Rankine cycle based waste heat recovery system applied to heavy duty vehicles: topological optimization and model based control

standards force the original engine manufacturers to search for innovative solutions in order
to reduce oil consumption. As an important part of the energy contained in the primary
carrier (the fuel) is lost to the ambient through heat, it seems convenient to recover a part
of this thermal energy and to turn it into fuel consumption reduction. Thermodynamic bottoming
cycle such as the Rankine cycle could be used to meet this objective. Its popular
use throughout the world for electricity generation makes it a natural candidate for on-board
implementation in vehicles. However, a certain number of hurdles are still present before the
system can be efficiently applied to heavy-duty trucks.
In the last thirty years, numerous studies heave been carried out to evaluate the real potential
of that kind of system on a vehicle but nothing has yet been commercialized. The heat
sources to recover from, the constraints relative to the on-board application and the long and
frequent transient behavior of the vehicle mean both the system architecture and its control
strategy need to be optimized. The system optimization leads to a choice in terms of working
fluid, heat sources and sinks, and components sizing in order to maximize power recovery
and hence the fuel saving. The control plays a major role by using the capability of such a
system to ensure an efficient and safe operation and limiting the interactions with the other
vehicle sub-systems.
In this thesis, a system design methodology is introduced to optimize the system architecture
using complete model-based vehicle simulation. The constraints relative to the mobile
application are taken into consideration to evaluate the potential of such a system. Modelbased
control strategies for on controlled variable, namely the superheat level, are developed.
Constrained by the implementation platform, different control frameworks ranging from PID
to model predictive controllers or observer based controllers are developed to fit into a normal
automotive electronic control unit. Most of these novel strategies were experimentally
validated on a test rig developed during the thesis.