Hasil untuk "Petroleum refining. Petroleum products"

Yujie Zheng, Zhuo Li, Shengtao Zhang et al.

Deploying Large Language Models to data-scarce programming domains poses significant challenges, particularly for kernel synthesis on emerging Domain-Specific Architectures where a "Data Wall" limits available training data. While models excel on data-rich platforms like CUDA, they suffer catastrophic performance drops on data-scarce ecosystems such as NPU programming. To overcome this cold-start barrier without expensive fine-tuning, we introduce EvoKernel, a self-evolving agentic framework that automates the lifecycle of kernel synthesis from initial drafting to continual refining. EvoKernel addresses this by formulating the synthesis process as a memory-based reinforcement learning task. Through a novel value-driven retrieval mechanism, it learns stage-specific Q-values that prioritize experiences based on their contribution to the current objective, whether bootstrapping a feasible draft or iteratively refining latency. Furthermore, by enabling cross-task memory sharing, the agent generalizes insights from simple to complex operators. By building an NPU variant of KernelBench and evaluating on it, EvoKernel improves frontier models' correctness from 11.0% to 83.0% and achieves a median speedup of 3.60x over initial drafts through iterative refinement. This demonstrates that value-guided experience accumulation allows general-purpose models to master the kernel synthesis task on niche hardware ecosystems. Our official page is available at https://evokernel.zhuo.li.

Yue Li Du, Ben Alexander, Mikhail Antonenka et al.

Retrieving semantically similar but visually distinct contents has been a critical capability in visual search systems. In this work, we aim to tackle this problem with Visual Product Graph (VPG), leveraging high-performance infrastructure for storage and state-of-the-art computer vision models for image understanding. VPG is built to be an online real-time retrieval system that enables navigation from individual products to composite scenes containing those products, along with complementary recommendations. Our system not only offers contextual insights by showcasing how products can be styled in a context, but also provides recommendations for complementary products drawn from these inspirations. We discuss the essential components for building the Visual Product Graph, along with the core computer vision model improvements across object detection, foundational visual embeddings, and other visual signals. Our system achieves a 78.8% extremely similar@1 in end-to-end human relevance evaluations, and a 6% module engagement rate. The "Ways to Style It" module, powered by the Visual Product Graph technology, is deployed in production at Pinterest.

Chenjie Li, Amir Gilad, Boris Glavic et al.

Programmatic weak supervision (PWS) significantly reduces human effort for labeling data by combining the outputs of user-provided labeling functions (LFs) on unlabeled datapoints. However, the quality of the generated labels depends directly on the accuracy of the LFs. In this work, we study the problem of fixing LFs based on a small set of labeled examples. Towards this goal, we develop novel techniques for repairing a set of LFs by minimally changing their results on the labeled examples such that the fixed LFs ensure that (i) there is sufficient evidence for the correct label of each labeled datapoint and (ii) the accuracy of each repaired LF is sufficiently high. We model LFs as conditional rules which enables us to refine them, i.e., to selectively change their output for some inputs. We demonstrate experimentally that our system improves the quality of LFs based on surprisingly small sets of labeled datapoints.

Ekansh Jauhari

The $n$-th symmetric product of a topological space $X$ is the orbit space of the natural action of the symmetric group $S_n$ on the product space $X^n$. In this paper, we compute the sequential topological complexities of (finite products of) the symmetric products of closed orientable surfaces, thereby verifying the rationality conjecture of Farber and Oprea for these spaces. Additionally, we determine the Lusternik-Schnirelmann category of (finite products of) the symmetric products of closed non-orientable surfaces. More generally, we provide lower bounds to the LS-category and the sequential topological complexities of the symmetric products of finite CW complexes $X$ in terms of the cohomology of $X$ and its products. On the way, we also obtain new lower bounds to the sequential distributional complexities of continuous maps and study the homotopy groups of the symmetric products of closed surfaces.

Lewis Stanton, Fedor Vylegzhanin

We develop a method for studying the pointed loop space of general polyhedral products, showing that many properties are determined by the moment-angle complex. To apply the method, we show that localised away from a finite set of primes, the loop space of a moment-angle complex is homotopy equivalent to a product of loops on spheres. As a consequence, we give p-local loop space decompositions of quasitoric manifolds, certain toric orbifolds and a wide family of polyhedral products. This verifies a conjecture of Anick for such spaces. We also describe the additive structure of loop homology of simply connected polyhedral products in terms of polynomials studied by Backelin and Berglund.

Zhang Jizhe, Qiao Lei, Zhang Yuhao et al.

The solutions for normal operation of logging tools at high temperature of 175 ℃ and above focus more on insulation.Vacuum flasks and multilayer insulation materials are usually used to isolate the entry of external heat.Scholars design vacuum flasks mainly based on theoretical formulas and empirical coefficients,and there is rare study on the temperature field of vacuum flasks by numerical simulation.In this paper,based on the structure of the magnetic probe and the parameters of the vacuum flask,together with temperature field coupling,the variation of the temperature field of magnetic steering tool over time were analyzed.The results show that the probe sensor needs to work for 8-10 hours underground at around 175 ℃,where the lengths of the upper and lower heat insulators should be 150 mm and 50 mm respectively,and the lengths of the upper and lower heat carriers should be 150 mm simultaneously.The increase of the heat carrier volume can extend the maximum working time,and the heat carrier volume should be reasonably determined by considering the factors such as working time,suitable working temperature of the sensor and external dimension of the insulation protection cabin.The research conclusions provide a theoretical basis for the design of insulation protection cabin for magnetic steering tool.

Gwanhyeong Koo, Sunjae Yoon, Ji Woo Hong et al.

Current image editing methods primarily utilize DDIM Inversion, employing a two-branch diffusion approach to preserve the attributes and layout of the original image. However, these methods encounter challenges with non-rigid edits, which involve altering the image's layout or structure. Our comprehensive analysis reveals that the high-frequency components of DDIM latent, crucial for retaining the original image's key features and layout, significantly contribute to these limitations. Addressing this, we introduce FlexiEdit, which enhances fidelity to input text prompts by refining DDIM latent, by reducing high-frequency components in targeted editing areas. FlexiEdit comprises two key components: (1) Latent Refinement, which modifies DDIM latent to better accommodate layout adjustments, and (2) Edit Fidelity Enhancement via Re-inversion, aimed at ensuring the edits more accurately reflect the input text prompts. Our approach represents notable progress in image editing, particularly in performing complex non-rigid edits, showcasing its enhanced capability through comparative experiments.

Shihan Kanungo, Jordan Schettler

In this note, we evaluate a multivariable family of infinite products which generalize Guillera's infinite product for $e$, and Ser's formula (rediscovered by Sondow) for $e^γ$. We describe formulas for the products in terms of special values of the Hurwitz zeta function $ζ(s,u)$ and its $s$ derivative. Additionally, we derive integral and double integral representations for the logarithms of these infinite products.

Ashish Kumar LOOMBA, Vinicius Eduardo BOTECHIA, Denis José SCHIOZER

We present an efficient and risk-informed closed-loop field development (CLFD) workflow for recurrently revising the field development plan (FDP) using the accrued information. To make the process practical, we integrated multiple concepts of machine learning, an intelligent selection process to discard the worst FDP options and a growing set of representative reservoir models. These concepts were combined and used with a cluster-based learning and evolution optimizer to efficiently explore the search space of decision variables. Unlike previous studies, we also added the execution time of the CLFD workflow and worked with more realistic timelines to confirm the utility of a CLFD workflow. To appreciate the importance of data assimilation and new well-logs in a CLFD workflow, we carried out researches at rigorous conditions without a reduction in uncertainty attributes. The proposed CLFD workflow was implemented on a benchmark analogous to a giant field with extensively time-consuming simulation models. The results underscore that an ensemble with as few as 100 scenarios was sufficient to gauge the geological uncertainty, despite working with a giant field with highly heterogeneous characteristics. It is demonstrated that the CLFD workflow can improve the efficiency by over 85% compared to the previously validated workflow. Finally, we present some acute insights and problems related to data assimilation for the practical application of a CLFD workflow.

ZHAO Yanting, SHEN Jian, ZHAO Sumin et al.

The Minghuazhen Formation in the Cangxian Uplift of Tianjin presents a unique challenge for geothermal resource development due to its sandstone reservoirs. These reservoirs are characterized by high shale content, suboptimal lithology, loosely cemented structures, and a tendency for sand flow, all of which complicate the process of reinjecting water back into the reservoir. This difficulty in reinjection poses a significant obstacle to the sustainable development of geothermal resources in the area. The existing well completion technology, which involves filter pipe gravel filling, is beneficial in maintaining the stability of the formation and facilitating water flow. However, improvements are needed in several aspects, including drilling diameter, drilling fluid composition, drilling assembly, and well washing processes, to enhance the effectiveness of reinjection. An optimized large-diameter gravel filling well completion technology has been developed to address these issues. This technology not only preserves the original sand retention and support functions but also increases the well completion diameter, thereby enlarging the flow area of the geothermal wells. Further, by optimizing the drilling fluid formula and drilling assembly, the contamination of the reservoir is minimized. Additionally, a combined well washing technique implemented after drilling operations helps remove plugging components from the formation, aiding in unclogging the permeation channels. Three groups of reinjection tests were carried out on two reinjection wells in Minghuazhen Formation, which were constructed by the new well completion technology. The maximum reinjection rates in the tests were 32.0 m³/h and 58.0 m³/h. Compared with the reinjection wells in the same layer, the reinjection volume increased by 3.48 times and more than 2.00 times respectively, which proved that the optimized gravel filling well completion technology of large-diameter filter pipe improved the reinjection effect effectively.

SHI Leiting, ZHAO Qiming, REN Zhenyu et al.

The fracture network developed in coal rock serves as the primary channel for gas migration, significantly influencing the seepage capacity of coal reservoir. The geometric characteristics of fracture plays a crucial role on determining the flow characteristics of coal-bed methane. To study this, a two-dimensional fracture network model of coal rock was established using COMSOL Multiphysics simulation software, focusing on the coal samples of Baode block as the research subject. The effects of fracture length, density, opening degree and angle on production were investigated, providing valuable theoretical guidance for enhancing coal-bed methane production. The results indicate that fracture length, density, and opening degree have a positive correlation with the seepage capacity of coal rock, while the angle with the flow direction negatively impacts it. However, with the increase of length, density and opening degree, the improvement in flow rate slows down, and the effect of increasing single factor to improve coal-bed methane mining can be neglected, making it difficult to control the cost-benefit ratio. Among the factors influencing outlet, angle and density exert a more significant effect than length and opening degree. Considering the surface directional well plus the high pressure hydraulic cutting method, we can enhance the efficiency of coalbed methane development. This approach connects the natural fracture system using directional borehole and hydraulic slot, fully utilizing the permeability advantage of parallel surface cutting direction. The high-pressure hydraulic cutting process induces cracks in the coal seam, increasing the number and connectivity of diversion channels, thereby bolstering the production of coal-bed methane.

Zhang Zhiguang, Pan Lingyong, Wang Junqiao et al.

The rolling rotor oil-gas mixed-flow pumps are rapidly gaining popularity in the natural gas extraction industry, and the experimental research on these pumps help to establish a foundation for theoretical analysis and forward design of pump heads, which is of great value to the research and development of high-performance pump heads and the promotion of natural gas development.By constructing an internal circulation experimental platform with wide working conditions, high stability and high precision, testing experiments on the operating temperature, pressure, displacement and power consumption of a certain type of rolling rotor pump were performed, and the working characteristics of the pump were analyzed.In addition, the working mechanism of the pump head was comprehensively established, proving the superior performance of this type of pump head.The research findings indicate that, in the context of oil-gas mixed-flow, the gas compression process within the pump's cylinder tends towards isothermal compression, and the temperature rise within the working chamber remains limited, maintaining the exhaust temperature at approximately 55 ℃.Although the rolling rotor pumps are used as compressors in the natural gas development field, they primarily exhibit pump-like working characteristics: the actual pump head displacement is directly proportional to the rotational speed and intake pressure, and linearly inversely proportional to the intake-exhaust pressure difference; the pump head volumetric efficiency is linearly inversely proportional to the intake-exhaust pressure difference; the pump head power consumption is linearly proportional to the intake-exhaust pressure difference, but unrelated to the intake pressure, furthermore, the magnitude of power consumption increases continuously with higher rotational speeds.Notably, the pressure pulsation at the rolling rotor pump's outlet is significantly lower compared to reciprocating natural gas compressors, and the exhaust pressure pulsation at the booster skid is almost eliminated, highlighting the substantial advantages of this equipment.These research results may provide valuable insights for the design and application of oil-gas mixed-flow pumps.

Mohammad Reza Mansouri Daneshvar, Parviz Mansouri-Daneshvar, Reza Moussavi-Harami et al.

Abstract Despite the previous interpretations of the evolutionary scenarios of the Qatar Arch, the present study aimed to present a new complementary scenario to lay out the probable effects of another basement fault, corresponding to the physical shape of the Arch. A remote sensing study was carried out based on geological and geophysical data interpretations, obtained from some national and international databases, including GIS-based global geology data of Harvard University, the free-air gravity anomaly data from the World Gravity Map (WGM), and the airborne geophysical project of the National Geosciences Database of Iran (NGDIR). Results revealed that offshore extrapolation of the Trans Arabian-Bostaneh Fault (TABF) lineament through the morphological boundary of Hauterivian–Barremian (HB) formations and the offshore extrapolation of the Kazerun fault could produce a new complementary model for the Qatar Arch evolution. Based on the particular shape of the Arch (i.e., narrowing and uplifting to the south influenced by the Kazerun basement fault and the rotation of TABF), we can propose that these faults crossed each other after the Barremian age, created a particular shape of the Qatar Arch, and influenced beneath the Fars hinterland. The obtained geophysical gravity data also showed a good match between structural units and faults with maximum values corresponding to the Qatar Arch and minimum values relating to the thick sediment field in the Fars region. In conclusion, the present study contributes to a new proposition about exploring the possible faults and a new gas field situated in the interior territory of Iran. All of these interpretations will provide a new insight to improve the extant knowledge in the geology and petroleum evolution of the study area, such as the Plate tectonic events and the creation of possible reservoirs in future.

Kateřina Macková, Martin Pilát

The goal of product mapping is to decide, whether two listings from two different e-shops describe the same products. Existing datasets of matching and non-matching pairs of products, however, often suffer from incomplete product information or contain only very distant non-matching products. Therefore, while predictive models trained on these datasets achieve good results on them, in practice, they are unusable as they cannot distinguish very similar but non-matching pairs of products. This paper introduces two new datasets for product mapping: ProMapCz consisting of 1,495 Czech product pairs and ProMapEn consisting of 1,555 English product pairs of matching and non-matching products manually scraped from two pairs of e-shops. The datasets contain both images and textual descriptions of the products, including their specifications, making them one of the most complete datasets for product mapping. Additionally, the non-matching products were selected in two phases, creating two types of non-matches -- close non-matches and medium non-matches. Even the medium non-matches are pairs of products that are much more similar than non-matches in other datasets -- for example, they still need to have the same brand and similar name and price. After simple data preprocessing, several machine learning algorithms were trained on these and two the other datasets to demonstrate the complexity and completeness of ProMap datasets. ProMap datasets are presented as a golden standard for further research of product mapping filling the gaps in existing ones.

Yousef Ibrahem, V.P. Morozov, V. Sudakov et al.

The Upper Devonian Semiluksk (Domanik) and Mendymsk carbonate successions have been identified as the most promising unconventional oil and gas deposits in Russia's Volga-Ural petroleum province. Three lithofacies were identified: limestones, organic matter-rich siliceous carbonate, and dolomitic limestones/dolomite. Their sediments are thought to have accumulated in the relatively deep-sea shelf depositional environment. According to X-ray analysis, calcite is the most common mineral phase in the limestones. The siliceous carbonate rocks' major mineral phases were identified to be quartz, followed by calcite. The rocks have relatively poor reservoir qualities, with porosity ranging from 0.1 to 5% and permeability ranging from 0.01 to 10 mD. Petrographic examination of thin sections reveals four major types of pores: 1. interparticle pores, which are abundant in all of the examined lithofacies; 2. intercrystalline pores formed between dolomite crystals and the very fine components of the siliceous carbonate rocks. 3. vugs as a result of leaching, and 4. fracture as a result of tectonic and/or diagenetic events. Plotting the obtained data from the petrographically examined thin sections on the ternary pore plot diagram resulted in the identification of six pore facies and their related diagenetic patterns. Pore facies 1 depicts the pores that formed during the depositional process. Pore facies 4 displays moldic pores. The vugs are represented by pore facies 6. Pore facies 2 is a mixture between pore facies 1 and pore facies 4. Pore facies 3 is a mixture between pore facies 1 and pore facies 6. Pore facies 5 is a mixture between pore facies 4 and pore facies 6. Comparing the petrographic results and the identified pore facies show that pore facies 6 have touching pores and they played the most important role in improving the reservoir quality. The pore facies 4 which have separate pores have the most influence in the reduction of the reservoir quality. The diagenetic origin secondary pores are the most pores identified in the studied carbonate section. This reflects the strong modification of the pores by the diagenetic overprinting. The massive dissolution that forms the touching vugs, and the dolomitization process are the most diagenetic processes that affect the pore system within the studied rocks.

Lianhua Hou, Fan Yang, Chun Yang et al.

The Gaoshiti-Moxi Gas Field has been discovered recently in the central part of Sichuan Basin, which might be the biggest gas field in China. In this gas field, a large amount of gas is from Neoproterozoic Sinian (Ediacaran) karstic carbonate rocks. The gas was originated from cracking of oil, and the oil source rocks are the Sinian and Cambrian strata. In this study, the characteristics of these carbonate reservoir rocks have been studied based on subsurface data and outcrop observation. The carbonate rocks are at the Sinian Dengying Formation, and the reservoir properties are controlled by sedimentary facies and diagenesis. The Dengying Formation is composed of algal mounds and grainstone build-ups, which were developed in a restricted platform setting from glacial period to interglacial period. Due to tectonic movement and eustatic cycle in Sichuan Basin during Neoproterozoic Era, five sedimentary cycles of transgression and regression have been developed in the Dengying Formation. According to the sedimentary facies and carbon-oxygen isotope variation, five sedimentary cycles have been recognized. Due to the penecontemporaneous dolomitization, together with hypergene karstification and late burial dissolution, the reservoir properties of the carbonates rocks have been improved greatly.

Zengwei Yao, Wenjie Pei, Fanglin Chen et al.

The crux of single-channel speech separation is how to encode the mixture of signals into such a latent embedding space that the signals from different speakers can be precisely separated. Existing methods for speech separation either transform the speech signals into frequency domain to perform separation or seek to learn a separable embedding space by constructing a latent domain based on convolutional filters. While the latter type of methods learning an embedding space achieves substantial improvement for speech separation, we argue that the embedding space defined by only one latent domain does not suffice to provide a thoroughly separable encoding space for speech separation. In this paper, we propose the Stepwise-Refining Speech Separation Network (SRSSN), which follows a coarse-to-fine separation framework. It first learns a 1-order latent domain to define an encoding space and thereby performs a rough separation in the coarse phase. Then the proposed SRSSN learns a new latent domain along each basis function of the existing latent domain to obtain a high-order latent domain in the refining phase, which enables our model to perform a refining separation to achieve a more precise speech separation. We demonstrate the effectiveness of our SRSSN by conducting extensive experiments, including speech separation in a clean (noise-free) setting on WSJ0-2/3mix datasets as well as in noisy/reverberant settings on WHAM!/WHAMR! datasets. Furthermore, we also perform experiments of speech recognition on separated speech signals by our model to evaluate the performance of speech separation indirectly.

Mingwen WANG, Gang LUO, Yunqiang SUN et al.

We simulated the stress changes around a salt basin using a static salt structure model under compressive stress background to investigate the stress perturbation caused by different salt body shapes and amplitudes. We designed a two-layer salt model with three bulges and sags using finite element methods to calculate the stress perturbation around the salt. The results show that salt shape is closely related to the stress perturbation in the sediments around the salt, and the fluctuations of the bulge and sag (smooth or steep) can also affect the stress perturbation magnitude. Extrusion horizontal stress, normal stress, and out-of-plane stress on the plane would occur near the salt uplift in a compressive tectonic stress environment. In contrast, tensile horizontal stress, out-of-plane stress, and vertical stress would occur near the salt sag. In addition, smoother bulges are associated with smaller produced stress perturbations, and steeper sags are associated with a greater reduction of stress perturbation in the sediment. The stress of a salt structure in western Kelasu of the Kuqa depression was simulated and the applicability of previous conclusions regarding this structure was verified. These conclusions provide scientific basis for the prediction of stress perturbations around salt basin systems. Key words: numerical simulation, salt structure, stress perturbation, Kuqa depression, oil drilling

Yongliang Gao, Zhigang Wen, Yaohui Xu et al.

Abstract In order to study the geochemical characteristics of fine-grained sedimentary rocks and clarify its relationship with the tight oil reservoir, the Chang7 Member of Triassic Yanchang Formation in Longdong area was taken as an example to be studied by rock pyrolysis, isotopic determination, maceral analysis, gas chromatography and gas chromatography–mass spectrometry. Lacustrine organic-rich fine-grained sedimentary rocks can be divided into organic-rich shale and organic-rich mudstone according to their lithology and texture. The results show that the organic-rich shale was deposited in a quiet and anoxic deep lacustrine environment, the organic matter was of high abundance and mainly sourced from plankton and benthic algae (type I–II1). Organic-rich mudstone mainly deposited in the semi-deep lacustrine environment, where the water was relatively shallow and the salinity was low, the organic matter was of low abundance and came from the mixed source of plankton and terrestrial clastic (type II1–II2). The Chang7 tight oil is widely distributed in the turbidite sand bodies, and it was mainly sourced from the Chang7 organic-rich shale.

Yu Qiannan, Liu Yikun, Yao Di et al.

Experiments for seepage flow patterns in thin and poor reservoirs could be carried out by the plate models made by artificial core designed based on the similarity theory. Pressure gradient distribution in fractured thin and poor reservoir by fine controlling was revealed based on the pressure data, and the sketch maps of seepage flow sections were drew to study on seepage flow patterns of fine controlled fractured thin and poor reservoir qualitatively and quantitatively. Experiment results showed that seepage flow patterns in thin and poor reservoirs were greatly influenced by permeability, heterogeneity and fine controlled fracturing cracks while the thin and poor reservoirs had low permeability and strong homogeneity. Fine controlled fracturing could reduce the negative influence of low permeability and strong heterogeneity. Thin and poor reservoirs could be divided into the non seepage flow section, the nonlinear seepage flow section and the quasi linear seepage flow section. The non seepage flow section was reduced by over 72 % after fine controlled fracturing, the area for fluid flowing through increased correspondingly, and the proportion of quasi linear percolation area which was more conducive to fluid flow raised by at least 86 %.