B-rep Model Research Articles

Relational descriptors for retrieving design features in a B-rep model using the similarity-based retrieval approach

Design features refer to local shapes or regions within a part that perform specific functions such as fastening and force transmission. These design features must be identified from product design results to conduct design verification, manufacturing evaluation, and process planning. Design features are formed by combining various form features, which poses a challenge when using existing methods to retrieve individual features. Therefore, this study introduced a relational descriptor that describes the relational characteristics between topological elements to retrieve design features in boundary representation (B-rep) models. In addition, a method to retrieve design features by combining the relational descriptor with shape descriptors was proposed. Experiments were performed to identify specific design features to validate the proposed method. The experimental results successfully retrieved all the design features included in the B-rep model.

ECAD-Net: Editable Parametric CAD Models Reconstruction from Dumb B-Rep Models Using Deep Neural Networks

This paper introduces a novel framework capable of reconstructing editable parametric CAD models from dumb B-Rep models. First, each B-Rep model is represented with a network-friendly formalism based on UV-graph, which is then used as input of eCAD-Net, the new deep neural network-based algorithm that predicts feature-based CAD modeling sequences from the graph. Then, the sequences are scaled and fine-tuned using a feature matching algorithm that retrieves the exact parameter values from the input dumb CAD model. The output sequences are then converted in a series of CAD modeling operations to create an editable parametric CAD model in any CAD modeler. A cleaned dataset is used to learn and validate the proposed approach, and is provided with the article. The experimental results show that our approach outperforms existing methods on such reconstruction tasks, and it outputs editable parametric CAD models compatible with existing CAD modelers and ready for use in downstream engineering applications.

BrepGen: A B-rep Generative Diffusion Model with Structured Latent Geometry

This paper presents BrepGen , a diffusion-based generative approach that directly outputs a Boundary representation (B-rep) Computer-Aided Design (CAD) model. BrepGen represents a B-rep model as a novel structured latent geometry in a hierarchical tree. With the root node representing a whole CAD solid, each element of a B-rep model (i.e., a face, an edge, or a vertex) progressively turns into a child-node from top to bottom. B-rep geometry information goes into the nodes as the global bounding box of each primitive along with a latent code describing the local geometric shape. The B-rep topology information is implicitly represented by node duplication. When two faces share an edge, the edge curve will appear twice in the tree, and a T-junction vertex with three incident edges appears six times in the tree with identical node features. Starting from the root and progressing to the leaf, BrepGen employs Transformer-based diffusion models to sequentially denoise node features while duplicated nodes are detected and merged, recovering the B-Rep topology information. Extensive experiments show that BrepGen advances the task of CAD B-rep generation, surpassing existing methods on various benchmarks. Results on our newly collected furniture dataset further showcase its exceptional capability in generating complicated geometry. While previous methods were limited to generating simple prismatic shapes, BrepGen incorporates free-form and doubly-curved surfaces for the first time. Additional applications of BrepGen include CAD autocomplete and design interpolation. The code, pretrained models, and dataset are available at https://github.com/samxuxiang/BrepGen.

BrepMFR: Enhancing machining feature recognition in B-rep models through deep learning and domain adaptation

Feature Recognition (FR) plays a crucial role in modern digital manufacturing, serving as a key technology for integrating Computer-Aided Design (CAD), Computer-Aided Process Planning (CAPP) and Computer-Aided Manufacturing (CAM) systems. The emergence of deep learning methods in recent years offers a new approach to address challenges in recognizing highly intersecting features with complex geometric shapes. However, due to the high cost of labeling real CAD models, neural networks are usually trained on computer-synthesized datasets, resulting in noticeable performance degradation when applied to real-world CAD models. Therefore, we propose a novel deep learning network, BrepMFR, designed for Machining Feature Recognition (MFR) from Boundary Representation (B-rep) models. We transform the original B-rep model into a graph representation as network-friendly input, incorporating local geometric shape and global topological relationships. Leveraging a graph neural network based on Transformer architecture and graph attention mechanism, we extract the feature representation of high-level semantic information to achieve machining feature recognition. Additionally, employing a two-step training strategy under a transfer learning framework, we enhance BrepMFR's generalization ability by adapting synthetic training data to real CAD data. Furthermore, we establish a large-scale synthetic CAD model dataset inclusive of 24 typical machining features, showcasing diversity in geometry that closely mirrors real-world mechanical engineering scenarios. Extensive experiments across various datasets demonstrate that BrepMFR achieves state-of-the-art machining feature recognition accuracy and performs effectively on CAD models of real-world mechanical parts.

Graph Neural Network-Based Finite Element Feature Recognition from B-rep Model

Graph Neural Network-Based Finite Element Feature Recognition from B-rep Model

Brep2Seq: a dataset and hierarchical deep learning network for reconstruction and generation of computer-aided design models

Abstract Three-dimensional (3D) reconstruction is a significant research topic in the field of computer-aided design (CAD), which is used to recover editable CAD models from original shapes, including point clouds, voxels, meshes, and boundary representations (B-rep). Recently, there has been considerable research interest in deep model generation due to the increasing potential of deep learning methods. To address the challenges of 3D reconstruction and generation, we propose Brep2Seq, a novel deep neural network designed to transform the B-rep model into a sequence of editable parametrized feature-based modeling operations comprising principal primitives and detailed features. Brep2Seq employs an encoder-decoder architecture based on the transformer, leveraging geometry and topological information within B-rep models to extract the feature representation of the original 3D shape. Due to its hierarchical network architecture and training strategy, Brep2Seq achieved improved model reconstruction and controllable model generation by distinguishing between the primary shape and detailed features of CAD models. To train Brep2Seq, a large-scale dataset comprising 1 million CAD designs is established through an automatic geometry synthesis method. Extensive experiments on both DeepCAD and Fusion 360 datasets demonstrate the effectiveness of Brep2Seq, and show its applicability to simple mechanical components in real-world scenarios. We further apply Brep2Seq to various downstream applications, including point cloud reconstruction, model interpolation, shape constraint generation, and CAD feature recognition.

Robust numerical integration of embedded solids described in boundary representation

Embedded and immersed methods have become essential tools in computational mechanics, as they allow discretizing arbitrarily complex geometries without the need for boundary-fitted meshes. One of their main challenges is the accurate numerical integration of cut elements. Among the various integration schemes developed for this purpose, moment fitting has proven to be a powerful technique that provides highly efficient and accurate integration rules.This publication presents a framework for the robust and efficient numerical integration of embedded solids described by oriented boundary meshes using moment fitting. The developments include an intersection algorithm that aims to drastically accelerate the computation of the necessary moments while achieving high accuracy. A closed surface parameterization of each cut domain is computed to facilitate the direct application of the divergence theorem. The algorithm is subject to a single quality criterion that guarantees the accurate evaluation of boundary integrals. At the same time, it allows to disregard classical mesh criteria, such as high aspect ratios, strongly varying angles, etc., resulting in extremely fast runtimes. In addition, an existing robust flood fill-based element classification scheme is further developed to initiate filling from arbitrary seed elements and to enable parallel execution, increasing its flexibility and efficiency.The successful application of all proposed algorithms to 4948 valid and flawed STLs from the Thingi10K database (Zhou and Jacobson, 2016) demonstrates their extraordinary robustness. In all cases, the wall-clock time scales at most linearly with the number of elements in the background mesh. We show that higher-order quadrature rules on the boundary elements enable efficient computation of the moments via the divergence theorem with near-machine precision. Finally, the presented methodologies are used to perform direct FE analyses on clean and flawed B-Rep models. All proposed algorithms are publicly available in the open-source C++ framework QuESo – Quadrature for Embedded Solids (https://github.com/manuelmessmer/QuESo), where the moment fitting equations are assembled and solved.

BRepGAT: Graph neural network to segment machining feature faces in a B-rep model

Abstract In recent years, there have been many studies using artificial intelligence to recognize machining features in three-dimensional models in the computer-aided design (CAD)/computer-aided manufacturing field. Most of these studies converted the original CAD data into images, point clouds, or voxels for recognition. This led to information loss during the conversion process, resulting in decreased recognition accuracy. In this paper, we propose a novel deep learning model called the boundary representation graph attention network (BRepGAT) to segment faces in an original boundary representation (B-rep) model containing machining features. We define descriptors that represent information about the faces and edges of the B-rep model from the perspective of feature recognition. These descriptors are extracted from the B-rep model and transformed into homogeneous graph data, which are then passed to graph networks. BRepGAT recognizes machining features from the graph data input. Our experimental results using the MFCAD18++ dataset showed that BRepGAT achieved state-of-the-art recognition accuracy (99.1%). Furthermore, BRepGAT showed relatively robust performance on other datasets besides MFCAD18++.

Development of a Tetrahedral Mesh Generator Based on Constrained Delaunay Triangulation

The article describes the basic principles of operation of the finite element mesh generator SMCMGrid, which is an integral part of the SMCM software package designed to solve problems in the mechanics of composite materials and structures, which was developed at the Scientific-educational center on supercomputer modeling and software engineering of Bauman Moscow State Technical University (BMSTU SIMPLEX CENTER). The paper describes a software pipeline for constructing high-quality tetrahedral unstructured meshes for arbitrary shape B-rep models based on constrained Delaunay triangulation. We have presented the main algorithms for generating meshes on the edges and surfaces of the model, which make it possible to create the so-called simplicial complex, for which it is then possible to construct a constrained Delaunay triangulation and create the final tetrahedral mesh of the required quality. The mesh is generated in a Riemannian space with a given metric that takes into account the required characteristics of the mesh at each point of the geometric model. We gave examples of finite element meshes for various geometric models of engineering structure elements, including composite materials, and also demonstrated the adequacy of the results of solving the elasticity problem on a mesh built using the SMCMGrid generator. The results of the work led to the conclusion that the presented generator can be used to solve problems of engineering analysis and modeling.

Direct immersogeometric fluid flow and heat transfer analysis of objects represented by point clouds

Immersogeometric analysis (IMGA) is a geometrically flexible method that enables one to perform multiphysics analysis directly using complex computer-aided design (CAD) models. While the IMGA approach is well-studied and has a remarkable advantage over traditional CFD, IMGA still requires a well-defined B-rep model to represent the geometry. Obtaining such a model can sometimes be equally as challenging as creating a body-fitted mesh. To address this issue, we develop a novel IMGA approach for the simulation of incompressible and compressible flows around complex geometries represented by point clouds in this work. The point cloud representation of geometries is a direct method for digitally acquiring geometric information using LiDAR scanners, optical scanners, or other passive methods such as multi-view stereo images. The point cloud object’s geometry is represented using a set of unstructured points in the Euclidean space with (possible) orientation information in the form of surface normals. Due to the absence of topological information in the point cloud model, there are no guarantees for the geometric representation to be watertight or 2-manifold or to have consistent normals. To perform IMGA directly using point cloud geometries, we first develop a method for estimating the inside–outside information and the surface normals directly from the point cloud. We also propose a method to compute the Jacobian determinant for the surface integration (over the point cloud) necessary for the weak enforcement of Dirichlet boundary conditions. We validate these geometric estimation methods by comparing the geometric quantities computed from the point cloud with those obtained from analytical geometry and tessellated CAD models. In this work, we also develop thermal IMGA to simulate heat transfer in the presence of flow over complex geometries. The proposed framework is tested for a wide range of Reynolds and Mach numbers on benchmark problems of geometries represented by point clouds, showing the robustness and accuracy of the method. Finally, we demonstrate the applicability of our approach by performing IMGA on large industrial-scale construction machinery represented using a point cloud of more than 12 million points.

User-Defined Machining Feature Recognition Based on Semantic Reasoning for B-rep Models

Computer-Aided Design and Applications is an international journal on the applications of CAD and CAM. It publishes papers in the general domain of CAD plus in emerging fields like bio-CAD, nano-CAD, soft-CAD, garment-CAD, PLM, PDM, CAD data mining, CAD and the internet, CAD education, genetic algorithms and CAD engines. The journal is aimed at all developers and users of CAD technology to ptovide CAD solutions for various stages of design and manufacturing. The journal publishes all about Computer-Aided Design and Computer-Aided technologies.

Propagating Design Updates to Structured Analysis Meshes

Generating structured meshes is often an expensive process, limiting the use of high-fidelity numerical simulation methods for design, especially when the design is not yet fixed and design updates are likely. For hexahedral meshes generated by decomposing a B-Rep model into regions for which simple meshing strategies are known, robustly propagating design modifications to the decomposed representation and subsequent mesh is very challenging. In this paper, an approach is presented to propagate parametric update and feature changes to structured meshes. Geometric and topological modifications on the design model are identified first, enabling the equivalent modifications to be identified on the decomposition of the design model. Virtual topology operations used to generate the initial decomposition are combined with the meshing constraints of individual sub-regions to update the decomposition and associated mesh locally. The approach is demonstrated for a number of design updates.

Sketch-Based Modeling in Mechanical Engineering Design: Current Status and Opportunities

Sketch-Based Modeling (SBM) is a field of study that focuses on the automatic creation of 3D models from freehand drawings. Today, two related branches coexist: one aimed at facilitating input for 3D content creation, and the other aimed at routing engineering designs into CAD/CAM/CAE. The latter is the goal of this position paper. Early attempts concentrated on the problem of line-drawing interpretation , but efforts switched toward geometric reconstruction as the “inverse projection” problem became the most challenging step to produce 3D models from 2D line-drawings. The term SBM was popularized when sketches began to displace line-drawings as main input. In the context of engineering design, interest in SBM has somewhat decreased, as studies have shown that current SBM tools are not as usable as paper and pencil sketches, nor do they yet provide any additional value to traditional media. Furthermore, engineers feel reasonably comfortable with current Mechanical CAD (MCAD) paradigms based on parametric modeling, and fail to recognize the potential benefits of interacting with computers via sketches. However, new technological trends such as personal fabrication and the democratization of CAD and manufacturing can significantly benefit from improved SBM tools. In this paper, we conduct a meta-review of the SBM literature which we view as a combination of three elements: geometry (shape), psychology (perception), and engineering (function). We advocate for a new approach to SBM based on reformulating the weights of these elements as an approach for searching the set of intentions in sketches conveyed through cues which, when perceived, reveal regularities and features of the object. Finally, we consider quality of CAD models not just as error-free models, reusable models, or even models that convey design intent, but models that consider these interrelated aspects as a whole. B-Rep models produced by current SBM approaches are “dumb models” without parametrization or procedures required to enable reusability and ensure that the design intent of the sketch is properly conveyed. We advocate for improved analysis approaches aimed at revealing higher-level design information embedded in engineering sketches, as a critical stage to generate richer 3D MCAD models. • A review of Sketch-Based Modeling (SBM) for Engineering Design is provided. • Two different branches of SBM research coexist: one geared toward “artistic” purposes and the other governed by “engineering” principles. • The two branches must be studied separately, because of the fundamentally different goals of artistic vs. engineering sketches and the different nature of the 3D models that must be generated. • While the automatic conversion of artistic sketches into 3D models has made significant progress in recent years, the problem of producing CAD models with controllable, semantic, and parametric solid geometry has not yielded significant contributions to the scholarly literature. • Future opportunities, research directions, and recommendations on SBM approaches for engineering design are discussed.

ComplexGen

We view the reconstruction of CAD models in the boundary representation (B-Rep) as the detection of geometric primitives of different orders, i.e. , vertices, edges and surface patches, and the correspondence of primitives, which are holistically modeled as a chain complex, and show that by modeling such comprehensive structures more complete and regularized reconstructions can be achieved. We solve the complex generation problem in two steps. First, we propose a novel neural framework that consists of a sparse CNN encoder for input point cloud processing and a tri-path transformer decoder for generating geometric primitives and their mutual relationships with estimated probabilities. Second, given the probabilistic structure predicted by the neural network, we recover a definite B-Rep chain complex by solving a global optimization maximizing the likelihood under structural validness constraints and applying geometric refinements. Extensive tests on large scale CAD datasets demonstrate that the modeling of B-Rep chain complex structure enables more accurate detection for learning and more constrained reconstruction for optimization, leading to structurally more faithful and complete CAD B-Rep models than previous results.

Partial Shape Recognition of Boundary-Representation Models using Deep Neural Networks.

Shape recognition of the 3D shapes in computer-aided design (CAD) can help automate the product modelling and reduce the time to develop a new product. In this paper, in order to recognize partial shapes of 3D CAD models such as bolts and fillets, we propose a shape recognition method based on graph-neural-networks, which learns geometrical and topological information from data of annotated 3D CAD models. By using the graph of face adjacencies of the BRep model as input to the learning model, it is possible to predict recognition target directly from the BRep models. We applied this method to machine part data and verified its effectiveness. We demonstrate that our method can classify the BRep models in FabWave dataset with higher accuracy than the methods based on the point clouds.

Assessment of predictive probability models for effective mechanical design feature reuse

Abstract This research envisages an automated system to inform engineers when opportunities occur to use existing features or configurations during the development of new products. Such a system could be termed a "predictive CAD system" because it would be able to suggest feature choices that follow patterns established in existing products. The predictive CAD literature largely focuses on predicting components for assemblies using 3D solid models. In contrast, this research work focuses on feature-based predictive CAD system using B-rep models. This paper investigates the performance of predictive models that could enable the creation of such an intelligent CAD system by assessing three different methods to support inference: sequential, machine learning, or probabilistic methods using N-Grams, Neural Networks (NNs), and Bayesian Networks (BNs) as representative of these methods. After defining the functional properties that characterize a predictive design system, a generic development methodology is presented. The methodology is used to carry out a systematic assessment of the relative performance of three methods each used to predict the diameter value of the next hole and boss feature type being added during the design of a hydraulic valve body. Evaluating predictive performance providing five recommendations ( $k = 5$ ) for hole or boss features as a new design was developed, recall@k increased from around 30% to 50% and precision@k from around 50% to 70% as one to three features were added. The results indicate that the BN and NN models perform better than those using N-Grams. The practical impact of this contribution is assessed using a prototype (implemented as an extension to a commercial CAD system) by engineers whose comments defined an agenda for ongoing research in this area.

Concurrent topology optimization of shells with self-supporting infills for additive manufacturing

This paper presents a systematic design optimization approach for shells with self-supporting infills for Additive Manufacturing (AM). The design workflow includes a concurrent Topology Optimization (TO) of the shells and the infills using a density based TO approach, followed by a model reconstruction that is tailored to guarantee a precise overhang control on the output Boundary-represented (B-rep) geometry. In the TO model, a new overhang constraint is developed to ensure that the TO result satisfying an overhang angle is defined on a basis of the discrete mesh elements. The TO process is stabilized by utilizing an adaptive parameter updating strategy and a two-field based formulation is proposed to control the minimum length scale. The topology optimized result is re-constructed into a B-rep model afterwards, in which the maximum overhang span of the explicit geometry is strictly controlled to satisfy the manufacturability requirement of an actual AM process. By applying the proposed TO and the associated geometry reconstruction together, the optimized infills are not only self-supporting but also can support the optimized shells from the structural inside. Truly manufacturable shell–infill structures with desirable mechanical properties can be obtained regardless of the voxel granularity of TO. Numerical examples and AM prototypes are given to demonstrate the effectiveness and the applicability of the proposed approach.

Machining feature recognition based on deep neural networks to support tight integration with 3D CAD systems

Recently, studies applying deep learning technology to recognize the machining feature of three-dimensional (3D) computer-aided design (CAD) models are increasing. Since the direct utilization of boundary representation (B-rep) models as input data for neural networks in terms of data structure is difficult, B-rep models are generally converted into a voxel, mesh, or point cloud model and used as inputs for neural networks for the application of 3D models to deep learning. However, the model’s resolution decreases during the format conversion of 3D models, causing the loss of some features or difficulties in identifying areas of the converted model corresponding to a specific face of the B-rep model. To solve these problems, this study proposes a method enabling tight integration of a 3D CAD system with a deep neural network using feature descriptors as inputs to neural networks for recognizing machining features. Feature descriptor denotes an explicit representation of the main property items of a face. We constructed 2236 data to train and evaluate the deep neural network. Of these, 1430 were used for training the deep neural network, and 358 were used for validation. And 448 were used to evaluate the performance of the trained deep neural network. In addition, we conducted an experiment to recognize a total of 17 types (16 types of machining features and a non-feature) from the B-rep model, and the types for all 75 test cases were successfully recognized.

Analysis of Assembly Tolerance Based on Assembly Constraint Information Model

Mechanical products are composed of two or more parts. The geometric tolerance and dimensional tolerance of each feature in part will affect the assembly performance of the product, which are accumulated and propagated between assembly fit and parts. In this paper, through the secondary development of CAD software, the B-rep model of parts is obtained. The model information is decomposed and simplified based on geometric features to obtain the key information of parts in the assembly process, simplify the operation, and improve the accuracy. Through a directed graph network, the transmission model of assembly error information based on geometrical and dimensional tolerances (GD&T) on the surface of parts is established. Combined with the error transfer characteristics of different geometric surfaces and different error sources, guided by the breadth-first search algorithm and the shortest path theory, the search and establishment of a three-dimensional assembly chain are realized. Finally, the three-dimensional chain is simulated by the Monte Carlo method. The calculation results are compared with the error range obtained by the traditional method to prove the effectiveness of the method.

A robust direct modeling method for quadric B-rep models based on geometry–topology inconsistency tracking

Boundary representation (B-rep) model editing plays an essential role in computer-aided design, and has motivated the very recent direct modeling paradigm, which features intuitive push–pull manipulation of the model geometry. In mechanical design, a substantial part of B-rep models being used are quadric models (composed of linear and quadric surfaces). However, push-pulling such models is not trivial due to the possible smooth face–face connections in the models. The major issue is that, during push–pull moves, it is often desirable to preserve these connections for functional, manufacturing, or aesthetic reasons, but this could cause complex inconsistencies between the geometry and topology in the model and lead to robustness issues in updating the model. The challenge lies in effectiveness towards detecting the instants when geometry–topology inconsistencies occur during push–pull moves. This paper proposes a novel reverse detection method to solve the challenge and then, based on it, presents a robust method for push–pull direct modeling while preserving smooth connections. Case studies and comparisons have been conducted to demonstrate the effectiveness of the method.