CIRED 20th International Conference on Electricity Distribution Prague, 8-11 June 2009

Paper 0796

APPLICATION OF BIDIRECTIONAL ENERGY MANAGEMENT INTERFACES FOR

DISTRIBUTION GRID SERVICES

Jan RINGELSTEIN David NESTLE

Institut für Solare Energieversorgungstechnik (ISET e.V.) – Germany
jringelstein@iset.uni-kassel.de dnestle@iset.uni-kassel.de

predictably and reliably. At the same time every BEMI

ABSTRACT provides metering data and measurement information from
the LV grid connection point, such as grid voltage and
In this paper we present a strategic approach for energy impedance.
management of distributed electric resources in low-voltage
networks with special consideration of services for the Thus, grid supervision becomes possible. The Pool-BEMI
distribution system operator (DSO). A system comprising aggregates the BEMI functions thus enabling new
Bidirectional Energy Management Interfaces (BEMIs) applications including distribution grid services, which the
developed and tested at ISET, carrying out decentralized ESP offers to the DSO.
decisions and grid supervision is introduced. The paper
presents operational states and possible applications of the Transmission Balancing
system group Balancing
proposed system in the distribution grid as well as a operator Balancing
manager energy market
energy
simulation designed for researching the system’s behaviour. Hour
Finally first results are given on the simulation-based reserve
development of an algorithm implementing a new approach ces Energy service
rvi EEX
for incentive-based voltage control, which is one of the se provider
id
gr
n d Pool-BEMI
Spot market
system’s advanced applications. utio g an trading
r ib ri n g R e
t te in
Dis m
Me Bill t
n
me ot
em
on
ge

Distribution
ma rgy

INTRODUCTION ito
rin
na

system
Ene

g
operator ~
Since the share of distributed generation (DG) in the grid is =

growing, there has been increasing awareness of arising grid BEMI BEMI
BEMI BEMI
~ ~
problems, but also potentials for optimizing DG operation. ~
~

The low voltage (LV) grid, where up to 50% of electricity

consumption takes place, offers potential for management
of controllable distributed electrical (CDE) units (loads or
generators) [1]. Up to now most of this remains unused due
to missing online control and grid supervision devices.
Energy management systems solving this problem need to
consider the requirements of the DSO as well as the large
number of independent grid customers operating CDE units.
STRATEGIC APPROACH
The Bidirectional Energy Management Interface (BEMI)
developed at ISET is an energy management system
designed for installation at LV grid connection points [2].
BEMI economically optimizes the operation of locally
connected CDE units according to variable tariffs for
consumption and generation given by a Pool-BEMI
operated by an energy service provider (ESP). This BEMI
system is shown in Fig. 1. Grid customers can obtain
information about the variable tariffs, energy cost and
device schedules, while BEMIs automatically carry out
CDE unit switching and operation supervision. The ESP
also acts as energy supplier and sells energy generated by
DG on higher-level markets. It could be shown that if many
BEMIs (>1000) are installed in the network, the ESP can
influence overall CDE unit operation by variable tariffs
Energy trading
Energiehandel
Services
Dienstleistung
Meter reading
Figure 1 : BEMI System in the liberalized energy
market © ISET
The patented BEMI concept was implemented and tested in
ISET’s DeMoTec laboratory with two household test
benches. Within this test, it could also be shown that the
BEMI is able to manage a house island grid using local
CDE units. A field-test in Mannheim showed the
applicability of the approach in real households. Further
field-tests are ongoing.
BEMI SYSTEM OPERATIONAL STATES
In order to technically approach the BEMI system, a
classification of operational states is needed. The system
uses information and communication technology (ICT) and
internet-based transmission of tariff and measurement data
based on the IEC61850 standard [3]. BEMI and Pool-BEMI
itself are ICT systems as well. Though these new
components give a broad range of possibilities, they also
give rise to new fault sources. The same is valid for
distributed generators, which at times may have to be
derated due to grid congestions.

CIRED2009 Session 4 Paper No 0796 Page 1 / 4

CIRED 20th International Conference on Electricity Distribution Prague, 8-11 June 2009

Paper 0796

Considering this context, Fig. 2 proposes a BEMI System
state chart from the ESP’s perspective. Operation states are
divided into three groups:
Normal (green): all customers are connected, all ICT
elements are operational, CDE unit operation as planned
Compromised (yellow): e.g. one or more BEMI out of
operation
services in normal state - where they are to support the DSO
in order to retain this state - and services in compromised or
disturbed state, where they should support the DSO in
leading system state back to normal. Fig. 3 proposes a
classification of possible applications in accordance to the
state chart.
More critical operational states

Disturbed (red): at least one customer cut off from public Normal Compromised Disturbed

grid Optimal Secure Adapted Derated ICT-

disturbed
Isolated Reconstr.
with island
Reconstr.
w/o island
Load/generation profile influencing
Peak load reduction DSO blackout notify

Optimal Secure Adapted Secure DER switch off for

Balancing energy provision
grid maintainance
O A V I P B D O A V I P B D O A V I P B D Fault detection and
Automatic grid state and customer supply supervision, grid diagnosis
localization
Congestion supervision and management
Pool-BEMI or repair BEMI or repair grid load-flow
Customer
ICT failure ICT failure congestion change Island grid
blackout
mode
inform.
Isolated ICT Disturbed Derated Grid recon-
O A V I P B D O A V I P B D O A V I P B D struction
Local voltage control and power quality optimization
Customer information on consumption, generation mix and system state
grid
failure

Grid Disturbed Grid & ICT Disturbed Disturbed & Isolated Figure 3: BEMI System applications according to state
O A V I P B D O A V I P B D O A V I P B D
© ISET
grid reconstruction by DSO with grid reconstruction by DSO
house- or residential island mode

Reconstruction
with island grid
O A V I P B D
O: Optimal operation
A:
V:
No quick load adaption needed
No grid regions blacked out
I: No island grids
P: Pool-BEMI in full operation
B: All BEMI in full operation
D: No DG derated or off-line
Figure 2: BEMI System state chart (not all state
transitions shown) © ISET
Note that the system state changes due to failures on the one
hand and repair or reconstruction on the other hand. For
example, state could change from optimal to derated due to
a grid congestion and the subsequent derating of DG. If a
grid failure were to occur in this situation and some BEMI
would cease operation, state would change to grid and ICT
disturbed. If blacked out customers were allowed to build
island grids in accordance to the DSO, another state change
would occur to reconstruction with island grid. After grid
reconstruction and restoration of all ICT functions, system
state would return to secure or optimal.
BEMI SYSTEM APPLICATIONS
An energy management solution aiming at economical and
technical integration of loads and DG typically causes
significant changes in grid operation when compared to
today’s situation. Technical impacts of a high share of DG
on grid operation have been researched in various studies,
focusing on specific DG technology or systematically
reviewing overall impacts [4]. BEMI System distribution
grid services are principally to support grid operation under
these circumstances. They can roughly be devided in
Reconstruction
without island grid
It should be noted that some applications in disturbed state
O A V I P B D need communication within the BEMI System during power
failure. This can be achieved by using uninterruptibe power
Normal state supplys for the BEMI’s computing core and ICT systems
equipped with backup power. E.g. broad-range radio
Compromised state
frequency communication could still be available during
power outage.
Some applications can be implemented by mere data
exchange and usage of BEMIs local intelligence. For
example, supervision of grid voltage can be carried out by
BEMIs alarming the Pool-BEMI when measurement values
exceed voltage thresholds set by the DSO. The Pool-BEMI
again can assess the overall network state by requesting
voltage measurements from additional BEMIs and inform
the DSO’s grid control center about the situation. Since
voltage rise is the first sign of a possible line overload in the
low voltage grid, the DSO could perform a switching
action, e.g. perform a ring closing. With this example it can
be seen that using BEMI and Pool-BEMI intelligence is
crucial in order to reduce the need for wide-area
communication, which can otherwise become a bottleneck
for the entire system operation.
There are also more advanced applications possible that
need a combination of distributed measurements and energy
management by variable tariffs. However, it has to be kept
in mind that for security applications with tough real-time
requirements solutions have to be found that do not need
communication, since ICT quality-of-service is typically
limited. This applies, for example, to selectively switching
of distributed generators that feed a grid fault during
automatic reclosing.

CIRED2009 Session 4 Paper No 0796 Page 2 / 4

CIRED 20th International Conference on Electricity Distribution Prague, 8-11 June 2009

Paper 0796

BEMI SYSTEM SIMULATION
Algorithms for advanced applications typically can not be
developed using BEMIs installed in real-field due to the
number of BEMIs required and the security of supply of
customers. Hence, a simulation was developed at ISET that
around 12:00 due to the changed tariff. It can be expected
that a considerable efficiency increase in using locally
generated energy is possible if photovoltaic generators were
installed in the considered grid region.
30 30

Tariff [ct/kWh]

Tariff [ct/kWh]
is able to model BEMI behavior in the distribution network. 20 20

This simulation consists of three modules. The first module 10 10

implements a BEMI and household simulation. Fig. 4 shows

1200
a simplified UML diagram for this module, which was 900
Standard Profile
Total
Standard
Standard Profile
Total
Total
Profile

800 1000 SOC

SOC
implemented using C++. It includes different device classes 700
SOC
FPS
FPS
FPS

Total load [kW]

800

Total load [kW]

for typical households: state-of-charge (SOC) devices like 600
500
freezers and fixed-program-schedule (FPS) devices like 400
600

washing machines and dish cleaners. Customer usage of 300 400

200
devices was simulated using statistical approaches, which 100
200

ensure that many individual household load profiles add up Total Total
to known standard profiles. Photovoltaic generators and 4 SOC
FPS
4 SOC
FPS

Individual load [kW]

Individual load [kW]
common heat and power plants are also modeled. For CDE 3 3
unit optimization, the simulation module uses the BEMI’s
original energy management algorithms. A second module 2 2

implemented in C# models the considered functions of the 1 1

Pool-BEMI and incorporates a user interface. The third

module interfaces to a professional network calculation 00:00 04:00 08:00 12:00 16:00 20:00 24:00 00:00 04:00 08:00 12:00 16:00 20:00 24:00

software, for which PowerFactory from DIgSILENT was Day 1 Day 2

used for flexible distribution grid modeling.
Figure 5: Example simulation results © ISET

A NEW APPROACH TO VOLTAGE CONTROL

Figure 4: UML class diagram of BEMI simulation
module © ISET
Fig. 5 shows an example result. Two weekdays were
simulated with 1000 households. Approx. 50% of the total
household energy consumption was attributed to managed
loads. From the results the LV grid’s high potential for
energy management can be seen. The first day’s tariff was
flat, so the independent loads behaved matching the VDEW
H0 standard profile [5]. At the second day shown BEMI
management leads to a significant load shift to the time
In LV grids, feed-in from DG typically rises voltage levels
at grid connection points. This mainly depends on active
power because of the mainly ohmic impedances of grid
lines. Also tariff-based energy management influences grid
voltage due to its influence on CDE unit operation. High
voltages can damage connected loads. Furthermore, they are
a first sign for a possible line or transformer overload [6].
But voltage control in the low voltage network can hardly
be done by supplying reactive power. Instead, in order to
lower voltage in a grid line, active power feed-in -
preferably from DG that uses storable fuel - has to be
derated or additional loads have to be connected. In this
way the usage of DG using renewable energy sources can
be improved and grid operation with high share of DG can
be supported. For this voltage measurements distributed in
the grid are required as well as appropriate incentives for
CDE unit operators. The BEMI System includes the means
to give incentives for quick load adaption by immediate
tariff change. Therefore, a method for incentive-based
voltage control was developed comprising the following
steps:
1. The Pool-BEMI transmits voltage thresholds according to
the DSO’s specifications to every BEMI.
2. Every BEMI supervises the voltage level at its grid
connection point and alarms the Pool-BEMI if a threshold is
exceeded.
3. For switching loads on or DG off, the Pool-BEMI has to
determine an appropriate tariff change. It transmits requests

CIRED2009 Session 4 Paper No 0796 Page 3 / 4

CIRED 20th International Conference on Electricity Distribution Prague, 8-11 June 2009

Paper 0796

for tariff change proposals to every BEMI. voltage control as distribution grid service. This service can
4. BEMIs send individual tariff change proposals based on help improve grid operation with high share of DG.
local CDE unit states to the Pool-BEMI Apart from ongoing simulation-based studies, the BEMI is
5. The Pool-BEMI assesses the proposals and carries out the currently further developed in two projects funded by the
tariff change german environmental ministry (BMU) and the european
6. BEMIs reoptimize CDE unit operation after receiving the project “Smart House/Smart Grid”. The projects also
new tariffs. This leads to immediate voltage reduction by include further field testing. Current development includes
load connection or DG deration. increasing BEMI energy efficiency and testing new ICT
systems for radio based in-house communication.
The method was implemented as automatic algorithm
within the BEMI Simulation. Herefor a small network
246
shown in Fig. 6 was modeled, which resembles a quite

Voltage average L1,L2,L3 [V]

weak rural area grid containing 98 households equipped 244

with BEMI and 5 kWp photovoltaic generators each. 242

3m 39 m 240
Feeder D
HH 89 BEMI BEMI BEMI BEMI HH 98 238

236
3m 39 m
External Feeder C
Grid
234
HH 69 BEMI BEMI BEMI BEMI HH 88

8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00
0,4 kV 3m 39 m 3m 39 m
Slack- Feeder B Daytime
Node
HH 35 BEMI BEMI BEMI BEMI BEMI BEMI HH 68 Figure 7: Voltage at HH 34 without (upper) and with
3m 39 m 3m 39 m (lower curve) incentive-based voltage control © ISET
Feeder A
HH 1 BEMI BEMI HH 2 BEMI BEMI BEMI BEMI HH 34
REFERENCES

Figure 6: Simulation test grid © ISET [1] M. Braun, P. Strauss 2008, “A review on aggregation
approaches of controllable distributed energy units in
Fig. 7 shows the voltage level at household HH 34 in grid electrical power systems”, International Journal of
line A rising substantially due to high active power feed-in Distributed Energy Resources, Volume 4 Nr. 4, pp.
(upper curve). The voltage after application of the voltage 297-319
management method is also shown (lower curve). It can be [2] C. Bendel, D. Nestle, J. Ringelstein 2007, “Integration
observed that the method has led to voltage reduction of distributed generation in the future energy system in
between approx. 10:15 and 12:30. Due to the parameters of a multi-player situation – common legal and technical
the loads, voltage reduction cannot be maintained any interface for grid operation”, 19th International
longer. However, the time the voltage exceeds the 6% Conference on Electricity Distribution, CIRED,
threshold shown in Fig. 7 was reduced by 33%. This again Vienna, Austria
reduces the need for DG deration due to exceeding the [3] C. Bendel, D. Nestle, J. Ringelstein, K. Schwarz 2007,
threshold. A first result of a comparative simulation showed “Data models for the integration of the low voltage
that approx. 30% of the energy deration from photovoltaic level into grid communication and control using
generators could be saved by the method when compared to decentralised decision”, Workshop on International
the same scenario without any voltage control measures. Standardization for Distributed Energy Resources,
Herewith it could be shown that the method can support Oldenburg, Germany
avoiding LV grid congestions. [4] T. Degner, A. Shustov, J. Jäger, T. Keil 2007,
„Changing Network Conditions Due to Distributed
Generation - Systematic Review and Analysis of their
CONCLUSION AND OUTLOOK Impacts on Protection, Control and Communication
A strategic approach for enabling the energy management Systems”, 19th International Conference on Electricity
potential in the LV grid using a system based on BEMI was Distribution, CIRED, Vienna, Austria
introduced. New operational states that arise with [5] Vattenfall Trading Services GmbH, “The wholesale
introduction of this system were defined and applications product GH0”, http://www.ohv.nl/uploads/files/
were identified. The simulation system introduced has GH0%20brochure%20engels.pdf?PHPSESSID=ac221
already proven to be of high importance for BEMI System 5d27aca57b24db54bfa0c3f2462, Jan 09, 2009
algorithm development. It was shown that a method could [6] R. Witzmann, G. Kerber 2007, “Aufnahmefähigkeit
be developed that combines the BEMI System’s capabilities der Verteilnetze für Strom aus Photovoltaik”,
such as to implement the application of incentive-based Elektrizitätswirtschaft 4/2007, pp. 50-54

CIRED2009 Session 4 Paper No 0796 Page 4 / 4