1,721,047 research outputs found
GPU-accelerated depth map generation for X-ray simulations of complex CAD geometries
No full textInteractive x-ray simulations of complex computer-aided design (CAD) models can provide valuable insights for better interpretation of the defect signatures such as porosity from x-ray CT images. Generating the depth map along a particular direction for the given CAD geometry is the most compute-intensive step in x-ray simulations. We have developed a GPU-accelerated method for real-time generation of depth maps of complex CAD geometries. We preprocess complex components designed using commercial CAD systems using a custom CAD module and convert them into a fine user-defined surface tessellation. Our CAD module can be used by different simulators as well as handle complex geometries, including those that arise from complex castings and composite structures. We then make use of a parallel algorithm that runs on a graphics processing unit (GPU) to convert the finely-tessellated CAD model to a voxelized representation. The voxelized representation can enable heterogeneous modeling of the volume enclosed by the CAD model by assigning heterogeneous material properties in specific regions. The depth maps are generated from this voxelized representation with the help of a GPU-accelerated ray-casting algorithm. The GPU-accelerated ray-casting method enables interactive (> 60 frames-per-second) generation of the depth maps of complex CAD geometries. This enables arbitrarily rotation and slicing of the CAD model, leading to better interpretation of the x-ray images by the user. In addition, the depth maps can be used to aid directly in CT reconstruction algorithms.This proceeding may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This proceeding appeared in Grandin, Robert J., Gavin Young, Stephen D. Holland, and Adarsh Krishnamurthy. "GPU-accelerated depth map generation for X-ray simulations of complex CAD geometries." In AIP Conference Proceedings, vol. 1949, no. 1, p. 190002. AIP Publishing LLC, 2018, and may be found at
DOI: 10.1063/1.5031636.
Copyright 2018 Author(s).
Posted with permission
Incorporation of composite defects from ultrasonic NDE into CAD and FE models
No full textFiber-reinforced composites are widely used in aerospace industry due to their combined properties of high strength and low weight. However, owing to their complex structure, it is difficult to assess the impact of manufacturing defects and service damage on their residual life. While, ultrasonic testing (UT) is the preferred NDE method to identify the presence of defects in composites, there are no reasonable ways to model the damage and evaluate the structural integrity of composites. We have developed an automated framework to incorporate flaws and known composite damage automatically into a finite element analysis (FEA) model of composites, ultimately aiding in accessing the residual life of composites and make informed decisions regarding repairs. The framework can be used to generate a layer-by-layer 3D structural CAD model of the composite laminates replicating their manufacturing process. Outlines of structural defects, such as delaminations, are automatically detected from UT of the laminate and are incorporated into the CAD model between the appropriate layers. In addition, the framework allows for direct structural analysis of the resulting 3D CAD models with defects by automatically applying the appropriate boundary conditions. In this paper, we show a working proof-of-concept for the composite model builder with capabilities of incorporating delaminations between laminate layers and automatically preparing the CAD model for structural analysis using a FEA software.This proceeding may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This proceeding appeared in Bingol, Onur Rauf, Bryan Schiefelbein, Robert J. Grandin, Stephen D. Holland, and Adarsh Krishnamurthy. "Incorporation of composite defects from ultrasonic NDE into CAD and FE models." AIP Conference Proceedings 1806, no. 1, (2017): 150004. , and may be found at DOI: 10.1063/1.4974728. Posted with permission.</p
Automated Construction and Insertion of Layer-by-Layer Finite Element Sub-Models of Damaged Composites
No full textFinite element models of composite structures are generally shell-based and modeled at the laminate level. More detailed layer-by-layer lamina-level models are sometimes needed for representing joints or for modeling defect growth processes. We describe a method and toolkit for automating the creation and insertion of layerby-layer finite element sub-models of composite laminate. We focus in particular on representing damage captured from nondestructive evaluation (NDE) measurements. The method is based on scripting existing simulation and solid modeling tools (ABAQUS and ACIS). It works even on complicated, curved CAD models. The submodel location is identified by the intersection of a cylinder with the structure. We then execute a series of instructions to generate a new shell with the submodel region removed, generate the layer-by-layer submodel, and bond together the layers and models with desired boundary conditions and defects. The instructions represent the steps of lamination and bonding for creating the composite. The output of the method includes CAD models of the new shell and each lamina within the submodel, and a Python script for ABAQUS that will load the CAD models, bond them together, and apply the specified boundary conditions.This proceeding appeared in Holland, Stephen D., Adarsh Krishnamurthy, Onur Bingol, and Robert Grandin. "Automated Construction and Insertion of Layer-by-Layer Finite Element Sub-Models of Damaged Composites." In Proceedings of the American Society for Composites—Thirty-third Technical Conference (2018). Lancaster, PA: DEStech Publications, Inc. DOI: 10.12783/asc33/26002. Posted with permission.</p
Multi-level voxel representation for GPU-accelerated solid modeling
No full textSolid models traditionally use boundary-representation (B-rep) to define and model their geometry. However, performing modeling operations such as Boolean operations or computing point membership classification with B-rep is computationally intensive, since B-reps do not have volumetric information. Voxelized representations, on the other hand, can be extended to include volumetric information of solid models. However, in order to use voxelized representations for solid modeling, efficient methods for voxelizing a B-rep solid model needs to be developed. In this thesis, GPU-accelerated methods are presented for creating and rendering a multi-level voxelization of a solid model that can be used along with the existing B-rep for modeling operations. Two GPU-accelerated algorithms are described; one for creating a multi-level voxelization given a B-rep of a solid model and another for ray casting to render the multi-level voxelization of the solid model. Compact and flat data structures are described that can be used to store the multi-level voxelization data and can be efficiently retrieved in parallel using GPU-algorithms for rendering and modeling operations. The GPU-accelerated multi-level voxelization method can generate models with an effective voxel count of up to 8 billion voxels. In addition, the GPU voxelization algorithm is more than 40x faster than the CPU implementation in generating the voxelization. Finally, we outline a few applications for the hybrid representation, which include fast point-membership classification, volume computation, and collision detection.</p
Animated rendering of cardiac model simulations
No full textHeart disease has been the leading cause of death both in the world and the United States
in the past decade. Computational cardiac modeling and simulation, especially patient-specific
cardiac modeling has been recognized as one of the best ways to improve diagnosis of heart
disease by providing insights in individual disease characteristics that cannot be obtained by
other means. However presenting the results of cardiac simulations to cardiologists in an
interactive manner can considerably improve the utility of cardiac models in understanding
the heart function. In this work, we have developed virtual reality and animated volume
rendering techniques to render the results of cardiac simulations. We have developed a GPU
accelerated algorithm that produces time varying voxelized representation of the quantities of
interest in a cardiac model, which can then be interactively rendered in real time. We voxelize
the different time frames of the analysis model and transfer the time-varying data to the GPU
memory using a flat data structure. This technique allows us to visualize and interact with
animation in real time. As a proof-of-concept, we test our method on interactively rendering
the simulation results of cardiac biomechanics simulations. We also present the timing results
on post-processing and rendering two different cardiac IGA at different resolutions. We achieve
an interactive frame rate of over 50 fps for all test cases.</p
Deep learning and GPU-accelerated algorithms for computer-aided engineering
No full textComputer-Aided Engineering (CAE) is necessary for fast and efficient product development in design and manufacturing. Historically, CAE played a vital role in the drastic improvement of production time and production cost over the past two to three decades. While it used to take several months to years to realize a product earlier, the same process takes only a few hours to days now, due to the availability of state-of-the-art CAE techniques. This thesis is a contribution to the CAE ecosystem with the motivation to integrate advanced ideas from machine learning and computational sciences with current CAE tools to improve the product development cycle. These ideas can be broadly classified into (i) data-driven/machine learning-based approaches, (ii) GPU acceleration for massively parallelizable tasks. These ideas have been adopted extensively and have become prevalent over the past five to eight years, with a substantial focus on other cyber-physical systems in the aerospace, automotive, agriculture, healthcare, and transportation sectors. However, applications to the manufacturing sector have not seen as much advancement compared to the other areas. In this work, we focus on our contributions to CAE, which is an integral part of the manufacturing cyber-physical system.
Over the past decade, transition to a new economy driven by automation and revolutionary changes in manufacturing technologies has enabled highly sophisticated, creative, and customizable products to be manufactured on demand by flexible robotic systems. Consequently, the demand for designing and customizing products for each user has grown exponentially. However, end-users who wish to customize their products or designs often do not have sufficient knowledge, experience, or expertise about manufacturing technologies and computational methods to analyze the design. This limitation requires CAE systems to be intelligent in making decisions with fewer interventions from the end-users (often termed as Industry 4.0). While several cyber-physical systems have embedded advanced data-driven tools such as deep learning and reinforcement learning into their workflow, their usage in the manufacturing systems is still very sparse. This dissertation is an attempt to fill this lacuna.
Another critical idea explored for integration to the state-of-the-art CAE tools is the development of algorithms accelerated with graphical processing units (GPUs). Almost a decade ago, GPUs were meant for accelerating the rendering pipeline of an application by computing a the geometric transformations of triangles and rendering them interactively (at a rate of more than 30 frames per second). However, the trend to use GPUs for general-purpose computations (GPGPUs) has been catching on. Many CAE applications have started to embrace the use of GPUs for general computations since they can perform certain typical computations in less than a minute, while the traditional computational methods take minutes/hours to do the same. Such low latency provides users with the opportunity to perform interactive custom designs, which are necessary for the current age of personalized products rather than mass production-based systems. In this dissertation, we leverage the speed of GPU-accelerated algorithms to accelerate data-driven CAE tools.
Specifically, we develop four CAE tools in this dissertation. First, we propose an intelligent decision-making system that can be applied in the product development process. This tool is useful in developing designs without iterative design reviews involving design or manufacturing engineers. Next, we develop a tool for performing quick design analysis to validate the designs. We then develop a tool for design exploration in situations where the computation of physics is trivial using CAE, but the inverse (finding the design satisfying the desired physical phenomenon) is an intractable problem. Third, we develop a tool for optimizing and obtaining designs with creativity while maintaining the design intent. Finally, we present frameworks developed for scaling deep learning to distributed manufacturing cyber-physical systems.</p
A framework for geometric modeling and structural analysis of composite laminates
No full textLaminated fiber-reinforced polymer (FRP) composites show considerable promise in structural applications due to their good combination of low weight and high strength. However, the manufacturing costs of laminated composites is significantly higher than their metallic counterparts. As a consequence, estimating the residual life of composites becomes critical, and can enable reusability in applications that demand lower mechanical strength requirements. One of the major factors affecting the residual life of the laminated composites is the defects introduced during manufacturing or in service. A common way of determining defects in the composite laminates is using non-destructive evaluation (NDE) techniques. In this study, a framework for modeling and structural analysis of composite laminates is presented. The framework follows the laminate manufacturing process and incorporates structural elements, such as stiffeners, as well as defects, such as delaminations, determined using NDE techniques. Each layer composing the laminate is modeled separately and combined to generate the final laminate. The layer combination process is called bonding and involves computation of boundary conditions for the constitutional model being selected for the analysis. Then, the final laminate model and the computed boundary conditions are used during the structural analysis. The initial framework used commercial off-the-shelf (COTS) software, i.e. 3D ACIS Modeler for 3-dimensional modeling and SIMULIA Abaqus for structural analysis via finite element modeling. The framework was then extended to use the NURBS library, NURBS-Python, and the isogeometric analysis library, gIGA, which were developed as a part of this study and released as free and open-source software on GitHub. Using NURBS for modeling and isogeometric analysis for structural analysis provide several advantages, such as directly operating on the exact geometry, and therefore; achieving better estimations on interlaminar and intralaminar stresses and strains, which has significant importance in determining the residual life of the composite laminates.</p
A cybermanufacturing framework incorporating deep learning and multi-resolution voxel representations
No full textCybermanufacturing (CM) is a modern concept involving predictive analytic operations and information technology to aid the manufacturing industry in better decision making for design and manufacturing processes. This thesis presents a data-driven intelligent Cybermanufacturing framework for the effortless design and manufacturing of a product.
While traditional manufacturing systems are iterative and especially require skilled operators in the process, CM systems alleviate this issue by making intelligent predictions without specialists' involvement. CM systems operate with a network and data-rich environment involving interaction within and between virtual and physical spaces resulting in an effective decision support system. The broad objective of this research is to define and establish a framework of a cyber-physical system consisting of such virtual and physical systems to confront various departments of a manufacturing process.
The first stage in most of the iterative manufacturing processes is a product design that is compliant with certain design specifications and requirements. However, this is not a one-stop solution; to realize the final design, a product goes through multiple iterations between design and manufacturing stages to be compliant with the existing manufacturing paradigm. To tackle this issue, we have developed data-driven decision support for an intelligent design for manufacturing (DFM) framework using a volumetric representation (voxels) of 3D CAD models and deep neural networks to make high-quality predictions of the manufacturability of a part or product without requiring domain expertise of the user. We have developed a manufacturing process planning framework that detects such features irrespective of its size by hierarchically representing 3D CAD models as volumes on multiple scale levels (multi-level voxels) and facilitating scale-variant feature learning through the implementation of a multi-level Deep Neural Network to make decisions from hierarchical data.
Along with virtual decision support systems for design and manufacturing, CM systems also involve actual manufacturing in the physical space using machines and robotic environments. We have developed an automated manufacturing module that includes an algorithm for direct 3D printing from voxels and optimization based robust reinforcement algorithm.</p
Immersogeometric analysis with point cloud geometry towards practical applications
No full textRecently, immersogeometric analysis (IMGA) was successfully applied to simulate compressibleand incompressible fluid flows over CAD models represented using triangles, non-uniform rational B-splines (NURBS), and analytic surfaces. However, performing flow analysis over real-life objects requires CAD model reconstruction, which can be as tedious as the mesh generation process itself. In a point cloud geometry, the object is represented as an unstructured collection of points. Point cloud representation has proliferated as a form of acquiring geometric information in digital format using LIDAR scanners, optical scanners, or other passive methods like multi-view stereo images. In this work, we perform IMGA directly on point cloud representation of geometry, thus enabling flow analysis over as-manufactured components. Due to the absence of topological information in a point cloud, there are no guarantees that the geometric representation is watertight, which makes performing inside-outside tests on the background mesh challenging. To address this, we first develop methods for generating topological properties on a point cloud and compute inside- outside information directly from the resulting topology. Then, validations are performed for these geometric estimation methods, as well as for point cloud IMGA (PC-IMGA) incompressible flow results. We finally demonstrate additional features and scalability of our approach by performing PC-IMGA on large construction machinery represented by a dense cloud of more than 12 million points, along with our other PC-IMGA developments, including weak thermal boundary conditions and transient boundaries.</p
A framework for multi-physics simulations using four-chamber cardiac models
No full textComputational cardiac models have been extensively used to study different cardiac biomechanics; specifically, finite-element analysis has been one of the tools used to study the cardiac wall's internal stresses and strains during the cardiac cycle. Cubic-Hermite finite element meshes have been used for simulating cardiac biomechanics due to their convergence characteristics and their ability to capture smooth geometries compactly- fewer elements are needed to build the cardiac geometry-compared to linear tetrahedral meshes. Such meshes have previously been used only with simple ventricular geometries with non-physiological boundary conditions due to challenges associated with creating cubic-Hermite meshes of the complex heart geometry. However, it is critical to accurately capture the different geometric characteristics of the heart and apply physiologically equivalent boundary conditions to replicate the in vivo heart motion. In this work, we created a four-chamber cardiac model utilizing cubic-Hermite elements and simulated a full cardiac cycle by coupling the 3D finite element model with a lumped circulation model. The myocardial fiber-orientations were interpolated within the mesh using the Log-Euclidean method to overcome the singularity associated with the interpolation of orthogonal matrices. Physiologically equivalent rigid body constraints were applied to the nodes along the valve plane. The accuracy of the resulting simulations was validated using open source clinical data. Based on this validated four-chamber model, we studied different disease states and complex interactions of the different boundary conditions on the cardiac function. We simulated a complete cardiac cycle of a heart with the acute myocardial infarction. We also assessed the effect of the Myocardial Infarction and Pericardium sac on the ventricular pumping ability and cardiac motion, respectively.A fluid-structure interaction (FSI) simulation on the left cardiovascular system (including the left ventricle, atria, and the aorta) was also performed using a hybrid ALE/IMGA framework. The Arbitrary Lagrangian-Eulerian method models the left chambers' moving walls. The immersogeometric analysis couples the bioprosthetic heart valves with the intraventricular and interatrial flow. The simulation results show two openings for the Mitral valve, one major and one minor, during the cardiac cycle due to the atrial kick. The reproduction of these detailed hemodynamics and structural features of the cardiac cycle demonstrates the ability of our framework to replicate the in-vivo hemodynamics of the left cardiovascular system.</p
- …
