Smart Metering for Low Voltage Electrical

Distribution System using Arduino Due

Dario De Santis, Domenico Aldo Giampetruzzi, Gaetano Abbatantuono, Massimo La Scala (Fellow IEEE)
Dipartimento di Ingegneria Elettrica e dellInformazione
Politecnico di Bari
Bari, Italy
dario.desantis@poliba.it, giampetruzzi.domenico@gmail.com, massimo.lascala@poliba.it

AbstractIn recent years, low voltage electrical grids are many different applications and this makes it very adaptable
experiencing the growing implementation of distributed for further additional improvements.
generation systems such as solar and wind generators, EV
charging systems and storage devices. For this reason the use of a Furthermore, the implemented system can acquire time and
smart management system for every active component becomes position data for creating a time data logger and geo-localizing
necessary to acquire and provide real time measurements of each device, thus offering the possibility of real time fault
main parameters of power grids and execute automatic actions. localization. Fast identification and correction of faults and
In this work, the design and implementation of a cheap, flexible failures, together with other issues normally requiring the
and reliable smart monitoring system is proposed. attention of utility companies, can reduce the overall power
outage duration in a significant way [8]. An excellent
KeywordsArduino Due, grid control, low voltage electric reliability has been reached using the Arduino Due Wi-Fi
grids, smart grids, smart metering shield, because it makes possible to get real time
measurements in two different ways: by an external web
I. INTRODUCTION server [9] or through a local one in case of failure of the first.
With the widespread diffusion of renewable and The overall reliability and robustness of the developed device
distributed energy sources such as solar or wind power plants, is further increased by the galvanic isolation between power
the energy flows through different grids, and in particular on circuits and electronic circuits.
medium and low voltage systems, have become bidirectional.
Distribution systems therefore have become active grids The smart monitoring prototype has been realized into
needing appropriate means to measure and manage extremely three distinct phases: firstly an OPAL RT OP 5600 real time
variable operating conditions. Many important progresses in digital simulator has been used to simulate the electrical grid
this direction have been made on high and medium voltage and loads, and test the developed software; then, custom made
grids, where great research and development efforts have been electronic boards have been implemented to allow the
made to realize modern power electronic devices that can truly connection between transducers and Arduino Due ADC
help systems operators to face a great range of operational inputs. Finally, the meter has been tested on possible
issues such as primary and secondary regulation, voltage application scenarios.
support, problems due to flow inversions, load unbalancing, II. SMART MONITORING SYSTEM DESCRIPTION
congestions and so on. Low voltage grids, instead, have been
still very slightly touched by this deep empowerment of The monitoring system is made of two sections: a first one
electric networks, although they probably constitute the most which connects the power circuit to the ADC inputs, and a
important infrastructure of energy distribution system inside second that implements digital signal processing. LEM LV25-
urban environments. These electrical networks, moreover, P voltage transducers with a 2.5 kV insulation voltage at
have very large extensions, so it is very important to use industrial frequency are used to acquire voltage measurements.
cheap, flexible and reliable monitoring systems [1,2]. The connection between grid and voltage transducer has been
realized through a 30 k resistor, which allows to gain the
The main goal of this paper is the realization of a smart maximum measurable voltage level (300 Vrms). In order to
monitoring system [3,4] based on an embedded solution which realize the current transduction system, LEM LA55-P
has been developed using Arduino Due, a very flexible and transducers have been used; they can acquire current
cheap microcontroller, especially for data processing purposes measurements up to 50 Arms with a frequency bandwidth of
[5]. Previous versions of this microcontroller have been 200 kHz. Each of these transducers uses closed loop Hall
proposed for similar applications [6,7]; the attention has been effect to ensure galvanic isolation between power and
focused also on the possibility to make automatic electronic circuits and provide high bandwidth and linearity
interventions on the grid by means of the same device used for [10].
monitoring purposes. The Arduino Due controller has been
chosen also for this reason: it has a large memory available for The first section of the device adapts output measurement
signals, given from voltage and current transducers, and the

978-1-5090-2370-7/16/$31.00 2016 IEEE

 analog input signals of the microcontrooller. Appropriate
attenuation and offset values must be set beccause the Arduino
ADC admits unipolar signals and maximuum instantaneous
values equal to 3.3 V. The developed electroonic boards of the
device allow a voltage range of 0-3 V for both voltage and
current measurements.
The second section holds the Arduino Duue microcontroller,
the Wi-Fi shield and the GPS module. Arduino Due is
characterized by 84 MHz clock speed, 512 kB k flash memory,
54 I/O digital and 12 analog inputs. The Wi-Fi module uses a
802.11 b/g connection [11] and two encrypption types: WEP
and WPA2. The connection with Arduino Due has been made
via a SPI port. The GPS module used in this t device is the
EM506 GLOBALSAT, having a very high sensitivity and an
extremely fast Time To First Fix (TTFF). The link between
GPS module and the microcontroller has beenn realized through
the serial pins at 4800 baud rate. The protottype is fed by two
TRACOPOWER power supplies. In tab.1 thhe main technical Fig.2. Smart Meter prototype
parameters of the realized monitoring system m are shown. The
simplified schematic representation is shownn in fig. 1, while in
fig. 2 the prototype of the device can be seen.
III. MATHEMATICA
AL FORMULATION
TABLE I. TECHNICAL PARAMETTERS The main operational flow chart of the monitoring system
Parameters Vaalues
is shown in fig.3. After the variables initialization, a first
control is executed to searchh for possible anomalies; the
Nominal Voltage 2330 V algorithm is then cyclically reppeated for the whole duration of
Nominal Frequency 500 Hz the acquisition process.
Current Frequency Bandwidth 2000 kHz

Nominal Current 5 A
50

Nominal Power 34.55 kVA

Arduino Due Clock Speed 84 MHz

Arduino Due Memory Size 5112 kB

ADC Resolution 122 bit

GPS Positional Accuracy 2.5 m

Fig.1. Smart Meter schematization

This work was developed within the project PON Res Novae
N (Reti, Edifici,
Strade, Nuovi Obiettivi Virtuosi per lAmbiente e lEnerggia), supported by the Fig.3. Main flow chart
Italian Ministry for Education, University and Researchh under the research
and competitiveness program to promote Smart Citiees Communities and
Social Innovation.
A. Phase Calculation Algorithm dqpll (k) = Kp pll v q (k) + ui (k) + ref (k) (5.2)
Calculation is iterated with a sampling time T equal to 1
ms. After acquiring discrete values corresponding to their
dq (k) = dq (k 1 ) + dqpll (k)T (5.3)
respective analog inputs, the conversion is made to provide
numerical values of voltage inputs also considering the offset
value introduced by acquisition boards. In the following In order to eliminate the effects of continuous or high
equations, x value is referred to instantaneous values of frequency harmonic components into voltage or current, two
voltage or current for the ith phase. Equations (1.1-.2) give two low-pass digital filters described by transfer functions (6.1-.3)
instantaneous values in the reference system alpha-beta,
are adopted. They extrapolate the continue component for d-
starting from a single real quantity x(k).
axis and q-axis to obtain the values of the fundamental
component for phase voltage and phase current as:
x (k ) = x (k )
(1.1)
m( z + 1)
x ( z ) = xd ( z) (6.1)
x (k ) = x (k ) d zn
(1.2)
m( z + 1)
x q ( z) = x q ( z) (6.2)
Introducing a 5 ms delay (equivalent to 90 electrical zn
degrees) the imaginary component beta x can be created. 2 1 q 1
Consequently the value can be calculated as follows: q= , m= , n= (6.3)
T q +1 q +1

0.005
= (2) The dq transfer functions of the filters are implemented in
T the phase calculation algorithm by equations (7.1) and (7.2).

In this way it is possible a fast calculation of voltage or x d ( k ) = x d ( k 1) n + x d (k)m + x d (k 1)m (7.1)

current magnitude in the frame, but is hard to be selective
towards a particular harmonic component using this reference
frame. By representing magnitudes in a rotating frame having x q ( k ) = x q ( k 1) n + x q (k)m + x (k 1)m (7.2)
q
the same pulsation of the grid voltage, it is possible to work on
constant values to examine the fundamental component for
currents and voltages. The matrix equation which allows the The time constant of the filter is set equal to 0.1 s to have
change in the reference system is shown in expression (3), a total calculation time of 0.5 s. For this reason, the algorithm
where dq(k) is the angle of the synchronous rotating frame at is repeated phase by phase every 0.5 seconds. Once obtained
kth step. the average values for both direct and quadrature components,
voltage and current amplitudes can be evaluated by means of
equation (8).
x d (k ) cos( dq(k)) sin ( dq (k)) x (k )
= (3)
x q (k ) - sin( (k)) cos( (k)) x ( k ) 2 2
dq dq x dq ( k ) = x d (k ) + x q ( k ) (8)

dq can be evaluated by a PLL according to (4.1-.2), and Finally, the calculation of active and reactive powers for
coincides with the angle of the rotating vector associated to the each phase can be easily obtained as shown in equations (9.1-
phase voltage according to the Voltage Orientated Control .2). The phase calculation algorithm is showed in figure 4.
(V.O.C.) technique [12]. Thus, phase of the measured voltage
is:
1
P( k ) = (v d (k )i d (k ) + v q (k )i q (k)) (9.1)
dqpll (t) = Kp pll vq(t) + Ki pll vq(t)dt + ref (t) (4.1) 2

1
Q(k ) = (v q(k )i d (k) v d (k)i q (k)) (9.2)
dq (t) = (t)dt (4.2) 2
dqpll

The following equations (5.1-.3) are solved after

discretizing integral operator in (4.1-.2) using the backward
difference discretization method, where ui is the output of
integral contribute in the PI regulator.

ui (k) = ui (k 1 ) + Ki pll v q (k)T (5.1)

 quadrature component of phase a voltage and applying the
same technique applied in the Phase Calculation algorithm.
(k)
dqpll
f(k) = (11)
2
The digital low-pass filter expressed by the equations
(12.1-.2) has been adopted in order to eliminate the numerical
ripple due to computational errors and to extrapolate the
frequency value free from error.

m(z + 1 )
f(z) = f(z) (12.1)
zn

f(k) = f(k 1 )n + f(k)m + f(k 1 )m (12.2)

D. Post Processing
The post processing algorithm calculates: phase and line
rms voltages and rms current values starting from module
values calculated for each phase; phase apparent power
starting from active and reactive powers calculated for each
phase as shown in equations (13.1-.4) and finally active,
reactive and apparent three-phase power Pt, Qt, At values and
the equivalent power factor P.F. by means of equations (14.1-
.4). A gain constant is applied for obtaining correct values and
compensating attenuation introduced by the acquisition board.
v idq
Vi = (13.1)
rms 2

i idq
Ii = (13.2)
rms 2
Fig.4. Phase calculation alghoritm flow chart

B. Concatenate Magnitudes Calculation v ijdq

Vij = (13.3)
The previously described algorithm has been adopted even rms 2
to calculate line voltages, but in this case equations (1.1-.2) are
applied to the difference between voltage values of analog
related inputs: Ai = P 2 + Q 2 (13.4)
i i
x (k ) = vi (k ) v j (k )
(10)
Pt = Pa + Pb + Pc (14.1)
In this way, for example, for the calculation of the voltage
between phase a and phase b vab the value of x(k) will be equal
to the difference between the value of va(k) and vb(k) at kth Q t = Qa + Qb + Qc (14.2)
step. The PLL calculates, by means of equations (5.1-.3), the
phase of the line voltages for synchronizing the rotating At = Aa + Ab + Ac (14.3)
reference system with the rotating vector associated to the line
voltage. By means of equation (8), the module of the line
P . F . = P /A (14.4)
voltage is calculated. The same procedure is repeated for each t t
of the three concatenate voltages combinations. The equations
(9.1-.2) were not used. E. Wi-Fi Comunication
The collected data are finally sent to an external web
C. Frequency Calculation server using the Wi-Fi module. The web server used in this
Frequency is given by equation (11), which uses the work is the free one offered by www.emoncms.org; a
pulsation dqpll calculated by means of the PLL. Pulsation is dashboard has been also realized by using tools offered the
obtained by equations (5.1-.2) at kth step, after calculating the same site. The user interface is accessible using PC, tablets or
smart phones.
 IV. TEST CASES the smart meter refer to the values
v displayed on the serial
The smart monitoring system was tested t connecting monitor of Arduino board.
balanced and unbalanced loads to the low voltage
v grid. Tests
TABLE III. COMPAR
RISON OF MEASUREMENTS
have been carried on four different operatiing conditions, in
order to prove the sensitiveness of the prrototype on many Phase Fluke 2887 Smart meter Difference[%]
possible scenery. For Case 1, the monitoringg device has been A 220.7 220.9 0.09
connected to a simple, single-phase load reqquiring about 700
W just on phase a; in Case 2 the connection wasw made with an Voltage [V] B 217.7 218.4 0.32
unbalanced three-phase load having P.F. equals to 1. In Case 3 C 219.1 219.8 0.32
an induction motor with star-connected wiindings acts as a
Current [A]
three-phase balanced load, with no addittional mechanical (Case 1)
A 3.32 3.33 0.30
loads. Finally, in Case 4 the same induuction motor was
A 3.38 3.39 0.30
connected to a line through a single-phase fed drive. In this Current [A]
case the motor has delta connected windiings and still no (Case 2) B 7.05 7.09 0.57
mechanical loads supplied. Tab. II shows nuumerical values of
C 7.77 7.73 0.51
active and reactive power for study cases, obtained
o from the
data sheets of the different equipments actingg as loads. A 1.56 1.57 0.64
Current [A]
(Case 3) B 1.59 1.60 0.63
TABLE II. ACTIVE AND REACTIVE POWER VALU
UES FOR TEST CASES
C 1.60 1.61 0.63
Phase of supply P [kW] Q [kVAR]
Current [A]
A 0.7 0 C 0.81 0.82 1.23
(Case 4)
Case 1 B 0 0

C 0 0

A 0.7 0

Case 2 B 1.6 0

C 1.7 0

A 0.06 0.34

Case 3 B 0.06 0.34

C 0.06 0.34

A 0 0

Case 4 B 0 0

C 0.18 0.05

V. TEST RESULTS
Fig.5. Case 1: single-phase load
The following figures 5-8 represent the dashboard
d built on
the web server reporting the active (red gauuges) and reactive
(blue gauges) power measurements collectedd and processed by
the developed smart monitoring system. It can
c be appreciated
how these values are very close to the nomiinal ones reported
in table II. The other reported measuremennts concern: phase
and line voltages, phase currents, apparennt power on each
phase (green gauges), electrical frequency and power factor
values. In order to identify the exact loocalization of the
acquired measurements, the dashboard alsso provides their
relative geographical coordinates (longitude and latitude) and
a small map. In order to assess the accuraccy of the obtained
data, a Fluke 287 multimeter, whose volltage and current
precision ranges are respectively equal to 0.44% and 0.6%, has
been assumed as reference device. Tabble III shows a
comparison between a selection of variablles (voltages and
currents) measured using Fluke 287 and thhe proposed smart
meter. The per-cent deviation between the twwo meters resulted
in the worst case less than 1.30%. Measureements taken with Fig.6. Case 2: three-phase unbalancedd load
 The perceptual use of microocontroller memory during tests
was lower than 10%, so, for future works, it could be possible
to implement additional algoritthms for real time managing of
low voltage grids, or to adjust the proposed devices as smart
power quality meters, reachingg the goal to obtain a flexible
and smart low voltage grid.

REFER
RENCES
[1] G. Aurilio, D. Gallo, C. Landi, M.
M Luiso, V. Cigolotti, G. Graditi, "Low
cost combined voltage and cuurrent transducer for Smart Meters",
Proceedings of IEEE Internationnal Instrumentation and Measurement
Technology Conference I2MTC C, pp. 1459-1464, 12-15 May 2014,
Montevideo, Uruguay
[2] G. Aurilio, D. Gallo, C. Landi, M. M Luiso and G. Graditi , "A low cost
smart meter network for a sm mart utility" , Proceedings of IEEE
International Instrumentation andd Measurement Technology Conference
I2MTC, pp.380 -385, 12-15 May 2014, Montevideo, Uruguay
[3] G. R. Barai, S.Krishnan, B.Venkatesh,
B Smart Metering and
Functionalities of Smart Meters in
i Smart Grid - A Review, 2015 IEEE
Fig.7. Case 3: three-phase balanced induction motor in no load condition Electrical Power and Energy Coonference (EPEC), pp. 138-145, 26-28
October 2015. London, UK
[4] D. J. Zheng, W. Gao, L. Lin, "Smart Meters in Smart Grid: An
Overview," 2013 IEEE Green Teechnologies Conference , pp.57-64, 4-5
April 2013, Denver (CO)
[5] W.J. Esposito, F.A Mujica,.D.G. Garcia, G.T.A. Kovacs, The Lab-In-
A-Box project: An Arduino comppatible signals and electronics teaching
system in Signal Processingg and Signal Processing Education
Workshop (SP/SPE), 2015 IEEE pp. 301-306, 9-12 August 2015, Salt
Lake City (UT)
[6] R.W. Fransiska et al., Electrticaal Power measurement Using Arduino
Uno Microcontroller and Labview, 3rd International Conference on
Instrumentation, Communicatioons, Information Technology and
Biomedical Engineering, 7-8 Novvember 2013, Bandung, Indonesia
[7] M. Masudur Rahman et al., Arrduino and GSM Based Smart Energy
Meter for Advanced Metering and a Billing System, 2nd International
Conference on Electrical Engineering and Information &
Communication Technologi, 21-223 May 2015, Dhaka, Bangladesh
[8] S. Depuru, L. Wang, V. Devabhaaktuni, and N. Gudi, "Smart Meters for
Power Grid-Challenges, Issues,, Advantages and Status" in Power
Systems Conference and Expositioon (PSCE), pp. 1-7, 20-23 March 2011,
Phoenix (AR)
[9] A. Arif, M. Al-Hussain et al., "EExperimental study and design of smart
Fig.8. Case 4: single-phase induction motor in no load condition energy meter for the smart gridd", 2013 International Renewable and
Sustainable Energy Conference (IRSEC),
( pp. 515-520, 7-9 March 2013,
VI. CONCLUSIONS Ouarzazate, Morocco
In this paper, a prototypal smart meterinng and monitoring [10] C. D. Oancea, C.Dinu, LEM Transducers Interface for Voltage and
device implementing Arduino Due board haas been presented, Current Monitoring in 9th Internaational Symposium on Advanced Topics
in Electrical Engineering (ATE EE), pp. 949-952, 7-9 March 2015,
for real time acquisition and visualization of the main Bucharest, Romania
parameters of low voltage electric grids. The values are [11] Z. Lipocak, M. Bokovic, Survvey of Smart Metering Communication
displayed on a web browser and can be coontroller with any Technologies in IEEE EUROC CON 2013, pp. 1391 1400, 1-4 July
mobile device connected to internet. Test reesults confirm that 2013, Zagreb, Croatia
the monitoring system guarantees good perfoormances in terms [12] S. Lee, J.-H.Lee, H. Cha, G Grid synchronization PLL robust to
of accuracy, reliability and simplicity of thhe user interface. frequency variation, unbalanced and distorted voltage, IEEE Energy
Thanks to LEM transducers used for volltage and current Conversion Congress and Expoosition (ECCE), pp.1150-1155, 17-22
September 2001, Phoenix (AR)
acquiring and the presence of external poower supplies the
realized system also owns a remarkable robbustness. Another
very important feature is the low cost: each single
s device, able
to monitor in real time just a single point of measure, has an
approximate total cost of 200 , and this value
v could easily
decrease in case of large scale production.