Professional Documents
Culture Documents
Status of PV in Germany
As mentioned above, as of September 2012 more than
31 GWp of PV capacity was installed in the German energy
system. The installed capacity of PV systems is now higher
than that of any other RES, as well as that of any conventional
power source (see Figure 1). Nevertheless, the cumulative
produced energy from PV only accounted for about 6% of
istockphoto.com/Mark Soskolne
march/april 2013
1540-7977/13/$31.002013IEEE
08
Se
20
06
20
04
20
02
20
00
20
98
20
96
19
94
19
92
19
19
19
90
80
Impact of High PV
Penetration on the Grid
Three main issues can be identified regarding high PV penetration in distribution grids:
1) Reverse power flows in the distribution system: In
distribution systems where the local PV generation
exceeds the local load demand, reverse power flows
In what follows, we highlight possible solutions to the technical issues outlined in the previous section. Note that the
technical challenges also have an economic impact on the
stakeholders involved.
0.4
1020
PV
110
Wind
Biomass and Others
220380
Hydro
0
10
15
20
Installed RES Capacity (GW)
25
30
figure 2. Distribution of RES over the typical nominal voltage levels of the
German power system (source: energymap.info).
56
60
Electricity Generation
and Load (GW)
50
40
Maximum Generation of
PV: 21 GW~39%
30
20
10
a.
m
2:
00 .
a.
m
4:
00 .
a.
m
6:
00 .
a.
m
8:
00 .
a.
m
10
.
:0
0
a.
m
12
.
:0
0
p.
m
2:
00 .
p.
m
4:
00 .
p.
m
6:
00 .
p.
m
8:
00 .
p.
m
10
.
:0
0
p.
m
.
0:
00
PV
Others Sources
Load
Time (h)
10
PR
PS
-10
PT
-20.5 kW|~12:00 p.m.
-20
0
Time, t (h)
(a)
24
245
239 V|~12:00 p.m.
Voltage, V (V)
Reducing Congestion
235
VR
VS
VT
225
220 V|~6:30 a.m.
215
0
Time, t (h)
24
(b)
figure 4. Example of (a) load flow and (b) voltage at the transformer in one
require communication devices.
LV distribution system with high PV penetration over the period of one day
The distributed energy resource
(source: University of Stuttgart).
(DER) unit reacts to specific
grid situations according to predefined parameters
controllable substations. These substations enable
and droop functions, as well as measurements (e.g., of
autonomous control of the LV grid sections, which can
voltage or frequency) at its point of common coupling.
increase reliability and economic efficiency. Furthermore, parameters from the surrounding systems and
Decentralized control strategies are communicationthe higher-voltage levels can be included.
based control strategies that interact with an intermediate level in the grid. For example, LV systems
Central control strategies describe communicationcan be clustered into different cells using intelligent,
based control approaches from the distribution
march/april 2013
51.4
51.2
51
50.8
50.6
50.4
50.2
50
49.8
49.6
Zone West
Zone South East
Zone North East
49.4
49.2
10:20:00.0 p.m.
10:19:30.0 p.m.
10:19:00.0 p.m.
10:18:30.0 p.m.
10:18:00.0 p.m.
10:17:30.0 p.m.
10:17:00.0 p.m.
10:16:30.0 p.m.
10:16:00.0 p.m.
10:15:30.0 p.m.
10:15:00.0 p.m.
10:14:30.0 p.m.
10:14:00.0 p.m.
10:13:30.0 p.m.
10:13:00.0 p.m.
10:12:30.0 p.m.
10:12:00.0 p.m.
10:11:30.0 p.m.
10:11:00.0 p.m.
10:10:30.0 p.m.
10:10:00.0 p.m.
10:09:30.0 p.m.
49
figure S1. Frequency recordings after the split of the ENTSO-E control area in 2006 (source: ENTSO-E, final report
on the disturbances of 4 November 2006).
50.251.5 Hz
Frequency
of the Grid
Frequency
of the Grid
3P
Power Reduction 3P =
40%* Actual PV Power Output/Hz
Power Reduction
of PV System
Fraunhofer
IWES
Control
Center
Control Signal
Optional Control Signal
Communication Interface
Distribution
Management
System (DMS)
HV
MV
MV
MV
PV
LV
Micro
DMS
LV
LV
Load
Central
Decentral
Local
60
Strategy
Description
Regulatory Framework
march/april 2013
3,000
Only Grid
Reinforcement
Costs Can Be
Lowered by
Self-Consumption
2,500
Strategy E
2,000
4
1,500
Strategy B
Strategy C
4
Strategy D
4
1,000
500
1
5
12
5
5
12
figure 8. Cost comparison of different voltage-control strategies (source: Stetz, Marten, and Braun, 2012).
In the research project called Active, Intelligent LowVoltage System, the Fraunhofer IWES and its partners
investigate decentralized control methods. The aim of
the components and control approaches developed is to
assure that the voltage within an LV system is kept within
the allowed voltage tolerance bandunder the boundary
condition of minimal grid reinforcement.
The controllable medium- and low-voltage (MV/LV) distribution transformer is able to change the transformer ratio using
an integrated OLTC. The voltage curves of the MV and LV
are thus decoupled. In this way, a voltage control method for
the complete LV system can be achieved that is independent
of grid topology. The compact design of the newly developed
OLTC transformer allows the mounting form of standard distribution transformers to be retained. Additionally, a robust and
maintenance-free design guarantees about 700,000 switching
operations. These advantages are accompanied by additional
hardware and a marginal increase in no-load losses.
ieee power & energy magazine 61
0.4 kV
20 kV
MV
Network
CP
CP
CP
.
=
Transformer
with OLTC
CP
CP
CP
.
=
CP
CP
.
=
CP
CP
.
=
.
=
CP
CP
CP
.
=
Voltage
Control
CP
.
=
CP
.
=
.
=
Control
Substation
CP = Connection Point
figure 9. Principal design and components of an active, intelligent LV system (source: Blo, 2012).
System Concept
Conventional
System Concepts
Depending on the particular case, the available components can be used in different combinations, from which
several appropriate system concepts can be derived (see
Figure 10). The voltage rise caused by DER in the LV grid
is not allowed to exceed 3%. Installing an OLTC transformer allows the voltage rise to be mitigated. A voltage
rise of up to 10% is therefore possible because the voltage
at the busbar at the transformer can be controlled over a
large range.
Substation
PV Inverter
Type of Communication
Passive
Passive
None
Passive
Active
Passive
Active
Active
Passive
Active
Active
P
Active Inverter
P, Q
U, P
Smart Substation
Q, P
P,Q
Active Substation
U
Active and Smart
Substation
U, P
Q, P
U
P, Q
figure 10. Various concepts for an LV system, including the MV/LV substation being considered for decentralized control
approaches (with communication) and local control approaches (without communication) (source: Blo, 2012).
62
march/april 2013
SCADA
from the system operator. Dynamic information (e.g., information on bus voltages and on active and reactive power
levels of DERs) is acquired from distributed state estimation of the grid. This is because such information is usually
only available for a few locations in the system, where sensors are installed.
Figure 11 outlines the architecture of the control mechanism developed. The voltage controller contains a grid
simulation unit and an optimization unit. The optimization
1.07
Umax
Umin
Umax
Umin
1.03
0.97
0.95
0:00
a.m.
6:00
a.m.
12:00
p.m.
Time (h)
(a)
6:00
p.m.
0:00
a.m.
0:00
a.m.
6:00
a.m.
12:00
p.m.
6:00
p.m.
0:00
a.m.
Time (h)
(b)
figure 12. Minimum and maximum voltage magnitude (a) without and (b) with central voltage control (source: Yan
et al., 2012).
march/april 2013
unit uses the grid simulation unit to assess power quality under different grid settingsin this case, tap positions and reactive-power feed-in of distributed generators.
This is done via load flow calculations, which require the
above-mentioned static and dynamic information of the
system. Given the desired quality of a grid setting, the
optimization unit will apply a heuristic search strategy to
find a grid setting that minimizes the number of voltage
violations in the grid with minimal reactive power and
tap changes.
Such a heuristic-based central voltage controller was
demonstrated to be able to resolve voltage violation in distribution systems. The controller was tested given the topology
of a real rural distribution system, as well as real solar radiation data and synthetic load profiles (which were generated
using real profiles). The grid contains two MV branches with
a total of 40 MV/LV substations. A total of 11 distributed
generatorsnine PV generators and two hydro generators
were installed in the grid. The voltage tolerance band was
assumed to be !3%.
The maximal and minimal voltage profiles for the simulation of one summer day are depicted in Figure 12. Without
any sort of voltage control, as shown in Figure12(a), the grid
experiences overvoltages around noon, which are caused by
the high feed-in of the PV generators. In addition, peak loads
in the early evening cause undervoltages. Using the voltagecontrol strategy, as shown in Figure 12(b), it is possible to
hold the voltages in the grid within the tolerance band of
!3% over the entire day. These results are promising, as they
demonstrate that voltage violations within a distribution system can be solved utilizing the provision of reactive power
by distributed generation.
64
Biographies
Jan von Appen is with Fraunhofer Institute for Wind Energy
and Energy System Technology.
Martin Braun is with Fraunhofer Institute for Wind
Energy and Energy System Technology and the University
of Kassel.
Thomas Stetz is with Fraunhofer Institute for Wind
Energy and Energy System Technology.
Konrad Diwold is with Fraunhofer Institute for Wind
Energy and Energy System Technology.
Dominik Geibel is with Fraunhofer Institute for Wind
p&e
Energy and Energy System Technology.
march/april 2013