21, rue d’Artois, F-75008 PARIS 142 BOLOGNA 2011

http : //www.cigre.org

Flexible Power Management in MV Networks with High Penetration of

Electric Vehicles

R.CALDON, L. CARRADORE, R.TURRI

University of Padova
Italy

SUMMARY

The envisaged increasing electrification of private transport (e.g. electric vehicles) will
modify further the energy demands of customers. In this perspective, significantly more
system monitoring and intelligent control will be needed in order to securely manage the
electrical network ensuring optimal conditions meanwhile enabling active customer
participation to system management through the provision of ancillary services. In this
perspective Vehicle-to-Grid (V2G) technology will allow bi-directional flows of on-board
batteries, thus enabling Electric Vehicles (EVs) participation in services for the grid support
as flexible distributed energy storages. A rather interesting feature of the bidirectional EV
battery charger is its inherent capability of providing reactive power without actually affecting
the battery life, which makes it a valuable resource for the network as distributed reactive
support.
In this paper, EVs aggregators, such as residential and office and car parking, are considered
to participate with other distributed resources to the network voltage regulation through a
recently proposed novel network management procedure where participants are activated
sequentially by means of a control token circulating around the responsive resources present
in suitably defined network regulating areas. The token, generated and controlled by the
network operator, represents a reactive power request and each customer-recipient may decide
to provide only part of the required power variation, the total amount or nothing, depending
on its own availability and economical benefit.
Through a case study example, it is shown that, provided a suitable and flexible regulation
scheme is adopted, EVs aggregators could meet responsively the grid power requirements and
actively participate to the network voltage regulation.

KEYWORDS

Active distribution networks, Electric Vehicles, Voltage regulation.

[email protected]
1. INTRODUCTION
In a smart grid scenario, the Information and Communication Technology (ICT) infrastructure
superimposed to the power grid is envisaged to radically modify the paradigm of network operation,
which would evolve towards an internet like model, where every node has local intelligence and can
be responsive to different types of signals, and thus opening to interesting new management criteria in
order to fully exploit distributed resources.
This scenario will foster the establishment of local energy markets, with the possibility to make
resources price responsive to network needs. In fact, distributed resources will actively participate in
regulation if the system operator is able to provide suitable incentives for network ancillary services.
However, the opportunities provided by smart grids will also increase the complexity of the system,
making its management even more challenging. In particular, traditional centralized control schemes,
even if modern optimization techniques are applied, may not be adequate for this liberalised scenario.
In this view, a novel procedure has been recently proposed for a distributed coordinated power
management of MV electrical networks, aimed at prompting a price-responsive participation of active
customers in the voltage and frequency network regulation [1]. Responsive elements are activated
sequentially like in a token ring computer network, where data is transmitted sequentially from one
node to the next with a control token circulating around the ring nodes.
This work focuses on the potentiality of Electric Vehicles aggregators (EVa), such as swapping
stations and car parking, to provide valuable reactive support to the grid. In fact, their expected large
penetration and their ability to provide reactive power through the converter interface of the
recharging infrastructures and bidirectional flows of on-board batteries provided by Vehicles-to-Grid
(V2G) technology [2] make EVs and their recharging infrastructures, if suitably managed, important
resources for the electrical network regulation [3].
The load curves for EVs aggregators have been assumed considering statistical data, available in
literature [4] and adapted to typical Italian car transport scenarios. The electric vehicles battery
recharging management defines an active power profile for the EVs aggregators (which depends on
the specific transport requirements of the served electric vehicles), on the basis of which their time
varying reactive power availability may be assessed and they can participate as responsive players in
the proposed distributed voltage regulation scheme.
Through a case study application, it is shown that EVs aggregators could meet responsively the grid
power requirements and actively participate to the network regulation.

2. ELECTRIC VEHICLES AGGREGATORS: A NEW RESOURCE FOR SMART GRIDS

EVs are expected to become important actors for the electrical grid, due to to their high penetration in
the next years and their need to plug-in to the grid to recharge the on-board batteries. This means
further, not easily predictable load to be supplied by the grid. On the other hand, EVs also represent a
new opportunity for the electric system, being able to exchange active and reactive power with the
grid thus opening to the possibility of participating in services for the grid support [5,6].
A high penetration of EVs strongly depends on the establishment of a suitable coordinated
management of all resources in the electrical network and their active participation on this
management. To avoid critical situations and incentivize a virtuous behavior of elements, EVs
shouldn’t be considered as passive loads, i.e. just batteries to be recharged according to driver needs.
Considering that cars are parked on average more than 90% of the time, EVs should take advantages
from their potentiality as distributed storage devices, participating in Distribution System Management
and network regulation, benefiting from the remuneration provided by the grid for these services.
The possibility of employing EV batteries for providing ancillary services by exchanging active power
with the grid has been studied for more than a decade, and it is recognized that the main drawback is
the detrimental effect on the batteries lifetime.
Considering the expected changes in ancillary service markets scenario and degradation aspects of
batteries, the reactive power provided by recharging infrastructures seems to be more interesting to
exploit. In fact, the reactive power market is expected to develop in the future, introducing new
opportunities of remuneration, whereas the reactive power exchanged with the grid does not affect the
state of charge of on-board batteries and thus their lifetime [7].

1
For either passively supply a high number of EVs or actively use them for grid support services, a
coordinated management of the resources is fundamental. Smart meter applications will allow
capillarity and differentiation of price signals in a smart grid environment and transmission of
information regarding plug-in vehicles, such as State of Charge (SOC) and driver requirements
(mainly when a fully charged battery is needed). A centralized control will have to coordinate the
presence of EVs, SOCs and time constraints in order to match active and reactive power requirements.
Even so, the coordination of single EV owners and their participation in energy markets appear a hard
goal, whereas it appears more likely that aggregation of EVs, such as car parking and swapping
stations, will actively participate in the network management. Firstly, they have the critical mass for
providing suitable infrastructures for these services. Secondly, a coordinated management of resources
and participation in ancillary services cannot be dependent only on the availability of vehicles
connected to the grid relying on driver needing. Driver needing is a priority, but it is possible to
motivate a desired behavior trying to change trends with suitable price signals or variable tariffs, that
represent a key factor to modify the unpredictability of EVs which move from/to different locations,
referring in particular to ancillary services that are remunerated depending on a contracted available
capacity. Consequently, the availability of different price signals for ancillary services would be more
feasible with aggregation of vehicles, even if smart meter applications will allow a capillary
communication in a smart grid vision.
Three types of EVs aggregators have been identified in previous studies suitable for providing
ancillary services, namely swapping station for coaches and cars, office car parking and residential car
parking [4,5].
A swapping station (SS) is more similar to traditional electrical storage devices because the
availability of batteries is always ensured, but the actual power availability depends on the recharging
batteries service. In fact, swapping stations are sites for replacing depleted batteries with fresh ones. If
used to provide a service to coaches, a SS very close to traditional storage, because the swap of
batteries occurs regularly (usually twice a day, when daily and nightly services start), whereas the
management of SSs for private cars is conditioned by drivers requirements.
Availability of batteries in car parking depends on drivers' habits and on the location of car parking.
For example, an “office” car parking receives cars during the day, but it has not the possibility to
provide active power during the night. On the other hand, a “residential” car parking has the highest
availability of batteries during the night. These characteristics make car parking less appropriate for
active power regulation due to the high number of constraints (less batteries, variability of incoming
and outgoing EVs and status of charge of batteries). The availability of batteries, that depends on
drivers' habits, requirements of recharged batteries and the degradation aspects of batteries makes
active power management difficult, even if some services match with these characteristics.
For this reason, the EVs participation on reactive power management, although not deeply
investigated yet, appears to be more interesting, particularly because the reactive power does not affect
the state of charge of batteries and even the converters losses could be withdrawn from the supply
network, although at any time the available reactive power would inversely dependent on the actual
active power flowing through the converter. From this perspective, car parking are suitable to provide
voltage regulation, thanks to the lower recharging requirements than SSs, which are dedicated to
recharging service. This means, higher reactive power bounds within which the reactive power
required by the grid could vary.

3. FLEXIBLE MANAGEMENT PROCEDURE FOR MV DISTRIBUTION SMART GRIDS

A novel flexible power management procedure applicable to smart grids has been recently proposed
which has a token ring like approach and appears to be particularly suitable to manage distributed
resources. Without going into deeper details, which may be found in [1, 8], it is worth outlining here
the principles of the method. To this aim, a pictorial representation of the distributed coordinated
network management (the meaning of the graphic symbols are given in the figure caption) is given in
Fig. 1.
The network supervisor is the higher level which coordinates areas at the lower level. Depending on
type and location of violations, responsive elements are grouped in so-called “regulating areas”.

2
Figure 1: Distributed coordinated power management scheme. Blue and green areas are voltage
and current areas respectively with their concerning active elements. Red circles and
triangles represent voltage and current tokens respectively. Yellow and violet arrows
show token movements, whereas dark yellow arrows represent exchange of
information with the network supervisor of each areas.

Referring to Fig. 1, the network management procedure can be schematically summarized by the
following steps:
1. some violations occur and the relevant regulating areas are identified;
2. tokens are simultaneously sent by the network supervisor to the concerned areas (tokens are
pictured as red circles and triangles for voltage and current regulation respectively);
3. tokens go to the first customer of each area;
4. each customer receives its own token and replies to the network supervisor,
5. the network supervisor receives the answer back from each customer-recipient and checks the
network status;
6. if violation still exists, another token is sent to the subsequent customer of the area, otherwise the
goal has been achieved and the procedure stops;
7. when the last customer of each area replies to its token, if violations still exist remuneration is step
increased and the procedure from step 2 to 7 is repeated;
8. if new violations occur, new tokens are created and sent to the relevant areas concurrently with the
other tokens (step from 1 to 8).
Within each regulation area each recipient receives the token, for example an active or reactive power
variation request against a remuneration, and may provide a reply action to the system operator. The
customer-recipient may decide to provide only part of the required power variation, the total amount
or nothing, depending on its own economical benefit deriving from such action. In this way, the
complexity of the management problem is reduced, avoiding complex optimisation problems that
requires a large amount of data.
A key element for a stable and efficient regulation is the choice of the regulating areas which depends
on the type of violation to be solved, the network structure and the availability of distributed
regulating power in the area itself. Different criteria have been developed for dynamically define the
regulating areas, and work is in progress for further improving the identification procedure of adaptive
areas, details of which are beyond the scopes of the present paper.
In this work the analysis is focussed on reactive power management for network voltage regulation,
and the active power management (current regulation) has been intentionally inhibited, although the
procedure can inherently perform the two types of regulation concurrently.
The criteria here adopted for the identification of voltage regulating areas are (i) close electric distance
of the nodes, (ii) similar initial voltage values and (iii) availability of responsive devices with
sufficient reactive power margins in order to avoid useless token circulation among recipients unable

3
to provide the required reactive power. As regards the latter requirement, it may happen that this
requirement is not satisfied during the day, in which case that area is aggregated with the closest area
satisfying the criteria. Conversely, if further distributed resources becomes available during the day,
the procedure regulating areas may be identified enabling a finer regulation.
The substation on load tap-changer transformer is coordinated with the management procedure.
Namely, before starting the regulation procedure, the transformer tap is adjusted so as the downstream
network bus voltages are centred around 1 pu. It may easily be demonstrated that this action
significantly reduces the reactive power required from the distributed resources in order to achieve the
desired regulation target.

APPLICATION
The case study is focused on the investigation of potential benefits of Electric Vehicles recharging
infrastructures to the distributed reactive power management of the system. The electric network,
shown in Fig. 2, is a 20 kV 32-bus radial network, connected to the bulk grid (slack bus N1, 132 kV)
through an on-load tap-changer transformer (OLTC).

Figure 2: 32-busbars case study network. Coloured rectangles indicate the regulating areas
whose pilot nodes are identified by dashed red circled labels - yellow dashed line
identifies extra regulating area in Scenario c).

Feeder A (busbars N3 through N18, 8.65 km) has a surplus of generation summing up to 6.75 MW/2.5
MVAr load and 17 MW/0 MVAr generation, whereas feeder B (busbars N19 through N32, 10.5 km)
loads and generators summing up to rated power of 7.25 MW/3.8 MVAr load and 4.7 MW/0 MVAr
generation. Two electric vehicles charging aggregators (EVa) have been considered, supplied
respectively by feeder A (office type, sized 3 MVA, connected to bus 15) and feeder B (residential
type, sized 3 MVA, connected to bus 31), whose daily active load profiles are reported in Fig. 3.

a) Reference scenario
Without any voltage regulation management, this network is operating under unbearable voltage
conditions, mainly caused by the presence of several DGs, as described below.
Loads vary during the day according to residential, commercial, tertiary and industry load profiles,
whereas generators vary with wind, photovoltaic, and co-generation production profiles, giving rise to
a daily bus voltage variation as shown in Fig 4(a), which is the result of a 24-hour load flow analysis
for the reference case without any voltage regulation.

4
 Figure 3: Daily load profiles of residential and office car parkings

a) b) c)
Figure 4: Bus voltage variations with distributed voltage regulation:
(a) Unregulated reference scenario
(b) Regulated scenario without EVa participation
(c) Regulated scenario with EVa participation

For each busbar the horizontal bars indicate the voltage variation range (green or red filled when
voltage is within or outside allowed limits respectively).
It clearly appears that feeder A, characterized by a surplus of generation, have unacceptably high
voltage levels on a great part of his busbars, whereas feeder B bus voltages have a large daily
variation, reaching levels just below the minimum allowed (N32).

5
b) Regulated scenario without EVa participation
The token ring like distributed regulation procedure outlined above has been implemented and applied
to the system.
In the first instance the two EVa have been considered as mere loads, i.e. just absorbing the required
active power, according to the load profiles of Fig. 3, at unity power factor. Therefore the active
elements participating to the reactive power regulation are the 7 distributed generators; on the basis of
their generation profiles and the network voltage distribution, four regulation areas are identified,
pictorially represented by the coloured areas in Fig. 2.
Application of the distributed regulating procedure, coordinated with the tap-changer transformer,
produces the results of Fig. 4(b). It is noticeable how the voltage variation scenario is greatly improved
with respect to the reference case, although few buses on both feeders still experience voltage levels at
times trespassing the permitted limits. Such voltage violations occur at different times of the day,
caused by shortage of reactive power availability by the distributed generators, which have to respect
their own capability limits.
It should be noted that in the reference scenario (without regulation) feeder B busbars experienced
voltage problems only marginally, whereas with the regulation their levels result more depleted. This
is the effect of the tap-changer, which operates, as mentioned above, trying to center the network
voltage variation with respect to 1 pu, in order to facilitate the subsequent reactive power regulation
action that is aimed at generally improving the network voltage distribution, although single busbar
voltages may deteriorate (as happens in this case for N31 and N32).

c) Regulated scenario including EVa participation

In this case the two EVa have been enabled to participate to the regulation game with their available
reactive power. As a consequence, an extra regulating area is automatically identified, indicated in Fig.
2 by the yellow dashed rectangle (pilot node N32).
The active power load profiles of Fig. 3 have been used to assess the time varying reactive power
bounds, introduced as upper and lower constraints used in the decisional step of these distributed
reactive power sources participating in token ring like procedure.
The reactive power bounds are represented in Fig. 5 by the red bars (a – residential, b - office),
whereas the light green bars indicate the actual hourly reactive power supplied by the two EVa . It can
be noted that, given the different voltage requirements of the two feeders, the two EVa are called to
provide reactive power of opposite sign.

Figure 5: Hourly reactive power bounds (red bars) and actual reactive power supplied (light
green bars) by the EVa.

6
Application of the distributed regulating procedure, coordinated with the tap-changer transformer, and
including EVa participation produces the results in Fig. 4(c). It’s interesting to note that EVa
contribution not only allows all bus voltages to stay within the limits at any time but also to
considerably reduce the daily voltage variation of the majority of the busbars. In addition, the new
regulating area (pilot node N32) is particularly beneficial to solve the local low voltage problems
arising in such region and highlighted in Fig. 4(b).
Finally, Fig. 6 reports a comparison of the voltage profiles along the two feeders at 13.00 hour,
representing the critical time period with the of highest voltage level reached by busbar N18,
respectively without regulation (plain curves, reference scenario) and with regulation (regulated
scenario including EVa participation, dashed curves), from which the effectiveness of the proposed
procedure may be qualitatively appreciated.

Figure 6: Feeders voltage profiles at 13:00 hour without regulation (plain curves,
reference scenario) and with regulation (regulated scenario including
EVa participation, dashed curves)

CONCLUSION
Electric Vehicles aggregators, such as car parking and swapping stations, might have a great
potentiality for voltage regulation services in distribution networks.
The possibility given by 4-quadrant charger converters to exchange reactive power with the grid
without affecting the SOC of batteries could be efficiently exploited in a smart grid scenario, where
novel flexible management procedure specifically for coordinating distributed resources (such as the
token ring approach with adaptive regulation areas employed in this work) may be designed and
implemented.
Depending on service characteristics of EV aggregators, management of aggregator could differ and
focus on different aspects. If swapping stations service may be easily forecasted, “residential” and
“office” car parking highly depend on the transport requirements and habits of the drivers, making
their management as ancillary services providers more complex.
However, the introduction of regulatory and incentivizing schemes for the distributed reactive power
support services would likely stimulate the necessary investments and development of the
infrastructure required exploit this potential resource.

7
BIBLIOGRAPHY
[1] L. Carradore, R. Turri, “A Token Ring Procedure for Distributed Co-ordinated Regulation of
Active MV Networks”, ICHQP 2010, Bergamo, Italia, 26 −29 Sept . 2010
[2] W. Kempton, J. Tomi´c, Vehicle-to-grid power implementation: From stabilizing the grid to
supporting large-scale renewable energy Journal of power sources, ISSN 0378-7753, CODEN
JPSODZ, 2005, vol. 144, no1, pp. 280-294
[3] A. Simpson, A New Dimension for ITS: Plug in Vehicles with Smart Grids, presentation
available online at http://www.itssummit.com.au/ Portals/58/Andrewn%20Simpson.pdf,
Australian ITS Summit, 20 November 2009
[4] L. Carradore, R. Turri, “Electric Vehicles Participation in Distribution Network Voltage
Regulation”, 45th Universities Power Engineering Conference 2010, Cardiff, UK, 31 August −3
Sept. 2010
[5] Carradore, R. Turri, L.M. Cipcigan, P. Papadopoulos, “Electric Vehicles as flexible energy
storage systems in power distribution networks”, EVER 2010, Montecarlo, Monaco, 25-28
March 2010
[6] P. Papadopoulos, L.M. Cipcigan, N. Jenkins, I. Grau, “Distribution Networks with Electric
Vehicles”, 44th UPEC 2009, Glasgow, 1-4 September, 2009
[7] M.C. Kisacikoglu, B. Ozpineci, L.M. Tolbert, “Examination of a PHEV Bidirirectional Charger
System for V2G Reactive Power Compensation”, Applied Power Electronics Conference and
Exposition (APEC), 2010 Twenty-Fifth Annual IEEE, 21-25 Feb. 2010
[8] L. Carradore, R. Turri, “Distributed Co-ordinated Power Management in MV Electrical
Networks”, 45th Universities Power Engineering Conference 2010, Cardiff, UK, 31 August −3
Sept. 2010