Search

Well known oil recovery factor estimation techniques such as analogy, volumetric calculations, material balance, decline curve analysis, hydrodynamic simulations have certain limitations. Those techniques are time-consuming, require specific data and expert knowledge. Besides, though uncertainty estimation is highly desirable for this problem, the methods above do not include this by default. In this work, we present a data-driven technique for oil recovery factor (limited to water flooding) estimation using reservoir parameters and representative statistics. We apply advanced machine learning methods to historical worldwide oilfields datasets (more than 2000 oil reservoirs). The data-driven model might be used as a general tool for rapid and completely objective estimation of the oil recovery factor. In addition, it includes the ability to work with partial input data and to estimate the prediction interval of the oil recovery factor. We perform the evaluation in terms of accuracy and prediction intervals coverage for several tree-based machine learning techniques in application to the following two cases: (1) using parameters only related to geometry, geology, transport, storage and fluid properties, (2) using an extended set of parameters including development and production data. For both cases model proved itself to be robust and reliable. We conclude that the proposed data-driven approach overcomes several limitations of the traditional methods and is suitable for rapid, reliable and objective estimation of oil recovery factor for hydrocarbon reservoir.

SummaryWe studied the applicability of a gradient-boosting machine-learning (ML) algorithm for forecasting of oil and total liquid production after hydraulic fracturing (HF). A thorough raw data study with data preprocessing algorithms was provided. The data set included 10 oil fields with more than 2,000 HF events. Each event has been characterized by well coordinates, geology, transport and storage properties, depths, and oil/liquid rates before fracturing for target and neighboring wells. Each ML model has been trained to predict monthly production rates right after fracturing and when the flows are stabilized. The gradient-boosting method justified its choice with R2 being approximately 0.7 to 0.8 on the test set for oil/total liquid production after HF. The developed ML prediction model does not require preliminary numerical simulations of a future HF design. The applied algorithm could be used as a new approach for HF candidate selection based on the real-time state of the field.

The approaches of building and methods of using the digital core are currently developing rapidly. The use of these methods makes it possible to obtain petrophysical information by non-destructive methods quickly. Digital rock physics includes two main stages: constructing models and modeling various physical processes on the obtained models. Our work proposes using deep learning methods for mineral and pore space segmentation instead of classical methods such as threshold image processing. Deep neural networks have long been able to show their advantages in many areas of computer vision. This paper proposes and tests methods that help identify different minerals in images from a scanning electron microscope. We used images of rocks of the Achimov formation, which are arkoses, as samples. We tested various deep neural networks such as LinkNet, U-Net, ResUNet, and pix2pix and identified those that performed best in segmentation.

To reduce inefficient costs and environmental risks, oil companies strive to optimize the process of hydrocarbon production at all stages of field development, including geological and technical works at wells. In particular, it is important to predict fluid production with high accuracy. 3D hydrodynamic modeling is a generally accepted technique for solving this problem. It provides reliable results but requires many input data, computational resources, and time for calculations. Since the decision-making process has to be reactive, it is necessary to develop a simultaneously precise and prompt predictive instrument for quick forecasts of liquid production. The most promising tools for these purposes are proxy models based on solving the material balance equation. They adapt to the existing historical data even without PVT properties and reservoir data. Some of the most popular approaches are proxy models such as Capacitance Resistance Models (CRM). CR-type model is a material balance-based flow model, which provides preferable transmissibility trends, the presence of sealing or leaking faults with compressibility effects in consideration, and dissipation between injector-producer pairs. It is a data-driven model with adjustable time constants and interwell connectivity parameters. Before the model tuning, all parameters must be initialized with analytical or random approximations, and then they can be found by an appropriate optimization procedure. Historical-based Capacitance Models can be applied to poorly studied fields. Besides, they give an opportunity to rapidly optimize field development strategy by making calculations with different well exploitation parameters. They only require historical data of hydrocarbon production volumes, injection profiles, and bottom-hole pressure dynamics as input data. One of the main is that properties in the interwell space are estimated approximately and considered to be constant throughout the entire development history. However, this is a weak assumption in the case of including well interventions and stimulations. Thus, the main goal of this work is to adjust coefficients online to changes in well operation modes, introducing new wells or shut-in the existing ones. Since the governing equation includes the considered CRM improvement, users can perform optimization over different timespans, including "special" intervals. As a result, weighting connectivity parameters of the model can be depicted on a map of well interactions versus time.

DIII-D L-mode experiments with local boron powder injection for real-time wall conditioning have been interpreted for the first time with the 3D plasma edge transport Monte Carlo code EMC3-EIRENE. Local B sourcing in plasma scenarios with upstream densities 1 . 5 ⋅ 1 0 19 m−3 and 2.2 MW heating results in a non-axisymmetric B distribution in the scrape-off layer (SOL) and on the divertor. The SOL frictional flows at high plasma density cause a strong inboard drag of injected impurities ( ≈ 90 % ), while lower background plasma densities tend to result in a more uniform distribution. The thermal forces prevent B deposition in the near SOL while the frictional force causes B fluxes to cover the divertor plasma-facing components in a region 7–10 cm beyond the strike line. Radiative dissipation occurs for B influxes above 1 ⋅ 1 0 20 s−1 and causes a moderate, non-axisymmetric reduction of the far SOL divertor heat fluxes. A comparison of top and midplane B injection shows no substantial difference in inboard vs. outboard asymmetries of the B distribution. On the other hand, erosion or recycling at the strike line may distribute the boron more uniformly in the SOL.

Summary The automatization is a modern trend in various field of geology. In this work we present a system which were constructed based on convolutional neural network (CNN) for automated core description. The system was successfully applied to production data. The application of the system speeds up the core description process in 7x. A sedimetnologist spent 40 minutes to describe 60 meters of core in a scale of 1:10cm instead of 5 hours. The results are stored in digital format which removes all paperwork. The system helps to describe most of required lithologic types (rock type and its structure). In case of missed rare lithotype – user can add it to the system. A pipeline to prepare and train the CNN model described.

The approach to handle the unresolved pores at 3D X-ray Micro Computed Tomography (μCT) images of core samples is developed. It enables a sufficient widening of digital rock studies for tight rocks. The μCT images of a low-permeable sandstone with a resolution of 1.2 μm/voxel have been generated. Pore Size Distribution shows the presence of a significant amount of sub-resolution pores. Downsampling has been applied to estimate the actual porosity with extrapolation. Visual noise, artifacts, and roundoff errors are the major factors affecting the quality of μCT images. We apply transform and spatial domain filtering to minimize all the artifacts. Regarding the overall concept of porosity and through running a geometrical histogram analysis, the Random Walker segmentation as a robust mathematical algorithm has been applied to turn the greyscale μCT images into binary ones resembling pores and grains. Next, the porosity of the binary images with a resolution of 1.2 μm/voxel has been calculated. The procedure continues with downsampling to artificially reduce the resolution and calculate the corresponding porosity. It has been observed that the calculated porosity for the highest resolution of 1.2 micrometer is still lower than the experimental value which is due to the existence of pores which their sizes are less than 1.2 micrometer, and cannot be seen in the CT images. In order to take the effects of sub-resolution pores into account, an extrapolation relying on the downsampling technique has successfully been applied. The implemented technique is based on the fact that the porosity of the reservoir rock sample is not a function of resolution. However, plotting of the calculated porosities versus their relevant resolutions indicates that the value of porosity has an inverse relationship with the voxel size. In other words, it could be interpreted that the closest values of the calculated porosity to the laboratory reports will be the output of the image processing when the size of voxel moves towards zeros as much as possible, which is technically impossible. Instead, a trendline can be fitted into the scatter plot of porosity versus resolution and find its extrapolation value for the voxel size of zero, which provides the porosity as close as possible to the experimental value. The main logic behind the digital core analysis is to calculate the properties only according to the digital images. Although there are some studies in which modifications have been done to consider the effects of sub-resolution pores, they are severely suffering from mathematical complexities, and they are mainly based on the global thresholding. The proposed technique can provide an accurate value of porosity when there are no additional data about the pore structure of the sub-micron scale.

The objective of this work is to study the applicability of various machine learning algorithms for the prediction of some rock properties which geoscientists usually define due to special laboratory analysis. We demonstrate that these special properties can be predicted only basing on routine core analysis (RCA) data. To validate the approach, core samples from the reservoir with soluble rock matrix components (salts) were tested within 100 + laboratory experiments. The challenge of the experiments was to characterize the rate of salts in cores and alteration of porosity and permeability after reservoir desalination due to drilling mud or water injection. For these three measured characteristics, we developed the relevant predictive models, which were based on the results of RCA and data on coring depth and top and bottom depths of productive horizons. To select the most accurate machine learning algorithm, a comparative analysis has been performed. It was shown that different algorithms work better in different models. However, two-hidden-layer neural network has demonstrated the best predictive ability and generalizability for all three rock characteristics jointly. The other algorithms, such as support vector machine and linear regression, also worked well on the dataset, but in particular cases. Overall, the applied approach allows predicting the alteration of porosity and permeability during desalination in porous rocks and also evaluating salt concentration without direct measurements in a laboratory. This work also shows that developed approaches could be applied for the prediction of other rock properties (residual brine and oil saturations, relative permeability, capillary pressure, and others), of which laboratory measurements are time-consuming and expensive.

The main objective of this work is to study formation damage phenomenon associated with in-situ migration of fines in rock samples from terrigenic reservoirs of Eastern Siberia experimentally. Results of oil flooding for the wide range of flow rates and permeabilities show that oil does not lead to fine mobilization and its consequent retention with permeability reduction. The permeability decrease is observed only when water is flowing in the porous space. Water itself is shown to be a strong driver of self-colmatation, which is likely because of its intensive physicochemical reaction with pore surface. Performing multiple core flooding tests and analyzing the data, we define the range of rock properties and flow conditions corresponding to development of the situ formation damage (self-colmatation). We classify self-colmatation over 5 types characterized with different behaviors of permeability decrease. We comprehensively study the dependencies of permeability reduction specifics on rock properties and flooding conditions. The dependencies account for multiple parameters correlated with permeability reduction: absolute permeability, residual oil saturation, three characteristic pore diameters, four characteristic grain diameters and four major minerals within the rock structure. We discovered that the increase of formation damages are associated with high absolute permeability and low residual oil saturations. It was shown that characteristic size of migrating fines is 0.01–0.005 m m and the diameter of pores where these fines can be strained is 12 μ m . We derive a single numerical criterion to estimate the influence of 13 parameters on colmatation process. The absolute permeability and the fraction of pore throats with diameters higher than 30 μ m are shown to be the strongest influencers on colmatation.

The structure of the non-axisymmetric heat load distribution at the divertor plates is determined not only by the toroidal but also from the poloidal spectrum of non-axisymmetric field perturbations. Whether they are intrinsic, like error fields, or they are applied through 3D coils, the non-axisymmetric fields produce complex 3D edge magnetic topologies (footprints) that alter the properties of the heat and particle flux distributions on the divertor target plates. In this manuscript, a study of the impact of applied 3D field poloidal spectrum on the footprint size and structure is done for the DIII-D tokamak using the resistive MHD code M3D-C1 coupled with the field line tracing code TRIP3D. To resolve the impact of the poloidal spectrum of the magnetic perturbation, the relative phase of the two rows of in-vessel 3D coils used to produce both a n = 2 and a n = 3 perturbation is varied, where n is the toroidal harmonic of the magnetic perturbation. This shows that the largest footprint is predicted when the relative phase of the two rows is close to zero, which is also where the resonant coupling with the plasma is maximized. These results suggest that it will be challenging to decouple the footprint size from the requisite resonant coupling for RMP–ELM control. The correlation between the measured heat load and particle flux distributions at the outer divertor plates in DIII-D and the magnetic measurements is in good agreement with the predicted dependence of the magnetic footprint size on the amplitude of the resonant component of the plasma response.

Summary The algorithms for the classification of images are well developed in recent years. They work well when the classes are clearly defined between each other. The geological classes are not like that because they can be observed in different scales and classification paradigms. To tackle this problem, we compare different feature extraction techniques and classification (semi- and supervised) algorithms. We present methods which help to increase the accuracy of rock type classification.

Summary In petroleum systems, geological research studies include the classification of formation lithology by using weblogging methods. With the development of computer technology and decision-making systems based on machine learning methods, interest to automate lithology classification tasks from well logging is growing. This is especially important when manually processing large amounts of information. In this paper, we aim to define the most effective machine learning techniques for well log-based determination of lithology on the example of oil field in western Siberia, Russia. There are 86 wells each of which contains 6 physical parameters: electrical resistivity, gamma-ray, density, well diameter, photoelectric factor and neutron porosity. We applied a Linear model, k-nearest neighbors, Artificial Neural Network, Gradient Boosting, Random Forest, AdaBoost, and recently developed gradient boosted decision tree (GBDT) systems, namely, XGBoost, LightGBM, and CatBoost on real data. The performance of algorithms was compared and evaluated using metrics such as accuracy and macro F1-score (F1) on the test set after hyperparameter tuning. As a result, among the applied algorithms, we found that CatBoost possessed the highest metrics. In our work, we have demonstrated the high generalizing ability of Machine learning methods within the chosen research object for well-logging based lithology classification problem.

A framework for porous media topology reconstruction from petrographic thin sections for clastic rocks is proposed. The framework is based on two sequential stages: segmentation of thin sections imagesinto grains, porous media, cement (with further mineralogical classification of segmented elements) and reconstructing a three-dimensional voxel model of rock at pore scale. The framework exploits machine learning algorithms in order to segment2D thin section images, perform structural and mineralogical classification of grains, cement, pore space, and reconstruct 3D models of porous media. Segmentation of petrographic thin section images and mineral classification of the segmented objects are performed by the means of combination of image processing methods and Convolutional Neural Networks (CNNs). The 3D porous media reconstruction is done by means of the Generative Adversarial Networks (GANs) are applied to the segmented and classified 2D images of thin sections. As the criteria of the reconstruction quality, the following metrics were numerically calculated and compared for original and reconstructed synthetic 3D models of porous rocks: Minkowski functionals (porosity, surface area, mean breadth, Euler characteristic) and absolute permeability. Absolute permeability was calculated using pore network model. The 3D reconstruction framework was tested on a set of thin sections and CT tomograms of the clastic samples from the Achimovskiy formation (Western Siberia). The results showed the validity of the goodness-of-fit metrics based on Minkowski functionals for reconstruction the topology of porous media. The combined usage of CNN and GAN allowed to create a robust 3D topology reconstruction framework. The calculated poroperm characteristics are correlated with laboratory measurements of porosity and permeability. The developed algorithms of automatic feature extraction from petrographic thin sections and 3D reconstruction based on these features allow to achieve the following goals. First is the reduction of the amount of the routine work done by an expert during petrographic analysis. Second leads to the reduction of the number of expensive and time-consuming CT scannings required for each physical sample in order to perform further absolute and relative permeability calculations. The proposed method can bring the petrographic thin section and CT data analysis to a new level and significantly change traditional core experiments workflow in terms of speed, data integration and rock sample preparation.

More than a half of world's hydrocarbon reserves is presented in carbonate reservoirs. Conventional waterflooding leads to inefficient oil recovery from these reservoirs, because majority of them have mixed or oil-wet wetting properties. It is well documented in literature, that the main reason of oil wetness of carbonate rocks is adsorbed components from crude oil. Although progress has been made in determination of oil components, which have a tendency to react with carbonates, carbonate reservoirs development still remains challenging. Hence, in this study we investigated the distribution of adsorbed oil components on rock surfaces in order to define their influence on fluids flow through porous carbonate samples. This work presents the results for several carbonate core samples taken from the oil zone of an oil reservoir, which mostly consist of calcite with the small impurities of magnesite and quartz. The work provides the standard study of pore structure of samples to assess the solvents influence on pore network of samples using μCT; the method of evaluation of the amount of organic matter adsorbed on calcite using Rock - Eval pyrolysis; the visualization of such matter distribution through samples; and also the results of kinetics experiments in order to evaluate the bond disruption energy between organic matter and surface. Studies have shown that combination of pyrolysis and μCT provides comprehensive and improved data about organic matter.

Nowadays, the advantages of Digital Rock Physics (DRP) are well known and widely applied in comprehensive core analysis. It is also known that the quality of the 3D pore scale model drastically influences the results of rock properties simulation, which makes the preprocessing stage of DRP very important. In this work, we consider the application of Deep Convolutional Neural Networks (CNNs) for the preprocessing of CT images, specifically for denoising, in two setups - conventional fully-supervised learning and the self-supervised learning, when the only available data is the noisy images. To train CNNs in a supervised setup, we use images processed by a combination of bilateral and bandpass filters. We trained CNNs of the same architecture with different loss functions to find out how the choice of a loss function influences the model's performance. Some of the obtained CNNs yielded the highest quality in terms of full-reference and no-reference metrics and significant histogram effect (bimodal intensity distribution). Images denoised with these models were qualitatively and quantitatively better than the reference “ground truth” images used for training. We use the Deep Image Prior algorithm to train denoising models in a self-supervised setup. The obtained models are much better than ones obtained in fully-supervised setup, but are too slow, as they are optimization-based rather than feed-forward. Such an algorithm can be used in the dataset generation for feed-forward meta-models. These results could help to develop an AI-based instrument to build high-quality 3D segmented models of rocks for DRP applications.

A newly developed plasma response model, combining the nonlinear two-fluid MHD code TM1 and toroidal MHD code GPEC run in ideal mode, quantitatively predicts the narrow isolated q95 windows (Δq95 ∼ 0.1) of edge-localized mode (ELM) suppression by n = 1, 2, and 3 resonant magnetic perturbations (RMPs) in both DIII-D and KSTAR tokamaks across a wide range of plasma parameters. The key physics that unites both experimental observations and our simulations is the close alignment of essential resonant q-surfaces and the location of the top of the pedestal prior to an ELM. This alignment permits an applied RMP to produce field penetration due to the lower E × B rotation at the pedestal top rather than being screened. The model successfully predicts that narrow magnetic islands form when resonant field penetration occurs at the top of pedestal, and these islands are easily screened when q95 moves off resonance, leading to very narrow windows of ELM suppression (typically Δq95 ∼ 0.1). Furthermore, the observed reduction in the pedestal height is also well captured by the calculated classical collisional transport across the island. We recover observed q95, βN and plasma shape dependence of ELM suppression due to the effect of magnetic islands on pedestal transport and peeling-ballooning-mode stability. Importantly, experiments do occasionally observe wide windows of ELM suppression (Δq95 > 0.5). Our model reveals that at low pedestal-top density multiple islands open, leading to wide operational windows of ELM suppression consistent with experiment. The model indicates that wide q95 windows of ELM suppression can be achieved at substantially higher pedestal pressure with less confinement degradation in DIII-D by operating at higher toroidal mode number (n = 4) RMPs. This can have significant implications for the operation of the ITER ELM control coils for maintaining high confinement together with ELM suppression.

Pedestal collapse (i.e., the complete loss of the edge transport barrier (ETB)) in DIII-D H-mode plasmas occurs when resonant magnetic perturbations (RMPs) penetrate the steep gradient region at the plasma edge. Normally, RMP driven magnetic islands can occur at the top and bottom of the H-mode pedestal and these islands generate conditions consistent with edge-localized-mode (ELM) suppression and density pump-out, respectively, based on nonlinear two-fluid MHD simulations. In contrast, MHD simulations show that the steep pressure gradient region between the top and bottom of the DIII-D pedestal is generally immune to resonant field penetration due to large local E × B and diamagnetic flows. By this fortuitous circumstance, the edge-transport-barrier and H-mode confinement can be maintained while achieving ELM suppression. However, pedestal collapse can occur in DIII-D when the screening flows are inadequate to prevent field penetration in the steep gradient region of the pedestal. Non-linear two-fluid MHD simulations support the role of resonant field penetration in pedestal collapse for DIII-D H-mode plasmas with weak edge E × B and diamagnetic screening flows. ITER will likely have weaker edge screening flows than present experiments due to its much larger size, making it more susceptible to resonant field penetration in the steep gradient region of the pedestal. Analysis of model ITER equilibria demonstrates that resonant field penetration in the steep pressure gradient region is possible for RMP levels of the order required for ELM suppression. The effect of such penetration on the ITER pedestal will depend sensitively on the resulting degree of island overlap.

The development of tight hydrocarbon resources has a significant impact on the future of the energy market. Digital Rock Physics (DRP) is technically recognized as an effective method for characterizing reservoir rock properties. However, there is uncertainty about what result the user-defined chain of Digital Image Processing (DIP) and Pore-Scale Simulation (PSS) will produce when applying DRP. The uncertainty surrounding this technology faces even a greater deal in tight formations where a considerable portion of pores are sub-resolved and cannot explicitly be seen in the reconstructed micro-scale x-ray Computed Tomography (μxCT) images. Regarding the μxCT images of Achimovskiy formation, which is classified as a tight sandstone resource, the addressed uncertainty has been evaluated based on various scenarios of well-known DIP and PSS approaches. The results show that image processing and segmentation impact on the absolute permeability values stronger than calculation methods. Furthermore, the cross-plot analysis shows that simple image processing and segmentation indicate the same trend as precise image processing and segmentation. The generated results for the studied formation show that if the experimental open porosity of a core sample is higher than 5 % , μxCT images could feasibly resolve the system of connected pores.

A convective heat flux model for perturbed plasmas, based on guiding center ion drift in vacuum fields (Wingen, et al 2014 Phys. Plasmas 21 012509), has been updated. The old model only considered ion heat flux, while here also electron heat flux is included. The updated model predicts divertor heat flux distributions in non-axisymmetric (3D) plasmas with applied resonant magnetic perturbation fields, and includes electric scalar potentials. It is found that a radial electric field in the near scrape-off layer can considerably shift the footprints toroidally, leading to a smearing out effect of the incident heat flux, while a simple model for sheath potential has little impact on footprints. Various approaches to model electron heat flux are studied. A convective electron model, based on collisionless free streaming, is found to yield the best agreement with measurements, while a conductive model requires a flat temperature gradient inside lobes to yield acceptable peak heat flux values. A heuristic heat flux layer approach, based on a fixed layer width also requires a limited heat flux inside the last closed flux surface (LCFS); by selecting various locations of the LCFS, the results of the conductive or convective model can be recovered respectively. The sum of ion and electron heat fluxes, both obtained by the convective model, is compared to experimental data for multiple time slices in DIII-D. Strike point splitting is observed with peak heat fluxes and layer widths that compare well to infrared camera measurements.

DIII-D research is addressing critical challenges in preparation for ITER and the next generation of fusion devices through focusing on plasma physics fundamentals that underpin key fusion goals, understanding the interaction of disparate core and boundary plasma physics, and developing integrated scenarios for achieving high performance fusion regimes. Fundamental investigations into fusion energy science find that anomalous dissipation of runaway electrons (RE) that arise following a disruption is likely due to interactions with RE-driven kinetic instabilities, some of which have been directly observed, opening a new avenue for RE energy dissipation using naturally excited waves. Dimensionless parameter scaling of intrinsic rotation and gyrokinetic simulations give a predicted ITER rotation profile with significant turbulence stabilization. Coherence imaging spectroscopy confirms near sonic flow throughout the divertor towards the target, which may account for the convection-dominated parallel heat flux. Core-boundary integration studies show that the small angle slot divertor achieves detachment at lower density and extends plasma cooling across the divertor target plate, which is essential for controlling heat flux and erosion. The Super H-mode regime has been extended to high plasma current (2.0 MA) and density to achieve very high pedestal pressures (~30 kPa) and stored energy (3.2 MJ) with H 98y2 ≈ 1.6–2.4. In scenario work, the ITER baseline Q = 10 scenario with zero injected torque is found to have a fusion gain metric independent of current between q 95 = 2.8–3.7, and a lower limit of pedestal rotation for RMP ELM suppression has been found. In the wide pedestal QH-mode regime that exhibits improved performance and no ELMs, the start-up counter torque has been eliminated so that the entire discharge uses ≈0 injected torque and the operating space is more ITER-relevant. Finally, the high- (⩽3.8) hybrid scenario has been extended to the high-density levels necessary for radiating divertor operation, achieving ~40% divertor heat flux reduction using either argon or neon with P tot up to 15 MW.

Summary The main aim of this work is to study the applicability of Machine Learning (ML) techniques for prediction of rock properties, which are commonly defined via special core analysis (SCAL). The mechanism of SCAL prediction on the basis of routine core analysis (RCA) was developed and validated. The possibility of application of ML methods for estimation of some rock characteristics was demonstrated. The comparative analysis of different ML techniques was provided to choose the most stable and accurate forecast methods. It was shown that Gradient Boosting algorithm and Artificial Neural Network allow to create the most robust and accurate models for considered rock properties.

Edge-localized mode (ELM) suppression by resonant magnetic perturbations (RMPs) generally occurs over very narrow ranges of the plasma current (or magnetic safety factor q_{95}) in the DIII-D tokamak. However, wide q_{95} ranges of ELM suppression are needed for the safety and operational flexibility of ITER and future reactors. In DIII-D ITER similar shape plasmas with n=3 RMPs, the range of q_{95} for ELM suppression is found to increase with decreasing electron density. Nonlinear two-fluid MHD simulations reproduce the observed q_{95} windows of ELM suppression and the dependence on plasma density, based on the conditions for resonant field penetration at the top of the pedestal. When the RMP amplitude is close to the threshold for resonant field penetration, only narrow isolated magnetic islands form near the top of the pedestal, leading to narrow q_{95} windows of ELM suppression. However, as the threshold for field penetration decreases with decreasing density, resonant field penetration can take place over a wider range of q_{95}. For sufficiently low density (penetration threshold) multiple magnetic islands form near the top of the pedestal giving rise to continuous q_{95} windows of ELM suppression. The model predicts that wide q_{95} windows of ELM suppression can be achieved at substantially higher pedestal pressure in DIII-D by shifting to higher toroidal mode number (n=4) RMPs.

The description of rocks is one of the most time-consuming tasks in the everyday work of a geologist, especially when very accurate description is required. We here present a method that reduces the time needed for accurate description of rocks, enabling the geologist to work more efficiently. We describe the application of methods based on color distribution analysis and feature extraction. Then we focus on a new approach, used by us, which is based on convolutional neural networks. We used several well-known neural network architectures (AlexNet, VGG, GoogLeNet, ResNet) and made a comparison of their performance. The precision of the algorithms is up to 95% on the validation set with GoogLeNet architecture. The best of the proposed algorithms can describe 50 m of full-size core in one minute., Comment: 25 pages, 9 figures, 3 tables, submitted to Computers and Geosciences Journal. Keywords: Core Image; Description; Convolutional Neural Networks; Representation; Geology; Lithotypes

The presence of error fields in a tokamak device is inevitable, so it is of fundamental importance to understand their impact on the operation of the device and up to what level they can be tolerated. In this paper a prediction of the impact on the magnetic footprints on the divertor plates due to the misalignment of NSTX-U equilibrium coils is presented. Resistive MHD simulations are used to predict the magnetic field perturbations experienced by the field lines due to the presence of error fields. A linear relation between the magnitude of the misalignment and the area of the magnetic footprints is found, as well as an inverse proportionality between the size of a magnetic footprint and the plasma penetration of the field lines composing the footprint. This study is intended as a starting point in the process of investigating divertor footprints associated with the plasma response to intrinsic error fields with resistive MHD codes. A basic interpretation of the results suggests that large footprints resulting from equilibrium coils misalignment may be desirable, to some extent, to decrease the peak heat load on the divertor components. A more complete interpretation of the results presented here will involve a number of complex physics issues, for example pedestal transport and stability, along with the role of various scrape-off layer and divertor processes, as well as experimental data to confirm the predictions.

A method for determining relative permeabilities from data on unsteady filtration is proposed based on analytical solution of the inverse problem of filtration in the Buckley-Leverett approximation (Johnson-Bossler-Naumann method), verification of the results by numerical solution of the direct problem, and developing recommendations for changing the filtration conditions in subsequent experiments. It is shown that to improve the agreement between natural phase permeabilities and those obtained under laboratory conditions, it is advisable to combine physical and numerical experiments.

This study concerns the effect of the zone of a plane surface energy deposition on a gas-dynamic flow with a shock wave of M = 2.3–2.7 in a channel with a rectangular cross section. The source of the pulsed energy is a distributed sliding nanosecond discharge that develops in an approximately 1-mm-thick layer on a surface of 100 × 30 mm2. The results of a 3D numerical simulation of the problem on the basis of the Navier-Stokes equations for a compressed gas are presented and compared with the experimental shadow images.

In this work we evaluated the ITER ELM coils design based on two metrics: the Chirikov vacuum magnetic island overlap parameter, and the vacuum Field Line Loss Fraction. The study was performed for a range of current amplitudes for three different n = 4 waveforms: square, cosine and sine. The results indicated that ITER ELM coils are designed with a high level of flexibility to accommodate different operation scenarios (H-mode and Steady State) with different values of q95 and q-profiles. The magnetic island overlap analysis showed that ITER ELM coils are capable of matching the DIII-D I-coil spectrum. The Field Line Loss analysis showed that edge vacuum stochastization might be achieved that is similar or greater than in DIII-D. Fault analysis of the coils indicated that ITER ELM coils are robust and show good characteristics even with 11% of dead coils.

As shown on DIII-D edge localized modes (ELMs) can be either completely eliminated or mitigated with resonant magnetic perturbation (RMP) fields. Two infrared cameras, separated 105° toroidally, were used to make simultaneous measurements of ELM heat loads with high frame rates. Without the RMP fields ELMs display a variety of different heat load dynamics and a range of toroidal variability that is characteristic of their 3D structure. Comparing radial averages there is no asymmetry between two toroidal locations. With RMP-mitigated ELMs, the variability in the radially averaged power loads is significantly reduced and toroidal asymmetries in power loads are introduced. In addition to RMP ELM suppression scenarios an RMP scenario with only very small ELMs and very good confinement has been achieved.

Non-stationary plane shock wave interaction with localized nanosecond-lasting surface discharges was investigated. Pulse discharge plasma glow evolution was recorded with a CCD camera and gas flow evolution was recorded with laser shadowgraphy. CFD simulations of pulse energy deposition near the horizontal wall surface in front of the shock wave were carried out. Non-stationary 2D symmetrical flow dynamics were studied and analysis of the instantaneous surface discharge energy rate through CFD and shadow images matching was carried out.

A way of effectively affecting the gasdynamic structures of a transonic flow over a surface by means of instantaneous local directed energy deposition into a near-surface layer is proposed. Experimental investigations into the influence of a pulsed high-current nanosecond surface discharge of the “plasma sheet” type on gas fast flow with a shock wave near the surface are carried out. The self-localization of energy deposition into a low-pressure region in front of the shock wave is described. Based on this effect, a facility for automated energy deposition into a dynamic region bounded by the moving shock front can be designed. The limiting value of the specific energy deposition on the surface in front of the shock wave is found. With the help of the direct-shadow method, an unsteady quasi-two-dimensional discontinuous flow arising when a plasma sheet is initiated on the wall in a flow with a plane shock wave is studied. By numerically solving the two-dimensional nonstationary equations of gas dynamics, the influence of the energy of a pulsed nanosecond discharge, which is applied in the frequency regime, on the aerodynamic characteristics of a high-lift profile is investigated. It is ascertained that the energy delivered to the gas before the closing shock wave in a local supersonic region that is located in the neighborhood of the profile contour in zones extended along the profile considerably decreases the wave drag of the profile.

Atmospheric-pressure dielectric barrier discharges for aerodynamic flow control are investigated. The properties of the plasma discharge during positive and negative half-cycles of the driving voltage are studied experimentally and numerically. The simulations are performed in nitrogen. It is shown that during the positive half-cycle the plasma operates in the streamer mode, whereas the negative half-cycle is characterized by a glow-type discharge. The computational results are compared with experimental optical measurements and found to be in good agreement.

We present the results of a series of measurements in which an array of capacitive V-dot probes are used to determine the surface potential in a single dielectric barrier discharge plasma operated as an aerodynamic plasma actuator. From these measurements, we determine the longitudinal electric field in the plasma. The results show that the surface immediately (within the first alternating current cycle of the plasma discharge) acquires a net positive surface charge, so that the electric field in the plasma is asymmetric, the magnitude of the maximum field being more than 50% larger in the negative than the positive polarity. The measurements also show that there is a region several millimeters downstream from the exposed electrode edge over which the electric field maintains a constant positive (downstream) polarity over the course of the discharge cycle.

Recent experiments on DIII-D have increased confidence in the ability to suppress edge-localized modes (ELMs) using edge-resonant magnetic perturbations (RMPs) in ITER, including an improved physics basis for the edge response to RMPs as well as expansion of RMP ELM suppression to more ITER-like conditions. Complete ELM suppression has been achieved utilizing n = 3 RMPs in the ITER baseline scenario. In addition, RMP ELM suppression has been expanded to include plasmas with helium concentrations near 25% and the use of n = 2 RMPs. Analysis of the kinetic profile response suggests that ELM suppression is correlated with the co-alignment of the ω⊥e = 0 location, an n = 3 rational surface, and the top of the pedestal. Modelling predicts that such a co-alignment could potentially lead to island (or island chain) formation just inside the top of the pedestal, inhibiting the growth of the pedestal and thereby maintaining the ELM-free state. Detailed analysis of data obtained during toroidal phase variations of the applied n = 3 RMPs have provided further evidence of an island-like structure at the top of the pedestal. In addition, nearly matched discharges with co-neutral-beam-injection (co-NBI) and counter-NBI have demonstrated the importance of the presence of the ω⊥e = 0 location for ELM suppression. In the counter-NBI cases, the toroidal rotation profile is such that there is no ω⊥e = 0 location and ELMs are not suppressed in conditions in which ELM suppression is generally observed with co-NBI.

Progress in the definition of the requirements for edge localized mode (ELM) control and the application of ELM control methods both for high fusion performance DT operation and non-active low-current operation in ITER is described. Evaluation of the power fluxes for low plasma current H-modes in ITER shows that uncontrolled ELMs will not lead to damage to the tungsten (W) divertor target, unlike for high-current H-modes in which divertor damage by uncontrolled ELMs is expected. Despite the lack of divertor damage at lower currents, ELM control is found to be required in ITER under these conditions to prevent an excessive contamination of the plasma by W, which could eventually lead to an increased disruptivity. Modelling with the non-linear MHD code JOREK of the physics processes determining the flow of energy from the confined plasma onto the plasma-facing components during ELMs at the ITER scale shows that the relative contribution of conductive and convective losses is intrinsically linked to the magnitude of the ELM energy loss. Modelling of the triggering of ELMs by pellet injection for DIII-D and ITER has identified the minimum pellet size required to trigger ELMs and, from this, the required fuel throughput for the application of this technique to ITER is evaluated and shown to be compatible with the installed fuelling and tritium re-processing capabilities in ITER. The evaluation of the capabilities of the ELM control coil system in ITER for ELM suppression is carried out (in the vacuum approximation) and found to have a factor of ∼2 margin in terms of coil current to achieve its design criterion, although such a margin could be substantially reduced when plasma shielding effects are taken into account. The consequences for the spatial distribution of the power fluxes at the divertor of ELM control by three-dimensional (3D) fields are evaluated and found to lead to substantial toroidal asymmetries in zones of the divertor target away from the separatrix. Therefore, specifications for the rotation of the 3D perturbation applied for ELM control in order to avoid excessive localized erosion of the ITER divertor target are derived. It is shown that a rotation frequency in excess of 1 Hz for the whole toroidally asymmetric divertor power flux pattern is required (corresponding to n Hz frequency in the variation of currents in the coils, where n is the toroidal symmetry of the perturbation applied) in order to avoid unacceptable thermal cycling of the divertor target for the highest power fluxes and worst toroidal power flux asymmetries expected. The possible use of the in-vessel vertical stability coils for ELM control as a back-up to the main ELM control systems in ITER is described and the feasibility of its application to control ELMs in low plasma current H-modes, foreseen for initial ITER operation, is evaluated and found to be viable for plasma currents up to 5–10 MA depending on modelling assumptions.

This paper explores the phenomena associated with pulsed discharge energy deposition in the near-surface gas layer in front of a shock wave from the flow control perspective. The energy is deposited in 200 ns by a high-current distributed sliding discharge of a 'plasma sheet' type. The discharge, covering an area of mm2, is mounted on the top or bottom wall of a shock tube channel. In order to analyse the time scales of the pulsed discharge effect on an unsteady supersonic flow, we consider the propagation of a planar shock wave along the discharge surface area 50–500 μs after the discharge pulse. The processes in the discharge chamber are visualized experimentally using the shadowgraph method and modelled numerically using 2D/3D CFD simulations. The interaction between the planar shock wave and the discharge-induced thermal layer results in the formation of a lambda-shock configuration and the generation of vorticity in the flow behind the shock front. We determine the amount and spatial distribution of the electric energy rapidly transforming into heat by comparing the calculated flow patterns and the experimental shadow images. It is shown that the uniformity of the discharge energy distribution strongly affects the resulting flow dynamics. Regions of turbulent mixing in the near-surface gas are detected when the discharge energy is deposited non-uniformly along the plasma sheet. They account for the increase in the cooling rate of the discharge-induced thermal layer and significantly influence its interaction with an incident shock wave.

Shock waves have traditionally been studied and characterized by their movement through media. And alternatively, the parameters of a medium have been successfully diagnosed by its influence on a shock-wavesmotion. The propagation of shock waves through media with various types of inhomogeneity has always been of particular interest for a wide range of application in different fields of physics, like astrophysics, laser physics, aerodynamics and many others. There are lots of experimental, theoretical and numerical studies concerning interactions of shock waves with turbulence [1], vortices [2], near-wall fine particle or dust layers [3], preshock inhomogeneities induced by non-uniform energy deposition [4], isolated bubbles [5], thermal layers [6, 7], plasma formations [8, 9], etc. These interactions are apparently mutual - not only are the dynamics and the structure of waves altered but also the properties of the medium are changed in a post shock flow. The present work studies the structure of the gas-dynamic flow resulting from the propagation of a plain shock wave along a near-wall expanding layer of the gas excited by a high-current sliding surface discharge (“plasma sheet”).

Discontinuity breakdown conditions were realized experimentally by instant homogenous energy input in front of shock wave. Energy deposition was conducted using nanosecond gas discharges in spatial and surface area. A shock tube and special type of nanosecond electric discharge with plasma electrodes were used for this research. One and two-dimensional numerical simulation under experimental conditions has been undertaken. Pressure, density, temperature and velocity fields have been examined. Comparison of numerical data and schleiren images of shock wave after its interaction with discharge area was conducted. The configuration of disturbed shock wave was found to be in a good correspondence with one from numerical calculations. Numerical results showed possibility to achieve Richtmyer-Meshkov type instability without membrane.