Search Results

In order to predict reality as accurately as possible leads to the fact that numerical models in automotive vibroacoustic problems become increasingly high dimensional. This makes applications with a large number of model evaluations, e.g. optimization tasks or uncertainty quantification hard to solve, as they become computationally very expensive. Engineers are thus faced with the challenge of making decisions based on a limited number of model evaluations, which increases the need for data-efficient methods and reduced order models. In this contribution, variational autoencoders (VAEs) are used to reduce the dimensionality of the vibroacoustic model of a vehicle body and to find a low-dimensional latent representation of the system.

Substructuring approaches are helpful methods to solve and understand vibro-acoustic problems involving systems as complex as a vehicle. In that case, the whole system is split into smaller, simpler to solve, subsystems. Substructuring approaches allow mixing different modeling “solvers” (closed form solutions, numerical simulations or experiments). This permits to reach higher frequencies or to solve bigger systems. Finally, one of the most interesting features of substructuring approaches is the possibility to combine numerical and experimental descriptions of subsystems. The latter point is particularly interesting when dealing with subdomains that remain difficult to model with numerical tools (assembly, trim, sandwich panels, porous materials, etc.). The Patch Transfer Functions (PTF) method is one of these substructuring approaches. It condenses information (impedance matrix) of subsystems on their coupling surfaces.

The Noise Vibration and Harshness (NVH) characteristics and requirements of vehicles are changing as the automotive manufacturers turn their focus from developing and producing cars propelled by internal combustion engines (ICE) to electrified vehicles. This new strategic orientation enables them to offer products that are more efficient and environmentally friendly. Although electric powertrains have many advantages compared to their established predecessors they also bring new challenges that increase the difficulty of matching the high quality requirements of premium car producers especially regarding NVH. Electric motors are one of the most important sources of vibrations in electric vehicles.

In times when automated driving is becoming increasingly relevant, dynamic simulators present an appropriate simulation environment to faithfully reproduce driving scenarios. A realistic replication of driving dynamics is an important criterion to immerse persons in the virtual environments provided by the simulator. Motion Cueing Algorithms (MCAs) compute the simulator’s control input, based on the motions of the simulated vehicle. The technical restrictions of the simulator’s actuators form the main limitation in the execution of these input commands. Typical dynamic simulators consist of a hexapod with six degrees of freedom (DoF) to reproduce the vehicle motion in all dimensions. Since its workspace dimensions are limited, significant improvements in motion capabilities can be achieved by expanding the simulator with redundant DoF by means of additional actuators.

Modern exhaust systems contain not only a piping network to transport hot gas from the engine to the atmosphere, but also functional components such as the catalytic converter and turbocharger. The turbocharger is common place in the automotive industry due to their capability to increase the specific power output of reciprocating engines. As the exhaust system is a main heat source for the under body of the vehicle and the turbocharger is located within the engine bay, it is imperative that accurate surface temperatures are achieved. A study by K. Haehndel [1] implemented a 1D fluid stream as a replacement to solving 3D fluid dynamics of the internal exhaust flow. To incorporate the 3D effects of internal fluid flow, augmented Nusselt correlations were used to produce heat transfer coefficients. It was found that the developed correlations for the exhaust system did not adequately represent the heat transfer of the turbocharger.

As computational methodologies become more integrated into industrial vehicle pre-development processes the potential for high transient vehicle thermal simulations is evident. This can also been seen in conjunction with the strong rise in computing power, which ultimately has supported many automotive manufactures in attempting non-steady simulation conditions. The following investigation aims at exploring an efficient means of utilizing the new rise in computing resources by resolving high time-dependent boundary conditions through a series of averaging methodologies. Through understanding the sensitivities associated with dynamic component temperature changes, optimised boundary conditions can be implemented to dampen irrelevant input frequencies whilst maintaining thermally critical velocity gradients.

With the huge improvements made during the last years in the area of integrated safety systems, they are one of the main contributors to the massively rising complexity within automotive systems. However, this enormous complexity stimulates the demand for methodologies supporting the efficient development of such systems, both in terms of cost and development time. Within this work, we propose a co-simulation-based approach for the validation of integrated safety systems. Based on data measurements gained from a test bed, models for the sensors and the distributed safety system are established. They are integrated into a co-simulation environment containing models of the ambience, driving dynamics, and the crash-behavior of the vehicle. Hence, the complete heterogeneous system including all relevant effects and dependencies is modeled within the co-simulation.

According to upcoming legislative regulations in certain countries, electric and hybrid-electric vehicles (EVs and HEVs) will have to be equipped with devices to compensate for the lack of engine noise needed to warn pedestrians against the vehicles. This leads to the question of appropriate sound design which has to meet specific psychoacoustic requirements. The present paper focuses on auditory features of warning sounds to enhance pedestrians' safety with a major focus on the detectability of the exterior noise of the vehicle in an ambient noise. For the evaluation of detectability, the psychoacoustic model developed by Kerber and Fastl will be introduced allowing for the prediction of masked thresholds of the approaching vehicle. The instrumental assessment yields estimates of the distance of an approaching vehicle at the point it becomes audible to the pedestrians.

The validity of numerical simulations is still limited by the unknown failure of materials when nonlinear load paths in successive stamping and crash processes occur. Localized necking is the main mechanism for fractures in ductile sheet metal. The classical forming limit curve (FLC) is limited to linear strain paths. To include the effects of nonlinear strain paths a theoretical model for instability (algorithm CRACH) has been used. The algorithm has been developed on the basis of the Marciniak model [8]. The calibration and validation of this approach is done by a set of multistage experiments under static and dynamic strain rates for a mild steel.

The relationship of the airbag fire-distribution as a function of impact velocity to the airbag fire-time is studied through the use of an optimization procedure. The study is conducted by abstracting the sensor algorithm and its associated constraints into a simple mathematical formulation. An airbag fire objective function is constructed that integrates the fire-rate and fire-time requirements. The function requires the input of a single acceleration time history; it produces an output depending on the airbag fire condition. Numerical search of the optimal fire threshold curve is achieved through parameterizing this curve and applying a modified simplex search optimization algorithm that determines the optimal threshold function parameters without computing the complete objective function in the parameter space. Numerical results are given to show the effectiveness and potential difficulties with the automatic search scheme.

This paper describes the approximation of the engine torque of an SI-engine using recurrent neural networks. As modern engine control units today are based on engine torque management, there is a need for an accurate determination of the produced engine torque. Since direct measurements of this value are not possible in series applications due to resulting high costs, a fast and accurate algorithm for the approximation of these values from other sensor signals has to be found. Dynamic neural networks are a promising method to fulfill these requirements.

Future emission regulations and customer needs require revolutionary new approaches to engine management systems. In the EC part-funded AENEAS program the partners Ricardo, Kistler and DaimlerChrysler formed a consortium to investigate the application of a new combustion pressure sensor concept and innovative algorithms for engine management systems. This paper describes the general scope and the basic concepts of the system.