An innovative approach to dynamic driving simulations for vehicle thermal management processes

Extending from steady state simulation cases allows engineers to investigate the "real" on road conditions that are typically associated with transient characteristics. The ability to predict time dependent thermal behavior of vehicle environments leads to improved product development (with optimised modification potential) and an inherent reduction in experimental dependency and cost. Additionally many of the critical thermal cases experimentally validated are coupled to non-equilibrium boundary conditions, resulting either in an over-prediction through steady state 3-D CFD modeling or an inaccurate prediction through 1-D modeling. Today’s current industrial standard for virtual vehicle thermal management is mainly limited to steady-state conditions with some semi dynamic cases. Previous attempts have been proposed in the literature by industrial experts, to simulate unsteady thermal behavior of these time dependent vehicle conditions (e.g. thermal soak and hill climb). However, some of these proposals require a dependency on strong computing power, complete 3-D exhaust flow simulations, and complex multi-model coupling arrangements. Even though these methods incorporate realistic modeling approaches (due to the envisioned availability in computational resources), they lack the potential to accommodate complex boundary conditions, and to date no methodology has been able to simulate dynamic driving scenarios. Which brings up the question, why? The research proposes that the vehicle simulation field has overlooked the possibility to attempt highly transient cases, such as dynamic driving, due to the extreme volatility existing within vehicle boundary conditions, coupled to the traditional approaches which attempt to simulate these conditions. Therefore for numerical stability the simulation time scales have to be drastically reduced in order to resolve the high boundary condition fluctuations. This naturally exaggerates the calculation time for reasonable results and consequently compromises the application of the methodology for industry (considering the vehicle development time constraints). Therefore due to the simulation paradigm, “that original boundary conditions need to be modeled”, dynamic driving scenarios for full vehicle configurations is considered unfeasible. Simulation simplification is typically made on the geometrical modelling side, the implementation of physics or the assumptions of certain sub-system parameters, but significant alteration has never been made to the fundamental vehicle boundary conditions. However it is clear that the mass of the component can dampen the speed of its surface averaged thermal response to a given set of boundary conditions. Therefore the component essentially reacts at a different time scale to that of the exposed energy flux. Through this understanding it is evident that there is a potential to simplify the boundary conditions (therefore accelerating the calculation) based on the response nature of components. The following research aims at challenging the current simulation paradigm by introducing a methodology which simplifies the vehicle boundary conditions. The research goal is firstly the proof that dynamic driving profiles are possible to simulate with the current computational resources for full vehicle configurations. And secondly, the simplification of boundary conditions coupled to a new methodology to simulate these conditions, can replicate the time dependent component temperature behaviors, without significant depreciation in simulation accuracy. In the following research, a technique is presented which simplifies the high frequency changes within vehicle boundary conditions through the utilisation of wavelet transform based signal decomposition. Through identifying the thermally relevant frequency ranges (which correspond to component temperature change) a simplified boundary condition signal is derived. This simplified signal is then utilised in a quasi-transient approach to simulate the time dependent thermal behavior of the vehicle under highly dynamic loading conditions (e.g. Race-Track). A sensitivity study was also conducted to identify the propagation of error corresponding to the utilisation of simplified boundary conditions compared to the original input signals. Then a quasi-transient approach was compared to a fully transient approach in order to identify the error associated with the utilisation of multiple steady state CFD solutions. Additionally several investigation were conducted on the type of CFD solution necessary to achieve appropriate convection conditions for a quasi-transient approach. Ultimately the optimal parameters from these prior sensitivity studies corresponding to the least introduced error were implemented on a full vehicle configuration. Here profile independence was explored by transferring the proposed methodology to alternative dynamic driving profiles. Simulation results on multiple profiles were validated through experimental climatic wind tunnel tests, where a strong correlation was attained. These results were achieved with very little penalty on resources, whereby the methodology yielded a fraction of the calculation times (using 1/5th of the computing capacity) published in literature by alternative methods on simpler driving profiles. Additionally it was found that the simulation accuracy achieved was independent of driving profile, with discrepancies corresponding to modeling limitations on the vehicle. These results further confirmed the transferability capacity of the proposed methodology and the potential for wide scale application beyond the focus of this investigation