Fines migration during subterranean CO2 storage is considered as one of major contributors to formation damage and injectivity decline. The main formation damage mechanism by migrating fines is rock clogging by straining of particles in thin pore throats. The main mobilisation mechanism of attached natural reservoir fines is their detachment by capillary forces exerting from the advancing water-gas menisci. The mathematical model for displacement of water by injected CO2 from layer-cake reservoirs under fines mobilisation, migration and straining extends the classical vertical capillary-gravity equilibrium model, where hydrostatic pressure gradient is assumed in each phase. The analytical model generalises classical Buckley-Leverett problem, where the extended fractional flow accounts for permeability damage due to fines migration. We show the significant effect of fines migration on water displacement - while it decreases well injectivity, fines migration simultaneously increases reservoir sweep, increasing CO2 storage capacity.
One of the key risks for a Carbon Capture Storage (CCS) is injectivity decline. Evaporation of the connate brine in near-wellbore region during CO2 injection may result in drying-up the rock yielding the mobilisation and migration of clay particles leading to decline rock permeability and consequent loss of well injectivity. Influx of the reservoir brine into the dried-up zone yields accumulation of precipitated salt and injectivity decline. This paper presents the results of eight coreflooding experiments aiming investigation of the effect of rock dry-out, fines migration, and salt precipitation during CO2 injection. Pressure drops across the cores, brine saturation and produced clay fines concentration versus Pore Volume Injected (PVI) have been measured. All lab tests exhibit the following features: intensive fines production at the very beginning of gas-water production period following reduced-rate fines production during overall evaporation period and continuous fines disappearance at the late stage; abrupt increase in gas permeability in the middle of evaporation, and non-monotonic evaporation rate and pressure drop. To explain these phenomena, we distinguished three sequent
Numerical simulations of geological CO2 storage in deep saline aquifers have demonstrated that vertical equilibrium (VE) models offer a robust and computationally efficient framework for reservoir optimization and upscaling. These studies emphasize the influence of fines migration and partial miscibility between water and CO2 on the evolving storage behaviour. Specifically, capillary-driven mobilization of clay and silica fines can impair injectivity, while water evaporation into the CO2 phase leads to near-wellbore drying, salt precipitation, and additional injectivity loss. However, conventional VE models do not account for these coupled physical and geochemical mechanisms. This work introduces an extended VE model that incorporates water vaporization into CO2, CO2 dissolution into the displaced brine, and fines migration leading to permeability reduction. For layer-cake aquifers, the model admits an exact analytical solution, providing closed-form expressions for both injectivity decline and sweep efficiency. The analysis identifies two distinct fronts during displacement: a leading displacement-dissolution front and a trailing full-evaporation front. Vertical model downscaling
Suspension-colloidal-nano transport in porous media encompasses the detachment of detrital fines against electrostatic attraction and authigenic fines by breakage, from the rock surface. While much is currently known about the underlying mechanisms governing detachment of detrital particles, including detachment criteria at the pore scale and its upscaling for the core scale, a critical gap exists due to absence of this knowledge for authigenic fines. Integrating 3D Timoshenkos beam theory of elastic cylinder deformation with CFD-based model for viscous flow around the attached particle and with strength failure criteria for particle-rock bond, we developed a novel theory for fines detachment by breakage at the pore scale. The breakage criterium derived includes analytical expressions for tensile and shear stress maxima along with two geometric diagrams which allow determining the breaking stress. This leads to an explicit formula for the breakage flow velocity. Its upscaling yields a mathematical model for fines detachment by breakage, expressed in the form of the maximum retained concentration of attached fines versus flow velocity -- maximum retention function (MRF) for breakage
In subterranean coal seam gas CSG reservoirs, massive amounts of small-sized coal fines are released during the production and development stages, especially during hydraulic fracturing stimulation. These coal fines inevitably cause mechanical pump failure and permeability damage due to aggregation and subsequent pore throat blockage. This aggregation behavior is thus of key importance in CSG production and needs to be minimized. Consequently, such coal fines dispersions need to be stabilized, which can be achieved by the formulation of improved fracturing fluids. Here, we thus systematically investigated the effectiveness of two additives; ethanol, 0.5 wt percent and SDBS, 0.001 and 0.01 wt percent, on dispersion stability for a wide range of conditions: pH 6 to 11, salinity of 0.1 to 0.6 M NaCl brine. Technically, the coal suspension flowed through a glass bead proppant pack, and fines retention was measured. We found that even trace amounts of sodium dodecylbenzene sulfonate SDBS i.e. 0.001 wt per cent drastically improved dispersion stability and reduced fines retention. The retention was further quantified by fractal dimensional analysis, which showed lower values for suspensi
The General Data Protection Regulation (GDPR) was enforced in 2018. After this enforcement, many fines have already been imposed by national data protection authorities in the European Union (EU). This paper examines the individual GDPR articles referenced in the enforcement decisions, as well as predicts the amount of enforcement fines with available meta-data and text mining features extracted from the enforcement decision documents. According to the results, articles related to the general principles, lawfulness, and information security have been the most frequently referenced ones. Although the amount of fines imposed vary across the articles referenced, these three particular articles do not stand out. Furthermore, good predictions are attainable even with simple machine learning techniques for regression analysis. Basic meta-data (such as the articles referenced and the country of origin) yields slightly better performance compared to the text mining features.
The determination of the mechanical properties of soils containing particles larger than the allowable size of standard laboratory equipments is complex. It is indeed necessary to remove the coarsest fraction to carry out the tests. This scalping poses a problem of reliability of the results at the scale of the structure. Parallel gradation is the method commonly used for estimating the shear strength of heterogeneous granular soils from tests on their finer fraction. However, the effect of high fines content on the estimation of shear strength by this method is not well understood. The results of this study showed that the parallel gradation method could predict the friction angle of the initial soil with high fines content when the modelled soil had a similar skeleton as the initial soil. However, the cohesion of the initial soil was overestimated.
The General Data Protection Regulation (GDPR) came into force in 2018. After this enforcement, many fines have already been imposed by national data protection authorities in Europe. This paper examines the individual GDPR articles referenced in the enforcement decisions, as well as predicts the amount of enforcement fines with available meta-data and text mining features extracted from the enforcement decision documents. According to the results, three articles related to the general principles, lawfulness, and information security have been the most frequently referenced ones. Although the amount of fines imposed vary across the articles referenced, these three particular articles do not stand out. Furthermore, a better statistical evidence is available with other meta-data features, including information about the particular European countries in which the enforcements were made. Accurate predictions are attainable even with simple machine learning techniques for regression analysis. Basic text mining features outperform the meta-data features in this regard. In addition to these results, the paper reflects the GDPR's enforcement against public administration obstacles in the Eu
We study the bulk properties of isotropic bidisperse granular mixtures using discrete element simulations. The focus is on the influence of the size (radius) ratio of the two constituents and volume fraction on the mixture properties. We show that the effective bulk modulus of a dense granular (base) assembly can be enhanced by up to 20% by substituting as little as 5% of its volume with smaller sized particles. Particles of similar sizes barely affect the macroscopic properties of the mixture. On the other extreme, when a huge number of fine particles are included, most of them lie in the voids of the base material, acting as rattlers, leading to an overall weakening effect. In between the limits, an optimum size ratio that maximizes the bulk modulus of the mixture is found. For loose systems, the bulk modulus decreases monotonically with addition of fines regardless of the size ratio. Finally, we relate the mixture properties to the 'typical' pore size in a disordered structure as induced by the combined effect of operating volume fraction (consolidation) and size ratio.
Fines migration behavior can either promote or obstruct fluid flow within the reservoir and is crucial for productivity optimization that needs fundamental understanding. The present work focuses on the contributions from effective stress build-up due to reservoir depletion and decreases in brine salinity from low-salinity water injection. Particle detachment is studied analytically by using the critical retention concentration function modified with stress and salinity dependents. Increase in formation effective stress leads to deformations within the reservoir configurations due to micro-cracks and reduced pore dimensions. Decreased size of the travelling channels promotes particle detachment, while the critical retention concentration decreases. A sensitivity study reveals that, under influence of effective stress, particle size and fluid velocity are dominant parameters controlling the fines migration by influencing the particle detaching forces. With decreasing brine salinity (e.g. via low-salinity water injection), clay particles that are attached on the pore surface increasingly swell, leading to reduction in effective pore space and flow channel. Decreased pore space direct
Percolation of fine particles (fines) in a static bed of larger particles is central to many industrial and natural processes. Non-cohesive fines either pass through the bed or become trapped depending on multiple factors including particle sizes, friction and restitution coefficients, and size-polydispersity. Here we consider the additional factor of cohesion. We use the discrete element method to simulate gravity-driven percolation of cohesive fine particles through a static bed of randomly packed large particles; fines interact with bed particles but not with each other. A large-to-fine particle diameter ratio of 7 geometrically permits non-cohesive fines to pass the narrowest pore throats formed by the large particles so they can freely percolate. However, sufficiently large cohesion and friction lead to non-geometric trapping. Fines are trapped when they fail to rebound after a collision, due to large cohesion, low restitution, and low collision velocity, and any subsequent rolling or sliding is insufficient to cause detachment. This establishes a sequence of local interactions -- collision, adhesion, and post-contact motion -- that governs the ultimate fate of a fine particle
The fine curve complex of a surface is a simplicial complex whose vertices are essential simple closed curves and whose $k$-simplices are collections of $k+1$ disjoint curves. We prove that the fine curve complex is homotopy equivalent to the curve complex. We also prove that the fine arc complex is contractible.
High-quality and open datasets remain a major bottleneck for text-to-image (T2I) fine-tuning. Despite rapid progress in model architectures and training pipelines, most publicly available fine-tuning datasets suffer from low resolution, poor text-image alignment, or limited diversity, resulting in a clear performance gap between open research models and enterprise-grade models. In this work, we present Fine-T2I, a large-scale, high-quality, and fully open dataset for T2I fine-tuning. Fine-T2I spans 10 task combinations, 32 prompt categories, 11 visual styles, and 5 prompt templates, and combines synthetic images generated by strong modern models with carefully curated real images from professional photographers. All samples are rigorously filtered for text-image alignment, visual fidelity, and prompt quality, with over 95% of initial candidates removed. The final dataset contains over 6 million text-image pairs, around 2 TB on disk, approaching the scale of pretraining datasets while maintaining fine-tuning-level quality. Across a diverse set of pretrained diffusion and autoregressive models, fine-tuning on Fine-T2I consistently improves both generation quality and instruction adhe
Earth observation (EO) is crucial for monitoring environmental changes, responding to disasters, and managing natural resources. In this context, foundation models facilitate remote sensing image analysis to retrieve relevant geoinformation accurately and efficiently. However, as these models grow in size, fine-tuning becomes increasingly challenging due to the associated computational resources and costs, limiting their accessibility and scalability. Furthermore, full fine-tuning can lead to forgetting pre-trained features and even degrade model generalization. To address this, Parameter-Efficient Fine-Tuning (PEFT) techniques offer a promising solution. In this paper, we conduct extensive experiments with various foundation model architectures and PEFT techniques to evaluate their effectiveness on five different EO datasets. Our results provide a comprehensive comparison, offering insights into when and how PEFT methods support the adaptation of pre-trained geospatial models. We demonstrate that PEFT techniques match or even exceed full fine-tuning performance and enhance model generalisation to unseen geographic regions, while reducing training time and memory requirements. Addi
General-purpose Large Language Models (LLMs) are frequently fine-tuned through supervised fine-tuning (SFT) to enhance performance in specific domains. Better results can be achieved by distilling the chain-of-thought of a larger model at the cost of numerous expensive calls and a much greater amount of data. We propose a novel blueprint for efficient fine-tuning that uses reasoning only for complex data identified by entropy. Specifically, across three small open models ($\approx 3B$) we split the training data into complexity categories by a single token answer entropy (ROC AUC $0.73$), fine-tune large language models (LLMs) via SFT and distillation, and show that our pipeline significantly outperforms the standard SFT approach ($0.58$ vs $0.45$ average accuracy) and outperforms the distillation approach ($0.58$ vs $0.56$ average accuracy) while using $81\%$ less data.
Fine-tuning policies learned offline remains a major challenge in application domains. Monotonic performance improvement during \emph{fine-tuning} is often challenging, as agents typically experience performance degradation at the early fine-tuning stage. The community has identified multiple difficulties in fine-tuning a learned network online, however, the majority of progress has focused on improving learning efficiency during fine-tuning. In practice, this comes at a serious cost during fine-tuning: initially, agent performance degrades as the agent explores and effectively overrides the policy learned offline. We show across a range of settings, many offline-to-online algorithms exhibit either (1) performance degradation or (2) slow learning (sometimes effectively no improvement) during fine-tuning. We introduce a new fine-tuning algorithm, based on an algorithm called Jump Start, that gradually allows more exploration based on online estimates of performance. Empirically, this approach achieves fast fine-tuning and significantly reduces performance degradations compared with existing algorithms designed to do the same.
Adaptation of foundation models using low-rank methods is a widespread approach. Another way to adapt these models is to employ orthogonal fine-tuning methods, which are less time and memory efficient despite their good generalization properties. In this work, we propose Householder Orthogonal Fine-tuning (HOFT), a novel orthogonal fine-tuning method that aims to alleviate time and space complexity. Moreover, some theoretical properties of the orthogonal fine-tuning paradigm are explored. From this exploration, Scaled Householder Orthogonal Fine-tuning (SHOFT) is proposed. Both HOFT and SHOFT are evaluated in downstream tasks, namely commonsense reasoning, machine translation, subject-driven generation and mathematical reasoning. Compared with state-of-the-art adaptation methods, HOFT and SHOFT show comparable or better results.
Fine-tuning has become the standard practice for adapting pre-trained models to downstream tasks. However, the impact on model robustness is not well understood. In this work, we characterize the robustness-accuracy trade-off in fine-tuning. We evaluate the robustness and accuracy of fine-tuned models over 6 benchmark datasets and 7 different fine-tuning strategies. We observe a consistent trade-off between adversarial robustness and accuracy. Peripheral updates such as BitFit are more effective for simple tasks -- over 75% above the average measured by the area under the Pareto frontiers on CIFAR-10 and CIFAR-100. In contrast, fine-tuning information-heavy layers, such as attention layers via Compacter, achieves a better Pareto frontier on more complex tasks -- 57.5% and 34.6% above the average on Caltech-256 and CUB-200, respectively. Lastly, we observe that the robustness of fine-tuning against out-of-distribution data closely tracks accuracy. These insights emphasize the need for robustness-aware fine-tuning to ensure reliable real-world deployments.
Ensuring compliance with international data protection standards for privacy and data security is a crucial but complex task, often requiring substantial legal expertise. This paper introduces LegiLM, a novel legal language model specifically tailored for consulting on data or information compliance. LegiLM leverages a pre-trained GDPR Fines dataset and has been fine-tuned to automatically assess whether particular actions or events breach data security and privacy regulations. By incorporating a specialized dataset that includes global data protection laws, meticulously annotated policy documents, and relevant privacy policies, LegiLM is optimized for addressing data compliance challenges. The model integrates advanced legal reasoning methods and information retrieval enhancements to enhance accuracy and reliability in practical legal consulting scenarios. Our evaluation using a custom benchmark dataset demonstrates that LegiLM excels in detecting data regulation breaches, offering sound legal justifications, and recommending necessary compliance modifications, setting a new benchmark for AI-driven legal compliance solutions. Our resources are publicly available at https://github.