Coarse Layer Research Articles

High-Precision Image Editing via Dual Attention Control in Diffusion Models Without Fine-Tuning

Existing diffusion models outperform generative models like Generative Adversarial Networks in image synthesis and editing. However, they struggle with high-precision image editing while preserving image details and the accuracy of editing instructions. To address these challenges, we propose a dual attention control method to achieve high-precision image editing. Our approach includes two key attention control modules: (1) cross-attention control module, which combines the cross-attention maps of the original and edited images through weighted parameters, ensures that the synthesized edited image retains the structure of the input image. (2) Self-attention control module, which varies based on the editing task, applied at “coarse” and “fine” layers, since the coarse layers help maintain input image details and the fine layers are better suited for style transformations. Experimental evaluations have demonstrated that our approach achieves excellent results in detail preservation, content consistency, visual realism, and semantic understanding, making it especially suitable for tasks requiring high-precision editing. Specifically, compared to the editing outcomes under no control conditions, the introduction of dual visual attention control has led to an increase of 6.19% in CLIP scores, a reduction of 29.3% in LPIPS, and a decrease of 24.7% in FID. These significant improvements not only validate the effectiveness of the dual attention control but also attest to the method’s substantial flexibility and adaptability across different scenarios. Notably, our approach is a zero-shot solution, requiring no user optimization or fine-tuning, facilitating real-world applications.

Zoom Text Detector.

To pursue comprehensive performance, recent text detectors improve detection speed at the expense of accuracy. They adopt shrink-mask-based text representation strategies, which leads to a high dependence of detection accuracy on shrink-masks. Unfortunately, three disadvantages cause unreliable shrink-masks. Specifically, these methods try to strengthen the discrimination of shrink-masks from the background by semantic information. However, the feature defocusing phenomenon that coarse layers are optimized by fine-grained objectives limits the extraction of semantic features. Meanwhile, since both shrink-masks and the margins belong to texts, the detail loss phenomenon that the margins are ignored hinders the distinguishment of shrink-masks from the margins, which causes ambiguous shrink-mask edges. Moreover, false-positive samples enjoy similar visual features with shrink-masks. They aggravate the decline of shrink-masks recognition. To avoid the above problems, we propose a zoom text detector (ZTD) inspired by the zoom process of the camera. Specifically, zoomed-out view module (ZOM) is introduced to provide coarse-grained optimization objectives for coarse layers to avoid feature defocusing. Meanwhile, zoomed-in view module (ZIM) is presented to enhance the margins recognition to prevent detail loss. Furthermore, sequential-visual discriminator (SVD) is designed to suppress false-positive samples by sequential and visual features. Experiments verify the superior comprehensive performance of ZTD.

Microstructural evolution and dynamic recrystallization mechanisms of additively manufactured TiAl alloy with heterogeneous microstructure during hot compression

Microstructural evolution and dynamic recrystallization (DRX) mechanisms of a Ti−48Al−2Cr−2Nb (at.%) alloy prepared by selective electron beam melting (SEBM) during hot deformation at 1150 °C and 0.1 s−1 were investigated by hot compression tests, optical microscope (OM), scanning electron microscope (SEM), electron back-scattered diffraction (EBSD) and transmission electron microscope (TEM). The results show that the initial microstructure of the as-SEBMed alloy exhibits layers of coarse γ grains and fine γ+α2+(α2/γ) lamellar mixture grains alternately along the building direction. During the early stage of hot deformation, deformation twins tend to form within the coarse grains, facilitating subsequent deformation, and a small number of DRX grains appear in the fine-grained regions. With the increase of strain, extensive DRX grains are formed through different DRX mechanisms in both coarse and fine-grained regions, involving discontinuous dynamic recrystallization mechanism (DDRX) in the fine-grained regions and a coexistence of DDRX and continuous dynamic recrystallization (CDRX) in the coarse- grained regions.

Quantification of spatial heterogeneity and its influence on particle migration

Suffusion experiments were utilised to quantify the evolving degree of heterogeneity in gap-graded soils. Upward flow was imposed on test specimens comprising a mixture layer with varying finer fraction contents overlaid by a coarse layer. A coaxial permeameter cell enabled the local permeability to be obtained using spatial time domain reflectometry. Three methods (coefficient of variation, Dykstra-Parsons coefficient, and Lorenz coefficient) were employed to quantify heterogeneity based on spatial variability in local permeability. For the two-layered specimen in this study, all three methods demonstrated that suffusion had a homogenising effect with particle migration from the mixture layer to the coarse layer. Detailed insights were obtained from a multi-layered approach, where the Dykstra-Parsons coefficient was found to be more sensitive to spatial variations, while the Lorenz coefficient was less dependent on the amount of data. The key observation was that an increase in heterogeneity led to a reduction in particle migration. This is an important finding as prior studies focused on homogeneous specimens, while this study demonstrates that small amounts of heterogeneity can significantly impact particle migration characteristics. This reinforces the need to quantify the evolving degree of heterogeneity.

Effect of Hydrate Saturation on Permeability Anisotropy for Hydrate-Bearing Turbidite Sediments Based on Pore-Scale Seepage Simulation

The permeability of natural gas hydrate (NGH) turbidite reservoirs typically exhibits significant anisotropy, with anisotropy being a crucial basis for evaluating reservoir production. The presence of hydrates, as a crucial constituent of the solid framework, not only impacts the overall permeability but also influences the permeability anisotropy. To investigate the saturation sensitivity of permeability anisotropy, a series of simulations are performed by integrating particle flow and computational fluid dynamics methods to construct the homogeneous and layered numerical samples and compute the evolution of permeability anisotropy. It is shown that the permeability is isotropic for homogeneous sediments and the isotropy remains unchanged regardless of variations in hydrate saturation. The permeability of layered sediments, in contrast, exhibits significant anisotropy due to the presence of dominant channels within the coarse layer. For uniformly distributed hydrates, the more effective blockage in coarse layers results in a reduction in anisotropy. While for preferentially distributed hydrates, the excess blocking of coarse layers makes the dominant channels transfer to the fine layers, the further blocking causes a U-shaped anisotropy–saturation curve characterized by a decrease–increase transformation. During the reservoir production process, the preponderance channels blocked by hydrates will be cleared and the horizontal permeability will significantly increase. As a result, the production efficiency of horizontal wells may exceed expectations. The findings offer a parameter support for production estimation and environmental assessment.

Effect of Median Soil–Particle Size Ratio on Water Storage Capacity of Capillary Barrier

Capillary barriers are widely used as a cover system to enhance the upper-soil-layer water storage capacity and reduce water infiltrate into the lower soil layer. In this paper, the effects of the median soil–particle size ratio on the water storage capacity of capillary barriers were studied using a series of indoor one-dimensional soil column infiltration tests. The results show that the water storage capacity rises with an increase in the median soil–particle size ratio until it exceeds 10. The variation in the total water storage capacity is related to not only the median soil–particle size ratio but also the particle size of coarse-grained soil or fine-grained soil. When the fine-grained soil-layer particle size is constant, the total water storage first increases, then decreases, and finally remains constant after increasing the median soil–particle size ratio. In contrast, when the coarse-grained soil layer particle size is constant, the relationship between the capillary barrier’s total water storage and median soil–particle size ratio can be defined as a power function. Using the capillary barrier can increase coarse-grained sand by 90% in water storage capacity and can only increase fine-grained sand by 7% in water storage capacity. The breakthrough time increases with the increase in the median soil–particle size ratio. The presence of the coarse and fine-grained soil layer interface in the capillary barrier can affect the fine-grained soil layer infiltration rate.

Pavement Binder Course Optimization

The main objective of this study is to design the asphalt pavement structure for the Alkut-Mayssan highway in Iraq as a case study by using PCASE application to determine the necessary thickness of CR-modified asphaltic layers. Additionally, Marshall tests were used to investigate the effect of the dry process crumb rubber modifier on the thickness of the binder coarse layer based on various variables, such as the crumb rubber modifier with different content (1, 2, 3, 4, 5)% and sizes (0.3mm, 2.36mm , 4.75mm, and mix gradation). The test results show that while pavement thickness reduced as crumb rubber gradation increases, MR values increase. Additionally, as compared to the control mixture, the addition of 1% of CR in various gradations could decrease the thickness of the asphalt binder layer.

Relationships Between Macrostructure and Microstructure in Lamellar Graphite Iron Castings

AbstractSpherical sheet steel molds filled with gray iron melts of varying chemical compositions and metallurgical conditions were air-cooled until solid, followed directly by austempering to preserve the austenite grain structure. The castings were studied using a combination of cooling curves and quantitative metallography, in order to clarify control of the austenite grain structure and its impact on the local microstructure. A novel method utilizing fast Fourier transform provided visual overview of macroscopic trends in the scale of the flake graphite structure. Castings inoculated with Sr-containing ferrosilicon featured finer eutectic cell structure but coarser equiaxed structure of austenite, emphasizing that melt treatments applied to control the graphite structure may have unintended effects on the austenite grain structure. In most non-inoculated castings, the microstructure was banded, with alternating layers of coarse and fine flake graphite with distance from the casting surface. The extent of the columnar zone of austenite grains showed no correlation with the graphite structure nor the volume fraction of dendrites. The volume-to-surface ratio of dendrites was more uniform in the columnar zone, but increased toward the center in the equiaxed zone. The casting with the highest carbon equivalent (4.34), featured zones containing finer dendrites and graphite. These zones appear to be gaps in the early solidification structure which filled later by secondary dendritic growth from surrounding austenite. This highlights that high carbon equivalent may lead to poor dendrite coherency which can make the microstructure less uniform and less predictable.

Brain MRI and CT Image Fusion Using Multiscale Local Extrema and Image Statistics

In medical applications such as radiotherapy and guided-image surgery, data fusion for diagnostic imaging has emerged as a critical issue. Because the objective of medical image fusion is to improve patient diagnosis accuracy, the fused image is created by preserving the source images' prominent details and features. It has been demonstrated that the Multi-Level Local Extrema representation has numerous advantages over traditional image modeling approaches. We propose an innovative MLE-based fusion method for multimodal medical images in this paper. In the MLE schema, the proposed algorithm decomposes source images into coarse and detailed layers, then fuses the source images using weights calculated from these detail images using image statistics. We visually and quantitatively compared the efficacy of the suggested approach to that of existing methods using five different types of medical images from various sources. The experimental results showed that the proposed scheme outperforms other current typical schemes in terms of both qualitative image quality and objective evaluation.

Exploration of KCl deposition dynamics for the formation of coarse and fine layer deposits

In biomass and waste combustion, the release of KCl into the gas phase results in its deposition on heat exchanger surfaces, enhancing their stickiness and accelerating deposit buildup from incoming ash particles. Despite this, the mechanisms governing KCl deposition remain inadequately explored. KCl has been observed to form either a fine, powdery layer through thermophoresis of nucleated KCl particles or a crystal-like, coarse layer through heterogeneous condensation of KCl vapor. However, the quantitative assessment of coarse and fine layer formation has never been performed. In this study, an entrained flow reactor and mechanistic model were employed to analyze KCl deposition behavior, examining the effect of probe temperature and exposure time. Quantification of both the coarse and fine deposit layers was accomplished. Probe temperature was found to have no influence on the deposition in the range from 300 °C – 595 °C. For coarse layer formation, the critical factor was the development of a fine layer upon which the crystal-like structure of the coarse layer grows. Lower probe temperatures accelerate initial layer formation, initiating the coarse section sooner. Additionally, it slightly increases nucleation and condensation, however, nucleation is increased stronger resulting in a lower ratio of coarse to fine deposit. The quantification of the coarse mass, in conjunction with the simulation, provides a first estimation of the critical mass required to initiate heterogeneous condensation on a steel tube. In this study, a critical mass of 1.26e-3 g/cm2 was determined.

The 2021 floods in the Netherlands from a river engineering perspective

In July of 2021, large areas in the catchment of the Meuse River in Belgium, the Netherlands and Germany were affected by extreme rainfall and floods. In the Netherlands, measured peak discharges in the upstream part of the Meuse and regional tributaries were the highest ever recorded. This resulted in extreme water levels in the upstream reaches, but downstream of Roermond (about 80 km downstream of the border with Belgium) water levels were significantly lower than during previous flood events with similar discharge levels near the border with Belgium. The lower water levels resulted from the implementation of the large-scale room for the river program, in combination with very strong peak attenuation of the flood wave. Peak attenuation was especially high in the so-called Lake Meuse in the Netherlands, due to large storage areas in the floodplains. The floods in the Netherlands came in certain parts of the Meuse basin rather unexpected. The flood forecasts in the upstream part of the Meuse in the Netherlands depended heavily on rainfall forecasts and rainfall-runoff modelling and underestimated the peak water levels up to 36 hours before the flood actually peaked. Further downstream, the lead time increases and forecasts are based on discharge levels that are measured in upstream parts of the catchments. This results in more accurate estimates. The estimated probability of occurrence of the measured peak water level is around 1:100 per year for the Meuse at Borgharen (rkm 16) and decreases to 1:15 per year downstream near Gennep (rkm155). In the tributaries in Limburg the probabilities of occurrence of the measured water levels vary widely: at many locations along the rivers Geul, Geleenbeek and Roer, exceedance probabilities are estimated to be between 1:100 and 1:1000 per year. The floods have also resulted in unprecedented morphological changes. The armour layer in the riverbed of the ‘Common Meuse’, consisting of very coarse gravel, was mobilized and layers of fine sand quickly eroded. This resulted in multiple scour holes with depths of 3 to 15m. Riverbank erosion was observed and large quantities of sand were deposited on the river banks.

Main results of permafrost monitoring in the French Alps through the PermaFrance network over the period 2010–2022

AbstractThis study presents data from the first years of permafrost monitoring in boreholes in the French Alps that started at the end of 2009 in the framework of the PermaFrance network. Nine boreholes are instrumented, among which six monitored permafrost temperature and active layer thickness (ALT) over >10 years. Ice‐poor and cold permafrost in high‐elevation north‐facing rock walls has warmed by up to >1°C at 10 m depth over the reference decade (2011–2020), whereas ice‐rich permafrost (rock glacier) temperatures remained stable. ALT has increased at four of the five boreholes for which decadal data are available. Summer 2015 marks a turning point in ALT regime and greatest ALT values were observed in 2022 (available for six boreholes), but thawing intensity did not show an obvious change. At one site with a layer of coarse blocks about 2 m thick, ALT was stable over 2018–2022 and response to the hottest years was dampened. Linear trends suggest an ALT increase of 2 m per decade for some ice‐poor rock walls, independently of their thermal state. The data reveal a variety of permafrost patterns and evolution with significant intraregional and local differences. Snow modulates the response to air temperature signal in various ways, with an important effect on near‐surface temperature trends and ALT: early snow melting in spring favors an ALT increase in rock walls. Maintaining these monitoring systems and understanding the physical processes controlling heterogeneous responses to climate signals is crucial to better assess permafrost dynamics and to adapt to its consequences.

Enhancing Water Quality Management in Ornamental Fish Aquariums with Pot and Bamboo Filter Systems

The filtration of aquarium wastewater composed of a pot and bamboo is constructed for the purification of microbial activities and to reduce turbidity. The aquarium wastewater is treated with sand to reduce BOD and COD; the usage of coconut husk and vetiver act to decrease the turbidity, nitrite, and pH while raising DO level. Initially, sand was sieved to produce 12 size ranges between 180 and 2000 m. The effectiveness of each range in removing turbidity, BOD, and COD through sand beds 4 or 5 cm deep were tested. The critical size (300 & 425 m) at which no solid particles could penetrate through the sand bed was observed at 3 and 4 cm in depth, respectively. Moreover, the flow of solids through sand filter beds had all particles smaller than 180 m removed from them. A 2 cm bed of sand allowed 0.2-0.5% of solids to flow through, and a 4 cm bed allowed 0.07-0.1 %. Both fine and coarse sand layers are supported by a layer of gravel. In the column approach, the thickness and flow rate of each layer were fixed. The thickness of the layers of sand and charcoal were fixed at 4 cm and 6 cm, respectively. The findings of the column analysis, the combined filter showed decreases in turbidity by 48%, COD by 52%, BOD by 30%, and increasing DO by 48%, respectively. The results reveal that absorbing nutrients can serve as bio-filters to create better conditions for fish growth. The final weight, length, and SGR of fishes in the three aquariums were significantly different after seven weeks of trial, the treated aquarium showed the highest value than the control.

Effect of interlayer mixed zone and effective stress on permeability anisotropy of NGH turbidite reservoir

Resulting from the bedding structure of turbidite reservoir, permeability anisotropy is an essential parameter for natural gas hydrate (NGH) exploitation. Fundamentally, the evolution of permeability anisotropy under increasing effective stress are investigated in this study. Taking the turbidite sediments in northern Cascadia as an example, a series of experimental tests and co-simulations of discrete element method (DEM) and computational fluid dynamics method (CFD) are conducted. For layered turbidite sediments, an interlayer mixed zone will be formed in the interface between fine-grained layer and coarse-grained layer after loading. Fine particles will migrate to the coarse layer, blocking the horizontal preponderance flow paths and further reduce the permeability anisotropy. Moreover, the development of interlayer mixed zone mainly occurs in initial compaction stage and slows down in higher stress level, for which the permeability anisotropy shows a staged reduction with increasing effective stress. Considering the effect of interlayer mixed zone, a predictive model of permeability anisotropy is proposed and verified by the experimental results, which has higher accuracy than traditional “layered cake” model. The finding of this research also confirms the mixing effect in deep sediments, which may affect the geological history records in target sedimentary sequence.

Examining the criteria for contact erosion in granular soil foundations

Contact erosion, or interfacial erosion, often occurs in layered foundations composed of coarse and fine layers; it refers to the phenomenon that the particles of the fine layer are detached by the flow parallel to the interface and transported through the pore channels of the coarse layer. To visualise the initiation and progression of contact erosion at the internal interface of soil strata, transparent soil model tests were employed to simulate real soil, and a series of contact erosion tests was conducted on double-layer foundations composed of different coarse and fine layers. Based on the test results, the commonly used geometric and hydraulic criteria for contact erosion were examined. It is shown that the geometric criteria are reliable, while hydraulic criteria give diverse results. For the same fine soil layer, the critical hydraulic gradient varies greatly with the coarse soil layer, while the critical flow velocity varies in a relatively small range since flow velocity has an intrinsic relationship with the dragging force imposed on particles by seepage flow. More efforts should be invested on the criteria based on critical velocity in further studies of contact erosion.

Facilely fabricating large-area robust heterogeneous wettability surface by mask-patterned ultrafine anode scanning electrodeposition for efficient water collection

Water collection has been highly demanded in water-deficient areas, and heterogeneous wettability surfaces have been proven to be significantly advantageous in meeting this demand. However, the currently available large-area metallic surfaces fabricated are less efficient, cost-effective, and mechanically robust. To overcome these shortages, a unique mask-patterned ultrafine anode scanning electrodeposition (MUAS-ECD) technology was proposed in this work, and simultaneously a nonconventional deposit design and preparation process strategy was developed. With the MUAS-ECD, a functional heterogeneous wettability surface containing a number of parallel superhydrophilic convex strips and hydrophilic concave grooves can be constructed concurrently on a large-area metallic substrate. With the design and preparation strategy, the robust convex nickel strips with alternately stacked compact and coarse material structural layers could be facilely formed only by adjusting the processing voltage. It was found that the achieved 5-layer deposit patterned strip-groove structural surfaces (5-LDPS) demonstrate an incredibly high water collection rate of 1.61 g h−1cm−2. Surprisingly, these surfaces still exhibit excellent water-collection ability even after 100 abrasion cycles with sandpaper. The proposed MUAS-ECD and accompanying process strategy show great potential in structuring large-area robust heterogeneous wettability metallic surfaces for efficient fog harvesting in water-starved environments.

Transcrustal magma plumbing process in the Cenozoic Dali potassic lamprophyre dyke formation in the southeastern Tibet Plateau

(Ultra-)potassic lamprophyres are volumetrically rare, but are widely distributed in diverse geological settings and extensively applied to trace the mantle source and melting processes. However, it is still debatable whether the lamprophyre formation involved transcrustal magmatic processes. In this study, we describe the diverse zoning patterns and textures of clinopyroxene (Cpx) from a Cenozoic potassic lamprophyre dyke in southeastern Tibet Plateau. Three types of Cpx macrocrysts with normal- and reverse oscillatory zoning and sieved textures, and Cpx microcrysts from Cpx–Phl nodules are recognized. The Cpx macrocrysts have highly variable Mg# (62–89) and chemical compositions. Their oscillatory-zoned mantle has alternating coarse layers or closely-packed layers, with distinct resorption surface shown in backscattered electron (BSE) images. High-amplitude oscillation of Mg# (±5 to ±12), Al, Cr and Sr at the boundary layers of mantle may reflect multiple magma mixing/recharge events. In contrast, low-amplitude and high-frequency oscillation of Mg# (varying from < ±3) across the closely-packed layers could be attributed to local magma mixing or kinetic effect at the crystal-melt interface. The sieved domains occur in the Cpx macrocrystic core with spongy space and linking the external groundmass by cracks, and have major and minor compositions and 87Sr/86Sr ratios similar to those of the rims. This indicates that the sieved domains are probably originated from dissolution of the Cpx macrocrysts by groundmass melt and later reprecipitation. The estimated pressure and temperature for the Cpx macrocryst high-Mg# cores and low Mg# rims are 580–1160 MPa (avg. 919 MPa) and 30–1270 MPa (avg. 681 MPa), and 1116–1266 °C (avg. 1216 °C) and 955–1215 °C (avg. 1110 °C), respectively, suggesting polybaric differentiation or Cpx-saturated magmas. The estimated pressure and temperature for the reverse-zoned Cpx marcocryst cores are 450–1310 MPa (avg. 953 MPa) and 1097–1209 °C (avg. 1166 °C), respectively, indicating that they are likely formed in equilibrium with a more evolved magma batch, and then were entrapped by more primitive mantle-derived potassic magmas during magma upwelling. The La/Yb and Dy/Yb modelling of equilibrium melts suggest that magmas of the Cpx–Phl nodules and their lamprophyre host were derived from similar mantle sources. Therefore, the complex zoning patterns and compositions of Cpx macrocrysts could indicate transcrustal magmatic processes in the potassic lamprophyre dyke formation.

Experimental study of the inrush of fines events in caving mining

Cave mining has inherent risks, which include rock bursts, collapses, subsidence, and inrushes. Among the inrush events that can occur are the inflows of water, mud, and dry-fine material. The latter type of inflow is one about which we have the least amount of information and little research, despite the fatal consequences that have occurred associated with this phenomenon. In this work, the inrush of fine events was simulated at a laboratory scale to study the main variables that influence these events. An experimental setup consisting of a drawbell filled with different layers of coarse and medium granular material was used as a base. A hang-up was induced in the setup and then fine material was added on top to observe the consequences. Three experimental stages were developed applying three distinct fine materials: glass microspheres, fragmented concrete, and finally clay. It was observed that the magnitude of the inrush event was directly related to lower friction and higher sphericity of the fine material. Finally, the inrush of fine events did not apparently occur due to shear failures in the experimental setup; rather in the simulated scenarios with the presence of coarse material, the removal of a hang-up was always required as a trigger mechanism to generate an inrush of fines event.

Thermal Treatment of Trichloroethene by Electrical Resistance Heating: Visualization of Gas Production in Coarse Layers

The effective implementation of in situ thermal treatment (ISTT) technologies requires understanding of gas production and migration in heterogenous media. However, investigations of the effects of high permeability contrast on gas formation, accumulation, and migration, as well as its potential effect on the redistribution of dense non-aqueous phase liquid (DNAPL), are relatively rare. In this study, electrical resistance heating (ERH) experiments were conducted in a thin sand-packed cell to simulate common yet not well-studied scenarios encountered during ISTT applications, such as coarse lenses surrounded by finer material. Two packing configurations were employed: 2 mm glass beads surrounded by 20/30 silica sand and 20/30 silica sand overlaying 40/50 silica sand. Each experiment contained an emplaced pool of trichloroethene (TCE) within the coarse material. If permeable material or pathways were present between the coarse lens and the upper cell boundary, the gas migrated along these pathways, and local DNAPL redistribution was limited to near the top of the pool before it vaporized. In contrast, if the coarse material was surrounded by finer material and contained a sufficient volume of DNAPL, the gas accumulated inside the coarse lens leading to DNAPL displacement from the lens. For five selected DNAPLs, this volume was estimated to be 0.1% to 0.5% of the total pore volume of the coarse material. The conceptual model developed in this study improves our understanding of this common geological scenario, demonstrating the importance of considering both lower- and higher-permeability material and their effects on multiphase flow during co-boiling, as well as the design of gas extraction systems during ISTT applications.

Pore network modeling of capillary barrier effects: impact of pore sizes

Although the working principles of capillary barrier effects (CBEs) have been well explained based on unsaturated soil mechanics, the selection of materials for engineering structures with CBEs mostly relies on designers’ experience or previously reported data, largely due to the lack of understanding of the microscale behavior of CBEs. This study explores the impact of pore sizes on CBEs based on pore network modeling. The results indicate that the pore size variability in either the fine or coarse layer has a remarkable influence on CBEs, with the latter having a dominating influence. A larger coarse-to-fine mean pore size ratio results in more effective CBEs. In addition to selecting materials with more uniform pore sizes, the coarse-to-fine mean pore size ratio is recommended to be &gt;6 (median particle size ratio of 30) to ensure the performance of CBEs.