationships can be captured at log scale. Using dimensionality reduction improves the neural network relationships and thus data population into the petrophysical wells.
The result is a more robust model capable to capture over 80% of the core relationships and able to predict permeability and rock types while preserving the geological features of the reservoir. The application of this method makes possible to determine the relevance of core and log data sources to address rock typing and permeability prediction uncertainties. The applied workflows also show how to break the autocorrelation of variables and maximize the usage of logs.
This work demonstrates that the introduced data-driven methods are useful for rock typing determination and address several of the challenges related to core to log properties derivation.
Simonov, Maksim (Gazpromneft Science & Technology Center, Peter the Great St. Petersburg Polytechnic University) | Shubin, Andrei (Saint Petersburg State University) | Penigin, Artem (Gazpromneft Science & Technology Center) | Perets, Dmitrii (Gazpromneft Science & Technology Center) | Belonogov, Evgenii (Gazpromneft Science & Technology Center) | Margarit, Andrei (Gazpromneft Science & Technology Center)
The topic of the paper is an approach to find optimal regimes of miscible gas injection into the reservoir to maximize cumulative oil production using a surrogate model. The sector simulation model of the real reservoir with a gas cap, which is in the first stage of development, was used as a basic model for surrogate model training. As the variable (control) parameters of the surrogate model parameters of gas injection into injection wells and the limitation of the gas factor of production wells were chosen. The target variable is the dynamics of oil production from the reservoir. A set of data has been created to train the surrogate model with various input parameters generated by the Latin hypercube.
Several machine learning models were tested on the data set: ARMA, SARIMAX and Random Forest. The Random Forest model showed the best match with simulation results. Based on this model, the task of gas injection optimization was solved in order to achieve maximum oil production for a given period. The optimization issue was solved by Monte Carlo method. The time to find the optimum based on the Random Forest model was 100 times shorter than it took to solve this problem using a simulator. The optimal solution was tested on a commercial simulator and it was found that the results between the surrogate model and the simulator differed by less than 9%.
BinAbadat, Ebtesam (ADNOC Offshore) | Bu-Hindi, Hani (ADNOC Offshore) | Al-Farisi, Omar (ADNOC Offshore) | Kumar, Atul (ADNOC Offshore) | Zahaf, Kamel (ADNOC Offshore) | Ibrahim, Loay (ADNOC Offshore) | Liu, Yaxin (ADNOC Offshore) | Darous, Christophe (Schlumberger Oil Company) | Barillas, Luisa (Schlumberger Oil Company)
Reservoir Rock Typing and saturation modeling need a two-sided methodology. One side is the geological side of the rock types to populate properties within geological concepts. The other side is addressing reservoir flow and dynamic initialization with capillary pressure. The difficulty is to comply with both aspects especially in carbonates reservoirs with complex diagenesis and migration history. The objective of this paper is to describe the methodology and the results obtained in a complex carbonate reservoir.
The approach is initiated from the sedimentological description from cores and complemented with microfacies from thin sections. The core-based rock types use the dominant rock fabrics, as well as the cementation and dissolution diagenetic processes. The groups are limited to similar pore throat size distribution and porosity-permeability relationships to stay compatible with property modeling at a later stage.
At log-scale, the rock typing has a focus on the estimation of permeability using the most appropriate logs available in all wells. Those logs are porosity, mineral volumes, normalized saturation in invaded zone (Sxo), macro-porosity from borehole image or Nuclear Magnetic Resonance (NMR), NMR T2 log mean relaxation, and rigidity from sonic logs. A specific calculation to identify the presence of tar is also included to assess the permeability better and further interpret the saturation history. The MICP data defined the saturation height functions, according to the modality of the pore throat size. The log derived saturation, and the SHFs are used to identify Free Water Level (FWL) positions and interpret the migration history.
The rock typing classification is well connected with the geological aspects of the reservoirs since it originates from the sedimentological description and the diagenetic processes. We identified a total of 21 rock types across all the formations of interest. We associated rock types with depositional environments ranging from supra-tidal to open marine that controls both the original rock fabrics and the diagenetic processes. The rock typing classification is also appropriate to model permeability and saturation since core petrophysical measurements were in use during the classification. The permeability estimation from logs uses multivariate regressions that have proven to be sensitive to permeability after a Principal Component Analysis per zones and per lithologies. The difference between the core permeability and the permeability derived from logs stays within one-fold of standard deviation as compared to the initial 3-fold range of porosity-permeability. We assigned the rock types with three Saturation Height Function (SHF) classes; (unimodal-dolomite, unimodal- limestone & Multimodal-Limestone). The log derived water saturation (Sw) from logs and SHF shows acceptable agreement.
The reservoir rock typing and saturation modeling methodology described in this paper are considerate of honoring geological features and petrophysical properties to solve for complex diagenesis and post-migration fluid alteration and movement processes.
A statistical screening methodology is presented to address uncertainty related to main geological assumptions in green field modeling. The goals are the identification of the entire range of uncertainty on production, learning which are the most impacting geological uncertain inputs and understanding the relationships between geological scenarios and classes of dynamic behavior.
The paper presents the methodology and an example application to a green field case study. The method is applied on an ensemble of reservoir models created by combining geological parameters across their range of uncertainty. The ensemble of models is then simulated with a selected development strategy and dynamic responses are grouped in classes of outcome through clustering algorithms. Ensemble responses are visualized on a multidimensional stacking plot, as a function of the geological input, and the most influential parameters are identified by axes sorting on the plot. Geological scenarios are then classified on dynamic responses through classification tree algorithms. Finally, a representative set of models is selected from the geological scenarios.
The example study application shows a final oil recovery uncertainty range of 4:1, which is reasonable for a green field in lack of data. Such high range of uncertainty could hardly be found by common risk assessment based on fixed geological assumptions, which often tend to underestimate uncertainty on forecasts. Ensemble outcomes are grouped in four classes by oil recovery, plateau strength, produced water, and breakthrough time. The adoption of such clustering features gives a broad understanding of the reservoir dynamic response. The most influential geological inputs among the examined structural and sedimentological parameters in the example application result to be the fault orientation and channel fraction. This screening result highlights the main drivers of geological uncertainty and is useful for the following scenario classification phase. Classification of the geological scenarios leads to five classes of geological parameter sets, each linked to a main class of dynamic behavior, and finally to five representative models. These five models constitute an effective sampling of the geological uncertainty space which also captures the different types of dynamic response.
This paper will contribute to widen the engineering experience on the use of machine learning for risk analysis by presenting an application on a real field case study to explore the relationship between geological uncertainty and reservoir dynamic behavior.
Weijermans, Peter-Jan (Neptune Energy Netherlands B.V.) | Huibregtse, Paul (Tellures Consult) | Arts, Rob (Neptune Energy Netherlands B.V.) | Benedictus, Tjirk (Neptune Energy Netherlands B.V.) | De Jong, Mat (Neptune Energy Netherlands B.V.) | Hazebelt, Wouter (Neptune Energy Netherlands B.V.) | Vernain-Perriot, Veronique (Neptune Energy Netherlands B.V.) | Van der Most, Michiel (Neptune Energy Netherlands B.V.)
The E17a-A gas field, located offshore The Netherlands in the Southern North Sea, started production in 2009 from Upper Carboniferous sandstones, initially from three wells. Since early production history of the field, the p/z plot extrapolation has consistently shown an apparent Gas Initially In Place (GIIP) which was more than 50% higher than the volumetric GIIP mapped. The origin of the pressure support (e.g. aquifer support, much higher GIIP than mapped) and overall behavior of the field were poorly understood.
An integrated modeling study was carried out to better understand the dynamics of this complex field, evaluate infill potential and optimize recovery. An initial history matching attempt with a simulation model based on a legacy static model highlighted the limitations of existing interpretations in terms of in-place volumes and connectivity. The structural interpretation of the field was revisited and a novel facies modeling methodology was developed. 3D training images, constructed from reservoir analogue and outcrop data integrated with deterministic reservoir body mapping, allowed successful application of Multi Point Statistics techniques to generate plausible reservoir body geometry, dimensions and connectivity.
Following a series of static-dynamic iterations, a satisfying history match was achieved which matches observed reservoir pressure data, flowing wellhead pressure data, water influx trends in the wells and RFT pressure profiles of two more recent production wells. The new facies modeling methodology, using outcrop analogue data as deterministic input, and a revised seismic interpretation were key improvements to the static model. Apart from resolving the magnitude of GIIP and aquifer pressure support, the reservoir characterization and simulation study provided valuable insights into the overall dynamics of the field – e.g. crossflows between compartments, water encroachment patterns and vertical communication. Based on the model a promising infill target was identified at an up-dip location in the west of the field which looked favorable in terms of increasing production and optimizing recovery. At the time of writing, the new well has just been drilled. Preliminary logging results of the well will be briefly discussed and compared to pre-drill predictions based on the results of the integrated reservoir characterization and simulation study.
The new facies modeling methodology presented is in principle applicable to a number of Carboniferous gas fields in the Southern North Sea. Application of this method can lead to improved understanding and optimized recovery. In addition, this case study demonstrates how truly integrated reservoir characterization and simulation can lead to a revision of an existing view of a field, improve understanding and unlock hidden potential.
Birnie, Claire Emma (Equinor ASA) | Sampson, Jennifer (Equinor ASA) | Sjaastad, Eivind (Equinor ASA) | Johansen, Bjarte (Equinor ASA) | Obrestad, Lars Egil (Equinor ASA) | Larsen, Ronny (Equinor ASA) | Khamassi, Ahmed (Equinor UK LTD)
To ensure safe and efficient operations, all offshore operations follow a plan devised to take into account current operation conditions and identify the optimum workflow with the minimum risk potential. Previously, planners had to manually consult eight data sources, each with a separate UI, and summarise the plan in a.pdf document. Equinor's Operation Planning Tool (OPT) has been developed to easily present the planners with the technical conditions of a platform, identify potentially dangerous combinations of concurrent activities, and propose learnings from eight years’ worth of incident recordings – all relevant to the current list of planned activities. The tool aims to answer questions such as ‘are other activities planned for the same time which would make this activity unsafe?’ or ‘have incidents previously occurred whilst performing similar tasks on this equipment type?’.
This paper details the development of the OPT with a particular focus on the application of Natural Language Understanding for extracting equipment types and tasks involved in previous incidents and relating these to planned activities. Utilising natural language processing techniques, a system has been developed that mines the content of Equinor's incident database, and assigns context to incidents, by identifying the systems, activities and equipment involved and the conditions on the asset at the time of the incident. The same context is also discovered from the content of planned activities. These key concepts are organised into a knowledge graph synthesising Equinor's institutional safety and operational experience.
The OPT has reduced time spent planning by providing a single interface detailing a plant's technical conditions, all planned work orders and relevant lessons learned from previous incidents. By reducing the reliance on personal experience, the tool has provided subjectively improved risk identification and handling, plus faster knowledge transfer to new employees as well as focussed cross-platform knowledge sharing. The success of the tool highlights the strength of combining data and leveraging the vast quantities of historic data available both in unstructured and structured forms to create a safe, offshore work environment.