IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO.

5, MAY 2016 999

LV Measurement Device Placement for Load Flow

Analysis in MV Smart Grids
Antonio Cataliotti, Member, IEEE, Valentina Cosentino, Dario Di Cara, and Giovanni Tinè, Member, IEEE

Abstract— This paper deals with the issue of measurement and power flows on the distribution feeders. To perform load
device placement for load flow analysis in medium-voltage (MV) flow analysis in MV distribution networks, two types of
distribution networks. The study is carried out using an innov- measurement instruments can be used: 1) power quality
ative measurement algorithm for load flow analysis developed
by the authors. It is based on low-voltage (LV) load power analyzers (PQAs) [5] and 2) phasor measurement
measurements applied on a backward/forward algorithm for units (PMUs) [6]. They should be able to provide electrical
the load flow resolution. The final aim is to identify the most data, such as currents on each distribution feeder and voltages
suitable number and placement of the LV measurement points at the substation buses [7]. These devices are generally
in order to limit the uncertainty on the power flow estimation to a connected to the MV network by means of proper voltage
target value. The proposed approach is based on the uncertainty
evaluation of the estimated power flows; a sensitivity analysis and current transducers. However, such instruments are not
is performed in order to assess the uncertainty contribution widespread diffused over distribution networks; thus, in
of each LV power measurement to the estimated power flow many cases, high instrumentation and installation investments
upstream of each MV feeder of the distribution network. The are needed, such as the costs of measurement instruments
LV measurement device placement is presented and discussed and transducers, MV cabinet and connections, and energy
in a real case study, i.e., the distribution network of the Ustica
Island. interruption for installation. This has limited the wide use
Index Terms— Advanced metering infrastructure, load
of both PMUs and PQAs in MV distribution networks.
flow (LF) analysis, meter placement, power measurements, In order to reduce the number of MV measurement points,
power quality analyzer (PQA), power system measurements, state estimation (SE) algorithms have been proposed in the
power system monitoring, smart grids. literature; such methods make use of few actual measurements
and they integrate the missing data with pseudomeasurements,
I. I NTRODUCTION exploiting the historical information (or a priori estimates) of
the active and reactive power magnitudes of each load [8], [9].
T HE INCREASING presence of distributed generation and
energy storage systems, at both medium-voltage (MV)
and low-voltage (LV) levels, is leading to important changes
Since pseudomeasurements are high-variance variables, the
quality of the estimated state variables depends on the
in the distribution network structure and management [1], [2]. number of pseudomeasurements. If the SE errors are too high
The modern smart grids are characterized by bidirectional and they do not ensure reliable network control, additional
power flows and an increased number of monitoring, com- load measurements are required [10], [11]. As regards
munication, and management systems and devices, which can this, different approaches can be found in the literature
guarantee many functionalities, such as voltage and frequency concerning the meter placement issue for distribution system
regulation, anti-islanding protection, fault location and isola- SE [10], [12]–[18].
tion, service restoration, power quality monitoring, and system Cataliotti et al. [19], [20] have proposed an innovative
diagnostic [3], [4]. measurement approach for LF analysis in MV distribution
In this framework, a fundamental role of the smart grid networks. The method makes use of only one measurement
monitoring system is the load flow (LF) analysis, which at the MV level (the voltage at the MV bus bar of the
evaluates a lot of network electrical data, such as node voltages primary substation), while the loads powers are measured at
the LV level by means of PQAs (installed in the secondary
Manuscript received June 20, 2015; revised August 25, 2015; accepted substations, the LV side of the MV/LV power transformers).
September 30, 2015. Date of publication November 6, 2015; date of current Starting from such measurements, a backward/forward algo-
version April 5, 2016. This work was supported by the University of Palermo
within the Research Grant FFR 2012-2013 through the Research Project rithm is used to evaluate the power system load flow.
entitled Sviluppo di Dispositivi di Misura Innovativi di Potenza, Energia e The use of LV measurements instead of MV ones allows
Power Quality per la Gestione delle Smart Grid Attraverso Un’opportuna to avoid or reduce the installation costs of the measurement
Infrastruttura di Comunicazione under Project 2012-ATE-0395. The Associate
Editor coordinating the review process was Dr. Edoardo Fiorucci. points. In fact, in many cases, PQAs are already installed in the
A. Cataliotti and V. Cosentino are with the Department of Energy, secondary substations to compute the total amount of energy
Information Engineering and Mathematic Models, University of Palermo, absorbed by LV loads supplied by the power transformer.
Palermo 90133, Italy (e-mail: [email protected]; [email protected]).
D. Di Cara and G. Tinè are with the Institute of Intelligent System In such cases, no further installation costs are needed. On the
for Automation, National Research Council, Palermo 90139, Italy (e-mail: other hand, in the case of new installations, the costs are
[email protected]; [email protected]). lower than those for MV installations (voltage transducers
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. are not required and LV current transducers are cheaper than
Digital Object Identifier 10.1109/TIM.2015.2494618 MV ones; the energy interruption is shorter and it involves
0018-9456 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
1000 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 5, MAY 2016

a lower number of customers). Apart from the installation losses, and these are evaluated starting from the rated data of
costs, the practical implementation of the proposed solution the short circuit and no-load tests. The LF algorithm starts
requires the acquisition of a high number of measurements. with the acquisition of the LV load power measurements and
In comparison with SE algorithms, this can lead to long the calculation of the active, reactive, and apparent powers
updating times and limited real-time performances. In the of the equivalent MV loads. After this, the voltage at each
distribution system operator (DSO) viewpoint, the aforesaid node is set equal to the value measured at the MV bus bars of
considerations can lead to the need of reducing the number the primary substation and the iteration process for load flow
of measurement points, making a tradeoff between real-time calculation begins. In brief, it consists of a current summation
features, accuracy requirements, and costs. In such cases, it method, based on the ladder iterative technique. Bus voltages
is still important to identify the most suitable number and and branch currents are updated by means of backward and
location of voltage and power measurements in the distribution forward sweeps through the network, using Kirchoff’s voltage
network. and current laws. A full description of the algorithm can be
In the light of these considerations, this paper presents a found in [20].
study aimed at identifying the most suitable placement of The measurement uncertainties of the estimated power
LV PQAs in order to limit the uncertainty on the power flow flows can be evaluated starting from the uncertainties on the
estimation to a target value when the LF algorithm of [20] measurement input variables, i.e., the LV load powers and the
is used and a reduced number of meters are considered. The slack bus MV voltage [23]. Such uncertainties are obtained by
proposed approach is based on the uncertainty evaluation on taking into account the uncertainty contributions of the mea-
the estimated power flows and a sensitivity analysis concern- surement instruments and transducers [24], [25]. The uncer-
ing the uncertainty contribution of each input variable (the tainty on the LF estimations can be obtained by means of a
LV power measurements) on the algorithm outputs (the power Monte Carlo analysis. In each iteration of the LF calculation,
flows) [21], [22]. On the basis of this analysis, a method the input variables are randomly changed within their uncer-
is proposed to identify which is the minimum number of tainty ranges; at the end of the simulation, the frequency
LV PQAs and their placement in order to keep the uncertainty distributions of all the branch power flows are obtained and the
on the power flow estimation under the desired target value. mean values and standard deviations are calculated for each
The LV meter placement method is presented and discussed distribution.
for the real case study of the distribution network of the
Ustica Island (Italy). B. Results for the Network Under Test
The paper is organized as follows. Section II summarizes the The LF algorithm was tested in the distribution network
load flow algorithm [19], [20] and the uncertainty evaluation of the Ustica Island, whose scheme is reported in Fig. 1.
procedure; the related results are presented for one of the two Two MV radial feeders depart from the diesel power plant:
MV feeders of the MV–LV Ustica network (the results for the Vittorio Emanuele and AUSL; they supply 13 and
other feeder can be found in [23]). The proposed LV meter 11 secondary substations, respectively. Two PQAs are installed
placement method is presented in Section III. Furthermore, the at the MV bus bars of the generating station at the
results are presented and discussed in Section IV, showing the beginning of each feeder; each of them is connected by
effectiveness of the proposed approach, the marginal accuracy means of three current transformers (CTs) and three voltage
adding or removing meters, and the influence of the accuracy transformers (VTs). Each substation is equipped with a
of pseudomeasurements and PQA measurements on the meter MV/LV power transformer, which is coupled to the MV bus
placement results. bar through a bypass connection. The LV load active and
reactive powers are measured by a PQA, which is connected on
II. L OAD F LOW AND U NCERTAINTY A NALYSIS the LV side of the power transformer, by means of three CTs.
A. LF Algorithm and Uncertainty Evaluation Procedure All the measurements are collected and transmitted to the
The algorithm for LF calculation of the MV distribution DSO control center by means of a High Performance Radio
network [19], [20] makes use of the measurement of the LAN (HiperLAN) communication system. The MV lines and
following quantities, which are used as input data: the three- power transformers data can be found in [20]. The PQAs have
phase voltage at the MV bus bars of the primary substation accuracy classes of 0.5 and 2 for the active and reactive power
of the distribution network and the three-phase load active measurements, respectively; CTs and VTs have an accuracy
and reactive powers for each secondary substation of the class of 0.5.
network, where such powers are measured at the LV side The LF analysis and the uncertainty evaluation have been
of the substation MV/LV power transformer. As regards the carried out by reproducing some actual conditions of power
MV network model, the secondary MV/LV substations are consumption at the LV side of each secondary substation.
the nodes and the MV lines between the substations are the The data were provided by the local DSO [20]. In such
branches. The lines are modeled considering both longitudinal operating conditions, the uncertainties on the load power
impedances and shunt admittances. In order to use the LV measurements have been calculated, considering the actual
measurements of the load powers, the network model includes values of the load power factors and currents and the spec-
the secondary substations’ MV/LV power transformers. The ifications of the measurement instruments installed in the
MV/LV power transformers and their LV loads are modeled distribution network. The whole procedure and formulas for
by their equivalent MV loads, which include the transformer the uncertainty evaluation can be found in [23], together with

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
CATALIOTTI et al.: LV MEASUREMENT DEVICE PLACEMENT FOR LOAD FLOW ANALYSIS IN MV SMART GRIDS 1001

Fig. 1. Scheme of the Ustica MV network and the details of LV PQA insertion (capital letter indicate the MV branches, while the numbers indicate the
MV nodes).

TABLE I TABLE II
F EEDER AUSL: U NCERTAINTIES ON P OWER M EASUREMENTS C OMPARISON B ETWEEN THE E STIMATED AND THE M EASURED P OWER
F LOWS AT THE B EGINNING OF THE F EEDER AUSL

UPmeas and UQmeas ) of the power flows at the beginning of

the feeder AUSL. The measurements were recorded by the
PQAs actually installed at the MV bus bars of the generating
station. Such a comparison shows the compatibility between
the simulation results and the measurements, thus confirming
the reliability of the LF algorithm, even considering the
measurement uncertainties.
the results obtained for the feeder Vittorio Emanuele in one It has to be observed that the effectiveness of the proposed
reference load condition (August 18, 2012, at 12:30). In the algorithm can depend on different factors, such as the accuracy
aforesaid reference condition, the results of the uncertainties of the network model and the data synchronicity. As regards
on the LV load power measurements for the feeder AUSL the first aspect, in [23], some simulations were carried out
are summarized in Table I. Fig. 2 shows the power fluxes in order to estimate the influence of the uncertainty on
calculated with the proposed LF algorithm and the absolute the network parameters; such an analysis showed a slight
expanded uncertainties (confidence level of 95%) calculated influence of the network model parameters on the power
for all the branch active and reactive power flows of the feeder flow uncertainties. As regards the data synchronicity, for the
AUSL. These results were obtained by means of the Ustica network, this is obtained by means of the HiperLAN
Monte Carlo analysis (105 iterations) considering the active communication system and the Supervisory Control And Data
and reactive power measurements as uncorrelated. The Acquisition system [26], which allows the DSO to collect the
Monte Carlo simulations were repeated in the hypothesis of the measurement data and to synchronize the PQAs by means
total correlation between active and reactive power measure- of an NTP server. In this sense, the critical element could
ments obtained by the same PQA at each node. In such a con- be related to the load stationarity during the measurements’
dition, no significant variations were found in the calculated acquisition time interval. As regards this, it can be assumed
power flow uncertainties [23]. Finally, Table II shows the com- that the PQA measurements are synchronized with a negligible
parison between the estimated values (Pest , Q est and related time jitter and the measurements’ acquisition time interval is
expanded uncertainties UPest and UQest ) and the measured sufficiently short (in comparison with the updating time of
values (Pmeas , Q meas and related expanded uncertainties the measurements). Such an assumption is even more valid

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
1002 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 5, MAY 2016

Fig. 2. Feeder AUSL. (a) Power fluxes and (b) related absolute expanded uncertainties (95% confidence level) in the reference load condition.

because of the reduced number of measurements (see the

next section).

III. LV M EASUREMENT D EVICE P LACEMENT

As already mentioned, when a reduced number of
measurements are considered (because of DSO requirements,
unavailability of measurement points, real-time performance
constraints, and so on), a suitable meter placement technique is
required to properly determine the number and location of the
measurement points, in order to obtain a sufficiently accurate
knowledge of the distribution system. Most existing meter
placement approaches have been formulated for the distrib-
ution system SE, where pseudomeasurements are integrated
with real measurements (often performed at the MV level by
means of a PMU), in order to ensure a suitable quality of
the estimated state variables. The proposed approaches make
use of optimization algorithms, which are aimed at determin-
ing the minimum number and position of measurements to
maintain the desired accuracy level or to reach an optimal or
pseudooptimal configuration with respect to a given accuracy
index [12]–[18]. In some cases, the algorithms take into
account further elements, such as cost/benefit ratio, equip-
ment costs, redundancy, and the impact of measurement Fig. 3. Flow chart of the meter placement procedure.
uncertainty [10], [27], [28].
In this paper, the meter placement issue is faced with the aim
to identify the most suitable number and placement of the uncertainties, the sensitivity analysis is carried out to evaluate
LV measurement points in order to limit the uncertainty on the the uncertainty contribution of each LV power measurement
power flow estimation to a desired target value, when the load on the estimated UPF. Such an analysis allows defining
flow algorithm of [20] is used and a reduced number of meters a sensitivity order (SO) for the meter placement (see the
are considered. In the following, according to the requirements next section). On the basis of this analysis, the measure-
of the local DSO of the Ustica Island, the analysis is carried ment point configuration (number and location) is identified
out with respect to the uncertainty of the power flow at the by eliminating, one by one, the measurements that have a
beginning of each feeder of the distribution network [hereafter lower contribution on the UPF uncertainty (i.e., lower SO),
indicated as upstream power flow (UPF)]. For this quantity, until reaching the predetermined uncertainty target. Thus,
a maximum uncertainty of 10% is considered as the reference the measurement points are placed only in certain nodes,
target value. whereas in the remaining, the LV load powers are estimated
as pseudomeasurements with an associated uncertainty value.
In the nodes where the presence of measurement instruments
A. Meter Placement Approach is considered, the related uncertainties are those previously
The proposed meter placement approach is based on the estimated.
uncertainty evaluation on the estimated power flows and a A flow chart of the proposed approach is reported in
sensitivity analysis concerning the uncertainty contribution of Fig. 3. Starting from the configuration with the measurements
each input variable (LV load power measurements) on the points located in all network nodes, the first point is elimi-
algorithm output variables (UPF). Starting from the estimated nated, which is the one with the lowest contribution on the

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
CATALIOTTI et al.: LV MEASUREMENT DEVICE PLACEMENT FOR LOAD FLOW ANALYSIS IN MV SMART GRIDS 1003

UPF uncertainty, i.e., the lowest SO (1). The power measure- TABLE III
ment of the selected node is then replaced by a pseudomea- U NCERTAINTY C ONTRIBUTIONS OF THE LV M EASUREMENTS AT
E ACH N ODE ON THE P OWER F LOWS AT THE B EGINNING
surement with the associated uncertainty value. In such a new OF THE F EEDERS AND R ELATED SOs
condition, the Monte Carlo simulation is executed and the new
UPF uncertainty value is estimated. If the value is still below
the reference target, another PQA is removed (the one with
SO 2) and so on (following the SO) until the uncertainty
remains below the reference target. When it rises above the
target value, the last replaced measurement point is restored,
thus obtaining the final measurement configuration.
In the following, the sensitivity analysis and the meter
placement approach are detailed for the case study of the
Ustica distribution network.

B. Sensitivity Analysis
The sensitivity analysis was carried out by evaluating the
uncertainty contribution of each LV load power measurement
on the estimated UPF. According to [21], a Monte Carlo
approach was used for the analysis. In each algorithm iter-
ation, the input data (LV load measurements) were randomly
changed within their uncertainty range, thus obtaining a proba-
bility distribution of each output variable (UPF). The standard
deviation of this distribution was assumed as the uncertainty
of the output variable. In order to evaluate the uncertainty
contribution of each LV measurement, UPF was evaluated
considering the presence of only that measurement, assuming
that in all the other nodes, the loads powers are known without
uncertainty. In such a condition, the Monte Carlo analysis was
carried out (105 iterations), randomly changing the considered
power measurement within its uncertainty range. All the other
LV measurements were kept constant at the values of the
reference load condition. In this way, the standard deviation
of UPF was obtained. The procedure was repeated for each
node of the network, thus analyzing the contribution of each
LV measurement uncertainty on the algorithm output variables.
As an example, the Monte Carlo analysis was carried out for
the reference load condition used for the uncertainty estimation
of Section II and [23]. The obtained results are reported in
Table III for both feeders; u P_L , u Q_L , and u S_L are the
uncertainty contributions of the LV measurements at each
node on active, reactive, and apparent power flows (UPF)
at the beginning of each feeder. They are listed according
to the decreasing order of u S_L ; according to this SO, the
highest uncertainty contributions on UPF are given by the LV
measurements at nodes 8 and 7, for the feeders AUSL and Fig. 4. Power flow uncertainties at the beginning of the feeder versus the
Vittorio Emanuele, respectively, and the lowest contributions number of measurement points. (a) AUSL. (b) Vittorio Emanuele.
are those of LV measurements at nodes 11 and 14. The
substation at node 9 has not been taken into account since it
were estimated as pseudomeasurements, the uncertainty
is actually disconnected. The results of the sensitivity analysis
value was set equal to 100% (assuming a Gaussian distri-
show that the SO depends on both the load power (see [20])
bution). On the other hand, in the measurement nodes, the
and the measurement instrument uncertainty, with particular
related uncertainties were set to the estimated values (see
respect to the CT contribution (see [23] and Table I).
Section II and [23]). As already mentioned, the value for the
target uncertainty was set equal to 10% (maximum uncertainty
C. LV Measurement Point Placement for active, reactive, and apparent powers at the beginning of
The SO identified in the previous section was used to iden- each feeder, as required by the local DSO).
tify the number and location of the measurement points for the Fig. 4(a) and (b) shows the uncertainties of the active,
Ustica MV network. For the nodes where the LV load powers reactive, and apparent UPFs for the feeders AUSL and

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
1004 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 5, MAY 2016

Fig. 5. Power flow uncertainties in all branches of the feeder in the case Fig. 6. Feeder AUSL. Power flow uncertainties at the beginning of the feeder
when seven PQAs are installed. (a) AUSL. (b) Vittorio Emanuele. calculated for all the possible combinations of seven PQAs.

Vittorio Emanuele, respectively. They were obtained by

varying the installed number of PQAs (following the SO).
These results show that the placement of seven PQAs allows
obtaining the estimation of UPF with an uncertainty lower
than 10%. From Fig. 4, it is also possible to deduce the impact
of adding or removing meters on the UPF uncertainties. It can
be observed that for the feeder AUSL, a marginal accuracy
increment can be obtained by adding a meter (eight meters
instead of seven); for the feeder Vittorio Emanuele, such an
increment is higher for active and apparent UPF values.
Fig. 5(a) and (b) shows the uncertainties on the power flows
in all branches of the feeders in the presence of seven PQAs.
Fig. 7. Power flow uncertainties in all branches of the feeder AUSL, in the
It can be seen that in all branches, the uncertainties are below case when six PQAs are installed.
the target value of 10%.

IV. VALIDATION AND D ISCUSSION flow uncertainties. Starting from the meter placement solution
In order to verify that the proposed strategy allows obtaining obtained with the proposed strategy (seven PQAs for each
the PQA placement solution with the lowest uncertainties on feeder, in the case study of the Ustica distribution network),
the UPFs, and a further analysis was carried out by considering the branch power flow uncertainties were evaluated in the
all the possible PQAs combinations. An iteration process was configurations obtained by adding and removing one meter
implemented in order to compare all the placement combina- (following the nodes’ SO). As an example, Figs. 7 and 8 show
tions of seven PQAs (330 combinations for the feeder AUSL, the results obtained for the feeder AUSL in the cases of
being 11 the nodes of the feeder). For each combination, six and eight measurement points, respectively. It can be
a Monte Carlo simulation was performed to calculate the observed that when a meter is removed (i.e., six PQAs are
active, reactive, and apparent power uncertainties on the placed instead of seven), the uncertainties are higher than 10%
branches power flows. The results are reported in Fig. 6; in different branches (including the one at the beginning of the
the minimum values of uncertainty were found for the feeder). On the other hand, if a meter is added (eight PQAs),
49th combination, which corresponds to the positions found a marginal uncertainty reduction is obtained.
with the proposed strategy, i.e., PQAs placed in nodes 8 , 2 , Finally, the influence of the accuracy of pseudomeasure-
12 , 7 , 6 , 9 , and 4 . ments and measurement instruments on power flow uncertainty
Further investigations were carried out in order to study and meter placement was also investigated. As regards the first
the impact of adding or removing meters on the power aspect, the branch power flow uncertainties were evaluated

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
CATALIOTTI et al.: LV MEASUREMENT DEVICE PLACEMENT FOR LOAD FLOW ANALYSIS IN MV SMART GRIDS 1005

Fig. 8. Power flow uncertainties in all branches of the feeder AUSL, in the Fig. 10. Feeder AUSL. Branch power flow uncertainties in the case with
case when eight PQAs are installed. seven measurement points and worse accuracy classes of PQAs and CTs (in
comparison with those actually installed).

this shows a relatively slight impact of the instrumentation

accuracy on the meter placement.

V. C ONCLUSION
In this paper, a study has been carried out to find the
most suitable number and placement of the LV measure-
ment points, in the case of the measurement approach for
load flow analysis in MV distribution networks proposed
in [19] and [20]. The strategy evaluates the number and
Fig. 9. Feeder AUSL. Power flow uncertainties in all branches of the AUSL placement of the LV measurement points by assuming a target
feeder, in the case with six PQA measurements and five pseudomeasurements
with an uncertainty of 50%. value for the uncertainty on the power flow estimation. The
LV measurement device placement method has been presented
and discussed in a real case study, i.e., the distribution network
by varying the pseudomeasurement uncertainties. For example, of the Ustica Island. The effectiveness of the proposed meter
Fig. 9 shows the results obtained for the AUSL feeder, placement solution has been discussed by considering different
by considering six measurement points (nodes 8 , 2 , 12 , aspects, such as the impact of varying the measurement points
7 , 6 , and 9 , according to the SO of Table III) with the on power flow uncertainties and the influence of the accuracy
uncertainties of Table I and five pseudomeasurements (in the of pseudomeasurements and PQA measurements on the meter
remaining nodes) with an uncertainty of 50%. The comparison placement results. It has been shown that the contribution of
between the results of Figs. 5 and 9 (where a pseudomeasure- the quality of pseudomeasurements is more significant than
ment uncertainty of 100% was assumed) shows a significant that of the accuracy of measurement instruments. In fact, a
reduction in the power flow uncertainties, which is comparable thorough knowledge of all nonmeasured loads knowledge is
with that achieved by adding a meter (see Fig. 8). Thus, it needed to reduce the number of measurement points and/or to
can be concluded that to obtain a significant reduction in the obtain a significant reduction in the power flow uncertainties.
power flow uncertainties, a considerably better estimation of On the other hand, the addition of only one meter can allow
all the nonmeasured loads has to be performed. Otherwise, one to obtain a suitable quality of the estimated power flows.
the addition of only one meter allows obtaining the same In this sense, the possibility of reaching a good tradeoff
uncertainty reduction. between the reduction in the measurement points and the
As regards the influence of the measurement instrument load flow estimation accuracy confirms the reliability of the
accuracy on the power flow uncertainty, the uncertainty analy- developed load flow algorithm for its practical implementation
sis and the meter placement procedure was repeated by in MV distribution systems.
considering worse accuracy classes of both PQAs and CTs
(in comparison with those actually installed in the Ustica ACKNOWLEDGMENT
distribution network). More in detail, the PQA uncertainties The authors would like to thank S. Russotto of the
were assumed equal to 1% and 3% for the active and reac- Ustica DSO Impresa Elettrica D’Anna e Bonaccorsi s.n.c. for
tive powers, respectively (instead of 0.5% and 2%, respec- all the data on the Ustica distribution network.
tively); moreover, CTs of class 1 were considered (instead of
class 0.5). As an example, Fig. 10 shows the results for the
R EFERENCES
configuration with seven measurement points for the feeder
AUSL. In comparison with Fig. 5, a marginal increase in the [1] K. Moslehi and R. Kumar, “A reliability perspective of the smart grid,”
IEEE Trans. Smart Grid, vol. 1, no. 1, pp. 57–64, Jun. 2010.
power flow uncertainties is observed and the 10% accuracy [2] J. Fan and S. Borlase, “The evolution of distribution,” IEEE Power
requirement is still fulfilled for all the branch power flows; Energy Mag., vol. 7, no. 2, pp. 63–68, Mar./Apr. 2009.

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.
1006 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 5, MAY 2016

[3] A. Kulmala, S. Repo, and P. Järventausta, “Coordinated voltage control [24] A. Cataliotti, D. Di Cara, A. E. Emanuel, and S. Nuccio, “A novel
in distribution networks including several distributed energy resources,” approach to current transformer characterization in the presence of
IEEE Trans. Smart Grid, vol. 5, no. 4, pp. 2010–2020, Jul. 2014. harmonic distortion,” IEEE Trans. Instrum. Meas., vol. 58, no. 5,
[4] H. E. Z. Farag, E. F. El-Saadany, and R. Seethapathy, “A two ways pp. 1446–1453, May 2009.
communication-based distributed control for voltage regulation in smart [25] A. Cataliotti, D. Di Cara, A. E. Emanuel, and S. Nuccio, “Current
distribution feeders,” IEEE Trans. Smart Grid, vol. 3, no. 1, pp. 271–281, transformers effects on the measurement of harmonic active power in LV
Mar. 2012. and MV networks,” IEEE Trans. Power Del., vol. 26, no. 1, pp. 360–368,
[5] A. Venturi et al., “The role of accurate measurements within smartgrids,” Jan. 2011.
in Proc. 2nd IEEE PES Int. Conf. Exhibit. Innov. Smart Grid [26] A. Cataliotti, V. Cosentino, D. Di Cara, S. Guaiana, N. Panzavecchia,
Technol. (ISGT Europe), Dec. 2011, pp. 1–6. and G. Tinè, “A new solution for low-voltage distributed generation
[6] P. Castello, P. Ferrari, A. Flammini, C. Muscas, and S. Rinaldi, “A new interface protection system,” IEEE Trans. Instrum. Meas., vol. 64, no. 8,
IED with PMU functionalities for electrical substations,” IEEE Trans. pp. 2086–2095, Aug. 2015.
Instrum. Meas., vol. 62, no. 12, pp. 3209–3217, Dec. 2013. [27] M. Pau, P. A. Pegoraro, S. Sulis, and C. Muscas, “Uncertainty sources
[7] D. Di Cara, M. Luiso, G. Miele, and P. Sommella, “A smart measurement affecting voltage profile in distribution system state estimation,” in
network for optimization of electrical grid operation,” in Proc. 19th Proc. IEEE Int. Instrum. Meas. Technol. Conf., Pisa, Italy, May 2015,
IMEKO TC 4 Symp., 17th IWADC Workshop, Adv. Instrum. Sensors pp. 109–114.
Interoperability, Barcelona, Spain, Jul. 2013, pp. 649–654. [28] R. G. Milbradt, L. N. Canha, P. B. Zorrilla, A. R. Abaide, P. R. Pereira,
[8] M. Baran and T. E. McDermott, “Distribution system state estimation and S. M. Schmaedecke, “A multicriteria approach for meter placement
using AMI data,” in Proc. IEEE/PES Power Syst. Conf. Expo. (PSCE), in distribution systems,” in Proc. 10th Int. Conf. Eur. Energy
Mar. 2009, pp. 1–3. Market (EEM), May 2013, pp. 1–7.
[9] M. Karimi, H. Mokhlis, A. H. A. Bakar, A. Shahriari, M. A. Faradonbeh,
and H. M. Rosli, “Impact of load modeling in distribution state
estimation,” in Proc. IEEE Int. Power Eng. Optim. Conf. (PEOCO),
Jun. 2012, pp. 67–71.
[10] P. A. Pegoraro and S. Sulis, “Robustness-oriented meter placement for
Antonio Cataliotti (M’01) received the M.S. and Ph.D. degrees in electrical
distribution system state estimation in presence of network parameter engineering from the University of Palermo, Palermo, Italy, in 1992 and 1998,
uncertainty,” IEEE Trans. Instrum. Meas., vol. 62, no. 5, pp. 954–962, respectively.
May 2013. He has been an Associate Professor of Electrical and Electronic Mea-
[11] M. Pau, P. A. Pegoraro, and S. Sulis, “Branch current state estimator
surements with the Department of Energy, Information Engineering and
for distribution system based on synchronized measurements,” in Proc. Mathematic Models, University of Palermo, since 2005. His current research
IEEE Int. Workshop AMPS, Aachen, Germany, Sep. 2012, pp. 1–6. interests include power quality measurements, power line communications,
[12] M. E. Baran, J. Zhu, and A. W. Kelley, “Meter placement for real-time and smart grids.
monitoring of distribution feeders,” IEEE Trans. Power Syst., vol. 11,
no. 1, pp. 332–337, Feb. 1996.
[13] H. Wang and N. N. Schulz, “A revised branch current-based distribution
system state estimation algorithm and meter placement impact,” IEEE
Trans. Power Syst., vol. 19, no. 1, pp. 207–213, Feb. 2004.
[14] R. Singh, B. C. Pal, and R. B. Vinter, “Measurement placement in Valentina Cosentino received the M.S. and Ph.D. degrees in electrical
distribution system state estimation,” IEEE Trans. Power Syst., vol. 24, engineering from the University of Palermo, Palermo, Italy, in 2001 and 2005,
no. 2, pp. 668–675, May 2009. respectively.
[15] R. Singh, B. C. Pal, R. A. Jabr, and R. B. Vinter, “Meter placement for She is currently an Assistant Professor of Electrical and Electronic Mea-
distribution system state estimation: An ordinal optimization approach,” surements with the Department of Energy, Information Engineering and Math-
IEEE Trans. Power Syst., vol. 26, no. 4, pp. 2328–2335, Nov. 2011. ematic Models, University of Palermo. Her current research interests include
[16] P. Janssen, T. Sezi, and J.-C. Maun, “Meter placement impact on power, energy and power quality measurements, detection of disturbances
distribution system state estimation,” in Proc. 22nd Int. Conf. Exhibit. sources in power systems, digital signal processing, virtual instrumentation,
Electr. Distrib. (CIRED), Jun. 2013, pp. 1–4. and smart grids.
[17] J. Liu, F. Ponci, A. Monti, C. Muscas, P. A. Pegoraro, and S. Sulis,
“Optimal meter placement for robust measurement systems in active
distribution grids,” IEEE Trans. Instrum. Meas., vol. 63, no. 5,
pp. 1096–1105, May 2014.
[18] M. G. Damavandi, V. Krishnamurthy, and J. R. Marti, “Robust meter Dario Di Cara received the M.S. and Ph.D. degrees in electrical engineering
placement for state estimation in active distribution systems,” IEEE from the University of Palermo, Palermo, Italy, in 2005 and 2009, respectively.
Trans. Smart Grid, vol. 6, no. 4, pp. 1972–1982, Jul. 2015. He is currently a Researcher with the Institute of Intelligent System
[19] A. Cataliotti, P. Russotto, D. Di Cara, E. Telaretti, and G. Tinè, for Automation, National Research Council, Palermo. His current research
“New measurement procedure for load flow evaluation in medium interests include power quality measurements, characterization of current
voltage smart grids,” in Proc. IEEE Int. Instrum. Meas. Technol. Conf., transducers in nonsinusoidal condition, harmonic power measurements, power
Minneapolis, MN, USA, May 2013, pp. 517–522. line communications, and smart grids.
[20] A. Cataliotti, V. Cosentino, D. Di Cara, P. Russotto, E. Telaretti, and
G. Tinè, “An innovative measurement approach for load flow analysis
in MV smart grids,” IEEE Trans. Smart Grid, May 2015. [Online].
Available: http://dx.doi.org/10.1109/TSG.2015.2430891
[21] Evaluation of Measurement Data—Supplement 1 to the ‘Guide to
the Expression of Uncertainty in Measurement’—Propagation of Giovanni Tinè (M’04) received the M.S. degree in electronic engineering and
Distributions Using a Monte Carlo Method, document JCGM 101, 2008. the Ph.D. degree in electronics, computer science, and telecommunications
[22] Uncertainty of Measurement—Part 3: Guide to the Expression of engineering from the University of Palermo, Palermo, Italy, in 1990 and 1994,
Uncertainty in Measurement (GUM:1995), ISO/IEC Guide 98-3, 2008. respectively.
[23] A. Cataliotti, V. Cosentino, E. Telaretti, D. Di Cara, and G. Tinè, He is currently a Researcher with the Institute of Intelligent System for
“Uncertainty evaluation of a backward/forward load flow algorithm for Automation, National Research Council, Palermo. His current research inter-
a MV smart grid,” in Proc. IEEE Int. Instrum. Meas. Technol. Conf., ests include electromagnetic compatibility of power electrical drive systems,
Pisa, Italy, May 2015, pp. 1279–1284. power-line communications, and smart grids.

Authorized licensed use limited to: Universidad de Costa Rica (UCR). Downloaded on May 20,2023 at 04:39:43 UTC from IEEE Xplore. Restrictions apply.