Supervisor Pr Olivier Deblecker
Co-supervisor Dr Zacharie De Grève

Abstract :

Wireless power transfer (WPT) refers to the remote transfer of electrical supply power between systems
separated in space, without resorting to any wired connection. Although seeming futuristic, WPT has
been envisioned in the end of the 19th century, but has been tributary of the contemporary technological
limitations. In the last decades, the genesis and the progress of power electronics have progressively
pushed these technical boundaries by allowing the conversion of high power (up to tens of kilowatts)
with high frequencies (up to tens of kilohertz), paving hence the way to the implementation of WPT to
energy-greedy applications such as electric vehicle (EV) battery charging. Bringing more convenience
and more safety, the latter application is a popular and trending topic in the scientific and industrial
communities. Using a resonant inductive coupling (yielding the resonant inductive power transfer or
RIPT), the wireless charging for EVs reaches currently comparable or even better performances than
classical wired connections.

Nevertheless, based on a coupled-mode resonance, RIPT systems are per se only effective in a restricted
area of the space of operating points, and present performances sensibly affected by their windings
design and by their operating conditions. Moreover, as high-power applications are considered, each
percent of the efficiency represents in absolute terms a non-negligible amount of power, which must be
saved to the extent possible. Consequently, this thesis has two main objectives.
The first objective is the development of a fast and accurate electromagnetic model of the windings
involved in the transfer, despite the modeling challenges ensuing from the occurrence of eddy currents
in the conductors and from the absence of any magnetic core channeling the useful flux lines. Based on
the assessment and the analysis of the computational burden associated with the state-of-the-art finiteelement
(FE) and generalized partial element equivalent circuit (PEEC) methods, lightening solutions
are proposed. Notably, an original method for optimizing the mesh of conductors incurring eddy currents
is exposed and validated in order to alleviate the two-dimensional FE model of the windings. Moreover,
an original FE formulation using surface impedance boundary conditions is proposed and results in a
drastic decrease of the burden associated with the three-dimensional FE model of the windings, which
was not practicable using state-of-the-art formulations. All the models developed in this part of the thesis
are confronted successfully with experimental results and are gathered in a virtual laboratory. The latter
consists in a fully parametrized open-source software tool embedding a user-friendly interface
permitting quick and easy virtual tests on RIPT windings, saving the time and money associated with
the experimental prototyping.

The second objective is the elaboration of a topology and of an optimal command strategy for the RIPT
system power converters, in order to extend the flexibility of the resonant coupling despite deviations
in the circuit parameters. An original optimal command methodology for bidirectional series-series
RIPT systems is proposed. Based on a resonant dual active bridge topology, the proposed strategy
considers proceeding to non-synchronous active rectification for simultaneously and independently
controlling the active and the reactive power flows in order to achieve the maximum efficiency despite
deviations in the circuit parameters and/or detuning of the resonant circuits. The method effectiveness
is demonstrated on an EV test-case via frequency- and time-domain simulations. The robustness of the
proposed command with respect to severe parameters deviations is increased by extending the
converters voltage range. For doing so, a transition to the innovative Z-source topology is considered
and compared to the conventional inclusion of an additional DC-DC converter in the system. Further,
the scope of applicability of the proposed methodology is enlarged to the four elementary and most
usual topologies for RIPT systems, improving hence the universality of the proposed strategy. Finally,
the proposed topology and optimal command are experimentally validated via the presentation of a
successful low-power (20 W) experimental proof-of-concept, entirely conceived in our Unit laboratory.