- Home
- Conferences
- Conference Proceedings
- Conferences
Second EAGE Digitalization Conference and Exhibition
- Conference date: March 23-25, 2022
- Location: Vienna, Austria
- Published: 23 March 2022
61 - 72 of 72 results
-
-
Optimization Algorithm for Surfactant Flooding: Comparative Study of Brent’s Hybrid and Adapted Newton-Rapson Method
Authors O. Akinyele and K. StephenSummaryAn optimization algorithm was deployed to solve the single-objective function for uncertainty modeling of surfactant flooding. The cost function had an appearance of discontinuity, which may result in convergence problems. Brent’s hybrid root finding method can be applied to the nonlinear equations of the fractional-flow curves to find the oil bank water saturation. Due to a complicated set of rules that prevent modification of Brent’s hybrid method to easily select which root-finding method is used, the Newton-Rapson was adapted and considered as an alternative. The adapted Newton-Rapson method was compared with Brent’s hybrid method for the overall performance of the optimization algorithm.
Various optimization search intervals were used, and each was constrained to 3600 test evaluations. The Newton-Rapson method converges faster with a lower value of the cost function and an average execution time of 8.79 seconds for the case studies. On the other hand, Brent’s method converges at an average execution time of 41.04 seconds with more intervals on the cost function with no convergence. We concluded that Newton’s method was a better solution capable of minimizing a noisy cost function with fewer iteration and faster execution times. Therefore, it should be considered the most effective optimization algorithm.
-
-
-
Quantitatively evaluate a machine learning algorithm for petrophysical modeling
Authors X. Wang and L. TrueloveSummaryThe results of a machine learning algorithm for petrophysical modeling are often evaluated by a qualitative measure, such as the visual check of a blind well test. However, different researchers may have different interpretations and conclusions for the same blind well test just based on the visual inspection. Meanwhile, the visual inspection of a blind well test does not well capture the lateral heterogeneity of the subsurface which is very important for the reservoir characterization. These subjective assessments can hinder us from better evaluating and improving the algorithm. More quantitative measures are required to effectively evaluate the results of a machine learning algorithm for petrophysical modeling. In this paper, we use three synthetic models whose petrophysical distributions are well known to quantitatively evaluate a machine learning algorithm for petrophysical modeling.
-
-
-
Mitigation of Geothermal Induced Seismicity Through Data Integration and 3D Geomodeling in a Cloud-Hosted Environment
Authors A. Plougoulen and A. HaouesseSummaryInduced seismicity events, with magnitudes large enough to be felt by local communities and impact surface infrastructures, are an undesirable potential result of geothermal operations, and significantly impact social acceptance. Mitigating the possible adverse effect of induced seismicity on the environment, human health, social acceptance and project economics, is of upmost importance if one wishes to increase the number of geothermal projects in a safe and sustainable manner. Increasing understanding of the subsurface through the creation of an accurate 3D earth model is key to risk mitigation, since induced seismicity is closely related to local geology and fault structure. As the project progresses and more data is collected, 3D visualization and 3D seismicity analysis are needed to refine understanding of fracture and fault systems and to identify spatial and spatiotemporal relationships between operational parameters and induced seismicity. A cloud-hosted digital twin of the subsurface unlocks the possibility for geothermal teams and partners to access the model from anywhere at any time. It supports communication around a shared understanding of the subsurface and the ability to quickly react to the latest insights as operations progress.
-
-
-
Data driven 3D Basin Modeling and Lithology Prediction using Seismic Attributes in Central Sumatra Basin
SummaryIn this study, we use various machine learning classifier algorithm to predict the lithology distribution in the Central Sumatra Basin. Some of the classification methods used are:
- MLPClassifier
- AdaBoostClassifier
- GradientBoostingClassifier
- GaussianNB
- KNeighborsClassifier
- DecisionTreeClassifier
- RandomForestClassifier
Each of these machine learning methods have advantages and disadvantages depending on the classification problem. Method selection is needed to determine the most appropriate classifier based on the AUC value (the higher the AUC value indicates the most suitable method for the classification problem).
The predictors used are 26 Post Stack 3D Seismic Attributes, while the target is the VSH value contained in each well. VSH values have been previously calculated using petrophysical calculations and have been grouped based on various cutoff.
After the lithology predictions were made, we performed 3D petroleum system modeling on the basin scale. This process can validate whether the lithology prediction results are relevant to geological conditions by looking at several sections.
-
-
-
Autoencoder-based generation of subsurface models
Authors A. Sabzi Shahrebabaki, S.H. Fouladi, E. Holtar and L. VynnytskaSummarySubsurface understanding is based on data analysis and interpretation. Seismic data is one of the main sources for model building, however, there are a lot of uncertainties and, as a result, possible interpretations. Good subsurface understanding and representation is vital for successful drilling operations in oil and gas exploration and production. One way to better represent the real outcome space of the subsurface models is to construct a set of scenarios that all agree with the data instead of relying on a single model. We propose a method to create all realizations between the substantially different scenarios by applying a data-driven approach enabling linear interpolation in latent space constructed using deep learning.
-
-
-
Implementation of Digital Transformation to Support Exploration Investment in Indonesia
Authors F.T. Amir, A.D. Wibisono, O. Oktariano and A.N. FadhillahSummaryUpdates on Digital transformation in SKK Migas as the government institution that supervised upstream oil & gas industry, especially for exploration investment in order to achieve 1 million BOPD in 2030. This paper covers 6 digitalization projects and their impacts in supporting exploration activities monitoring.
-
-
-
Enhanced collapse feature extraction from high-resolution seismic data using convolutional neural network
Authors M. Saadat Dastenayi, E. Salehi, M. Etminan, A. Yousefzadeh and H. NezamoleslamiSummaryFor the purpose of developing a challenging marine field in the Persian Gulf, a high-resolution seismic survey was acquired with the aim of evaluating different geohazards elements prior to the drilling. Carbonate rocks were dominant lithology in the area of study and as a result, a large number of collapse features have been developed which were evaluated as the major geohazards elements. We applied neural network classification for extraction of collapse geobodies for subsequent geohazards analysis. The acquired seismic survey consists of several densely spaced 2D lines in different azimuths which were finally merged into an integrated 3D seismic cube for further analysis. Strong time-distorted acquisition footprints were the main challenging issue regarding the classification procedure using high-resolution seismic data. They remained in the final processed data despite several remedial actions in the processing steps. Footprint artifacts share similar seismic character with the collapse features and were classified as geohazards using neuron-based neural network classification. We successfully examined and applied convolutional neural network to discriminate them from true collapse features.
-
-
-
Hybrid Physics-Based Data-Driven Methods… the Future for Petroleum Engineering?
Authors B. Moradi, L. Alessio and O.J. SandalSummaryData-driven methods are widely and successfully used in social sciences, an area where no discernible physics exists, but many observations exist. As a result, there has been a drive to implement such approaches into our industry.
However, in petroleum engineering, we are faced with the opposite scenario: limited observations with ambiguous interpretation and relatively well-known physics and equations. Over the last 50 years, we have seen the establishment of physics-based approaches, with numerical simulation at its core.
We are seeing the emergence of hybrid approaches: data-driven but physics compliant. In this work, we contend that these methods may provide the best of both worlds, and we provide two specific examples of such hybrid methods. First, we introduce a novel method for zonal production allocation, which allows for the inclusion of physics and observation data. We demonstrate through a case study how this approach significantly improves production allocation and therefore reservoir management decisions. Secondly, we discuss the integration of the published remaining oil compliant mapping algorithm with machine learning methods for the purpose of ‘locate-the-remaining-oil’ activities and determining behind casing and infill drilling opportunities. Drilling results from recent projects are examined, and the method’s accuracy is evaluated.
-
-
-
Pioneering Subsurface Data Management Studio and Asset Retirement Obligation Retrenchment in Upstream South Sumatra Region, Indonesia
Authors A. Permana, A. Sulaksono, N. Hikmah, I. Dwiyono, R. Pratama, Y. Gustian, P. Suseno and S. ParsaulianSummaryAccording to the geoscientist work profile in Oil & Gas Journal, 1998, inadequate data management practice has caused 60% proportion of the time to be spent for data searching and compilation. The oil and gas industry in the South Sumatra region is categorized as a matured working area and always becoming a challenge to operate efficiently, not only because of well-developed fields and depleted reservoirs, but also commonly deals with a huge amount of old raw data set, either well coordinates, seismic, trajectory, or log data, even data duplication itself. However, since its first discovery in 1896, the South Sumatra region has already transformed into one of the most productive basins in Indonesia until the present through immense exploration and development activities. Additionally, more than 4000 wells has been drilled in 86 oil fields of the South Sumatra block within an 18,181 km2 working area. Massive operational jobs such as drilling programs, workover, and well intervention involved a lot of subsurface data. Invalid data has proven a serious problem that caused several operational failures and enormous financial losses. Proper implementation of subsurface data management for the last several years has generated significant progress for future company milestones.
-
-
-
Forecasting the Cost of Surface Facilities Based on Crude Oil Prices Using Linear Regression Method
More LessSummaryThis paper will evaluate and analyze the relationship between the project cost of surface facilities (US$/bbl) and the oil price (US$/bbl). The steps taken in this paper are as follows: collecting the project cost of surface facilities (2003–2020), oil price history (2003–2020), and calculating the deviation between the cost of surface facilities and oil prices. Then developing a formula in the form of an estimate of the project cost of surface facilities at a certain oil price based on 224 oil projects in Indonesia.
Based on the evaluation and statistical analysis of 224 onshore - oil projects in Indonesia, the formula (equation) showing the correlation between the project cost of surface facilities (US$/bbl) and oil price (US$/bbl) is z = 0.007x + 2.1148, where z = cost of surface facilities and x = oil price. By using the oil price year 2020 (39 US$/bbl), the estimated cost of surface facilities in onshore area in Indonesia is 2.38 US$/bbl.
-
-
-
Automating legacy borehole geology image processing and interpretation workflows to one click on cloud and on-premise
Authors A. He, B. Qamhiyeh, M. Carles, L. Comparon, N. Bize-Forest and I. Le NirSummaryBorehole imaging tool processing and interpretation workflows are among the first run after acquisition, and getting their results as early as possible is especially valuable when these results can drive near-real time decisions. However, these workflows are generally a sequence of highly complex and time-consuming operations which require geoscientists with prior training and expertise, as well as their full attention for the workflow duration. Hence, there is an opportunity to automate these workflows down to a single click.
As the oil and gas community is shifting towards cloud technologies, lifting these legacy workflows to tap into newer cloud ecosystems is becoming a necessity. But we do not forget that high-speed and secure internet connectivity is sometimes a luxury, and that not everything should necessarily run on the cloud.
This paper describes how we built fully automated workflows to process multiphysics borehole image logs and to extract the bed surfaces from them, down to a single click and with no supervision. Its software architecture is both deployable at the wellsite and on the cloud.
-
-
-
LIBERATING & STANDARDIZING DATA IN A UNIQUE PLATFORM
By M. TERRISSESummaryGeoscientists need to collaborate and iterate on data. In current eco-systems, softwares run in silo, following their own standards and communicate with complex import/export process.
Sismage-CIG, TotalEnergies’ Geosciences and Reservoir integration platform has started a digital transformation program around OSDU and cloud-based solutions with the objective to liberate and make data easily available and standardized within a unique data platform.
The approach followed and described in this paper, as an e-poster, LIBERATING & STANDARDIZING DATA IN A UNIQUE PLATFORM, develop the idea of separating data and the software, store data in a unique place in open way usable by anyone, simplifying and accelerating data exchanges.
The platform in this case is based on OSDU which is a new standard from the industry to simplify collaboration, data exchanges and traceability. In 2021 started the first implementation of OSDU-based use cases illustrating the value and benefits associated with such a transformation.
Two examples of implementation are described to prove and illustrate the concept
Digital transformation around data liberation and standardization is a key aspect for the future. The road will be long. First examples show that it is possible and very promising for valorising data, simplifying, and accelerating the decision-making process.
-