Due to the limited supply of fossil fuels and the common interest of politicians, OEMs and the general public to reduce vehicle-related C02-emissions, plug-in hybrid electric vehicles (PHEVs) are one promising step towards the electrification of the automotive powertrain. Numerous possibilities exist to choose, arrange and size the different powertrain components, i.e. the internal combustion engine, electric machine(s), transmission(s) and electric energy storage. This contribution presents a framework for the holistic evaluation of powertrain concepts which is suited to handle manufacturing constraints and also include a cost model. Key elements of this framework are a synchronized parameterization as well as surrogate modeling and multi-criteria optimization techniques. While numerous studies exist about simulation approaches which put emphasis on the calculation of powertrain efficiencies, a holistic approach covering aspects from multiple domains, including manufacturing and cost is currently missing. In case incompatibilities occur in one or more of the neglected domains the cost to resolve these issues rise disproportionately as the product development process is getting more refined. According to the concept of frontloading the costs to resolve problems in a development process can be decreased by providing as much information as possible in the early concept phase. In this context data-mining studies as proposed in [2] and modular simulation approaches have been used. Potential data sources like proprietary data, pre-existing models and component catalogs have been sourced in order to have a data basis that provides the relevant information to meet the requirements. The requirements can be rather high-level and more customer-oriented or more detailed, when incorporating the engineers’ preferences. Thus they determine the level of detail of the parameterization and are consequently synchronized with the latter. A parameterization approach with two types of numeric values has been chosen: on the one hand continuous numerical values to ensure scalability of components for searching the design space in an unbiased way and on the other hand discrete numerical values, e.g. to take advantage of standard parts. Further categorical parameters are introduced and allow the switching between already existing components or concepts. The key parameters that influence the manufacturing process and the cost have been identified for each component by consulting expert knowledge for each component, e.g. the parameterization of the electric motor(s) follows the ideas laid out by [3], whereas the internal combustion engine has been parameterized according to [4]. A modular simulation model allows the calculation of results for different layers of complexity, according to the respective set of parameters. A design of experiments approach is used in order to minimize the number of required simulations runs and cover the design space with training data points in an efficient way. The fit of data is done by regression models or artificial neural networks, depending on the number and quality of training data. Statistical analyses and cross-validation approaches are used to validate the fit and generate a valid surrogate model of the respective architecture concept. Once these models exist the decision making phase starts. Depending on the requirements a single- or a multi-criteria objective function has to be solved by optimization or search routines. As surrogate models have a relatively simple algebraic structure the optimization runs and the creation of Pareto-frontiers is comparatively quick. The decision making process is further supported by the possibility to navigate the design space, so that the resulting concept fulfills the requirements. The framework finally is validated by two exemplary concepts which were generated based on high-level requirements on the one hand and manufacturing constraints on the other hand.     «

Due to the limited supply of fossil fuels and the common interest of politicians, OEMs and the general public to reduce vehicle-related C02-emissions, plug-in hybrid electric vehicles (PHEVs) are one promising step towards the electrification of the automotive powertrain. Numerous possibilities exist to choose, arrange and size the different powertrain components, i.e. the internal combustion engine, electric machine(s), transmission(s) and electric energy storage. This contribution presents a f...     »