Publikationen - Netze für Erneuerbare Energien
Arjun Brück, Silvan Faßbender, Eberhard Waffenschmidt,
"Single- and multi-objective parameter optimization in a tool
for designing PV-diesel-battery systems",
7th International Energy and Sustainability Conference 2018 (IESC
2018), Cologne, 17.-18. May 2018
In many isolated off-grid areas diesel generators are the common way
of providing electricity. The high energy cost and CO2 emissions might
be reduced by implementing PV plants with an attached battery storage
into the microgrid. However, the correct dimensioning of both PV and
battery storage is crucial. Using a MATLAB/Simulink tool based on previous
work, such PV-Diesel systems can be calculated for variable storage
capacity, PV sizing and dispatch strategies.
To find a preferably efficient optimization method in MATLAB, a genetic
and a simplex algorithm are compared. Optimization objectives were low
levelized cost of electricity (LCOE) or carbon dioxide (CO2) emissions,
by sizing photovoltaics and battery of the system. The specific algorithms
were chosen since they don’t rely on derivatives as the Simulink calculation
is discrete and non-linear.
It is shown that the simplex algorithm converges within a couple of
minutes and quiet faster than the genetic algorithm. Furthermore, a
multi-objective optimization is implemented using an epsilon-constraint
method. The user is able to identify appropriate dimensioning with emphasis
on different targets by calculating distinct pareto optimal solutions.
> Pre-Print of the paper: PDF-Document
The different operations of the Nelder-Mead algorithm.