The optimization of mechanical behavior in safety systems during crash scenarios consistently poses challenges in vehicle development. Hence, a reinforcement learning-based approach for optimizing restraint systems in frontal impacts is proposed. The trained agent, which adjusts five parameters simultaneously, is capable of minimizing loads on a seen and unseen anthropomorphic test device on the co-driver position and is thus able of transferring knowledge. A hundred times higher rate of convergence to reach a similar optimum compared to a global optimization algorithm has been achieved.

Quotation of engineer-to-order products provides substantial challenges in effectively managing engineering resources. This paper describes an approach that rationalizes this process by integrating multi-disciplinary design analysis and optimization with a new open-source library for managing engineering knowledge before and after optimization. The approach is applied and evaluated on mechanical rock excavation machines. Adapting the approach and considering the user feedback gathered can lead to an enhanced design space overview during quotation and thus more competitive product offerings.

To achieve higher functionality in mechatronic systems, the management of disturbance factors plays a crucial role. For this purpose, a method was developed in prior works to address this management via the optimisation of product structures. However, this method lacks applicability due to the complexity of one step of the method. It is the goal of this paper to present a software tool, utilizing cluster-analysis to sort components into assemblies, with which this step is supported. Additionally, the method is generally adapted to address a wider spectrum of phenomena in mechatronic systems.

In design education, technical drawing training requires a large amount of resources. The aim of this paper is to propose a concept for an artificial intelligence-based tutoring system that partly automates technical drawing education. The educational needs of the students are defined via an error analysis of 100 corrected drawing exercises and the definition of 3 error clusters with 134 different error types. Three sub-concepts with a collection of training exercises are proposed for the tutoring system to mitigate these errors. The resulting concept is validated by a survey with 29 students.

A digital twin is the mapping of a physical twin between hybrid spaces. The lifecycle of digital and physical twins occurs through the concepts of foetal, child, and adult twins. This technology can be used to assist clients and designers with real-time data. The use of digital twin technology in architectural design can be realised at various stages, from design to operation. Designers will be able to gain knowledge of the past, present, and future using this technology. This will reveal possible design scenarios. In this study, a hypothetical scenario is designed, in which designers build a building while already having a digital twin template. To do this, Building Information Modelling (BIM) is used as a reference model for digital twins, along with the fidelity levels of digital twins and the level of detail-development of BIM. When designers want to design a new project related to their predecessors, they already use the same type of digital twin-building portfolio they can use for their new design. A digital twin will help optimise the new process. Therefore, the digital twin of a building with a similar building type can be used to extract relevant data for the design process.

This paper explores the suitability of Artificial Neural Networks (ANNs) as an enabler of Design Automation in the turbomachinery industry. Specifically, the paper provides 1) a preliminary estimation of the effectiveness of ANNs to define values for design variables of reciprocating compressors (RC) and 2) a comparison of ANNs performance with traditional and more computationally demanding methods like CFD. A tailored ANN trained on a dataset composed by 350+ Baker Hughes’ RC automatically assigns values to 8 geometrical variables belonging to multiple parts of the RC in order to satisfy two target conditions linked to their thermodynamic performance. The results highlight that the ANN-assigned parameters return an optimal solution for RC also when the target values do not belong to the training dataset. Their predictive capacity for RC thermodynamic performance, with respect to CFD, are comparable (i.e. less than 2% in terms of calculated absorbed power) and the approach enables a significant gain in terms of computational time (i.e. 2 minutes vs 10 hours). Future perspectives of this work may involve the integration of this tool in an advanced DA method to lead Design Engineers (DEs) during the whole design process.

Conceptual design synthesis, which focuses on generating solution alternatives, has a significant impact on the cost and quality of the final product. The development of radically new and significantly better solutions requires the generation and exploration of a large solution space. This work deals with the conceptual design synthesis of multi-state mechanical devices (MSMD). A scheme for representing a MSMD design task is described. Empirical studies have been carried out to develop a common understanding of the MSMD design synthesis process and use this knowledge for developing a prescriptive model. In order to support the effective and efficient use of the proposed prescriptive model, a web-based computational tool is developed.

Currently, engineers need to manually analyse requirement specifications for determining parameters to create geometries in generative engineering. This analysis is time-consuming, error-prone and causes high costs. Generative engineering tools (e.g. Synera) cannot interpret natural language requirements directly. The requirements need to be formalised in a machine-readable format. AI algorithms have the potential to automatically transform natural language requirements into such a formal, machine-readable representation. In this work, a method for formalising requirements for generative engineering is developed and implemented as a prototype in Python. The method is validated in a case example using three products of an automotive engineering service provider. Requirements to be formalised are identified in the specifications of these three products, which are used as a test set to evaluate the performance of the method. The results show that requirements for generative engineering are formalised with high performance (F1 of 86.55 %). By applying the method, efforts and therefore costs for manually analysing requirements regarding parameters for generative engineering are reduced.

Over the last 20 years, finite element analysis (FEA) has become a standard analysis tool for metal joining processes. When FEA tools are combined with design of experiments (DOE) methodologies, academic research has shown the potential for virtual DOE to allow for the rapid analysis of manufacturing parameters and their influence on final formed products. However, within the domain of bulk-metal joining, FEA tools are rarely used in industrial applications and limit DOE trails to physical testing which are therefore constrained by financial costs and time.

This research explores the suitability of an FEA-based DOE to predict the complex behaviour during bulk-metal joining processes through a case study on the staking of spherical bearings. For the two DOE outputs of pushout strength and post-stake torque, the FEA-based DOE error did not exceed ±1.2% and ± 1.5 Nm respectively which far surpasses what was previously capable from analytically derived closed-form solutions. The outcomes of this case study demonstration the potential for FEA-based DOE to provide an inexpensive, methodical, and scalable solution for modelling bulk-metal joining process

Automating modelling activities in computer-aided design (CAD) systems is no exception within design automation, one of the current research endeavours aiming to use and transform design-related data in design decision-making processes and the generation and evaluation facilitation of new design solutions. The paper explores the differences between CAD models based on their feature-based CAD modelling sequences that lead to the final models' design. The dataset collected and structured for the study contains more than 1400 CAD models clustered on two levels by using an unsupervised K-means clustering algorithm. The algorithm is performed on the number (total and unique) and the first-order Markov model transition matrices of the CAD modelling operations and their sequential order, respectively. Therefore, three and ten groups (clusters) of CAD models are obtained regarding the level of clustering. The results show that most of the obtained groups are specified by the dominant transition between particular modelling operations. In addition, the study also provides insight into the potential of using feature-based CAD modelling operations' sequences as a first step toward automating the user interaction with the CAD system.

Design aesthetics are one of the most important factors affecting the attractiveness of industrial products. Psychological theory suggests that a moderate level of novelty and complexity yields pleasant feelings in users. A design that is initially surprising to consumers and acceptable over time requires aesthetic interest associated with its novelty and complexity. In this study, we formulated the perceived novelty and complexity of a closed contour shape. Based on this formulation, we developed “Hybrid-GAN,” which is a shape-generation system capable of generating a variety of shapes of arbitrary novelty and complexity. In a series of experiment, we obtained subjective evaluations of novelty and complexity, as well as beauty and interest, for the generated shape samples. The results indicated that our novelty and complexity formulations had significant positive correlations with subjective evaluations. The sum of the formulated novelty and complexity also had a significant positive correlation with interest. The results of this study are expected to be used to support the design of attractive shapes by providing feedback to designers regarding the degrees of novelty and complexity that users find most pleasant.

Designing vibrating systems is challenging due to component interaction. One approach to reduce the resulting complexity is top-down design where requirements on components are formulated such that the overall system achieves the design goal. Previous work showed how to derive quantitative and solution-neutral requirements on components of a vibrating system, expressed as permissible ranges of impedance. This work adapts the methodology to a practical use case and provides a concrete technical solution: A hammer drill that can cause white finger syndromes to users is equipped with an appropriate vibration absorber. The hammer drill is represented by a lumped mass model and validated using experimental data of a reference design. Solution-neutral and quantitative component requirements on the overall dynamics of the vibration absorber expressed by impedance are derived. They provide a clear target for the component design. A vibration absorber in form of a Tuned Mass Damper (TMD) is designed accordingly. The final design is validated experimentally and shown to reduce the vibration by 47%.

The as-built geometry and material properties of parts manufactured using Additive Manufacturing (AM) can differ significantly from the as-designed model and base material properties. These differences can be more pronounced in thin strut-like features (e.g., in a lattice structure), making it essential to incorporate them when designing for AM and predicting their structural behaviour. Therefore, the aim of this study is to develop a numerical model with realistic characteristics based on a thin strut-based test artefact and to use it accurately for estimating its compressive strength. Experiments on test samples produced by selective laser sintering in PA 1101, are used to calculate geometrical deviations, Young's modulus, and yield strength, which are used to calibrate the numerical model. The experimental and numerical results show that the numerical model incorporating geometrical and material deviations can accurately predict the peak load and the force-displacement behaviour. The main contributions of this paper include the design of the test artefact, the average geometrical deviation of the struts, the measured material data, and the developed numerical model.

Finite element (FE) simulations can be used both in the early product development phase to evaluate the performance of developed components as well as in later stages to verify the reliability of functions and components that would otherwise require a large number of physical prototype tests. This requires calibrated material cards that are capable of realistically representing the specific material behavior. The necessary material parameter identification process is usually time-consuming and resource-intensive, which is why the direct inverse method based on machine learning has recently become increasingly popular. Within the neural network (NN) the generated domain knowledge can be stored and retrieved within milliseconds, which is why this method is time and resource-efficient. This research paper describes advantages and potentials of the direct inverse method in the context of the product development process (PDP). Additionally, arising transformation opportunities of the PDP are discussed and an application scenario of the method is presented followed by possible linkage potentials with existing development methods such as shape optimization.

Simulation is fundamental to many engineering design processes and powers the field of computational design. Simulation inherently consumes energy resulting in CO2 emissions that impact our environment. While one can source energy from renewable sources and use energy efficient hardware, efforts need to also be made in how we can use simulation in a sustainable manner.

This paper presents a sustainable simulation framework that borrows concepts from web services. The framework makes it easy for engineering firms to adopt and embed sustainable simulation practices thereby removing the burden from the designer tin thinking about how to design sustainably. An illustrative example reveals a 25% reduction in computational effort can be achieved by adopting the framework.

Dual ontological products are a physical construction and an emotional construction. Multitude of human factors must be considering when designing dual ontological products. To increase the product's impact and reach, designers should also understand the requirements of potential users. A design stage conflict exists between the emotional construction and the physical construction of a product when considering human factors. Designers find it difficult to achieve the right compromise between these constructions and hence, the balancing of the two is crucial. This research therefore contributes a novel harmonistic knowledge-based framework which makes designers aware of design stage conflicts and consequences of commitments made on human factors in the use-phase of the artefact. This approach was implemented in a machine learning based computational tool which exploits harmonistic knowledge and information collected from potential users to proactively assist, guide, and motivate product designers. This paper also presents a descriptive study for the evaluation of the framework and its implementation as a computer-based prototype tool. Results show the necessity and beneficial use of the tool for design engineering practice.

Drive units of eBikes are used in every type of bicycle and for different riding scenarios and riders. Due to the different riders and bike types, an enormous variety of influencing parameters and load spectra must be considered during the design process. Therefore, in this paper, a systematic approach for the optimization of the drive unit is presented, which adopts and combines several approaches from design theory. The focus is on efficient modeling and simulation of the relevant parameters and load spectra to minimize uncertainties in the design process.

Based on a system analysis, dimension-reduced parameter spaces are formed for the simulation of the system, meta-models are integrated into the simulation model and the results of the simulation are transferred into a data-based surrogate model to cover the parameter space in an efficient way with a minimum number of time consuming FE simulations. Furthermore, a coordinate-based evaluation method is presented for the FE model in order to form the input for the surrogate model, reduces the amount of data, and to allows a geometry- and mesh-independent evaluation to compare different models.

Representation of design information using causal ontologies is very effective for creative ideation in product design. Hence researchers created databases with models of engineering and biological systems using causal ontologies. Manually building many models using technical documents requires significant effort by specialists. Researchers worked on the automatic extraction of design information leveraging the computational techniques of Machine Learning. But these methods are data intensive, have manual touch points and have not yet reported the end-to-end performance of the process. In this paper, we present the results of a new method inspired by the cognitive process followed by specialists. This method uses the Knowledge Graph with Rule based reasoning for information extraction for the SAPPhIRE causality model from natural language texts. Unlike the supervised learning methods, this new method does not require data intensive modelling. We report the performance of the end-to-end information extraction process, which is found to be a promising alternative.

The ideation process has a significant impact on the initial concept generation and final product creativity of the design. Visual stimuli play an important role in the process of innovative product design. With the increase in computing capability, generative design methods are widely implemented. In this paper, features of design targets and combinational objects in 2 combinational design tasks are fused using adversarial neural generative networks to form the generated stimuli. It is also used with combinational object pictures to investigate the impact on creativity in design ideation. The study invited designers to use and subjectively self-evaluate the two stimuli in a design task. Through analysis of participant data (n=20), the results showed that the generative stimuli had an advantage over the combinational image stimuli in terms of the smoothness of creativity in the design ideation of outcomes. And there is a positive correlation between designers' years of design education and their tendency to prefer generative stimuli. Based on the results obtained, ideas are provided for the study of the influence of visual and generative stimuli on the designer's ideation process.

Topology Optimization (TO) is an established method for the development of high strength and lightweight structural components. However, its results have to be geometrically revised to obtain a computational model that meets product development requirements. Design has to be in accordance with manufacturing constraints. Geometry reconstruction therefore still is a typically manual and tedious task, but increasingly supported by computational and automated approaches. In this paper, the inclusion of a casting constraint in an automated Medial Axis based reconstruction method is presented. Since the Medial Axis provides cross-section values by the computation of maximally inscribed spheres, this information is used for geometry reconstruction and even further for the purposeful adaption of the cross-section to match Heuver's circle method. Thereby, the directed solidification of molten material is considered. With a predefined feeder position, the demonstrator of a suspension control arm is used to show the application of the method. Resulting CAD-models are also structurally evaluated for their stiffness characteristics.