Available online at www.sciencedirect.

com

Energy Procedia 25 (2012) 82 – 92

PV Asia Pacific Conference 2011

PV Grid Integration – System Management Issues

and Utility Concerns
Raymond Hudsona,*, Gerd Heilschera,b
a
BEW Engineering, a DNV Company,
2303 Camino Ramon, Suite 220, San Ramon, CA 94583 USA
b
Ulm University of Applied Sciences, Eberhard Finckh Str. 11, D-89075 Ulm, Germany

Abstract

Solar PV systems have evolved from small stand-alone applications through residential and commercial systems to
MW-scale installations and groups of systems that feed significant energy into the utility grid. High PV penetration
levels create a number of challenges for the management of the operation of the utility grid. This paper will cover the
current and historic status of the integration of renewable energies into utility grids over a variety of global regions.
This includes grid control and network planning. The technology for the integration of PV is described including the
codes and standards governing the utility interactivity requirements. This will include short term dynamic responses
as well as power ramp rates. Geographic, technological, and meteorological factors will be discussed. The
coordination of meteorology and energy production is a key factor that is examined in detail including real world
data. The paper includes a discussion of future utility grid features to support increasing levels of PV and other
renewable generation sources.

© 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Solar Energy
© 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Solar Energy Research
Research Institute of Singapore (SERIS) – National University of Singapore (NUS). The PV Asia Pacific
Institute of Singapore (SERIS) – National University of Singapore (NUS). The PV Asia Pacific Conference 2011
Conference 2011 was jointly organised by SERIS and the Asian Photovoltaic Industry Association
was jointly organised by SERIS and the Asian Photovoltaic Industry Association (APVIA).
(APVIA).
Open access under CC BY-NC-ND license.
Keywords: photovoltaic; utility grid; integration; variability; solar resource; grid codes

1. Introduction

With Europe aiming at 20% renewable energy generation [1] and Germany or the State of California
targeting above 30% of their energy demand by 2020 [2, 3], renewable energy sources have to step into a

* Corresponding author. Tel.: +1 925 327 3011; fax: +1 925 867 3331
E-mail address:raymond.hudson@dnv.com

1876-6102 © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Solar Energy Research Institute of Singapore
(SERIS) – National University of Singapore (NUS). The PV Asia Pacific Conference 2011 was jointly organised by SERIS and the Asian Photovoltaic
Industry Association (APVIA). Open access under CC BY-NC-ND license.
doi:10.1016/j.egypro.2012.07.012
 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92 83

new role as a major energy source of the near future. The present and near future topology of the utility
grid is presented in Section 2. Section 3 introduces the variability of the solar energy resource in time and
area. It is the aim of this article to explain the challenges of integrating high penetration of photovoltaic to
the utility grid with the experiences gained in countries like Spain, Germany and the USA (Section 4).
The mitigation strategies that have been developed so far will be presented in Section 5.

2. The utility grid topology

The topology of the utility grid will have to change in a way that it facilitates the introduction of
additional renewable sources including PV. This is illustrated in Fig. 1 below. The communications and
control logic has to be extended to the low voltage grid to support distributed generation.

Fig. 1. The utility grid today (left) and tomorrow (right) [4] - the communication and control logic has to be extended to support the
integration of distributed generation.

3. The variability of the solar energy source

While the fact that the sun rises and sets each day is a certain event, a cloud crossing the sun can
potentially reduce the sunlight by 80% within one second. A solar module and the inverter which are
converting the sunlight into electricity are capable to follow these fast changes. This high variability of
the solar energy source is a concern of many utilities facing higher penetration of photovoltaic energy [5].
An example of an irradiance measurement on a variably cloudy day in Hawaii [6] is shown in Fig. 2.

The evaluation of a one year dataset with one second time resolutions from Hawaii [6] shows up and
down ramps of up to 1000 W/m2 (Fig. 3, left graph). In total, however, only about 400 up or down ramps
higher than 500 W/m2 occur. Changing to a 10 second timescale almost 7000 up or down ramps have
been counted higher than 500 W/m2, reflecting more than 50% change in power.
84 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92

Fig. 2. Global horizontal solar irradiation (blue) at Oahu Hawaii with 1 second time resolution over one day and change of
irradiation (ramps in red) [6].

100,000000%
Events of binned ramps, relative [%]

10,000000%

1,000000%

0,100000%

0,010000%
900 second

0,001000% 60 second

0,000100% 10 second

1 second
0,000010%

0,000001%
00
-9 0

0
0
0
0
0
0
0
0
1 0
> 000
0
-8 0
-7 0
-6 0
-5 0
-4 0
-3 0
-2 0
-1 0
00
00

10
20
30
40
50
60
70
80
90
0
0
0
0
0
0
0
0

10
-1

Change of irradiance (ramps) W/m²

Fig. 3. Frequency distribution of ramps in different time scales of a single sensor (left) and smoothing effect with area (right). The
average of 17 sensors within 1 km2 reduces the maximum 1 second ramp below 200 W/m2.

Wind speed and cloud patterns are the driving forces of this variability. Solar photovoltaic systems
collect the energy from the sun in a spatial dimension as well. Therefore it is important to understand how
the variability changes with the area distribution of the solar systems. Figure 3 shows in the right graph
how the maximum 1 second ramps and the events of high ramps during one year are significantly reduced
by looking at several, distributed sensors. Averaging just 2 single point solar sensors in a distance of 100
metre reduces the occurrence of ramps higher than 500 W/m2 from over 400 to about 40 events per year.
Averaging four sensors with distances from 100 metre to 1 km, no ramps higher than 500 W/m2 are
observed. The average of all of the 17 sensors spread over 1 km2 at the measurement site reduces the
maximum ramp to 200 W/m2 within one second (data source [6], evaluation BEW). The reduction of
variability due to the averaging of the resource in time and area is beneficial however the remaining
absolute ramp rates of up to 200 W/m2 are still a concern for utilities.
 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92 85

4. Challenges of photovoltaic utility grid integration

Utilities are concerned about the safe, efficient, and reliable operation of the power grid. There are
many examples of high penetration PV installations that are operating well. There are, however, some
cases which illustrate the general concerns with this issue.

PV is a modular energy source with system sizes ranging from just 1 kW to multi MW. In Germany
80% of the PV power is interconnected on the distribution network [9]. The distribution networks
historically have been designed to transfer energy from the high voltage level down to the residential load
which is connected at the low voltage level. Detailed measurements are available at the connection points
between the distribution system (DSO) and the transmission system (TSO) and visible at the distribution
grid control centres. The fluctuations of load at the residential level are estimated by standardised load
profiles. With the integration of decentralised power generation such as PV into the low voltage level, the
distribution control centres looses visibility and control over the grid, as network operators have no
measurements at the low voltage level today. Manual yearly peak current readout had been sufficient up
to now (Fig. 4a).

How the actual load profile changes at a low voltage transformer due to photovoltaic has been
analysed in [7]. Figure 4b shows how the load curve even turns to negative values, which indicates the
feeding of energy back to the medium voltage network at this test site at Ulm, Germany. The black line in
Fig. 4b represents the estimated load calculated with the standard load profiles of these customers. At this
residential area, 133 houses and just 14 PV systems with a total power of 108 kWp are connected to the
low voltage transformer. Note that if all of the houses were to be covered with PV, a total power of more
than 1 MW would feed into the installed 630 kVA transformer.

Fig. 4. (a) Distribution network - no actual measurement at the low voltage level; (b) High resolution three phase power
measurement (PL1, PL2, PL3) at a low voltage transformer shows negative load flow during a sunny day [7] .
86 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92

That the load forecast based on standard load profiles no longer applies with distributed PV power
input is explained in Fig. 5. The 2-second measurements of load and PV power on a sunny day are
displayed for a residential customer (Fig. 5a). Part of the load can be covered by the PV system (green).
The surplus power (magenta) has to be fed into the grid.

W
120

100

80

Energy [Wh]
60

40

20

00:00
00:45
01:30
02:15
03:00
03:45
04:30
05:15
06:00
06:45
07:30
08:15
09:00
09:45
10:30
11:15
12:00
12:45
13:30
14:15
15:00
15:45
16:30

17:15
18:00
18:45
19:30

20:15
21:00
21:45
22:30

23:15
Average load (150 households) Photovoltaic system
Direct use of PV Surplus PV fed into grid
Residual load of average household

Fig. 5. (a) Real time residential load (blue) and PV power input (red); (b) Resulting yearly load profiles [8].

Designing the PV system (Fig. 5b in red) with an energy output equal to the energy demand of a
residential customer (dark blue), only one third of the PV power could be used directly (green). This
fraction might be higher in hotter climates with strong cooling load demand. The load profile of this
customer however changes substantially (light blue). Night time load becomes higher than midday load
and the energy peak time is in the evening [8].

Fig. 6. (a) Voltage band influenced by distributed PV input. Without PV input the voltage decreases over the feeder line. With PV
input also the voltage increase has to be considered in the grid design [9]; (b) Network planning and operation at transmission and
distribution level are not interlinked today. The graph on the right states different tasks, influencing parameters and suitable tools
needed for a complete system analytical model [10].
 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92 87

The dimensioning of the distribution network is determined by the power to be transmitted and the
voltage limits that have to be met at every connection point. Because historically there has been only load
in the low voltage network, the voltage drop from the transformer to the customer has been evaluated
once during the network planning. With distributed PV power input to the low voltage network this
equation changes in terms of transformer short circuit power, phase shift at the connection point, actual
load and the active and reactive power output of the PV inverter (Fig. 6a). In many cases the voltage band
today is the limiting factor for the possible degree of power input to the low voltage network. A real
world example of this is shown in Fig. 7 below [10].

Fig. 7. The two graphs show the results of a low voltage network calculation with (left graph) and without (right graph) distributed
generation. The local power input leads to a voltage rise above 1.05 p.u. in some grid sections [10].

To enable additional connection of photovoltaic power expensive grid extensions are undertaken. For
the technical and financial optimisation of the integration of photovoltaic to the distribution grid the
network planning tools and the network operation have to be adapted (Fig. 6b). Validated models of
photovoltaic inverters for the power flow and the dynamic simulation are needed to accomplish this.
Present network planning and operation of the transmission and the distribution grid have been more or
less independent. With a dynamic power flow in two directions the information exchange between
transmission and distribution systems has to be adapted [10].

5. How to prepare for the integration of high penetration of photovoltaic

There are several important considerations in the integration of high penetration levels of PV into the
utility grid. These include the operation of the utility grid including balancing and control, the interaction
of the PV equipment with the grid as required by grid codes, and forecasting of the PV power generation
(energy meteorology).

5.1. Balancing renewable energy input

Much of the uncertainty about the potential of variable renewables to contribute to power portfolios
stems from limited understanding of the balancing capability of existing flexible resources. To address
this, the International Energy Agency (IEA)'s GIVAR Project (Grid Integration of Variable Resources)
88 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92

has developed the Flexibility Assessment (FAST) Method described in detail in [11]. The FAST method
determines the present potential for a variable renewable energy share in electricity demand along a four
steps process (Fig. 8).

Fig. 8. FAST method proposed by the IEA to prepare for the balancing challenge. With four dedicated steps an assessment of the
potential of variable renewable energy systems to contribute to power portfolios is demonstrated for different network areas [11].

Step 1 assesses the maximum technical ability of the four flexible resources to ramp up and down over
the balancing time frame. This is the Technical Flexible Resource.
Step 2 captures the extent to which certain attributes of the power area in question will constrain the
availability of the technical resource, to yield the Available Flexible Resource.
Step 3 is to calculate the maximum Flexibility Requirement of the system, which is a combination of
fluctuations in demand and the Variable Renewable Energy (VRE) output (the net load), and
contingencies.
Step 4 brings together the requirement for flexibility and the available flexible resource to establish the
Present VRE Penetration Potential (PVP) of the system in question.

5.2. Observation and control

Due to the efficient organisational structure of the network operation in Spain (RED) the major part of
the PV power systems are connected to the control centre for renewable energies and deliver actual status
and power information to the grid operator every 12 seconds (Fig. 9). Systems with an installed power
over 10 MW are under the direct control of the network operator for curtailment if feed in power exceeds
the load [12]. Based on these operational data of the renewable energy input network operators are able to
distinguish between load and renewable energy input. In addition the uncertainty of load forecast and
solar and wind power forecast are decreasing.

The feed-in law in Germany asks for actual power measurements and power curtailment functions for
systems over 100 kWp since two years and will extend this control down to system size of 1 kWp from
2012.
 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92 89

Fig. 9. Measures of the TSO in Spain to gain observability and controllability to the feed in from photovoltaic power plants [12].

5.3. Grid codes

The adaptation of grid codes to the decentralised input of renewable power to the utility network is an
essential measure to enable grid stability also with high fractions of solar and wind power input. Several
steps have been taken from the regulatory authorities in many countries. In the first phase of the
integration of distributed renewables fixed voltage and frequency limits have been set up and the systems
have been obliged to immediately shut down in case the grid is outside of these limits. Often this is also
stipulated the “Get out of the way” phase, where PV systems still proved to have a negligible influence on
the electrical system. Therefore today, grid codes are developed in a way that photovoltaic inverters have
to support the system stability and assist in the cost-effective integration of decentralised renewable
energy input. This is a more “Utility Friendly” approach. Figure 10 explains one grid-friendly feature,
that the inverter reduces power in the case of frequency conditions higher than allowed.

Fig. 10. Regulation of active power at over-frequency situations for PV systems connected to the medium voltage in Germany [13].

Beside frequency stabilisation solar inverters also have to support voltage stability by integrating Low
Voltage Ride Through (LVRT) capability (Fig. 11). In Germany since April 2011 all PV systems
connected to the medium voltage network have to be certified to provide these functions defined in the
latest grid codes [14].
90 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92

Fig. 11. Low-Voltage-Ride-Through definition for PV systems connected to the medium voltage in Germany [15].

For the integration of photovoltaic systems into the low-voltage network, grid extension is the most
expensive way but still the standard solution. Solar inverters providing reactive power, however, may be a
more cost effective way to increase the possible fraction of renewable power input to the distribution grid.

Fig. 12. (a) No voltage control assistance from PV inverters – voltage (green) rises over 1.04 p.u. (b) With voltage control assistance
from solar inverters with reactive power input, voltage is reduced significantly (right graph shows reduction below 1.03 p.u.) and
thus existing grid infrastructure is able to host higher amount of PV [9].

Figure 12 explains how voltage band limits can be met with reactive power input. The figure on the
left shows the system with no reactive power applied and the right side shows the reduction in voltage
(green dashed line) when reactive power functionality is utilised [13]. Still it is in discussion whether an
independent local voltage regulation by the individual inverters is sufficient or a central control of the
parameters from the distribution grid operators is necessary.

5.4. Energy meteorology

After explaining several measures for the successful integration of photovoltaic power systems into the
utility grid the main remaining obstacle is the variability of this new energy sources. To overcome this
barrier three different disciplines have first to understand each other and second start working together.
These efforts today are covered under the new scientific topic energy meteorology.
 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92 91

Fig. 13. Energy meteorology time lines. The graph compares the main topics, the heartbeat and the forecast horizon of the
photovoltaic, meteorology and utility network domain.

Figure 13 explains the starting points of the energy meteorology players. In respect to grid operation
and management a reliable solar power forecasting system is needed. In Fig. 14 the setup of such a
regional photovoltaic power prediction system is described [16].

Fig. 14. Design of a regional solar power forecasting system [16].

Based on the ECMWF weather forecast local irradiance and PV power measurements are needed to
improve the accuracy of the solar power forecast. For that, the detailed knowledge of all system locations
connected to the network is also an important prerequisite to reduce the uncertainty of the forecast.

6. Conclusion

x Short term variability of solar is significantly smoothed with area

92 Raymond Hudson and Gerd Heilscher / Energy Procedia 25 (2012) 82 – 92

x Standards for grid connection and operation of variable renewable energy systems need to be
modified, accepted, and implemented due to high penetration of PV
x Utility friendly inverter features are required for high penetration levels to assist network stability
and reduce integration costs
x Maximum capacity of distributed generation and reinforcement of appliances in the distribution
system have to be evaluated
x Information on power flow and voltage is needed in distribution networks
x Balancing the variable and increasing input of renewable becomes a major issue for transmission
systems
x Forecast methods and products are needed that assist the integration of high penetration of variable
renewable energy input

References

[1] Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the promotion of the use of
energy from renewable sources and amending and subsequently repealing Directives 2001/77/EC and 2003/30/EC.
[2] The German government’s Energy Concept, Long/term strategy for future energy supply. 28.09.2010.
From http://www.bmu.de/english/energy_efficiency/doc/46516.php
[3] California Renewable Energy Ressources Act. From http://www.leginfo.ca.gov/pub/11-12/bill/sen/sb_0001-
0050/sbx1_2_bill_20110412_chaptered.html, April 2011.
[4] Roadmaps - Smart Grids © OECD/International Energy Agency 2011, figure 1, page 2
[5] Ogimoto K.; Power System Evolution - System wide issues related to massive penetration of PV in the electricity systems.
IEA Task 14 meeting Valencia September 2010.
[6] Renne D. NREL data set shows clouds’ effects on solar power. NREL News Release Nr 4411, August 31, 2011.
[7] Heilscher G, Ruf H, Mayer O, Grottke M, Rabe W. Protection of the system management at high penetration of
photovoltaic energy in the low voltage grid. Proc. 26th European Photovoltaic Systems Conf., Hamburg, Germany; 2011, p.
4466-4468.
[8] Heilscher G. A real-world business model for solar electricity. Photon Utility Conf., Berlin, Germany; 2011.
[9] Stetz Th. Integration of PV plants in distribution grids. OTTI PV Monitoring Workshop, Munich, Germany; 2011.
[10] Stewart E. Connecting to the grid without causing disturbance. Solar Power International Conference, Dallas, USA; 2011.
[11] Chandler, H. Harnessing Variable Renewables - A Guide to the Balancing Challenge © OECD/International Energy
Agency 2011, ISBN 978-92-64-11138-7, figure 8, page 6
[12] Hidalgo J. Expanding visibility and controllability requirements: The solar PV case. IEA Task 14 meeting Lisboa May
2011.
[13] Technical Guideline – Generating Plants Connected to the Medium-Voltage Network. BDEW, figure 2.5.3-1 at page 27,
June 2008. From www.bdew.de.
[14] Technical Guideline – Generating Plants Connected to the Medium-Voltage Network. BDEW, chapter 6, June 2008 and
latest updates from Feb. 2011. From www.bdew.de.
[15] Technical Guideline – Generating Plants Connected to the Medium-Voltage Network. BDEW, figure 2.5.2.2-2 at page 23,
June 2008. From www.bdew.de.
[16] Lorenz E, Heinemann D. Regional Photovoltaic Power Prediction based on ECMWF Forecasts. 31st Annual International
Symposium on Forecasting, Prague, June 2011.