Fuzzy Control of an Electrostatic Separation Process

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 44, NO. 1, JANUARY/FEBRUARY 2008

9

Fuzzy Control of an Electrostatic Separation Process Mohamed Younes, Amar Tilmatine, Karim Medles, and Lucian Dascalescu, Senior Member, IEEE

Abstract—Process control is one of the most promising fields of application of fuzzy systems theory. This paper aims at analyzing the particularities of fuzzy logic as applied to the control of a standard insulation–metal electrostatic separator. The optimum operation of such an installation requires the adjustment of a control varaible: the high voltage applied to its electrode system, so as to maximize the quantities of high-purity insulation and metal products collected after separation from the input granular mixture. The input and the output variables were fuzzified in accordance with the authors’ expertise in the field of electrostatic separation. A matrix correlated the applied high voltage to the state of the output variables. The control algorithm was tested by numerical simulation, using the mathematical model established with the design of experiments methodology. Index Terms—Electrostatic processes, electrostatic separation, fuzzy controller, fuzzy logic. Fig. 1. Industrial roll-type corona-electrostatic separator. (1) Feeder. (2) Grounded rotating roll electrode. (3) High-voltage electrodes.

I. INTRODUCTION HE AVAILABILITY of a sound mathematical model is a prerequisite for the implementation of traditional control systems. Therefore, the modeling of electrostatic separation processes has been an active research topic during the recent years [1], [2]. One of the main conclusions of the various studies is that these processes depend on a large number of variables [3], not all of them controllable [4], and that each application is characterized by a different model [5]. Any traditional approach to the control of such processes would fail, especially if also taking also into account the complexity of “online” monitoring of the output variables (mass and/or purity of the separated products).

T

Paper MSDAD-07-05, presented at the 2006 ESA/IEEE/IEJ/SFE Joint Conference on Electrostatics, Berkeley, CA, June 20–23, and approved for publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Electrostatic Processes Committee of the IEEE Industry Applications Society. Manuscript submitted for review June 30, 2006 and released for publication June 12, 2007. This work was supported by the French and Algerian Governments within the framework of a TASSILI project. M. Younes and K. Medles were with the Electronics and Electrostatics Research Unit, Laboratoire d’Automatique et d’Informatique IndustrielleEcole Sup´erieure d’Ing´enieurs de Poitiers (LAII-ESIP), University of Poitiers, University Institute of Technology at Angoulˆeme, 16021 Angoulˆeme Cedex, France. They are now with the Electrostatics and High Voltage Research Unit, Interaction R´eseaux Electriques-Convertisseurs-Machines (IRECOM), University of Djillali Liabes, 22000 Sidi-bel-Abb`es, Algeria (e-mail: [email protected]; [email protected]). A. Tilmatine is with the Electrostatics and High Voltage Research Unit, Interaction R´eseaux Electriques-Convertisseurs-Machines (IRECOM), University of Djillali Liabes, 22000 Sidi-bel-Abb`es, Algeria (e-mail: amar_tilmatine@ yahoo.fr). L. Dascalescu was with the Electronics and Electrostatics Research Unit, Laboratoire d’Automatique et d’Informatique Industrielle-Ecole Sup´erieure d’Ing´enieurs de Poitiers (LAII-ESIP), University of Poitiers, University Institute of Technology at Angoulˆeme, 16021 Angoulˆeme Cedex, France. He is now with the Electrostatics of Dispersed Media Research Unit, Electrohydrodynamic (EHD) Group, Laboratory of Aerodynamic Studies, University of Poitiers, University Institute of Technology at Angoulˆeme, 16021 Angoulˆeme Cedex, France (e-mail: [email protected]). Digital Object Identifier 10.1109/TIA.2007.912802

Under these circumstances, a system based on empirical rules may be more effective. No automated control system has been developed for the industrial electrostatic separators, yet. In most applications, the operator has the possibility to adjust two control variables: highvoltage level [9] and—in the case of roll-type separators—the speed of the rotating electrode [10]. The adjustment is performed in accordance with a set of rules based on expert knowledge of the best operating conditions for each application. This paper aims at analyzing the particularities of fuzzy logic as applied to the control of an insulation–metal standard electrostatic separator. The target was the simultaneous maximization of the high-purity conductor product and the minimization of the middling fraction obtained after separation. At this stage of the research, the separator was assumed to dispose of only one control variable: the high voltage energizing the electrode system. II. PROBLEM FORMULATION In a standard roll-type electrostatic separator [10], the granular mixture to be separated is fed with a certain speed on the surface of a rotating roll electrode, connected to the ground (Fig. 1). A high-intensity electric field is generated between this roll and one or several electrodes connected to a highvoltage supply [11]. The insulating particles are charged by ion bombardment in the corona field zone [12] and are pinned to the surface of the rotating roll electrode by the electric image force [10], [13]. The conducting particles are not affected by the corona field; they charge by electrostatic induction in contact with the grounded roll and are attracted to the high-voltage electrode. Thus, the list of factors influencing the electrostatic separation process should include the high-voltage level, the electrode configuration, the feed rate, the granule size, and the roll speed [3], [4].

0093-9994/$25.00 © 2008 IEEE

10

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 44, NO. 1, JANUARY/FEBRUARY 2008

Optimum operation of an electrostatic separator requires simultaneous maximization of several performance indexes: 1) metal recovery in the conductor product; 2) purity of the conductor product; 3) insulation recovery in the nonconductor product; and 4) purity of the nonconductor product. In the case of the electric wire waste recycling application considered in this paper, it has been established [2] that any action contributing to minimizing the quantity of the middling fraction also increases the recovery of insulating particles in the nonconductor product, the purity of which depends only on the quality of the mechanical conditioning (fine chopping) of the in-feed material. On the other hand, under normal conditions, the purity of the conductor product is not a very constraining issue (all product containing more than 97% copper is usually accepted by the recycling industry). Consequently, a good separation process is the one that maximizes the mass of the conductor product and minimizes the middling. With these considerations in view, the problem to solve consists in finding the value of the only control variable of the process, i.e., the applied high voltage U , that would minimize the percentage of middling mM and maximize the percentage of conductor mC in the output of the separator. The mass of conductor MC and of nonconductor MNC , as well as the mass of middling MM can be continuously monitored and employed for computing the feedback variables mM and mC of the fuzzy controller as mM =

MM MC + MNC + MM

mC =

MC . MC + MNC + MM (1)

As a rule, the value U1 that would minimize mM and the value U2 that would maximize mC are different: U1
= U2 . Therefore, when turning on the machine, the controller will start by looking for the value U1 , memorize the value mM 1 when mM was minimized, then look for U2 , memorize the value mM 2 when mC was maximized, and finally obtained a value U ∗ situated between U1 and U2 , for which mM = mM 1 + p(mM Z − mM 1 )

(2)

where p is a coefficient between 0 and 1 (the limit values correspond to the minimization of mM and to the maximization of the conductor product, respectively). For this paper, p = 0.5.

Fig. 2.

Block diagram of the fuzzy control system of an electrostatic separator.

A. Input and Output Variables The two output variables of the electrostatic separation process considered in this study, i.e., the percentage of conductor product (mC ) and the percentage of middling (mM ) are fed back to the input of the fuzzy controller (Fig. 2). At any given time, only one of the variables is selected in accordance with the rules defined in Section II and compared to a reference value Mr . This reference value may be Mr = 0, when the controller is set to minimize mM , and Mr = 100, when the objective is the maximization of mC . The difference between the measured values (mC or mM ) and the reference value (Mr ) is denoted by X and represents one of the two input variables of the controller. Thus X(i + 1) = Mr − mC (i)

Fuzzy controllers are very simple conceptually [14], [15]. They consist of an input stage, a processing stage, and an output stage. The input stage maps sensors or other inputs, such as switches, thumbwheels, etc., to the appropriate membership functions and the truth values. The processing stage invokes each appropriate rule and generates a result for each then combines the results of the rules. Finally, the output stage converts the combined result back into a specific control output value.

X(i + 1) = Mr − mM (i) (3)

where i is the cycle counter of the weighting system that processes the information given by sensors placed at the outlet of the separator. The other variable of the fuzzy controller is denoted by Y , and is defined as Y (i + 1) =

[mC (i) − mC (i − 1)] ∆u(i)

Y (i + 1) =

[mM (i) − mM (i − 1)] ∆u(i)

or (4)

with ∆u being the output variable of the controller and representing the increment of the applied high voltage u. It should be noted that u designates the normalized centered value of the actual voltage U as u=

III. FUZZY LOGIC CONTROLLER

or

2(U − U0 ) Um ax − Um in

(5)

where U0 = (Um ax + Um in )/2, with Um ax and Um in being the upper and lower limits, respectively, between which U can vary in a given application. The fuzzy controller is activated only when the following data are available: mM (0), mM (1), and ∆u(1). At the initial moment, when the separator is turned on (cycle i = 0), the fuzzy controller is not yet activated and the electrode system of the separator is energized at a predefined high voltage u. The weighting system provides the values mM (0) and mC (0), with a delay ∆t imposed by the characteristics of the sensors. During

YOUNES et al.: FUZZY CONTROL OF AN ELECTROSTATIC SEPARATION PROCESS

Fig. 3. “Membership functions” defining the states of the input and output variables of the fuzzy controller. (a) Percentage of middling X = m M . (b) Percentage of conductor X = m M . (c) Ratio Y = ∆x/∆u. (d) Increment ∆u of the normalized central value u of the applied high voltage.

the next cycle (i = 1), the applied voltage is u + ∆u(1), with ∆u(1) having a predefine value in this case, ∆u(1) = 0.02. At the end of this cycle, the fuzzy controller will be activated: the values mM (1) and mC (1) are available, together with mM (0), mC (0), and ∆u(1). It is, thus, possible to calculate X(2) and Y (2), using (3) and (4), for i = 1. During the cycle i = 2, the fuzzy controller uses X(2) and Y (2) to calculate the new increment ∆u(2) of the high voltage, and the weighting system will provide the values mM (2) and mC (both the values for equation (2)). At the end of any cycle i, either mM (i) or mC (i) will serve, together with ∆u(i) for computing the input variables X(i + 1) and Y (i + 1), with (3) and (4), respectively. B. Fuzzification Each variable X, Y , and ∆u can be divided into a range of states “negative” (N), “zero” (Z), “positive” (P). These states can be defined using three sets of membership functions (µXN , µXZ , µXP ), (µYN , µYZ , µYP ), and (µ∆ u N , µ∆ u Z , µ∆ u P ), respectively. The most common shape of membership functions is the combination of triangles and shouldered ramps shown in Fig. 3. With this scheme, the input or output variable no longer jumps abruptly from one state to the next. Instead, as its value increases, the variable loses value in one membership function, while gaining value in the next membership function. In the example presented in Fig. 4, the “fuzzification” process converts the value X1 = 0.6% into (µXP = 0.6, µXZ = 0.4) and Y1 = −1 into (µYN = 0.25, µYZ = 0.75). The number of membership functions in a set and their placement expressed the preferences of the process operator who served as a consultant for this research. Thus, 2% was considered appropriate for the width of the triangle representing the µXZ membership function for the fuzzification of X, when X is the percentage of middling mM (knowing that the middling could not represent more than 15% of the output). In the situations when X = mC , the width of the same function was established at 1.6% for an application where the maximum percentage of separated conductor product mC could not exceed 50%.

11

Fig. 4. Fuzzification of the input variables: calculation of the “truth values” of the “membership functions.” (a) X = 0.6. (b) Y = −1.

For the other input variable Y , the width of the triangular membership function µYZ was fixed at 8 [-]. The choice of a smaller value would increase the duration of any process of minimization/maximization. C. Inference Rules The processing stage of the fuzzy controller is based on a collection of logic rules in the form of IF-THEN statements, where the IF part is called the “antecedent” and the THEN part is called the “consequent”. Rule 1: IF [(X is negative) AND (Y is negative)] THEN (∆u negative). Rule 2: IF [(X is negative) AND (Y is positive)] THEN (∆u positive). Rule 3: IF [(X is positive) AND (Y is negative)] THEN (∆u positive). Rule 4: IF [(X is positive) AND (Y is positive)] THEN (∆u negative). Rule 5: IF [(X is zero) OR (Y is zero)] THEN (∆u is zero).

is is is is

For the values X1 = 0.6% (µXP = 0.6, µXZ = 0.4) and Y1 = −1 (µYN = 0.25, µYZ = 0.75) considered in the example discussed before IF (µXP

= 0.6 AND µYN = 0.25) THEN µ∆ u P = MIN(0.6, 0.25) = 0.25

IF

(6)

[(µXP = 0.6 AND µYZ = 0.75)]

OR[(µXZ

= 0.4 AND µYN = 0.25)]

OR[(µXZ

= 0.4 AND µYZ = 0.75)]

THEN

µ∆ uZ , = max[min(0.6, 0.75), min(0.4, 25) min(4.0, 0.75) = 0.6.

(7)

12

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 44, NO. 1, JANUARY/FEBRUARY 2008

Fig. 5. Defuzzification of the output variable of the fuzzy controller. (a) Truth values of the output membership functions generated using the inference rules. (b) Calculation of the actual setting of the output variable ∆u.

The aforementioned computations were done using the “min– max” inference method. According to this method, the output membership function is given the minimum weight of all the antecedents, when an AND relationship exists between them, as in (6). In the case of OR operator, the output membership function is given the maximum weight, as in (7). D. Defuzzification At this stage, each combination of output membership functions generated using the inference rules is converted back into a control output value. For the example discussed earlier, the output values corresponding to µ∆ u P = 0.25 and µ∆ u Z = 0.6 are ∆u1 = 0.35 and ∆u2 = 0.08, respectively. The calculation formula [15, p. 54] was used to obtain the final control output (Fig. 5): ∆u = 0.063.

(8)

IV. SIMULATION RESULTS AND DISCUSSION

Fig. 6. Results of the first simulation of fuzzy controller operation. (a) Adjustment of the applied voltage. (b) Modification of the mass (in percent of the feed) of “middling.” (c) “Conductor” products collected at the output of the separator. (I) Minimization of m M . (II) Maximization of m C . (III) Optimization. (IV) Stable operation at constant feed rate (100 kg/h, i.e., m = 0). (V) Steep variation of the feed rate from 100 to 150 kg/h (∗: the controller is active all the time; •: the controller is deactivated after the cycle #40).

A. Process Model The operation of the fuzzy controller is tested by numerical simulation. The process model employed for the simulations was established by using the design of experiments methodology [5]. Thus, the percentage of middling mM and the percentage of conductor mC were expressed as functions of the normalized centered value of the applied voltage u and the feed rate m as mM = 2.89 U 2 − 1.35 U − 4.92 U m + 53 m + 7.85 mC = 1.15 U + 1.82 U m − 2.34 m + 44.58.

(9) (10)

In the earlier formulas, u is given by (5), with Um in = 26 kV, Um ax = 30 kV, while m=

2(M − M0 ) Mm ax − Mm in

(11)

where M0 = (Mm ax + Mm in )/2, with Mm in = 50 kg/h and Mm ax = 150 kg/h being the upper and lower limits, respectively, between which the feed rate M can vary in the application considered here. These models enable the prediction of mM and mC , when the separator operates at a given u and m. For instance, at U = 29 kV and M = 125 kg/h, i.e., u = 0.5 and m = 0.5, mM = ym M (0.5, 0.5) = 9.31%. Thus, it is possible to simulate the response of the separator at any adjustment of the applied high voltage U , as well as the effect of uncontrollable variations of the feed rate M . B. Optimization of the Process Response A first set of simulated results (Fig. 6) was obtained for a feed rate M = 100 kg/h (i.e., m = 0). The fuzzy controller needed

YOUNES et al.: FUZZY CONTROL OF AN ELECTROSTATIC SEPARATION PROCESS

13

C. Response of the System at Uncontrolled Variations of the Feed Rate In response at a steep variation of the feed rate from 100 to 150 kg/h (phases IV and V, in Fig. 6), the high voltage was adjusted at a new optimum value U = 30 kV, in less than four cycles. If the voltage remains U = 28.8 kV, then mM = 10%, and mC = 44.5%. This demonstrates the efficiency of the fuzzy controller to optimize the operation of the electrostatic separator in conditions that are different from those initially set for a given application. The simulation of the case when the feed rate dropped down very fast from 150 to 100 kg/h is displayed in Fig. 7, phases IV and V. The three-step operation of the fuzzy controller can be easily observed on the curves of U, mM , and mC . As soon as the drop of the feed rate caused the modification of mM and mC (cycle 33), the controller started to adjust the applied high voltage U , so as to minimize mM (up to cycle 50), then maximize mC (cycles 51–56), and finally, attain the target expressed by (2) (cycles 57–60). In the absence of voltage control (i.e., U remains at 30 kV), the diminution of the feed rate is followed by a tiny reduction of the percentage of middling and a slight increase of the percentage of conductor. With the voltage adjusted by the fuzzy controller, the final set point is identical with the one obtainedat the end of phase III in Fig. 6. This demonstrates that, at a given feed rate, the controller is capable to retrieve the same optimum, no matter what are the initial conditions. V. CONCLUSION

Fig. 7. Results of the second simulation of fuzzy controller operation. (a) Adjustment of the applied voltage. (b) Modification of the mass (in percent of the feed) of “middling.” (c) “Conductor” products collected at the output of the separator. (I) Minimization of m M . (II) Maximization of m C . (III) Optimization. (IV) Stable operation at y-constant feed rate (150 kg/h, i.e., m = 1). (V) Steep variation of the feed rate from 150 to 100 kg/h (∗: the controller is active all the time; •: the controller is deactivated after the cycle #40).

ten cycles to minimize mM (phase I), followed by six cycles to maximize mC (phase II), and five cycles to attain the target expressed by (2) (phase III). The optimum response is obtained at a high voltage U = 28.8 kV. The second simulation (Fig. 7) was carried out for a feed rate, M = 150 kg/h (i.e., m = 1). In that case, the phases I–III were confounded, as maximization of mC was achieved at the same time with the minimization of mM , at U = 30 kV, which is the limit high voltage for the application under study. The percentage of middling mM = 9.7%, was higher than at the lower feed rate M = 100 kg/h, where mM = 8.5%, and the percentage of conductor was slightly lower (mC = 45.2%, as compared with mC = 45.5%). This points out that, as expected, at higher feed rates, it is not possible to maintain the same performance indexes as at lower ones.

Fuzzy controllers might represent an appropriate solution for optimizing such complex electrostatic processes as the separation of granular solids. Expertise in the field of electrostatic separation is a prerequisite for defining the membership functions and the inference rules the design of such a controller is based upon. The results obtained by adjusting one control variable, the high voltage applied to the electrode system, in accordance with the state of two output variables (the masses of conducting and middling products) were satisfactory. Nevertheless, a different choice of fuzzification and defuzzification methods employed is likely to improve the dynamic response of the control system. Therefore, further researches should focus on this issue, as well as on the possibility of simultaneously adjusting a second control variable: the speed of the rotating roll electrode connected to the ground. Though numerical simulations proved the feasibility of fuzzy controllers, their efficiency remains to be demonstrated by experimental investigation on actual electrostatic separation processes. However, taking into account the availability of “online” continuous weighting devices and the simplicity of their interfacing with a digital controller, it is expected that no difficulties will impede the industrial application of such controllers. REFERENCES [1] M. Mihailescu, A. Samuila, A. Urs, R. Morar, A. Iuga, and L. Dascalescu, “Computer-assisted experimental design for the optimisation of electrostatic separation processes,” IEEE Trans. Ind. Appl., vol. 38, no. 5, pp. 1174–1181, Sep./Oct. 2002.

14

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 44, NO. 1, JANUARY/FEBRUARY 2008

[2] L. Dascalescu, A. Tilmatine, F. Aman, and M. Mihailescu, “Optimisation of electrostatic separation processes using response surface modeling,” IEEE Trans. Ind. Appl., vol. 40, no. 1, pp. 53–59, Jan./Feb. 2004. [3] L. Dascalescu, A. Mihalcioiu, A. Tilmatine, M. Mihailescu, A. Iuga, and A. Samuila, “Electrostatic separation processes,” IEEE Ind. Appl. Mag., vol. 10, no. 6, pp. 19–25, Nov./Dec. 2004. [4] L. Dascalescu, A. Samuila, A. Mihalcioiu, S. Bente, and A. Tilmatine, “Robust control of electrostatic separation processes,” IEEE Trans. Ind. Appl., vol. 41, no. 3, pp. 715–720, May/Jun. 2005. [5] K. Medles, A. Tilmatine, F. Miloua, A. Bendaoud, M. Younes, M. Rahli, and L. Dascalescu, “Set point identification and robustness testing of electrostatic separation processes,” IEEE Trans. Ind. Appl., vol. 43, no. 3, pp. 618–626, May/Jun. 2007. [6] I. Kiss, L. Pula, E. Balogh, L. T. K´oczy, and I. Berta, “Fuzzy logic in industrial electrostatics,” J. Electrostat., vol. 40/41, pp. 561–567, 1997. [7] I. Kiss and I. Berta, “New concept of ESP modelling based on fuzzy logic,” J. Electrostat., vol. 51/52, pp. 206–211, 2001. [8] N. Grass, “Fuzzy-logic-based power control system for multifield electrostatic precipitators,” IEEE Trans. Ind. Appl., vol. 38, no. 5, pp. 1190–1195, Sep./Oct. 2002. [9] A. Iuga, V. Neamtu, I. Suarasan, R. Morar, and L. Dascalescu, “Highvoltage supplies for corona-electrostatic separators,” IEEE Trans. Ind. Appl., vol. 34, no. 2, pp. 286–293, Mar./Apr. 1998. [10] A. Iuga, R. Morar, A. Samuila, and L. Dascalescu, “Electrostatic separation of metals and plastics from granular industrial wastes,” in Proc. IEEE Sci. Meas. Technol., 2001, vol. 148, pp. 47–54. [11] K. Haga, “Applications of the electrostatic separation technique,” in Handbook of Electrostatic Processes, J. S. Chang, A. J. Kelly, and J. M. Crowley, Eds. New York: Marcel Dekker, 1995, pp. 365–386. [12] L. Dascalescu, R. Morar, A. Iuga, A. Samuila, V. Neamtu, and I. Suarasan, “Charging of particulates in the corona field of roll-type electroseparators,” J. Phys. D, Appl. Phys., vol. 27, pp. 1242–1251, 1994. [13] L. Dascalescu, A. Mizuno, R. Tobaz´eon, A. Iuga, R. Morar, M. Mihailescu, and A. Samuila, “Charges and forces on conductive particles in roll-type corona-electrostatic separators,” IEEE Trans. Ind. Appl., vol. 31, no. 5, pp. 947–956, Sep./Oct. 1995. [14] H. J. Zimmermann, Fuzzy Set Theory and Its Applications. Boston, MA: Kluwer, 1991. [15] H. B¨uhler, R´eglage par logique floue. Lausanne, France: Presses Polytechniques et Universitaires Romandes, 1994.

Mohamed Younes was born in Mostagnem, Algeria, in 1965. He received the M.S. degree from the University of Sciences and Technology, Oran, Algeria, in 1989, and the Magister (Dr. Eng.) degree from the Department of Electrical Engineering, in 1998, and the Ph.D. degree from the University Djilali Liabes, Sidi-Bel-Abbes, Algeria, in 2007, all in electrical engineering. He was a Research Fellow with the University Institute of Technology, Angoulˆeme, France. He is currently a Senior Lecturer of electrical engineering at the University Djilali Liabes. He is the author or coauthor of several scientific papers published in various international or national journals, as well as in conference proceedings. His current research interests include high-voltage engineering, computational electrostatics, and fuzzy-logic. Dr. Younes is a member of the Electrostatics and High-Voltage Engineering Research Unit, Interaction R´eseaux Electriques-Convertisseurs-Machines (IRECOM) Laboratory, University Djilali Liabes.

Amar Tilmatine received the M.S. degree in electrical engineering and the Magister (Dr. Eng.) degree from the University of Science and Technology, Oran, Algeria, in 1988 and 1991, respectively, and the Ph.D. degree from the University Djilali Liabes, Sidi Bel Abbes, Algeria, in 2004. Since 1991, he has been teaching electric field theory and high-voltage techniques at the Department of Electrical Engineering, University Djilali Liabes. He was the Chairman of the Scientific Committee of the Department of Electrical Engineering, University

Djilali Liabes, from November 2002–November 2005. He is currently the Head of the Electrostatics and High-Voltage Engineering Research Unit, Interaction R´eseaux Electrique-Convertisseurs Machines (IRECOM) Laboratory, University Djilali Liabes. From 2001 to 2007, he visited the Electronics and Electrostatics Research Unit of the University Institute of Technology Angoulˆeme, at least once a year, as Invited Scientist, to work on a joint research project on new electrostatic separation technologies. His current research interests include high-voltage insulation, gas discharges, and electrostatic precipitators.

Karim Medles was born in Tipaza, Algeria, in 1972. He received the M.S. degree, the Magister (Dr. Eng.) degree, and the Ph.D. degree in electrical engineering from University Djilali Liabes, Sidi Bel Abbes, Algeria, in 1994, 1999, and 2006, respectively. In 1999, he joined the University Djilali Liabes, where he was initially a Senior Lecturer, and then, became an Assistant Professor. He was invited as a Visiting Scientist in France. He is a member of the Electrostatics and High-Voltage Engineering Research Unit, Interaction R´eseaux Electrique-Convertisseurs Machines (IRECOM) Laboratory, University Djilali Liabes. He is the author or coauthor of several scientific papers published in various international journals, as well as in conference proceedings. His current research interests include power systems, high-voltage engineering, and electrostatics.

Lucian Dascalescu (M’93–SM’95) received the M.S. degree (Hons.) in electrical engineering from the Technical University of Cluj-Napoca, Cluj-Napoca, Romania, in 1978, the Dr. Eng. degree in electrotechnical materials from the Polytechnic Institute of Bucharest, Bucharest, Romania, in 1991, and the Dr. Sci. degree and the “Habilitation a` Diriger de Recherches” diploma both in physics from the University “Joseph Fourier,” Grenoble Cedex, France, in 1994 and 1996, respectively. From 1978 to 1982, he was with the Combinatul de Utilaj Greu [CUG (Heavy Equipment Works)], Cluj-Napoca. In 1983, he joined the Technical University of Cluj-Napoca, where he was initially an Assistant Professor, and then, became an Associate Professor of electrical engineering. From October 1991 to June 1992, he was with the Laboratory of Electrostatics and Dielectric Materials (LEMD), Grenoble, France, as a Research Fellow. He was an Invited Research Associate and a Lecturer at Toyohashi University of Technology, Toyohashi, Japan, and also a Visiting Scientist at the University of Poitiers, Poitiers, France. From 1994 to 1997, he was with the University Institute of Technology, Grenoble, where he taught electromechanical conversion of energy. In September 1997, he was appointed as a Professor of Electrical Engineering and Automated Systems, and the Head of the Electronics and Electrostatics Research Unit at the University Institute of Technology, Angoulˆeme, France. From 1999 to 2003, he was the Head of the Department of Management and Engineering of Manufacturing Systems. Currently, he is the Head of the Electrostatics of Dispersed Media Research Unit, which is part of the Electrohydrodynamic (EHD) Group, Laboratory of Aerodynamic Studies, University of Poitiers. He is the author of several textbooks in the field of electrical engineering and ionized gases. He holds 14 patents and has authored or coauthored more than 70 journal papers, and was invited to lecture on the electrostatics of granular materials at various universities and international conferences. Prof. Dascalescu is a Senior Member of the IEEE Industry Applications Society (IAS) and the Chair of the Electrostatics Processes Committee. He is a member of the Electrostatics Society of America, the Electrostatics Society of Romania, the Soci´et´e des Electriciens et Electroniciens (SEE), and the Club Electrotechnique, Electronique, Automatique (EEA), France.