Nguyen, Hieu Trung
(2017).
Decision making for smart grids with renewable energy.
Thèse.
Québec, Université du Québec, Institut national de la recherche scientifique, Doctorat en télécommunications, 188 p.
Résumé
We are at the dawn of the smart grid era where there is significant increase in the demandside participation
in the grid’s operations. One important smart grid research topic concerns active demandside
management which can potentially result in great benefits for different involved grid entities (e.g.,
electricity customers, utility, resource aggregators) and enable to support increasing penetration of distributed
renewable energy resources. However, efficient design for different decision making problems
must be conducted to to realize the potential impacts of the active demandside management, which
is the focus of this dissertation. Specifically, we study three decision making problems of the corresponding
demandside smart grid entities considering integration of renewable energy sources, which
are smart home energy management, smart load serving entity (LSE) pricing design, and cost allocation
for cooperative demandside resource aggregators (DRAs) under the virtual power plant (VPP)
framework. Our research has resulted in three major contributions, which are presented in three main
chapters of this dissertation.
The first contribution is related to the efficient energy scheduling design for smart homes equipped
with solar assisted thermal load. This design is conducted under the timevarying dynamic pricing
scheme which can potentially bring great demand response benefits for home electric consumers.
Specifically, we develop a rolling twostage stochastic programming based algorithm, which aims
to minimize the electricity cost, guarantee user comfort, and efficiently utilize renewable energy resources.
We also propose to exploit the solar assisted thermal load for the energy management and
analyze the impacts of different parameters on the smart home economic improvements.
The second contribution concerns the development of a dynamic pricing scheme for a load serving
entity (LSE) that can incentivize electric customers to provide demand response services. The design
can effectively encourage participation of electric customers with flexibilities in energy consumption
while not negatively affecting other electric customers lacking flexibilities in changing their energy
consumption. Moreover, the proposed pricing scheme is compatible with the current market structure.
Toward this end, we consider the pricing design as a bilevel optimization problem where the grid operator
is the leader, who determines the demand response price, and the flexible customers are followers,
whose energy consumption is adjusted in response to that price signal. We describe how to transform
the proposed bilevel optimization problem, which is difficult to be solved directly, into an equivalent
single objective mixed integer linear program (MILP), which can be solved efficiently by a branch and
cut algorithm. Numerical results show that our proposed pricing design can be beneficial to both grid
operator and electric customers.
The third contribution aims to develop an efficient cost allocation scheme for cooperative demandside
resource aggregators (DRA), which are coordinated under an emerging smart grid concept, namely, the virtual power plant (VPP). We address this problem by using the core based cooperative game
theoretic approach. Since the core of the underlying game can contain many cost allocation solutions,
our design enables us to choose an appropriate cost allocation solution inside the core that optimizes
both stability and fairness metrics. This core based cost allocation problem is formulated as a largescale
biobjective optimization problem with an exponential number of implicit constraints related
to the core definition. In particular, the parameters of these constraints are the function values of
coalitions of DRAs, which are the outcomes of the optimal bidding strategies of the corresponding
coalitions of DRAs. To tackle this highly complex biobjective optimization problem, we propose
to employ the constraint and row constraint generation methods, which exploit the fact that the
number of optimization variables can be much smaller than the number of optimization constraints.
Numerical studies show that the proposed algorithm allows to construct the Pareto front with a large
number of Pareto points for a VPP consisting of a large number of DRAs. Moreover, the proposed
framework enables the VPP to determine a suitable cost allocation for its members considering the
tradeoff between stability and fairness.
Type de document: 
Thèse
Thèse

Directeur de mémoire/thèse: 
Le, Long Bao 
Motsclés libres: 
smart grid; efficient energy scheduling design; smart homes; dynamic pricing scheme; load serving entity; cost allocation scheme; cooperative demandside resource aggregators; virtual power plant 
Centre: 
Centre Énergie Matériaux Télécommunications 
Date de dépôt: 
12 févr. 2018 21:27 
Dernière modification: 
28 sept. 2020 13:35 
URI: 
http://espace.inrs.ca/id/eprint/6534 
Actions (Identification requise)

Modifier la notice 