Post on 25-May-2018
transcript
SECTION DE
Génie mécanique
Multi-Objective Optimization applied To MicroGrid
Author
Guillaume Ridoux
Supervisor
Dr François Maréchal
Acknowledgements Shrikant Bhat
Dr François Maréchal
ABB Global Industries &
Services Ltd. Bangalore, India
Motivation & Objectives The MicroGrid technology is part of the ”smartgrid” technological concept. While it aims at improving the efficiency of energy production, the smartgrid technology is expected to achieve a fast deployment in the forthcoming years. In this study, a MicroGrid management problem is formulated, from the modeling of the system, to the definition of the objectives to optimize.
The aim of the project is to evaluate the opportunities offered by Genetic algorithm for solving multi-objective problem in the case of MicroGrid system operation.
In this study, a MicroGrid system previously studied in the literature is analyzed. The possibility to produce combined heat and power has been added to the original model in order to obtain a better overall efficiency.
A Proton Exchange Membrane Fuel Cell (PEMFC), an Internal Combustion Engine (ICE) and a MicroTurbine (MT) are used for cogeneration; a battery, a Wind Turbine, Photovoltaic solar panels and the grid also supply electricity. A boiler has been added to be sure that the thermal demand is met at any time.
MicroGrid Model
Formulation of the multi-objective optimization problem
Decision making method
Sensitivity analysis
When dealing with more than two objectives the Pareto set becomes much more complex. It is worth studying the objective space and relating it with the decision variables space in order to get a better overview of the solutions available before taking decision. A GUI has been used to identify the solutions of the Pareto set.
The possible deviations from the reference case are studied through various sensitivity analysis and two options for the trading of electricity are compared. The battery storage is an important component of the system when stability enters into play. It allows to compensate the losses due to the stable operation of the units. The comparison between the solutions shows the range of performances reachable under the conditions studied.
A preferred solution is selected using a Multi-Attribute Decision-Making method that allows obtaining weights inherent to the distribution of the solutions. This method has the advantage to be valid even if the objectives are expressed in different units.
The Genetic Algorithm gives the possibility to find optimum for objectives expressed in different units. The economic/environmental dispatch problem is implemented and a third objective, improving the operation of the utility with respect to the effects caused by load transient, is added.
Study of the Pareto set
3 Dimensional Pareto Set with observed solutions
Architecture of the studied Microgrid
Projections of the Pareto set
Evolution of the battery level along the day
for five different solutions
Representation of the electricity production profile for the
E-TOPSIS solution.
Min (WAverage,i !Wit )2
Wit"0
i=1
N
#t=1
T
# (kWh2 )
if (T-toff ,i )>0 WAverage,i =Wi
t
t=1
T
#(T ! toff,i )
(kWh)
else WAverage,i = 0.
Conclusion & future work
1. Ref Case. 2. -20% Demand load. 3. 15kWh battery. 4. Flat electricity import price
Min Expenditurest ! Incomest
Wsuppliedt
t=1
T=24
" ($/kWh)
Min (eGWP, FCt + eGWP, ICE
t + eGWP, MTt + eGWP, Grid
t + eGWP, Boilert )
t=1
T
!
(kgCO2eq/ day)
This study contributes in the development of MG by providing further information to the decision maker on the quality of its operation strategy. The results have shown that buying the electricity at the market price was enabling better benefits and higher stability, because of the lower price. However, it involves higher imports, and therefore higher emissions. The formulation of the objective for the operation stability brings an important feature to the optimization. Indeed the production profiles obtained can become more stable, and it creates a new strategy for the use of storage devices.