Definition and verification of a set of reusable reference architectures for hybrid vehicle development

Abstract

Current concerns regarding climate change and energy security have resulted in an increasing demand for low carbon vehicles, including: more efficient internal combustion engine vehicles, alternative fuel vehicles, electric vehicles and hybrid vehicles. Unlike traditional internal combustion engine vehicles and electric vehicles, hybrid vehicles contain a minimum of two energy storage systems. These are required to deliver power through a complex powertrain which must combine these power flows electrically or mechanically (or both), before torque can be delivered to the wheel. Three distinct types of hybrid vehicles exist, series hybrids, parallel hybrids and compound hybrids. Each type of hybrid presents a unique engineering challenge. Also, within each hybrid type there exists a wide range of configurations of components, in size and type. The emergence of this new family of hybrid vehicles has necessitated a new component to vehicle development, the Vehicle Supervisory Controller (VSC). The VSC must determine and deliver driver torque demand, dividing the delivery of that demand from the multiple energy storage systems as a function of efficiencies and capacities. This control component is not commonly a standalone entity in traditional internal combustion vehicles and therefore presents an opportunity to apply a systems engineering approach to hybrid vehicle systems and VSC control system development. A key non-­‐functional requirement in systems engineering is reusability. A common method for maximising system reusability is a Reference Architecture (RA). This is an abstraction of the minimum set of shared system features (structure, functions, interactions and behaviour) that can be applied to a number of similar but distinct system deployments. It is argued that the employment of RAs in hybrid vehicle development would reduce VSC development time and cost. This Thesis expands this research to determine if one RA is extendable to all hybrid vehicle types and combines the scientific method with the scenario testing method to verify the reusability of RAs by demonstration. A set of hypotheses are posed: Can one RA represent all hybrid types? If not, can a minimum number of RAs be defined which represents all hybrid types? These hypotheses are tested by a set of scenarios. The RA is used as a template for a vehicle deployment (a scenario), which is then tested numerically, thereby verifying that the RA is valid for this type of vehicle. This Thesis determines that two RAs are required to represent the three hybrid vehicle types. One RA is needed for series hybrids, and the second RA covers parallel and compound hybrids. This is done at a level of abstraction which is high enough to avoid system specific features but low enough to incorporate detailed control functionality. One series hybrid is deployed using the series RA into simulation, hardware and onto a vehicle for testing. This verifies that the series RA is valid for this type of vehicle. The parallel RA is used to develop two sub-­‐types of parallel hybrids and one compound hybrid. This research has been conducted with industrial partners who value, and are employing, the findings of this research in their hybrid vehicle development programs.