Bigoni, Francesco (Eni S.p.A) | Pirrone, Marco (Eni S.p.A) | Trombin, Gianluca (Eni S.p.A) | Vinci, Fabio Francesco (Eni S.p.A) | Raimondi Cominesi, Nicola (ZFOD) | Guglielmelli, Andrea (ZFOD) | Ali Hassan, Al Attwi Maher (ZFOD) | Ibrahim Uatouf, Kubbah Salma (ZFOD) | Bazzana, Michele (Eni Iraq BV) | Viviani, Enea (Eni Iraq BV)
The Mishrif Formation is one of the important carbonate reservoirs in middle, southern Iraq and throughout the Middle East. In southern Iraq, the formation provides the reservoir in oilfields such as Rumaila/West Qurna, Tuba and Zubair. The top of the Mishrif Formation is marked by a regional unconformity: a long period of emersion in Turonian (ab. 4.4 My) regionally occurred boosted by a warm humid climate, associated to heavy rainfall. In Zubair Field, within the Upper interval of Mishrif Formation, there are numerous evidences of karst features responsible of important permeability enhancements in low porosity intervals that are critical for production optimization and reservoir management purposes.
In the first phase, the integration of Multi-rate Production logging and Well Test analysis was very useful to evaluate the permeability values and to highlight the enhanced permeability (largely higher than expected Matrix permeability) intervals related to karst features; Image log analysis, on the same wells, allowed to find out a relationship between karst features and vug densities, making possible to extend the karst features identification also in wells lacking of well test and Production logging information. This approach has allowed to obtain a Karst/No Karst Supervised dataset for about 60 wells.
In the second phase different seismic and geological attributes have been considered in order to investigate possible correlations with karst features. In fact there are some parameters that show somehow a correlation with Karst and/or NoKarst wells: the Spectral Decomposition (specially 10 and 40 Hz volumes), the detection of sink-holes at top Mishrif on the Continuity Cube and its related distance, the sub-seismic Lineaments (obtained from Curvature analysis and subordinately from Continuity), distance from Top Mishrif. In the light of these results, the most meaningful parameters have been used as input data for a Neural Net Process ("Supervised Neural Network") utilizing the Supervised dataset both as a Trained dataset (70%) and as a Verification dataset (30%). A probability 3D Volume of Karst features was finally obtained; the comparison with verification dataset points out an error range around 0.2 that is to say that the rate of success of the probability Volume is about 80%.
The final outcomes of the workflow are karst probability maps that are extremely useful to guide new wells location and trajectory. Actually, two proof of concept case histories have demonstrated the reliability of this approach. The newly drilled wells, with optimized paths according to these prediction-maps, have intercepted the desired karst intervals as per the subsequent image log interpretation, which results have been very valuable in the proper perforation strategy including low porous intervals but characterized by high vuggy density (Karst features). Based on these promising results the ongoing drilling campaign has been optimized accordingly.
There are a vast number of reservoirs with drill cuttings and core images that have classification problems associated with them. This could be due to the images not being classified in the first place, or the images may be available but the interpretation reports could be missing. Another problem is that images from different wells could be interpreted by different wellsite geologists/sedimentologists and hence result in an inconsistent classification scheme. Finally, there could also be the problem of some images being incorrectly classified. Ergo it would be desirable to have an unbiased objective system that could overcome all of these issues. Step in convolutional neural networks. Advances during this decade in using convolutional neural networks for visual recognition of discriminately different objects means that now object recognition can be achieved to a significant extent. Once the network is trained on a representative set of lithological classes, then such a system just needs to be fed the raw drill cuttings or core images that it has not seen before and it will automatically assign a lithological class to each image and an associated probability of the image belonging to that class. In so doing, images below a certain probability threshold can be automatically flagged for further human investigation. The benefit of such a system would be to improve reservoir understanding by having all available images classified in a consistent manner hence keeping the characterization consistent as well. It would further help to reduce the time taken to get human expertise to complete the task, as well as the associated cost.
Waterflooding is the main technic to recover hydrocarbons in reservoirs. For a given set of wells (injectors and producers), the choice of injection/production parameters such as pressures, flow rates, and locations of these boundary conditions have a significant impact on the operating life of the wells. As a large number of combinations of these parameters are possible, one of the critical decision to make is to identify an optimal set of these parameters. Using the reservoir simulator directly to evaluate the impact of these sets being unrealistic considering the required number of simulations, a common approach consists of using response surfaces to approximate the reservoir simulator outputs. Several techniques involving proxies model (e.g., kriging, polynomial, and artificial neural network) have been suggested to replace the reservoir simulations. This paper focalizes on the application of artificial neural networks (ANN) as it is commonly admitted that the ANNs are the most efficient one due to their universal approximation capacity, i.e., capacity to reproduce any continuous function. This paper presents a complete workflow to optimize well parameters under waterflooding using an artificial neural network as a proxy model. The proposed methodology allows evaluating different production configurations that maximize the NPV according to a given risk. The optimized solutions can be analyzed with the efficient frontier plot and the Sharpe ratios. An application of the workflow to the Brugge field is presented in order to optimize the waterflooding strategy.
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.
Masini, Cristian (Petroleum Development Oman) | Al Shuaili, Khalid Said (Petroleum Development Oman) | Kuzmichev, Dmitry (Leap Energy) | Mironenko, Yulia (Leap Energy) | Majidaie, Saeed (Formerly with Leap Energy) | Buoy, Rina (Formerly with Leap Energy) | Alessio, Laurent Didier (Leap Energy) | Malakhov, Denis (Target Oilfield Services) | Ryzhov, Sergey (Formerly with Target Oilfield Services) | Postuma, Willem (Target Oilfield Services)
Unlocking the potential of existing assets and efficient production optimisation can be a challenging task from resource and technical execution point of view when using traditional static and dynamic modelling workflows making decision-making process inefficient and less robust.
A set of modern techniques in data processing and artificial intelligence could change the pattern of decision-making process for oil and gas fields within next few years. This paper presents an innovative workflow based on predictive analytics methods and machine learning to establish a new approach for assets management and fields’ optimisation. Based on the merge between classical reservoir engineering and Locate-the-Remaining-Oil (LTRO) techniques combined with smart data science and innovative deep learning algorithms this workflow proves that turnaround time for subsurface assets evaluation and optimisation could shrink from many months into a few weeks.
In this paper we present the results of the study, conducted on the Z field located in the South of Oman, using an efficient ROCM (Remaining Oil Compliant Mapping) workflow within an advanced LTRO software package. The goal of the study was to perform an evaluation of quantified and risked remaining oil for infill drilling and establish a field redevelopment strategy.
The resource in place assessment is complemented with production forecast. A neural network engine coupled with ROCM allowed to test various infill scenarios using predictive analytics. Results of the study have been validated against 3D reservoir simulation, whereby a dynamic sector model was created and history matched.
Z asset has a number of challenges starting from the fact that for the last 25 years the field has been developed by horizontal producers. The geological challenges are related to the high degree of reservoir heterogeneity which, combined with high oil viscosity, leads to water fingering effects. These aspects are making dynamic modelling challenging and time consuming.
In this paper, we describe in details the workflow elements to determine risked remaining oil saturation distribution, along with the results of ROCM and a full-field forecast for infill development scenarios by using neural network predictive analytics validated against drilled infills performance.
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.
This study examines which is the margin of usability for Artificial Intelligence (AI) algorithms related to the rock properties distribution in static modeling. This novel method shows a forward modeling approach using neural networks and genetic algorithms to optimize correlation patterns among seismic traces of stack volumes and well rock properties. Once a set of nonlinear functions is optimized in the well locations, to correlate seismic traces and rock properties, spatial response is estimated using the seismic volume. This seismic characterization process is directly dependent on the error minimization during the structural seismic interpretation process, as well as, honoring the structural complexity while modeling. Previous points are key elements to obtain an adequate correlation between well data and seismic traces. The joint mechanism of neural networks and genetic algorithms globally optimize the nonlinear functions and its parameters to minimize the cost function. Estimated objective function correlates well rock properties with seismic stack data. This mechanism is applied to real data, within a high structural complexity and several wells. As an output, calibrated petrophysical time volumes in the interval of interest are obtained. Properties are used initially to generate a geological facies model. Subsequently, facies and seismic properties are used for the three-dimensional distribution of petrophysical properties such as: rock type, porosity, clay volume and permeability. Therefore, artificial intelligence algorithms can be widely exploited for uncertainty reduction within the rock property spatial estimation.
A new real-time machine learning model has been developed based on the deep recurrent neural network (DRNN) model for performing step-down analysis during the hydraulic fracturing process. During a stage of the stimulation process, fluids are inserted at the top of the wellhead, while the flow is primarily driven by the difference between the bottomhole pressure (BHP) and reservoir pressure. The major physics and engineering aspects involved are complex and, quite often, there is a high level of uncertainty related to the accuracy of the measured data, as well as intrinsic noise. Consequently, using a machine learning-based method that can resolve both the temporal and spatial non-linear variations has advantages over a pure engineering model.
The approach followed provides a long short-term memory (LSTM) network-based methodology to predict BHP and temperature in a fracturing job, considering all commonly known surface variables. The surface pumping data consists of real-time data captured within each stage, such as surface treating pressure, fluid pumping rate, and proppant rate. The accurate prediction of a response variable, such as BHP, is important because it provides the basis for decisions made in several well treatment applications, such as hydraulic fracturing and matrix acidizing, to ensure success.
Limitations of the currently available modeling methods include low resolution BHP predictions and an inability to properly capture non-linear effects in the BHP/temperature time series relationship with other variables, including surface pressure, flow rate, and proppant rate. In addition, current methods are further limited by lack of accuracy in the models for fluid properties; the response of the important sub-surface variables strongly depends on the modeled fluid properties.
The novel model presented in this paper uses a deep learning neural network model to predict the BHP and temperature, based on surface pressure, flow rate, and proppant rate. This is the first attempt to predict response variables, such as BHP and temperature, in real time during a pumping stage, using a memory-preserving recurrent neural network (RNN) variant, such as LSTM. The results show that the LSTM can successfully model the BHP and temperature in a hydraulic fracturing process. The BHP and temperature predictions obtained were within 5% relative error. The current effort to model BHP can be used for step-down analysis in real time, thereby providing an accurate representation of the subsurface conditions in the wellbore and in the reservoir. The new method described in this paper avoids the need to manage the complex physics of the present methods; it provides a robust, stable, and accurate numerical solution throughout the pumping stages. The method described in this paper is extended to manage step-down analysis using surface-measured variables to predict perforation and tortuosity friction.
In this paper, we present a water-cut estimator utilizing the function approximation capability of an artificial neural network (ANN). The inputs to the ANN are optical sensor readings in a Red-Eye water-cut meter, which features the near-infrared (NIR) absorption spectroscopy technology. The initial training of the ANNwas done with a data set acquired from our multiphase flow-loop test facility, which was filled with live oil, water and gas. The test fluid stream was adjusted with good ranges of water-cut and gas-volume fractions which were supposed to cover the situations that can be foreseen in real production. However, clear discrepancies between the outputs of the ANN and the water-cut values from BS&W measurmentswere observedwhen the ANN was applied to actual production data measured by Red-Eye meters installed at two offshore wells. To address this issue and equip the ANN with self-adapting capability in real application, we propose a Bayesian approach to update the parameters of the ANN based on both initial flow-loop data and collected field data. The performance of the adapted ANN on both the data sets shows the effectiveness of the method.