Professional Documents
Culture Documents
Reactive Power Supplied by PV Inverters - Cost-Benefit-Analysis
Reactive Power Supplied by PV Inverters - Cost-Benefit-Analysis
net/publication/43607141
Article
CITATIONS READS
42 3,419
1 author:
Martin Braun
Universität Kassel
223 PUBLICATIONS 2,124 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Martin Braun on 08 August 2014.
M. Braun
Institut fuer Solare Energieversorgungstechnik e. V. (ISET), Koenigstor 59, D-34119 Kassel, Germany
Phone +49(0)561/7294-118, Fax +49(0)561/7294-400, E-mail: mbraun@iset.uni-kassel.de
ABSTRACT: This paper introduces in a first step the capability and availability of reactive power supply by PV
inverters. In a second step this potential is studied concerning its economic usability. This is done by a proposed approach of
allocating additional costs to reactive power supply and assessing the benefits for network operation. Based on cost-benefit-
analyses the paper presents an economic potential of using reactive power from PV inverters for distribution network
operation.
Keywords: Economic Analysis, Photovoltaic, Inverter Losses, Reactive Power, Ancillary Services
Table I: Available Reactive Power Potential of a capabilities and availabilities of reactive power supply.
110 kWp PV generator in Kassel, Germany in 2005 This will increase the total additional costs for reactive
power supply. Neglecting this option, the presented paper
Availability with Availability with starts using the most economic potentials of the standard
Q in kVar
100 kVA inverter 110 kVA inverter sizing.
100 55.0% 89.1%
90 88.4% 94.7% 4.2 Additional operational costs
80 93.7% 98.3% PV generators have low operational costs as little
70 97.1% 99.8% maintenance is required in normal circumstances and no
60 99.3% > 99.9% fuel costs occur. Similar to all other power plants they
50 99.9% > 99.9% have a certain self consumption. This is assumed to be
40 > 99.9% 100% the most important cost component which is analysed in
30 > 99.9% 100% detail in this paper.
20 > 99.9% 100% The additional self consumption of PV generators
10 > 99.9% 100% due to reactive power supply can be attributed to the
additional losses of the PV inverter. This simplification
This table also shows that it is not possible to assumes that the inverter has no external power supply
guarantee 100% availability of the reactive power supply for self consumption and is installed at the point of
without oversizing the inverter. However, more than common coupling without additional losses by wiring.
99.9% is possible. The determination of the self consumption of PV
inverters starts with the available data on the inverter’s
efficiency
4 COSTS OF REACTIVE POWER SUPPLY BY PV PAC PAC
INVERTERS eta = =
PDC PAC + P'loss
Standard PV inverters are often lacking the
functionality to supply reactive power. However, from a depending on the active power output of the AC side PAC
technical point of view they are capable to provide this and the active power input on the DC side PDC. The
service. The functionality is required rarely because difference between PDC and PAC are the losses P’loss of the
owners of PV generators often do not get incentives for inverter.
providing reactive power supply. Before analysing the The losses of an inverter can be approximated by a
benefits in the next section, this section analyses the second polynomial function [2]:
costs of this functionality.
Costs for reactive power supply can be separated into P'loss (PAC ) = c self + cVloss ⋅ PAC + c Rloss ⋅ (PAC )
2
Based on these results the inverter losses can be The graph is limited by the semicircle of the device’s
assumed independent from the power factor and PAC can nominal apparent power and shows the symmetry of
be substituted by S with only small errors. This leads to negative and positive reactive power supply. Three areas
the polynomial approximation of the losses proposed in can be distinguished:
the presented paper:
1. PAC = 0, Q = 0: The inverter is deactivated if no
Ploss (S ) = cself + cVloss ⋅ S + c Rloss ⋅ (S ) power is supplied:
2
ΔPloss (t ) = 0
According to a set of datasheets of different inverters
(not representative) an efficiency curve has been derived 2. PAC > 0, Q ≠ 0: With increasing reactive power
as an example for a PV inverter of the year 2005 (with a supply the additional losses increase with a
maximum efficiency etamax of 95.6% and a European second order polynomial function:
efficiency etaEuro of 94.7%). Further improvements of PV ΔPloss (t ) =
inverter efficiencies are assumed for the year 2015 c self + cVloss ⋅ PAC
2
(t ) + Q 2 (t )
(etamax = 98.1% and etaEuro = 97.7%) and 2030 (etamax =
99.1% and etaEuro = 98.7%). These three inverter (
+ c Rloss ⋅ PAC
2
(t ) + Q 2 (t ) )
efficiency curves are displayed in Figure 3 and should
show a trend of future improvements.
(
− c self + cVloss ⋅ PAC (t *) + c Rloss ⋅ PAC
2
)
(t *)
100% 3. PAC = 0, Q ≠ 0: Without active power supply
98%
the inverter would be deactivated. Hence, the
standby losses cself have to be attributed to the
96% reactive power supply in contrast to the situation
with P > 0:
Efficiency
94%
2005
ΔPloss (t ) = c self + cVloss ⋅ Q(t ) +c Rloss ⋅ Q 2 (t )
92% 2015
2030
90% These additional losses can be attributed to the
88%
reactive power supply to get the related value in
kW/kVAr:
86%
ΔPloss (t )
0 10 20 30 40 50 60 70 80 90 100
Apparent Power (%Smax) Ploss (t ) =
Q(t )
Figure 3: Example inverter efficiency curves for the
years 2005, 2015 and 2030 At this stage it is helpful to distinguish two cases:
With these basic assumptions of the proposed 1. Daytime (P > 0): Active power is provided by
approach we can now calculate the additional losses by the PV generator. In addition reactive power is
taking the difference of the inverter’s losses with reactive supplied. Only additional losses are attributed
power supply Q(t) ≠ 0 and without Q(t*) = 0 for the same to reactive power supply. The related
DC link power PDC at time t as well as reference time t*: additional losses are displayed in Figure 5
showing
ΔPloss (t ) = • the decline of additional losses of a
Ploss (PDC (t ), Q(t )) − Ploss (PDC (t *) = PDC (t ), Q(t *) = 0 ) certain reactive power supply with
increasing active power supply and
With the loss curve of 2005 this leads to the • an increase of additional losses with
additional losses caused by reactive power supply as increasing reactive power supply.
displayed in Figure 4.
Figure 4: Additional inverter losses of a 100 kVA PV Figure 5: Related additional inverter losses [W/kVAr]
inverter due to reactive power supply Q with different due to reactive power supply Q [%Smax] during active
active power output P (assumed inverter performance of power supply P [%Smax] at Daytime (assumed inverter
the year 2005) performance of the year 2005)
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3 - 7 September 2007, Milan, Italy
2. Nighttime (P = 0): The PV generator would be • Estimated costs (if active power supply P>0) of
deactivated if only active power needs to be active power generation by PV generators in
considered. All losses have to be attributed to Germany (cf. [6]):
reactive power supply. The related additional 2005: 50 c€2005/kWhPV,
losses are displayed in Figure 6 showing very 2015: 30 c€2005/kWhPV, and
high losses for small quantities of reactive 2030: 15 c€2005/kWhPV
power supply due to the attributed standby
losses. These are then reduced with increasing Table II: Costs of reactive power supply in c€2005/kVArh
reactive power supply passing a point of for different P/Q set points over the assumed years 2005,
minimum related additional losses. 2015 and 2030 considering a PV inverter of 100 kVA
Their benefit can be assumed to be as high as the costs of reactive power supply by PV inverters would cost less in
the substitutable reactive power sources. This section most cases, further improving in the long-term future.
provides a comparison with capacitors, network In Spain an incentive system according to [9] defines
purchase, and an analysis of the benefits of reactive three load situations (peak, plateau, and off-peak). If
power based ancillary services. generating units provide the corresponding power factors
they receive an incentive if they counteract they have to
5.1 Comparison with Capacitor Banks pay a penalty. This incentive is attractive for PV plant
A standard network component to compensate operators, particularly, because the power generation
reactive power is a capacitor bank. They are often costs of PV are far less than in Germany. As the
applied in industrial facilities and at certain network incentive is paid per kWh it cannot be compared
nodes. Capacitors are normally sufficient because in most reasonably here.
of the network’s situations the load has an inductive
character mainly due to transformers and induction 5.3 Comparison with Benefits for Network Operation
generators. The analysis of costs of capacitor banks for Reactive power is necessary for an optimized
industrial facilities results in the kVArh prices displayed network operation. Capacitive reactive power increases
in Figure 7. They depend on the used full load hours on a the voltage level while inductive reactive power
diminishing scale. The cost estimation is based on the decreases the voltage level. However, the voltage needs
following assumptions: 20 years lifetime, 5% discount to stay within certain limits demanding for distributed
rate, 15% additional installation costs, losses of reactive power compensation. Reactive power is used in
1.5 W/kVAr [7] and power purchase costs of network operation mainly for three ancillary services:
10 c€2005/kWh for an industrial facility (close to [4]). The 1. voltage control,
figure shows that with higher full load hours the costs 2. reduction of grid losses, and
decrease below 0.2 c€2005/kVArh. 3. reduction of congestions.
The value of these services is analyzed in the following
Costs of Capacitors [c€/kVArh]
Table III: Loss savings in c€2005/kVArh due to reactive Figure 8: Relative Reduction of Loading due to reactive
power compensation with different load power factors power compensation by PV inverters considering
cos(phi) and average network losses with costs for losses different PV penetration levels PPV and different load
of 6 c€2005/kWh power factors cos(phi)
Network Losses
cos(phi) 2% 4% 6% 8% 6 CONCLUSIONS
0.95 0.039 0.079 0.118 0.158
0.9 0.058 0.116 0.174 0.232
This paper introduces in a first step the capabilities
and availabilities of reactive power supply of PV
0.85 0.074 0.149 0.223 0.297
inverters showing an interesting potential. In a second
step this potential is studied concerning its economic
usability. This is done with a proposed approach of
5.3.3 Reduction of Congestions allocating additional costs to reactive power supply and
By active compensation of reactive power it is assessing the benefits for network operation. The cost-
possible to reduce the reactive power flows in the benefit-analysis shows that PV inverters already have
network. Particularly at peak load situations this can some opportunities at present but many more in the
reduce the loading of the network helping to avoid future with further generation cost reductions and
congestions. Figure 8 displays the relative reduction of improvements of the inverter’s efficiency. Even today
the loading of the considered network element (e.g. line PV inverters are in some cases cheaper than capacitors
or transformer) by reactive power compensation. The and cheaper than the reactive power purchase costs by
network element is assumed to operate at 100% rated the network operator in Germany and incentives paid in
capacity Sn considering different load power factors Spain. Moreover, in the future an active distribution
cos(phi). In these situations the reactive power flow is network management allows using the reactive power
compensated by PV inverters with 100% of their rated supply of PV inverters economically for reducing
capacity. Their penetration Ppv is assumed to be 5%, network losses and congestions.
10% and 20% of the rated capacity Sn. The paper presents an economic potential of using
Figure 8 shows that the loading can be reduced by reactive power supplied by PV inverter for network
15% (cos(phi)=0.98), 30% (cos(phi)=0.94) or 45% operation. Particularly in the future this becomes more
(cos(phi)=0.87) of the inverter’s capacity at a penetration and more interesting. This potential should be used to
level of 10%. This reduction is significant. If the range of optimize the quality, economy and security of
15-45% is taken and the network costs are assumed to be distribution network operation.
50-150 €2005/kWa the benefit can be calculated as 7.5- The integration of distributed reactive power supply
67.5 €2005/kVAPV. Relating this benefit to some hours of by PV inverters requires further analysis of the economic
reactive power compensation, say 10-30 hours/year, the
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3 - 7 September 2007, Milan, Italy
ACKNOWLEDGEMENT
REFERENCES