Investigation of AIMD Based Charging Strategies for EVs Connected to a Low-Voltage Distribution Network - Intelligent Computing for Sustainable Energy and Environment - page 439

each EV is automatically disconnected from the grid when it is fully charged.
Updates are performed every 5 minutes (i.e. ΔT = 300s).
From Irish trac survey data [10] it can be concluded that the majority of
commuters arrive home between 4pm-8pm each day. To capture this in our simu-
lation we generate EV home arrival times from a normal distribution centered at
6 pm with a standard deviation of 1 hour. It should be noted that this falls within
the period normally associated with peak-power on the Irish grid (5-7pm), hence
unregulated EV charging has the potential to substantially increase peak-power
on the grid. Fig. 2(a) shows the voltage profile of the minimum voltage level in
the distribution network for three different scenarios: (i) no EVs on the grid; (ii)
uncontrolled charging of EVs; and (iii) charging of EVs with our proposed AIMD
smart charging algorithm. The corresponding power flows at the substation are
plotted in Fig. 2(b). As expected the impact of uncontrolled EV charging is to
increase the voltage drops significantly. The Minimum non-EV voltage on all
buses was 0.92 pu during peak-periods, but with uncontrolled EV charging co-
inciding with peak-power, bus voltages drop to 0.87 pu, leading to voltage issues
on the network. In addition, transformers are overloaded with demanded power
exceeding the available power by 65%. Thus, for our test distribution network
uncontrolled charging at 50% EV penetration cannot be supported. Simulations
conducted for different EV penetration levels (not included) show that the max-
imum level that can be sustained under these conditions is 10%. In contrast,
with AIMD smart charging voltage sags are substantially reduced (0.93pu com-
pared to 0.87pu) and capacity and infrastructure constraints are maintained
while making best use of available power.
Fig. 2. Comparison of uncontrolled and AIMD smart charging of EVs for a typical LV
network over a 72 hour period in mid-winter (0 = midnight): (a) Minimum voltage on
the distribution network; (b) substation power flow
The daily price variation signal E ( k ), set according to [11], is plotted in Fig.
3(a). The impact of the inclusion of this term to modulate the available power
signal in the AIMD algorithm is clearly evident when comparing the voltage
sags for uncontrolled and smart EV charging. In the latter the significant voltage
sags have been postponed until after the peak-time (around 9 pm) reflecting a
corresponding shift in the EV charging load. This is further highlighted in Fig.
3(b), which shows a comparison of the EV load profile obtained using AIMD