Coarse Cell Research Articles

Cut‐Cell Microstructures for Two‐scale Structural Optimization

AbstractTwo‐scale topology optimization, combined with the design of microstructure families with a broad range of effective material parameters, is widely used in many fabrication applications to achieve a target deformation behavior for a variety of objects. The main idea of this approach is to optimize the distribution of material properties in the object partitioned into relatively coarse cells, and then replace each cell with microstructure geometry that mimics these material properties. In this paper, we focus on adapting this approach to complex shapes in situations when preserving the shape's surface is essential.Our approach extends any regular (i.e. defined on a regular lattice grid) microstructure family to complex shapes, by enriching it with tiles adapted to the geometry of the cut‐cell. We propose a fully automated and robust pipeline based on this approach, and we show that the performance of the regular microstructure family is only minimally affected by our extension while allowing its use on 2D and 3D shapes of high complexity.

Using graph neural networks for wall modeling in compressible anisothermal flows

Abstract Compressible anisothermal flows, which are commonly found in industrial settings such as combustion chambers and heat exchangers, are characterized by significant variations in density, viscosity, and heat conductivity with temperature. These variations lead to a strong interaction between the temperature and velocity fields that impacts the near-wall profiles of both quantities. Wall-modeled large-eddy simulations (LESs) rely on a wall model to provide a boundary condition, for example, the shear stress and the heat flux that accurately represents this interaction despite the use of coarse cells near the wall, and thereby achieve a good balance between computational cost and accuracy. In this article, the use of graph neural networks for wall modeling in LES is assessed for compressible anisothermal flow. Graph neural networks are a type of machine learning model that can learn from data and operate directly on complex unstructured meshes. Previous work has shown the effectiveness of graph neural network wall modeling for isothermal incompressible flows. This article develops the graph neural network architecture and training to extend their applicability to compressible anisothermal flows. The model is trained and tested a priori using a database of both incompressible isothermal and compressible anisothermal flows. The model is finally tested a posteriori for the wall-modeled LES of a channel flow and a turbine blade, both of which were not seen during training.

Development method of splitting the radio-reflective large-sized reflector surface into cutting patterns made of mesh with coarse cells

An important design component of space antenna reflectors is a reflecting surface. At present, metallic knitted mesh fabric is most often used as a reflective surface; it fully meets the requirements for such surfaces in terms of physical and mechanical properties. Particularly important characteristics for large-sized reflectors are low specific gravity and high light transmission coefficient, which a mesh with coarse cells provides. However, the structural features of this mesh fabric impose additional requirements on its use as a reflective surface for large-sized transformable reflectors. The article deals with a special approach to designing patterns for a reflective surface made of mesh with coarse cells. The developed pattern cutting option allows to obtain a reflective surface with the least distortion (error), in accordance formulated criteria. The authors have suggested the option of obtaining cutting patterns of metallic mesh fabric, described the advantages of the proposed option over the classical one. The cutting pattern of the parabolic surface for the offset reflector was obtained. The proposed cutting pattern option was analyzed. Furthermore, optimization was carried out, which made it possible to maintain the integrity of the cellular structure of patterns along one loop row. The compared results show that the developed cutting pattern option completely complies with the requirements and it can be used to form a reflective surface from the metallic knitted mesh with coarse cells.

A Switched-Huygens-Subgridding-Based Combined FDTD–PITD Method for Fine Structures

A new combined method is proposed for the problems containing fine structures. In this method, the precise-integration time-domain (PITD) method is adopted for fine structures with fine cells, and the finite-difference time-domain (FDTD) method is used for the rest with coarse cells. The fields in the coarse and fine cells are updated synchronously, which makes the proposed method fast. Besides, since the PITD method is only applied locally, the limitation of its large memory requirement is also alleviated. The numerical results confirm that this combined method has a fast computational speed.

An improved calculation scheme of electron flow in a propagator method for solving the Boltzmann equation

In order to accurately evaluate the electron acceleration process in the calculation of the time evolution of the electron velocity distribution function (EVDF) based on the Boltzmann equation, an improved scheme blending upwind and central differences is introduced into the propagator method (PM). While the previous PM based on the upwind scheme needs fine cells to obtain an accurate EVDF at low electric fields, the improved PM is robust against coarse cells, which allows the reduction of cell resolution. Calculations of the EVDF in Ar under RF electric fields demonstrated that the blending scheme can provide satisfactorily accurate results even with cells about tenfold larger than the upwind case at low reduced electric fields below 1 Td, which leads to much shorter computational time because the reduction in the number of cells satisfactorily compensates for the complexity of the blending scheme. This technique has been built into a new user-friendly PM software named BOSPROM.

3-D forward modelling for DC resistivity method based on smooth multiscale finite-element algorithm

SUMMARYWhen the conventional finite-element method is used to simulate the 3-D direct-current (DC) resistivity response over a conductive earth with large complex structures, it requires finely discretized mesh to accurately represent the underground structures. Directly solving the current-conduction problem on the fine mesh will lead to a huge amount of calculation. In this paper, we develop a fast 3-D forward modelling method for DC resistivity method based on smooth multiscale finite-element algorithm. Instead of using the conventional polynomial basis functions, we construct the multiscale basis functions by solving the local boundary-value problems of partial differential equation in parallel on the multiscale meshes. The multiscale basis functions can capture the small-scale heterogeneous information in coarse cells and reflect it to the large scale by assembling macro matrix of coarse mesh. Thus, it enables us to quickly obtain the accurate solution by solving the original current-conduction problem with complex structures on coarse mesh. We further adopt the oversampling technology to improve the forward modelling accuracy. Besides, by combining with the gradient smoothing technology, we avoid establishing the continuous form of multiscale basis functions and their spatial derivative integral operation to rapidly assemble the macro matrix. Finally, the reliability of the proposed algorithm is verified by applying our code to the 3-D complex models and comparing it with the conventional finite-element method.

What is a static head measurement? Evaluating structural error and implications for regional groundwater modeling

AbstractMost groundwater modelers avoid using static heads measured from active production wells because they can introduce a bias into model calibration. However, in the deep confined Cambrian‐Ordovician Sandstone Aquifer System in the Central Midcontinent of North America, dedicated observation wells are sparse and remote from areas of most concentrated pumping. As a result, in areas where drawdown is the greatest and modeling is most needed, only static heads from production wells are available for calibration. This paper evaluates two leading sources of discrepancies in using production well data, spatial and temporal structural error (S.E.). A simple Theis solution is used to evaluate the potential magnitude of spatial S.E. when calibrating a regional MODFLOW model with coarse cell resolution. Despite theoretical analyses indicating that spatial S.E. could be significant, statistical analysis of the model results suggests that temporal S.E. is dominant. Long (ranging over decades) or frequent (monthly) head datasets are key in understanding temporal S.E., to better capture water‐level variability. In this study, the range in static head observations impacted estimates of the remaining time a well could extract water from the aquifer by 0.1 to 16.0 years. This uncertainty in future water supply is highly relevant to stakeholders and must be assessed in hydrographs depicting risk.

Intuitive approach for design of inner structure by mimicking magnetic effect

Abstract Three-dimensional (3D) printers enable the realization of parts with complex shapes, particularly parts with an internal structure as well as an external shape. They are beginning to be used in real production as well as for prototyping. Because 3D printers allow shapes for parts that could not be imagined when conventional machining processes were the only fabrication methods, designers need to design porous parts with minimum weights that can maintain the required load-bearing capability without concerns about the realization of those parts. However, designers cannot design the internal structure of the parts using the current computer-aided design systems as freely as they can design the external shape. Therefore, in this paper, an intuitive design tool is proposed for users to interactively design an internal structure inside given external shape. To create a porous structure inside given closed volume, we aimed to generate a honeycomb-like structure comprising cells whose size and crowdedness can be intuitively manipulated by a designer. Thus, fine cells exist in certain regions, while coarse cells exist in the remaining regions according to the design. To realize this aim, the phenomenon in which more iron particles are attracted near a magnet and fewer are attracted further away from the magnet is imitated. More and finer honeycomb-like cells are attracted near a magnet located on the external surface of the closed volume. The designer can add magnets and move them on the external surface until the desired internal honeycomb-like structure is obtained.

Pyramidal Semantic Correspondence Networks

This paper presents a deep architecture, called pyramidal semantic correspondence networks (PSCNet), that estimates locally-varying affine transformation fields across semantically similar images. To deal with large appearance and shape variations that commonly exist among different instances within the same object category, we leverage a pyramidal model where the affine transformation fields are progressively estimated in a coarse-to-fine manner so that the smoothness constraint is naturally imposed. Different from the previous methods which directly estimate global or local deformations, our method first starts to estimate the transformation from an entire image and then progressively increases the degree of freedom of the transformation by dividing coarse cell into finer ones. To this end, we propose two spatial pyramid models by dividing an image in a form of quad-tree rectangles or into multiple semantic elements of an object. Additionally, to overcome the limitation of insufficient training data, a novel weakly-supervised training scheme is introduced that generates progressively evolving supervisions through the spatial pyramid models by leveraging a correspondence consistency across image pairs. Extensive experimental results on various benchmarks including TSS, Proposal Flow-WILLOW, Proposal Flow-PASCAL, Caltech-101, and SPair-71k demonstrate that the proposed method outperforms the lastest methods for dense semantic correspondence.

Effect of artificial aging on the strength, hardness, and residual stress of SLM AlSi10Mg parts prepared from the recycled powder

This study investigated the effect of artificial aging on the hardness, strength, and residual stress of SLM (Selective Laser Melting) AlSi10Mg parts prepared from recycled powder used for 30 SLM printing cycles. First, an analysis was conducted for a comparison of recycled and virgin powders; it revealed that the powders have the same shape and phase and chemical composition. The only significant differences were the slightly increased average particle size and higher oxygen content in the recycled powder, especially on the powder surface. Typical molten pools were found in the microstructure, and a fine cellular type of structure was observed. The peripheral areas of the molten pools contained coarser cells with an average size of 0.6 μm2, and the middle part of the molten pools contained finer cells with an average size of 0.17 μm2. Fine Si precipitates with an average size of 25 nm were observed inside the cells. In terms of microstructure, no significant differences were observed when compared to the published data relating to SLM using virgin powders. This suggests that recycled powder after 30 printing cycles can be utilized for SLM component preparation to achieve cost-effectiveness. However, comparison of products with the use of virgin and recycled powders showed that the use of recycled powder affects the submicron characteristics (precipitate size increase), density and the resulting mechanical properties. Artificial aging of samples from the recycled powder at 170 °C caused an increase in hardness, Young's modulus and strength, while ductility remained essentially the same after aging. Peak values of hardness and strength were determined after 6 h of artificial aging. The increase in strength of the prepared samples can be explained by the precipitation of fine Si particles and their coarsening inside the cells. However, after longer aging times, the precipitates continued to coarsen, coalesced and displayed a plate-like shape, which led to the reduction in strength. The residual stress measured by the bridge curvature method (BCM) showed a significant decrease with increasing artificial aging time. The lowest residual stress was determined for an aging time of 100 h; however, the difference between residual stress for 6–7 h and 100 h was not significant. In order to obtain higher strengths and remove internal stresses after SLM during one heat treatment, the most suitable time for artificial aging at 170 °C seems to be 6 h.

Overlapping Cell Segmentation of Cervical Cytology Images Based on Nuclear Radial Boundary Enhancement

The accurate segmentation of cervical cell images is one of the key steps of the cervical cancer computer-aided diagnosis system. For the problem of overlapping cell and boundary blurring in cervical cell clusters, the researchers propose a segmentation algorithm based on the nuclear radial boundary enhancement for overlapping cell of cervical cytology images. This method not only suppresses the noise of cervical cytology images but also preserves the contrast of overlapping cell boundary. The researchers generate the weight graph by the candidate contour points and contour line segment attributes and utilize the dynamic programming algorithm to find the shortest path in the weight graph. The shortest path corresponds to the coarse segmentation contour in the cell image. The level set model is used to finely segment the obtained coarse cell segmentation boundary, so as to obtain the final cervical cell boundary. Through the quantitative and qualitative evaluation results, such as dice similarity coefficient, true positive rate, and false positive rate, it can be seen that the overlapping cell segmentation algorithm in this paper has achieved better segmentation results. Compared with other current overlap cell segmentation algorithms, the segmentation results obtained in this paper have greater advantages.

Mixed Generalized Multiscale Finite Element Method for flow problem in thin domains

In this paper, we construct a class of Mixed Generalized Multiscale Finite Element Methods for the approximation on a coarse grid for an elliptic problem in thin two-dimensional domains. We consider the elliptic equation with homogeneous boundary conditions on the domain walls. For reference solution of the problem, we use a Mixed Finite Element Method on a fine grid that resolves complex geometry on the grid level. To construct a lower dimensional model, we use the Mixed Generalized Multiscale Finite Element Method, which is based on some multiscale basis functions for velocity fields. The construction of the basis functions is based on the local snapshot space that takes all possible flows on the interface between coarse cells into account. In order to reduce the size of the snapshot space and obtain the multiscale approximation, we solve a local spectral problem to identify dominant modes in the snapshot space. We present a convergence analysis of the presented multiscale method. Numerical results are presented for two-dimensional problems in three testing geometries along with the errors associated to different numbers of the multiscale basis functions used for the velocity field. Numerical investigations are conducted for problems with homogeneous and heterogeneous properties respectively.

하이브리드 난류 모델을 이용한 전류고정덕트 후류의 고정도 수치 해석

A hybrid turbulence model has developed by combining a sub-grid scale model using dynamic k equation in LES with k-ω SST model of RANS equation. To ascertain potential applicability of the hybrid turbulence model, fully developed turbulent channel flows at ReSUBτ/SUB=180 have been simulated of which computational domain has a top wall with coarse cells and a bottom wall with fine cells. The streamwise mean velocity and turbulent intensity profiles showed a good agreement with DNS data when using the hybrid model rather than using a single model in k-ω SST or dynamic k equation models. Computational simulations of turbulent flows around KVLCC2 with a pre-swirl duct have been mainly performed using the hybrid turbulence model. Compared to the results obtained from RANS simulation with k-ω SST model as well as LES with dynamic k equation SGS model, turbulent wakes of the duct in the present simulation using the hybrid turbulence model were very similar to that of LES. Also, the resistances acting on hull, rudder and duct in hybrid turbulence model were similar to those in RANS simulation whereas the viscous forces acting on the hull in LES had a significant error due to coarse cells inappropriate to the sub-grid scale model.

Multi‐layer random walker image segmentation for overlapped cervical cells using probabilistic deep learning methods

A method for overlapping cell image segmentation is presented with a focus on multi-layer image processing in a three-phase scheme. In the first phase, a convolutional neural network is developed to provide a coarse cell segmentation with multiple output layers to identify cell cytoplasm, locations of cell nuclei, and the background, all as probabilistic image maps for the layer outputs. In the second phase, the probabilistic image maps from the convolutional neural network are used to identify locations of cell nuclei and cell cytoplasm. Then, multi-layer random walker image segmentation is used with cell nuclei as hard initial seeds and the cytoplasm estimates as soft seeds in a diffusion graph-based segmentation of the cells. With rough cell segmentation from both the trained convolutional neural network and the multi-layer random walker graph-based technique, a third phase combines and refines the cell segmentation using the Hungarian algorithm to optimise the assignment of individual pixel locations for the final cell segmentation. We evaluate the proposed method on cervical cell images generated from the International Symposium on Biomedical Imaging 2014 dataset with results that give a Dice similarity coefficient of 97.2% (compared to 93.2% for competitors) when trained on the generated dataset.

Effects of rescanning parameters on densification and microstructural refinement of 316L stainless steel fabricated by laser powder bed fusion

• The effect of laser power and scan speed on microstructure of 316L SS is explored. • Cell size of single scan can guide microstructural expectation of multilayer scan. • Rescanning with a 2nd parameter achieved 70.2 % cell size refinement. • Laser rescanning increased the relative density by 0.21−0.42%. • Spatially heterogeneous microstructure was achieved by applying rescanning. A challenge with microstructural control and refinement in laser powder bed fusion (LPBF) is maintaining high density when choosing parameters for desired microstructures. Rescanning during LPBF has been reported to improve densification and decrease surface roughness for many different alloys. However, little has been reported regarding the effects of locally rescanning with varying processing parameters on sub-grain cell size refinement for 316L stainless steel (SS). This study presents a novel solution to enable high densification with microstructural control in 316L SS by using a set of initial scanning parameters to achieve densification and a different set of rescanning parameters to refine the microstructure. Results showed that rescanning resulted in heterogeneous microstructure with coarse cell size of 0.84 μm and locally refined cell size of 0.35 μm, while maintaining a high level of densification (99.96 %), therefore enabling potential variations in component strength and hardness. The spatial distribution of local microstructure refinement was dictated by the melt pool dimensions of initial scanning and rescanning relative to the powder layer thickness. To better understand the link between LPBF process parameters and microstructure, the Wilson-Rosenthal equation was used to predict cooling rate ( G × R ) and correlate with sub-grain cell size. Such variation in properties may be useful for applications requiring parts with hardened surfaces, or localized strengthening at stress concentrations and sites of expected failure.

Influence of different pretreatments on the quality of wheat bran-germ powder, reconstituted whole wheat flour and Chinese steamed bread

This paper studied the influences of low-temperature (temp) extrusion (nonexpanded), high-temp extrusion (expansion), explosion puffing and high-temp jet milling on the modification of wheat bran-germ powder (WBGP) and on the dough properties, storage stability and Chinese steamed bread (CSB) quality in relation to reconstituted whole wheat flour (WWF). The four pretreatments significantly increased the soluble dietary fiber content of WBGP by 79.0%, 72.8%, 47.6% and 70.9%, respectively, and the lipase and lipoxygenase activities were inactivated to varying degrees. The extrusion technologies did not cause significant changes in the gluten indexes and pasting properties of WWF, while the low-temp extrusion had better thermal stability and anti-retrogradation. The gas retention coefficients of WWFs were pretreated by extrusion was higher than the other WWFs. The specific volume of CSB pretreated by low-temp extrusion was slightly larger than that of control, and the cell diameters and coarse cell volumes were smaller. Pretreatment can obviously inhibit the increase in fatty acid contents of WWF during storage and inhibit enzyme activity. The storage stability of extrusion technologies was better than that of wheat flour. Altogether, low-temp extrusion was more suitable for pretreating WBGP.

Effect of Nano-Si3N4 Reinforcement on the Microstructure and Mechanical Properties of Laser-Powder-Bed-Fusioned AlSi10Mg Composites

Laser powder bed fusion (LPBF) technology is of great significance to the rapid manufacturing of high-performance metal parts. To improve the mechanical behavior of an LPBFed AlSi10Mg alloy, the influence of nano-Si3N4 reinforcement on densification behavior, microstructure, and tensile property of AlSi10Mg was studied. The experimental results show that 97% relative density of the 3 vol.% nano-Si3N4/AlSi10Mg composite was achieved via optimization of the LPBF process. With the increase in the nano-Si3N4 content, the tensile strength and the yield strength of the composite steadily increase as per the Orowan strengthening mechanism while the elongation decreases. In addition, nano-Si3N4 reinforcement reduces the width of the coarse cell structure region and the thermal influence region of the AlSi10Mg matrix. After annealing, the tensile strength of the nano-Si3N4/AlSi10Mg composite decreases and the elongation increases significantly.

A porosity-based flood inundation modelling approach for enabling faster large scale simulations

Floods are among the most devastating natural hazards in the world. With climate change and growing urbanisation, floods are expected to become more frequent and severe in the future. Hydrodynamic models are powerful tools for flood hazard assessment but face numerous challenges, especially when operating at a large scale. The downside of discretising an area using a fine mesh yielding more accurate results, is the expensive computational cost of simulations. Moreover, critical input information such as bathymetry (i.e, riverbed geometry) are required but cannot be easily collected by field measurements or remote sensing observations. During the past few years, the development of sub grid models has gained a growing interest as these enable faster simulations by using coarser cells and, at the same time, preserve small-scale topography variations within the cell. In this study, we propose and evaluate a modelling framework based on the shallow water 2D model with depth-dependent porosity enabling to represent floodplain and riverbed topography through porosity functions. To enable a careful and meaningful evaluation of the model, we set up a 2D classical model and use it as a benchmark. We also exploit ground truth data and remote sensing derived flood inundation maps to evaluate the proposed modelling framework and use as test cases the 2007 and 2012 flood events of the river Severn. Our empirical results demonstrate a high performance and low computational cost of the proposed model for fast flood simulations at a large scale.

Knowledge-based classification of fine-grained immune cell types in single-cell RNA-Seq data.

Single-cell RNA sequencing (scRNA-Seq) is an emerging strategy for characterizing immune cell populations. Compared to flow or mass cytometry, scRNA-Seq could potentially identify cell types and activation states that lack precise cell surface markers. However, scRNA-Seq is currently limited due to the need to manually classify each immune cell from its transcriptional profile. While recently developed algorithms accurately annotate coarse cell types (e.g. T cells versus macrophages), making fine distinctions (e.g. CD8+ effector memory T cells) remains a difficult challenge. To address this, we developed a machine learning classifier called ImmClassifier that leverages a hierarchical ontology of cell type. We demonstrate that its predictions are highly concordant with flow-based markers from CITE-seq and outperforms other tools (+15% recall, +14% precision) in distinguishing fine-grained cell types with comparable performance on coarse ones. Thus, ImmClassifier can be used to explore more deeply the heterogeneity of the immune system in scRNA-Seq experiments.

Retinal Ependymoma: A Rare Extra-Axial Presentation

Abstract Introduction/Objective Ependymomas are well-demarcated and slow-growing neuroepithelial neoplasms that comprise 3–9% of primary CNS tumors. The vast majority of ependymomas arise either intracranially, mostly in children, or in the spinal cord and are associated with ependymal lining. Histologic hallmarks are perivascular pseudorosettes, ependymal rosettes and alternating zones of nuclear crowding and nuclear free zones composed of coarse cell processes. In high grade ependymomas increased mitoses, necrosis and nuclear pleomorphism may be seen. Methods We present the case of a 63-year-old woman in with a past medical history of retinopathy of prematurity, glaucoma and right eye enucleation. She presented with a painful blind left eye refractory to medical treatment and subsequently underwent left eye enucleation. Results On histologic examination, an incidental retinal ependymoma was identified. The neoplastic cells were fusiform and had long coarsely fibrillar cell processes. Characteristic periodicity of nuclear crowding and scarcity was observed. In places, neoplastic cell processes extended radially to delicate and sometimes hyalinized blood vessels, forming so-called perivascular pseudorosettes. Definite ependymal rosettes were not recognized in the examined sections. Stigmata of chronicity were found, such as ischemic type necrosis, blood vessels with dystrophic calcifications and foci of ossification replete with fibroadipose tissue in the marrow spaces. The neoplastic cells labeled with antibodies against glial fibrillary acidic protein (GFAP), thus confirming glial lineage in the neoplastic cell. Additionally, there was scattered intracytoplasmic expression of epithelial membrane antigen (EMA) in a dot-like or circular pattern. The latter is well described in ependymomas and often used to support the diagnosis. Conclusion Ependymomas rarely occur at extracranial sites, such as the chest, abdomen and pelvis. We are presenting the fourth case of retinal ependymoma reported in literature. The tumor had classical immunomorphologic findings.