Page 473 - Decision Making Applications in Modern Power Systems
P. 473
432 Decision Making Applications in Modern Power Systems
EV fleet in this study. Simulations to investigate the impact of varying
threshold values on revenue are presented here. As the threshold value
increases in Fig. 16.12, the initial revenue drop gets smaller; however, the
revenue sharply decreases as the threshold increases from 20 to 60 kW, indi-
cating that most commitments failed to satisfy the constraints defined by
Eqs. (16.22) and (16.23) because the required power adjustments exceeded
the capacity of the EV fleet.
16.5 Distributed optimization with asynchronous ADMM
and V2G capabilities
Sections 16.3 and 16.4 have summarized the modeling approaches and the
performance metrics of the market integration strategies of the controllable
EVs within microgrid scenarios. Various types of information have been
considered in the planning problems, including monthly projection of load
profile, renewable generation, and the aggregate EV energy demand.
However, the resultant optimal power profile, that is, PtðÞ; tA½1; 2; ... ; T,
obtained from various market integration strategies, cannot be directly
applied to control the individual vehicles, so additional steps are needed to
disaggregate the aggregate power profile. In this section, we extend the
previous evaluation approach on a monthly basis to an implementable day-
by-day operation, whose objective is to allow multiple EV agents to follow
the prescheduled aggregator’s power consumption profile in a decentralized
and asynchronous fashion, using ADMM.
Fig. 16.13 indicates the implementation architecture of the decentralized
load following program, where each EV computes a local optimization prob-
lem but exchanges information with the aggregator by limited amount of
communication. The overall problem is defined as follows:
Aggregator
Signal 1
Signal 2 Signal N
...
EV 1 EV 2 EV N
FIGURE 16.13 Distribute optimization paradigm.
