Defect inspection is one of the most critical tasks in the industry as it can reduce risks of production stops and assure quality control. In recent years, multiple industries have been adopting computer vision systems, especially based on deep learning techniques, as their main detection methods to improve efficiency, reduce risks and human resources, and enhance real-time performance. However, its adoption in the industry is still limited by the labor-intense and time-consuming process of collecting high-quality custom training datasets. At the same time, many industries have access to the CAD models of the components they want to detect or classify as part of the design process. Taking this into account, in the present work, we analyze the performance of various image classification models to visually detect defects in production. Our method systematically generates synthetic datasets from CAD models using Blender to train neural networks under different settings. The proposed method shows that image classification models benefit from a diversity of the range of defect values during training but struggle to identify low-resolution defects, even for state-of-the-art architectures like Vision Transformer and ConvNext or SqueezeNet, which proved to have comparable performance to these networks. Similarly, adding background, texture, and camera pose to training examples provides more contextual information to image classification models but does not necessarily help them detect the defects accurately. Finally, we observed that using a unique tolerance value for all flange pipe sizes can negatively impact the detection accuracy because, for larger pipe flanges, minor defects are not as perceptible as for small flanges.

This report is the documentation of a master’s thesis which was conducted at Scania CV AB in Södertälje. In this study, the benefits, challenges and conditions of using parametric CAD-models for aiding CFD-simulations and performance-optimization in the product development within internal combustion engines have been assessed. The goal of the thesis included developing and proposing a methodology for design engineers at Scania which will aid them in creating parametric CAD-models which are robust, flexible, comprehensible and intelligent. The study also included assessing the benefits and pre-requisites of such methodology with both practical and theoretical approaches. The ultimate goal of this entire study was to create value for the organization by reducing lead-time in the design process while promoting the production of high quality products. A case-based approach was applied in the study in which modeling strategies resulting from practical investigations and analyzing existing methodologies were implemented on a CAD-model representing the gas-volume of a turbine volute. The chosen strategies were evaluated and subsequently documented as a part of the methodology or discarded depending on its support for the parameterization. The final methodology itself was evaluated based on the quality of the parameterized CAD-model, the time required to create the model and its compatibility to the present design process at Scania CV AB. Finally the methodology was discussed with respect to the different evaluations, and the defined research questions were discussed and answered. The results of the thesis revealed that if parametric CAD-models are made in a structured, standardized and conscious manner, they are able to be highly robust and flexible which gives the models the ability to assume a big set of different forms. The methodology is recommended to be tested in a pilot project and be implemented through internal courses at the company. It was concluded that a methodology which aids the design engineers in creating parametric CAD-models will be the key towards implementing parametric CAD-models in the company and also enabling the many benefits of parameterization, which includes reduced lead-time, enhanced component performance, increased knowledge about the component, and promotion of collaboration among engineers. It was also concluded that parametric models are best suited when the existing design freedom is big and when the time permits performance analyses via optimizations, while challenges include ensuring that the model is parameterized correctly with respect to the CFD-engineers’ wishes while taking into account the requirements from other disciplines. Therefore it is very important to establish a communication between the different engineers. Ultimately, when parametric models are established in the organization, they are recommended to be implemented eventually in both short-term and long-term projects within Scania for its beneficial properties.

This thesis describes the design and implementation of an application that converts CAD models into polygon models. When going from CAD models to 3D polygon models a conversion of the file type has to be performed. XperDI uses these polygon models in their tool, called sales configurator, to create a photo realistic environment to be able to have a look at the end product before it is manufactured. Existing tools are difficult to use and is missing features that is important for the Sales Configurator. The purpose of this thesis is to create a proof of concept application that converts CAD models into 3D polygon models. This new lightweight application is a simpler alternative to convert CAD models into polygon models and offers features needed for the intended use of these models, that the alternative products do not offer.

[pt] Modelos CAD 3D são ferramentas utilizadas na indústria para planejamento e simulações antes da construção ou realização de tarefas. Em muitos casos, como por exemplo na indústria de óleo e gás, esses modelos podem ser massivos, ou seja, possuem informações detalhadas em larga escala no intuito de que sejam fontes de informações precisas. Para obtenção de navegação interativa nesses modelos é necessária uma combinação de hardware e software adequados. Mesmo hoje com GPUs mais modernas, a renderização direta desses modelos não é eficiente, sendo necessárias abordagens clássicas como descarte de objetos não visíveis e LOD antes de enviar os dados à GPU. Logo, para renderização em tempo real de modelos CAD massivos são necessários algoritmos e estruturas de dados escaláveis para processamento da cena de forma eficiente. O trabalho dessa tese propõe o MCAD (Massive Computer-Aided Design) Shape grammar, uma gramática expansiva que gera objetos para criar cenas 3D de modelos massivos de forma procedimental. Nos últimos anos, modelagem procedimental tem ganhado atenção para criar cenas 3D rapidamente utilizando uma representação compacta, que armazena regras de geração ao invés de representação explícita da cena. MCAD Shape grammar explora repetições e padrões presentes em modelos massivos para renderização de cenas, reduzindo o consumo de memória e processando a cena procedimentalmente de forma eficiente. Convertemos modelos reais de refinarias em MCAD Shape grammar e implementamos um renderizador para os mesmos. Os resultados mostraram que esta solução é escalável com alto desempenho, além de ser a primeira vez que modelagem procedimental é utilizada nesse domínio. / [en] 3D CAD models are tools used in the industry for planning and simulations before construction or completion of tasks. In many cases, such as in the oil and gas industry, these models can be massive, that is, they have large-scale detailed information in order to be sources of accurate information. Interactive navigation in these models requires a combination of appropriate hardware and software. Even nowadays with modern GPUs, the direct rendering of these models is not efficient, requiring classic approaches such as culling non-visible objects and LOD before sending the data to the GPU. Therefore, for real-time rendering of massive CAD models, we need scalable algorithms and data structures to efficiently process the scene. The work of this thesis proposes MCAD (Massive Computer-Aided Design) Shape grammar, an expansive grammar that procedurally generates objects to create 3D scenes of massive models. In recent years procedural modeling has drawn attention for quickly creating 3D scenes using a compact representation, which stores generation rules rather than explicit representation of the scene. MCAD Shape grammar explores repetitions and patterns present in massive models for rendering scenes, reducing memory footprint and procedurally processing the scene efficiently. We converted real refinery models into MCAD Shape grammar and implemented a renderer for them. Results showed that our solution is scalable with high performance, also it is the first time that procedural modeling is used in this domain.

With constantly changing regulations on emissions, heavy commercial vehicles manufacturershave to adapt for their products to preserve their quality while meetingthese new requirements. Over the past decades, noise emissions have become a greatconcern and new stricter laws demand companies to decrease their vehicle pass-bynoise target values.To address the requirements from different disciplines, Scania follows a simulationdriven design process to develop new concept models EATS. The collaboration amongengineers from different fields is thereby necessary in order to obtain higher performancesilencers. However, the pre-processing step in terms of acoustic simulationsis time-consuming, which can slow the concept development process.In this thesis, a new method was introduced to automate the pre-processing of silenceracoustic models and allow for design optimisation based on acoustic performanceresults. A common Scania product study case was provided to several theseswithin the NXD organisation. The collaboration among the master thesis workersaimed to demonstrate the benefits of KBE and MDO and how they can be integratedwithin Scania’s current concept development and product introduction processes.The performed work was divided in the following steps: data collection, methoddevelopment and concluding work. The first step consisted in gathering sufficientknowledge by conducting a thorough literature review and interviews. Then, an initialmethod was formulated and tested on a simplified silencer model. Once approvedand verified, the method was applied to the study case EATS.The study case showed that a complex product can have its acoustic pre-processingstep automated by ensuring a good connectivity among the required software anda correct denomination of the geometrical objects involved in the simulations. Themethod investigated how morphological optimisations can be performed at bothglobal and local levels to enhance the transmission loss of a silencer. Besides optimisingthe acoustic performance of the models, the method allowed the identificationof correlations and inter-dependencies among their design variables and ouput parameters.

[pt] Modelos CAD 3D têm desempenhado um importante papel no gerenciamento de projetos de engenharia. Em vários desses modelos é perceptível a presença de diversos objetos com representações implícitas sendo representados por malhas triangulares. Ainda que apropriada para o rendering, a utilização de malhas triangulares traz consigo algumas desvantagens, como a ambiguidade em objetos pouco discretizados. A engenharia reversa visa a reconstrução dessa representação discreta em sua representação contínua original. Neste trabalho, propomos uma metodologia para a reconstrução de geometrias em modelos CAD utilizando Support Vector Machines e Descritores de Forma. / [en] 3D CAD Models have played an important role in engineering projects management. It is noticeable in many of these files the presence of several objects with implicit representation that end up being represented as triangular meshes. Although suitable for rendering, the triangular mesh representation brings some drawbacks, such as the ambiguity in objects with low discretization rate. The reverse engineering aims to reconstruct this discrete representation to its original continuous representation. In this work, we propose a novel methodology for geometry reconstruction in CAD models using Support Vector Machines and Shape Descriptors.

This thesis addresses the problem of constructing 3D CAD models from 2D engineering drawings of the object. Procedures for each step in the popular bottom-up approach are described. In each of the steps in this approach, the capability is extended to handle classes of parts not handled by present art. These include auxiliary views, parts with curved entities that are inclined to all viewing directions, and views that result in non-manifold edges in the surface model. In addition this thesis also presents procedures to extract features from the solid model constructed so that the 3D model can be manipulated and modified within current CAD systems. In order to enable extracting the construction history, detecting symmetry in the part is essential. This thesis presents an algorithm to detect symmetry (both global and partial) in the part given the solid model of the part. Working of the procedures is illustrated using benchmark parts from the literature. The thesis concludes with a discussion on scope for further work.

This master thesis has been executed at Scania CV AB in Södertälje, Sweden. Scania is a manufacturer of heavy transport solutions, an industry which is changing rapidly in order to meet stricter regulations, ensuring a sustainable future. Continuous product improvements and new technologies are required to increase performance and to meet markets requirements. By implementing design optimization in the design process it enables the potential of supporting design exploration, which is beneficial when products with high performance are developed. The purpose was to show the potential of design optimization supported by simulation driven design as a tool in the development process. To examine an alternative way of working for design engineers, elaborating more competitive products in terms of economical and performance aspects. Furthermore, to minimize time and iterations between divisions by developing better initial concept proposals. The alternative working method was developed iteratively in parallel with a case study. The case study was a suction strainer and were used for method improvements and validation, as well as decision basis for the included sub-steps. The working method for implementing design optimization and simulation driven design ended up with a procedure consisted of three main phases, concept generation, detail design and verification. In the concept generation phase topology optimization was used, which turned out to be a beneficial method to find optimized solutions with few inputs. The detail design phase consisted of a parameterized CAD model of the concept which then was shape optimized. The shape optimization enabled design exploration of the concept which generated valuable findings to the product development. Lastly the optimized design was verified with more thorough methods, in this case verification with FE-experts. The working method was tested and verified on the case study component, this resulted in valuable knowledge for future designs for similar components. The optimized component resulted in a performance increase where the weight was decrease by 54% compared with a reference product.

Interactive applications with 3D graphics are used everyday in a lot of different fields, such as games, teaching, learning environments and virtual reality. The scenarios showed in interactive applications usually tend to present detailed worlds and characters, being the most realistic as possible. Detailed 3D models require a lot of geometric complexity. But not always the available graphics hardware can handle and manage all this geometry maintaining a realistic frame rate. Simplification methods attempt to solve this problem, by generating simplified versions of the original 3D models. These simplified models present less geometry than the original ones. This simplification has to be done with a reasonable criterion in order to maintain as possible the appearance of the original models. But the geometry is not the only important factor in 3D models. They are also composed of additional attributes that are important for the final aspect of the models for the viewer. In the literature we can find a lot of work presented about simplification. However, there are still several points without an efficient solution. Therefore, this thesis focuses on simplification techniques for 3D models usually used in interactive applications.

Aus der Einleitung: "Im Modell und Formenbau des Maschinenbaus beschreibt Reverse Engineering den Prozess der 3D-Erfassung eines Objektes, die Aufbereitung der Digitalisierungsdaten zu CAD-Modellen und die weitere Nutzung dieser Daten in einer CAD/CAM-Umgebung. Ziel dieser Arbeiten ist es dann weiterführend, physische Objekte durch CNC-Fräsen oder mittels Generativer Fertigungsverfahren herzustellen. Die Maßkontrolle der gefertigten Produkte gegenüber dem CAD ist ebenfalls eine Aufgabestellung des Reverse Engineering (Schöne 2009, Wang 2011)."