JEE-Vol 6, No 3,2018 PDF

Journal of Electrical
Engineering
Volume 6, Number 3, May-June 2018 (Serial Number 23)
David Publishing
David Publishing Company

www.davidpublisher.com
Publication Information:
Journal of Electrical Engineering is published bimonthly in hard copy (ISSN 2328-2223) by David Publishing
Company located at 616 Corporate Way, Suite 2-4876 Valley Cottage, NY 10989, USA.
Aims and Scope:

Journal of Electrical Engineering, a bimonthly professional academic journal, covers all sorts of research on
electronics and microelectronics, power engineering, energy materials, signal processing, communications and
networking, automatic control system, optoelectronics, bioelectrical engineering and other relevant fields.
Editorial Board Members:

Anjan Biswas (USA), Marian Apostol (Romania), Gurevich Yuri Genrijovich (Mexico), Nader Shehata (Egypt),
Radian G. Belu (USA), Ivan Yukhymovych Protsenko (Ukraine), Mário Pedro Guerreiro Marques da Silva (Portugal),
Vladimir A. Levchenko (Russia), Waqas Ahmed Imtiaz (Pakistan), A. Sertap Kavasoglu (Turkey), Amar Rouane
(France), Joon Hoon Park (Korea), Adel Bouallegue (Tunisia), Yuri Ivanovich Choni (Russia), Sameir M. Ali Hamed
(Sudan), Tran Cong Hung (Vietnam), Behrooz Shahsavari (USA), Miryala Muralidhar (Japan), Fethi CHOUBANI
(Tunisia), Igor Bdikin (Portugal), Hazry Desa (Malaysia), Slava Yordanova (Bulgaria), Jimsher Aneli (Georgia),
Takashiro Akitsu (Japan), Cheng Song (China), Jinsoo Song (South Korea), Ibrahim Fathy El-Said Tarrad (Egypt),
Nan Wang (USA), Salvatore Magazu (Italy), Patrick VAUDON (France), Aexksi Gerasimovi (Georgia) and so on.
Manuscripts and correspondence are invited for publication. You can submit your papers via web submission, or
E-mail to electrical@davidpublishing.com. Submission guidelines and web submission system are available at
http://www.davidpublisher.com.
Editorial Office:
616 Corporate Way, Suite 2-4876 Valley Cottage, NY 10989, USA.
Tel: 1-323-984-7526, 323-410-1082; Fax: 1-323-984-7374, 323-908-0457.
E-mail: electrical@davidpublishing.com
Copyright©2018 by David Publishing Company and individual contributors. All rights reserved. David Publishing
Company holds the exclusive copyright of all the contents of this journal. In accordance with the international
convention, no part of this journal may be reproduced or transmitted by any media or publishing organs (including
various websites) without the written permission of the copyright holder. Otherwise, any conduct would be
considered as the violation of the copyright. The contents of this journal are available for any citation. However, all
the citations should be clearly indicated with the title of this journal, serial number and the name of the author.
Abstracted/Indexed in:
Cambridge Science Abstracts (CSA) J-Gate
China National Knowledge Infrastructure (CNKI) Polish Scholarly Bibliography (PBN)
Google Scholar Ulrich’s Periodicals Directory
CrossRef WorldCat
Subscription Information:
Price (per year): Print $300.

616 Corporate Way, Suite 2-4876 Valley Cottage, NY 10989, USA.
Tel: 1-323-984-7526, 323-410-1082; Fax: 1-323-984-7374, 323-908-0457.
E-mail: order@davidpublishing.com
D DAVID PUBLISHING

www.davidpublisher.com
Journal of Electrical
Engineering
Volume 6, Number 3, May-June 2018 (Serial Number 23)
Contents
Investigation and Analysis
129 Availability of Lithium Ion Batteries from Hybrid and Electric Cars for Second Use: How to
Forecast for Germany until 2030
Enrique Machuca, Fabian Steger, Johanna Vogt, Katja Brade and Hans-Georg Schweiger
144 The Study of Demagnetization of the Magnetic Orientation of Permanent Magnets for IPMSM
with Field-Weakening Control under Hot Temperature
Noriyoshi Nishiyama, Hiroki Uemura and Yukio Honda
151 A Mixed-Mode Biquad Employing OTAs and Grounded Capacitors

Takao Tsukutani, Yasutomo Kinugasa and Noburu Yabuki
156 A Simple Traffic Model Including Possibility of Overtaking

Reuben Thieberger
Techniques and Methods
161 Protection Coordination Optimization for FREEDM (Future Renewable Electric Energy Delivery
and Management) System
Mohamed F. Kotb, Magdi El-Saadawi and Eman H. El-Desouky
177 Control Algorithm for Non-isolated Supercapacitor Based Kinetic Energy Recovery System
Luka Lopin, Slobodan Vukosavić and Nikola Lepojević
186 An Intelligent Robot with Infinity Controls Making Life Easier for People with Disabilities and
Aging
Lara. Al Chamy, Mohamad Saad El Dine, Abd. El Kader El Rafeii and Ahmad. El Rafhi
Journal of Electrical Engineering 6 (2018) 129-143
doi: 10.17265/2328-2223/2018.03.001
D DAVID PUBLISHING
Availability of Lithium Ion Batteries from Hybrid and

Electric Cars for Second Use: How to Forecast for
Germany until 2030
Enrique Machuca*1, Fabian Steger*2,3, Johanna Vogt3,4, Katja Brade1 and Hans-Georg Schweiger1,3
1. Center of Automotive Research on Integrated Safety Systems and Measurement Area (CARISSMA), Technische Hochschule
Ingolstadt, Ingolstadt 85049, Germany
2. School of Engineering, Royal Melbourne Institute of Technology (RMIT), Melbourne, VIC 3000, Australia
3. Faculty of Electrical Engineering and Computer Science, Technische Hochschule Ingolstadt, Ingolstadt 85049, Germany
4. Fraunhofer Institute for Transportation and Infrastructure Systems IVI, Dresden 01069, Germany
* Contributed equally
Abstract: Due to growing numbers of sold HEV (hybrid electric vehicles), PHEV (plug-in hybrid electric vehicles), and BEV (battery
electric vehicles), new market opportunities to reuse or recycle old lithium ion batteries arise. Thus, a forecast of available batteries
caused by accidents or from end-of-life vehicles was carried out using a mathematical model. Input data were obtained from an estimate
of newly registered hybrid and electric vehicles in Germany from 2010 until 2030, from the accident rate of cars in Germany, and from
the average cars’ lifetime. The results indicate that (a) the total amount of available second use batteries in 2030 will be between
130,000 units/year and 500,000 units/year, (b) the highest amount of batteries will be obtained from end-of-life vehicles not from
accident vehicles, although most batteries from accident vehicles will be suitable for 2nd use, and (c) the quantity of hybrid, plug-in
hybrid, and electric car batteries available for reuse will continue to rise after 2030.
Key words: Forecast, model, lithium ion batteries, second use, recycling, HEV, PHEV, BEV.
1. Introduction second life [4]. The second use of low and high voltage
lithium ion batteries coming from car accidents or from
The sales of HEV (hybrid electric vehicles), PHEV
end-of-life vehicles is becoming a reality, which is
(plug-in hybrid electric vehicles), and BEV (battery
already explored on an initial stage by several
electric vehicles) are growing and this trend is expected
companies including OEM like Daimler and BMW [5].
to continue [1]. One important factor is the current
intention to ban diesel engines from cities due to their 2. Purpose
high level of pollutant emissions [2]. An even more
Aware of this situation, the purpose of the present
crucial factor for the future of hybrid and electric cars
research was to create a forecast for the German market
besides decreasing prices may come from politics via
based on existing literature and a market research
EU directives or German Government incentives [3].
analysis. It aims to predict the amount of available
Considering the increasing production of HEV,
lithium ion batteries from accidents and end-of-life
PHEV, and BEV worldwide and therefore in Germany
vehicles until 2030.
too, a new business opportunity arises: the batteries’
To quantify the amount of expected batteries
Corresponding author: Enrique Machuca, Dipl.-Ing., available for second use, a mathematical model was
industrial engineer, research fields: second use of lithium ion
created and implemented using the software Matlab
batteries, lithium ion battery testing, and battery system
development. (The Mathworks, Inc., USA). The model received three
130 Availability of Lithium Ion Batteries from Hybrid and Electric Cars for Second Use:
How to Forecast for Germany until 2030
information about the age and size of the stock in the

future is necessary. To obtain a picture of
sold/registered cars containing lithium ion batteries in
the next years, different literature sources like scientific
papers, publications, and reports from companies and
public institutions were carefully analyzed.
Fig. 1 Input and output parameters of the model.
To further obtain information about different battery
different input parameters and allowed obtaining capacities, this study aimed to solicit the amount of
two different outputs, as can be seen in Fig. 1. The input new registered HEV, PHEV, and BEV separately per
parameters for the model were: year. But, as not all found predictions specified the cars’
(1) Number of first registrations of hybrid and technology as HEV, PHEV, or BEV, also EV, as a
electric vehicles per year in Germany from 2010 until global term, has been included as a separate category in
2030; the analysis.
(2) Accident rate; As the study was focused on the German market,
(3) Vehicle life duration1. forecasts referring to different markets (e.g. USA,
Due to a big uncertainty about future initial China, Japan or even Europe or worldwide) were
registration numbers for electric propelled cars, three rejected. Applying a correction factor to those values to
fundamental scenarios (e.g. optimistic, medium and get numbers for Germany would have been imprecise
pessimistic) were considered for the input “First and unreliable.
registrations of cars”. The starting point to build the future scope of
For the input parameters “Accident rate” and “Car available lithium ion batteries until 2030 was to
life duration”, it was assumed that electric vehicles evaluate known (not forecasted) information provided
behave similarly to average cars during the last years. by the Kraftfahrt-Bundesamt (KBA, German Federal
This information will be detailed in the next chapter. Motor Transport Authority). The first registration
Finally, the model provides two different outputs: numbers sorted by technology (e.g. Diesel, Gasoline,
(1) Number of batteries available for second use per HEV, PHEV, BEV, etc.) were available monthly from
year from accident vehicles from present until 2030; 2010 up to 2016 [6]. Based on this information, a
(2) Number of batteries available for second use per relatively exact availability of 2nd use batteries of
year from end-of-life vehicles from present until 2030. different technologies could be stablished for the very
near future.
3. Approach
Additionally, different research [6-11] aiming at
Within this chapter it will be explained how the input prospective hybrid and electric vehicles in Germany
data for the model were obtained. Furthermore, the was consulted.
proposed mathematical model to calculate the number 3.1.1 HEV, PHEV and BEV
of available batteries will be presented in detail. The following sources were analyzed for the study
for HEV, PHEV and BEV:
3.1 New Registered Hybrid and Electric Cars in
(a) Shell (2014) [7], KBA: HEV, PHEV and BEV
Germany from 2010 until 2030
initial registrations for the years 2013, 2020, 2030 and
To forecast the behavior of stock in the future 2040 for Trend and Alternative scenarios, where:
(1) Trend: continue passenger car trends and
1
It was assumed that the vehicle’s life period is similar to the consider an increasing use of biofuels.
batteries’. Thus, not more than one battery per car life was
taken into account (2) Alternative: examine the potential impacts of
Availability of Lithium Ion Batteries from Hybrid and Electric Cars for Second Use: 131
increased electrification of propulsion systems and (HEV, PHEV, and BEV), which are given in a
fuels. percentage based on the initial registrations in
(b) eMAP project (electromobility-scenario based Germany. These 9 combinations were reduced by
market potential, assessment, and policy options, weighted summing the segments accordingly (small:
www.project-emap.eu): 28%, medium: 59%, and large: 13%). The solicited
(b1) Shell (2009) [8]: passenger vehicle registrations percentages were separately multiplied with both first
from 2000 until 2030 in % (forecast offered every five registration scenarios given by Shell /KBA [7], giving
years). The information considered for the study was 6 data series per car type.
from 2010 on. To convert this information into real 3.1.2 EV as a global term
numbers, the passenger car new registrations offered Regarding the amount of EV, Ref. [6] was used
by Shell and KBA [7] (for the years 2013, 2020, 2030 (total amount of HEV, PHEV, and BEV) for known
and 2040), considering the trend and the alternative data of the past. The analyzed sources were:
scenarios, were employed. From 2010 to 2016, the (c) German Government plan until 2030 [11]: the
information used was the one provided by the KBA, goal proposed by the German Government
and for the rest of the years the information provided contemplates to reach 1 million sold EV by 2020 and 6
by Shell was employed. Only for the year 2025, Shell million by 2030 (cumulative). The input data here were
did not provide any information. Thus, the value for obtained by extrapolating the provided accumulative
2025 was obtained via interpolation between 2020 and EV sales for 2016-2030, so that the sales for this period
2030. of time amount to 6 million in total.
(b2) Kugler [9]: German vehicle sales per year for (d) eMAP project: KE-CONSULT [10]: EV sales
the period between 2010 and 2030 for HEV, PHEV, per year from 2010 until 2030 for the scenarios BaU,
BEV differentiating between a small, medium, and TeD and PoD.
large segment depending on the vehicle size for three The above-analyzed data provided information for
different scenarios: initial vehicle registrations between 2010 and 2030 for
(1) BaU (Business as Usual): it models the effect of new cars coming to the market in Germany sorted by
current policies and EV technologies. HEV, PHEV, BEV, and EV (Figs. 2-5).
(2) TeD (Technology Driven scenario): higher Using this information, three scenarios (optimistic,
investments into traction battery research and medium, and pessimistic) were assumed for the input
development are assumed. data of the model. To generate the three curves, a cubic
(3) PoD (Policy Driven scenarios): it explores Bezier function was employed, as presented in Eq. (1):
EU-wide (PoD-EU) and country-specific pathways B(t) = P0·(1-t)3+3P1·t·(1-t)2 +
where additional policies are incorporated to enhance 3P2·t2·(1-t) +P3·t3, t є [0, 1] (1)
the reduction of CO2 emissions and to promote where B(t) was the number of available cars, P0, P1, P2
electrified vehicles. and P3 are the four points that define the cubic Bezier
To create more scenarios, and therefore acquire curve, and t shows how the curve increases, being t = 0
more input data, the information provided from Shell for all years before the rise and t = 1 for all years
and KBA [7] regarding the new car registration for the thereafter (steady state), increasing linearly in between:
Trend and Alternative scenarios was further employed. t(y) = 0, year < y_start;
Kugler [9] provides in total 9 combinations of 3 _
t(y) = , y_start < year < y_end;
scenarios (BaU, TeD, and PoD) and 3 car sizes (small, _ _
medium, and large segment) for each of the 3 car types t(y) = 1, year > y_end;
The curve starts at P0 going toward P1 and arrives at million/year) will be from HEV, PHEV, and EV type in
P3 coming from the direction of P2. As the known rise saturation, which will happen in 2035. This steady state
between 2010 and 2016 is relatively small, P0 = P1 can sum was distributed to the three car types according to
be assumed, being renamed as CS (cars at start of the the sources. The medium and pessimistic cases were
slope), which will be always close to 0. The ending chosen with reduced numbers fitted to the sources
point was also fixed and it was the information assuming a steady state for battery car first registrations
provided by the different sources for the year 2030, in 2040. Finally, the three created scenarios obtained
considering three scenarios, which were manually with Eq. (2) were used as input for the model. They are
fitted to the data points derived by the source. As a colored in green (optimistic case), yellow (medium
market saturation (a steady state of orders) was case) and red (pessimistic case).
assumed giving P2 = P3, being renamed as CE (cars at
3.2 Car Accident Rates
steady state at the end of the slope). The applied
changes can be seen on Eq. (2): To include damaged batteries from accidents, a car
accident rate was required to serve as model input.
B(t) = CS·(2·t3 -3·t2+1)+CE·(3·t2 -2·t3), t є [0,1] (2)
Therefore, the EUSka (Elektronische
Figs. 2-5 represent the forecast for the registration Unfalltypen-Steckkarte) database on accident
numbers of HEV, PHEV, BEV and EV. Every colored information was examined [12]. EUSka was developed
line represents a different combination of sources. To by German insurance companies (Unfallforschung der
estimate future developments, different sources and Versicherer) and distributed by PTV Group. It contains
trends for the registered cars in 2010-2016 (KBA) were information on accidents from years 2010 to 2016 with
considered. The optimistic case is based on the reference to the German federal states Saxony, Hessen,
assumption that all new registered cars (3.5 Hamburg, Bremen, Brandenburg, and Saxony-Anhalt.
Fig. 2 Newly registered HEV in Germany 2010-2030.

Fig. 3 Newly registered PHEV in Germany 2010-2030.
Fig. 4 Newly registered BEV in Germany 2010-2030.

Fig. 5 Newly registered EV in Germany 2010-2030.
In terms of geographical, infrastructural, population-, as damaged car (5.9% of all accidents). The remaining
and automotive-based circumstances this source can be crashes (94.1%) were assumed to have only little
seen as representative for Germany [13], although only impact on the battery—allowing its future use for
data of four federal states could be used for the second life applications. These percentages and
evaluation as no data of Hamburg and Bremen were absolute amounts of accidents showed no
available prior to 2014. time-dependent trend in any of the categories in the
It was then necessary to decide, which accidents validated time span. Thus, and because of the
were relevant for damaged batteries. Although the comparatively small effect of the rate on the final result,
EUSka database provides a wide variety of information a constant accident rate was assumed for the forecast.
on recorded accidents, it lacks data on the severity of Additionally, it was necessary to multiply the
deformation. It was thus necessary to analyze number of accidents by the average rate of cars
additional information sources for being able to split participating in an accident (Factor 1.2, based on
the cases according to accident severity. In addition, as EUSka), giving the final factor of 7.0% damaged cars
the actual usage of electrical propelled cars is very low, per accident.
it was necessary to extrapolate information based on all To extrapolate this result to the total amount of cars
car types. in Germany, a statistic of the Statistisches Bundesamt
The EUSka database provides information about the [14] was used, which states that in average 2,442,000
injury grade of the accident participants. This accidents per year happened between 2010 and 2016.
information was used to estimate the influence of an The KBA [15] further provided data, that the total
accident on the battery. Car accidents with passengers, amount of cars in Germany was 43,969,000 in average,
which suffered fatal (0.07% of all recorded accidents), resulting in a total of 0.056 accidents per registered car
severe (1.1%), or minor injuries (4.7%), were validated and year in Germany. As an estimation this rate was
evaluated to be similar for EVs. being the only variable of the equation.
Multiplying this accident rate of 0.056 accidents per To identify a scenario for the generic car lifetime in
registered car and year in Germany with the portion of Germany, information about the Weibull parameters
accidents of 7.0%, which lead to deregistration, for a car’s lifetime in Germany in 1995 [19] was
approximately 0.004 cars per registered car and year chosen as the starting point:
are deregistered after accidents. θ = -5 years; α = 5; β = 15.7 years;
This amount contains on one hand several batteries In Ref. [19], the authors found as a general rule that
that were not deformed and thus appropriate for a choosing θ in years similar to -α fits well to real data for
second use. On the other hand, there are batteries most of the countries in their research.
included that are damaged and therefore not To update that relatively old information for the
appropriate for a second use. In Ref. [16] a forecast, calculations were made for determining the
superposition (penetration depths and distribution of Weibull parameters for a car’s lifetime in Germany
typical battery installation positions) was realized to based on information from “Vehicle registrations
estimate the number of deformed batteries from (FZ15): Stock of motor vehicles and trailers by vehicle
accident cars. With an extrapolation, it was possible to age (KBA)” [20, pp. 6-7].
estimate the number of deformed batteries within the Stock data were given depending on the time of first
amount of accident EV. This portion of damaged registration of individual cars, allowing recalculating
batteries (0.8% of all accident batteries) was so small, this information based on the age of the cars. In year
that all batteries from accidents were seen as applicable 2008 the statistics changed in categorizing ambulances
for second use. and caravans, so data from earlier years had to be fitted
(by a factor reducing the step) to the number of cars in
3.3 Ageing Function for the Car Lifetime
2008. The stock of the first year was used as absolute
The third input parameter of the model was the number (100%) for calculating the percentage of still
ageing function of the car. To describe the probability registered cars depending on their age, because directly
of a car of still being registered along the years from its using the first registration numbers would have caused
first registration [18], the Weibull reliability function a 9% loss in the first year. One reason for this
was used as suggested by Ref. [17], represented in Eq. difference is short-term registrations of cars, which are
(3), and seen in Fig. 6. exported to other countries shortly after registration,
thus causing a registration but no stock at the first of
Weibull , , , 3 January, when the statistic is recorded. The solicited
where: results are marked in Fig. 6 using different colored dots
θ is the location, being a minimum value of t that the depending on the first registration years.
end of life can occur, which will be the age at which The Weibull parameters were fitted manually to this
scrappage starts; in unit time. graph to minimize the error relative to the average of
α is the shape, being the failure steepness for a the available values at each age, to meet the
generic car. It is always > 0 and determines the Weibull requirement to have θ (in years) similar to -α, and third
distribution slope (in that case > 1 as the failure rate that the integral of the reliability function should be
increases with age); no dimensional parameter. nearby 13.45, which was the ratio of the 2016 stock
β is the scale, a parameter indicating the (45,071,000 cars) by 2016 new registrations (3,350,000
characteristic service life for a generic car; in unit time. cars). In this way, the calculated number of
t is the time, the actual age of a car in units of time, existing cars in steady state will fit to the actual reality.
136 Availability of Lithium Ion Batteries
s from Hybrid
d and Electric
c Cars for Second Use:
H
How to Forec
cast for Germany until 203
30
10
00%
Percentage of still registered cars
8
80%
6
60%
4
40%
2
20%
0%
0 10 20 30 40
Age of Car after firstt registration // years
Weibull AVG 1980

0
1985 1990 1995
5
2000 2005 2010
0
2012 2014 2016
6
Fig. 6 Regisstration rate deepending on age of the cars (depending on

n year of first registration) aand fitted Weib
bull reliabilityy
function for the
t car’s probaable lifetime.
Although the t past treend showedd an increaasing To find the curvve for registeered cars in alla years, thee
reliability off cars, no basse to forecastt this trend too the reliability vectorr has to be connvoluted (* in
n Eqs. (4)-(6)))
future was found.
fo withh the registraation of new ccars to the market
m as seenn
The followwing results were
w obtainedd and used forr the in Eq.
E (4):
forecast: registered new registered
θ = -3.0 years;
y α = 3.5;; β = 17.7 yeaars; typ life 4
The accorrding functionn is shown as black line in Fig. The
T emission of aged battteries could be
b calculatedd
6. by a convolutiion with a (still unkno own) vectorr
describing the probability of deregistrration as off
3.4 Mathemmatical Modeel to Calculaate the 2nd Use
ageeing accordingg to Eq. (5):
Battery Emisssion to the Market
M
ageed nnew registered
The param meters typicaal life (ageinng) and accident tyyp ageing 5
rate will likeely fit to battery cars as well.
w Nevertheeless The
T amount of o defect battteries caused by accidentss
the result strongly
s depeends on the not well-knnown cou
uld be calculaated using thhis convolution shown inn
registration rate of cars in future. Therefore,
T it was Eq. (6):
decided noot to differeentiate betw ween car tyypes, acciident new registered
suppliers annd so on. Thuus, a generic car life couldd be typ accident 6
represented by a typicall curve of thhe probabilityy of The
T describeed model assumes the followingg
being still inn use/registereed (abovemenntioned reliabbility paraameters as knnown to start the forecast:
function). (1) Simulationn time step: tiime format off all input andd
In the proposed time diiscrete forecaast, all vectorss are outp
put data, time distance beetween time points
p m andd
indexed withh the parametter m, a representative of time. m+1;
(2) Accident rate accident while step of a car, in unit Weibull PDF results was used to achieve consistent
1/time step; time-discrete data:
(3) The emission of new cars to the market typ life 0 1 (10)
registered as a vector, in nr. of cars in the certain
time step; typ life 1 deregister while time step 1
(4) A statistic representing the typical life of a
generic car. Eq. (11) now allows to separate both reasons for
The time step was selected to be one year, which is deregistration (accident, age):
accurate enough, considering the weak assumptions for deregister while step ageing while step
initial car registrations. In the present model a car’s life accident while step ageing while step · accident while step
was calculated based on the Weibull distribution from (11)
Eqs. (7) [21] and (8) [22]: To calculate the probability of ageing in a certain
time step, Eq. (11) was transformed to Eq. (12):
Weibull , , , · · (7)
ageing while step
deregister while step accident while step

, , , 1 8 12
Weibull 1 accident while step
The PDF (probability density function) describes the Due to float rounding errors, the results should be
probability that a car will be deregistered at a specific limited to values below 1. Additionally, if the accident
date in its life. After calculating the time discrete PDF, rate was above the probability of deregistration based
the vector was normalized to 1, to avoid calculation on the Weibull statistics (negative result) in the first
errors as of the limit not to calculate in infinite future months, these values must be set to zero. Using the
(error < 0.5%). This vector was recalculated to the actual input values, the error in additionally generated
probability, that a still registered car will be batteries was below 0.5 % and was removed
deregistered in a certain time step: completely by linear reducing the output vectors
Weibull , , , typ ageing and typ accident .
deregister while step ∑
(9)
, , ,
Weibull
As there was a probability of aging and accident in
∑ Weibull , , , is a term for the the same time step, this amount was considered
probability of a car deregistration before the certain (subtracted) accordingly in the above-mentioned Eq.
time step m. Here, a time discrete representation (11). To generate the output for aged and accident
of the CDF (cumulative distribution function) of batteries separately, it was necessary to decide where to
the Weibull distribution based on the PDF was used put this amount. Even when years are used as time step,
to settle rounding differences. Due to calculation this error was very low (< 0.2% of absolute cars)
errors with float values, the results need to be limited compared to the error of the input estimations. Thus,
to 1. the authors chose to distribute this error equally
The Weibull CDF represents the probability for a employing Eqs. (13) and (14):
deregistered status of a generic car as of both reasons typ ageing typ life · ageing while step
(age, accident) in a certain time step. 1
To calculate , which represents the · 1 accident while step 13
typ life 2
probability for a car still being registered starting from typ accident typ life · accident while step 1
its registration, a time discrete version (Eq. (10)) of the 1
Weibull reliably function (Eq. (3)) deduced from the ageing while step 14
2
s from Hybrid
d and Electric
H
How to Forec
30
4. Results and discusssion comming from acccidents (betw ween 15,0000 and 70,0000
per year in 20330). Among end-of-life vehicles thee
The abovee described model
m and data allow to preedict
bigggest share off batteries will be obtained
d from hybridd
the number of o available batteries
b per year
y from preesent
elecctric vehicles (between 70,000 and 240,000 per yearr
until 2030 coming from accidents
a on one side, andd the
in 2030,
2 Fig. 8)), followed bby plug-in hyybrid electricc
number of available
a battteries per yeear coming from
f
vehhicles (betweeen 30,000 and 130,00 00 per year,,
end-of-life electric
e vehiccles (HEV, PHEV,
P BEV, and
Fig. 10). The sm mallest amounnt will arise from
f accidentt
EV) on the other
o side. Figgs. 7-14 summ marize the ressults
batttery electric vehicles
v (betw
ween 2,000 annd 10,000 perr
for all the studied technoologies. As not n all referennces
yeaar, Fig. 11), off which the aamount of battteries, whichh
specified thee different vehhicles technologies uniform mly,
are not available for 2nd use after accid dents, will bee
HEV, PHEV, and BEV V were treaated as sepaarate
sources fromm EV for the model.
m relaatively low.
The resuults show thhat the numbber of availlable The
T quality of this predicttion about thee availabilityy
batteries is rising
r with thhe biggest slope between 2020
2 of batteries
b for second life strongly dep
pends on thee
and 2025. This finding reesults from ann assumed steeady quaality of the inpput data. Anallysis of differrent literaturee
state of initiial registrationns in 2035 annd from a typpical sou
urces during thhis study to oobtain input data
d showed a
lifetime for cars of 15 yeears (Fig. 6). In I 2030, the total
t com
mplex situatioon concerninng the stated technologiess
amount of available seecond use batteries b willl be (HE
EV, PHEV, and
a BEV verrsus EV). In addition, thee
between 135,000 units/yyear and 5000,000 units/yyear. quaality of the listed figures like future new
n EV saless
The amounnt of batteries coming from end-off-life stro
ongly dependds on severall influencing
g factors likee
vehicles (beetween 120,0000 and 430,,000 per yeaar in poliicy directivess from the EU
U or German Governmentt
2030) will be much biggeer than the nuumber of batteeries inceentives, as well as on electriic vehicle pricces. Updating
Fig. 7 Available batteries from accidentss for HEV 2010-2030.

Availability of Lithium Ion Batteries
s from Hybrid
d and Electric
c Cars for Second Use: 1399
H
How to Forec
30
Fig. 8 Available batteries from end-of-life for HEV 2010-2030.
Fig. 9 Available batteries from accidentss for PHEV 20010-2030.

s from Hybrid
d and Electric
H
How to Forec
30
Fig. 10 Avaiilable batteriess from end-of-llife for PHEV 2010-2030.
Fig. 11 Avaiilable batteriess from acciden

nts for BEV 20110-2030.
Fig. 12 Available batteries from end-of-life for BEV 2010-2030.
Fig. 13 Available batteries from accidents for EV 2010-2030.

Fig. 14 Available batteries from end-of-life for EV 2010-2030.
the model with newly available data will allow a more status correctly, organize safe transportation, and
reliable picture about the number of batteries from ensure assembly of similar cells in a most economical
accidents and end-of-life vehicles after 2030. way to succeed in this new market—a challenge that
While German automotive companies like Daimler will surely need a significant amount of batteries
or BMW already explore the 2nd use lithium ion available to be profitable, but that will gain importance
battery market [23], form consortiums and establish in the future.
cooperations aiming at innovating projects to give a
Acknowledgment
second life to batteries coming from electric vehicles,
several challenges to this very specific market should The authors would like to thank Peter Meißner, Max
not be disregarded [24]. Auernhammer and Reinhard Büchl from ELOGPlan
In today’s vehicles, a plethora of different battery GmbH (Ingolstadt, Germany) and Dr. Christian
types concerning chemistry or formats is applied [25]. Erbsmehl from Fraunhofer Institut für Verkehrs-und
It can be expected that this complexity will increase as Infrastruktursysteme IVI (Dresden, Germany) for
new technologies enter the market and the obtained discussion. This work was financed by the
values of available batteries for second use will be split Bundesministerium für Bildung und Forschung
into small portions of similar battery types. In the same (BMBF) grant FH-Impuls SAFIR-SEANCe
time, the efficiency of stationary storage systems (Förderkennzeichen: 13FH7I04IA).
strongly depends on a most homogenous distribution of
References
battery types. Also, different aging behavior of cells in
[1] “International Energy Agency (IEA). 2017. Global EV
2nd use systems is a challenge when combining these
Outlook 2017: Two million and counting”.
sub-systems into bigger units. Potential providers will [2] Petroff, A. 2017. “These Countries Want to Ban Gas and
have to collect the cells from the user, identify the cells’ Diesel Cars.” CNN Money. Accessed November 30,
2017. https://www.kba.de/DE/Statistik/Fahrzeuge/Bestand/Fahr
http://money.cnn.com/2017/09/11/autos/countries-bannin zeugklassenAufbauarten/b_fzkl_zeitreihe.html?nn=652402.
g-diesel-gas-cars/index.html. [16] Johanna, V. 2017. Verfügbarkeit von Traktionsbatterien
[3] Reiner, R., Cartalos, O., Evrigenis, A., and Viljamaa, K. aus verunfallten Elektro-und Hybridfahrzeugen:
2010. “Challenges for a European Market for Electric Prädiktion vor dem Hintergrund des aktuellen
Vehicles.” European Parliament. Unfallgeschehens. Bachelorarbeit.
[4] Lih, W., Yen, J., Shieh, F., and Liao, Y. 2012. “Second [17] Kolli, Z., Dupont-Kieffer, A., and Hivert, L. 2010. “Car
Use of Retired Lithium-ion Battery Packs from Electric Survival in a National Car Fleet: Non-parametric and
Vehicles: Technological Challenges, Cost Analysis and Parametric Approaches Applied to French Data.’
Optimal Business Model.” In Proceedings of Conference Presented at the 12th WCTR, Lisbon, Portugal, July
International Symposium on Computer, Consumer and 11-15, 2010.
Control, pp. 381-4. [18] Wikipedia. 2018. “The Weibull Distribution.” Accessed
[5] Reid, G., and Julve, J. 2016. “Second Life-Batteries as February 20, 2018.
Flexible Storage for Renewables Energies.” http://reliawiki.org/index.php/The_Weibull_Distribution.
Bundesverband Erneuerbare Energie e.V. (BEE). [19] Zachariadis, T., Samaras, Z., and Zierock, K.-H. 1995.
[6] Kraftfahrt Bundesamt (KBA) Fahrzeugzulassungen (FZ): “Dynamic Modeling of Vehicle Populations: An
Neuzulassungen von Kraftfahrzeugen und Engineering Approach for Emissions Calculations.”
Kraftfahrzeuganhängern-Monatsergebnisse (2010-2016). Technological Forecasting and Social Change 50:
[7] Adolf, J., Balzer, C., Joedicke, A., Schabla, U., Wilbrand, 135-49.
K., Rommerskirchen, S., et al. 2014. “Shell Passenger Car [20] Kraftfahrt-Bundesamt (KBA): Fahrzeugzulassungen
Scenarios for Germany to 2040.” Shell Deutschland Oil (FZ15) Bestand an Kraftfahrzeugen und
GmbH and Prognos AG, Hamburg. Kraftfahrzeuganhängern nach Fahrzeugalter (1980 bis
[8] VTT Technical Research Centre of Finland. 2013. 2017), 2017.
“eMAP Project: D2.1-Analytic Framework: Parameters, [21] 2018. “Weibull Formulas and Plots.” In Engineering
Data & Methodology.” Statistics Handbook. Accessed February 20, 2018.
[9] Kugler, U. 2015. “eMAP Project: D6.2-Scenario Report, http://www.itl.nist.gov/div898/handbook/apr/section1/apr
with an in-Depth Description of the Scenarios' 162.htm.
Background.” [22] MathWave. 2018. “Cumulative Distribution Function.”
[10] Ke-Consult. 2015. “eMAP Project: D7.1-Evaluation of MathWave. Accessed February 20, 2018.
Scenarios for Electromobility with an Integrated http://www.mathwave.com/help/easyfit/html/analyses/dis
Quantitative Assessment and Multi Criteria Analysis.” tributions/weibull.html.
[11] Catenacci, M., Verdolini, E., Bosetti, V., and Fiorese, G. [23] Daimler/Mercedes-Benz:
2013. “Going Electric: Expert Survey on the Future of https://www.mercedes-benz.com/en/mercedes-benz/next/
Battery Technologies for Electric Vehicles.” Energy e-mobility/the-second-lives-of-lithium-ion-batteries;
Policy 61: 403-13. BMW AG:
[12] EUSka Data Analyzes of Police Accident Data by https://greentransportation.info/news/2017/10/bmw-2nd-l
Fraunhofer Institute for Transportation and Infrastructure ife-battery-farm.html.
Systems IVI, Dresden. [24] Recharge—The European Association for Advanced
[13] Internal Research Study by Fraunhofer Institute for Rechargeable Batteries. “Re-use and Second Use of
Transportation and Infrastructure Systems IVI, Dresden. Rechargeable Batteries.” October 2014.
[14] Statistisches Bundesamt, Fachserie 8, Reihe 7, Verkehr, [25] Fraunhofer-Institut für System-und Innovationsforschung
Verkehrsunfälle 2016, Seite 54. ISI. 2017. “Energiespeicher-Roadmap (Update
[15] Kraftfahrt-Bundesamt. “Bestand in den Jahren 1960 bis 2017)–Hochenergie-Batterien 2030+ und Perspektiven
2017 nach Fahrzeugklassen.” Online zukünftiger Batterietechnologien.”
doi: 10.17265/2328-2223/2018.03.002
D DAVID PUBLISHING
The Study of Demagnetization of the Magnetic

Orientation of Permanent Magnets for IPMSM with
Field-Weakening Control under Hot Temperature
Noriyoshi Nishiyama1, Hiroki Uemura2 and Yukio Honda2

1. Panasonic Corporation, Moriguchi City, Osaka 570-8501, Japan
2. Osaka Institute of Technology, Osaka City, Osaka 530-0013, Japan
Abstract: In this study, we investigate the demagnetization resistance of a concentrated winding IPMSM (interior permanent magnet
synchronous motor) accounting for field weakening control by changing the magnetization direction of the permanent magnet under a
high-temperature environment. IPMSMs are investigated by FEA (finite element analysis) using the same volume of the permanent
magnet while changing the magnet’s width, thickness and magnetic field orientation angle. FEA found that a V-shaped angle Va = 100°
and a changed magnet length of 97% using an oblique magnetic-field-oriented magnet strike a good balance between demagnetization
resistance and torque at 180 °C. Comparison between demagnetization of negative d-axis current (current phase β = 90°) and
demagnetization of field weakening control (β = 80°) using concentrated winding IPMSM with V-shaped angle Va = 100° is conducted.
With the demagnetization factor at β = 80° for β = 90°, the demagnetization factor 0.39 (2.6 times) at α = 0° decreases to 0.23 (4.3 times)
at α = 20°. The demagnetization resistance in the field weakening control is further improved.
Key words: IPMSM, demagnetization, concentrated winding, high temperature, field weakening.
1. Introduction For obliquely distributed magnets with an oblique

orientation angle α, the magnetic field orientation
The high efficiency of a small-sized permanent
inclines from the plate thickness direction with respect
magnet synchronous motor is one of its key features.
to the magnet width Wm and the magnet thickness tm.
On the other hand, one of its disadvantages is that the
Furthermore, the operating point of the magnet does
coercive force decreases at high temperature and then
not change much from the permeance equation, since
easily demagnetizes. In recent years, motors for
the magnetic field H is acting on the magnet. On the
vehicles, high-temperature heat pumps, and
other hand, the reverse magnetic field component
extreme-environment-compatible robots have had to
affecting the magnetization direction of the magnet
operate in severe load environments, which are often
can be reduced to Hm, thus improving the
extremely high-temperature environments where
demagnetization resistance [1, 2].
permanent magnets are demagnetized. In this work, we
Few studies have attempted to examine the
examined a parallelogram-shaped permanent magnet
magnetic field orientation of permanent magnets
whose magnetic field orientation was made oblique
detailed demagnetization characteristics under high
with respect to the plate thickness direction (see Fig.
temperature environments [3-8]. A field weakening
1). Lm is the magnet length, Am is the magnet
control that energizes the negative d-axis current
cross-sectional area.
component can suppress the induced voltage of the
motor and can operate in the wide rotational speed
Corresponding author: Noriyoshi Nishiyama, doctor of
philosophy in Engineering, research fields: permanent magnet range under a limited power supply voltage [9]. The
synchronous motor design. demagnetization in field-weakening control is likely to
The Study of Demagnettization of the
e Magnetic Orientation
O of Permanent M
Magnets for 1455
IPMSM with
h Field-Weake ening Contro
ol under Hot Temperature
T
arraangement anggle Va while also changing the magnett

lenggth Lm and magnet
m width Wm under a fixed magnett
voluume at 1800 °C. A cuurrent causin ng a reversee
mag gnetic field to
t act on the rotor’s magn net is passedd
throough the windding and the aair gap, and magnetic
m fluxx
dennsity before annd after energgy conductio
on is obtainedd
by electromagnetic fielld analysiss softwaree
Fig. 1 que magnetic orientated
Obliq o maggnet. (JM
MAG-Designeer). A neoddymium sinteered magnett
(rev
versible data of
o NMX-S 366 UH) is used d. [10]
occur at a current
c phasee β = 80° whhere the negaative
T analysis conditions arre shown in Table 1, andd
The
d-axis currennt component is very largee.
the analysis moddel is shown in Fig. 2. Th he evaluationn
In this papper, we invesstigate the opttimum designn for
indeexes are demmagnetization limit current and torque.
improving the demagnnetization resistancer of a
T demagnettization ratio Dr n is defined as Eq. (2)::
The
concentratedd winding IPMSMI (intterior permaanent
Thee demagnetizzation improovement ratiio Dir n iss
magnet synnchronous motor)m accouunting for field
f
defi
fined as Eq. (33):
weakening control by changing thhe magnetizaation
D n = 100 × (1 - B2/B1)
Dr (2))
direction of o the perrmanent maagnet underr a
Dir n = DDr n / Dr 0 (3))
high-temperrature environnment.
The
T torque rattio Tr n is deffined as Eq. (4):
(
2. Demagn
netization Analysis
A Tr n = T n / T0 (4))
wheere B1 is thhe air gap’ss magnetic flux densityy
A concenntrated windinng motor is advantageous
a s for
befo
fore reversed magnetic fielld is applied. B2 is the airr
use in highh-temperaturee environmennts. In addittion,
gapp’s magnetic flux densityy after reverssed magneticc
IPMSM, in which
w a magnnet is embeddded in a rotor core,
field is applied. Dr
D 0 is the deemagnetizatioon ratio at thee
is generally arranged inn a rectangular parallelepiped
permanent magnet,
m and by
b deeply embbedding a maagnet Tab
ble 1 Analysiss conditions.
in the rotorr core, the magnetic
m fielld acting onn the Staator ID 56 mm Ressidual magneticc
1.16 T
magnet caan be relaxxed, which improves the Rottor OD 54..4 mm fluxx density Br
Staack length 32 mm Coeercivity Hcj 2,387 kA/m
demagnetizaation resistancce.
Maagnet 11335 mm3 Eleectromagnetic
Permeancce coefficiennt Pc, whichh determines the 35A300
vollume (at one pole) steeel sheet
operating pooint of the peermanent maggnet, is expressed Win nding 1500 turn, 3Y Temmperature 180 °C
as Eq. (1):
LmAg σ
Pcc = (1)
AmLg ｆ
where Ag is the air acrosss-sectional area,
a Lg is thee air
gap length, σ is the leeak coefficient, and f is the
magneto mootive force losss factor.
The improovement of thhe demagnetiization resistaance
is greatly aff
ffected by thee angle differeence betweenn the
direction of the working magnetic
m fielld and the maagnet
orientation direction. The
T aim of demagnetizaation
analysis is to
t study the optimum dessign by changging
the magneticc field orienttation angle α and the maagnet Fig.. 2 V-shaped magnet arrangement motorr model.
146 The Study of Demagnettization of the
O of Permanent M
Magnets for
IPMSM with
T
parallel-orieented magneet (α = 0°) placed in flat incrreases in the V-shaaped arrang gement, thee
arrangementt (Va = 1800°). T0 is thhe torque at the dem
magnetizationn improvemeent ratio alsso decreasess
parallel-orieented magneet (α = 0°) placed in flat as the
t magnet length ratio decreases. Further, if thee
arrangementt (Va = 180°).
The demaagnetization limit
l current is the maxim mum
current at which
w reduction of the noon-conductingg air
gap’s magnnetic flux density, beefore and after a
application ofo the reversee magnetic fiield current, does
d
not yet reachh 1%.
The torquue is calculaated with ratted current (7.07
Arms) and current
c phase β = 20°.
3. Magneet Length and Dem

magnetizattion
Resistancee
Here, we consider dem magnetizationn analysis byy an
analytical model
m in whhich the maggnet arrangem ment Fig.. 3 Magnet arrangement
a an
ngle Va vs. deemagnetization
n
ratio and torque ratio.
r
angle Va is changed
c from
m 180° to 60°.
Based on a 1% demaggnetization lim mit current off the
flat plate moodel (Va = 1880°), this vallue is increased 3
times for thhe Va = 100° model andd 8 times forr the
spoke modeel (Va = 60°)). As the maggnet arrangem ment
angle Va iss changed, demagnetizati
d ion improvem ment
ratio Dr n and torque ratio r Tr n arre in a tradee-off
relationship (see Fig. 3).
In the anaalysis results, the horizontal axis showss the
magnet lenggth ratio andd the verticall axis showss the
demagnetizaation resistannce improvem ment ratio orr the
torque ratio (see Figs. 4 and
a 5). Fig.. 4 Magn net length rratio vs. deemagnetization
n
For the demagnetizattion resistannce improvem ment imp
provement ratio.
ratio, the demagnetizatioon limit currennt of a rotor with
w
the flat-pllate magneet arrangem ment using a
parallel-orieented magnet is 100% (ciircle within solids
line; 1). TheT torque ratio is thhe ratio of the
parallel-orieented magneets to the flat f arrangem ment
rotor’s torquue (circle within solid linne; 1). Underr the
condition off a constant magnet voluume, at a maagnet
length of lessl than 1000%, the magnet
m widthh is
increased. OnO the other hand, at a magnet
m length of
100% or moore, the magnnet width is constant
c using an
oblique maggnetic-field-orriented magnnet.
Although the demagnnetization im mprovement ratior Fig.. 5 Magnet leength ratio vs. torque ratio
O of Permanent M
Magnets for 1477
IPMSM with
T
demagnetizaation improvement ratio is improved,, the Tab

ble 2 Magnet parameters.
torque ratioo decreases. Aiming for demagnetizaation Magnet Magnet magnetic magnet
No. length width orientation length
improvemennt ratio of 2000% or more,, the torque ratio r Lm [mm] Wm [mm] angle α [deg.] ratio
becomes 90% or less (circle in dottedd line; 2.3). They T 1 1.65 10.75 0 100%
are in a tradee-off relationnship. 2 1.60 11.1 0 97%
3 1.50 11.8 0 91%
Next, we investigate the possibiliity of improvving
4 1.50 11.8 20 97%
the demagnnetization ressistance by obliqueo magnnetic 5 1.40 12.7 20 90%
field oriented magnets whhen the magnnet length ratiio of
Va = 100° model
m is less than
t 100%.
In the obllique magnetiic-field-oriennted magnet, evene
if the magneet thickness tm t is small, thhe magnet lenngth
Lm can bee increased, and thus improvementt in
demagnetizaation resistance can be exppected. Usingg the
Va = 100° model
m offers a lot of latituude in the maagnet
arrangementt, so we inveestigated the demagnetizaation
resistance annd torque usiing the magneet length Lm as a
parameter.
A list of the studied magnets
m is given
g in Tablle 2,
based on a motor
m model in which maagnet 1 is useed as Fig.. 6 Magn net length rratio vs. deemagnetization
n
imp
provement ratio and torque rratio.
a reference. Demagnetizaation is evaluuated as the ratio r
of the demaagnetization limit currentt ratio to the 1% mag gnet 5 with thhe magnetic ffield orientatiion angle α =
demagnetizaation limit cuurrent. Torquue is evaluateed as 20° is used, the demagnetizaation limit cu urrent ratio iss
the ratio of the magnet’s torque ratiio to the average impproved over that
t of the mmotor using magnets
m withh
value of the air gap’s maagnetic flux density at the timet the magnetic field
f orientattion angle α = 0°. Thee
of non-condduction. The torque of thhe IPMSM iss the chaange in the torque
t ratio increases ass the magnett
sum of the magnet torquue and the reeluctance torrque, widdth Wm increases, but it iss smaller thaan the changee
since the maaximum torquue per currennt is about 200° in in th
he demagnetiization limit ccurrent ratio.
the current phase and thhis is a motoor model maainly
4. Demaagnetization
n Rattio att
based on magnet
m torquue. In orderr to clarify the
Fieeld-Weaken
ning Control
influence off using differeent magnets, here we com mpare
the torque with
w the magnnet torque ratiio. The
T demagneetization ratioo of the field d weakeningg
In the annalysis resultss (see Fig. 6), 6 the horizoontal conntrol (β = 80°)8 decreasses with resspect to thee
axis is the magnet
m lengtth ratio, and the vertical axis neggative d-axis of the currennt (β = 90°), and
a this ratioo
shows the demagnetizati
d ion limit currrent ratio andd the is defined
d as thee current phaase demagnetization ratioo
torque ratio.. The demagnnetization lim mit current rattio is Cr n.n
greatly reduuced for a motor using magnet 2 or maagnet Cr 80 = Dr 80 / Dr 90 (5))
3 with a magnetic
m fielld orientationn angle α = 0° wheere, Dr 80 is the demagneetization ratio o at β = 80°,,
obtained byy increasingg the magnnetic width Wm Dr 90 is the dem magnetizationn ratio at β = 90°.
9
through redducing the magnet thicckness tm (i.e., ( The
T smaller thhe current phhase demagneetization ratioo
reduction off the magnet length
l Lm). Cr, the more margin
m is giveen to the dem
magnetizationn
On the othher hand, in a motor in which
w magnet 4 or resiistance in the field-weakenning control.
148 The Study of Demagnettization of the
O of Permanent M
Magnets for
IPMSM with
T
Fig. 7 shows the relationshipp between the impprovement rattio Dir = 90%% (magnet 5)), Cr is aboutt
demagnetizaation ratio Drr with respectt to the currennt at the same value as
a magnet 1.
the magneticc orientation angle α = 0° and α = 200° in At
A the demagnnetization immprovement raatio Dir = 900%
the Va = 130°
1 model. At the maggnetic orientaation or more,
m the currrent phase deemagnetizatio
on ratio Cr iss
angle α = 0°, when the t current is 42.5 A, the also
o improved.
9 is 0.91%, and
demagnetizaation ratio at β = 90° (Dr 90)
6. Measureme
M ent of Maggnetic Flux Density
the demagneetization ratioo at β = 80° (Dr
( 80) is 0.779%.
When the current phhase differss by 10°, the We
W evaluate the t magnetic flux density of prototypee
demagnetizaation ratio is reduced
r to 0.87 times. mag gnets with oblique orieentation. Th he prototypee
At the maagnetic orienttation angle α = 20°, whenn the mag gnets were prroduced by obbliquely slicing samariumm
current is 70 A, the dem magnetizationn ratio at β = 90° cobbalt sintered magnet withh respect to thet magneticc
(Dr 90) is 0.91%, and thhe demagnetizzation ratio att β = field orientationn direction. Since theree is a largee
80° (Dr 80)) is 0.67%. IfI the current phase differrs by marrket demand for neodymiuum sintered magnets,
m it iss
10°, the deemagnetizatioon factor is reduced to 0.67 diffficult to usee them for sspecial proto otypes underr
times. devvelopment; thherefore, wee used samaarium cobaltt
At α = 200° with respecct to α = 0°, in
i addition too the sinttered magneets to evaluuate the difference
d inn
large dem magnetizing resistance current, the orieentation anglees.
demagnetizaation ratio when
w β is shiffted from 900° to
80° is smalll, the demaggnetization reesistance ratiio is
high.
5. Magneet Length and Dem magnetizattion

Ratio at Field
F Weakeening Conttrol
Fig. 8 shhows the cuurrent phase demagnetizaation
ratio Cr withh respect to thhe magnet length ratio, which
w
is 1, 2, 3, 4, and 5 in Tabble 2. With thhe magnetic field
f
orientation angle
a α = 0°°, the motor using magnnet 1
(magnet lenngth 100%) inndicates Cr = 0.39. Withh the
magnet lenggth ratio of 977%, the motoor using magnnet 2 Fig.. 7 Current vs.
v demagnetizaation ratio.
indicates Crr = 0.34. On O the other hand, with the
magnetic fieeld orientatioon angle α = 20°, the motorm
using magneet 4 indicates a smaller Crr = 0.23. Withh the
magnet lenggth ratio of 90%9 and, thhe magnetic field
f
orientation angle
a α = 0°, the motor ussing the magnnet 3
indicates Cr C = 0.54; whereas with w the maagnet
orientation angle
a α = 200°, the motorr using magnnet 5
indicates a smaller
s Cr = 0.44.
0
At the maagnet length raatio of 97% (m magnet 2, maagnet
4), the currrent phase demagnetizat
d tion ratio Cr
C is
improved. At A the magnett length ratio of 90%, Cr is i as Fig.. 8 Magn net length rratio vs. cu
urrent phasee
low as 74% % (magnet 3), and at demagnetizaation dem
magnetization ratio.
r
O of Permanent M
Magnets for 1499
IPMSM with
T
Two typees of magnetss have the folllowing magnnetic

field orientaation angles: α = 0°, 20°°. The prototype
magnet size is magnet thhickness tm = 1.8 mm, maagnet
width Wm = 10.8 mm, and a axial lenngth L = 31 mm.m
The magneetic flux deensity measuuring devicee is
composed of o a magnet analyzer
a (MAAD-300R, DM MT),
a tesla meterr (TM-4700, DMT), and an a ultrafine prrobe
(w 0.75 mm m - t 0.28 mm F-075, DMT) witth a
resolution off 21,600 (see Fig. 9).
Next, the prototype magnets
m were assembled innto a
Fig.. 10 Air gap magnet
m flux deensity (measurred).
rotor with magnet
m arranggement anglee Va = 130°, and
the magneticc flux densityy was measurred. Furtherm more,
the rotor waas assembled in a cylindriccal dummy sttator
made of a magnetic body b having the same innner
diameter as the stator core,c and thhe magnetic flux
density of thhe air gap wass measured (ssee Fig. 10).
We discusss the results of the air gapp’s magnetic flux
density meaasured for the t rotor asssembled in the
dummy statoor. As a resuult of measurring the air gap’s
g
magnetic fluux density, itt is found thhat the maxim mum
magnetic field orientation angle deecreased slighhtly,
but the maggnetic flux deensity distribuution approacched Fig.. 11 Air gap magnet
m flux deensity (analysiss).
a sinusoidaal wave form m and is thhus effective for 90° on the horizzontal axis is distorted by the open slott
reducing thee torque ripplee. of the
t stator. Thhe air gap magnetic flux density
d of α =
Fig. 11 shhows the air gap magnetiic flux densitty of 20° is as small asa 7% as com mpared with α = 0°, closee
the motor by b FEA. Thhe magnetic field orientaation to a sinusoidal wave
w from 666° to 72° on thhe horizontall
angle α = 0°0 indicated by b the solid line
l and a = 20° axiss without the effect of opeen slot. It is similar
s to thee
indicated byy the broken line
l are non-eenergized airr gap actu
ual measurem ment at the air gap magnetic
m fluxx
magnetic fluux densities. dennsity of the cyylinder dummmy stator.
The magnnetic flux dennsity of 47° tot 53° and 877° to
7. Conclusion
C ns
Inn this paperr, we examinned the optiimum designn
accounting forr field-weakkening conttrol obliquee
mag gnetic field orientation
o off IPMSM driven under a
highh-temperature environmeent, changing g the magnett
thicckness tm andd magnet widdth Wm whilee maintainingg
a constant
c maggnet volume. We obtain ned that thee
demmagnetizationn factor Cr at β = 80° for β = 90°, Cr =
0.399 (2.6 times) at α = 0° deecreases to Cr
C = 0.23 (4.33
times) at α = 20°° by FEA (finnite element analysis)
a andd
Fig. 9 Air gaap magnet fluxx density meassuring device. inveestigation off our prototyype. The demmagnetizationn
150 The Study of Demagnetization of the Magnetic Orientation of Permanent Magnets for
IPMSM with Field-Weakening Control under Hot Temperature
resistance of the oblique magnetic orientation magnets Magnet in Arbitrary Directions.” In Proceedings of 2016
Ann. Meet. Rec. IEEJ, II, pp. 124-5. (in Japanese)
IPMSM in the field-weakening control is further
[6] Akune, R., Akatsu, K., Kume, K., Yamamoto, T., and
improved. Saito, S. 2016. “The Anti Demagnetization Method for
Permanent Magnet Synchronous Motor Focused on
References Magnetized Direction of Permanent Magnet and Basic
[1] Nishiyama, N., and Honda, Y. 2016. “Development of Experiment.” In 2016 IEE-Japan Industry Applications
IPMSM for High Temperature Conditions Using Inclined Society Conference (JIASC2016), 3-54. (in Japanese)
Magnetic Field Orientation and V-Shape Magnet [7] Peng, P., Xiong, H., Zhang, J., Li, W., Leonardi, F., Rong,
Arrangement.” In Proceedings of IEEJ Joint Technical C., Degner, M. W., Liang, F., and Zhu, L. 2016. “Effects
Meeting on “Magnetics” and “Linear Drives” of External Field Orientation on Permanent Magnet
MAG-16-2010, LD-16-145, pp. 23-8. (in Japanese) Demagnetization.” In Proceedings of IEEE Energy
[2] Uemura, H., Nishiyama, N., and Honda, Y. 2017. “The Conversion Congress and Exposition 2016, 7855067.
Effeteness of Magnetic field directions of V shaped [8] Galea, M., Papini, L., Zhang, H., Gerada, C., and Hamiti,
embedded magnets for IPMSM.” In Proceedings of 2017 T. 2015. “Demagnetization Analysis for Halbach Array
Ann. Meet. Rec. IEEJ, V, p. 15. (in Japanese) Configurations in Electrical Machines.” IEEE Trans. on
[3] Nishiyama, N., Uemura, H., and Honda, Y. 2017. “The Magn. 51 (9): 8107309.
Study of Highly Demagnetization Performance IPMSM [9] Kawano, S., Murakami, H., Nishiyama, N., Ikkai, Y.,
under Hot Environments.” In International Conference on Honda, Y., and Higaki, T. 1997. “High Performance
Electrical Machines and Systems 2017 Proceedings, Design of an Interior Permanent Magnet Synchronous
8056505. Reluctance Motor for Electric Vehicles.” In Proceedings
[4] Asano, Y., Honda, Y., Takeda, Y., and Morimoto, S. 2001. of Power Conversion Conference, Nagaoka, Vol. 1, pp.
“Reduction of Vibration on Concentrated Winding 33-6.
Permanent Magnet Synchronous Motors with Considering [10] Nishiyama, N., Uemura, H., and Honda, Y. 2017. “The
Radial Stress.” IEEJ Trans. on Ind. Appli. l21 (11): Optimum Design of the Magnetic Orientation of
1185-91. (in Japanese) Permanent Magnets for IPMSM under Hot Environments.”
[5] Maeda, Y., Urata, S., and Nakai, H. 2016. “The Evaluation In Proceedings of IEEE International Conference on
of Demagnetizing Characteristics of the Permanent Power Electronics and Drive Systems, pp. 380-5.
doi: 10.17265/2328-2223/2018.03.003
D DAVID PUBLISHING
A Mixed-Mode Biquad Employing OTAs and Grounded

Capacitors
Takao Tsukutani1, Yasutomo Kinugasa1 and Noburu Yabuki 2

1. National Institute of Technology, Matsue College, Matsue 690-8518, Japan
2. National Institute of Technology, Tsuyama College, Tsuyama 708-8509, Japan
Abstract: This paper introduces a mixed-mode biquad employing OTAs (operational trans-conductance amplifiers) and grounded
capacitors. The circuit can perform mixed-mode operation by selecting the input and output terminals. Additionally, the circuit
enables low-pass, band-pass, high-pass, band-stop and all-pass transfer functions suitably choosing the input terminals. The circuit
parameters ω0 and Q can be tuned orthogonally through adjusting the trans-conductance gains of the OTAs. The biquad enjoys very
low sensitivities with respect to the circuit active and passive components. The achievement examples are given together with
simulation results by PSPICE.
Key words: Active circuit, mixed-mode operation, biquad characteristics, OTAs.
1. Introduction above-mentioned performances has not yet been

studied sufficiently.
High performance active circuits have received
This paper introduces a mixed-mode biquad
much attention. It is well known that the OTA
employing five OTAs and two grounded capacitors.
(operational trans-conductance amplifier) provides
The biquadratic circuit can perform the current-mode,
highly linear electronic tunability and wide tunable
voltage-mode, trans-admittance-mode and
range of its trans-conductance gain. Additionally,
trans-impedance-mode operations by selecting the
OTA-based circuit requires no external resistors,
input and output terminals. Additionally, the circuit
hence it is very suitable for monolithic integration.
enables LP (low-pass), BP (band-pass), HP
The current-mode and voltage-mode biquads using the
(high-pass), BS (band-stop) and AP (all-pass)
OTAs have been discussed previously [1-6].
characteristics suitably choosing the input terminals.
It is much desirable for the biquad circuit design
The circuit parameters ω0 and Q can be tuned
that various circuit characteristics can be realized with
orthogonally through adjusting the trans-conductance
no component matching conditions. Additionally, it is
gains of the OTAs. It is made clear from sensitivity
required to set the circuit parameters ω0, Q and H
analysis that the circuit enjoys very low sensitivities to
orthogonally or independently. In applications to
the circuit active and passive components. The
analogue signal processing, it may be desirable to
achievement examples are given together with
synthesize mixed-mode biquad with input current or
simulation results by PSPICE.
voltage and output current or voltage. The
mixed-mode biquad using six OTAs and two 2. OTA
grounded capacitors has already been reported in the
Fig. 1 shows the symbol for the OTA. This shows
past [1]. However, the mixed-mode biquad with
dual current output OTA.
The current output Io is given by:
Corresponding author: Takao Tsukutani, Dr., research
field: analog signal processing. I o = ± g m ( V+ − V− ) (1)
input terminals suitably.

In current-mode operation, the way to realize the
LP, BP, HP, BS and AP transfer functions is as
follows:
Current-mode operation (Vin1 = Vin2 = Vin3 = Vin4
=0)
Fig. 1 Symbol for OTA. LP: Iin1 = Iin, Iin2 = Iin3 = 0
I out (s) g m1g m 2 / C1C 2
where gm denotes the trans-conductance gain. TLP (s) = = (8)
I in (s) D(s)
In Eq. (1), the sign “±” shows the polarity of the
BP: Iin2 = Iin, Iin1 = Iin3 = 0
current output.
I out (s) (g / C )s
The OTA [2] with MOS transistors is shown in TBP (s) = = − m2 2 (9)
I in (s) D(s)
Fig. 2. The trans-conductance gain gm can be
HP: Iin3 = Iin, Iin1 = Iin2 = 0
characterized by:
I out (s) s2
W THP (s) = = (10)
g m = μ n Cox I b (2) I in (s) D(s)
L
BS: Iin1 = Iin3 = Iin, Iin2 = 0
where μn, Cox, W/L and Ib are the electron mobility of
I out (s) s 2 + g m1g m 2 / C1 C 2
NMOS, gate oxide capacitance per unit area, transistor TBS (s) = = (11)
I in (s) D(s)
aspect ratio and bias current, respectively. The
AP: Iin1 = Iin2 = Iin3 = Iin
transconductance gain gm is adjustable by a supplied
Iout (s) s 2 − (g m 2 / C2 )s + g m1g m 2 / C1C2
bias current Ib. TAP (s) = = (12)
Iin (s) D(s)
3. Circuit Configuration and Analysis

Fig. 3 shows the mixed-mode biquad circuit
configuration. The circuit is constructed with five
OTAs and two grounded capacitors.
Routine analysis yields the voltage and current
outputs (Vout(s) and Iout(s)) given by
N v (s) N (s)
Vout (s) = I out (s) = i
D(s) (3) D(s) (4)
where
1 g Fig. 2 OTA with MOS transistors.
N v (s) = [{I in 3 (s) + g m 2 Vin 2 (s)}s 2 − m 2 {I in 2 (s)
g m3 C2
(5)
g m1g m 2
+ g m1Vin1 (s) − g m 4 Vin 3 (s)}s + {I in1 (s) + g m5 Vin 4 (s)}]
C1C 2
Ni (s) = g m3 N v (s) (6)

g m2 g m4 g g g
D(s) = s 2 + s + m1 m 2 m5 (7)
C 2 g m3 C1 C 2 g m 3
It is found from the equations above that the circuit

can perform the mixed-mode operation by selecting
the input and output terminals. And various circuit
transfer functions can easily be realized choosing the Fig. 3 Mixed-mode biquad circuit configuration.
A Mixed-Mode Biquad Employing OTAs and Grounded Capacitors 153
The circuit parameters ω0 and Q can be expressed as: the current-mode circuit characteristic with the cut-off
g m1 g m 2 g m 5 1 C 2 g m1 g m 3 g m 5 frequency f0(=ω0/2π) = 1 MHz, quality factor Q = 1.0
ω0 = , Q= (13)
C1 C 2 g m 3 g m4 C1 g m 2 and gain constant H = 1.0. In this simulation, we have
The circuit parameters ω0 and Q are tuned used a macro model of the OTA shown in Fig. 2.
orthogonally by adjusting the trans-conductance gains To realize the circuit characteristic above, we have
of the OTAs. determined that the bias currents and capacitors were
The sensitivities with respect to circuit active and Ib1 = Ib2 = Ib3 = Ib4 = Ib5 = 40 μA and C1 = C2 = 17 pF,
passive components (i.e. trans-conductance gains and respectively. Also, we have set the supply voltages
capacitors) are shown in Table 1. We can find from and input current at VDD =‐ VSS = 1.85 V and Iin = 10
these values that the biquad enjoys very low μA.
sensitivities to the circuit components. It noted that the Fig. 4 shows the simulation responses with PSPICE.
sensitivities do not depend on the circuit component Fig. 4a shows the LP, BP, HP and BS responses. The
values. AP response is shown in Fig. 4b. The simulation
In the following, we consider to realize the responses are favorable enough over a wide frequency
voltage-mode circuit transfer functions. In range. The power dissipation was 4.83 mW.
voltage-mode operation, the LP, BP, HP, BS and AP Fig. 5 shows the simulation responses with
transfer functions are obtained by selecting the input f0-tuning (i.e. f0 = 500 kHz, 1 MHz and 2 MHz),
terminals as follows: keeping Q = 1.0 and H = 1.0. In this case, the
Voltage-mode operation (Iin1 = Iin2 = Iin3 = 0) capacitors and bias currents were C1 = C2 = 17 pF and
LP: Vin4 = Vin, Vin1 = Vin2 = Vin3 = 0 Ib3 = Ib4 = Ib5 = 40 μA, Ib1 = Ib2 = 9.6 μA, 40 μA and
Vout (s) g m1g m2 g m5 / C1C 2 g m3 210 μA, respectively. Fig. 6 shows the simulation
TLP (s) = = (14)
Vin (s) D(s) responses with Q-tuning (i.e. Q = 0.707, 1.0, 2.0 and
BP: Vin1 = Vin, Vin2 = Vin3 = Vin4 = 0 5.0), keeping f0 = 1 MHz. In this case, the capacitors
Vout (s) (g g / C g )s and bias currents were C1 = C2 = 17 pF and Ib1 = Ib2 =
TBP (s) = = − m1 m 2 2 m3 (15)
Vin (s) D(s) Ib3 = Ib5 = 40 μA, Ib4 = 80 μA, 40 μA, 10 μA and 1.6
HP: Vin3 = Vin, Vin1 = Vin3 = Vin4 = 0 μA, respectively. It is found that the circuit parameters
Vout (s) g m 2 s 2
f0 and Q can be tuned electronically adjusting the bias
THP (s) = = (16)
Vin (s) g m3 D(s) currents of the OTAs.
BS: Vin2 = Vin4 = Iin, Vin1 = Vin3 = 0 In the following, we have considered about the
voltage-mode biquad circuit. The circuit parameters f0,
Vout (s) g m 2 s 2 + g m1g m5 / C1C 2
TBS (s) = = (17) Q and H are same as the current-mode ones. Also, we
Vin (s) g m3 D(s)
have set that the values of the bias currents and
AP: Vin1 = Vin2 = Vin4 = Vin, Vin3 = 0
capacitors were all the same as the current-mode circuit.
Vout (s) g m 2 s 2 − (g m1 / C2 )s + g m1g m5 / C1C2
TAP (s) = = (18)
Vin (s) g m3 D(s) Table 1 Sensitivity to circuit components.
In addition, the trans-admittance-mode or x Sωx 0 SQx
trans-impedance-mode operation is obtained from gm1 0.5 0.5
gm2 0.5 -0.5
selecting the input terminal Iin(s) or Vin(s) and output
gm3 -0.5 0.5
terminal Vout(s) or Iout(s). gm4 0.0 -1.0
gm5 0.5 0.5
4. Design Examples and Simulation Results
C1 -0.5 -0.5
As a design example, we consider a realization of C2 -0.5 0.5
(a) Fig. 6 Q-tuning responses (current-mode circuit).
(b)
Fig. 4 Simulation responses (current-mode circuit). (a)
(b)
Fig. 5 f0-tuning responses (current-mode circuit).
Fig. 7 Simulation responses (voltage-mode circuit).
Fig. 7 shows the simulation responses. You can see
In this simulation, the size of all the MOS
that the simulation responses are good enough as well
transistors have W = 4 μm and L = 2 μm. And we
as the current-mode ones. Here, we have set the input
have used the parameters of MOSIS 0.5 μm for other
voltage Vin = 100 mV. The power dissipation was 4.83
device parameters.
mW.
A Mixed-Mode Biquad Employing OTAs and Grounded Capacitors 155
5. Conclusions characteristics. The solution on this will be discussed

in the future.
A mixed-mode biquad employing five OTAs and
two grounded capacitors has been proposed. We have References
demonstrated that the circuit can perform the [1] Abuelma’atti, M. T. et al. 2005. “A Novel Mixed-Mode
mixed-mode operation by selecting the input and OTA-C Universal Filter.” International Journal of
output terminals, and that the circuit enables LP, BP, Electronics 92 (7): 375-83.
[2] Bhaskar, D. R. et al. 2005. “New OTA-C Universal
HP, BS and AP transfer functions by suitably
Current-mode/Trans-admittance Biquads.” IEICE
choosing the input terminals. Additionally, the circuit Electronics Express 2: 8-13.
parameters ω0 and Q can be tuned orthogonally [3] Ramirez-Angulo, J. et al. 1992. “Current-mode
through adjusting the trans-conductance gains of the Continuous-time Filters: Two Design Approaches.” IEEE
Transactions on Circuits and Systems 39 (6): 337-41.
OTAs. It has been made clear that the circuit has very [4] Tao, Y., and Fidler, J. K. 2000. “Electronically Tunable
low sensitivities to the circuit active and passive Dual-OTA Second-Order Sinusoidal Oscillators/Filters
components. with Non-interacting Controls: A Systematic Synthesis
Approach.” ibid. 47: 117-29.
The achievement examples have been given
[5] Tsukutani, T. et al. 2003. “A Realization of Multiple
together with simulation results by PSPICE. The Circuit Transfer Functions Using OTA-C Integrator Loop
simulation responses have been appropriate enough Structure.” IEICE Transactions on Fundamentals E86-A,
over a wide frequency range. The circuit configuration (2): 509-12.
[6] Tsukutani, T. et al. 2014. “Electronically Tunable Inverse
is very suitable for implementation on both bipolar
Active Filters Employing OTAs and Grounded
and CMOS technologies. Capacitors.” International Journal of Electronics Letters,
The non-idealities of the OTA may affect the circuit online.
doi: 10.17265/2328-2223/2018.03.004
D DAVID PUBLISHING
A Simple Traffic Model Including Possibility of

Overtaking
Reuben Thieberger
Department of Physics, Ben Gurion University, Beer Sheva 84105, Israel
Abstract: Previously we examined different parameters relevant to traffic flow. For illustrative purposes we considered a specific case
of approaching a city. The case involves a traffic light where one continues on the main road, into which additional cars are entering at
the light. At this intersection an alternative route begins, which is longer but into which no additional cars are entering. In addition to the
“Slow to move” model we add here a possibility of “overtaking”, which is quite artificial as our model is one dimensional. Still we think
that this gives fair results.
Key words: Cellular automata, traffic flux, overtaking.
1. Introduction The most basic model, for traffic, was the paper by
Biham et al. [3]. This model consists of a one
Empirical observations of traffic show that at high
dimensional chain of sites on which particles are
enough densities the behaviour of traffic becomes quite
distributed according to their density. At each time step
complex. Therefore, Cellular Automata is one of the
each particle can move to the right if that site is not
most used methods for evaluating traffic and that is
occupied. This model enabled to distinguish between
because of their speed and complex dynamic behaviour.
free flow at low densities and a jammed region for high
Cellular automata were first studied by Ulam and von
densities of particles. A more realistic model was
Neumann [1]. An important contribution to the field
suggested by Nagel and Schreckenberg [4] who
was in the work of Wolfram [2] who introduced
introduced several modes for the car velocity. A
classifications, used in the present study. The
different approach was taken by M. Takayasu and H.
elementary Cellular Automaton is a collection of cells
Takayasu [5] who introduced the effect of delay by
arranged on a one dimensional array. Each cell can
allowing a car to move only if the next and after the
obtain just two possible numbers: one and zero. The
next site are empty. More recently introducing
“time” is discreet and at each time step all the cell
additional connection sites into essentially the
values are updated synchronously. The value of each
Nagel-Screckenberg model was suggested by Nassab
cell depends just on the values in the previous step of
et al. [6].
that cell and its two neighbours. Wolfram names each
In a previous study [7], we examined a specific
elementary Cellular Automaton with a binary numeral,
traffic problem. In that study we added to previous
which he calls: “rule”. This value results from reading
calculations the “slow to move” model [8]. This gives
the output when the inputs are lexicongraphically
us a delay just as in the previously described models
ordered. This will become clearer when we explain the
and seems to us simpler. The feature which we added in
rules which we use.
this study is a sort of “overtaking” which is somewhat
artificial as we are considering just one lane traffic. We
Corresponding author: Reuben Thieberger, Ph.D.,
professor of physics, research fields: statistical
will explain later the method. It seems to us that this
physics,computational physics. model gives the main features of previous studies but is
A Simple Traffic Model Including Possibility of Overtaking 157
simpler. which are used in our calculations are implemented.

To make our exposition clearer we describe again
2.1 Measures
the procedure given in our previous study. The rules we
used are taken from the cellular automata model as The density, p, is given by the ratio between the
proposed by Gershenson and Rosenblueth [9]. Our number of “ones” (i.e. vehicles) and the total number
main interest in this paper is to see how the “overtaking” of cells. Initially we take this value to be the same for
changes the traffic flux. all regions. We check how this value changes in the
different regions. Here we are interested only in the
2. The Model
equilibrium values. The velocities, v, denoted by vt, for
Our highways are represented by an array of cells, the velocity after the second traffic light, is given by the
each cell has the values zero or one. One represents a number of cells which change in one step from 0 to 1.
vehicle and zero an empty portion of the highway. We In our calculation, space and time are just abstract
assume that the magnitude of a cell corresponds to the quantities. Still if concrete numbers are desired, one
average length of a vehicle. In Fig. 1, we show the
layout of our model. At a certain point we have a
bifurcation where there are two different ways to
proceed and at a later point where they merge again.
This model represents in a simplistic way the
possibility of using two alternative routes (the main
route and the “bypass”) when approaching a city from a
certain direction of suburbs. We add the possibility that
additional cars are coming into the main road and are Fig. 1 The movement of vehicles.
removed when approaching the city. So that overall the
Table 1 Wolfram rules used in this model.
number of vehicles is preserved.
t-1 t184 t252 t136
In Fig. 1, “IN 1” is the main car movement and we
000 0 0 0
assume that the main outflaw (“OUT 1”) of cars is the 001 0 0 0
same as the number entering. The first junction is light 010 0 1 0
1 and after the intersection additional cars enter (“IN 011 1 1 1
2”), this is the same number of cars as leaving at “OUT 100 1 1 0
101 1 1 0
2”. This procedure keeps the overall number of cars
110 0 1 0
constant. At light 1 we have the possibility of cars to 111 1 1 1
move to the bypass at light 2 they again combine with
the main road.
In this paper we have one modification, we assume
that for a certain percentage of the cars there is a
possibility to “jump” three spaces in advance if in front
of the car there is one car and two empty spaces before
it.
The rules, which are the same as used by Gershenson
and Rosenblueth [9], are given in Table 1.
In Fig. 2, we show schematically how the rules Fig. 2 The movement of vehicles.
can quote [9] were the “length” of one cell is given by 3. Results and Discussion
five meters, and a time step is given by a third of a
We used a fixed grid: The main road was comprised
second. As the usual movement of “cars” is one cell per
of 1,200 cells, the “by pass” 300 cells and the distance
time step, these values give us a speed of about 50
between the two lights was 120 cells. We used the
km/hour, roughly the speed limit within a city.
“green wave” regime. As we have just two lights it was
The time going on the bypass may be longer or
shown by Gershenson and Rosenblueth [9], that in this
shorter than the one going straight on the main road.
case one does not get different results using the
We will check the ratio between these two times and
“self-organizing” regime.
denote it by “q:pe”.
The number of cycles we have to use in order to
We have a parameter telling us the amount of “cars”
obtain significant results depends on the parameter we
added to the main road at the junction of the bypass.
wish to calculate. For calculating the velocity or the
This same amount is deducted from the “main road”
flux 32,000 cycles give quite accurate resuls. We
farther away and is done in order to preserve the total
determine the number of cycles needed by taking a
number of vehicles. The actual addition of cars is
second sample, with different random numbers, and
governed by a random number which depends on a
comparing the values thus obtained. To be on the safe
parameter (i.e. the percentage of cycles when a car is
side we used 64,000 cycles on the here presented
added). We have a parameter telling us what part of
results.
cars do not move even when they could move
We introduce a vehicle on the first intersection for
according to our rules. We denote this parameter by crr,
40% of the steps and we eliminate the same number of
and it changes between zero and one. The last
vehicles on the last point of our main route, again per
parameter which we add in this section is the percent of
unit time.
cars which can jump: “pc-skip”.
In Fig. 4 and Fig. 5 we show the results, for a
2.2 The Grid specific density, of the velocity and flux as a function
of the part of cars which are allowed to overtake. In
A general view of the grid is given in Fig. 3.
these figures the density was taken as ρ = 0:6 and the
The figure showed that we denote the cells on the
part of cars which are slow to move by crr = 0:5.
main route by ip and the cells on the bypass by iq. The
Fig. 6 shows the ratio of time it takes to go on the
cells between ip = istop1 and ip = istop2 we denote by
main road compared to the time to go by the “bypass”.
ipe. The cells after ip = istop1 we denote by ipt. At ip =
It is interesting to note the big difference in the result as
istop the vehicles move on the main road or on the
a function of the part of cars which are allowed to
bypass according to the “lights”.
overtake.
The understanding of the change in velocity as a
function of the “pc-skip” shows us the possible gain
compared to the danger of overtaking on rural roads. It
shows us that in many cases the gain is quite small
compared to the increase in the danger of collision.
In conclusion, we can say that our calculations give
us a wide range of information which can be applied
for specific cases. It shows us the importance of
modifying the simple rule by adding modifications to
Fig. 3 The grid. the study presented by Gershenson and Rosenblueth [9].
A Simple Traffic Model Including Possibility of Overtaking 159
Fig. 4 The change in velocity as function of the cars which are allowed to overtake.
Fig. 5 The change in flux as function of the cars which are allowed to overtake.
Fig. 6 The values of q:pe as a function of pc-skip.

References [6] Nasaab, K., Schreckenberg, M., Ouaskit, S., and

Boulmakou, A. 2005. “1/f Noise in a Cellular
[1] von Neumann, J. 1966. Theory of Self-reproducing Automaton Model with Traffic Flow with Open
Automata (edited and completed by Arthur Burks), Boundaries and Additional Connection Sites.” Physica A
University of Illinois Press. 354: 597-605.
[2] Wolphram, S. 1986. Theory and Application of Cellular [7] Thieberger, R. 2016. Influence of Various Traffic
Automata. World Scientific. Densities On 1/F Noise, Traffic and Granular Flow’15,
[3] Biham, O., Middleton, A. A., and Levin, D. 1992. “Self edited by Knoop, V. L., and Daamen, W. Springer
Organization and a Dynamical Transition in Traffic Flow International Publishing Switzerland, 467-73.
Models.” Phys. Rev. A 46: R6124. [8] Nishinari, K., Fukui, M., and Schadschneider, A. 2004. “A
[4] Nagel, K., and Schreckenberg, M. 1992. “A Cellular stochastic cellular automata model for traffic flow with
Automaton Model For Freeway Traffic” J. Phys. I France multiple metastable states” J. Phys. A 37: 3101.
2: 2221. [9] Gershenson, C., and Rosenblueth, D. A. 2009. “Modeling
[5] Takayasu, M., and Takayasu, H. 1993. “1/F Noise in Self Orgenizing Traffic Lights With Elementary Cellular
Traffic Model” Fractals 1: 860-6. Automata” Complex Systems 19: 385.
doi: 10.17265/2328-2223/2018.03.005
D DAVID PUBLISHING
Protection Coordination Optimization for FREEDM

(Future Renewable Electric Energy Delivery and
Management) System
Mohamed F. Kotb1, Magdi El-Saadawi2 and Eman H. El-Desouky2

1. Department of Electrical Engineering, Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt
2. North Delta Electricity Distribution Co., Mansoura 35511, Egypt
Abstract: The FREEDM (future renewable electric energy delivery and management) system is a smart distribution system that
facilitates seamless integration of high-penetration DRER (distributed renewable energy resources) and DESD (distributed energy
storage devices) with the existing distribution system. Protection schemes have been proposed to detect the overcurrent faults
throughout the FREEDM system, according to its requirements. In this paper the time inverse directional over current protection
coordination scheme is developed as a backup protection when the primary protection communication failed. The proposed scheme is
applied to FREEDM network using conventional mathematical model. To speed up the fault clearing time without coordination loss,
the settings of the proposed relays in the two directions are minimized using genetic algorithm. The developed methods are validated
using ETAP software. The results ensure that the faults throughout the FREEDM system sections are detected and the relays tripping
time are minimized.
Key words: FREEDM, protection coordination, genetic algorithm, ETAP software.
Nomenclature Minimum pickup current of relay i

Maximum pickup current of relay i
Future Renewable Electric Energy Delivery and Minimum Time Multiplier Setting for relay i
FREEDM
Management
Maximum Time Multiplier Setting for relay i
DRER Distributed Renewable Energy Resources
NSF National Science Foundation
1. Introduction
SST Solid-state transformer
FID Fault Isolation Device FREEDM (future renewable electric energy delivery
IFD Intelligent Fault Detection
and management) concept created by NSF is a smart
IEM Intelligent Energy Management
PC Protection coordination
grid system with novel features used to improve the
PCS Pickup current setting integration between green DRERs (distributed
OCRs Overcurrent relays renewable energy resources) and DESDs (distributed
PCO Protection-coordination optimization energy storage devices) with the existing conventional
TDS Time dial setting
power networks. This integration can be utilized to
CTM Coordination time margin
The ith Relay operating time provide a reliable source to the loads in addition to
,
The curve set-related parameters (inverse, very storing energy [1-3]. Three key components are
inverse, extremely inverse, etc.)
introduced by FREEDM system to increase the
Fault current value
Pickup (set start) current value effectiveness of the protection system and
Pickup current for relay i consequently improve service quality to customers.
These devices are SST, FID and IFD. SST is a digitally
Corresponding author: Mohamed F. Kotb, Dr., , research controlled converter used instead of the standard
fields: optimum electric protection for microgrid, FREEDM
system. transformer. This device guarantees controlling the
162 Protection Coordination Optimization for FREEDM
(Future Renewable Electric Energy Delivery and Management) System
flow of power in two directions, voltage magnitude and problems/failures [1, 2]. The directional inverse time
various phase relationships of current and voltage. It overcurrent relays are applied to detect the fault in the
allows active controlling of DRER and loads. Other system as in Ref. [2].
than the conventional transformer, the SST is capable The PC is a very important subject in power systems
to limit the fault current inherently to 2.0 pu as it stops extending of the fault and prevent electric
significantly by reducing the voltage for the whole equipment from damage. OCRs are specifically
network [3-5]. FID is a new power electronic applied to attain precise PC to evade electrical
equipment used in FREEDM system, which is capable equipment strain and miss operation of the primary and
to break asymmetrical current with large capacity backup protection elements [6]. Many studies
within microseconds [2, 5]. IFD, IFM and IEM are previously exerted to properly set PCO and can be
incorporated to quickly detect and manage the faults, classified as conventional and analytical
and to achieve operative power flow control methodologies and computerized methods [7].
respectively [4]. The FREEDM system topology and The conventional methods can be characterized as
its major components are shown in Fig. 1 [2]. trial and error, topological analysis and optimization
FREEDM networks are mesh systems causing two methods. Linear programming methodologies [8-10],
major problems, higher short circuit level leading to nonlinear programming [11-13], mixed integer
voltage dip and bi-directional flow of power [2]. So, programing [11], sequential quadratic programming
the pilot-differential protection using communication [14], simplex and dual simplex [15, 16], and analytical
is used as primary protection capable to detect the technique [17] require large number of iterations to
faults in cycles. The overcurrent protection is used as a solve the PC settings. The over-all optimal solution
backup protection in case of communication cannot be obtained when the conventional methods are
Fig. 1 FREEDM system topology and components [2].

Protection Coordination Optimization for FREEDM 163
used [7]. In addition, its convergence is another 3. Proposed Conventional Protection

problem when applied to large systems. To expedite Coordination Algorithm
the convergence time, the topological analysis
To only identify, detect and isolate the faulty section
approaches are used [18, 19]. The artificial
in the FREEDM system, which is a loop system, two
intelligence-based techniques prove good solution for
sets of directional over current relays are selected and
these inadequacies [7]. Genetic algorithm [20-25] and
installed for each zone in two opposite directions as
modified firefly algorithm [6], were appropriately
indicated in Fig. 2.
applied in resolving the PCO problem.
When a fault occurs at any zone, the direction of
In this paper the time inverse directional over current
fault current is detected by both the two zone relays
protection coordination is designed/modeled and
with opposite directions. Each relay has a backup relay
the relays settings are optimized using GA for
as its direction to support the detection process. Relays
FREEDM system. The proposed method can speed up
R1-R2-R3 can detect the fault in anticlockwise direction
the fault detection and isolation, so it can improve
while relays R4-R5-R6 can detect faults clockwise
the reliability by reducing the outages time of faulty
direction indicted in Fig. 2. For any fault at any section,
zones.
the closet two relays in the faulty zone should be
2. Problem Statement activated due to the clockwise and anticlockwise
The main aim of PCO between OCRs is to find out currents fed from both sides. For example, if a fault
the optimal TDS and PCS of relays under system exists at zone 3, the fault currents passes through all
constraints [26]. The time margin between the primary relays but R5 acts in clockwise direction, while R2
and backup OCRs is critical decision to sustain the detect the fault current in anticlockwise direction to
accuracy of the discrimination practice [27]. isolate zone 3. Relays R4 and RX will be back up for R5
Interference between the main and backup relays is not respectively while relays R1, RX are back up for R2
permitted. CTM between the relays are generally taken respectively. Both the relay operating time, pickup
between 0.2 to 0.5 S depending on the relay type and current setting and time multiplier setting are
manufacturer [28]. The PCS can be adjusted between calculated as following:
50% and 200% with 25% step [9]. PCS has upper and A- Relay Operating Time
lower limits based on the minimum and maximum fault The over current relays setting can be expressed in
current successively [6]. Eq. (1) according to IEC 60255-151 [29].
In this paper the PC is designed for the FREEDM .
where (1)
system using the conventional technique and
simulating the system in ETAP software for validation. B- Relay Pickup Current Setting
The time dial setting and plug setting of the selected (2)
over current relays are designed according to the
C- Relay Time Multiplier Setting
coordination procedures and constraints. To speed up
the tripping times and to isolate the faulty section as (3)
soon as possible, GA is applied to minimize the relays
4. Proposed GA-based Optimal Protection
TDS using the same rules and constraints. The
Coordination
minimum values are also reapplied to ETAP simulator
to validate and proof the results. The model of the OCR coordination presented in the
Fig. 2 Proposed locations and directions of overcurrent relays in FREEDM system.
previous section, requires a special optimization tool to same constraints mentioned in Eqs. (1), (2) and (3).
handle the linearity of such highly-constrained model. Objective Function
Most of the deterministic optimization methods are not Minimize: ∑ (4)
suitable to calculate and solve the presented
where: is relay (i) time delay for the nearest
optimization problem due to many mathematical
downstream maximum fault and N is the number of
difficulties. The search algorithm is a probabilistic
relays.
intelligent method, which searches a population of
points in parallel. GA gives correct effective and 4.2 Genetic Algorithm
practical solutions to define the optimal decisions for The GA is a probabilistic intelligent search
any processes and tasks. GA can be applied with algorithm, which searches a population of points in
discrete, discontinuous and linear functions. parallel. GA differs from other traditional optimization
4.1 Problem Formulation methods in three significant points [31]. It searches a
population of points in parallel, it uses probabilistic
In this paper, GA is applied to minimize the relays rules rather than deterministic ones, and it can process
time dial setting using the same rules and constraints. an encoding set of parameters. The real coding is used
The main objective of the proposed method is to in this research since it provides better performance
minimize the relays tripping time of the previously and faster conversion compared to other coding
planned arrangement to isolate the faulty zone as soon methods. With other coding, it is required to alternate
as possible using GA. The proposed methodology to the real coding in the phase of calculating the total
procedure is explained by Fig. 3. The results of this daily cost, which results in additional processing time.
method are proved using ETAP software. The It is required to examine the capability of the GA to
objective function is formulated as in Eq. (1) under the solve the coordination problem compared to the traditional
start
Generate primary/pickup pairs
Perform load flow study
Perform faulty study
Setting of overcurrent relays
No
Check coordination using
ETAP package
Yes
Perform optimal selection of time dial setting using GA
Record outputs setting
End
Fig. 3 A flow chart of the proposed methodology procedure.
technique. The success of the GA means that it will be augmenting additional cost terms with the main
possible to implement GA for other problems, where the objective function. The additional terms assign
traditional methods are not valid [23]. nonlinear costs for solutions that violate the constraints
To employ GA with a tightly-constrained problem, depending on their relative locations with respect
such as the coordination problem, it is possible to to the feasibility boundaries [24, 25]. The flowchart
generate only feasible solutions by avoiding indicated in Fig.4 shows the GA evolution process
individuals which violate the given constraints. Since [32].
the infeasible solutions mostly cover the search space
5. Implementation of Protection
at the initial generation, the complete avoidance of the
Coordination for FREEDM System
infeasible solutions gives a high possibility for missing
the area of global minimum. Another approach is The FREEDM system indicated in Fig. 1 is used to
achieved by moving the infeasible individuals to the apply and verify the proposed methodology [30].
nearest feasible area. For the highly-constrained Conventional circuit breakers are used instead of FIDs.
problem, this approach would be too complex and a The FREEDM system is represented in ETAP software
very time-consuming process. The penalty function as in Fig. 5 to perform load flow, short circuit and
approach is another alternative that converts the protection coordination studies without and with
constrained problem to an unconstrained one by minimization.
Start
Chromosome population Initialization
Objective function value per member using the random setting. The
penalty values are considered if found.
Fitness is estimated then accomplish the fitness scaling
Select parent chromosomes
Produce new offspring “Crossover and Mutation”
Evaluate offspring and insert the best replacing worst parents
Yes Migrate individuals

Number of generations between
migration reached? between subpopulations
No
No
Maximum number of generation
reached?
Yes
Save the optimal settings
End
Fig. 4 Flowchart of the GA evolution process.
Fig. 5 FREEDM system load flow current represented in ETAP.

5.1 Description of System under Study Table 4.

5.2.1 Selection of Relay Current Pick up Values
The system is composed of 12.47 kV/2 MVA
The current pick up and time dial values are
distribution loop network, three main sections, four
important characteristics of any rely. The results
circuit breakers, local generation and energy storage
obtained in Table 4 are used to obtain the relays pickup
devices capable to supply AC and DC currents with
currents and time dial (TDS) by applying Eqs.
data as in Table 1 [1]. (1), (2) and (3). The minimum pickup current of
5.2 Conventional Protection Coordination any relay is considered as 125 to 150% of the full load
current while the maximum pickup current of
The proposed conventional protection coordination any relay is considered as 50% of the minimum fault
algorithm is applied to the tested FREEDM network current.
modeled in ETAP program. The load flow and short In order to determine time dial settings, maximum
circuit for the three zones are performed. The load flow fault current seen by relays is used for calculation to
currents in the FREEDM system represented in ETAP ensure quick operation of relays for all type of faults.
are expressed in Fig. 5. The load flow currents through Table 6 shows the maximum currents seen by the
different relays are given in Table 2. relays during fault at different zones. Normal inverse
The short circuit currents passing by certain circuit over current relay is selected taking , as 0.14
breakers at the three zones are recorded in Table 3. and 0.02 respectively [29]. is considered as 0.5
Whereas, Fig. 6 shows the short circuit current flow for the primary relays while the time margin between
due to a fault at zone 1 conducted by ETAP. primary and backup relay is considered as 0.2 sec. The
The maximum and minimum fault currents obtained relays operating time and the adjustable time
after performing the three lines to ground and line to multiplier settings for all relays can be calculated as
ground faults using ETAP simulation are shown in indicated in Table 7.
Table 1 FREEDM network data [1].
Feeder lines data
line R1(ohm) R0 (ohm) L1(mh) L0 (mh) Length (km)
1 3.971 3.971 6.894 9.842 7.141
2 1.257 1.257 2.183 3.116 2.261
3 1.634 1.634 2.837 4.051 2.939
4 1.634 1.634 2.837 4.051 2.939
Feeder SST’s data
Distribution SST Phase A Phase B Phase C Loads at 0.24 kV
D-SST1 0.116 MVA 0.194 MVA 0.102 MVA 0.100 MVA
Table 2 Load flow current through relays using ETAP.

Relays R1 R2 R3 R4 R5 R6 Rx
Current Ampere 18 8.1 0.3 8.1 0.3 8.1 26.2
Table 3 Short circuit current at the three zones passing by circuit breakers.
Zone SC current at CB1 (A) SC current at CB2 (A) SC current at CB3 (A) SC current at CB4 (A) SC current at CBx (A)
1 0.751 0.151 0.141 0.136 0.886
2 0.555 0.563 0.257 0.248 0.803
3 0.374 0.38 0.388 0.398 0.771
Fig. 6 FREEDM system short circuit current flow due to a fault at zone 1.
Table 4 Fault current magnitudes for faults at different zones.

Fault current sensed by R1 at Fault current sensed by R2 Fault current sensed by R3 Fault current sensed by R4 at
CB1 (kA) and R6 at CB2 (kA) and R5 at CB3 (kA) CB4 (kA)
Fault LLL-G SLG LLL-G SLG LLL-G SLG LLL-G SLG
Zone 1 0.572 0.657 0.105 0.12 0.105 0.12 0.105 0.12
Zone 2 0.555 0.457 0.563 0.457 0.257 0.205 0.248 0.205
Zone 3 0.374 0.303 0.38 0.303 0.388 0.303 0.388 0.317
Table 5 Relay pick up settings.

R1 R2 and R6 R3 and R5 R4
(A) 1.25 1.487 1.4 1.4
(A) 0.225 0.304 0.0113 0.304
CT ratio 600/5 200/5 200/5 200/5
(A) 1 1.2 1.3 1.4
Table 6 Clockwise and anticlockwise relays SC currents.

Relays acting as primary protection Relays acting as backup protection
Zones Clockwise Anti-clockwise Clockwise Anti-clockwise
Tripped CB Tripped CB
Relay ISC (A) Relay ISC (A) Relay ISC (A) Relay ISC (A)
1 R6 105 R1 572 2 and 1 R5 105 RX 569 3 and X
2 R5 257 R2 257 3 and 2 R4 248 R1 248 4 and 1
3 R4 388 R3 388 3 and 4 RX 252 R2 380 X and 2
Table 7 Coordination settings obtained by conventional method.

Clock wise coordination settings Anti-clock wise coordination settings
1.4 =1.0 = 0.7 = 1.0 = 0.8

1.3 = 0.9 = 0.6 = 0.8 = 0.62
1 = 0.7 = 0.5 = 0.63 = 0.5
1.4 =1.2 = 1.15 = 1.2 = 1.15
The coordination study is verified using ETAP • Full load current is chosen according to load flow
package. The Star View Coordination TCC curves for report (maximum current for normal scenario).
the relays operating in anti-clockwise and clockwise • Maximum value of short circuit current is taken
directions are shown in Figs. 7 and 8 respectively. The from Table 6.
Coordination TCC curves before optimization for zone • Operating time of relay is calculated according to
2 and 3 are indicated in Figs. 9 and 10. The obtained Eqs. (1), (2) and (3).
results indicated in Table 7 and Figs. 7 to 10 ensure that 5.3.1 Optimization Criteria
the proposed method is capable to fully protect the The following assumptions are taken when applying
FREEDM network and coordinate its relays tripping the proposed GA-based method:
and enable to be backup protection for its primary y Coordination time delay interval between backup
protection when the communications failed. and primary relay is 0.2 s.
y The is maintained between 0.1 and 1.1
5.3 GA-based Optimum Protection Coordination
according to IEC 60255-151 [29].
Optimization coordination problem is solved as per y The minimum pickup current of a relay is
the following arrangements: considered as 125 to 150 % of the full load current.
• The development of primary and back up relays while the maximum pickup current of relay
pairs are accomplished according to the previous considered 50% of the minimum fault current
explained arrangement for clockwise and y K1 and K2 are taken as 0.14 and 0.02 respectively,
anti-clockwise. according to IEC 60255-151 [29].
Fig. 7 Star view coordination TCC curves for the relays operating in anti-clockwise direction.
Fig. 8 Star view coordination TCC curves for the relays operating in clockwise direction.
Fig. 9 Coordination TCC curves for R2, R5 at zone 2 before optimization.

Fig. 10 Coordination TCC curves for R3 R4 at zone 3 before optimization.
The GA-based optimum protection coordination is The TCC if a fault occurred at zone 2 and zone 3 are
applied to the FREEDM network test system. The indicated in Figs.13 and 14.
results are applied to the FREEDM network system A comparison between Ti and TDSi for both
modeled in ETAP software for verification purposes. clockwise coordination settings and anti-clock wise
Table 8 summarizes the parameters of the GA-based coordination settings is shown in Tables 10 and 11.
optimization process applied to the coordination From these tables the total tripping time in clockwise
optimization. and counter-clockwise for conventional coordination
5.3.2 Simulation Results and GA optimization coordination can be concluded in
The developed GA program is implemented to the Table 12.
same FREEDM network with data indicated in Table 1 It can be observed that the anticlockwise relays total
to minimize the relays tripping time and to isolate the tripping time is reduced by 47.2% and 44.7% in
faulty zone as soon as possible. The coordination clockwise direction. The proposed optimized
settings results are tabulated in Table 9 indicating the protection method is capable to fully protect the
optimum pickup currents with the minimum time dial FREEDM network and coordinate its relays
setting and tripping time. To verify the effectiveness of tripping. The method is able to be backup protection
the proposed GA, its coordination setting results are for its primary protection when the communications
applied and simulated using ETAP software. The failed within minimum time. The minimization
coordination is satisfied for all cases. Figs. 11 and 12 of tripping time guarantees reducing faults
indicate the TCC coordination curves for the consequences on the FREEDM system while achieving
anti-clockwise relays, R1-R2- R3-RX and the clockwise the protection requirements and protection
relays, R4-R5-R6-RX respectively due to optimization. coordination.
Table 8 GA-optimization parameters [32].

Total population size 200
Number of individuals per subpopulation 20
Maximum generations 1,000
Number of subpopulations 10
Generation gap 0.8
Migration rate between subpopulations 0.2
Insertion rate 0.9
Probability of crossover 0.9
Probability of mutation 0.01
Number of generations between migration 20
Table 9 Coordination settings after optimization using GA.

Clock wise coordination settings Anti-clock wise coordination settings
1.4 = 0.62 = 0.4848 = 0.53 = 0.4279

1.3 = 0.4 = 0.3001 = 0.31 = 0.2584
1 = 0.35 = 0.2405 = 0.36 = 0.3046
1.4 = 0.7 = 0.673 = 0.7 = 0.673
Fig. 11 Coordination TCC curves for R1-R2-R3 RX using GA optimization.

Fig. 12 Coordination TCC curves for R4-R5-R6 and RX using GA optimization.
Fig. 13 Coordination TCC curves for R3-R4 at zone 3 using GA optimization.

Fig. 14 Coordination TCC curves for R2 R5 at zone 2 using GA optimization.
Table 10 Clock wise coordination settings.

Ti
Conventional GA Conventional GA
1.4 = 1.0 = 0.62 = 0.7 = 0.4848
1.3 = 0.9 = 0.4 = 0.6 = 0.3001
1 = 0.7 = 0.35 = 0.5 = 0.2405
1.4 = 1.2 = 0.7 = 1.15 = 0.673
Table 11 Anti-clock wise coordination settings.

Ti
Conventional GA Conventional GA
1.4 = 1.0 = 0.53 = 0.8 = 0.4279
1.3 = 0.8 = 0.31 = 0.62 = 0.2584
1 = 0.63 = 0.36 = 0.5 = 0.3046
1.4 = 1.2 = 0.7 = 1.15 = 0.673
Table 12 Time between primary and backup relays (Sec).

Primary and backup relays Conventional Coordination GA Optimization
(Anti-clockwise) R1 R2 R3 RX 3.6 1.9
(Clockwise) R4 R5 R6 RX 3.8 2.1
6. Conclusion (ECTI-CON), Phetchaburi, Thailand. DOI:

10.1109/ECTICon.2012.6254294.
Directional over current protection scheme is [9] Ezzeddine, M., Kaczmarek, R., and Iftikhar, M. U. 2011.
“Coordination of Directional Overcurrent Relays Using a
proposed for the FREEDM system using the
Novel Method to Select Their Settings.” IET Gen., Trans.
conventional mathematical model as backup protection & Dist. 5 (7): 743-50. DOI: 10.1049/iet-gtd.2010.0603.
when the primary protection failed due to [10] Sueiro, J. A., Dorado, E. D., Míguez, E., and Cidrás, J.
communication fail. The coordination of the proposed 2012. “Coordination of Directional Overcurrent Relay
Using Evolutionary Algorithm and Linear Programming.”
scheme is achieved for clockwise and anti-clockwise International Journal Electrical Power & Energy Systems
directions. The total time to isolate the FREEDM 42 (1): 299-305.
system is minimized using GA. The total tripping time [11] Huchel, Ł., Zeineldin, H. H., and El-Saadany, E. F. 2017.
“Protection Coordination Index Enhancement Considering
in both anticlockwise and clockwise directions reduced
Multiple DG Locations Using FCL.” IEEE Trans. on
by 47.2% and 44.7% respectively. The protection Power Deliv. 32 (1): 344-50.
coordination is validated using ETAP software for both [12] Bedekar, P. P., and Bhide, S. R. 2011. “Optimum
conventional and optimized methods. Coordination of Directional Overcurrent Relays Using the
Hybrid GA-NLP Approach.” IEEE Transactions on
References Power Delivery 26 (1):109-19.
[13] Abyaneh, H. A., et al. 2003. “A New Optimal Approach
[1] Huang, A. 2010. “FREEDM System—A Vision for the for Coordination Overcurrent Relays in Interconnected
Future Grid.” IEEE Power and Energy Society General Power Systems.” IEEE Transactions on Power Delivery
Meeting, Providence, USA, 1-4 (DOI: 18 (2): 430-5.
10.1109/PES.2010.5590201). [14] Birla, D., Maheshwari, R. P., and Gupta, H. O. 2006. “A
[2] Sharma, N. 2015. “Novel Directional Protection Scheme New Nonlinear Directional over Current Relay
for the FREEDM Smart Grid System.” M.Sc. thesis Coordination Technique and Banes and Boons of Near
submitted to Arizona State University, Publ. Numb.: AAT End Faults Based Approach.” IEEE Trans. on Power
1596547; ISBN: 9781321980240; Source: Masters. Deliv. 21 (3): 1176-82.
[3] https://www.freedm.ncsu.edu/. [15] Bedekar, P. P., Bhide, S. R., and Kale, V. S. 2010.
[4] Mandava, P. 2014. “Design and Development of “Optimum Coordination of Overcurrent Relay Timing
Protection Schemes for FREEDM Smart Grid Systems.” Using Simplex Method.” Elect. Power Comp. and Syst. 38
M.Sc. thesis submitted to Arizona State University, by the (10): 1175-93.
Grad. Superv. Comm.: George Karady, Chair. Raja [16] Bedekar, P. P., Bhide, S. R, and Kale, V. S. 2009.
Ayyanar. “Optimum Time Coordination of Overcurrent Relays
[5] Vodyakho, O., et al. 2011. “Solid-state Fault Isolation Using Two Phase Simplex Method.” World Academy of
Devices: Application to Future Power Electronics-based Science Engineering & Technology 28: 1110-4.
Distribution Systems.” IET Electric Power Applic. 5 (6): [17] Mahari, A., and Seyedi, H. 2013. “An Analytic Approach
521-8. for Optimal Coordination of Overcurrent Relays.” IET
[6] Tjahjono, A., et al. 2017. “Adaptive Modified Firefly Gener., Transm. & Dist. 7 (7): 674-80.
Algorithm for Optimal Coordination of Overcurrent [18] Thangaraj, R., Pant, M., and Deep, K. 2010. “Optimal
Relays.” IET Gen., Trans. & Dist. 11 (10): 2575-85. DOI: Coordination of Over-current Relays Using Modified
10.1049/iet-gtd.2016.1563. Differential Evolution Algorithms.” Engineering Appl. of
[7] El-Fergany, A. 2016. “Optimal Directional Digital Artif. Intel. 23 (5): 820-9.
Overcurrent Relays Coordination and Arc-flash Hazard [19] Mohamed, M. M., Mekhamer, S. F., and El-Kharbawe, N.
Assessments in Meshed Networks.” Intern. Trans. on 2007. “A Modified Particle Swarm Optimizer for the
Elect. Ener. Syst. 26 (1): 134-54. DOI: 10.1002/etep.2073. Coordination of Directional Overcurrent Relays.” IEEE
[8] Niyomphant, S., Leeton, U., Kulworawanichpong, T., and Trans. on Power Deliv. 22 (3): 1400-10.
Chomnawan N. 2012. “Application of Linear [20] Singh, M., Panigrahi, B. K., and Abhyankar, A. R. 2011.
Programming for Optimal Coordination of Directional “Optimal Overcurrent Relay Coordination in Distribution
Over-Current Relays.” 9th Intern. Conference on System.” Inter. Conf. on Energy, Automation, and Signal
Electrical Engineering/Electronics, Computer, (ICEAS), Bhubaneswar, Odisha, India, 28-30 Dec. 2011,
Telecommunications and Information Technology 822-7.
[21] So, C. W., Li, K. K., Lai, K. T., and Fung, K. Y. 1997. for Optimal Coordination of Overcurrent Relays.” IET
“Application of Genetic Algorithm for Overcurrent Relay Gen., Trans. & Dist. 7 (7): 674-80. DOI:
Coordination.” 6th Internat. Conf. on Develop. in Power 10.1049/iet-gtd.2012.0721.
Systems Protection, Nottingham, UK, 66-9. DOI: [27] IEEE Std. 242. 2001. “IEEE Recommended Practice for
10.1049/cp:19970030. Protection and Coordination of Industrial and Commercial
[22] Uthitsunthorn, D., and Kulworawanichpong, T. 2010. Power Systems.”
“Optimal Overcurrent Relay Coordination Using Genetic [28] Ojaghi, M., Sudi Z., and Faiz, J. 2013. “Implementation of
Algorithms.” 2010 Intern. Conf. on Advances in Energy Full Adaptive Technique to Optimal Coordination of
Engineering (ICAEE), Beijing, China, 162-5. DOI: Overcurrent Relays.” IEEE Trans. on Power Delivery 28
10.1109/ICAEE.2010.5557589. (1): 235-44. DOI: 10.1109/TPWRD.2012.2221483.
[23] Koochaki, A., Asadi, M. R., Mahmoodan, M., and [29] IEC 60255-151. 2009. “Measuring Relays and Protection
Naghizadeh, R. A. 2008. “Optimal Overcurrent Relays Equipment—Part 151: Functional Requirements for
Coordination Using Genetic Algorithm.” 11th Inter. Conf. Over/under Current Protection.”
on Optimization of Electrical and Electronic Equipment [30] Tatcho, P. A. 2013. “Fault Protection Framework Designs
(OPTIM 2008), Brasov, Romania, 197-202. DOI: and Methods in Power Electronics Converter Based Power
10.1109/OPTIM.2008.4602366. Distribution Systems.” Ph.D. submitted to the Department
[24] Bedekar, P. P., Bhide, S. R., and Kale, V. S. 2009. of Electrical and Computer Engineering, Florida State
“Optimum Coordination of Overcurrent Relays in University College of Engineering.
Distribution System Using Genetic Algorithm.” Inter. [31] Mohammadi, R., Abyaneh, H. A., Mahdinia, H., Fathi, S.
Conf. on Power systems, JCPS ’09, Kharagpur, India. DOI: H., and Rastegar, H. 2011. “Overcurrent Relays
10.1109/ICPWS.2009.5442716. Coordination Constraints.” IEE Trans. on Power Delivery
[25] Singh, D. K., and Gupta, S. 2012. “Optimal Coordination 26 (3): 1927-38. DOI: 10.1109/TPWRD.2011.2123117.
of Directional Overcurrent Relays: A Genetic Algorithm [32] El-Desouki, E. H., Azmy, A. M. R., and Alaam, M. A.
Approach.” IEEE Students’ Conf. on Electrical, 2015. “Internal Networks of Generation Power Plants.”
Electronics and Computer Science (SCEECS), Bhopal, M.Sc. thesis submitted to Tanta University, Faculty of
India, 1-4. DOI: 10.1109/SCEECS.2012.6184808. Engineering, Electrical Power and Machines Eng.
[26] Mahari, A., and Seyedi, H. 2013. “An Analytic Approach Department.
doi: 10.17265/2328-2223/2018.03.006
D DAVID PUBLISHING
Control Algorithm for Non-isolated Supercapacitor

Based Kinetic Energy Recovery System
Luka Lopin1, Slobodan Vukosavić2 and Nikola Lepojević2

1. Computer Controlled Systems, Lola Institute, Belgrade 11030, Serbia
2. The Department of Power Converters and Drives, University of Belgrade, Belgrade 11120, Serbia
Abstract: This paper presents the control algorithm for supercapacitor based kinetic energy recovery system. Supercapacitor shares
energy with the motor drive DC link through bidirectional non-isolated converter. Two cascaded linear regulators control feedback
linearized states of the system. As a result, the DC link voltage is stabilized and the supercapacitor voltage and current are limited.
Control algorithm does not increase ripple of voltages and currents, and thus life time expectancy of supercapacitor is prolonged. In
torque-controlled motor drives, current on DC side can rapidly change with bandwidth as high as 2 kHz. Based on measurements of
KERS states, the algorithm estimates rapidly changing motor drive current on DC side and stabilizes the DC link voltage with
overshoot less than 2%. Controller is independent from the motor drive parameters and relies only on its own measurements. This
enables kinetic energy recovery system (KERS) to be a standalone device.
Key words: Supercapacitor KERS, energy storage, regenerative braking, energy recovery, control system.
1. Introduction the DC link of the electric drive. As a consequence, DC

link voltage increases above nominal value. At this
There are two types of supercapacitors (SC);
moment, switch turns on DBR and irreversibly
pseudocapacitor and electric double layer capacitor
dissipates excess of energy to heat. Fig. 1 shows
(EDLC) [1]. Emerging energy storage unit for power
topology of motor drive with the DBR.
applications are EDLC [2]. EDLC consists of two
Kinetic energy of a drive could be recuperated via
nanoporous active carbon electrodes with specific
active rectifier. But for end user, such recuperation is
surface area in range of 1,000-2,500 m2/g [3]. Such
same as if no energy is recuperated, since energy
capacitors have specific capacitance in range of
companies in most countries are not paying for
100-200 F/g [4]. SCs have approximately one hundred
intermittent bursts of energy.
time greater energy density than electrolytic capacitor
SC based kinetic energy recovery system (KERS)
and ten time greater power densities than chemical
transfers energy from the DC link to the SC through
batteries [5] and high life time expectancies—up to 10
bidirectional converter during braking. Later, during
years [6]. This makes SCs a promising device for
acceleration, braking energy is recovered from the SC
energy storage applications.
to the DC link and consequently to the electric machine.
There is a need for braking in the motor drive
Topology of motor drive with the KERS is shown on
systems. Motor drives most commonly break via
Fig. 2. Same KERS topology is used in Refs. [8-15].
dynamic breaking resistor (DBR). While braking, the
Control algorithm of KERS is expected to fulfill
electric machine changes from motor to generator
following basic tasks:
regime [7]. Kinetic energy of the shaft is than
y Charging SC while motor drive is braking;
transformed into electric energy and it’s transferred to
y Discharging SC while motor drive is accelerating;
y Limiting voltage and current of SC;
Corresponding author: Luka Lopin, M.A., research fields:
signals and systems. y Stabilizing DC link voltage.
178 C
Control Algorrithm for Non--isolated Sup
percapacitor Based Kinetic Energy Rec
covery System
works are prooposed for ennergy manag

netw gement of SCC
in hybrid vehiccles. In Refss. [14, 15] flatness andd
passivity based control are pproposed. Co ontrollers aree
dep
pendent on agging sensitivve parameterss of DC linkk
and
d SC capacitannce.
The
T aim of thhis paper is tto devise a novel
n controll
algo
orithm able tot achieve baasic tasks off KERS withh
impproved ripple,, robustness aand closed loo
op bandwidthh.
Fig. 1 Motor drive topologgy with DBR. Thiis paper prooposes feedbback linearizzation basedd
con
ntroller with estimation
e of motor drive current.
c
2. Control
C Strrategy
Control
C algorrithm stabilizzes the DC link voltage,,
whiile limiting SC voltage andd current. Geeneral controll
blocck diagram iss shown in F Fig. 3. Inner loop
l controlss
the SC current.. Output of inner regulaator controlss
Fig. 2 Motor drive topologgy with SC bassed KERS.
nsistors T1 annd T2 using thhe PWM method. In Refs..
tran
Achievingg these taskks is follow wed by cerrtain [8-113] all algoorithms com mprise of in nner currentt
difficulties. Increased vooltage and cuurrent ripple can conntroller. Outerr loop controols DC link vo oltage. Outerr
lead to irrreversible deegradation of o SC [5] and reguulator outputt represents tthe SC curreent reference..
consequentlyy to shorter MTBF of a KERS. While W Thee current refeerence is dynaamically limiited to ensuree
ripple of currrent and volttage is inhereent for the cirrcuit propper limiting of
o the SC volttage.
in Fig. 2 duee to switchingg operation off the transistorrs T1 Upper
U and loower SC currrent limits deepend on SC C
and T2, it is of interest noot to further increase ripplle by volttage. Similar dynamic currrent limits are a utilized inn
applying inaadequate conttrol algorithm m. Such algoriithm Reff. [12]. Curreent limits, foor current dirrection as inn
increases riipple due to local insttabilities in the Fig. 5, are as folllows:
regulator. Innstabilities aree result of a nonlinearity
n of the y If Usc [Uminn, Umax] => i** [-Imax, Imax]
controlled system or a change of the sysstem y If Usc>Umaxx => i* [0, Imax] and DBR is on
parameters. y If Usc<Umin => i* [-Imaxx, 0] and KER RS is off
In Refs. [88-9] modifiedd linear controller is propoosed. The
T DC link voltage
v reference is fixed slightly abovee
In Ref. [8] SC
S current rippple reaches 10% and sysstem the maximum vooltage of the ddiode rectifier.. This methodd
has low clossed loop banddwidth becausse of PLL. In Ref. is used
u in Ref. [88]. If regulatoor manages too stabilize thee
[9] regulattor relies on two aging sensiitive DC link voltagee, diode rectiffier is inversely polarizedd
parameters— —capacitancee of DC link and SC. Thuus it andd energy exchhange occurss only betweeen the motorr
requires occcasional tuninng of parametters. In Refs. [10, driv
ve and the SC C. As long as the voltage of the SC is inn
11] fuzzy loogic controllerr is proposedd. In Ref. [10]] SC segmment Usc [Uminm , Umax] and motor drive power is less s
current ripplle is slightly improved
i commpared to claassic thann KERS unitt maximum power, regulator will bee
controllers and
a it reaches 25% in the worst w case. In Ref. ablee to keep the DC link voltage at the refference point..
[11], controlller depends on o motor drivve measuremeents. Thiis enables KE ERS system to completeely dischargee
The ripple iss greatly impproved, but coontroller requuires eneergy of the SC during acceleration with a ratee
occasional fine tuning. In Refs. [12, [ 13] neeural dicttated by the motor
m drive.
C
covery System 1799
mple RC moodel. For thhe purpose of regulatorr

sim
syn
nthesis, such models
m are inaadequate. Mo
odeling SC ass
RC circuit has certain
c difficulties. It is impossible too
Fig. 3 General control bloock diagram off KERS.
glect combinned resistancce of inductor and SC
neg C
When SC C voltage is ata the minimuum value, currrent inteernal resistance. Great cappacitance of SC
S results inn
reference limits are changed
c to prevent furrther reall poles of thee LRCsc netw
work and poles of the LCscc
discharge off the SC. Thee DC link volltage drops too the netw
work are imaginary. SC capacitance and internall
nominal valuue, diode recctifier turns ono and the KE ERS resiistance are agging sensitivee. This fact co
ombined withh
unit turns off.
o When SC C voltage is at the maxim mum agin
ng sensitive DC
D link capaacitance results in seriouss
value, currennt reference limits are chhanged to prevent deteerioration of regulator
r robbustness over time.
further chargging of SC annd DBR is turrned on to prevent One
O should notice SC’ss dynamics are severall
dielectric ruppture of the DC
D link. decades slower than
t dynamiccs of inductor and DC link..
Often, SCCs are used to recover part ofo kinetic eneergy. Oveer several sam
mpling periodds of regulator, SC voltagee
If the brakinng torque of thhe motor drivve is too largee, the is roughly
r constant. For thhe reason ofo increasingg
DC link volltage will increase regardlless of the KE ERS robuustness of KERS
K unit, SC is modeeled as ideall
unit transferrring energy with maximuum current too the volttage source. Voltage
V of SC
C is measuredd and updatedd
SC. At this moment,
m the DBR shouldd be turned onn. In in each
e samplinng period off regulator. Fig.
F 5 showss
such scenariio, the DC linnk voltage is limited by DBR D circcuit with SC modeled
m as coonstant voltag
ge source andd
action and part of thee shaft’s kiinetic energyy is recttifier and inveerter modeledd as disturbannce current.
recovered inn KERS unit. State
S space model
m of circuuit shown in Fig.
F 5 can bee
During accceleration of o motor drivve, the DC link writtten as Eq. (11) and Eq. (2)), where:
voltage tendds to decreasee. If acceleratting torque off the y Id—inverterr and rectiffier modeled d as currentt
motor drivee is too largge, the DC link l voltage will distturbance;
decrease regardless of the KERS unit transferrring y E—SC moddeled as consstant voltage source;s
energy with maximum cuurrent from the t SC to thee DC y L—inductoor’s inductancce;
link. At one moment, thee DC link volltage drop will be y C—DC linkk capacitancee;
sufficient too turn on thee diode rectifier. Now motor m y i—inductorr’s current, allso SC’s curreent;
drive is joinntly powered from the KE ERS unit andd the y uc—DC linnk capacitor vvoltage;
diode rectifiier. The KER RS regulator is unsuccessffully
trying to inccrease the DC C link voltagee to the refereence
value with maximum
m SC
C current, thuss draining thee SC
voltage to the
t minimum m value. Once SC is draiined,
KERS unit turns
t off.
3. Controll-Oriented KERS Modeling

For the puurpose of reguulator synthessis, rectifier’ss and
inverter’s innfluence on the DC link are modeledd as
disturbance current. Succh circuit is shown
s in Figg. 4.
This currentt acts as a distturbance on DC
D link voltaage.
There aree several models
m of SC
C ranging from
f Fig.. 4 KERS circuit with inverrter and rectifiier modeled ass
non-linear model
m [16], RC
R network model [17] and currrent disturban
nce.
180 C
covery System
During
D operaation of intterest, diode rectifier iss
inveersely polariized, and disturbance cu urrent equalss
inveerters currentt on DC sidee. Measuring g disturbancee
currrent is not practicable.
p Current senssor increasess
pricce and hardw ware complexxity. Measureed current iss
pulssed and thereefore noisy. N No matter wh hich filteringg
techhnique is utiilized, noise remains and d jeopardizess
reguulator’s stability, hence inncreasing SC’s current andd
volttage ripple annd decreasingg expected liffetime of SC.
Disturbance
D c
current could be estimated from Eq. (2))
Fig. 5 KERS S circuit with inverter
i and rectifier modeleed as writtten as Eq. (3). Pulsed noise is significaantly reducedd
current distu
urbance and SC S modeled as a constant volltage and
d low noise siignal is attainned with sim mple low passs
source.
digiital filter.
y u*—coontrol signal representingg duty ratioo of
· (3))
PWM controolled transistoors; u*=1—uupper switch is
i on
(T1 on Fig. 2);
2 u*=0—low wer switch is on (T2 on Figg. 2); Final
F form of o disturbannce current estimator iss
u* [0,1] andd u* R obtaained by applying digital low pass filtter on Eq. (3))
with
h sampling period
p T coincciding sampliing period off
(1)
regu
ulator and baandwidth freqquency ω1, written
w as Eq..
(4). Control signnal u* is delayyed one samplling period too
(2)
obtaain causal esttimator.
This moddel describes dynamic of inductor and DC 1
· · ·
link. SC dynnamic is negllected on purrpose. Modell has 1
four parameeters. SC voltage E and dissturbance currrent
(4))
Id are measuurable. Inducctance L is stable
s parameter.
Parameter C is aging sensitive
s paraameter. DC link
5. State Trransformattion and Feedback
k
usually coonsists of electrolytic capacitor. Its
Lin
nearization
n
capacitance degrades in time
t and decrreases robusttness
of regulator.. Model
M of KERRS unit descrribed with Eq qs. (1) and (2))
is nonlinear.
n Because of veery wide ran nge of SC’ss
4. Disturb
bance Curreent Id Estim
mation
volttage and cuurrent and bidirectional power flow,,
High perrformance position
p conntrolled systtems lineear regulatorr will have regimes with increasedd
usually conssist of inner toorque controll loop. Positioon is currrent and vooltage ripple. For this reason, somee
controlled by
b applying maximum
m torque for brakking authhors use moddified linear controllers [8, [ 9], fuzzyy
and acceleraation. In such systems, inverter’s currennt on logiic regulators [10, 11], neuural networks [12, 13] orr
DC side cann change from m maximum positive
p currennt to flatnness and passsivity controll theories [14,, 15].
maximum negative
n curreent in 0.5 mss [18]. Thereffore, This
T paper prroposes feedbback linearization controll
influence off disturbance current is crrucial and sim mple metthod. Regulattor synthesis consists of th hree steps:
feedback DC C link voltage regulator iss not sufficiennt to y state transfformation too utilize non n-linear statee
stabilize voltage during intermittent mode of motor m space system in controllable ccanonic form m;
drive. y feedback linnearization of transformed d states;
C
covery System 181
y linear regulation
r of linearized
l staates. shows second trransformed sttate z2 is sum
m of power off
States i annd uc can be written
w in diffferent notationn, as all sources
s and drains
d in KER
RS model.
in Eq. (5). Now
N system coould be writteen as Eqs. (6),, (7), System
S withh partially transformed d states inn
(8) in vectorr form. conntrollable cannonic form iss given with Eqs. (9) andd
(17).
=
X= (5)
(17))
(6) By
B defining control
c signall u* as Eq. (18), system iss
feed
dback linearrized from nnew control input v too
nsformed outpput states z1 and z2 and target
tran t system
m
(7) from
m Eqs. (9) annd (10) is achiieved.
(18))
· (8)
System inn Eq. (8) is not in controllabble canonic foorm.
6. Linear
L Reggulator
State transfformed and feedback linearized
l taarget
system in coontrollable caanonic form is i written as Eqs. Estimated
E dissturbance currrent Id and measured
m SC
C
(9) and (10) with transforrmed states z1 and z2. volttage E are notn constant in time. Eaach samplingg
(9) periiod Id and E are
a updated aand feedback linearizationn
(10) andd state transfformation ouutput are alteered. With Id
Let’s noww introduce sttate z1 in funnction of statees x1 havving ability to rapidly change from m maximum m
and x2 writteen as Eq. (111). First Lie derivative off Eq. positive to maxim mum negativve value in 0.5 5 ms, processs
(11) is writteen as Eq. (12). of updated feedbackf liinearization and statee
(11) tran
nsformation acts as a feedforward d action onn
· (12) distturbance supppression of DC link vo oltage. Withh
Equalizing Eqs. (9) annd (12) we haave Eqs. (13) and feeddforward acttion, linear rregulator can n have singlee
(14). inteegral action. Proposed
P reguulator is showwn in Fig. 6. Itt
commprises of innner PI regulattor of state z2 and outer P
0 (13) reguulator of statee z1.
Control
C objecct is doublee integrator. The controll
prob blem is anaalogous to poosition contrrol of motorr
(14) drivves, thus simmilar controllling techniqu ues could bee
used. In Ref. [119] optimal P PID position controller iss
Solving partial differenntial Eq. (13) and later Eq.. (14) giveen. This papeer proposes ooptimal PPI reegulator withh
leads to transformed states system in controlllable mod deled compuutational delaay and same optimizationn
canonic formm. State transsformation iss given with Eqs. metthod as in Ref. [19].
(15) and (166). To
T obtain opptimal discrette PPI regullator, controll
objeect must be discretized.
d Fiirst integratorr is written inn
(15)
(16)
Eq. (15) shows
s first trransformed state
s z1 is sum
m of
energies of reactive com mponents of KERS. Eq. (16) Fig.. 6 Proposed PPI regulator of linearized KERS
K .
182 C
covery System
integral form
m as Eq. (19).. ; ; (25))
(19)
(26))
PWM conntrolled conveerters are usuually switchedd via
IGBT transiistors with freequencies in range of 10 kHz Eq.
E (27) deffines criterioon function as sum off
and higher. With such high h frequenccies, DSP brrings reguulator’s errorr samples. Iff closed loop p system hass
computationnal delay of one saampling perriod. polees in segmennt zpi [0, 1] annd no zeros, step
s responsee
Computationnal delay alteers closed loopp dynamics annd it willl be strictly aperiodic
a witth no oversho oot response..
has to be takken into accouunt. Under asssumption conntrol Theen by minimiizing criterionn function, faastest strictlyy
signal is constant
c durring one sw witching perriod, apeeriodic responnse is obtained.
discretized Eq.
E (19) with modeled com mputational delay
d ∑∞ ∆ (27))
is written as Eq. (20). Eqq. (20) is writtten in “z” dom
main Minimizing
M crriterion functtion [16] results in optimall
as Eq. (21). reguulator gains written
w as Eq.. (28).
0.0566640625
1 (20)
0.001953125
0.31640625 (28))
(21)
Final
F regulatoor gains in Eq. (29) aree five timess
Second inntegrator is written
w in inteegral form ass Eq. smaaller to obtainn robust regullator.
(22). Discrettized Eq. (22) is written ass Eq. (23). Unnder
; ; ; (29))
assumption state z2 chhanges lineaarly during one
switching peeriod, Eq. (233) is written in “z” domaiin as To
T obtain refeerence z1*, refference of DC
C link voltagee
Eq. (24). uc* has to be non-linearly
n transformed to space off
tran
nsformed stattes z1 and z2. This transsformation iss
(22)
writtten as Eq. (330).
1 (23) (30))
To
T limit curreent of SC, refeerence z2* of inner controll
(24)
loopp has to bee limited. Cuurrent limits have to bee
Closed loop
l block diagram with w discrettized tran
nsformed to space of trannsformed staate z1 and z2
linearized KERS
K model and discrete PPI regulatoor is writtten as Eqs. (331) and (32). When SC vo oltage reachess
shown in Fiig. 7. Proporrtional actionn in inner looop is max ximum or miinimum voltaage, Imax valu ue is changedd
displaced inn feedback branch to obtain o zero free as previously
p disscussed.
transfer funcction. Regulattor gains are normalized inn Eq. (31))
(25). Transffer function inn “z” domainn with normallized (32))
gains is writtten as Eq. (266). Anti-windup
A m
mechanism iss shown in Fiig. 7. Controll
sign
nal is limitedd u* [0, 1]. T To limit conttrol signal v,,
conntrol signal u** limits have tto be transforrmed to spacee
of states
s z1 and z2, written as Eqs. (33) and d (34).
(33))
Fig. 7 Block diagram of regulator with anti-win ndup
mechanism an
nd displaced proportional
p acction of inner loop.
l (34))
Control Algorithm for Non-isolated Supercapacitor Based Kinetic Energy Recovery System 183
7. Numerical Simulations In the second simulation, shown in Fig. 10,

robustness of regulator is tested. DC link has 10%
Computer numerical simulations were conducted in
smaller capacitance than capacitance parameter in
Matlab Simscape software package. Circuit from Fig. 2
regulator, thus simulating aging of device. KERS
was simulated for various torque profiles. Parameters
operates similar as in Fig. 8, but with slightly increased
of model are:
damped oscillations of SC current. At the end DC link
y L = 10 mH—inductor value;
working life, when capacitance drops 20%, SC current
y R = 37 mΩ—inductor resistance;
chatters and device operates with great losses.
y CDC_Link = 1,500 μF—DC link capacitance;
Third simulation, shown in Fig. 11, simulates
y ESRDC_Link = 50 mΩ—equivalent series
discharging of SC to minimum voltage, when drive is
resistance of DC link capacitor;
in motoring regime. When SC is discharged to
y CSup.Cap = 4 F—supercapacitor capacitance;
minimum voltage, SC current limits in regulator are
y RSup.Cap = 85 mΩ—supercapacitor internal
altered to prevent further discharge. At this moment,
resistance;
regulator is not able to keep DC link voltage at 600 V
y fpwm = 10 kHz—IGBT switching frequency;
y DeadTime =3 μs—–IGBT dead time;
y fs = 20 kHz—sampling frequency of regulator;
y Imax = 50 A—maximum current of inductor and
SC;
y USCmax = 200 V—SC maximum voltage;
y USCmin = 120 V—SC minimum voltage;
y Iinverter_PEAK = 13 A—peak value of inverter’s
output current;
y ω1 = 600 rad/s—low pass filter bandwidth in
disturbance current estimator.
Results of the first simulation are shown on Fig. 8.
First graph is inverter’s current on DC side. Blue line is
actual current and red line is estimated current. Second Fig. 8 Results of first simulation.
graph is DC link voltage. Third graph is SC voltage.
Fourth graph is SC current. At first moment, inverter’s
current is positive, simulating motoring regime of drive
and SC initial voltage is 150 V. When t = 1 s, drive
suddenly changes to generating mode, simulating
braking. Inverter’s current on DC side changes from
maximum positive to maximum negative in 1 ms.
Regulator suppresses disturbance of DC link voltage
with overshoot of 1.6% in less than 16ms. This moment
is shown in Fig. 9. When t = 2 s, drive changes back to
motoring mode in similar manner. When t = 3 s, KERS
is simulated in regime where inverter’s current on DC
side is changing in ramp. Inductor current ripple is less Fig. 9 DC link voltage disturbance when motor drive
than 2% and DC link voltage ripple is less than 0.5%. suddenly brakes.
184 Control Algorithm for Non-isolated Supercapacitor Based Kinetic Energy Recovery System
Fig. 10 Results of second simulation. DC link capacitance

is reduced by 10%, simulating aging of device and testing
robustness.
Fig. 12 Results of fourth simulation. SC charges to
maximum voltage.
turned on. Later, motor drive changes to motor regime

and DC link voltage stabilizes.
8. Conclusion
This paper presents a non-linear control algorithm
for supercapacitor based non-isolated kinetic energy
recovery system. It fulfils all KERS functionalities:
y Charging SC while motor drive is braking;
y Discharging SC while motor drive is
accelerating;
y Limiting SC current and voltage;
y Stabilizing DC link voltage and suppressing
sudden disturbances.
Fig. 11 Results of third simulation. SC discharges to SC current ripple is less than 2% and DC link
minimum voltage. Later, at t = 2 s, motor drive brakes.
voltage ripple is less than 0.5%. Control algorithm
reference. DC link voltage drops and diode rectifier relies only on two parameters; inductance of inductor
turns on. At this moment KERS operates with and capacitance of DC link. SC capacitance is not
increased current ripple and transistors should be parameter of regulator. Hence SC aging does not
turned off. deteriorate performance of KERS. Regulator is able
In fourth simulation, overcharging of the SC is to suppress influence of rapidly changing torque
simulated. Results of fourth simulation are shown in on DC link voltage, relying on estimation of
Fig. 12. When SC voltage reaches maximum value, inverter’s current on DC side. Control algorithm relies
current limits in regulator are changed to prevent only on its own measurement, thus enabling KERS
further charging of SC. As a consequence, DC unit to be standalone device independent of motor
link voltage increases. At this moment DBR should be drive.
Control Algorithm for Non-isolated Supercapacitor Based Kinetic Energy Recovery System 185
References [11] Jabbour, N., and Mademlis, C. 2017.

“Supercapacitor-based Energy Recovery System with
[1] Mittal, A. K., and Kumar, J. M. 2011. “Electrochemical Improved Power Control and Energy Management for
Double-Layer Capacitors Featuring Carbon Nanotubes.” Elevator Applications.” IEEE Trans. Power Electron 32
American Scientific Publishers, Encyclopedia of (12): 9389-99.
Nanoscience and Nanotechnology 13: 263-71. [12] Ortuzar, M., Moreno, J., and Dixon, J. 2007.
[2] Conway, B. E. 2009. Electrochemical Supercapacitor “Ultracapacitor-based Auxiliary Energy System for an
Scientific Fundamentals and Technological Application. Electric Vehicle: Implementation and Evaluation.” IEEE
New York: Kluwer Academic/Plenum. Trans. on Industrial Electronics 54 (4).
[3] Beguin, F., and Frackowiak, E. 2001. “Carbon Materials [13] Moreno, J., Ortuzar, M. E., and Dixon, J. W. 2006.
for the Electrochemical Storage of Energy in Capacitors.” “Energy-Management System for a Hybrid Electric
Carbon 39: 937. Vehicle, Using Ultracapacitors and Neural Networks.”
[4] Francois, B. Raymundo-Pinjero, E., and Frackowiak, E. IEEE transactions on Industrial Electronics 53 (2).
2009. “Electrical Double-Layer Capacitors and [14] Payman, A., Pierfederici, S., and Meibody-Tabar, F. 2007.
Pseudocapacitors.” Carbons for Electrochemical Energy “Implementation of a Flatness Based Control for a Fuel
Storage and Conversion Systems, Chapter 8. CRC Press, Cell/Ultracapacitor Hybrid System.” Power Electronics
329-75. Specialists Conference, IEEE.
[5] Goalous, H., Louahlia, H., and Gallay, R. 2011. [15] Becherif, M., Ayad, M. Y., and Miraoui A. 2006.
“Supercapacitor Characterization and Thermal Modelling “Modeling and Passivity-based Control of Hybrid Sources:
with Reversible and Irreversible Heat Effect.” IEEE Fuel Cell and Supercapacitors.” Industry Applications
Transaction on Power Electronics 26 (11). Conference, 2006. 41st Annual Meeting. Conference
[6] Maxwell 160V Module datasheet. Record of the 2006 IEEE.
[7] Vukosavić S. N. “Electrical Machines.” Power [16] Xu, N., and Riley, J. 2011. “Nonlinear Analysis of a
Electronics and Power System, Springer. Classical System: The Double Layer Capacitor.” Elsevier
[8] Kim, S.-M., and Sul, S.-K. 2006. “Control of Rubber B.V. Electrochemistry Communication. doi: 13,
Tyred Gantry Crane with Energy Storage Based on 10.1016/j.elecom.2011.07.003, 2011.
Supercapacitor Bank.” IEEE Transactions on Power [17] Zubieta, L., and Bonert, R. 2000. “Characterization of
Electronics 21 (5). Double-Layer Capacitors for Power Electronics Applicatins.”
[9] Azib, T., Bethoux, O., Remy, G., Marchand, C., and IEEE Transactions on Industry Applications 6 (1).
Berthelot, E. 2010. “An Innovative Control Strategy of a [18] Vukosavić, S. N., Perić, L. S., and Levi, E. 2016. “A
Single Converter for Hybrid Fuell Cell/Supercapacitor Three-Phase Digital Current Controller with Improved
Power Source.” IEEE Trans. on Ind. El. 57 (12). Performance Indices.” IEEE Transaction on Energy
[10] Jabbour, N., and Mademlis, C. 2016. “Improved Control Conversion 32 (1): 184-93.
Strategy of a Supercapacitor Based Energy Recovery [19] Vukosavić, S. N. 2007. “Digital Control of Electrical
System for Elevator Applications.” IEEE Trans. Power Drives.” Power Electronics and Power System. New York:
Electron 31 (12): 8398-408. Springer.
doi: 10.17265/2328-2223/2018.03.007
D DAVID PUBLISHING
An Intelligent Robot with Infinity Controls Making Life

Easier for People with Disabilities and Aging
Lara. Al Chamy, Mohamad Saad El Dine, Abd. El Kader El Rafeii and Ahmad. El Rafhi
Université de technologie et de sciences appliquées Libano-française, Route de Deddeh, EL-KOURA, Liban
Abstract: The evolution of technology, affecting all fields, requires artificial intelligence. In this article, this intelligence manifests
itself in the creation of a system that can be used as a solution for different problems in different domains. For this purpose, we have
created a control table with a robot brought by a control weapon via bluetooth. The control of the robot is based on the detection of the
simple red color of the laser and the weapon is controlled according to the detection of the finger touch.
Key words: Baro 4WD, touchscreen, BLE Link—a bluetooth 4.0 module, folding webcam, visual studio, image processing.
1. Introduction severe the handicap is, more complex are the devices,
implying increased efforts to simplify the interactions
From the 1950s, robotics developed thanks to the
between man and these devices [5].
rise of computer science and advances in artificial
Artificial intelligence is the intelligence exhibited by
intelligence research. Machines controlled by
machines or software. It is the subfield of computer
computers, more and more autonomous and able to
science. Artificial intelligence is becoming a popular
determine the behavior to adopt facing an event, are
field in computer science as it has enhanced the human
born [1].
life in many areas [6]. As name implies artificial
Now robots are going to develop which looks like
intelligence is making the machines like human that
human and scientist are trying to put emotions in them
can think like human and can do the work like human
[2].
[2].
Robots are poised to fill a growing number of roles
Application areas of artificial intelligence is having a
in today’s society, from factory automation to service
huge impact on various fields of life as expert system is
applications to medical care and entertainment [3].
widely used these days to solve the complex problems
The relation of man to the robot has always been
in various areas as science, engineering, business,
ambiguous: Man creates robots more and more
medicine, weather forecasting [6] fault diagnosis,
advanced and autonomous to relieve themselves of
automatic configuration, artificial vision and pattern
tedious tasks, repetitive or dangerous [1].
recognition, speech and language comprehension,
The successful introduction of robotics into human
supervision of industrial production systems, automatic
environments will reduce fatigue, increase precision,
planning of space missions.
and improve quality [4] and will rely on the
The areas employing the technology of Artificial
development of competent and practical systems that
Intelligence have seen an increase in the quality and
are dependable, safe, and easy to use [4].
efficiency [6].
Assistance to disabled people is still a domain in
There are many AI research challenges: providing
which a lot of progress needs to be done. The more
semantic Web technologies; extract knowledge from

Corresponding author: Mohamad Saad El Dine, Dr., large masses of data; realize artificial systems that
research fields: electronics, electronics, ronotics, image
processing, control system. adapt to a changing environment; realize systems
An intelligent robot with infinity controls making life easier for people with disabilities and aging 187
where intelligence is distributed among many artificial shield “BLE Link—A Bluetooth 4.0 module”, Lipo
agents that interact; to be able to reason about time and 2200mAh Battery, the weapon formed by a Pan and
space; develop robots that are autonomous or able to Tilt Kit, two Servos and Robot Gripper-Extra Width
assist the human being in his tasks, alone or in a (Fig. 2).
community; and above all, to scale, to confront the real  Our table design with a Folding Webcam. The
problems, the data arriving continuously and in table is considered like a screen touch.
quantity [7]. It is made simply of aluminum with brown glass
The importance of embedded applications on image plate (Figs.1a and 1b).
and video processing-used for faster
3. Components of Touchscreen
performance-communication and cryptography
domain has been taking a larger space in current A basic touchscreen has three main components: a
research era [8]. A few years ago touchscreens were
futuristic and only used in industrial applications.
Today we do not even think twice before tapping or
swapping a touchscreen around us [9]. A touchscreen is
any monitor, based either on LCD (liquid crystal
display) or CRT (cathode ray tube) technology that
accepts direct onscreen input. The ability for direct
onscreen input is facilitated by an external (light pen)
or an internal device (touch overlay and controller) that
relays the X, Y coordinates to the computer [10]. (a)
Touchscreen technology has the potential to replace
most functions of the mouse and keyboard. The
touchscreen interface is being used in a wide variety of
applications to improve humancomputer interaction.
As the technology advances, people may be able to
operate computers without mice and keyboards [10].
It’s easier for people to interact with a device using
their finger than it is to point a mouse to the right
location [9]. The touchscreen is an assistive technology.
(b)
This interface can be beneficial to those that have Fig. 1 Table design.
difficulty in using other input devices such as a mouse
or keyboard.
The ability to transform any surface in a touchscreen
means lower costs, making the technology more cost
effective [10].
2. Robot and Table Design

The project contains two parts:
 The robot “Barron 4WD”, including: the Arduino
“Romeo V2-All in one Controller”, the Bluetooth Fig. 2 Robot design.
188 An intelligent robot with infinity controls making life easier for people with disabilities and aging
touch sensor, a controller, and a software driver. The Our touch screen is a table of aluminum (high: 35 cm,
touchscreen is an input device, so it needs to be brown glass plate: length: 25cm, width: 19.2cm).
combined with a display and a PC or other device to We know that all new touch screens are made of
make a complete touch input system [10]. different materials and small devices very sensitive,
(1) Touch Sensor which makes their maintenance very difficult as well as
A touch screen sensor is a clear glass panel with a sometimes impossible.
touch responsive surface. The touch sensor/panel is If we compare our touch screen with one of the most
placed over a display screen so that the responsive area simple screen, “a simple low cost Touchscreen to
of the panel covers the viewable area of the video Control Home Automation System” [11].
screen. The sensor generally has an electrical current or Their project needs 3 components:
signal going through it and touching the screen causes a (1) Touchscreen device
voltage or signal change. This voltage change is used to In this section we will describe the design of the
determine the location of the touch to the screen [10]. touchscreen. At first, some laser beams are arranged
(2) Controller closely with vertical axis and on the opposite site of the
The controller is a small PC card that connects same axis some photo resistors are placed with the line
between the touch sensor and the PC. It takes of sight of the laser beams. Photo resistors act as
information from the touch sensor and translates it into sensors. Now the same things are done with horizontal
information that PC can understand. The controller is axis. Fig. 4 shows the diagram of the proposed
usually installed inside the monitor for integrated touchscreen.
monitors or it is housed in a plastic case for external (2) Required Software Development
touch addons/overlays. The controller determines what Software components are required to receive data
type of interface/connection you will need on the PC. from the touchscreen and process the data to extract
Controllers are available that can connect to a
Serial/COM port (PC) or to a USB port (PC or
Macintosh). Specialized controllers are also available
that work with DVD players and other devices [10].
(3) Software Driver
The driver is a software update for the PC system
that allows the touch screen and computer to work
together. It tells the computer’s operating system how
to interpret the touch event information that is sent Fig. 3 Components of touch screen.
from the controller. Some equipment such as thin client
terminals, DVD players, and specialized computer
systems either do not use software drivers or they have
their own built-in touch screen driver [10].
In our project, we do not use any touch sensor. The
detection of touch is made by image capture via camera.
The system does not contain any controller. We only
use a folding webcam connecting via serial port to our
pc. And the software used in our project is the VISUAL
STUDIO. Fig. 4 Description of touch screen design.
information relevant to determine whether a touch

occurs or not. The information is extracted to
determine the location of the touch and make decision
to send command to the output component. Fig. 5
summarizes the general design of the software
component.
(3) Interfacing of output circuit
Output circuits are connected with the desired
electrical devices. To design the output circuit we have
used relay switch. A relay is an electrical switch that
opens and closes under the control of another electrical
circuit. In the original form, the switch is operated by
an electromagnet to open or close one or many sets of
contacts. The interfacing circuit for four electrical
devices is shown in e Fig. 6.
Many such projects have been created. In each new
innovation we wish to achieve our goals. Different
advantages characterize our project.
Fig. 6 Interfacing circuit for four electrical devices.
In the case of normal screens, it is necessary to
replace the deteriorated devices. This operation is very the event of breakage or a replacement of the broken
complicated and requires a lot of precision. In our camera.
project, we are far from these problems. As well as the realization of table is very easy and
Our project does not require a complicated simple contrary to the normal screens. They require
maintenance. It is sufficient to change the glass plate in very thin and precise welding because imprecision
causes operational problems.
This table considered as a screen touch,
economically does not cost expensive. Even if we
compare it by the cheapest touch in the world,
(Momomo Ebook reader smart with card slot Extended
7inch Touch Screen digital E-book+recording
+Video+MP3 music Wifi player (41.90$), 12 Inch
Resistive Touch Screen Monitor with HDMI DVI USB
(98.0$), 21.5" Pcap Open Frame Touch Monitor, Kiosk
Touch Screen Monitor(300$)) it costs only 15$ and
with low weight.
On the other hand, the remote control is simply a
laser. It is not a professional laser. It is one that we can
buy at 3$. You can never find a remote control at this
price, easy to use, and does not need maintenance. Note
that it is easier to use a laser than using any other
remote control like a mouse for example to point to the
Fig. 5 Design of software component. right location, especially for the handicaps.
And we know that colors are very varied. We use A single entry is used to control the weapon. Our
this variety to maximize the number of input. Infinity finger touch is also the only input.
of commands using simple input, it is one of our goals. According to the finger position on the suitable area,
We need this simplicity especially for the disabled the weapon performs four functions: opening, closing,
and those suffering from mental and physical up and down (Fig. 8).
problems.
5. Design in Visual Studio
4. Description of the System
It was important to create an easy-to-use system
In our project, we control 2 parts: the robot and the especially for people with mental or physical
weapon. This control is done with the table considered disabilities. In this case, we approve that our project
like a touch screen. eliminates the problem of disability and it does not
It is divided into two areas, the first one to control demand a lot of effort to be applied.
the robot via the red color of the laser, and the second The system is as following:
one to control the weapon via the finger. In the “Form.cs [Design]” window, we created our
When we detect the laser red color, then we identify model:
the direction of the robot.  A serialPort1 to set the bluetooth shield.
The position of the red color (value of x and y)  A textbox containing the “COM” accessible,
indicates the direction to be follow (dx > 0 to the right, below which a label “Puerto serial”.
dx < 0 to the left, dy > 0 to the front, dy < 0 to  A textbox containing the “9600” transmission
backword). speed, below which a “Baudrate” label.
The input is only this red color. But the robot  A “connect” button to make the bluetooth
commands are multiples: accessible (serial-Port1.Open ()).
Forward, backward, to right, to left, forward to the  Four buttons for the different directions of the
right, forward to the left, backward to the right and robot, controlled via bluetooth, and a stop button that
backward to the left (Fig. 7). stops any command.
 A pictureBox1, of size 320; 240. And a
WebCamCapture of height 240, width 320 and
TimeToCapture 100.
In addition, two buttons:
 a “Start” button: which allows the beginning of
the capture of the video.
 a “Stop” button: which allows the capture of the
video to stop.
Fig. 7 Detection of red color.
The system starts capturing the video. It was now
necessary to detect the red color of the laser. For this
purpose, we created a bit-map, named “b”. In this
bitmap, the color detection was done according to RGB
(the three colors: red, green and blue (their values from
0 to 255)).
The programming for the robot to be controlled
Fig. 8 Detection of finger print. according to the red color of the laser starts from here.
First of all, it was necessary that each time we move

the laser, we specify the coordinates of the detected
point.
Then, we added a “textbox”, in which it was
specified x and y, each time we clicked the button
“test-filter”, with a button “clear” which allows each
time to delete the coordinates.
We specify the margin of each of the small areas
considered for the weapon on the table by detailing the
values of i and j of each.
We insisted on the white color of this area (r = 255, g
Fig. 9 Visual studio design.
= 255, b = 255), in which we could easily detect the
pixels of our fingers. A small “textbox” allowed us to
detect the presence of a key or its absence, on the same
“timer” of the webcam, with 3 buttons: “start”, for the
initialization of the timer, “stop” for his stop and
“clear1” to empty the contents of the textbox.
6. Process and Operation

Operation begins with an image capture. Necessarily,
one applies a normalization of luminosity and a
filtering of the colors in order to get rid of the parasites.
It needs detection, recording, and then decision of
order according to the obtained values. Fig. 10 System process.
Let us add here that the red color of the laser is only
one of the various inputs that can be used.
For example, a green color is detected on the same
detection zone. It works as follows: by drawing a
complete turn on the table by the green laser, the robot
in this case actually turns a complete turn.
This value will be saved in the databases and then
read to be ordered.
Now it’s a capture to area of the key. The correlation
with the previous image indicates the command to be
executed.
From here, a different command can apply.
Other than the finger touch, one can enjoy the
application of different colors on the fingers to vary the
controls.
For example, the blue color applied to one of the
fingers in one of the specified areas means that the Fig. 11 Filter system.
weapon must rotate 180°. Biblioghraphie sélective.

[2] International Conference on Recent Innovations in
We have four areas on the table. These four areas
Science, Engineering and Management YMCA, New
used for the simple touch of the finger, different colors Delhi, 30th May, 2015 ISBN: 978-81-931039-4-4
and different areas in parallel with different colors. ARTIFICIAL INTELLIGENCE IN MODERN TIME
Ekta Nehra Student of C R college of Education, Hisar,
7. Treatment and Filtration Haryana-125001, (India).
[3] Human-Robot Interaction, David Feil-Seifer, Maja J
One of the biggest problems was the parasites caused Matari´c.
by the light. [4] Robots in Human Environments O. Khatib, K. Yokoi, O.
The use of different types of filter, improves the Brock, K. Chang, and A. Casal Robotics Laboratory.
Department of Computer Science Stanford University,
detection and the color distinction necessary to use to
Stanford, California 94305.
control the robot. Such a treatment ensures a good [5] Intuitive human interaction with an arm robot for severely
quality/function essential to realize the concept. And to handicapped people - A One Click Approach Claire Dune,
have great results, it is still necessary to eliminate the Christophe Leroux, Eric Marchand.
parasites caused by the reflection camera. So we use [6] International Journal of Engineering and Innovative
Technology (IJEIT) Volume 4, Issue 10, April 2015 79
insulating layers.
Artificial Intelligence and its Application in Different
Areas Avneet Pannu, M. Tech Student Department of
8. Conclusions
Computer Science & Engineering DAV Institute of
End of our project, we have a robot ensuring a good Engineering and Technology, Jalandhar India.
[7] Intelligence Artificielle Et Robotique "Confluences de
report, quality/cost. We enjoy a simple entry “the color”
l'homme et des STIC" Les cahiers de l'ANR-n04mars
and a simple “touch of the finger” to order and 2012.
remotely a robot uses to facilitate life especially for [8] A brief experience on journey through hardware
handicaps. Solving the problem of disability was one of developments for image processing and its applications on
Cryptography.
the goals. All this is in a simple, enforceable, secure
[9] How Do Touchscreens Work?? A complete overview of
system. the most modern touch screen technologies.
Such a project is used in different fields. It can be [10] Comparative Study of Various Touchscreen Technologies,
used as a service for handicaps; it makes their daily life Mudit Ratana Bhalla, Dept. of Computer Science and
Applications, Dr.H.S.Gour Central University, Sagar
easy by staying in their place.
(M.P.)/ Anand Vardhan Bhalla, Dept. of Electronics and
It can be used to shorten the time needed to perform Telecommunication, B.T.I.R.T Engineering College Sagar
multiple tasks at the same time, especially in large (M.P.).
factories where the number of workers can be reduced. [11] Design and Development of a Simple Low-cost
And by replacing the camera with a satellite, we can Touchscreen to Control Home Automation System
M. Muntasir Rahman, Md. Aktaruzzaman, Computer
control the robot from one place to another.
Science & Engineering, Islamic University, Bangladesh
muntasir@cse.iu.ac.bd, mazaman@cse.iu.ac.bd Mst.
References
Ashrafunnahar Hena3
[1] Bibliothèque nationale de France, direction des collections, Applied Physics, Electronics & Communication
département Sciences et techniques, ROBOTS, Engineering, Islamic University, Bangladesh
ROBOTIQUES ET INTELLIGENCE ARTIFICIELLE, hena@aece.iu.ac.bd.

JEE-Vol 6, No 3,2018 PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

JEE-Vol 6, No 3,2018 PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Journal of Electrical

David Publishing Company

Aims and Scope:

Editorial Board Members:

David Publishing Company

David Publishing Company

151 A Mixed-Mode Biquad Employing OTAs and Grounded Capacitors

156 A Simple Traffic Model Including Possibility of Overtaking

Techniques and Methods

Availability of Lithium Ion Batteries from Hybrid and

information about the age and size of the stock in the

Fig. 2 Newly registered HEV in Germany 2010-2030.

Fig. 3 Newly registered PHEV in Germany 2010-2030.

Fig. 4 Newly registered BEV in Germany 2010-2030.

Fig. 5 Newly registered EV in Germany 2010-2030.

Weibull AVG 1980

Fig. 6 Regisstration rate deepending on age of the cars (depending on

deregister while step accident while step

Fig. 7 Available batteries from accidentss for HEV 2010-2030.

Fig. 8 Available batteries from end-of-life for HEV 2010-2030.

Fig. 9 Available batteries from accidentss for PHEV 20010-2030.

Fig. 10 Avaiilable batteriess from end-of-llife for PHEV 2010-2030.

Fig. 11 Avaiilable batteriess from acciden

Fig. 12 Available batteries from end-of-life for BEV 2010-2030.

Fig. 13 Available batteries from accidents for EV 2010-2030.

Fig. 14 Available batteries from end-of-life for EV 2010-2030.

The Study of Demagnetization of the Magnetic

Noriyoshi Nishiyama1, Hiroki Uemura2 and Yukio Honda2

1. Introduction For obliquely distributed magnets with an oblique

arraangement anggle Va while also changing the magnett

3. Magneet Length and Dem

demagnetizaation improvement ratio is improved,, the Tab

5. Magneet Length and Dem magnetizattion

Two typees of magnetss have the folllowing magnnetic

A Mixed-Mode Biquad Employing OTAs and Grounded

Takao Tsukutani1, Yasutomo Kinugasa1 and Noburu Yabuki 2

Key words: Active circuit, mixed-mode operation, biquad characteristics, OTAs.

1. Introduction above-mentioned performances has not yet been

input terminals suitably.

3. Circuit Configuration and Analysis

Ni (s) = g m3 N v (s) (6)

It is found from the equations above that the circuit

(a) Fig. 6 Q-tuning responses (current-mode circuit).

5. Conclusions characteristics. The solution on this will be discussed

A Simple Traffic Model Including Possibility of

Key words: Cellular automata, traffic flux, overtaking.

simpler. which are used in our calculations are implemented.

Fig. 6 The values of q:pe as a function of pc-skip.

References [6] Nasaab, K., Schreckenberg, M., Ouaskit, S., and

Protection Coordination Optimization for FREEDM

Mohamed F. Kotb1, Magdi El-Saadawi2 and Eman H. El-Desouky2

Key words: FREEDM, protection coordination, genetic algorithm, ETAP software.

Nomenclature Minimum pickup current of relay i

Fig. 1 FREEDM system topology and components [2].

used [7]. In addition, its convergence is another 3. Proposed Conventional Protection

Fig. 2 Proposed locations and directions of overcurrent relays in FREEDM system.

Generate primary/pickup pairs

Perform load flow study

Perform faulty study

Setting of overcurrent relays

Perform optimal selection of time dial setting using GA

Record outputs setting

Chromosome population Initialization