Promoteur : Prof. Jacques Lobry

Résumé :

The massive integration of Distributed Generation (DG) units in the electric
distribution systems has created serious technical challenges for the Distribution
System Operators (DSOs) including the voltage rise and local congestion problems.
In order to deal with such arising issues, the conventional passive manner of the
distribution network management based on the fit-and-forget policy is being replaced
by the Active Network Management (ANM) schemes aiming at operating network
in a safe and cost-effective way by taking advantage of the real-time communication
and optimal control of the available devices. This thesis addresses the voltage control
problem of the Medium-Voltage (MV) distribution systems under deterministic to
uncertain model.
Inspired by the ANM framework, in the first part of the thesis, a centralized
sensitivity-based voltage control approach is developed which manages the
transformer tap position and DG active and reactive powers in order to maintain the
node voltages and branch currents within their permitted limits. The sensitivity
analysis determines impacts of the control variables on the operational limits of the
system. Thanks to the information provided by the proposed sensitivity analysis
methods, the voltage control problem is simplified to a linear optimization
formulation that can remain tractable in almost real-time. Another important feature
of the sensitivity-based voltage control approach is that it does not require the state
estimation interface and with limited number of the voltage and power
measurements, it can manage the voltage constraints.
In the electric distribution systems, as an accurate and up-to-date model of the system
is not available, the calculations and analyses are performed relying on the simplified
deterministic model that can lead to erroneous analyses and eventually wrong control
decisions. In the second part of the thesis, it is considered that the network model is
not deterministic anymore but is rather uncertain varying within the predefined
bounds. A probabilistic framework is developed in order to evaluate impacts of the
uncertain models of the system components on the voltage control problem of the
MV distribution systems. In addition, a robust voltage control algorithm is designed
that considers the model uncertainty when taking corrective decisions of the control
variables. It determines a solution, which remains immunized against all possible
realizations of uncertainties associated with the network component models.