# A-Multiphase-TractionFast-Battery-Charger-Drive-for-Electric-or-Plug-in-Hybrid-Vehicles

A Multiphase Traction/Fast-Battery-Charger Drive for Electric or Plug-in Hybrid Vehicles Solutions for Control in Traction ModeA. Bruyère, L. De Sousa, B. Bouchez Valeo Engine and Electrical Systems VEES 14, avenue des Béguines, 95800 Cergy Saint-Christophe, France antoine.bruyere@valeo.com P. Sandulescu1,2, X. Kestelyn1,2, E. Semail1,2(1)Arts et Metiers PARISTECH, L2EP, Lille, France (2)Univ Lille Nord de France, L2EP, Lille, France 8 Bd Louis XIV, 59046 LILLE, France eric.semail@ensam.euAbstract— For Electric Vehicles (EV), the charger is one of the main technical and economical weaknesses. This paper focuses on an original electric drive [1]-[3] dedicated to the vehicle traction and configurable as a battery charger without need of additional components. This cheap solution can outfit either electric or plug-in hybrid automotive vehicles, without needing additional mass and volume dedicated to the charger. Moreover, it allows a high charging power, for short duration charge cycles. However, this solution needs specific cares concerning the electrical machine control. This paper deals with the control of this drive [1], focusing on traction mode. In introduction, a review is done about topologies of combined on-board chargers. Then, the studied topology is introduced; using a 3-phase brushless machine supplied with a 6-leg Voltage Source Inverter (VSI). A model for its control is defined in the generalized Concordia frame, considering the traction mode. Then, an analysis of this model is established using a multimachine theory and a graphical formalism (the Energetic Macroscopic Representation denoted EMR). Using EMR, a description of energy flows shows specific control constraints. Indeed, numerical simulations illustrate the perturbations on the currents and the torque when controlling the machine with standard control methodologies. An improved control, deduced from the previous analysis, shows good performances, strongly reducing currents and torque ripples. Keywords- Electric Vehicle, Plug-in Hybrid Vehicle, On-board Battery Charger, H-bridge Voltage Source Inverter, Multiphase Drive, Control I. INTRODUCTION For both electric and Plug-in hybrid vehicles [4], one of the main technical and economical weaknesses concerns the use of a charger. Indeed, technically, an on-board charger means loading extra volume and extra mass. Otherwise, an off-board charger does not allow recharging the battery anywhere. In both cases, economically, using a charger means extra cost due to a specific converter. In the last few years, several solutions have been tested, combining the motor converter with the motor windings to make on-board chargers [5]-[9]. This is possible since both charger and motor with its supply device are composed of windings, capacitors and power electronics. Moreover, the battery recharge only occurs when the car is stopped, so, the drive is not used for the two different operation modes at the same time. In [5], several of these solutions are described and are called “combination topologies”. Nevertheless, theses solutions are only suitable for a single-phase charge and cannot offer a faster charging option by a three-phase grid connection. Other solutions have been proposed in [6]-[7], but two main drawbacks are recurrent. The first one is the need of a high current relay to connect the AC grid on the electrical machine’s coils. This is still an over-cost that makes the solution less attractive. The second one is the generation of a rotating magnetic air-gap field, which is able to induce high voltage on the rotor’s windings or to move the rotor. This is a serious issue in case of Permanent-Magnet Synchronous Machine (PMSM). In this paper, an original combination topology battery charger is studied [1]-[3]. This solution ensures lack of air-gap field due to the stator windings when these windings are supplied in charge mode. Moreover, it allows both single-phase and three-phase (fast) charging modes. However, this topology induces specific cares for control, due to the need of controlling three independent currents. In a first part, this original topology is introduced. Note that more details about the topology, its assets and drawbacks, can be found in [1]. A model is then established, considering the drive control in traction mode. Two specific tools are used to analyze the model: first, the multimachine theory [10]-[12]. This tool has been developed specifically for studying multiphase drives. Then, a graphical formalism called Energetic Macroscopic Representation (EMR) [14]-[17]. EMR allows a representation of models fitted with an energy study, in view of controlling energy flows. Using both the multimachine theory and EMR, specific constrains appear for controlling the machine. In a third section, two kinds of control in traction mode are tested through numerical simulations. These simulations show that standard control methodologies, fitted with standard 3-phase drive control, can induce strong perturbation on currents and torque. Thus, an improved control is tested, showing a strong reduction of these perturbations. The influence of the Pulse Width Modulation (PWM) technique is also studied. II. DRIVE DESCRIPTION The drive is composed of a three-phase machine whose phases are not electrically coupled (no wye, no delta-coupling). Each phase is supplied by a full H-bridge Voltage Source Inverter (VSI). In comparison with classical three-phase wye coupled machines supplied by a three-leg VSI, this topology allows: • Imposing a higher voltage to each phase; • Using each one of the three phases of the machine as an inductance for achieving a battery charger. Of course, if it is possible with the classical topology to use only two power components and two drivers to achieve each one leg of the three legs, then the proposed topology suffers apparently of twice more power components and more complex control than the classical topology. Nevertheless, it must be remarked that the maximum current in each device will be half less, which allows the use of smaller and cheaper components [18]. In Figure 1. the VSI energy source is the DC/DC converter capacitor, imposing the voltage UC. This converter works as a voltage boost in traction mode. In the next discussion, UCis assumed to be constant and the study focuses on the machine control using the VSI. i2DC/DC converterBatteryGrid connection atmiddle-point stator windings:DC/AC converterUCu1u3u2i1i3UBatiLiDC/ACiDC/DCEMI Filtersthis gives a direct view of energy flows. Then, each energy sub-systems of the drive are connected together using the action/reaction principle. Moreover, the integral causality is always respected in order to fit with the physical reality. For example, in Figure 2. it is shown how connecting each others the energy sub-systems of the studied drive, using the action/reaction principle. Thus, Figure 2. means: the DC/DC converter is considered as the electrical energy source for the drive. It imposes the voltage UCas “action” to the drive. Then, the DC/DC converter is connected to the VSI, which imposes the associated reaction: the DC bus current iDC/AC. The VSI tuning input is ACDCm/K, representing the modulation functions. Next, the VSI also imposes the 3-dimensional voltage vector SuGto the machine stator windings. The associated reaction to SuGis the current vector SiG, imposed by an accumulation of energy block (representing the magnetic energy accumulation in windings). From this energy accumulation block point of view, the emf vector SeGis seen as a perturbation input. At last, the torque T is an output of the electromechanical conversion block. It is a function of the two inputs of this block: SiGand the rotation speed Ω (rotation speed is supposed to be imposed by system outside). Finally let us check that this scheme is a description of the energetic chain. Indeed, each couple of superposed arrows represents the power when multiplying each action with its associated reaction. ChassisTΩiSiSuSeSmDC/ACUCiDC/ACElectrical MachineDC/DCDC/ACconverterFigure 2. Machine and Converter EMR Representation in the Natural Reference Frame The representation of Figure 2. is given in the natural UVW frame, what refers to the expressions (1) and (2). Now, considering the expressions (3)-(5), defined using the multimachine theory, a new representation can be introduced for the electrical machine (Figure 3. ). In Figure 3. the Concordia transformation is represented with an electrical coupling. This illustrates the energy distribution between the two fictitious machines M0 and M1, stacked one above the other one (energy flows are represented with green double arrows in the figure. It is assumed that the main part of energy passes through the main fictitious machine M1). To be noted that the energy distribution operated with the Concordia transformation also naturally respects the harmonics distribution depicted with TABLE I. Concretely, this means, taking the example of the voltage vector SuG, potentially containing an infinity of harmonics, that M1 is only supplied with voltage harmonics of ranks 1, 2, 4, 5, 7, … Concerning the zero-sequence M0 machine, it is supplied with voltage harmonics of ranks 3, 6, 9, … There is no interaction between these harmonics families and both fictitious machine creates its own torque. The total torque T is the sum of TM0and TM1. These notions were introduced with (4) and (5). Here it is graphically expressed in terms of energy distribution into independent “fictitious machines”. Finally, this approach will help design and control analysis. iM0eM0iM0uM0iM1uM1uSiSiM1eM1TM0+TM1ΩTM0ΩTM1ΩZero-sequence fictive machine M0(harmonics ranks 0, 3, 6, 9, …)Concordia TransformationMain fictive machine M1(harmonics ranks 1, 2, 4, 5, 7, 8…)1100 MMMMSSiuiuiuGGGG⋅+⋅=⋅Energy flowFigure 3. Electrical Machine Representation in the Concordia Reference Frame To conclude with this chapter, Figure 4. shows the manner with which the control structure is organized using EMR. The formalism helps defining a control structure by an inversion of the energy chain. For conversion blocks, the inversion can be directly established. For energy accumulation blocks, the inversion needs a controller and the associated measurements. Here, controllers are used to control the fictitious machines currents. The representation of Figure 4. fits with the control of the torque T: torque reference is split into two components TM0 refand TM1 ref. Then, these references are transformed in currents references, inversing the model electromechanical conversion blocks. The currents are controlled using controllers, leading to the voltage references. At last, the inverse Concordia transformation leads to the voltage reference expressed in the natural UVW frame. To be noted that M1 currents can be controlled in a rotating Park frame, exactly as it is generally done for standard 3-phase drives. iM0eM0iM0uM0iM1uM1uSiSiM1eM1TM0+TM1ΩTM0ΩTM1ΩuSrefuM0 refuM1 refiM1 refiM0 refTrefTM0 refTM1 refM0 controlM1 controlConcordiainverseM0 and M1currents controllersDirectinversionTorque referencerepartitionFigure 4. 3-Phase Machine Control Structure Taking Into Account the Two Fictitious Machines IV. CONTROL AND DESIGN CONSTRAINTS POINTING OUT The multimachine theory, associated with EMR, has been developed and experimentally validated for 5-, 7- and 9- phase drives [10]-[13]. Now, it is used to illustrate the specific control constraints of the topology described with Figure 1. A. Tests Conditions In order to illustrate the influence of the zero-sequence fictitious machine M0 control, it is considered a machine with non-sinus electromotive forces (emf), characterized with 15% of emf 3rdharmonic. Then, for each test, the following operating mode is set: - The rotation speed is fixed: N = 1000 rpm. - M1 currents are controlled in a dq-Park rotating frame with Proportional + Integral (PI) controllers (as it is usually done for standard 3-phase drives). The following references are arbitrarily chosen: iM1d-ref= 10 A, iM1q-ref= -30 A. - Initial conditions of currents: 0 A. - M1 currents closed loop time constant tuning: 2.1 ms. - The simulations are carried in continuous mode (without discrete sampling effects). Taking this operation mode as reference, the studies will focus on controlling M0. Step after step, we will answer the following questions: - What is the influence of the emf waveform? - What is the influence of the voltage modulation? B. Standard Control A first control is carried out. This control uses a standard control methodology (developed for standard 3-phase-machine/3-leg-VSI drives). Thus, the zero-sequence fictitious machine M0 is not taken into account and the control structure is designed only considering M1 (Figure 5. ). Then, the control is established in a standard way, controlling the dq-currents in the rotating Park frame associated with M1, as described in the tests conditions. The results of this first control are shown in Figure 9. (a): The currents in the natural UVW frame, (b): the currents in Park frame, (c): the torque. Because emf contains a 3rdharmonic and because M0 is not controlled, the current also naturally contains a 3rdharmonic. The main drawbacks concern extra losses and torque ripples. This illustrates why a standard control is not fitted with the considered topology. iM0eM0iM0uM0iM1uM1uSiSiM1eM1TM0+TM1ΩTM0ΩTM1ΩuSref0uM1 refiM1 refTref= TM1 refM0 controlM1 controlM0 is ignoredFigure 5. 3-Phase Machine Control Structure in Usual Case C. Controlling the zero-sequence current M0 being at the origin of pulsating torque when iM0is not controlled, we will now use the control structure described in Figure 4. A focus is brought to M0 current control in Figure 6. In this second test, it is considered an ideal control of iM0, with a perfect compensation of the perturbation eM0(Figure 6. ) To cancel the M0 pulsating torque, it is chosen as reference for iM0: iM0 ref= 0 A. In Figure 10. (b), iM0perfectly follows its reference. Then, iM1currents are always controlled as previously. The global behavior is identical to an electrically coupled machine and the torque is smooth (Figure 10. (c)). So it is demonstrated that controlling iM0is a solution to compensate the lack of electrical coupling. sM0M01Kτ++-iM0eM0+-iM0-ref ++uM0C(s)iM0-ref uM0iM0iM0eM0uM0-refFigure 6. Focus on M0 Fictitious Machine Control D. Influence of the Pulse Width Modulation (PWM) Until this point, the VSI was modeled as amplifier that imposes average voltages. Now the influence of the Pulse Width Modulation (PWM) is studied. Indeed, it is generally assumed that PWM is at the origin of common mode voltage (6) which affects the zero-sequence M0 fictitious machine. In the standard case, with an electrical coupling, this common mode voltage can be ignored (except for EMC considerations or bearing current calculation). In the considered system, the common mode voltage,