A Single Artificial Neural Network Model Predicts Bubble Point Physical Properties of Crude Oils

Reservoir fluid properties at bubble points play a vital role in reservoir and production engineering computations. Ideally, the bubble point physical properties of crude oils are obtained experimentally. On some occasions, these properties are neither available nor reliable; then, empirically derived correlations or artificial neural network models are used to predict the properties. This study presents a new single multi-input multi-output artificial neural network model for predicting the six bubble point physical properties of crude oils, namely, oil pressure, oil formation volume factor, isobaric thermal expansion of oil, isothermal compressibility of oil, oil density, and oil viscosity. A large database comprising conventional PVT laboratory reports was collected from major producing reservoirs in the Middle East. The model input is constrained mathematically to be consistent with the limiting values of the physical properties. The new model is represented in mathematical format to be easily used as empirical correlations. The new neural network model is compared with popular fluid property correlations. The results show that the developed model outperforms the fluid property correlations in terms of the average absolute percent relative error.

Inline Drilling Fluid Property Measurement, Integration, and Modeling to Enhance Drilling Practice and Support Drilling Automation

Frequent, reliable, and repeatable measurements are key to the evolution of digitization of drilling information and drilling automation. While advances have been made in automating the drilling process and the use of sophisticated engineering models, machine learning techniques to optimize the process, and lack of real-time data on drilling fluid properties has long been recognized as a limiting factor. Drilling fluids play a significant function in ensuring quality well construction and completion, and in-time measurements of relevant fluid properties are key to automation and enhancing decision making that directly impacts well operations. This paper discusses the development and application of a suite of automated fluid measurement devices that collect key fluid properties used to monitor fluid performance and drive engineering analyses without human involvement. The deployed skid-mounted devices continually and reliably measure properties such as mud weight, apparent viscosity, rheology profiles, temperatures, and emulsion stability to provide valuable insight on the current state of the fluid. Real-time data is shared with relevant rig and office- based personnel to enable process monitoring and trigger operational changes. It feeds into real-time engineering analyses tools and models to monitor performance and provides instantaneous feedback on downhole fluid behavior and impact on drilling performance based on current drilling and drilling fluid property data. Equipment reliability has been documented and demonstrated on over 30 wells and more than 400 thousand ft of lateral sections in unconventional shale drilling in the US. We will share our experience with measurement, data quality and reliability. We will also share aspects of integrating various data components at disparate time intervals into real-time engineering analyses to show how real-time measurements improve the prediction of well and wellbore integrity in ongoing drilling operations. In addition, we will discuss lessons learned from our experience, further enhancements to broaden the scope, and the integration with operators, service companies and other original equipment manufacturer in the domain to support and enhance the digital drilling ecosystem.

A new method for logging identification of fluid properties in tight sandstone gas reservoirs based on gray correlation weight analysis-A case study of the Middle Jurassic Shaximiao Formation on the eastern slope of the Western Sichuan Depression, China

Energy exploration is becoming increasingly complex worldwide, and tight sandstone gas is an important field for the future development of the oil and gas industry. For the reservoir properties of the Shaximiao Gas Reservoir on the eastern slope of the Western Sichuan Depression in the Sichuan Basin, western China, it was found that the low-resistance characteristics of the reservoir complicate the gray characteristics among reservoir fluid property parameters. Some commonly used fluid property identification techniques, such as the flow zone index method, correlation analysis method of logging parameters, and traditional mathematical statistical methods, have poor fluid property evaluation results. Therefore, how to eliminate the influence of the gray features among the reservoir parameters on the identification of reservoir fluid properties and how to accurately identify the reservoir fluid properties are urgent problems that need to be solved. In this paper, we proposed a new method for identifying the fluid properties of tight sandstone reservoirs by combining gray system theory and multivariate statistical theory. This method can perform gray correlation weight analysis on parameters (combined parameters) closely related to fluid properties; furthermore, the logging identification method based on gray correlation weight analysis is used to identify reservoir fluid properties. The results show that the gray correlation weight analysis can accurately characterize the gray characteristics of reservoir fluid parameters, and the gray comprehensive correlation weight results are in good agreement with the production status of the studied gas reservoir. We used the method to identify the fluid properties of the target layer in 58 wells in the study area, and the discrimination rate of the model was 86.5%. In addition, the new model was used to predict the reservoir fluid properties of 12 newly drilled wells in the study area, and the accuracy of the reservoir fluid property prediction was 91.67%.

Predicting Reservoir Fluid Properties from Advanced Mud Gas Data

Summary In a recent paper, we published a machine learning method to quantitatively predict reservoir fluid gas/oil ratio (GOR) from advanced mud gas (AMG) data. The significant increase of the model accuracy compared to traditional modeling approaches makes it possible to estimate reservoir fluid GOR based on AMG data while drilling, before the wireline operation. This approach has clear advantages because of early access, low cost, and a continuous reservoir fluid GOR for all reservoir zones. This paper releases further study results to predict other reservoir fluid properties in addition to GOR, which is essential for geo-operations, field development plans, and production optimization. Two approaches were selected to predict other reservoir fluid properties. As illustrated by the reservoir fluid density example, we developed machine learning models for individual reservoir fluid properties for the first approach, similar to the GOR prediction approach in the previous paper. As for the second approach, instead of developing many machine learning models for individual reservoir fluid property, we investigated the essential properties for equation of state (EOS) fluid characterization: C6 and C7+ composition and the molecular weight and density of the C7+ fraction. Once these properties are in place, the entire spectrum of reservoir fluid properties can be calculated with the EOS model. The results of reservoir fluid property prediction are satisfactory with both approaches. The reservoir oil density prediction has a mean average error (MAE) of 0.039 g/cm3. The accuracy is similar to the typical density derived from the pressure gradient from wireline logging data. For the essential fluid properties required for EOS model prediction, the overall accuracy is less than the laboratory measurements but acceptable as the early phase estimations. The reservoir fluid properties predicted from the EOS model are similar to the predictions from individual machine learning models. We applied the field measured AMG data into the reservoir fluid property models and achieved good results, as illustrated by the reservoir fluid density example. The previous paper completed the methodology to predict all reservoir fluid properties based on AMG data. This work paves the way to generate a complete reservoir fluid log for all relevant reservoir fluid properties while drilling. The method has a significant business impact, providing full coverage of reservoir fluid properties along the well path in the early drilling phase. The advantage of providing reservoir fluid properties in all reservoir zones while drilling far outweighs the limitation of somewhat reduced reservoir fluid property accuracy.