This thesis focuses on powertrain modelling and simulation for next-generation mild hybrid buses. The problem area of focus for the work is in the design and development of a novel mild-hybrid parallel passenger bus architecture. Many unknowns arise from the study of this type of vehicle, in particular the difficulty in controlling and combining the propulsive energy delivered from multiple power sources. In early design stages a common solution is to develop a powertrain model that can capture the required characteristics to inform engineering decisions. However, many current powertrain models exhibit a high level of dependence between sub-system component models through kinematic linking of powertrain speeds which reduces model flexibility. There are also challenges in validating these models where a high level of understanding of instantaneous system behavioural characteristics is required to evaluate the complex control systems and energy management strategies associated with mild hybrid powertrain operation. Therefore, the first objective of this thesis is to establish a flexible and intrinsically decoupled modelling architecture where sub-model operation is not dependant on inputs from external sub-models at system boundaries. The second objective is to develop a methodology suitable for the validation of a complete vehicle system model which captures both global and instantaneous behavioural characteristics. The third objective is to show that both the novel modelling architecture and novel validation methodology can be applied to the design of a next-generation parallel mild hybrid bus. These objectives have been met through the development of a novel vehicle modelling architecture in the MATLAB Simulink environment which eliminated kinematic speed linking between sub-system components, allowing each subsystem component to intrinsically calculate its own rotational or linear speed. A novel, multi-fidelity validation algorithm has also been developed which captures global and instantaneous system characteristics, as well as assessing the presence or absence of anomalies in the simulation results. Combinatorial effects of signals can often produce, or mask, errors at different levels of fidelity within the model and a strategy to identify these is proposed. Time domain lead or lag behaviours were identified using a cross correlation function which to the authors best knowledge has not been used in this application previously. Finally, the validated models were used in a number of studies to assess the advantages of the mild hybrid vehicle architecture in comparison to the baseline micro-hybrid technology over a standard drive cycle. The optimum drive ratio between the hybrid electric components and the main powertrain was also assessed as well as the effects of replacing the standard battery system with an ultracapacitor energy storage system. The robustness of the proposed vehicle model has been assessed through a study similar to a system sensitivity analysis assessing the effects of inputting the expected extremes of parameters on simulation results stability, with particular focus on the impact of vehicle mass and driver aggressiveness on vehicle performance. The mild hybrid system was shown to decrease fuel consumption by 10.6% for the double deck vehicle configurations and 10.5% for the single deck vehicle configurations from simulation results on an LUB drive cycle. The drive ratio between the MGU’s and main powertrain that resulted in the lowest overall fuel consumption over an LUB drive cycle was 3.1. Simulations also showed no significant improvement in vehicle fuel consumption through replacement of the battery system with an equivalent ultracapacitor system. The model robustness analysis showed that the models are capable of simulating when input parameters are at the extremes of their expected values. For the single deck vehicle configurations, the mild hybrid vehicle architecture is more sensitive to mass variation than the MicroHybrid3 baseline and for the double deck vehicle configuration the mild hybrid vehicle is less sensitive to mass variation at lower acceleration rates but reverts to being more sensitive to mass variation at higher acceleration rates. The work in this thesis has supported engineering decisions and assisted with the technology demonstrator vehicles achieving low carbon emission bus status from the Low Carbon Vehicles Partnership.
|Date of Award||2019|
- Queen's University Belfast
|Supervisor||Juliana Early (Supervisor) & Geoff Cunningham (Supervisor)|