Ebook: Unconventional Methods for Geoscience, Shale Gas and Petroleum in the 21st Century

Today, geology encompasses cosmology as we bring materials from the asteroid Ryugu near Mars with the satellite Hayabusa. The materials have shed light on the possibility of life’s existence on Ryugu. This book aims to provide research directions for geology in the 21st century. However, writing a history of the present or future research directions is no easy task. Nevertheless, we endeavor to present the research directions of the early decades of the 21st century.

Drawing on past experiences of various misunderstandings in research directions, several papers in this book discuss fuzzy set theoretical approaches. Interestingly, this groundbreaking paper was initially rejected in the 1960s. Later, Professor Lotfy A. Zadeh’s paper on “fuzzy sets” was published in the International Journal of Science and Control. This paper introduced several important concepts that indicated various research directions.

The neural network was initially underestimated in the book published in 1969. This book had a significant influence that caused many researchers to shift away from neural networks. For several decades, many researchers abandoned research on neural networks. However, a few persevering individuals continued their research until the discovery of backpropagation, which not only revolutionized the field but also gave rise to deep learning.

In this book, we revisit heuristic approaches and meta-heuristic approaches. While H.A. Simon emphasized heuristic approaches for semi-structured problems, we are still searching for theoretical, mathematical, or logical approaches to problem-solving. In the 1960s, we discovered various heuristic approaches that were sought after during that era.

As Eric Hobsbawm wrote, it is difficult to write the history of one’s own days. As mentioned earlier, selecting influential methods is no easy task. We hope to include the most influential papers that represent the smart geology of the first few decades of the 21st century, with one chapter dedicated to deep learning.

We acknowledge the importance of theoretical approaches from physics, mathematics, and statistics underlying meta-heuristic methods, artificial intelligence, deep learning, and other human approaches. We must build upon human knowledge and approaches, as Herbert Alexander Simon (Nobel laureate, 1978) stated, there are infinitely many problems we cannot solve theoretically, or in other words, mathematically. However, we can make progress, as demonstrated by Andrew John Wiles (Professor, Oxford University) when he proved Fermat’s Last Theorem, left unresolved by Pierre de Fermat in 1623, in a 108-page proof published in the Annals of Mathematics in 1995. Similarly, the ABC conjecture, first proposed by Joseph Oesterlé in 1988 and D.W. Masser in 1985, was proved by Shinichi Mochizuki (Professor, Kyoto University) in approximately 600 pages in 2015 and at RIMS in March 2021. Mochizuki’s method allowed for the proof of Fermat’s problem in just a few lines, based on his method.

We expect that this book will contribute to the exploration of various possible solutions to challenging problems in both the Earth’s geometry and the cosmos, as H.A. Simon envisioned. The first paper serves as an introduction to biology, which is now expanding into the science of the cosmos.

The oil and gas industry has a wealth of information accumulated over an ex-tensive period of time. This huge information is now known as big data and needs to be managed effectively in order that the industry remain sustainable. Digital transformation provides many benefits including enhancing connectivity and even optimization of processes within the system. In the current situation, digitalization in the oil and gas industry would enable a faster solution provider and at the same time provide value to the stakeholders.

Challenges that have been identified include the possibility of the benefits of digitalization not reaching those who really need it and the increase in risk of compromising data privacy and security. Several recommendations can be identified that will benefit the industry and society. However, the successful implementation of digital transformation in the oil and gas industry requires a strong collaboration between the industry, decision-makers and the society.

The main objectives of geosciences is to find the current state of the Earth – i.e., solve the corresponding inverse problems – and to use this knowledge for predicting the future events, such as earthquakes and volcanic eruptions. In both inverse and prediction problems, often, machine learning techniques are very efficient, and at present, the most efficient machine learning technique is deep neural training. To speed up this training, the current deep learning algorithms use dropout techniques: they train several sub-networks on different portions of data, and then “average” the results. A natural idea is to use arithmetic mean for this “averaging”, but empirically, geometric mean works much better. In this paper, we provide a theoretical explanation for the empirical efficiency of selecting geometric mean as the “averaging” in dropout training.

Long-term pressure and flow rate history are important for reservoir characterization and reservoir management. However, a complete set of these data are often not available due to numerous technical difficulties. Currently, datasets with missing information are omitted and not considered for further analysis. In this study, we use machine learning algorithm via linear regression for flow rate history reconstruction. Only few studies have demonstrated the application of linear regression for well testing purposes. However, pressure and flow rate data in a producing field are comparatively longer and more complex. A combination of feature extraction and linear regression was applied for long term flow rate history reconstruction. The dataset used to evaluate the performance of the proposed method was obtained from a real producing field. This study indicated the high performance of linear regression at estimating missing flow rate history using available pressure readings in the dataset. Although linear regression has the benefits of high interpretability and fast computation time, it fails to perform well in reconstructing flow rate history when there is a significant degree of variation in the flow rate and pressure data.

Synthetic well testing is an important tool which can be utilised to understand the fractures’ influence on flow behaviour. Different fracture sets have been used to identify the well performance and the conductivity of the fracture network. Five models were built using outcrop fracture data sets with similar statistical properties, and their flow performances were analysed using the well-test response to evaluate the outcrop-related uncertainties.

The results of the aforementioned scenarios have shown remarkable differences in the pressure responses related to the degrees of fracture conductivity in each fracture set. This variation in the pressure response indicates that the higher fracture density may not necessarily result in a higher fracture conductivity.

The fracture conductivity effect was further investigated using a scenario of a producer completed in a matrix block trapped within fractures. The results have referred that the distance to the fracture and the fracture conductivity have a considerable influence on the dual-porosity signature, which may mask the radial flow response of the matrix in the derivative plot when the fractures are very close.

The outcome of this work can be used to understand the outcrop-related uncertainties as a pre-work of a full field fracture modelling and to calibrate the fracture conductivity at the well vicinity to improve the history matching. The results can also help in well-test interpretations when a similar pattern of pressure response is obtained from real well-test data.

The estimation of effective fracture permeability depends mainly on the geometry of the void space between the fracture surfaces. Sometimes, these void spaces are closed partially or totally for various reasons, which create a contact area between the fracture surfaces. These contact areas cause the fluid to follow a tortuous path around them, which reduces the permeability magnitude notably.

In this study, a digitised fracture network of a carbonate formation has been used to investigate the impact of contact areas and variable aperture width on the effective fracture permeability by using a discrete fracture networks approach. Moreover, a statistical analysis of a fracture width was used to build a stochastic aperture distribution to evaluate the fluid flow behaviour. Where, the properties of the aperture histogram are the only required parameters for the aperture modelling, in addition to several advantages in the current workflow compared with the former modelling approaches.

The results were represented by a correction curve plotted based on the 3D simulation results, which can be used to evaluate the reduction factor in fracture permeability by considering the impact of contact areas in fractures. Furthermore, it can also be utilised for the history-matching process by calibrating the fracture permeability, hence fluid flow behaviour at the well region or the reservoir scale.

Studies on fractured reservoir is becoming vital especially in shale formation due to the emerging of production for shale oil and gas globally, besides awareness of disasters due to incorrect estimation of subsurface overpressure and fracture pressure. Understanding the formation of fracture in relation to the structural evolution and its behavior in various lithofacies is important. Another major factor controlling the fracture development beside lithology and mineral composition, is abnormal pressure. Therefore, a study on fracture characterization and modelling of shale formations is still ongoing to characterize shale by evaluating lithofacies and fracture characteristics. In addition, analysis of subsurface rock properties to determine the fracture gradient in shales of gas wells and fracture models of selected shale formations are in progress. In these research, the study of fractures will be determined both by onshore and offshore data. In the onshore scale, the natural fracture will be evaluated, taking into account the natural fractures network data includes contribution of mineral composition and lithofacies in Miri region. While, offshore data will be analyzed through well log data and initial relationships of Eaton’s method in estimating fracture pressure along the wellbore. Outcrop studies was conducted around Miri vicinity mainly at Beluru and Long Lama route. Lithofacies identification was determined through X-ray diffraction (XRD), X-ray Fluorescence (XRF) and total organic carbon (TOC). Eminently, based on the XRD and XRF analysis, there are two types of shale identified; where the shale with high quartz percentage is classified as siliceous black shale and located nearby Long Lama whereas those with prominent calcite percentage is categorized as calcareous grey shale, which are found in the Beluru. In general, the type of shales around study area is referred as grey shale with good level of fracture development. The range of TOC percentages of the samples are from 2.03 to 2.28. Therefore, the result obtained from integration of well log analysis and fracture gradient prediction can be compared with natural fracture model in West Baram Delta.

Failure modes and effects analysis (FMEA) is a popular reliability tool in petroleum engineering. In FMEA, potential failure modes or corrective actions are evaluated, each assigned a Risk Priority Number (RPN) score, and prioritized for decision making. FMEA is also known as Failure Modes, Effects, and Criticality Analysis (FMECA), while focuses on failure modes prioritization. Despite of the popularity of FMEA and FMECA, it is not clear, how potential failure modes and corrective actions could be represented systematically, for effective decision making. In this paper, two new representations (i.e., a tree representation and a vector representation), for potential failure modes and corrective actions, are proposed. The tree representation for a potential failure mode allows its root cause(s), effect(s) and corrective action(s), together with their severity, occurrence and detection rating(s), to be represented as a three-layer tree model. The tree representation for a corrective action with similar contents is outlined too. The RPN model, together with its score, is represented as a node of the tree model. These tree models can also be represented as their associated equivalence layered-vector representations. In this paper, the usefulness of the proposed approaches is illustrated with benchmark FMEA worksheets pertaining to petroleum engineering.

The growing energy demand coupled with depletion of regular hydrocarbon reserves have greatly increased the significance of shale gas reservoirs. This study examines the fracture half-length and spacing affects in shale gas-reservoirs interpreted through the pressure drop rates and the production rates. This work aims to comprehend the variables, such as Klingenberg effects, Knudsen diffusion, non-Darcy flow, and the dual porosity caused by a fractured system, that influence the flow-behaviour in reservoirs of shale gas. The most fitting mathematical model for shale gas reservoirs was chosen after careful consideration of the several suggested mathematical models. Additionally, to examine and model the ideal half-length and spacing of the shale gas reservoir, suitable parameters for the reservoir system and simulation model were developed. This section discusses how the matrix permeability and the natural fracture-networks affect the fracture parameters designs. The study revealed that the reservoir parameters have a significant influence on the fracture half-length designs as well as the fracture-spacing plans. Similarly, whilst both the matrix permeability and the natural-fracture permeability effect fracture-spacing, the fracture half-length is impartial to matrix-permeabilities.

Machine learning through artificial intelligence have been successfully applied in solving variety of problems in several disciplines. In the energy sector, oil and gas industry there is potential and opportunities attract maximum investment. The benefits are (1) Reduction of operational costs, (2) Improvement in Efficiency, (3) Reduction of cycle time span, (4) Replacing knowledge and know-how of experienced staff, (5) Filling in information gap in company

A.I. Technology has tremendous potential in streamlining many processes in oil and gas both upstream and downstream.

The key benefit would be optimization and efficiency in scheduling, maintenance and product delivery. A.I. is also making inroads in Refinery Operations in corrosion detection and mitigation.

Low saline water and polymer injections are well established as effective enhanced oil recovery (EOR) techniques. However, the synergetic effect of combining the two methods is not fully studied and founded. In this study, three reservoir simulation case studies are carried out to compare the incremental oil recovery during the low flooding of saline water, polymer flooding and combined the low flooding of saline and polymer. The first two case studies employ different well patterns but of the same reservoir model. This was achieved by reconfiguring the placement of injector(s) and producer wells. In the first case, only one injector and one producer wells are located in the two far corners of the model. In the second case, four injectors and one producer wells are arranged in a 5-spot pattern. The numerical simulation results show that low salinity and high polymer concentration flooding has the highest oil recovery. It is also observed that the 5-spot well pattern has higher aerial sweep efficiency, with incremental recovery up to 67.3%. The third case study involves a heterogeneous reservoir model which is simulated using stochastically generated permeability. In all the cases, the incremental oil recovery observed during combined low salinity and polymer flooding is significantly higher than the incremental oil recoveries observed when the two are injected independently. The improved oil recovery is attributed to the synergy of the wettability alterations and better sweep efficiency of low salinity water flooding and polymer flooding, respectively.

Kick refers to uninvited influx flow from the formation into the wellbore during drilling operation. Undesired event such as non-productive time (NPT) and blowout may occur if the engineers ignore the positive indications of kick. The well should be shut-in immediately and well control procedures should take place after the kick is detected. In this study, a base model has been created in the simulation software, Drillbench. Besides, two types of shut-in methods have been evaluated and studied to investigate the relationship of different shut-in methods affecting the volume of pit gain using the software. Both shut-in methods have been simulated in the case with water-based mud and oil-based mud. The results of the studies with the quantitative difference in term of the volume of pit gain between two methods is included in this paper.

This work addresses the pressure transient behavior of horizontal gas injection well in low permeable reservoirs. Low permeable reservoirs such as shale oil reservoirs have been receiving great attentions lately which normally require hydraulic fracturing and horizontal well development to maximize the oil production. However, the primary recovery factor of shale oil reservoirs is still low and has been estimated to be below 10–15% due to tight nature of the shale formations. Enhanced oil recovery method such as miscible carbon dioxide (CO2) injection is said to be one of the most efficient and effective methods used to increase the oil recovery factor of a low permeable shale oil reservoir. The objective of this paper is to study the pressure transient behavior of the horizontal gas injection well in low permeable shale oil reservoirs using numerical simulator, CMG-GEM. Flow regimes and its significant reservoir parameters are investigated from the log-log plot of pressure-derivatives. It is found that a unit-slope line is developed on pressure-derivative log-log plot at early time due to the gas compressibility effect, followed by early radial flow and early linear flow regimes. The effect of various parameters such as gas injection rate, duration of gas injection, well location and well perforation length are studied and analyzed on the changes of pressure transient characteristics. It is identified that gas injection rate affects the pressure-derivative response significantly at middle time due to gas mobility and viscosity; whereas well location and well perforation length affect the late time pressure-derivative response which relate to dominant boundary effect; however, duration of gas injection is not able to show or prove any impacts on the pressure-derivative behavior due to numerical instability issue. Reservoir characteristics such as average permeability and skin can be identified from the flow regimes equations similar to the horizontal production well.

The movement of hydrocarbon in a shale formation is restricted due to the ultra-low permeability. The ultra-low permeability usually creates capillary entrapment due to high capillary pressure. Fracturing was introduced in shale formation which creates more cross-sectional area of flow for the oil to be produced from shale formation. Another approach to increase the flow of hydrocarbon is by decreasing the viscous force via the mean of viscosity reduction. The ability of Thermal Enhanced Oil Recovery (EOR) in recovering light oil from ultra-low permeability shale was not thoroughly studied. This case study studied the potential of injecting cyclic steam into a fractured shale formation in the variation of steam qualities and fracture configurations. The simulation study incorporated data obtained from Bakken and Eagle Ford Shale Formation to ensure a representable dynamic model. The simulation study had been conducted using in-house commercial software by constructing a dual permeability synthetic model due to stark contrast of matrix and fractures permeability. Manual numerical computation was incorporated in the model to assess the dependency of relative permeability oil in the event of change in reservoir temperature which was the case since the simulation covered Thermal EOR. Fracture half-length was determined as the most sensitive factor in contributing to increase in oil recovery among all the fracture properties. Significant recovery of additional 46 % in oil production for the most optimum case had been observed after incorporating recovery mechanisms of thermal expansion, wettability alteration and interfacial tension (IFT) reduction in the simulation. The amount of recovery increased from 1000 MSTB to 1460 MSTB when fracture half-length of 400 ft. and steam quality of 0.4 was used.

The radius of investigation is still ambiguous and there is uncertainty in radius of investigation calculation. Every changes of pressure in the reservoir will change the radius of investigation. Thus, these variations will make the maximum radius of investigation difficult to define. To analyze this uncertainty, the pressure changes in a reservoir is evaluated by using the Ei-Function equation to plot the pressure profile which is pressure versus distance of the well graph. Furthermore, the pressure profile graph can be used to set a cut off of pressure difference at the end of transient effect that can be defined as maximum radius of investigation. This project required Matlab software for analytical approach and Eclipse Simulator software for numerical approach. The numerical method is used to prove the analytical method. The analytical method will provide the pressure profile which indicate the pressure of reservoir reading further away from the well. Similarly, the numerical method will generate the pressure of reservoir numerically to indicate the same as analytical method. The homogeneous reservoir is used to analyze this ambiguity where the manipulated variable is the flowrate and production time. The preliminary interpretation showed that different flowrate will not affect the radius of investigation while different production time will affect the radius of investigation.

Multi-objective optimal power flow (OPF) focusing on fuel cost generation and emission with wind power integration is solved by proposed a SWTCM technique developing with non-dominated sorting (NS) and PSO algorithms. The SWTCM_NSPSO develops the balancing search solution competency of global best investigation and use of local best including the stochastic weight trade-off mechanism cooperating with coefficients with dynamistic trade-off technique. This improved algorithm combines with chaotic mutation to increase search efficiency, diversion, and prevent premature convergence problem. Crowding distance (CD) and NS approaches remarkably optimize the best Pareto front cooperating with Fuzzy selecting function for providing the local best solution. Two stages method is created to choose the best optimum resolution (global) from many trials of best compromise group (local). The modified IEEE 30 bus test system is investigated by using SWTCM_NSPSO. The simulation results clearly simulated a lower values set and better Pareto fronts distribution curve than other methods e.g. simple NSGAII and NSPSO, chaotic deviation combined with non-dominated sorting PSO, and trade-off stochastic method integrating with NSPSO leading to save operation fuel cost and reduce pollutant emission through provide a better multi- objective trade-off solution.

Sensor drift is a phenomenon which indicates unexpected variations in the sensory signal responses beneath the same working conditions. In this paper, a competitive co-evolutionary (ComCoE) Multilayer Perceptron artificial neural network (MLPN) is applied to detect chemical gas sensor drift. The efficiency of the ComCoE MLPN in detecting chemical gas sensor drift is evaluated as well as compared with the performance of other classification methods from the literature. The proposed ComCoE MLPN has shown promising preliminary results in this application.

Stackelberg games with simple recourses are formulated, in which the coefficients of equality constraints are represented as continuous type fuzzy random variables or discrete type ones. In general, these problems cannot be solved by applying the conventional methods, since simple recourses involving fuzzy random variables have not been defined. To cope with such fuzzy random variables, a possibility measure concept is applied. Then, two kinds of Stackelberg games with simple recourses are converted into usual optimization problems for some fixed degrees of permissible possibility levels specified by the first level’s player. For the transformed optimization problems, the Stackelberg solution concept of the first level’s player is introduced. By using an example, the property of the proposed Stackelberg solution concept is explained.

In this chapter, we explore the application of the derivative of disproportion functions in developing a cryptographic system and a pattern recognition technique. Firstly, we present an algorithm that utilizes focal functions, known as functions of disproportion, in a cryptographic system. The transmitted symbols are encrypted using the weighted sum of at least two of these functions with randomly generated coefficients. Numerical experiments demonstrate the robustness and reliability of the proposed procedure.

Furthermore, we demonstrate how the derivative of disproportion functions can govern a dynamic process, serving as a tool to identify the form of a real-valued function. This process enables us to determine the class to which the function belongs, independently of its unknown parameter values.

This paper presents Enumeration Method in gas condensate reservoir simulation to study the condensate banking complex physics phenomena. Initially, coarse scale grid is commonly used for gas condensate reservoir simulation study. Nevertheless, the coarse scale simulation disregards the condensate bank or it is not able to demonstrate the precise distribution and effects. By introducing Local Grid Refinement (LGR) in simulation model arguably brings a better representation of the condensate bank effect near wellbore but significantly increases the run time. This become severe especially in full field modeling with comingled production. Therefore, enumeration initialization approach was developed to divide the simulation explicitly in coarse scale simulation. During the stops, a region near wellbore was designed where condensate bank parameters were modified based on the history matching. Hence, the drastic change of well performance due to condensate banking could be captured. This drastic change could not physically described in conventional coarse scale simulation model, thus affect prediction accuracy. Comparison between enumeration ways with conventional approach were then investigated. It was found that enumeration method shows a better prediction in investigating the behavior. This is due to its ability to predict mobility changes due to condensate banking, consequently, improve the condensate bank characterization.

Murteree organic shale are one of most rich unconventional gas reservoirs in Cooper Basin, South Australia. In shale gas reservoirs, estimation of water saturation and porosity are very erroneous when calculated from the well logs. In Roseneath and Murteree shales of Copper Basin in South Australia, we investigated their fluids contents using Archie model, Simandoux, Indonesian and Dual water model. Later, brine estimation from Waxman-Smits was correlated with Archie, Simandox, Indonesian and Dual water model results. Due to presence of high amount of clay minerals, conductivity and clay bound water are very high, the changes in salinity level of the brine due to post depositional features, followed by physical and chemical changes effect the determination of water saturation consequently. High clay volume in shale gas reservoirs cause over estimation of water saturation when determined by Archie and other resistivity models. Waxmans-Smiths equation gives more reliable results because it accounts the cation exchange capacity and conductance of clay minerals present in the formation. Also it is noticed that as the clay mineralogy percentage, clay bound water also increases with same pattern respectively. In this paper, description of Waxmans-smith equation in determination of water saturation has been presented and also relation between clay minerals and clay bound water has been discussed. Form our analysis, it has been estimated that, Waxmans-Smit conductivity model gives estimation of water more reliable as compare to resistivity models like Archie, Simandoux and Indonesian in unconventional shale gas reservoir. Hydrocarbon potential are present with respect to water saturation and resistivity logs.

The main goal of using surfactants as a fracture agent in tight shale gas reservoirs is to minimize the capillarity, interfacial tension, modify contact angle and reservoir wettability. Most of the recent studies conducted similar experiments at ambient conditions. However, one of the limitations of different previously laboratory studies is the lack of measurements with gas as they were done using air or crude oil. The major component in shale gas is methane plus some lighter hydrocarbon. In this chapter we will investigate the surfactant solution behavior with some light hydrocarbons such as n-Heptane to better mimic reservoir hydrocarbon behavior. All necessary laboratory experiments had been conducted plus phase behavior for the selected surfactants. Anionic surfactants gave excellent aqueous stability results however; the impact of the salinity was observed carefully. An optimized formulation was achieved that resulted in type III microemulsion.