Wu, JinYong (Schlumberger) | Banerjee, Raj (Schlumberger) | Bolanos, Nelson (Schlumberger) | Alvi, Amanullah (Schlumberger) | Tilke, Peter Gerhard (Schlumberger - Doll Research) | Jilani, Syed Zeeshan (Schlumberger Oilfield UK Plc) | Bogush, Alexander (Schlumberger)
Assessing the waterflood, monitoring the fluids front, and enhancing sweep with the uncertainty of multiple geological realisations, data quality, and measurement presents an ongoing challenge. Defining sweet spots and optimal candidate well locations in a well-developed large field presents an additional challenge for reservoir management. A case study is presented that highlights the approach to this cycle of time-lapse monitoring, acquisition, analysis and planning in delivery of an optimal field development strategy using multi-constrained optimisation combined with fast semi-analytical and numerical simulators.
The multi-constrained optimiser is used in conjunction with different semi-analytical and simulation tools (streamlines, traditional simulators, and new high-powered simulation tools able to manage huge, multi-million-cell-field models) and rapidly predicts optimal well placement locations with inclusion of anti-collision in the presence of the reservoir uncertainties. The case study evaluates proposed field development strategies using the automated multivariable optimisation of well locations, trajectories, completion locations, and flow rates in the presence of existing wells and production history, geological parameters and reservoir engineering constraints, subsurface uncertainty, capex and opex costs, risk tolerance, and drilling sequence.
This optimisation is fast and allows for quick evaluation of multiple strategies to decipher an optimal development plan. Optimisers are a key technology facilitating simulation workflows, since there is no ‘one-approach-fits-all' when optimising oilfield development. Driven by different objective functions (net present value (NPV), return on investment (ROI), or production totals) the case study highlights the challenges, the best practices, and the advantages of an integrated approach in developing an optimal development plan for a brownfield.
This paper presents a novel implementation for evolutionary algorithms in oil and gas reservoirs history matching problems. The reservoir history is divided into time segments. In each time segment, a penalty function is constructed that quantifies the mismatch between the measurements and the simulated measurements, using only the measurements available up to the current time segment. An evolutionary optimization algorithm is used, in each time segment, to search for the optimal reservoir permeability and porosity parameters. The penalty function varies between segments; yet the optimal reservoir characterization is common among all the constructed penalty functions. A population of the reservoir characterizations evolves among subsequent time segments through minimizing different penalty functions. The advantage of this implementation is two fold. First, the computational cost of the history matching process is significantly reduced. Second, problem constraints can be included in the penalty function to produce more realistic solutions. The proposed concept of dynamic penalty function is applicable to any evolutionary algorithm. In this paper, the implementation is carried out using genetic algorithms. Two case studies are presented in this paper: a synthetic case study and the PUNQ-S3 field case study. A computational cost analysis that demonstrates the computational advantage of the proposed method is presented.
Determining the optimum location of wells during waterflooding contributes significantly to efficient reservoir management. Often, Voidage Replacement Ratio (VRR) and Net Present Value (NPV) are used as indicators of performance of waterflood projects. In addition, VRR is used by regulatory and environmental agencies as a means of monitoring the impact of field development activities on the environment while NPV is used by investors as a measure of profitability of oil and gas projects. Over the years, well placement optimization has been done mainly to increase the NPV. However, regulatory measures call for operators to maintain a VRR of one (or close to one) during waterflooding.
A multiobjective approach incorporating NPV and VRR is proposed for solving the well placement optimization problem. We present the use of both NPV and VRR as objective functions in the determination of optimal location of wells. The combination of these two in a multiobjective optimization framework proves to be useful in identifying the trade-offs between the quest for high profitability of investment in oil and gas projects and the desire to satisfy regulatory and environmental requirements. We conducted the search for optimum well locations in three phases. In the first phase, only the NPV was used as the objective function. The second phase has the VRR as the sole objective function. In the third phase, the objective function was a weighted sum of the NPV and the VRR. A set of four weights were used in the third phase to describe the relative importance of the NPV and the VRR and a comparison of how these weights affect the optimized NPV and VRR values is provided.
We applied the method to determine the optimum placement of wells using two sample reservoirs: one with a distributed permeability field and the other, a channel reservoir with four facies. Two evolutionary-type algorithms: the covariance matrix adaptation evolutionary strategy (CMA-ES) and differential evolution (DE), were used to solve the optimization problem. Significantly, the method illustrates the trade-off between maximizing the NPV and optimizing the VRR. It calls the attention of both investors and regulatory agencies to the need to consider the financial aspect (NPV) and the environmental aspect (VRR) of waterflooding during secondary oil recovery projects. The multiobjective optimization approach meets the economic needs of investors and the regulatory requirements of government and environmental agencies. This approach gives a realistic NPV estimation for companies operating in jurisdiction with requirement for meeting a VRR of one.
The Middle Minagish Oolite Formation is 450 to 550 feet thick interval of porous limestone reservoir, composed of peloidal/skeletal grainstones with lesser amount of packstone, oolitic grainstone, wackstone and mudstone in Umm Gudair field, West Kuwait. It is characterized by small scale reservoir heterogeneity, primarily related to the depositional as well as diagenetic features. Capturing reservoir properties in micro scale and its spatial variation needs special attention in this reservoir due to its inherent anisotropy. Reservoir properties will depend on the level that we are analyzing on reservoir (millimeter to meter scale). Here we used Electrical Borehole Image (EBI) and Nuclear Magnetic Resonance (NMR) to capture small scale feature of Umm Gudair carbonate reservoir and compared them with core data
In present work, reservoir properties (including texture, facies, porosity and permeability) interpreted by the EBI shows good match with NMR driven properties and core data. Textural changes in image logs also match well with pore size distribution from NMR logs. Further highly porous zones which are considered either due to primary porosity or vugs match with larger pores of NMR logs and these corroborates with also core derived porosity. A good match has been observed between EBI, NMR and cored derived porosity. Permeability calculations have also been made and compared with core data. A detail workflow has been developed here to interpret reservoir properties on un-cored wells, where only low vertical resolution data is available. This technique is quite useful to identify the characters and mode of origin highly porous zones in reservoir section which are generally not identifiable by low resolution standard logs. This workflow will allow us to interpret the heterogeneity at high resolution level in un-cored wells, as results are validated with integration of EBI, NMR and core data.
Vibrations are caused by bit and drill string interaction with formations under certain drilling conditions. They are affected by different parameters such as weight on bit, rotary speed, mud properties, BHA and bit design as well as by the mechanical properties of the formations. During the actual drilling process the bit interacts with different formation layers whereby each of those layers usually have different mechanical properties. Vibrations are also indirectly affected by the formations since weight on bit and rotary speed are usually optimized against changing formations (drilling optimization process). Therefore it can be concluded that for optimized drilling reduction of vibrations is one of the challenges.
A fully automated laboratory scale drilling rig, the CDC miniRig, has been used to conduct experimental tests. A three component vibration sensor sub attached to drill string records drill string vibrations and an additional sensor system records the drilling parameters. Uniform concrete cubes with different mechanical properties were built. Those cubes as well as a homogeneous sandstone cube were drilled with different ranges of weight on bit and bit rotary speed. The mechanical properties of all cubes were measured prior to the experiments. During all experiments, drilling parameters and the vibration data were recorded. Based on analyses of the data in the time and the frequency domain, linear and non-linear models were built. For this purpose the interrelations of sandstone and concrete mechanical properties, drilling parameters and vibration data were modeled by neural networks. Application of sophisticated attribute selection methods showed that vibration data in both, time- and frequency domain, have a major impact in modeling the rate of penetration.
Arnaout, Arghad (TDE Thonhauser Data Engineering GmbH) | Thonhauser, Gerhard (Montanuniversitat Leoben) | Esmael, Bilal (Montanuniversitat Leoben) | Fruhwirth, Rudolf Konrad (TDE Thonhauser Data Engineering GmbH)
Detection of oilwell drilling operations is an important step for drilling process optimization. If drilling operations are classified accurately, detailed performance reports not only on drilling crews but also on drilling rigs can be produced. Using such reports, the management can evaluate the drilling work more precisely from performance point of view.
Mud-logging systems of modern drilling rigs provide numerous sensors data. Those sensors measurements are considered as indicators to monitor different states of drilling process. Usually real-time measurements of the following sensors data are available as surface measurements: hookload, block position, flow rates, pump pressure, borehole and bit depth, RPM, torque, rate of penetration and weight on bit.
In this work, collected sensors measurements from mud-logging systems are used to detect different drilling operations. Detailed data analysis shows that the surface sensors measurements can be considered as a main source of information about drilling operations. For this purpose, a mathematical model based on polynomials approximation is constructed to interpolate sensors data measurements.
Discrete polynomial moments are used as a tool to extract specific features (moments) from drilling sensors data. Then we use these moments for each drilling operation as pattern descriptor to classify similar operations in drilling time series. The extracted polynomial moments describe trends of sensors data and behavior of rig's sub-systems (Rotation System, Circulation System, and Hoisting System). Furthermore, this paper suggests a method on how to build patterns base and how to recognize and classify drilling operations once sensors data received from mud-logging system. Drilling experts compare the results to manually classified operations and the results show high accuracy.
Stuck pipe has been recognized as one of the most challenging and costly problems in the oil and gas industry. However, this problem can be treated proactively by predicting it before it occurs.
The purpose of this study is to implement the two most powerful machine learning methods, Artificial Neural Networks (ANNs) and Support Vector Machines (SVMs), to predict stuck pipe occurrences. Two developed models for ANNs and SVMs with different scenarios were implemented for prediction purposes. The models were designed and constructed by the MATLAB language. The MATLAB built-in functions of ANNs and SVMs, and the MATLAB interface from the library of support vector machines were applied to compare the results. Furthermore, one database that included mud properties, directional characteristics, and drilling parameters has been assembled for training and testing processes. The study involved classifying stuck pipe incidents into two groups - stuck and non-stuck - and also into three subgroups: differentially stuck, mechanically stuck, and non-stuck. This research has also gone through an optimization process which is vital in machine learning techniques to construct the most practical models. This study demonstrated that both ANNs and SVMs are able to predict stuck pipe occurrences with reasonable accuracy, over 85%.
The competitive SVM technique is able to generate generally reliable stuck pipe prediction. Besides, it can be found that SVMs are more convenient than ANNs since they need fewer parameters to be optimized. The constructed models generally apply very well in the areas for which they are built, but may not work for other areas. However, they are important especially when it comes to probability measures. Thus, they can be utilized with real-time data and would represent the results on a log viewer.
Stationkeeping in ice-covered waters has become a large area of interest forresearch and development in light of heightened interest in Arctic oil and gasexploration. The performance of Dynamic Positioning (DP) control systems forstationkeeping purposes in ice conditions is a difficult challenge fornumerical modeling assessment. Given that full-scale validation data for DP inice operations is often scarce, physical modeling of stationkeeping in iceoffers the best method for assessing the performance of dynamically positionedvessels in these conditions. A series of model tests carried out at theNational Research Council of Canada's Ice Tank facility in August and Septemberof 2011 attempted to observe the effects of various managed ice conditions(i.e. ice floes which have been broken into manageable pieces by an icebreaker) on DP performance. Results from these tests are discussed. Ofparticular interest in this study is the observation of non-linear effects ofvarying ice conditions on DP performance. The use of machine vision-based dataproducts as potential estimators of ice loading is discussed. It is concludedthat simple statistical observations of these conditions will be unable tofully characterize the effects of various ice parameters on performance, andthat investigation into more advanced data products available from machinevision systems may be able to aide in characterizing these effects as well asin the development of models capable of predicting ice loads.
Synthetic aperture radar (SAR) has been extensively used for the derivationof valuable information regarding sea ice properties and conditions. This workfocuses on the use of RADARSAT-2 ScanSAR Wide images (500x500 km swaths with50x50 meter pixels) to provide sea ice information for operations support inthe Arctic. Our developed processes generate several products that supportnavigation and operations in ice infested waters: i) Sea ice images, i.e.delineating and mapping sea ice relative to the open water, ii) Seasonal trendcharts of sea ice over an area of interest and, iii) Automated ice featuretracking and pressure zone mapping.
Using the RADARSAT-2 dual-polarization images and automated techniques, seaice maps are generated to identify regions of open water and of sea ice. Fromthe sea ice maps, total ice concentration is derived and combined withhistorical concentration maps. The output seasonal trend charts can be used toassist in monitoring Arctic sea ice extent and sea ice identification to aidwith navigational safety operations. Finally, we develop an automated icefeature tracking that can track moving ice and from which pressure and driftzones are identified. Future work will involve the development of theprediction of movement of ice floes and packs, using the ice feature trackingtechnique as the foundation.
Blunt, J.D. (ExxonMobil Upstream Research) | Garas, V.Y. (ExxonMobil Upstream Research) | Matskevitch , D.G. (ExxonMobil Upstream Research) | Hamilton, J.M. (ExxonMobil Upstream Research) | Kumaran, K. (ExxonMobil Corporate Strategic Research)
Safe and economic hydrocarbon exploration, development and productionoperations in the high arctic deepwater require a nuanced understanding of thesea ice environment. Robust image analysis techniques provide methods bywhich this nuance can be more objectively characterized and used for decisionmaking while in operations. Morphological segmentation and windowedstatistical analysis are proposed as two approaches that provide usefulinformation on the tactical scale by rapidly characterizing floe fieldmorphology and relative surface roughness. Their use is demonstratedwithin the context of actual high arctic field program data. Results fromthe method application are shown and the benefits and limitations of their useare discussed.