LINE VOLTAGE REGULATOR BASED ON MAGNETIC-CONTROLLED INDUCTORS FOR LOW VOLTAGE GRIDS

The rapid expansion of distributed energy resources in the German low voltage network has led to voltage violations, especially in rural areas with little electric load. To address this issue, this paper presents an innovative line voltage regulator for low voltage networks which operates with a magnetic controlled inductor. Therefore, a software environment for simulating magnetic based regulators and a hardware test bench for line voltage regulators are developed. Results of the simulations and the laboratory tests of the first prototype device are provided.


INTRODUCTION
The energy concept of the German Federal Government plans to massively expand the amount of renewable energy.Until 2020 it should rise to 35% and until 2050 to 80% [1].Most of the renewable power units are installed in the distribution networks [2][3] which leads to a bidirectional power flow between the voltage levels.Due to that, especially in rural areas with a high penetration of distributed generation and less electric load, issues occur satisfying the demanded voltage range by EN 50160 [4].Hence there is a need for action to avoid voltage violations and to enable further expansion of distributed energy resources (DER).Network reinforcements can be partly avoided through the use of new technologies [5].Particularly for long branches, the use of line voltage regulators (LVR) can be beneficial [3] [6].An LVR introduces an additional control voltage to the line by using of a series transformer.This control voltage leads to a rise or a reduction of the line voltage.Figure 1 shows the voltage impact of an LVR along a feeder.Due to a high distributed generation, there would be a voltage violation without further action.The LVR provides opposite phase voltage to decrease the line voltage.To vary the control voltage different technologies can be used.

Figure 1 Operation Principles of a Line Voltage Regulator
This paper presents a research on an innovative LVR for low voltage networks.The LVR operates with a magneticcontrolled inductor (MCI) that enables a stepless regulation of voltage and high robustness due to lacking switch elements or other moving parts.Based on the first prototype by EBG and Magtech [7], further developments are considered in the InLiNe project.The goal of this project is to study and develop innovative voltage regulators for low voltage networks.For this purpose, a hardware test bench and a software environment for simulating magnetic based regulators are developed.Based on first simulations and laboratory tests the functional principles of the prototype, its performance and its influence on power quality are presented.

SETUP AND FUNCTIONALITY OF THE PROTOTYPE
In the considered case the intended purpose of the LVR is to reduce voltage in the event of voltage rise through a feed-in by DER.Therefore, the current prototype is not yet able to increase voltage at the feeder side.Figure 2a) shows a simplified single-phase model of the prototype.As seen in the figure, current flow is directed from the feeder side with DER via the prototype to the substation.The MCI and the primary winding of the series transformer build up the control circuit.Formula (1) and (2) represent the appropriate voltage equations.
=   +   (1)   =   +   =   +   + (2) The phasor diagrams based on the equations are shown in Figure 2b) and outline qualitatively the voltages by different inductance values of the MCI.results in a voltage rise of   .However, adjustments of the MCI inductance are not proportional to the voltage at the feeder side   due to saturation effects in the MCI and varying voltage angles.In the following, the remaining question how the MCI varies its inductance is answered.

Magnetic-Controlled Inductor
The used MCI consists of a toroidal transformer with a turns ratio of  1  2 ⁄ = 1.Additionally, there is a separate control winding inside of the iron core which is directed vertically to the main coils.Through the control winding flows a direct current which intentionally saturates the core of the transformer.Depending on the rising core saturation the inductance of the main coils decreases.Hence, the main coils' inductance is regulated by varying the control current.Figure 3 shows the setup of the MCI and the direction of the magnetic flux.

Three-Phase Circuit
The three-phase circuit distinguishes from the model in Figure 2a) since the MCIs are wired in an interconnected star winding connection.This has the advantage that the voltage   is split on two MCIs which allows the use of a smaller iron core with less iron losses and less material costs.Figure 4 shows a simplified circuit diagram of the prototype.The turns ratio of the series transformer is 9,8.

HARDWARE TEST BENCH FOR LVR'S
To study the functional principles of the prototype, its performance and its influence on the power quality, a test bench is developed in the laboratory which enables the integration of LVRs into an existing test platform.In this way, the prototype can be studied for several voltage and load situations (0V-500V, 0A-160A).Furthermore the test bench allows to study the different components of the prototype in particular.
Figure 5 shows the prototype integrated in the test bench.

MODELLING MAGNETIC BASED LVR'S
To model magnetic based LVRs the simulation software PLECS is chosen since it allows to combine magnetic and electric circuits easily and provides an interconnection to MATLAB/Simulink.First of all the purpose is to model a series transformer and an MCI in general so that several setups can be developed by different parameter specifications and variants of connection.Both components -series transformer and MCI -are transformer types, hence the standard model of a transformer is considered to model them.Referring to Figure 6a) the standard model contains winding resistances  1 , ′ 2 and leakage inductances  1 , ′ 2 of the primary and the secondary coil of a transformer.Furthermore, it contains the core losses resistance   and the main inductance   .
To identify these impedances for one component opencircuit and short-circuit tests are used (referring Figure 6b) and 6c)) [8].

Figure 6 a) Standard Model of a Transformer b) Short-Circuit Model of a Transformer c) Open-Circuit Model of a Transformer
In general, an LVR series transformer operates in a linear region, hence the standard model is able to represent it appropriately.In contrast, an MCI is working intentionally in saturation conditions with an additional dependency on the control current, hence, it is assumed that the standard model is not sufficient.An analysis of the prototype MCI shows that winding resistances and even the leakage inductances behave linear and are nearly independent to the control current.Consequently, they can be modelled by impedances.On the contrary, the core loss resistance and the main inductance show dependency to the control current and saturation effects.Therefore, the cross-branch of the MCI model cannot be designed with passive components.Instead of that, a magnetic circuit with a hysteretic core element is created that is interconnected with the electrical circuit and shown in Figure 7.For the hysteretic core model a PLECS component is used which operates with a Preisach model and is smoothed by a Lorentzian distribution function [9].The Preisach model generates a hysteresis curve based on the coercivity   , the remanence   , the magnetic field strength and the magnetic flux density at the point of saturation   ,   as well as the saturation permeability   .Figure 7b) shows the generated hysteresis curve with the associated defining points.Further specific parameters of the component that need to be provided for the magnetic AC DC

Saturation Effects of the MCI
To analyse the cross-branch of the MCI the open-circuit measurements have to be considered.Based on the currents and voltages of the MCI the H-B-hysteresis curves of the core are determined with Formula (3) and (4).
: cross-sectional area : mean length of flux path : magnetic flux density : number of turns : magnetic field strength  2 : voltage at the secondary  1 : current in the primary winding winding Thereby ,  and  are fixed parameters specified by the component,  1, and  2, are measured in open-circuit at various control currents.Figure 8 shows selected hysteresis curves and the impact of the control current   .The control current produces an additional magnetic flux that causes the core to saturate at a lower magnetic flux density in the main winding.The hysteresis curve is compressed with rising control current and the total inductance decreases.The defining points of the hysteresis curves are identified for one operation point.  and   correspond with the points of intersections.Furthermore, a point of saturation (  ,   ) is identified by observing the slope of the curve.The slope of the saturation region corresponds to   .To model the MCI for different operation conditions the permeability is changed which leads to different flux densities.Therefore, a functional relationship between the control current and the permeability is calculated via curve fitting.

IMPACT OF THE PROTOTYPE AND MODEL VALIDATION
In the following, the static behavior of the prototype is analyzed.The analysis is focused on the performance of the device and its influence on power quality.The voltage on the feeder side is fixed to 400V phase to phase.The grid side is connected to an electric load of 2,8Ω.This setup yields a current from the feeder to the grid which models feed in by DER.The resulting current depends on the voltage variation of the prototype and lies between 80A and 95A. Figure 9a) shows the voltage variation between the feeder and grid side RMS voltages.The maximal variation is about 19V.It can be seen that the voltage variation is not linear to the control current which should be considered for designing the regulator.The efficiency is seen in Figure 9b).It decreases with the rising control current since the current in the control circuit grows, due to the decreasing inductance of the MCI.The two last data points are outlier caused by a voltage drop of 5V of the feeder voltage during the measurements.The minimal efficiency is about 98%.The difference of the voltage angle decreases with the rising control current.The THD of the grid side voltage is represented in Figure 9d).The feeder side voltage already has a THD of 1,4%.Due to that, the prototype mainly adds distortions between a control current of 100mA and 650mA.In this domain, saturation effects of the MCIs can be noticed.Higher control currents result in a completely saturated core and accordingly to a linear behavior again.Summarized, due to the prototype there is an additional distortion of 1,9% in the worst case.Additionally, in Figure 9 the behavior of the simulated prototype is compared with the measurements.It can be noted that for control currents higher than 300mA, the voltage variations in the simulation are about 2V larger than in the measurements.Furthermore, the simulation model has 1% higher efficiency than the prototype in the laboratory tests.The voltage angle deviates about 0,3°.In general, the THD displays the same behavior in the simulation and the measurements.The modeled prototype adds mainly distortions between a control current of 200mA and 650mA.Beyond this region the THD is 1% smaller than in the measurements, because the grid voltage in the simulation is a pure sine wave without distortion.
To evaluate the model in more detail and to give statements regarding the deviations of the current   and voltages   and   of the control circuit (depicted in Figure 2) are compared for the simulation and the laboratory tests.Therefore, Figure 10 gives a graphical example for a control current of 300mA.

Figure 10 a) MCI Voltage b) Primary Voltages of the Series Transformer a) Current in the Control Circuit
First, it can be noted that the mean Pearson correlation between the different simulated and measured MCI voltages   is 0,953 which means that there is a high linear dependency.Deviations appear since the Preisach model does not fit the hysteresis curve of the MCI exactly.Furthermore, the measured voltage is smoother than the simulated one which suggests that there is an impact of parasitic capacities.The primary voltages of the series transformer   has a mean Pearson correlation of 0,973.However, the peak values of the simulated   are on average about 10% higher than in the measurements.According to that, the voltage variation in the simulation is larger than the in measurements.This deviations arise because of the errors at the MCI voltages and the negligence of parasitic impacts in the device.For the same reason there are deviations in the current in the control circuit  which has a Pearson correlation of 0,935.
According to this validation the simulation has a sufficient accuracy to represent the real behavior of the prototype.

CONCLUSION
On one hand, this paper reveals the functional principles of the Line Voltage Regulator prototype which is based on magnetic components.On the other hand, it presents its performance and its influence on power quality in static operating conditions.Its performance, i.e. the voltage variation and the efficiency, proves satisfactory.On the contrary its influence on power quality, i.e. the voltage angle and the THD, should be improved to avoid any malfunctions.A simulation model is developed which proves to be sufficiently accurate.By means of this model, further developments and optimizations can be done at the prototype.Alternative connection variants, adding filters as well as parameter changes, can be studied.Furthermore, different control algorithms can be developed.Finally, promising results will be transferred from the simulation model to the real prototype and analyzed in laboratory tests.

Figure 2 a
Figure 2 a) Simplified Single-Phase LVR Model b) Phasor Diagram to the LVR Model According to the phasor diagrams decreasing the inductance of the MCI lead to a voltage drop of   and The InLiNe Project is funded by the German Federal Ministry for Economic Affairs and Energy Figure 5 LVR Test Bench

Figure 7 a
Figure 7 a) MCI Model b) Hysteresis Model and Its Defining Points Subsequently, a model of the whole prototype is designed by combining the models of the series transformer and the MCI.They are parameterized and connected to the threephase circuit as shown in Figure 4. To identify the parameters of the series transformer and the MCI, voltage and current measurements at their primary and secondary windings are done by open-circuit and short-circuit tests.For the MCI this tests are carried out by different control currents.In the following, the parameter determination of the MCIs cross-branch is considered in more detail since only this part contains non-linear elements.

Figure 9 a
Figure 9 a) Voltage Variation b) Efficiency c) Voltage Angle d) THD Figure 9c) shows the angle between the feeder and grid side voltages.A high angle between these voltages can lead to issues when the device is switched on or off because there will be a voltage phase step in the network.