Hatziargyriou 等 - 2011 - Quantification of economic, environmental and oper

EUROPEAN TRANSACTIONS ON ELECTRICAL POWER
Euro. Trans. Electr. Power 2011; 21:1217–1237

Published online 10 September 2009 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etep.392
Quantification of economic, environmental and operational

benefits due to significant penetration of Microgrids in a typical
LV and MV Greek network
N. D. Hatziargyriou1, A. G. Anastasiadis1*,y, A. G. Tsikalakis1 and J. Vasiljevska2,3

1
School of Electrical and Computer Engineering, National Technical University of Athens,
9 Iroon Polytechniou Street, GR-157 80 Zografou, Athens, Greece
2
Faculty of Engineering, University of Porto, Portugal
3
INESC Porto, University of Porto, Portugal
SUMMARY
The effect of the use of a Microgrid Central Controller (MGCC) to achieve co-ordinated operation of various
DG sources, storage devices and controllable loads with regards to the potential economic benefits and the
power losses avoided in the local network is presented. Additionally, the effect of connecting numerous
typical Microgrids in a typical medium voltage (MV) network regarding power losses and the emissions
avoided is studied. The emissions avoidance calculation is based on the marginal emissions curve of the
upstream network, quantifying also the calculated losses reduction impact on emissions avoidance. The
presented results, based on a variety of realistic market prices time-series, various levels of renewable energy
resource (RES) production and demand from different seasons, show clear benefits in terms of both losses
and emissions avoidance. Finally, this paper performs a sensitivity analysis on the potential impact of the DG
topology within a Microgrid on losses and emissions avoidance. Copyright # 2009 John Wiley & Sons, Ltd.
key words: Microgrids; distributed generation; environmental assessment; power losses; renewable
energy sources (RESs); markets
1. INTRODUCTION
Microgrids are defined as low voltage (LV) or in some cases, e.g. Japan, as medium voltage (MV)
networks with distributed generation (DG) sources, together with storage devices and controllable
loads (e.g. water heaters, air conditioning) with a total installed capacity in the range of few kWs to
couple of MWs. The unique feature of Microgrids is that although they operate mostly interconnected
to the upper level voltage distribution network, they can be automatically transferred to islanded mode,
in case of faults in the upstream network.
From the grid’s point of view, a Microgrid can be regarded as a controlled entity within the power
system that can be operated as a single aggregated load [1] and, given attractive remuneration, as a
small source of power or ancillary services supporting the network. The installation of DG close to
loads will reduce flows in transmission and distribution circuits with loss reduction as a consequence.
Microgrids can provide additional benefits to the local utility by providing dispatchable power for use
during peak power conditions and alleviating or postponing distribution system upgrades [2,3].
They can also provide network support in times of stress by relieving congestions and aiding
restoration after faults. From a customer’s point of view, Microgrids, similar to traditional LV
distribution networks, provide their thermal and electricity needs, but moreover, enhance local
reliability [4], and improve some power quality indices by supporting voltage and reducing voltage
*Correspondence to: A. G. Anastasiadis, School of Electrical and Computer Engineering, National Technical University
of Athens, 9 Iroon Polytechniou Street, GR-157 80 Zografou, Athens, Greece.
y
E-mail: aanestis@power.ece.ntua.gr
Copyright # 2009 John Wiley & Sons, Ltd.

1218 N. D. HATZIARGYRIOU ET AL.
dips. Power quality impact regarding harmonic distortion, voltage flickers or voltage unbalance have
been and are being studied and proposed methodology for analysing, evaluating and proposing ways to
combat them are inductively presented in References [5–7].
This paper describes the results not only from the simulation of one Microgrid’s operation under
various combinations of realistic market prices and renewable energy sources (RESs) production but
also of many Microgrids under MV. Special focus is given on the impact that Microgrids operation can
have on a typical MV network when significant penetration of Microgrids is foreseen. For this purpose
a typical LV with DG operating co-ordinated as a Microgrid and a typical MV comprising of many LV
networks are used. Additionally, a program in Matlab environment [8] in order to perform a power flow
analysis and the consequent calculation of power losses of the system has been developed. One step
forward is the estimation of the avoided emissions due to Microgrids operation, based on the marginal
emissions curve of the upstream network. The above-calculated losses are also taken into account for
these calculations and the part of emissions avoided due to losses reduction is quantified.
The methodology of the above calculations is explicitly described in Section 3 and is applied to the
case study networks under realistic market prices all of them described in Section 4 for representative
periods. The results from this application are presented in Section 5 quantifying the benefits in terms of
end users cost reduction, active losses reduction in both the LV and MV network and finally emissions
avoidance. Based on these results, some main conclusions can be drawn regarding the potential
benefits of larger penetration of Microgrids in typical networks.
2. OPERATIONAL FRAMEWORK
Possible hierarchical control system architecture comprises the following three control levels, shown in
Figure 1 [9]:
Distribution Management System (DMS).
Microgrid Central Controller (MGCC).
Local Controllers (LCs).
The DMS consists of the Distribution System Operator (DSO) or Distribution Network Operator
(DNO) and one or more Market Operators (MOs), who can be the same or a different entity. The DSO
is responsible for the operation of medium and LV areas in which more than one Microgrid may exist.
The MO is responsible for the market functions, like determining the external clearing market prices
for the specific area operated by the DSO.
Figure 1. Hierarchical control structure for Microgrids.
Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2011; 21:1217–1237
DOI: 10.1002/etep
ECONOMIC, ENVIRONMENTAL AND OPERATIONAL BENEFITS OF MICROGRIDS 1219
The main interface between the DMS and the Microgrid is the Microgrid Central Controller
(MGCC), while each microsource and load, (if demand side options are to be applied),within the
Microgrid is equipped with LC, designated as Microsource Controller (MC) and Load Controller,
respectively.
The MGCC may assume different roles ranging from the main responsibility for the maximization of
the Microgrid value and optimization of its operation to simple co-ordination of the LCs. The
information exchange within a typical Microgrid is as follows: every m minutes, e.g. 15 minutes, each
DG source bids for production for the next hour in m minutes intervals. These bids are based on the
energy prices in the open market and the operating costs of the DG units plus the profit of the DG
owner. The MGCC optimizes the Microgrid operation according to the open market prices, the bids
received by the DG sources and the forecasted loads and sends signals to the MCs of the DG sources to
be committed and, if applicable, to determine the level of their production. In addition, consumers
within the Microgrid might bid for their loads supply for the next hour in same m minutes intervals or
might bid to curtail their loads. In this case, the MGCC optimizes operation based on DG sources and
load bids and sends dispatch signals to both the MCs and the load controllers. Figure 2 shows the
information exchange flow in a typical Microgrid operating under such conditions.
For the market operation within a Microgrid, two market policies can be considered:
Market Policy 1 where the MGCC aims to minimize the cost of energy for the end-users without
selling energy to the grid.
Market Policy 2, where the MGCC aims to maximize the value of the DG sources by selling
excess energy to the upstream network.
More details on these policies as well as options for participation of loads with demand side bidding
(DSB) are described in more detail in Reference [10].
3. METHODOLOGY FOLLOWED
3.1. Economic operation

Economic scheduling comprises unit commitment (UC) and economic dispatch (ED), so that the
production of the DG sources, whose outputs can be regulated, and the power exchanged with the grid
are determined.
UC is solved first using a priority list. The DG bids, the load bids and the market prices are placed in
one list according to their differential cost at the highest level of production for the specific period. This
list is sorted in ascending bid values, so that the total demand is met.
DG bids are considered quadratic as presented in Equation (1) according to the cost function of the
units, if any, the feedback from the market prices and the need for making some profit, necessary for the
annual depreciation of the installation cost. In Equation (1) term, ai is expressed in sct/kWh2, bi is
expressed in sct/kWh and ci in sct/hour. Finally xi is the output of the units. The load bids
Figure 2. Information exchange diagram.
DOI: 10.1002/etep
usually are linear and usually ai and ci in Equation (1) are zero.
active bidðxi Þ ¼ ai x2i þ bi xi þ ci (1)
ED is performed next, to determine the output of the regulated sources. The output of the RESs, i.e.
the wind turbine (WT) and PV, cannot be regulated and their output is determined by the availability of
the primary source, i.e. wind or sun radiation.
If the bids are continuous and convex functions, then mathematical optimization methods can be
applied such as the Sequential Quadratic Programming (SQP) utilized in our case [11]. Artificial
Intelligence Techniques, like can be also used, especially if scalar or discontinuous bids are considered
[12].
3.2. Power losses estimation

Modern distribution systems were designed to accept bulk power at the Bulk Supply Transformers and
to distribute it to customers. Thus, the flow of both real power (P) and reactive power (Q) was always
from the higher to the lower voltage levels. However, with significant penetration of DG the power
flows may become reversed and the distribution network is no longer a passive circuit supplying loads
but an active system with power flows and voltages determined by the generation as well as the loads.
The change in real and reactive power flows caused by Microgrids generation has important technical
and economic implications for the power system [6]. So, Microgrid generation alters the power flows in
the network and so will alter network losses.
If a small Microgrid generation is located close to a large load then the network losses will be
reduced as both real and reactive power can be supplied to the load from the adjacent Microgrid
generators. Conversely, if a large Microgrid generation is located far away from network loads then is
likely to increase losses on the distribution system.
Generally, there are active power losses in the transmission network. These losses depend on the
currents in the branches of this network which in turn depend on the voltages, and calculating these
voltages is the object of the power flow calculation.
The avoided losses are often pointed out as an important value to be credited to microgeneration.
Therefore, the quantification of the impact of microgeneration on losses of transmission and
distribution networks is an important issue. Several studies have been made to evaluate the impact of
DG on losses as well as to allocate the variations to generators and consumers. Most of those studies are
devoted to find out how losses vary in a specific feeder and DG plant connection or when a predefined
scenario of DG penetration is considered. Some of those studies propose algorithms to define the
optimal location to DG [13,14].
The Microgrid generation will generally choose to operate at unity power factor to minimize their
electrical losses and avoid any charges for reactive power consumption, irrespective of the needs of the
distribution network. If a Microgrid generation produces some power at unity power factor the voltage
profile is much more satisfactory [6,15]. The total injected complex power at bus i, denoted by Si, is
given by: Si ¼ Pi þ jQi ¼ Vi Ii. The summation of powers over all buses gives the total system losses:
X
n
PL þ jQL ¼ Vi Ii ¼ Vbus
T
Ibus (2)
i
where PL and QL are the real and reactive power losses of the system, Vbus is the column vector of the
nodal bus voltages, Ibus is the column vector of the injected bus currents and n is the number of buses.
3.3. Estimation of environmental benefits

Since the Microgrids penetration in the grids is expected to be relatively low, the initial UC schedule of
the centralized production is not expected to change. However, there will be modifications in the ED of
the most expensive, i.e. critical units of the upstream network, and the network losses as calculated
with the method described above. Both will alter the emissions of the upstream network. Therefore, in
DOI: 10.1002/etep
order to estimate the emissions avoided, using average yearly or even monthly values will lead to
misleading results, since very rarely will the base units be affected [16]. For this reason, the assessment
of environmental impact of Microgrids uses a monthly 24-hour typical emissions curve, Pol, depending
on the upstream network units’ characteristics as provided by the following formula:
P
unno
fcuðhr;mÞi emfðpoÞi
i¼1
Polðhr;m; poÞ ¼ (3)
dayðmÞ
where unno is the number of the units that may be affected by the introduction of the DG, fcu is the
frequency that the unit i is expected to be the critical unit of the system for the month m and the hour hr
and emf is the emission factor of the pollutant po for the unit i. The parameter day(m) is the number of
the days in month m.
Typical 24-hour emission curve from the island of Crete for different periods shows the application
of Equation (3) in Figure 3.
The knowledge of hourly marginal units is even more useful when PV installations are foreseen and
the production of DER depends on the market prices, as is in our case.
4. CASE STUDY NETWORKS
4.1. Typical LV network

Typical LV network is used in our study (Figure 4) [17]. LV network lines are either underground cable
lines, mainly encountered in urban areas with a high load density, or overhead lines mainly
Figure 3. Typical emission curve from the island of Crete and CO2.
Figure 4. Typical LV network used.
DOI: 10.1002/etep
Figure 5. Typical load curve for each feeder of the study case network.
encountered in rural areas with a comparatively low load density. Overhead lines are mainly
constructed with bare conductors made of Al or Cu. Variations do exist. Cables are usually enclosed in
either metallic or galvanized conduit. The residential and commercial sub-networks were represented
with both overhead lines and underground cables, while the industrial sub-network used only
underground cable. It is also possible to change these arrangements and use overhead lines (in our case)
instead of cables and vice versa. It would then be necessary to modify the line parameters in accordance
with information supplied by manufacturers. Resistances and reactances of the lines can be found in
the Appendix—Table AI.
The network in our study comprises three feeders, one serving a primarily residential area, one
industrial feeder serving a small workshop and one commercial feeder. Typical load curve for each
feeder of the study case network is shown in Figure 5. A variety of Distributed Energy Resource (DER),
such as one microturbine (MT), one fuel cell (FC), one directly coupled WT and several PVs are
installed in the residential feeder. It is assumed that all DER produce active power at unity power factor.
Table I provides the capacity of the installed DG sources and their fuel costs. Both, MT and FC are
assumed to run on natural gas, whose efficiency is 8.8 kWh/m3 and price 10 sct/m3 [18]. For the MT
the efficiency is assumed 26% [19], while for FC the findings of the analysis in Reference [20] for
proton exchange membrane (PEM) efficiency were used to calculate the bids reflecting the
operational costs. The corresponding bids submitted are provided in Table II, while emissions
data for the fuel-consuming units are provided in Table III [21]. For RES the bids are considered
Table I. Data for the capacity of the RES of the Microgrid.

Unit ID Unit name Max capacity (kW)
3 WT 15
4 PV1 3
5–8 PV2-5 2.5
Table II. Data for the bids considered for the fuel-fired DG units.
Unit Unit ai bi ci Min Max
ID name (sct/kWh2) (sct/kWh) (sct/hour) capacity (kW) capacity (kW)
1 MT 0 4.37 0 6 30
2 FC 0.033 2.41 0.8415 3 30
Table III. Data for the emissions considered for the installed DG units in g/kWh.
Unit name CO2 NOX SO2
MT 724.6 0.2 0.004
FC 489.4 0.014 0.003
DOI: 10.1002/etep
as equal to 0, reflecting their operating cost. The scope was to calculate the maximum potential
savings for the customers in Policy 1. Moreover, if the RES bids were not zero but a value
reflecting the necessary price for timely paying back the investment, then for some hours this
value would have been higher than the spot market prices used for the market environment.
Under such circumstances, RES production bid would not be always accepted by the MGCCs
and their owner would lose some income. A methodology for assessing the value of intermittent
RES under market environment and the circumstances under which such an investment would be
able to be paid back has been explicitly described in Reference [22].
For the monthly demand data, annual demand is distributed to each month according to the
Reliability Test System (RTS) weekly variation [23] and the typical demand curve of the Microgrid is
used [17], as well as, monthly demand and production of RES of the studied LV network [16]. Data
about the wind velocity time-series of the island of Crete were used and a typical WT of 15 kW. The
WT power curve is represented by a 3rd order polynomial [22]. For PVs, normalized time-series from
the PV installation, 1.1 kW, in the campus of the NTUA are used [24].
With regards to the market policies applied, focus is given on the market policies and a comparison
between (a) operation without DG sources are considered, i.e. all the demand has been met by the grid
and (b) Microgrid operation for Market Policy 1 applied is provided. A description of the market prices
is made in the following sub-paragraph.
Then the avoided losses in the MV/LV transformer and on the LV lines of the network in Figure 4 can
be calculated. The losses for each type of hour are combined with the typical emissions curve to
calculate the additional emissions avoided due to losses reduction. In the emissions calculations the
reduction of the demand in the network and any emissions by the DG units are taken into account. The
results are presented in the coming section.
4.1.1. Market operation context. In order to represent realistically the open market in which the LV
grid operates, electricity market prices from the Amsterdam Power Exchange (ApX) for 2003 [25] for
each of the 4 months have been taken as representative ones. Days with low, average (typical) and high
electricity market prices within each representative month have been identified. Figure 6 shows the
market prices for such days for October, as month with the highest electricity market prices among the
representative months. Similar graphs can be obtained for the other months as well. Additionally, the
averages per hour marginal system prices of the Hellenic Transmission System (HTS) [26] were used
in one scenario. These prices present rather low and much less volatile values compared to the
electricity market prices from the ApX as Figure 6 shows.
ApX prices have been combined with high, and low RES production level for each of the studied
month, indicating the days with highest and lowest wind and PV production respectively for the
presented results. HTS prices have been combined with the average per season and per hour wind and
PV production. Data from the Hellenic Transmission System Operator (HTSO) for 2006 have been
used to derive the 24-hours typical curve for each month.
Figure 6. ApX electricity market prices October 2003 (left) and HTS market prices (right).
DOI: 10.1002/etep
Figure 7. Typical MV network of 64 Microgrids.
4.2. Typical MV network

Here the impact of many Microgrids operating in a typical radial MV network, shown in Figure 7, is
studied. This network consists of two parallel transformers of 25 MVA, a typical size of many HV/MV
substations in many semi-rural areas in Greece. The typical MV distribution network is derived from a
physical MV network of Greece [26,27], which supplies a small town and the surrounding rural area at
a rated voltage level of 20 kV supplied from a 150 kV transformer station. The total number of LV
Microgrids similar to Figure 4 is 64 in eight feeders. The MV network structure consists of 130 nodes.
The base voltage may be modified to study different voltage levels as long as the chosen values are
realistic. The line length is 1 km from bus to bus and they can be modified as long as the typical MV
distribution network character is retained. Resistances and reactances of the MV network lines can be
found in the Appendix—Table AII.
It should be noted that the distribution installation common practices do not vary greatly among
countries of Europe. European MV distribution networks are three-phase and either of meshed or radial
structure, with the latter dominating rural installations. Each feeder includes numerous laterals at
which MV/LV transformers are connected. MV network lines are either underground cable lines,
mainly encountered in urban areas with a high load density, or overhead lines mainly encountered in
rural areas with a comparatively low load density (as in this case). Overhead lines are mainly
constructed with bare conductors made of Al or Cu [28–30].
In some countries simple design rules are used to give an indication of the maximum capacity of
dispersed generation which may be connected at different points of the distribution system. These
simple rules tend to be rather restrictive and more detailed calculations often show that more
generation can be connected with no difficulties. On 20 kV network and at buses the maximum
capacity of dispersed generator is 6.5–10 MVA [6].
5. RESULTS
5.1. Economic benefits due to Microgrids

Here the economic benefits from the proposed operation for each Microgrid are presented. If in the
Microgrid only RES was installed, then the annual savings by their operation in the market context can
be assessed using probabilistic analysis techniques, convolution of expected wind or solar energy with
the market prices. The methodology for such an analysis and representative results from its application
with ApX prices are presented in Reference [22].
DOI: 10.1002/etep
Figure 8. Cost reduction according to the ApX prices for high RES production for the study case.
Figure 9. Cost reduction according to the ApX prices for low RES production for the study case.
Figure 10. Cost reduction for average RES production for the scenario of HTS prices.
Since nine study cases have been studied, according to the level of RES production (lowest, average
or highest) and the electricity prices according to ApX prices, the final results are shown in Figures 8–
10. Figure 10 presents the results for the Greek study case scenario.
The cost reduction is expressed in terms of percentages in comparison with the case where no DG
sources are considered. Significant cost reduction can be noted for the cases with highest electricity
prices, especially for August as a month with highest electricity prices according to ApX prices value
(34.68–40.87% of cost reduction due to the case with no DG sources).
For the scenario of the Greek electricity tariff prices, the highest cost reduction was observed for
October as month with lowest demand from the four taken as representative ones. The MT was not fully
dispatched in the early morning hours due to the valley demand period and its operation cost being
rather higher in comparison with the FC one. July and January present lowest cost reduction as months
DOI: 10.1002/etep
Table IV. Summarized results for case study 1 and for Market Policy 1.
Case study 1 (LV network with the three feeders)
ApX prices RES production Cost reduction (%)
Min Max
High Low 7.610 34.68
High Average 11.92 38.54
High High 16.59 40.87
Average Low 0.700 16.16
Average Average 3.840 18.44
Average High 9.200 21.65
Low Low 0.040 3.080
Low Average 1.240 6.220
Low High 5.090 9.350
with highest demand, as presented in Figure 10. The cost reduction takes the biggest power effect in the
cases with highest electricity prices.
Table IV comprises the cases with high market prices, when the implementation of the Microgrid has
the greatest impact, when Market Policy 1 is considered.
5.2. Study of the active power losses according to the ApX prices
5.2.1. Study of the LV network. Nine cases have been studied according to the level of RES
production (highest, lowest and average) and the electricity prices according to ApX prices. The first
case is that where no DG sources are considered. The study has been done for the months January,
April, July and October (05:00 in the morning—lowest demand and 19:00 in the afternoon—highest
demand).
In this network (Figure 4), the slack bus is at the MV network side while the rest of the network is at
LV side. So, practically the total power losses are the power losses from Transformer MV/LV plus the
power losses from the LV network.
Characteristic graphical presentations of the results are shown in the following Figures 11–14 and
Tables V and VI.
In the following paragraphs, all the corresponding figures and tables will not be shown due to the
limited space of the paper.
As it was expected the DG power production of Microgrids reduces the power losses, Figures 11
and 12, of the LV network (for example: maximum active power losses reduction 51.13%—October,
19:00 a.m., scenario 6, Figure 13). The losses are considerably reduced because the generation is much
closer to the load and the lines carry much reduced flows.
Figure 11. January: the total 24-hours active losses for some scenarios. Similar graphs for the other months.
DOI: 10.1002/etep
Figure 12. January: total active power losses as DG production increases. Total demand 207.965 kW (19:00
in the afternoon) - similar graphs for the other months.
Figure 13. Comparison of maximum reduction of active power losses due to DG production. Maximum
total load demand is 207.965 kW (19:00 in the afternoon) - similar graphs for the other hours.
Figure 14. Active power losses reduction in a typical day of January, Scenario 6 (high ApX - high RES) vs
Scenario 1 (without DG) - similar graphs for the other months.
If the Microgrid produces some power at unity power factor, the voltage profile is much more
satisfactory (in all cases, voltage drop lower than 3%).
5.2.2. Study of the MV network. Three cases have been studied per month. Always, the first case of
each month is that where no DG sources are considered. The other two are for maximum and minimum
DOI: 10.1002/etep
Table V. January: PTotal load demand,max ¼ 207.965 kW (19:00 in the afternoon) (similar results for the other
months and hours).
Scenarios for January month Injection from the Total PDG (kW) Total losses P (kW)
grid (MV/LV)
P (kW) Q (kVar)
1 Without DG 212.3 5.67 0.00 4.357
2 Avg ApX–avg RES 141.3 2.75 68.8 2.220
3 Avg ApX–high RES 144.8 2.85 65.4 2.258
4 Avg ApX–low RES 148.7 2.97 61.5 2.310
5 High ApX–avg RES 141.3 2.75 68.8 2.220
6 High ApX–high RES 144.8 2.85 65.4 2.258
7 High ApX–low RES 148.7 2.97 61.5 2.310
8 Low ApX–avg RES 172.2 3.88 38.8 3.146
9 Low ApX–high RES 175.7 4.01 35.4 3.217
10 Low ApX–low RES 179.7 4.17 31.5 3.308
Table VI. LV network’s total power losses reduction (%) for January (similar tables for the other months and
hours).
Scenarios for January month Total losses—reduction (%)
PLoad,min ¼ 55.201 kW PLoad,max ¼ 207.965 kW
05:00 a.m. 19:00 p.m.
P (kW) Q (kVar) P (kW) Q (kVar)
1 Without DG 0 0 0 0
2 Avg ApX–avg RES 4.82 5.70 49.1 51.5
3 Avg ApX–high RES 18.3 24.1 48.2 49.7
4 Avg ApX–low RES 0.96 1.04 46.9 47.5
5 High ApX–avg RES 4.82 5.70 49.1 51.5
6 High ApX–high RES 18.3 24.1 48.2 49.7
7 High ApX–low RES 0.96 1.04 46.9 47.5
8 Low ApX–avg RES 4.82 5.70 27.8 31.5
9 Low ApX–high RES 18.3 24.1 26.2 29.2
10 Low ApX–low RES 0.96 1.04 24.1 26.4
DG penetration, respectively. The studies have been done for months January, April, July and October.
Some of the results are shown in the Table VII. Figure 15 presents the losses avoidance within the
24 hours horizon.
The total daily losses in both LV and MV network are summarized in Table VIII. The total power
losses of the whole network (LV and MV Networks) are given by the equations: total power losses of
the network (active or reactive) equals total power losses of the LV network plus total power losses of
the MV network.
Table IX, presents the allocation of losses between LVand MV. Clearly, the percentage of total active
power losses allocation of the LV network is increased because of DG production.
As it was expected the DG power production reduces the active power losses of the MV network.
The real power injection from the slack bus is always reduced with DG power production. The voltage
profile is very satisfactory (in all cases voltage drop is lower than 3%). Because of power loss reduction
environmental benefits are expected, quantified in detail in 5.4.
5.3. Sensitivity for the Greek scenario

In this scenario the average prices of the HTSO prices have been combined with the average monthly
production of PV and wind. Characteristic graphical presentations of the results for the LV and MV
DOI: 10.1002/etep
Table VII. MV network’s total active power losses reduction.

Scenarios Total power active losses—reduction (%)
January April July October
PLoad,min ¼ PLoad,max ¼ PLoad,min¼ PLoad,max ¼ PLoad,min ¼ PLoad,max ¼ PLoad,min ¼ PLoad,max ¼

55.201 kW 207.965 kW 48.954 kW 184.433 kW 55.354 kW 208.543 kW 46.291 kW 174.397 kW
05:00 p.m. 19:00 a.m. 05:00 19:00 a.m. 05:00 p.m. 19:00 a.m. 05:00 p.m. 19:00 a.m.
p.m.
P Q P Q P Q P Q P Q P Q P Q P Q
1 With 1.16 1.37 28.9 28.9 0 0 9.91 9.9 19.6 19.3 0.6 0.54 0 0 0 0
minimum
DG
penetration
2 With 26.7 26.5 56.5 56.5 37 37 9.91 9.9 32.3 32 57.9 57.8 28.1 27.5 65.1 65
maximum
DG
penetration
Figure 15. HTS prices. January: 24-hours curve of the active power losses for the MV network and for the
main three scenarios - similar graphs for the other months.
Table VIII. Percentage (%) reduction summation of active power losses at both LV and MV network for the
whole 24 hours and all months studied based on ApX prices.
Comparison between scenario 1 (without DGs production) and scenario 6 (high RES–high ApX)
Months LV network MV network
Only line losses Total losses
January 31.02 55.23 42.68
April 15.09 36.78 25.16
July 28.98 56.43 41.76
October 30.19 67.50 47.50
Table IX. Percentage (%) of total active power losses allocation between LV and MV networks.
Months Scenario 1: without DGs pro- Scenario 6: high RES–high
duction ApX
LV MV LV MV
January 51.81 48.19 62.36 37.64
April 53.56 46.44 60.77 39.23
July 53.43 46.57 65.16 34.84
October 53.62 46.38 71.30 28.70
DOI: 10.1002/etep
Figure 16. HTS prices. 24-hours curve of the active power losses for the LV network and for the scenario
without DG.
Figure 17. HTS prices. 24-hours curve of the active power losses for the LV network and for the scenario
with DG production.
Figure 18. Percentage total active power losses reduction for the LV network due to DG units’ production.
network (Figures 4 and 7, respectively) and for the studied period are shown in the following
Figures 16–22. The procedure is the same as the previous one, Section 5.2.
Figure 18 shows increase in losses of the LV network in respect to the scenario with no DGs
production (Figure 17), for the period 01:00–07:00 in the morning. This period corresponds to low
demand of the Microgrid and increased production from the DG sources. However, as shown in
Figure 4, DG sources are connected to one feeder only. The production exceeds the feeder demand but
not the demand of the Microgrid. In such a case the active power losses are increased compared to the
current situation. In order to study more the topology impact, it is assumed that it is possible to install
DOI: 10.1002/etep
Figure 19. Total active power losses reduction for the LV network due to change in the topology.
Figure 20. Total active power losses for the MV network with no DGs production.
Figure 21. Total active power losses for the MV network with DGs production.
the MT at bus no. 8 (in another feeder) instead of bus no. 7, presenting better picture in terms of LV
losses reduction, as shown in Figure 19.
In all the cases for the Greek scenario, MV losses are not increased, due to the much-reduced flow on
the MV lines as Figures 20–22 show.
Table X summarizes the results for the active power losses. Significant reduction of active power
losses between scenario with and without DGs production is observed in both LV and MV network. In
addition, topology of the DG units within the Microgrid, especially as they are more scattered within
the LV feeders, can lead to additional losses reduction if the DG penetration within the day is relatively
DOI: 10.1002/etep
Figure 22. Total active power losses reduction for the MV network due to DG units’ production.
Table X. Percentage (%) reduction summation of active power losses at both LV and MV network for the
whole 24 hours and all months studied.
Months LV network MV network
MT on Bus 7 MT on Bus 8 Only line losses Total losses
January 33.45 49.86 62.49 46.88
April 31.36 49.74 68.36 48.44
July 33.47 49.26 63.75 47.48
October 29.86 49.91 71.33 49.01
Figure 23. Maximum emissions percentage avoidance for ApX prices per month studied.
high. Furthermore, percentage reduction of active power losses is higher in case where the MT is
removed for the bus 7 to the bus 8. Similar results are concluded for the MV network. There are two
cases in the MV network. The first case is referred to the active power losses of MV lines, only. The
second one is referred to the whole active power losses where in active power losses of MV lines are
added the active power losses from the 64 Microgrids. In all cases, the active power losses represent the
total 24 hours active power losses for a typical day of month.
5.4. Environmental assessment

Knowing DG production, and the losses at MV, the emissions reduction for representative cases can be
calculated for the various market scenarios studied for meeting the same demand for the typical MV
network using methodology of Section 3.3. Both DG production and losses avoidance lead to
significant reduction in the emissions.
DOI: 10.1002/etep
Table XI. Summary of emissions reduction for the studied period and Microgrid—MV network for the
typical days studied.
Months DG penetration Emissions reduction due to Microgrids
CO2 (tn) NOx (kg) SO2 (kg) PM-10 (kg)
January Max ApX 21.76 69.06 41.92 647.8
Greek scenario 16.47 91.15 50.98 734.6
Min ApX 1.310 3.070 1.690 27.50
April Max ApX 15.65 41.69 21.32 400.1
Greek scenario 4.400 80.00 33.29 608.7
Min ApX 0.140 0.200 0.100 1.500
July Max ApX 16.76 66.26 31.12 432.7
Greek scenariO 10.01 86.30 39.28 513.1
Min ApX 1.360 2.000 0.900 12.60
October Max ApX 13.86 54.37 26.49 473.6
Greek scenariO 12.58 74.33 38.89 510.7
Min ApX 0.490 4.180 1.070 20.20
Figure 23 presents the emissions reduction percentages for each month studied for ApX. Regarding
CO2 the maximum reduction can be achieved during January and max ApX–max RES combination
while for the rest of pollutants this is achieved for the same combination during October.
Table XI summarizes the results for the emissions reduction for the studied months for the scenarios
of ApX and maximum and minimum penetration for this market. In the same table, results from the
Greek Scenario application are also provided. The minimum emissions reduction is achieved for
combination of low ApX–low RES during April for all the pollutants studied. The emissions reduction
is sensitive to the DG penetration but is also very sensitive to the upstream network emissions curve,
since DG penetration by itself is not sufficient to explain the difference in the month that maximum
emission reduction occurs. For CO2 the emissions reduction is much sensitive to the RES penetration.
Operation under the Greek scenario presents relatively high emissions reduction in CO2, closer to
the results of maximum prices and scenario for ApX. For the rest of the pollutants, operation under
HTS prices reduces more the emissions than all the case studied. This is due to the fact that the DG
production is increased, especially for units that consume fuel. These units do not present that so much
lower CO2 levels compared to the marginal units of the HTS but present significantly lower emissions
levels for the rest of pollutants justifying the results below.
Figure 24 presents a comparison between the emissions avoided as a percentage compared to the
emissions without Microgrid operation. The numbers refer to the summation of all the days studied.
Another issue that has been studied is the impact of losses reduction in emissions reduction. Without
DG, losses of both the LV and MV account for about 3% of the pollutants for the days studied. With
Microgrids, the percentage of emissions reduction due to losses reduction is significantly higher and is
shown in Figure 25: It is apparent that the reduction of emissions due to losses reduction constitutes a
Figure 24. Emissions reduction percentage for all the studied days.
DOI: 10.1002/etep
Figure 25. Percentage of emissions reduction due to losses reduction compared to the total losses avoided.
Figure 26. Further reduction of emissions due to change in the topology for the Greek scenario.
Table XII. Reduction of emissions due to change in the topology.

Months CO2 (tn) NOx (kg) SO2 (kg) PM-10 (kg)
January 0.49 0.6 0.4 5.4
April 0.33 0.5 0.2 4.1
July 0.36 0.5 0.2 2.7
October 0.34 0.4 0.2 3.0
significant percentage of the total losses avoided for all the cases studies, especially for CO2. Once
again the figure refers to the total emissions reduction for all the days studied. It should be noted that for
the Greek System, losses reduction accounts for the 39% of the avoided emissions during April. For the
ApX case and especially for October, reduction in CO2 can reach 50%.
For the Greek scenario, the impact of the topology was studied in Section 5.4. The losses reduction
has as impact on further reductions in the emissions avoided as Figure 26 depicts. Mainly CO2
emissions are affected while for the rest of pollutants the change is similar. In that case the losses
reduction accounts for the 20% of the avoided emissions. Table XII provides the absolute reduction
values for the pollutants studied.
6. CONCLUSIONS
This paper deals with potential benefits one can get exploiting the Microgrid and Multi-Microgrid
concept. Presented analyses are performed using the ApX electricity market prices, as well as the
Greek electricity tariff prices as a comparative scenario, with different RES penetration levels.
DOI: 10.1002/etep
The first conclusion, attributed as an economic benefit, is the cost reduction for the end users when
the demand is met by the Microgrid’s units and the electricity prices are high. Paying back the
investment of DG is highly dependent on the technology, the subsidy environment and the market
prices and is an issue of further investigation.
Additional value of the Microgrid operation is given to the total active power losses reduction due to
reduction of the real power injected from the slack bus. The highest total active power losses reduction
(%) was observed between scenarios with and without DGs production in the LV network (Greek
Scenario, October, 49.91%) and MV network (Greek Scenario, October, 49.01%), respectively. The
percentage of LV losses in the total network losses of the whole MV network is increased from 51 to
70% as DG penetration increases. The more scattered the DG units within the Microgrid, the higher
losses reduction is achieved if the DG penetration within the day is relatively high.
Both DG sources production, with lower emission level than the upstream network, and the losses
reduction can provide significant emissions savings, especially as the RES penetration gets higher. The
losses reduction can play significant role in emissions avoidance since they may lead up to 20% of the
CO2 emissions avoidance, if the DG sources are more scattered in the LV feeders.
After some necessary assumptions were made, an idea of the magnitude of the benefits was obtained.
But, we argue that the Microgrids will only mature as a viable market alternative for consumers and utilities
when all the benefits produced by a particular Microgrid are accounted for and credited to its owners.
7. LIST OF SYMBOLS AND ABBREVIATIONS
7.1. Symbols
active_bid(xi) the bid function of the i unit as a function of its production
ai quadratic bid coefficient of the unit i (sct/kWh2)
bi linear bid coefficient of the unit i(sct/kWh)
ci constant bid coefficient of the unit i(sct/h)
emf emission factor of the thermal unit
fcu frequency that each unit is expected to be the critical unit of the system
hr hour
mo month
n number of buses
P, Q real and reactive power of the system, respectively
PL,QL real and reactive power losses of the system, respectively
po pollutant
Si total injected complex power at bus i
unno the number of the units that may be affected by the introduction of the distributed
generation
Vbus, Ibus column vector of the nodal bus voltages and injected bus currents respectively
xi production of the i unit
7.2. Abbreviations
ApX Amsterdam power exchange
DER Distributed Energy Resources
DG Distributed Generation
DMS Distribution Management System
DNO Distribution Network Operator
DSB demand side bidding
DSO Distribution System Operator
ED economic dispatch
FC fuel cell
HTS Hellenic Transmission System
HTSO Hellenic Transmission System Operator
DOI: 10.1002/etep
HV high voltage
LC Local Controller
LV low voltage
MC Microsource Controller
MGCC Microgrid Central Controller
MO Market Operator
MT microturbine
MV medium voltage
NTUA National Technical University of Athens
PEM Proton Exchange Membrane
PV Photovoltaic
RES Renewable Energy Sources
RTS Reliability Test System
SQP Sequential Quadratic Programming
UC unit commitment
WT wind turbine
APPENDIX
Table AI. Lines of the study case LV network.

Sending node Receiving node R (pu) X (pu)
Grid 1 0.0025 0.01
1 2 0.0001 0.0001
2 3 0.0125 0.00375
3 4 0.0125 0.00375
4 5 0.0125 0.00375
5 6 0.0125 0.00375
3 7 0.021875 0.004375
1 8 0.033125 0.00875
1 9 0.0075 0.005
9 10 0.015 0.010625
10 11 0.02125 0.005625
11 12 0.02125 0.005625
9 13 0.010625 0.005625
13 14 0.010625 0.005625
10 15 0.023125 0.00625
15 16 0.023125 0.00625
R and X of the lines have been calculated in power base of 100 kVA and voltage base 400 V.
Table AII. Electrical characteristics of MV transmission lines (20 kV).

Conductors 3 50 Al þ 50 St (TC)
R (V/km) 0.823
X (V/km) 0.150
(jB) B (mS/km) 45.867
Imax (A) 123
ACKNOWLEDGEMENTS
This work was supported by the EU research project ‘More Microgrids’ (contract no. SES6 -019864). The authors
gratefully acknowledge the support received from the EU for this research.
DOI: 10.1002/etep
REFERENCES
1. MICROGRIDS—‘Large Scale Integration of Micro-generation to Low Voltage Grids’, EU Contract ENK5-CT-
2002-0061 0, Technical Annex, May 2002.
2. Hatziargyriou N, Asano A, Iravani R, Marnay C. Microgrids. IEEE Power Energy Magazine 2007; 5(4):78–94. DOI:
10.1109/MPAE.2007.376583
3. Gil HA, Joos G. On the quantification of the network capacity deferral value of distributed generation. IEEE
Transactions on Power Systems 2006; 21(4):1592–1599. DOI: 10.1109/TPWRS.2006.881158
4. Brown RE. Electric Power Distribution Reliability. Marcel Dekker: New York, 2002.
5. Tande JO, Jenkins N. International standards for power quality of dispersed generation. Paper 4/7 CIRED
International Conference on Electricity Distribution, Nice, France, 1–4 June 1999.
6. Jenkins N, Allan R, Crossley P, Kirschen D, Strbac G. Embedded Generation. The Institution of Electrical Engineers:
London, UK, 2000.
7. Soultanis N, Papathanassiou S, Hatziargyriou N. A stability algorithm for the dynamic analysis of inverter dominated
unbalanced LV microgrids. IEEE Transactions on Power Systems 2007; 22(1):294–304. DOI: 10.1109/
TPWRS.2006.887961
8. Mathworks Inc. Matlabr 5.3,Reference Guide http://www.mathworks.com/
9. Hatziargyriou N, Dimeas A, Tsikalakis A. Centralized and decentralized control of microgrids. International
Journal of Distributed Energy Resources 2005; 1(3):197–212.
10. Tsikalakis A, Hatziargyriou N. Centralized control for optimizing microgrids operation. IEEE Transactions on
Energy conversion 2008; 23(1):241–248. DOI: 10.1109/TEC.2007.914686
11. Rao SS. Engineering Optimization, Theory and Practice, 3rd edn. John Willey & sons: New York, 1996.
12. Lee KM, El-Sharkawi A. Tutorial Modern Heuristic Optimization Techniques with Applications to Power Systems,
IEEE PES Winter meeting 2002, New York, IEEE PES 02TP160.
13. Costa P, Matos M. Avoided losses on LV networks as a result of microgeneration. Electric Power Systems Research
2009; 79:629–634. DOI: 10.1016/j.epsr.2008.09.014
14. Barker P, Mello R. Determining the impact of distributed generation on power system. Part I. Radial distribution
systems. Proceedings of IEEE Power Engineering Society Summer Meeting, Seattle, 2000.
15. Saadat H. Power System Analysis. McGraw-Hill: Boston, MA, 1999.
16. Tsikalakis A, Hatziargyriou N. Environmental benefits of distributed generation with and without emissions trading.
Journal of Energy Policy 2007; 35(6):3395–3409. DOI: 10.1016/j.enpol.2006.11.022
17. Papathanassiou S, Hatziargyriou N, Strunz K. A Benchmark LV Microgrid for steady state and transient analysis.
Proceedings of the Cigre Symposium on Power Systems with Dispersed Generation, Athens, April 2005.
18. Hotworkct, Combustion Technology, 2006 (http://www.hotworkct.com/engineering/alcoa.htm).
19. Pepermans G, Driesen J, Haeseldonckx D, Belmans R, D’haeseleer W. Distributed generation: definition, benefits
and issues. Journal of Energy Policy 2005; 33(6):787–798. DOI: 10.1016/j.enpol.2003.10.004
20. Azmy AM. Simulation and management of distributed generating units using intelligent techniques. PhD
Dissertation, Universitat Duisburg-Essen, 2005.
21. US Environmental Protection Agency. Available at: http://www.epa.gov/globalwarming/greenhouse/greenhouse18/
distributed.html
22. Tsikalakis A, Hatziargyriou N. Financial evaluation of renewable energy source production in microgrids markets
using probabilistic analysis. In Proceedings of the IEEE Power Tech ’05 Conference, St.Petersburg, June 2005; paper
no. 133. DOI: 10.1109/PTC.2005.4524361
23. Chowdhury N, Billinton R. A reliability test system for educational purposes-spinning reserve studies in isolated and
interconnected systems. IEEE Transactions on Power Systems 1991; 6(4):1578–1583. DOI: 10.1109/59.117005
24. Barnes M, Dimeas A, Engler A, et al. Microgrid laboratory facilities. International Conference on Future Power
Systems, 2005, 16–18 November 2005; The Netherlands, 1–6, 10. DOI 1109/FPS.2005.204229
25. Amsterdam Power Exchange historical data. Available online at: http://www.ApX.nl.
26. Hellenic Transmission System Operator. Available at: www.htso.gr.
27. E.U. research project MORE MICROGRIDS—Advanced Architectures and Control Concepts for More Microgrids.
Contract no.: SES6-019864, 2004–2009.
28. Public Power Corporation of Greece. Available at: www.dei.gr.
29. Burke J. Power Distribution Engineering: Fundamentals and Applications. Marcel Dekker: New York, 1994.
30. Pansini A. Electrical Distribution Engineering. Fairmont Press: Lilburn, GA, 2007.
31. Rudion K, Styczynski ZA, Hatziargyriou N, et al. Development of benchmarks for low and medium voltage
distribution networks with high penetration of dispersed generation. Proceedings of MEPS Conference, Wroclaw,
Poland, 2006.
DOI: 10.1002/etep

Hatziargyriou 等 - 2011 - Quantification of economic, environmental and oper

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hatziargyriou 等 - 2011 - Quantification of economic, environmental and oper

Uploaded by

Copyright:

Available Formats

EUROPEAN TRANSACTIONS ON ELECTRICAL POWER

Euro. Trans. Electr. Power 2011; 21:1217–1237

Quantiﬁcation of economic, environmental and operational

N. D. Hatziargyriou1, A. G. Anastasiadis1*,y, A. G. Tsikalakis1 and J. Vasiljevska2,3

Copyright # 2009 John Wiley & Sons, Ltd.

Figure 1. Hierarchical control structure for Microgrids.

3.1. Economic operation

Figure 2. Information exchange diagram.

3.2. Power losses estimation

3.3. Estimation of environmental beneﬁts

4. CASE STUDY NETWORKS

4.1. Typical LV network

Figure 4. Typical LV network used.

Table I. Data for the capacity of the RES of the Microgrid.

Figure 7. Typical MV network of 64 Microgrids.

4.2. Typical MV network

5.1. Economic beneﬁts due to Microgrids

5.3. Sensitivity for the Greek scenario

Table VII. MV network’s total active power losses reduction.

January April July October

PLoad,min ¼ PLoad,max ¼ PLoad,min¼ PLoad,max ¼ PLoad,min ¼ PLoad,max ¼ PLoad,min ¼ PLoad,max ¼

5.4. Environmental assessment

Table XII. Reduction of emissions due to change in the topology.

7. LIST OF SYMBOLS AND ABBREVIATIONS

Table AI. Lines of the study case LV network.

Table AII. Electrical characteristics of MV transmission lines (20 kV).

You might also like