**An Alternative Carrier-Based Implementation of Space Vector Modulation to Eliminate Common Mode Voltage in a Multilevel Matrix Converter**

#### **Janina Rz ˛asa**

Department of Power Electronics and Power Engineering, Rzeszow University of Technology, 35-959 Rzeszow, Poland; jrzasa@prz.edu.pl; Tel.: +48-017-865-1976

Received: 13 December 2018; Accepted: 30 January 2019; Published: 6 February 2019

**Abstract:** The main aim of the paper is to find a control method for a multilevel matrix converter (MMC) that enables the elimination of common mode voltage (CMV). The method discussed in the paper is based on a selection of converter configurations and the instantaneous output voltages of MMC represented by rotating space vectors. The choice of appropriate configurations is realized by the use of space vector modulation (SVM), with the application of Venturini modulation functions. A multilevel matrix converter, which utilizes a multilevel structure in a traditional matrix converter (MC), can achieve an improved output voltage waveform quality, compared with the output voltage of MC. The carrier-based implementation of SVM is presented in this paper. The carrier-based implementation of SVM avoids any trigonometric and division operations, which could be required in a general space vector approach to the SVM method. With use of the proposed control method, a part of the high-frequency output voltage distortion components is eliminated. The application of the presented modulation method eliminates the CMV in MMC what is presented in the paper. Additionally, the possibility to control the phase shift between the appropriate input and output phase voltages is obtained by the presented control strategy. The results of the simulation and experiment confirm the utility of the proposed modulation method.

**Keywords:** multilevel matrix converter; rotating voltage space vector; common move voltage; space vector pulse width modulation; venturini control method

#### **1. Introduction**

A multilevel matrix converter (MMC) is a frequency converter, whose topology [1–8] was proposed by analogy to multilevel inverters and its aim is the reduction of the voltage rating of the switches with respect to the supply voltages and the further improvement of the synthesized current and voltage waveforms. Two scientific centres paid attention to the analysis of MMC operations and two different control methods were developed there. The authors of papers [5–8] concentrate on the use of the space vector modulation (SVM) method, whereas in papers [2–4], the implementation of the Venturini control method is presented. The main goal in using either of the modulation methods in controlling MMC is the synthesis of the referenced sinusoidal output voltage and sinusoidal input current by controlling the input displacement angle. The application of these methods is involved with appearing CMV on the output terminals. The problem with appearing CMV is concerned with all the converters being controlled by the use of the pulse width modulation (PWM) method, both indirect frequency converters with a DC link, as well as a direct matrix converter (MC) and MMC. Because the topology of MMC is the modification of conventional MC, the analysis of the cancelation methods of CMV used in MC would be valuable in finding the control method to eliminate the CMV in MMC.

As for MC, many different methods have been reported to mitigate the detrimental influences of the CMV. The majority of the methods are based on a modification of SVM by the elimination of

the zero-space vector, as a complement of the switching cycle and replacement of the zero vectors by rotating space vectors or by active vectors, with minimum absolute values [9–12]. The authors of References [9,10] found that, in MC controlled by the use of SVM using rotating vectors, instead of zero vectors, had 42% lower CMV. The voltage transfer ratio (VTR) was found to be higher than 0.5. Next, in Reference [12], the controlling method of CMV reduction is achieved by using the switch configurations that connect each input phase to a different output phase, which means that rotating space vectors are used or the configurations connect all the output phases to the input phase, with the minimum absolute voltage. The authors found that the result of the CMV peak value reduction was 45.4%, while the VTR was 0.5% and 42.3%, when the VTR was higher. The next is the method that targeted the operation of the drive for a higher modulation index range (0.577 ≤ m ≤ 0.866) [12]. This method eliminates the zero vectors but continues to use active voltage vectors, with normalized duty ratios. The elimination of zero vectors reduces the peak value of CMV by 42%. Even though these methods, with their own modulation strategies, can produce a sound output performance within the specified operating range, they are applicable only to a limited VTR range. To achieve a sound output performance for the whole VTR range, these different modulation strategies should be properly combined. However, it is inconvenient to combine each method with the different switching patterns, because each modulation method, with its own formulae, uses different vectors to calculate the duty cycles. The main goal of the article is to present the control method that results in the elimination of CMV in MMC. The entire elimination of CMV in MC and in MMC is possible only by the use of such configurations, resulting from on-off states of bidirectional switches, which realize the rotating voltage space vectors.

In the application to MC, the use of the only rotating voltage space vectors to obtain the entire elimination of CMV is presented in References [13–19]. The author of Reference [13] compares the CMV in MC, controlled by the use of the Venturini method, which solely applies rotating space vectors, using the scalar control method and SVM control method, applying active and zero-space vectors. To obtain the cancelation of CMV in MC, the authors of Reference [14] introduce the new SVM and develop the modification of four-step commutation. The modification of the four-step commutation is dictated by the fact that, during the four-step commutation, such switch configurations arise, which are represented by active space vectors, what results in the high value of CMV. The SVM technique, solely using rotating space vectors, is also applied by the authors of References [15,19] in the modulation of dual MCs. Next, in Reference [20], the authors present a carrier-based implementation of SVM for dual MCs using only rotating space vectors. The advantage of the proposed strategy is an alternative way to achieve SVM, which does not involve the knowledge of space vectors, when it is derived. Additionally, it avoids any trigonometric and division operations that could be needed to implement the SVM using the general space vector approach.

In reference to MMC, an elimination of CMV is discussed in References [21,22] but the method presented there does not rely on the use solely of rotating space vectors in the synthesis of the output voltage. The authors of References [21,22] use space vector modulation, applying active and zero space voltage vectors. This method demands many trigonometric and division operations. The method analysed in References [21,22] results in only a 50% reduction in the peak value of CMV.

To solve the problem of the elimination of CMV in MMC, the author of the paper proposes the application of the modulation method, using solely rotating space vectors. To determine the switch duty cycles of MMC, the carrier-based implementation of SVM was used, which makes the application of the modulation method easy. The advantage of the proposed strategy is an alternative way to achieve SVM, which does not involve the knowledge of space vectors, when it is implemented. Additionally, it avoids any trigonometric and division operations that could be needed to implement the SVM, using the general space vector approach. The elaboration of the proposed method required the analysis of admissible switch configurations and the characteristics of their corresponding space vectors. The switch configurations and space vectors, providing the synthesis of the required output voltage and the elimination of CMV, were selected and presented in the paper.

In the MMC controlled by the implementation of the proposed method, the output voltage containing the fundamental component and high-frequency distortion components, with considerably less amplitudes, compared with the distortion component in the output voltage of MC, is synthesized. The next advantage of the method is that it obtains, in simulation tests, the entire cancelation of CMV and the reduction of the peak value of CMV to 12% of the peak value of the output voltage. This paper also presents the possibility to control the phase shift between the output voltage and appropriate input voltage. Controlling the phase shift between the output and input voltage may be important when working with the same output and input frequency of MMC and it could be applied as a converter in a Flexible AC Transmission System (FACTS) [23] to control the power flow or compensate the voltage dips.

The paper is organized in a total of five sections. The second section describes the topology of MMC, the admissible configuration of the analysed converter and the output voltage space vectors. The proposed modulation method is described in the third section. The fourth section contains the simulation and experimental results to validate the proposed method and the last section has the discussion.

#### **2. Multilevel Matrix Converter**

#### *2.1. Topology of MMC*

A multilevel matrix converter (MMC) (Figure 1) consists of 18 bidirectional switches, SAa1 – SCc2 and nine clamp capacitors, C1 – C9. The bidirectional switches constitute two integral semiconductor modules. The manufacturers, Yaskawa, ABB, Alstom, Siemens and so forth, have shown their interest in the production of these semiconductor power modules, consisting of bidirectional switches. In fact, Yaskawa has introduced many standard units of matrix converters of medium voltages and several megawatts.

**Figure 1.** Scheme of multilevel matrix converter (MMC).

The clamp capacitors in MMC are connected by two switch matrixes and play a role analogous to flying capacitors in multicell converters, principally providing an additional intermediate voltage level in the process of synthesizing the output voltage. The capacitance of clamp capacitors depends on the value of the load current, switching the frequency and assumed value of the voltage ripple on the clamp capacitors [1–3]. Based on studies [1] and [24–26], concerned with the multicell converter, a capacitance value (1) of the capacitor should be determined using the admissible value of the voltage ripple on that capacitor Δ*UC*, number *p* of multilevel converter cells, load current *Io* and switching frequency *fs* of the converter switches.

$$\mathcal{C} = \frac{I\_o}{\Delta l l\_{\mathcal{C}} p f\_s} \tag{1}$$

To eliminate improper clamp capacitor voltages, shaping additional balancing circuits are used in MMC. The balancing circuit, providing an automatic maintenance of the clamp capacitor voltages, consists of *Rb* resistance, *Lb* inductance and *Cb* capacitance, connected in a series. The parameters of the balancing circuits are selected so that the resonance frequency *fo* of those circuits is equal to the switching frequency *fs* of the switches (2).

$$f\_{\mathcal{o}} = \frac{1}{2\pi\sqrt{L\_b C\_b}} = f\_{\mathcal{s}} \tag{2}$$

The resonance frequency is the most important parameter of the balancing circuit but the effectiveness of the balancing process, with a passive RLC circuit, depends also on the balancing circuit characteristic impedance, the converter parameters and the converter operating conditions [26,27].

#### *2.2. Admissible Switch Configurations of MMC*

Taking into account the voltage characteristics of the MMC supplying source, the switch configurations that do not make short circuit of the input phases and simultaneously assure the current path in the resistive-inductive load, are admissible. Therefore, each of the three output phases could be connected into one of the input phases directly through two series-connected bidirectional switches or through two switches and a clamp capacitor. Clamp capacitors C1 – C9 perform additional intermediate voltage levels in waveforms of the output voltage. In the ideal conditions of charging, voltages on the clamp capacitors should be equal to half of the appropriate line-line voltages (3).

$$\begin{aligned} \mu\_{\rm C1} = \mu\_{\rm C4} = \mu\_{\rm C7} &= (\mu\_{\rm a} - \mu\_{\rm b})/2 \\ \mu\_{\rm C2} = \mu\_{\rm C5} = \mu\_{\rm C8} &= (\mu\_{\rm b} - \mu\_{\rm c})/2 \\ \mu\_{\rm C3} = \mu\_{\rm C6} = \mu\_{\rm C9} &= (\mu\_{\rm c} - \mu\_{\rm a})/2 \end{aligned} \tag{3}$$

As a result of the use of clamp capacitors, there exist additional current flow paths between the input and output phases. Taking into account only the admissible states of the bidirectional switches of MMC, each of the output phases may be connected with the three supply phases in nine different ways, corresponding to the converter configurations. These configurations arise from the 'on' or 'off' states of the bidirectional switches, which, for the output phase A, are shown in Table 1.

**Table 1.** Switch configuration of MMC in the output phase A.


In the case of the first three configurations (1, 2, 3), shown in Table 1, every output phase is directly connected with one of the input phases, which is characteristic of a conventional MC. This means that the phase output voltage is equal the appropriate phase input voltage. The next six configurations (4, 5, 6, 7, 8, 9) implement the connections across clamp capacitors. For instance, while the switches SAa1 and SAb2 (configuration 4) or switches SAb1 and SAa2 (configuration 5) are 'on,' the current of the output phase A is running across the capacitor C1 and connected in parallel capacitors C2 and C3. The instantaneous voltage of the output phase A is the same in both cases and equals *uC*<sup>1</sup> *= (ua + ub)/*2, because of the voltage of the capacitor C1 and because it is connected in parallel capacitors C2 and C3, equalling *(ua* − *ub)/*2. The choice between configurations 4 and 5 is the choice between different capacitor current directions, allowing for the possibility to control the capacitor voltage. The same regularity is fulfilled in the next two couples of configurations, that is, configurations 6 and 7 or 8 and 9. Recapitulating, we can conclude that each output phase can be connected to *ua*, *ub*, *uc*, as the voltage supply (henceforth termed "full-amplitude voltage supply") or to (*ua* + *ub*)/2, (*ub* + *uc*)/2, (*ua* + *uc*)/2 (henceforth, "half-amplitude voltage supply").

Considering a three-phase to three-phase MMC, one has to take into account 93 = 729 possible switch configurations, which can be used practically in the process of the synthesis of output voltages and the synthesis of the voltages of clamp capacitors, determining the intermediate levels of the supply voltages.

#### *2.3. Output Voltage Space Vectors in MMC*

The output voltages in three-input to three-output circuit of MMC could be presented using space vectors, as defined by (4). Instantaneous output voltages *uA, uB* and *uC* in the relation (4) are appropriately equal to one of values defined in the output phase A in Table 1. The label *xxx* in the name <sup>→</sup> *Vxxx* of the space vector means the type of configuration appropriately chosen in the output phase A, B and C.

$$\stackrel{\rightarrow}{V}\_{xxx} = \frac{2}{3} \left[ u\_A + au\_B + a^2 u\_C \right] \tag{4}$$

Analysis of the output voltages, corresponding to 729 switch configurations, allows for it to be noticed that the instantaneous output voltage space vectors could be split into the following groups:


The active voltage space vectors correspond to the connection of two output phases to the same voltage supply. The zero vectors arise when the output phases are connected with the same "full-amplitude voltage supply" or the same "half-amplitude voltage supply."

#### *2.4. Output Voltage Space Vectors Reducing CMV in MMC*

Among 342 configurations of MMC, with instantaneous output voltages represented by rotating space vectors, only 54 could be considered to reduce CMV. This conclusion is drawn from the analysis of rotating space vectors, determined with the assumption that the MMC is supplied by a balanced input voltage (5).

$$
\begin{bmatrix} u\_d \\ u\_b \\ u\_c \end{bmatrix} = \begin{bmatrix} \mathcal{U}\_{\text{im}} \cos \omega\_i t \\ \mathcal{U}\_{\text{im}} \cos(\omega\_i t - 120^\circ) \\ \mathcal{U}\_{\text{im}} \cos(\omega\_i t + 120^\circ) \end{bmatrix} \tag{5}
$$

All 342 rotating voltage space vectors representing instantaneous output voltages could be split into the following five groups:

• Rotating voltage space vectors, with a constant module equal to the amplitude of the input voltage *Uim*. These vectors correspond to switch configurations, where three output voltages are synthesized by the use of three full-amplitude supplying voltages. Six vectors belonging to this group create two sets, consisting of three rotating space vectors shifted by 120º. One set rotates in a positive direction (CCW vectors) along a complex plane and the next set rotates in a negative one (CW vectors). Two of these vectors are represented by Equations (6) and (7), as well as the relation (6)—for CCW vectors and (7)—for CW vectors. The digits in the label of the voltage vector name should be interpreted as follows: the first digit defines the configuration of the switches in the output phase A, the second and third digits, the output phase B and C, appropriately. The application of the rotating space vectors belonging to this group in the modulation of switch duty cycles results in a zero value of CMV (8).

$$
\stackrel{\rightarrow}{V}\_{123} = \frac{2}{3} \left[ u\_a + a u\_b + a^2 u\_c \right] = \mathcal{U}\_{im} e^{i\omega\_i t} \tag{6}
$$

$$\overrightarrow{\dot{V}}\_{132} = \frac{2}{3} \left[ u\_a + a u\_c + a^2 u\_b \right] = \mathcal{U}\_{im} e^{-j\omega\_l t} \tag{7}$$

where: *a* = *ej*120◦ , *a*<sup>2</sup> = *ej*240◦

$$
\mu\_{CMV(123)} = \frac{\mu\_a + \mu\_b + \mu\_c}{3} = 0 \tag{8}
$$

• Rotating voltage space vectors, with a constant module equal to half of the input phase voltage amplitude that corresponds to 48 configurations with a connection of three output phases to three different half-amplitude voltage supplies. Half of these vectors complete 8 sets of three vectors rotating in a positive direction (CCW vectors) and half of them form 8 sets of vectors rotating in a negative one (CW vectors). Two of these vectors are assigned as (9) (CCW vectors) and (10) (CW vectors). The application of the rotating space vectors belonging to this group in the modulation of switch duty cycles also results in a zero value of CMV (11).

$$\stackrel{\rightarrow}{V}\_{468} = \frac{2}{3} \left[ \frac{1}{2} (\boldsymbol{u}\_{\boldsymbol{d}} + \boldsymbol{u}\_{\boldsymbol{b}}) + a \frac{1}{2} (\boldsymbol{u}\_{\boldsymbol{b}} + \boldsymbol{u}\_{\boldsymbol{c}}) + a^2 \frac{1}{2} (\boldsymbol{u}\_{\boldsymbol{d}} + \boldsymbol{u}\_{\boldsymbol{c}}) \right] = -\frac{1}{2} l I\_{\rm im} e^{i(\omega\_{\boldsymbol{t}} t - 120^{\circ})} \tag{9}$$

$$\stackrel{\rightarrow}{V}\_{486} = \frac{2}{3} \left[ \frac{1}{2} (u\_d + u\_b) + a \frac{1}{2} (u\_d + u\_c) + a^2 \frac{1}{2} (u\_b + u\_c) \right] = -\frac{1}{2} l l\_{\text{int}} e^{-j(\omega\_i t + 120^\circ)} \tag{10}$$

$$
\mu\_{\rm CMV(468)} = \frac{\frac{1}{2}(u\_a + u\_b) + \frac{1}{2}(u\_b + u\_c) + \frac{1}{2}(u\_a + u\_c)}{3} = 0 \tag{11}
$$

• Rotating voltage space vectors, with a constant module equal to half of the input phase voltage amplitude, which corresponds to 72 configurations, with a connection of two output phases to two different half-amplitude voltage supplies and a third output phase connected to the full-amplitude voltage supply. Two examples are shown: CCW vector, as Equation (12) and CW vector, as Equation (13). The application in the modulation of switch duty cycles the rotating space vectors, belonging to the group being discussed, results in a value of CMV (14) that is not zero.

$$\stackrel{\rightarrow}{V}\_{148} = \frac{2}{3} \left[ u\_a + a \frac{1}{2} (u\_a + u\_b) + a^2 \frac{1}{2} (u\_a + u\_c) \right] = \frac{1}{2} l l\_{im} e^{i\omega\_i t} \tag{12}$$

$$\stackrel{\rightarrow}{V}\_{184} = \frac{2}{3} \left[ u\_a + a \frac{1}{2} (u\_a + u\_c) + a^2 \frac{1}{2} (u\_a + u\_b) \right] = \frac{1}{2} l I\_{\text{im}} e^{-j\omega\_l t} \tag{13}$$

$$
\mu\_{CMV(148)} = \frac{u\_d + \frac{1}{2}(u\_d + u\_b) + \frac{1}{2}(u\_d + u\_c)}{3} = \frac{1}{2}u\_d\tag{14}
$$

• Rotating voltage space vectors, with a changeable module, that correspond to 72 configurations, with a connection of two output phases to two different full-amplitude voltage supplies and a third output phase connected to a half-amplitude voltage supply. The Equation (15) is an example of the vectors belonging to this group. The application of the rotating space vectors, belonging to this group, in the modulation of switch duty cycles, results in a value of CMV (16) that is not zero.

$$\begin{split} \stackrel{\rightarrow}{V}\_{126} &= \frac{2}{3} \Big[ u\_a + a u\_b + a^2 \frac{1}{2} (u\_b + u\_c) \Big] = \mathcal{U}\_{im} e^{j\omega\_l t} + \frac{1}{3} a^2 u\_{bc} \\ &= \mathcal{U}\_{im} e^{j\omega\_l t} + \frac{\sqrt{3}}{3} \mathcal{U}\_{im} \sin\omega\_l t \, e^{j240^\circ} \end{split} \tag{15}$$

$$
\mu\_{CMV(126)} = \frac{u\_a + u\_b + \frac{1}{2}(u\_b + u\_c)}{3} = \frac{1}{6}u\_{bc} \tag{16}
$$

• Rotating voltage space vectors, with a changeable module, that correspond to 144 configurations, with a connection of two output phases to two different half-amplitude voltage supplies and a third output phase connected to a full-amplitude voltage supply. The Equation (17) is an example of the vectors belonging to this group. The application of the rotating space vectors, belonging to this group, in the modulation of switch duty cycles, results in a value of CMV (18) that is not zero.

$$\begin{split} \overrightarrow{V}\_{146} &= \frac{2}{3} \left[ u\_a + a \frac{1}{2} (u\_a + u\_b) + a^2 \frac{1}{2} (u\_b + u\_c) \right] = \\ &= \mathcal{U}\_{im} e^{i\omega\_i t} - \frac{\sqrt{3}}{3} \mathcal{U}\_{im} \sin(\omega\_i t - 60^\circ) e^{i120^\circ} + \frac{\sqrt{3}}{3} \sin \omega\_i t \, e^{i240^\circ} \end{split} \tag{17}$$
 
$$\begin{split} \varepsilon\_{12} &= \varepsilon\_{12} \varepsilon\_{23} \end{split} \tag{18}$$

$$
\mu\_{CMV(146)} = \frac{u\_a + \frac{1}{2}(u\_a + u\_b) + \frac{1}{2}(u\_b + u\_c)}{3} = \frac{1}{6}u\_{ac} \tag{18}
$$

Finally, only 54 of the 729 switch configurations of MMC could be chosen, while the CMV elimination is required in the proposed control method. The lay-out of a complex plane of the rotating space vectors, belonging to the mentioned groups, is shown in Figure 2. In Figure 2, the initial position of the rotating space vectors is shown. The digits used in the label of vectors, instead of the full names of vectors, are shown in the figure.

**Figure 2.** Lay-out of a complex plane of the voltage rotating space vectors, whose implementation assures the elimination of CMV.

The performed analysis of all of the allowable configuration of MMC and the corresponding voltage space vectors allows the vectors synthesizing the sinusoidal output voltage, sinusoidal input current and elimination of CMV to be chosen. As a result, the proposed modulation method is based on the application solely of the switch configurations that correspond to the rotating space vectors, with a constant module. The rotating space vectors are not usually used in modulation strategies, because they lay in different positions, so it is difficult to create a repetitive pattern. However, in Reference [13], one can find that the implementation of the Venturini modulation function in the determination of switch duty cycles in conventional MC could provide the control using only rotating space vectors, while simultaneously eliminating CMV. By analogy, the Venturini modulation functions are used in the proposed strategy for modulation duty cycles in MMC.

#### **3. Proposed modulation method**

#### *Method of the Output Voltage Synthesis*

The synthesis of the output voltage by the use of rotating space vectors could be realized by applying a carrier-based implementation of SVM. The Venturini modulation functions are used in the proposed method to set out the switch duty cycles. The determination of the duty cycles for bidirectional switches of conventional MC and MMC by the use of Venturini modulation functions has been presented in References [13] and [2,3]. Here, the Venturini modulation function, in consideration of angle *ψ*, in the form of (19) or (20), is used. A value of angle *ψ* defines the phase shift between that defined by (5) input phase voltages and the appropriate output phase voltages. The application of the modulation function (19) results in CCW output voltage rotating space vectors and a lagging input displacement angle, whereas the modulation function (20) gives CW output voltage rotating space vectors and leading input displacement angle [2]. Both of them, that is, the modulation function (19), depending on the difference *ω<sup>o</sup>* − *ω<sup>i</sup>* of the output and input frequency and modulation function (20), depending on the sum *ω<sup>o</sup>* + *ω<sup>i</sup>* result in the output voltage amplitude, equal to half of the input voltage amplitude, at most.

$$\begin{array}{l} d\_1^- = m\_{Aa}^- = m\_{Bb}^- = m\_{Cc}^- = \frac{1}{5} \left( 1 + 2k\_{lI} \cos \left( \omega\_o - \omega\_i \right) \, t + \Psi \right) \\\ d\_2^- = m\_{Ab}^- = m\_{Bc}^- = m\_{Ca}^- = \frac{1}{5} \left( 1 + 2k\_{lI} \cos \left( \left( \omega\_o - \omega\_i \right) \, t - \frac{2\pi}{3} + \Psi \right) \right) \\\ d\_3^- = m\_{Ac}^- = m\_{Ba}^- = m\_{Cb}^- = \frac{1}{5} \left( 1 + 2k\_{lI} \cos \left( \left( \omega\_o - \omega\_i \right) \, t + \frac{2\pi}{3} + \Psi \right) \right) \end{array} \tag{19}$$

$$\begin{array}{l} d\_1^+ = m\_{Aa}^+ = m\_{Bc}^+ = m\_{Cb}^+ = \frac{1}{3} (1 + 2k\_{l\ell} \cos(\omega\_o + \omega\_i) \, t + \Psi) \\\ d\_2^+ = m\_{Ab}^+ = m\_{Ba}^+ = m\_{Ca}^+ = \frac{1}{3} (1 + 2k\_{l\ell} \cos\left(\left(\omega\_o + \omega\_i\right) \, t - \frac{2\pi}{3} + \Psi\right)) \\\ d\_3^+ = m\_{Ac}^+ = m\_{Bb}^+ = m\_{Ca}^+ = \frac{1}{3} (1 + 2k\_{l\ell} \cos\left(\left(\omega\_o + \omega\_i\right) \, t + \frac{2\pi}{3} + \Psi\right)) \end{array} \tag{20}$$

As carrier signals, two-phase shifted carrier signals are adopted. The displacement of carrier signals, involved in the control of switches Sij1 and switches Sij2, is *Ts*/2, where *Ts* is the carrier signal period. Duty cycles, in which switches Sij1 are switched-on, arise from the comparison of the corresponding modulation functions with one of the carrier signals. A carrier signal shifted by half of the switching cycle *Ts* determines the duty cycles of Sij2 switches (Figure 3). The digits in Figure 3, placed below the duty cycles, mean the numbers of configurations defined in Table 1. At the bottom in Figure 3, the names of the appropriate space vectors synthesizing the load voltage are placed. The mentioned vectors are the same as the rotating space vectors shown in Figure 2.

The application of modulation functions, with phase shift angle *ψ*, provides the phase shift between the output and the appropriate input voltage. This feature is important when the MMC works with the same input and output frequency and it allows the MMC to be used as a converter in Flexible AC Transmission System (FACTS) devices.

**Figure 3.** Switch duty cycles and rotating space vectors of the output voltage in MMC for CCW-type space vectors (**a**) and CW-type space vectors (**b**).

#### **4. Simulation and Experiment**

#### *4.1. Simulation*

Simulation tests were realized in an EMTP-ATP program. The matrix of the bidirectional switches was modelled as a matrix of the ideal bidirectional switches, controlled using signals generated in TACS subroutine. The supply grid is represented by ideal sinusoidal voltage sources, with an RMS value of 220 V and a frequency of 50 Hz, while the load consists in star-connected resistance and inductance elements, with values of 2 Ω and 10 mH. The carrier frequency was *fcarr* = 5 kHz. The capacitance of the clamp capacitors is equal 10μF. The performances obtained for MMC, controlled by the use of the carrier-based implementation of SVM combined with the Venturini modulation functions, are presented in Figures 4–7. The waveforms of the output voltage and CMV, as depicted in Figure 4 and in Figure 5, illustrate the control, with a lagging input displacement angle (CCW rotating space vectors). Waveforms in Figures 6 and 7 correspond to the leading input displacement angle, when the CW rotating space vectors are used. One can see that in both cases, the CMV is equal to zero. All these waveforms illustrate the control, with a value of the shift angle of *ψ* = 0. The Fourier analysis was performed, with an accuracy of 10 Hz. It could be observed that, besides fundamental harmonics, the output voltage consists of high frequency components, concentrated as sidebands around each multiple of the carrier frequency. However, it is also seen that, in comparison with conventional MC (Figure 8b), the amplitudes of the first group of these harmonics are significantly decreased. The comparison of the distortion components of the first groups in MC and MMC output voltages (Figures 4b and 8b), obtained with the same controlling parameters, indicates that the amplitude of these components decreased from 180.9 V in MC to 8.7 V in MMC. The waveforms chosen for presentation and FFT analysis prove that the applied modulation method results in a significant reduction of the distortion components of the MMC output voltage, compared with the MC output

voltage, which is a basic demand of a proper controlling method used for MMC. The second important feature of the proposed modulation method is the entire cancelation of CMV.

\ **Figure 4.** Waveforms of the output voltages (red lines) and CMV (green lines) (**a**) and Fourier analysis of the output voltage (**b**), synthesized in MMC and controlled by the use of CCW rotating space vectors, for the angle *ψ* = 0◦ and output frequency of 50 Hz. \

\ **Figure 5.** Waveforms of the output voltages (red lines) and CMV (green lines) (**a**) and Fourier analysis of the output voltage (**b**), synthesized in MMC and controlled by the use of CCW rotating space vectors, for the angle *ψ* = 0◦ and output frequency of 80 Hz.

\

\

\ **Figure 6.** Waveforms of the output voltages (red lines) and CMV (green line) (**a**) and Fourier analysis of the output voltage (**b**), synthesized in MMC and controlled by the use of CW rotating space vectors, for the angle *ψ* = 0◦ and output frequency of 50 Hz.

The next analysis (Figures 9 and 10) deals with the control of the phase shift between the output and input phase voltages. MMC works with the output frequency the same as it does with an input one. In Figure 9, the waveform of the output voltages, together with the appropriate input voltages, for different shift angle values *ψ* between the input and output voltages, is shown. Performed simulation analysis proves that the control of the phase shift between the output and input voltage in MMC, controlled by the use of the carrier-based implementation of SVM, with Venturini modulation functions, is possible and is characterized by a linear relation (Figure \10).

\ **Figure 7.** Waveforms of the output voltages (red lines) and CMV (green line) (**a**) and Fourier analysis of the output voltage (**b**), synthesized in MMC and controlled by the use of CW rotating space vectors, for the angle *ψ* = 0◦ and output frequency of 80 Hz.

**Figure 8.** Waveforms of the output voltage (red lines) and CMV (green line) (**a**) and Fourier analysis (**b**) of the output voltage, synthesized in conventional MC by the use of CCW rotating space vectors, for the output frequency of 50 Hz.

() **Figure 9.** *Cont*.

**Figure 9.** Waveform of the output voltages (red lines), input voltages (blue line) and CMV (green line) in MMC, controlled by the use of CCW rotating space vectors, for the output frequency of 50 Hz and angle *ψ* = 60◦ (**a**); angle *ψ* = 6−0◦ (**b**) and angle *ψ* = 180◦ (**c**).

**Figure 10.** Angle shift between the output and input voltage versus the reference angle *ψ*.

#### *4.2. Experiment*

To further verify the proposed control method, measurement tests were performed. The experimental parameters are shown in Table 2.


**Table 2.** Experimental Parameters.

Presented in Figures 11–13, waveforms represent the chosen results of measurements. They were registered in MMC, controlled by the use of modulation function (19) and were therefore the CCW-type rotating voltage space vectors. In Figure 11, the output voltage, together with the appropriate input voltage, is shown. The referenced angle shift is equal to 60◦, −60◦ or 180◦ and the measurements confirm the accomplishment of these values. In Figure 12, the synthesized output voltage and CMV are shown. The maximum instantaneous values of CMV appear when the angle shift *ψ* is equal to −60◦ (Figure 12b). These maximum values of CMV do not exceed 40 V (Figure 13b), which is less than 12% of the peak values of the output voltage and amplitude of the input voltage.

(a)

(b)

**Figure 11.** *Cont*.

(c)

**Figure 11.** Waveform of the input phase voltage (orange) and output phase voltage (blue) for the referenced shift angle *ψ* = 60◦ (**a**); *ψ* = 6−0◦ (**b**) and *ψ* = 180◦ (**c**).

(a)

**Figure 12.** *Cont*.

(c)

**Figure 12.** Waveform of the output phase voltage (blue) and CMV (orange), for the shift angle *ψ* = 60◦ (**a**); *ψ* = 6−0◦ (**b**) and *ψ* = 180◦ (**c**).

()

() **Figure 13.** *Cont*.

()

**Figure 13.** The enlarged waveform of the output phase voltage (blue) and CMV (orange), for the shift angle *ψ* = 60◦ (**a**); *ψ* = 6−0◦ (**b**) and *ψ* = 180◦ (**c**).

#### **5. Discussion**

The proposed carrier-based SVM, using Venturini modulation functions, is valuable because it realizes control whilst improving the waveform of the MMC output voltage compared with the MC output voltage. In MMC, controlled using the proposed method, the amplitudes of the first group of output voltage distortion components, concentrated near the first multiple of the carrier frequency, are only near 5% of the appropriate amplitude of output voltage distortion components in MC.

An important achievement, obtained by implementing the proposed modulation method, is the entire elimination of CMV, which was confirmed by the results of simulation tests. In the experiment, the peak value of CMV is less than 12% of the amplitude of the supplying voltage and the peak value of the output voltage. The occurrence of the CMV, measured higher than zero, may be explained by the noise activated by the four-step commutation process of the bidirectional switches. During the short period of the commutation steps, the active space vectors can appear and cause a higher than zero CMV. To avoid the problem with commutation noise, the four-step commutation in the experimental model of MMC should be modified in future research.

An additional advantage of the presented modulation method, not presented until now in the papers concerned with analysis of MMC, is the possibility to control the phase shift between the output voltage and the appropriate input voltage.

The drawback of the proposed controlling method is the fact that, using only rotating space vectors, the elimination of CMV is possible; this is concerned with the application of modulation functions, which determine the constant input displacement angle between the input voltage and the phase current. On the other hand, this feature may be utilized in FACTS devices that realize series compensation when the input terminals of MMC are connected with the supply network in a shunt manner. MMC, controlled using CW rotating space vectors (modulation function (20)) at the input terminals, draws the current which precedes the appropriate supplying voltage, so MMC works as a source of reactive power for the AC system and also realizes shunt compensation.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **Abbreviations**


#### **References**


© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **A Compound Current Limiter and Circuit Breaker**

#### **Amir Heidary 1, Hamid Radmanesh 2, Ali Bakhshi 1, Kumars Rouzbehi <sup>3</sup> and Edris Pouresmaeil 4,\***


Received: 18 April 2019; Accepted: 7 May 2019; Published: 16 May 2019

**Abstract:** The protection of sensitive loads against voltage drop is a concern for the power system. A fast fault current limiter and circuit breaker can be a solution for rapid voltage recovery of sensitive loads. This paper proposes a compound type of current limiter and circuit breaker (CLCB) which can limit fault current and fast break to adjust voltage sags at the protected buses. In addition, it can act as a circuit breaker to open the faulty line. The proposed CLCB is based on a series *L-C* resonance, which contains a resonant transformer and a series capacitor bank. Moreover, the CLCB includes two anti-parallel power electronic switches (a diode and an IGBT) connected in series with bus couplers. In order to perform an analysis of CLCB performance, the proposed structure was simulated using MATLAB. In addition, an experimental prototype was built, tested, and the experimental results were reported. Comparisons show that experimental results were in fair agreement with the simulation results and confirm CLCB's ability to act as a fault current limiter and a circuit breaker.

**Keywords:** circuit breaker; fault current limiter

#### **1. Introduction**

Faults in electrical power systems are inevitable. They can lead to high transients and thermal stresses on power system equipment such as overhead lines, cables, transformers, and switchgears. Therefore, the fault current protection schemes are important. The simplest solution to limit the short-circuit current would be the application of a source with high impedance. The main drawback of this solution is that it also influences the system during normal operation conditions, and it results in a considerable voltage drop for high current loads [1,2]. Therefore, electric networks require efficient and reliable equipment to limit the short-circuit current. Another solution to this problem is the use of technologies such as fault current limiters (FCLs). The FCL is one of the protection devices, which is used to limit the fault current. The FCL should limit the fault current passing through it within the first half-cycle and the best FCL should limit the fault current before the first peak [3]. However, high price, power losses, continuous current after fault current flow limitation, and harmonic distortion are some of the main problems of typical FCLs. Since the 1970s, several types of FCLs have been investigated such as fuses with fault-current limitation, series current limiting reactors [4], series transformers [5], superconducting fault current limiters (SCFCL) [6–8], solid-state FCLs (SSFCL) [9–13], and fault current limiting circuit breakers (FCLCB). In the recent years, researchers have focused on the SSFCLs and FCLCBs, such as: Purely resistive FCL [14], hybrid-resistive FCL [15], saturable core FCL [16], IGBTs controlled series reactor FCL [17], solid-state FCLCB (SSFCL-CB) [18], and bridge type

FCL [19]. These new protection devices usually use inductors to decrease the fault current. In these structures, the reactor is ignorable during the normal operation mode and has a fixed impedance during the fault episode, which decreases the system fault current and in some cases can improve the system stability [19]. The influence of the FCL on the short-circuit level of the substation bus bar splitter circuit breaker has been investigated in [20,21]. A rectifier-type SFCL with non-inductive reactor has been reported in [22]. In [23], the power electronic switches selection for 20 kV distribution network application are discussed. A DC circuit breaker for voltage source converter (VSC) has been proposed in [24]. The fast-closing switch application in solid-state circuit breaker and its optimization process has been studied in [25]. Application of current-limiting circuit breakers to control the arc-flash energy has been presented in [26]. Classification of solid-state circuit breakers and application of solid-state circuit breaker, to improve grid voltage quality during the fault is reported in [27]. In [28], the comparison of two control methods of power swing reduction in a power system with unidirectional power flow controller (UPFC) is discussed. Analysis and control of fault current by firing angle control of solid-state fault current limiter is an important issue which depends on the strategy of power electronic switch control [29].

This paper presents a new type of current limiter circuit breaker (CLCB) with series compensation. This protection device is invisible during the normal operation mode. During the fault period, it disconnects the loads from the source. The operational effectiveness of this device is verified by MATLAB simulations and confirmed by the developed experimental tests. The results show the fast-closing switch based CLCB has more advantages than the former FCLs with low cost and can improve the system protection against fault by fast current limiting and breaking.

Expected advantages of the proposed CLCB over other FCLs are as the following:


This paper has been organized as follows:

In Section 2, the system topology including proposed CLCB is discussed. In Section 3, the analytical analysis of the CLCB operation during normal and fault operation modes, voltage sag at sensitive bus, and power losses are studied. Then, in Section 4, the control system is studied. In the next section, the MATLAB software was used to simulate the operational behavior of the CLCB. In Section 6, experimental results are presented and finally a conclusion is drawn.

#### **2. Electrical Network Modeling**

Figure 1 shows a single line diagram of the power grid, in which CLCB connects bus 3 and bus 4 as bus coupler.

**Figure 1.** Single line diagram of the distribution network.

Bus 3 is assumed to be faulty and bus 4 is connected to the sensitive loads by feeders. The CLCB topology is shown in Figure 2.

**Figure 2.** Current limiting circuit breaker topology.

In this circuit, *D*<sup>1</sup> and *Sb* are power electronics diode and IGBT switch, respectively, and *C*<sup>s</sup> is a series capacitor bank. In addition, the primary side of transformer *T*<sup>1</sup> is connected in series to a line and its secondary is connected to two anti-parallel IGBTs. During normal operation mode, the resonance transformer and series capacitor form a series resonance *L-C* tank with resonance frequency equal to electrical network frequency. In this case, *D*<sup>1</sup> for positive half-cycles and *Sb* for negative half-cycles, are in on-state and voltage drop on the CLCB components is negligible. The CLCB configuration during normal operation mode is shown in Figure 3a.

**Figure 3.** Proposed CLCB topology (**a**) normal operation, (**b**) fault current limiting, and (**c**) fault current breaking.

During a fault, the fault current increases and passes the threshold current level (*I*L). In this case, the control circuit detects the fault and turns on the antiparallel IGBTs. Therefore, the secondary side of the resonance transformer is short-circuited and the resonance transformer shows negligible impedance. The series capacitor impedance then limits the fault current. Figure 3b shows the CLCB topology in the fault current limiting mode. To open the faulty line, the control circuit turns off *Sb* after one cycle delay. In this case, *D*<sup>1</sup> passes a positive half-cycle and the induced DC voltage on the series capacitor charges it. Then, the series capacitor opens the faulty line successfully. The CLCB topology in circuit breaker mode is shown in Figure 3c.

#### **3. Analytical Studies**

#### *3.1. CLCB Operation in Normal Mode*

In this mode, the secondary side of the transformer is open and the series resonance *LC* tank is in resonance condition. Therefore, the electrical network equivalent circuit in steady state condition is an *R-L* circuit where *R* and *L* equal to source, line, and load resistances and inductances, respectively. In addition, the source voltage is denoted with *V*s(*t*) and is equal to *V*<sup>m</sup> *sin*(ω*t*). By applying Kirchhoff law to the network, the line current for the steady-state condition,

$$V\_m \sin(\omega t) = L \frac{d i\_L(t)}{dt} + R i\_L(t) \tag{1}$$

then

$$\dot{q}\_L(t) = \frac{V\_{\text{ff}}}{\sqrt{R^2 + \alpha^2 L^2}} \sin\left(\omega t - \tan^{-1}\frac{\alpha L}{R}\right) \tag{2}$$

Equation (2) shows the sinusoidal nature of the line current during the normal operation mode.

#### *3.2. CLCB Operation in Fault Current Limiting Mode*

During a fault, the resonance transformer is by-passed via IGBTs and the equivalent circuit of the network is an *R-L-C* circuit, where *R* and *L* include the source, line, and CLCB (transformer leakage and magnetization) resistances and inductances, respectively, and *C* is the series capacitor bank. In this case, the *RLC* circuit current can be obtained using the Equation (3)

$$\text{LC}\frac{d^2V\_C(t)}{dt^2} + \text{RC}\frac{dV\_C(t)}{dt} + V\_C(t) = V\_s(t) - \left(V\_D + V\_{IGBT}\right) \tag{3}$$

where initial conditions for *L* is *iL*(0−) = *iL*(0+) = *I*0, for *C* is *VC*(0−) = *VC*(0+) = *V*0, and *V*<sup>D</sup> and *V*IGBT is IGBT voltage drop, respectively

$$\frac{dV\_C(0^-)}{dt} = \frac{i\_L(0^-)}{\mathcal{C}} = \frac{I\_0}{\mathcal{C}}\tag{4}$$

Solving this equation results in the following equation:

$$V\mathcal{E}(t) = e^{-\alpha t} \left( A\_1 \cos \beta t + A\_2 \sin \beta t \right) + \frac{V\_{\text{II}}}{\sqrt{\left(1 - L\mathcal{L}\omega^2\right)^2 + \left(R\mathcal{L}\omega\right)^2}} \sin \left(\omega t + \frac{\pi}{2} + \tan^{-1} \left(\frac{R\mathcal{L}\omega}{\left(1 - L\mathcal{L}\omega^2\right)}\right) \right) - \left(V\_{\text{D}} + V\_{\text{II}}\text{grr}\right) \tag{5}$$

where α = *<sup>R</sup>* <sup>2</sup>*<sup>L</sup>* , <sup>ω</sup><sup>0</sup> <sup>=</sup> <sup>1</sup> <sup>√</sup> *LC*, <sup>β</sup> <sup>=</sup> α<sup>2</sup> − ω<sup>0</sup> 2, and the value of *A*<sup>1</sup> and *A*<sup>2</sup> can be obtained using initial conditions. Then,

$$\dot{m}\_L(t) = \mathcal{C} \Big( \varepsilon^{-at} (A\_1 \cos \beta t + A\_2 \sin \beta t) \Big)' + \frac{\omega V\_m}{\sqrt{\left(1 - L\Omega \alpha^2\right)^2 + \left(R \Omega \omega\right)^2}} \cos \left(\omega t - \tan^{-1} \left(\frac{R \Omega \omega}{\left(1 - L\Omega \alpha^2\right)}\right)\right) \tag{6}$$

The obtained value for *i*L(*t*) includes two-term responses and one steady-state term. The transient responses are dampened after some milliseconds. The steady-state response includes the phase angle shift as shown in the simulation results.

#### *3.3. CLCB Operation in Circuit Breaking Mode*

In this case, the electrical network is in faulty condition and the suggested CLCB should open the faulty line. Therefore, the control system turns off *Sb* and induces the DC voltage on the series capacitor. The charged capacitor then opens the faulty line and the transmission line current reaches zero. In this case, we have:

$$\dot{a}\_L(t) = \mathcal{C} \Big( e^{-\alpha t} (A\_3 \cos \beta t + A\_4 \sin \beta t) \Big)' \tag{7}$$

The Equation (7) includes two exponential parts and, the line current reaches zero.

#### **4. Control Strategy**

The control block diagram of the proposed CLCB is shown in Figure 4. In the normal mode, the *Sb* was in on-state for negative half-cycles and IGBTs were in off-state. Therefore, the line current (*i*L) passed through the series resonance *LC* tank and the CLCB showed negligible impedance.

**Figure 4.** Bock diagram presentation of CLCB control logic.

At fault inception, the *I*<sup>L</sup> becomes greater than the maximum permissible current (*I*ref) and the control circuit turns on the anti-parallel IGBTs and turns off the *Sb* after the one cycle delay. Therefore, the resonance transformer is bypassed and the impedance of the series capacitor limits the fault current. By turning off the *T*s1, the faulty line is opened and the CLCB acts as a circuit breaker. After fault removal, the step generator resets the gates pulses of the power electronics switches and returns the network to the pre-fault condition.

#### **5. Simulation Results**

The single line diagram of the electrical network including CLCB and shown in Figure 1 is simulated. The parameters of the suggested CLCB and electrical network are listed in Table 1. The results are obtained considering a single-phase to the ground short-circuit fault at bus A. The simulation results are studied for the system with and without using the CLCB.


**Table 1.** Parameters of electrical network and CLCB.

In normal operation mode, both buses delivered power to the loads at half capacity (12.5 MVA). In this case, there was no voltage drop on the CLCB devices and because of the system symmetry; no current was circulated through the interconnected CLCB. In addition, it is assumed that there was no CLCB connected to the feeder and line current was in normal condition as shown in Figure 5. A fault at bus (A) could cause severe voltage sag, which would affect the sensitive load. In this case, the fault current increased and its amplitude reached 6.8 kA as shown in Figure 5.

**Figure 5.** Fault current at bus A without using CLCB.

It is assumed that the fault occurred at bus (A) which produced an increase of current in the interconnection CLCB and bus (A) experienced a transient voltage. To prevent the service interruption at a sensitive load, the CLCB was connected in series with the feeder and interconnection bus as shown in Figure 1. In fault case, the CLCB impedance increased and its series *LC* tank was in series with the interconnection bus during the increase of the current. Therefore, its impedance decreased the faulty line current to an acceptable level and compensated the voltage sag at bus (A). Figures 6 and 7 show the fault current and bus (A) voltage for both cases with and without using CLCB.

**Figure 6.** Fault current during the normal operation and fault with connected CLCB.

**Figure 7.** RMS value of bus A voltage with and without using CLCB.

3RVWIDXOW PRGH

In the case 1 (without CLCB), the fault current increased to the peak value of 6.8 kA but by using the CLCB, the fault current was limited to the peak value of 200 A. It is shown that in case 1, the voltage of the bus (A) decreased approximately to zero. However, CLCB not only reduced the voltage sag to 0.9 pu, but also it opened the faulty line and fixed the bus (A) voltage to 1 pu. During the normal operation mode, the impedance of the series resonance *LC* tank was close to zero and there was no voltage drop on it. During the fault, the resonance transformer was bypassed and a considerable voltage drop was seen on the series capacitor. In circuit breaking mode, the induced DC voltage on the series capacitor charged it higher than the peak voltage of the network and caused it to open the faulty line. Figure 8 shows the series capacitor voltage during normal operation and fault for AC and DC operation cases.

**Figure 8.** Series capacitor voltage during the normal operation and fault including fault current limiting mode (AC operation) and circuit breaking mode (DC capacitor charging).

As shown in Figure 8, after fault inception, the fault current increased but the impedance of the series capacitor in AC mode decreased the fault current. After one cycle delay, the controller turned off *T*s1 and induced DC voltage on the series capacitor charged it with DC voltage. In this case, the faulty line was opened via a series capacitor and the fault current reached zero.

The load voltage during normal and fault operation modes is shown in Figure 9. The fault occurred at instant (a) and the voltage of the load decreased to zero. At instant (b), the fault was cleared and the load voltage returned to the pre-fault value.

**Figure 9.** Load voltage during the normal and fault operation modes.

After the fault removal, the voltage of the electrical load was distorted for a first half-cycle. The stored energy on the series capacitor during the fault period caused this voltage fluctuation.

The CLCB operation and its effect on faulty line current are shown in Figure 10. These comparative plots show the CLCB influence on both decreasing the fault current and opening the faulty line. The dotted plot shows the fault current when there is no connected FCL in series with the feeder. By FCL utilization, the fault current was decreased as shown with dash line in Figure 10.

**Figure 10.** Comparison of line current during the normal and fault operation modes; with and without using CLCB.

In the instant of fault inception, the first peak of the fault current decreased and after that the limited fault current reached an acceptable level. The blue solid plot shows the line current during normal and fault operation modes affected by the proposed CLCB. At the first cycle of the line fault current, the proposed CLCB acted as a fault current limiter. Then it opened up the faulty line, and current decayed to zero.

#### **6. Experimental Results**

To verify the simulation results, a CLCB prototype was built as shown in Figure 11. The CLCB prototype was tested in normal and fault operation modes. Table 2 lists the experimental setup parameters.

**Figure 11.** Prototype of the proposed CLCB.


**Table 2.** Experimental setup characteristics.

Using a mechanical switch, a single line to ground fault was implemented. The controlling circuit included a voltage transformer, a current transducer (*LTS 25-NP*), IGBT gate drivers (TLP250), RC filter, and an Atmel XMEGA microcontroller. Measurements of line voltage and current in faulty condition were processed and detected by microcontroller and operation command was generated in two stages. In the first stage, by operating a switch of the transformer, secondary fault current magnitude was limited. In the second stage, by operating series IGBT, fault current was broken.

Figure 12 shows the line current during the normal and fault operation modes. In this plot, the phase to ground fault occurred at instant (a) via a mechanical switch and was cleared up at instant (b) by the opening of the mechanical switch. As shown here, after the fault occurrence, the CLCB limited the fault current, opened the faulty line, and decreased the fault current to zero. After fault clearance, CLCB recovered the faulty line in less than 20 ms. This measured curve is in fair agreement with Figure 6.

**Figure 12.** Line current during the normal and fault operation modes.

Figure 13 shows the protected bus voltage during the normal and fault operation modes. As shown in this figure, CLCB can successfully fix protected bus voltage to an acceptable level during the fault. This figure is in agreement with Figure 7. In this figure, the duration of the normal, fault operation modes, and its effect on the line current can be seen in the upper curve.

**Figure 13.** Protected bus voltage during the normal and fault operation modes.

The voltage of the series capacitor is shown in Figure 14. The series capacitor voltage during the normal operation mode was sinusoidal and this capacitor was in resonance with the series transformer primary. After the fault, by operating series IGBT operation, voltage changed to the DC voltage, which opened the faulty line. This figure is in agreement with Figure 8.

**Figure 14.** The voltage of series capacitor during the normal and fault operation modes.

Comparing the proposed CLCB with traditional CB and power electronic based CB, superiority of the proposed structure can be listed as follows:


#### **7. Conclusions**

In this paper, a new type of CLCB is proposed. This device acts by dual-function protection, not only limiting the fault current but also open the faulty line similar to a circuit breaker. In practice, its fast response to faults can successfully limit the first peak of the fault current. In addition, the proposed CLCB assists to recover the protected buses voltage to an acceptable level. Therefore, the sensitive loads do not experience a significant voltage sag. The CLCB can be placed as a solid-state circuit breaker (instead of the traditional circuit breakers) and behaves as a fault current limiter. Performance of proposed CLCB is proved by simulation and experimental test results.

**Author Contributions:** All authors contributed equally to this work and all authors have read and approved the final manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
