Increasing numbers in distributed generation units and growing
electrification of the transportation sector leads to new challenges in
electrical distribution grids. In this thesis, a holistic approach,
including different functional layers for the operation, monitoring, and
control of low voltage grids, is presented. Power flow methods are
usually used to monitor the grid, but for control purposes, their
implicit and non-linear character is quite challenging. This work
introduces a linear explicit power flow approximation. It exploits
on-line information combined with pseudo measurements to adapt to
operating points of the grid. The lack of this functionality is a
primary source of error in standard off-line methods. Needed grid
parameters for the approximation, are calculated with an approach that
combines a dynamic thermal model of the power cables with a mean value
estimation of the impedance. Thus, resistive parameter changes due to
load currents can be tracked during grid operation. The first
operational layer is designed as a distributed model predictive control
(DMPC). Its purpose is to better unify three-phase generation units,
charging facilities, and dominant consumers in low voltage grids. It
maximizes the transport capacity of the network, keeps sensitive data
from each controller private, and considers the limitation of grid
assets. A secondary layer deals with the inherent unbalance in low
voltage grids. The approach uses a Jacobi based distributed optimization
algorithm to coordinate local, flexible electric power. With the
developed power flow approximation, it is possible to formulate a local
optimization problem, that does not scale with grid size. Additionally,
it can directly reduce the negative- and zero-sequence components
without the need for additional measurements.
