Design Procedure of MMC-HVDC System: Comprehensive Consideration of Internal and External Dynamics

This paper proposes a systematic design procedure with comprehensive consideration of the internal and external dynamics in modular multilevel converter high voltage direct current (MMC-HVDC) transmission system. Previous studies on MMC parameter selection separately deal with each specific component such as energy storage capacity for voltage ripple of sub-module (SM) capacitor, arm inductance for second harmonic circulating current reduction, maximum allowable modulation index for MMC operating condition, which considered only a single purpose. However, the parameters respond dynamically to their characteristics and interact directly with the MMC performance, power system conditions, and specific requirements. In this study, we investigate the mutual relationships between the parameters and their performance. Then, we determine the parameter values based on a proposed systematic design procedure with the desired objectives and restricted conditions, which could be cumbersome and time-consuming to approach proper and acceptable parameter values. Therefore, this study could provide engineering evaluations and insights to help MMC-HVDC system engineers and project developers in intuitive approaches regarding the design aspects of the technology requirement challenges. The efficacy and accuracy of the analysis and design method for the MMC-HVDC system parameters were validated by PSCAD/EMTDC time-domain simulation and real-time digital simulation with hardware-in-loop system.


I. INTRODUCTION
Modular multilevel converter (MMC) application has been widely adopted by power utilities in recent years, as its inherent attributes bring clear advantages compared with two-level voltage-sourced converter (VSC) due to reduced losses and footprint of mega-scale applications such as high voltage direct current (HVDC) transmission system, flexible alternating current transmission systems (FACTS), medium voltage DC system, and motor drives, introduced in [1]- [10]. Previous research for MMC applications attempted to achieve stable performance, control, design, and its operation The associate editor coordinating the review of this manuscript and approving it for publication was Dwarkadas Pralhaddas Kothari. as a grid-tied converter [11]- [21]. Among these key studies, the methodology for parameter determination is an important subject to achieve enhanced MMC-HVDC systems. The purpose of designing the parameters in previous studies was to make MMC capable for steady-state operation, allowing it an improved ability through goal-oriented design solutions based on the precise analysis of internal dynamics. The previous design philosophy could lack a system integration and cause an inaccuracy between an anticipated performance and a required specification of MMC-HVDC system, since complicated interaction exists between the internal and external parameters due to large power system interconnection. Therefore, a comprehensive analysis of the parameter relationships and a systematic process should be established to improve the VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ MMC-HVDC system without adverse impacts on bulk power system. In general, there are main components in MMC development such as the energy storage capacity for capacitor voltage ripples and arm inductance for second harmonic circulating current reduction. The most important part of MMC design is the determination of sub-module (SM) capacitance. MMC utilizes the stored energy in SM capacitors, which are required in a larger number than conventional two-level VSC. Therefore, it must be designed carefully to maintain its performance. The method of estimating the energy storage capacity was introduced when MMC was proposed by Marquardt [22]. Similarly, calculation methods were developed in detail in some studies and used to calculate SM capacitance [23]- [25]. In [26]- [29], a design methodology considering capacitor voltage ripple according to the operating point was introduced based on the same approach. References [30]- [33] used simple equations based on the results empirically derived in [26]. [34] adopted a design method based on the detailed analysis of the MPPF-based capacitor in terms of aging and voltage ripple. Previous studies on energy storage capacity focused on attempting to design an available specification. However, the parameter should be comprehensively analyzed and then determined considering requirements and limitations due to the connected power system condition.
Arm inductance could be considered for an inevitable second harmonic circulating current, a sudden fault current, and unwanted resonance in MMC. To reduce the second harmonic circulating current, the design method of arm inductor as a harmonic filter was introduced [35]- [37]. This method is less preferable in terms of harmonic stability than a proper control scheme, which indicates that there is a disadvantage in the necessary arm resistance even if a virtual arm resistance is available to secure harmonic stability [38]. Hence, it is assumed that it is not used in actual MMC-HVDC systems to avoid undesired adverse impacts from a stability point of view. Reference [31] showed a design method through the analysis of resonance generation area between SM capacitance and arm inductance. This is the method to avoid areas where the resonance may occur at twice the fundamental frequency. Reference [39] introduced a method for suppressing the second harmonic circulating current generated by the capacitor voltage ripple and reducing the fault current rise rate by analyzing the internal dynamics of MMC. However, since the method is not considered for the circulating current suppression control (CCSC) scheme that eliminates the second harmonic circulating current, more careful selection with MMC auxiliary control effect is necessary by multi-faceted judgment.
In addition to the above parameters, the modulation index for MMC-HVDC system was analyzed in [40]- [42]. This study determines the PQ capability which is the basis of VSC principle. In general, the modulation index is determined according to the operating point of the VSC, and the modulation index equal to 1 is ideal without 3rd harmonic injection strategy. However, if the compensation voltage considering the circulating current introduced above is synthesized on the original reference voltage, the modulation index will be less than one. Therefore, the MMC-HVDC developer should consider the maximum modulation index to meet the PQ requirements accounting for SM capacitance and arm inductance in the initial design. For the MMC-HVDC system design concerning SM capacitance, the arm inductance, and modulation index mentioned above, the design and analysis were carried out in [43], [44]. However, operating conditions of MMC may be changed by parameters owing to the influence of the control techniques required for MMC conditions. Hence, MMC-HVDC system design should be based on a comprehensive analysis of the impact of additional control techniques of MMC such as voltage balancing control with sorting algorithm and CCSC [45]- [56]. Furthermore, voltage and current at AC and DC sides of the MMC-HVDC system should be selected in various ways according to the design of the previously introduced components as well as MMC voltage level considering connected power system conditions. Therefore, in this study, a systematic design procedure of MMC-HVDC system parameters is established regarding their characteristics for comprehensive and approachable development.
The remainder of this paper is organized as follows: Section II includes a brief overview of the internal dynamics of MMC. In Section III, the relationships between the parameters and the performance are properly investigated, and the MMC-HVDC system parameters determination methodology is proposed in Section IV. Finally, PSCAD/EMTDC time-domain simulation results for verifying the determined parameters are shown in Section V, followed by the conclusion in Sections VI. Fig. 1 indicates the configuration of MMC. V dc is the DC link voltage between positive and negative pole in HVDC system and I dc is the DC current. V c is capacitor voltage and i c is the capacitor current flowing into each individual capacitor. i arm is the arm current consisting of upper (i armp ) and lower (i armn ) arm currents including AC (i sj ) and circulating currents (i cir ) (phase j = a, b, c, and the subscript p denotes the upper arm and n denotes the lower arm), which can be expressed as [14]:

II. BASIC MMC DYNAMICS FOR PARAMETERS
where i 2f is second harmonic circulating current. V sm is terminal voltage generated from SM consisting of two-level halfbridge converter depicted in Fig. 1. According to [14], MMC is featured by the following equations where e j in (4) is the inner EMF generated in phase j and desired inner voltage as a sinusoidal waveform, which can be expressed as: The reference arm voltage could be defined from the above analysis as follows:

A. EXTERNAL PARAMETERS OF MMC-HVDC SYSTEM
The MMC parameters to be considered are listed in Table 1 for accomplishing a stable operation and control of MMC-HVDC system in Fig. 2. Active power (P s ) plays a key role in MMC-HVDC planning process as pre-determined variables, and is naturally determined by line impedance as well as angle differences between grid-side voltage and MMC-side voltage. Reactive power (Q s ) is also highly related to the modulation signal, which factors are influenced by internal and external conditions; thus, we will provide the guideline of parameter determination in this research. Reactive power (Q s ) value should be determined due to the voltage deviation and this kind of reactive power should be determined by power system study. In other words, active power is determined with transmission power capacity and the design specification of reactive power can be determined, based on power system study such as simulation.
Although the primary voltage of transformer is a predetermined parameter on the basis of grid-side rating, the secondary voltage should be properly determined and thoroughly studied, as it determines the characteristics of PQ capability directly associated with DC voltage and current. The secondary voltage also affects the modulation index, the number of SMs, SM capacitance, arm current, and arm inductance, and those factors should be considered for reliability, system loss, and stability requirements.
In sum, internal dynamics and parameters of MMC heavily rely on secondary voltage of AC side.

B. INTERNAL PARAMETERS OF MMC-HVDC SYSTEM 1) DC VOLTAGE AND CURRENT
In Table 1, the main parameters for MMC operation are shown. DC voltage and current are economic variables in relation to active power assuming that DC power is equal to AC power and additional power losses due to passive elements and power conversion. The DC voltage is inversely proportional to DC current for constant DC power, and it relates to switching device rating. Therefore, both parameters could be considered in terms of economic requirements. Especially, DC voltage highly relates to modulation index because MMC terminal voltage (V t ) is determined by (m · V dc )/2 which affects PQ capability. The DC current should be considered with the DC cable rating that is related to the cost of an important part of HVDC system where economical requirements have to be considered.

2) THE NUMBER OF SMs AND REDUNDANT SMs
The number of SMs relates to a blocking capability of a switching device for individual SM voltage (V sm ) with a determined DC voltage. It is an economic consideration variable that can be determined within a range to meet the DC link voltage and the rating of the switching device.
The number of redundant SMs depends on the redundancy methods [57]. For a passive mode (cold reserve state) for the redundancy, the number of redundant SMs is a variable independent of the MMC-HVDC system operation in which economics and reliability are to be considered together. Thus, it can be determined through optimization to enhance the MMC-HVDC system reliability after the system variables are determined. Otherwise, an active mode (hot spinning reserve state) has an effect on capacitor voltage ripple and switching frequency depending on power losses [57]. In the case of the active mode, the number of redundant SMs should be considered with reliability and economical requirement as well as system loss evaluation.

3) ARM INDUCTANCE
Arm inductor should be designed according to three factors. These are: to reduce the second harmonic circulating current VOLUME 8, 2020  flowing inside the MMC, the fault current rise rate and to consider the reactive power consumption while avoiding resonance in the MMC circuit. Particularly, since arm inductor is related to CCSC strategy, it should be designed for stability and reactive power loss of MMC-HVDC system. For CCSC, arm inductor only needs to consider reactive power loss and stability, and when CCSC is disabled, second harmonic circulating current reduction studied in many previous studies is the main purpose.

4) SM CAPACITANCE, VOLTAGE RIPPLE, SWITCHING FREQUENCY
Energy storage capacity is a fundamental parameter that allows MMC to operate in a stable region. SM capacitance is an important parameter related to voltage ripple that affects the lifespan of SM capacitor. Hence, it should be managed to be generated within the limitation that designates voltage ripple factor (k max ). The voltage ripple management is related to the switching frequency (f sw ), which influences the power conversion loss and power quality. As a result, the energy storage capacity should be designed considering reliability and economical requirements while guaranteeing minimum MMC operating conditions [58].

5) PQ CAPABILITY
The modulation index of conventional VSC is related to the operating point for desired active and reactive powers. Furthermore, MMC has additional factors affecting the modulation index, which are the compensation signal generated due to the control to reduce the second harmonic circulating current and additional reactive power consumption by the arm inductor. In this paper, the physical factors discussed in the previous research are not considered because their influences could be negligible [40], [41].

6) ARM CURRENT
Arm current is related to the rating of the switching device and includes AC and DC currents as shown in (1)-(3). When the above mentioned main parameters are determined, it can be selected by the modulation index (k) of AC current over DC current. In this study, even though the second harmonic circulating current increases the peak value of the arm current, it is assumed that it becomes zero by CCSC strategy.

IV. DESIGN OF MMC-HVDC TRANSMISSION SYSTEM A. CONSIDERATIONS OF DESIGN
MMC parameter should be designed with the trade-off relationship between cost-effectiveness and system reliability. In this study, however, a design method is developed from the performance perspective of MMC-HVDC system except for economical and reliable factors that are assumed as variables being optimized from manufacturers. In addition, it is assumed that the desired active and reactive power values depend on possible HVDC applications to current power systems, and no overload capability of the whole equipment is applied.
The initial MMC-HVDC design phase defines the P-Q rated values and AC voltage to be connected through the grid planning process. The internal and external parameters for MMC design should be calculated only with the minimum parameters obtained from the system planning. In this study, a systematic design process is proposed for obtaining the parameters introduced in Table 1. For reliability and economical requirement-based design, the optimization process is applied independently with the proposed methodology and is not considered in this paper.   are pre-defined parameters before the determination process of the main parameter. These parameters are provided after the power system study completion.

2) DC VOLTAGE AND CURRENT (V dc , I dc )
In general, DC voltage could be determined by a loss-cost optimization, and DC current could be determined by considering the cable cost corresponding to the long distance. Once DC voltage is optimized according to the HVDC transmission length, other MMC parameters could follow the design process provided in Fig. 3. In this study, it begins by selecting the DC voltage optimized from a cost assumption as the maximum value in Fig. 4, and redefined considering the . DC voltage and current determination method. VOLUME 8, 2020 rated current of the switching device as shown in the Step 9 of Fig. 3. As shown in Fig. 4, the DC voltage and current to meet the required DC power are inversely proportional. Hence, as the DC voltage is fixed, the DC current could be calculated, and the entire design process would be iterated while identifying the required conditions.
The variable of individual SM voltage determines the number of SMs for DC voltage of MMC which is related to the voltage rating of the switching device thereby individual SM voltage magnitude could be applied as an optimization target with the economic evaluation. In this paper, we suggest the recommended voltage utilization value (λ v ) of 60 % over the rating of the switching device is used in (9) and specifications suitable for the ratings of commercial products are adopted.
where V dc is DC voltage of MMC, N 0 is the number of SMs, V sm is the nominal voltage of SM, λ v is the voltage utilization factor, and V IGBT is the voltage rating of IGBT. To obtain the secondary voltage of the main transformer being determined in next section (Step 5), the necessary parameters should be estimated as an initial parameter. A primitive energy storage capacity could be obtained as follows [31], where C 0 is initial parameter of SM capacitor which would be updated during the process, EP 0 is energy per power ratio (20∼50 [kJ/MVA]) for the capacitance that meets the normal operation of MMC [26], [30]- [33], [57], and S n is the apparent power based on V dc I dc /cos(ϕ). A temporary arm inductance value could be obtained by assuming that the second harmonic circulating current (I 2f ) is a specific value, as follows [39], where L 0 is initial parameter of arm inductance which would be updated during the process, I 2f is second harmonic circulating current, and ω 0 is the nominal angular frequency.
To specify the second harmonic circulating current value for the variable, X Lpu could be empirically selected to an approximate 0.1 [p.u] as a desirable value due to reactive power consumption in the arm reactance [32]. From (11) with (9), (10), (12) and (13), per unit value of arm inductance could be defined as (14): where Z base is the impedance base, V LL is the line to line AC voltage, and m is the modulation index. Therefore, the specific second harmonic circulating current (I 2f ) could be calculated with assuming that the per unit value of X Lpu is 0.1 [p.u] and modulation index is 1 in (12) and (13) for (14) as follows, .
The initial parameter of (I 2f ) is determined with the brief value. Thus the value should be recalculated to find a more appropriate value for an important role of PQ capability, stability and fault current rising limitation. Leakage inductance (L tr ) of the main transformer can be determined by the transformer manufacturer as in [43]. In this study, however, we adopted the general value used in [26].

5) CONVERTER SIDE AC VOLTAGE (Secondary VOLTAGE, V c )
Secondary voltage of the main transformer should be determined within a restricted range that meets an operable modulation index of MMC while providing the required active and reactive power. Using the PQ capability curve formula, a maximum acceptable MMC voltage should be selected to define the secondary voltage. The formula for PQ capability is as follows, where X eq = ωL tr +ωL f +ωL arm /2 as depicted in Fig. 1 and 2.
In this study, however, we assumed that L f = 0 because the interface transformer is installed close to the terminal of MMC, thereby the line impedance between them is negligible, and V c is equal to V t . Moreover, since the primary voltage of the transformer has an allowable operating range regulated by the connected grid code, the secondary voltage has also the regulated voltage range. As a result, when MMC performs only pure reactive power compensation (STATCOM mode), which has a maximum voltage while supplying Q max within the rated power of MMC, the secondary voltage could range in the regulated AC voltage from 0.95 to 1.05 [p.u] assuming that the grid code is an arbitrary value. In (16) with the conditions of P s = 0, Q max and V spu = 1.05 [p.u], secondary voltage could be calculated as follows.
where V cpu is the converter side AC voltage, X eqpu is the equivalent total reactance, Q maxpu is the maximum reactive power requirement and V spu is the grid side AC voltage at per unit value, respectively. When the secondary voltage is considered with modulation index, it depends on the DC voltage that provides the AC terminal voltage of MMC and summed by individual SM voltage multiplying the number of activated SMs. Therefore, MMC DC voltage could be as follow, where k max is the voltage ripple factor and V ripple is k max · V sm . From (18), minimum DC voltage due to k min that is k max = −k min , is as follows [26], Since the value obtained in (17) (13),

6) ENERGY STORAGE CAPACITY (SM CAPACITANCE, C sm )
Energy storage capacity of SM could be adjusted with the previously determined parameters and a desirable voltage ripple factor based on a detailed analysis through (40) of [26]. The initial SM capacitance (C 0 ) is in the range of a stable operating condition due to the particularly conservative value. Moreover, the voltage ripple is determined by the natural response of the initially designed values. However, the CCSC control algorithm could reduce the voltage ripple, which affects the internal dynamics of MMC. Conversely, a voltage balancing algorithm for the voltage ripple could aggravate the voltage ripple, which would cause a forced ripple. Therefore, the required voltage ripple factor is associated with switching frequency depending on the MMC system loss, which should be examined with a specific switching principle adopted by the MMC developer and designed with the loss evaluation in the design step (Step 10) of the design process [46], [48], [49], [59], [60]. In this paper, the SM capacitance is only validated for analysis in whether stable operating point or not under the full-sorting algorithm.

7) ARM INDUCTANCE (L arm )
In case circulating current suppression control (CCSC) is disabled, arm inductance is determined by selecting the second harmonic circulating current mentioned in Step 4 of the design procedure. The CCSC, however, is conventionally enabled for MMC-HVDC transmission system. Arm inductance depends on whether the CCSC activates or deactivates, which should be considered.

a: CASE I WITHOUT CIRCULATION CURRENT SUPPRESSION CONTROL
• (a-1) second harmonic circulating current calculation (I 2f ): the I 2f could be calculated by (15) with a redefined energy per power ratio (EP 0 ) through the SM capacitance validated from previous step.

b: CASE II WITH CIRCULATION CURRENT SUPPRESSION CONTROL
• (b-1) Compensation voltage calculation (U 2f ): Equation (11) provides arm inductance determination method without CCSC, thereby the resulting second harmonic voltage across the arm inductor is equal to (32) in [39] assuming that (24) and (25) in [39] include the second harmonic current component in arm current. However, with the CCSC algorithm, the second harmonic circulating current term should be removed in (26) in [39]. Thus the equation should be derived without the second harmonic term, therefore, as follows, • (b-2) Arm inductance determination (L arm ): Arm inductance with the CCSC could be designed regarding fault current rise rate for gate turn-off delay while a resonance problem is investigated [31], [39], [44], because the second harmonic circulating current is negligible at this time.

8) MAXIMUM ALLOWABLE MODULATION INDEX (MAMI)
The initial maximum allowable modulation index (m max0 ) is based on (20) for the secondary voltage in the case without CCSC strategy. However, for the compensation voltage signal (m U2f ) obtained (21) owing to voltage ripple generated by SM capacitor natural dynamics, the MAMI could be acquired by where m max0 is the MAMI without CCSC and m U2f is the compensation signal index. In this study, 3rd harmonic injection strategy is not considered.

9) ARM CURRENT AND IGBT CURRENT (I arm , I IGBT )
The current rating of switching device (I IGBT ) has also a current utilization factor (λ i ) such as the individual SM voltage determination in (9) due to a thermal consideration, as follows, From (1) and (2) for arm currents (I arm ), a ratio of AC and DC currents (k) is as follows, Assuming that the DC and AC powers are equal with a lossless system as follows, From (13) and (24) in (25), the following equation is derived, From (1), (2), (24) and (26), arm current is as follows, For ease of the device current rating, γ 0 is used as the MMC arm current rating factor and with minimum value of 1 in case that m = 1 and ϕ = 0, as follows, However, m and ϕ depend on operating point from (16). Therefore, when the MMC-HVDC system operates to unity power factor(assuming the condition of PQ curve linear ramp rate), the modulation index that could be obtained in step (Step 5) is the smallest. The operating angle could be calculated under this condition, as follows, From (27) and (28), final arm current is as follows, As a result, the current rating of switching device could be determined considering specific products from many semi-conductor manufacturers, and inspected in terms of economical aspect. From (28) and (30), final IGBT current is as follows,

10) SYSTEM LOSS EVALUATION
System losses consists of the power losses on passive elements (interface transformers, lines, connection points, DC cables) and power conversion elements (active device).
The system loss evaluation in this paper focuses on switching loss highly related to the SM capacitance with the voltage ripple factor that limits the switching frequency and various capacitor voltage sorting algorithms [59], [60]. The voltage fluctuation of the SM capacitor is determined by a natural voltage ripple and a forced voltage ripple. When energy storage capacity is obtained from the design step (Step 6), the voltage ripple factor (k max ) is considered as a purely natural voltage ripple. Owing to the characteristics of MMC, however, voltage ripple is additionally generated by various capacitor voltage sorting techniques, which denote forced ripples generated by external factors [40]. Based on the characteristics, the active mode redundancy method could inversely reduce the capacitor voltage ripple [57], and thus the system loss evaluation should be investigated with the various relationship in terms of reliability and economical requirement. Therefore, for the system loss, the voltage ripple is determined according to the SM capacitance, the sorting algorithms, and redundancy methods. Finally, the switching frequency that can satisfy the system loss can be determined. Since this study was extensively scrutinized by previous studies [57], [59], [60], only the verification is conducted with an assumption that a required MMC-HVDC system adopts the passive mode redundancy method in this paper.

11) FINAL DESIGN PARAMETERS VERIFICATION AND DETERMINATION
The determined parameters were validated by PSCAD/ EMTDC simulation, which is shown in the next section.

V. VALIDATION OF DESIGN PARAMETERS FOR MMC-HVDC TRANSMISSION SYSTEM A. PSCAD/EMTDC SIMULATION RESULTS
For the MMC-HVDC system parameters, PSCAD/EMTDC simulation with a time step, 1/(2πf 0 mN 0 ), is based on [61], [62], allowing for accuracy of interacting between the HVDC transmission system and MMC performance. Furthermore, MMC control is focused on direct modulation associated with the dq-framed CCSC method [46]. This paper confines the parameters offered in Table 1 that provide the required prerequisite variables and necessary parameters for MMC-HVDC. The determined parameters are shown except for the complete design procedure because detailed economical and reliability evaluations are required to finalize the optimized MMC-HVDC system parameters. The economical requirements and reliability are out of the scope of this study. However, this research provides a design approach to expand an insight regarding the entire MMC-HVDC transmission systems. Table 2 indicates that necessary parameters for the MMC-HVDC system are determined by the proposed design procedure. Except for the economical requirements of the DC voltage and the switching device, it begins that V dc is fixed to ±120 kV. As shown in (a) of Fig. 9, the MMC-HVDC system has an operating point that the active and reactive power are 200 MW and 60 MVar at the required normal operation as shown in (a) and (b) of Fig. 5, when the reference commands are 1.0 and 1.5 s, respectively. Under these conditions, AC voltage at the grid side is 154 kV and DC current is 833.3 A. The ratings of switching devices for commercial product is 4 kV. However, the individual SM voltage is 2.4 kV considering a recommended voltage utilization factor (60%) as shown in (f) of Fig. 5. Hence, the number of SMs is 100 without redundancy. For initial energy storage capacity, 50.0 [kJ/MVA] is adopted and the SM capacitance is 6000 µF which is not considered for voltage ripple limitation. As a result, voltage ripple is 7.5% when arm inductance, which is X Lpu is 0.1 [p.u] and a specific I 2f is 300 A obtained from (15), is the main role of reducing the second harmonic circulating current component without the CCSC. The second harmonic circulating current is injected into the arm current thereby the arm current is distorted as depicted in (h) of Fig. 5. Based on the initial parameters, the temporary secondary voltage at the valve side is calculated as 128.317 kV through (15). The final reference signal and calculated modulation signal are well matched and the valve side voltage is generated within the defined voltage level.
Considering the CCSC strategy, the compensation voltage should be defined when the maximum reactive power support, in shown (b) of Fig. 9 is enabled as a STATCOM  mode at 1.5s as shown in (a) of Fig. 6. The arm current has only fundamental frequency component due to the CCSC as shown in (c) of Fig. 6. However, the reference signal plus the compensation signal that is 0.025 obtained from (21), shown in (c) of Fig. 7, is over-modulated as shown in (a) of Fig. 7. Therefore, the valve side voltage considering the MAMI should be re-defined through (20) and (22). The final valve side voltage is updated from 128.317 kV to 124.94 kV considering the MAMI due to the compensation voltage. As shown in (b) of Fig. 7, the maximum allowable modulation index is within the available range. With the CCSC strategy, the initial arm inductance could be validated in terms of undesirable resonance phenomenon through [44], which is 380.17 rad/s within the safe region. Finally, the arm current depicted in Fig.(8) is well matched compared with the value obtained from (30), where the γ 0 is 1.169, the I dc is 833.3 A, and hence the I arm is 973.9 A. VOLUME 8, 2020  An experimentation based on real-time digital simulator with hardware-in-loop system is performed for validation of proposed design methodology. The RTDS system consists of Giga Transceiver Workstation InterFace (GTWIF) and PB5 processor cards mounted on main rack, and Xilinx ML-605 FPGA board for the MMC-FPGA Unified Model (U5) control and operation, which is shown in Fig. 11 [59], [63], [64]. The MMC simulation model in the VSC bridge box of the RTDS is computed for 2.5 µsec, which is enough time for 512 SM capacitor voltages to transfer between ML-605 and the customized physical controller based on Xilinx's Aurora Protocol with 5.0 Gbps high speed serial communication. Sorting algorithm of 432 capacitor voltages can be completed in only 434 clocks; the operation clock was set to 200 MHz (5 nsec per clock). RTDS and HILS can be synchronized by the Grand Master Clock (GMC) as in [64]. The customized physical controller operates every time at 10 µsec for a control period. Table 3 indicates the main parameter values obtained from the proposed design method for verifying the efficacy in a similar way of PSCAD/EMTDC simulation. Fig. 12 shows the HILS-based experimental results to verify the model of the required MMC-HVDC system. Fig. 12(a) shows the active and reactive power that are 800 MW and 80 MVar, respectively. The value of 800 MW and 80 MVar (capacitive mode) can be transformed to the value of ±5 V scale in the RTDS based HILS testing, since a scaling factor is adopted due to a limited range of peak value for D/A output (it is possible to ±10 V , but ±5 V is selected to prevent an overstretch for the measured data) defined by Giga-Transceiver Analog Output (GTAO) card in RTDS [65]. Figs. 12(b), (c) and (d) are AC voltage/current at rectifier side and short term AC voltage/current of (b), and DC voltage/current, respectively,  which are identically applied with a scaling factor as depicted in Fig. 12(a). The MMC-HVDC system is identically configured in the RTDS as Fig. 2. Especially, the MMC station 2 (rectifier mode) is applied with the step 8 for final MAMI adding to the compensation signal obtained from the step 5. On the other hand, in case of MMC Station 1 (inverter mode), MAMI is not applied which is also the result of EMTDC simulation; consequently, we may notice that the application of MAMI indicates the improved performance as in Fig. 7.
Therefore, Fig. 13 shows the experimental results of the capacitor voltage ripple and arm current. As shown in Fig. 13(a), the capacitor voltage in the inverter side without proposed method has the voltage ripple factor as 8 %; on the other hand, the rectifier side with proposed method has 3.35 % as an anticipated result. As shown in Fig. 13(b), the peak value of arm current calculated by (30) is 1.128 kA, while γ 0 is 1.128 and I dc is 1000 A. Fig. 14 shows the experimental results regarding the MAMI. As mentioned above, both waveforms depicted in Fig. 14(a) as Initial and Final MAMI depends on the CCSC, resulting in the improvement of arm current in Final MAMI as shown in Fig. 13(b). For more detailed validation, the zoom-in waveform in Fig. 14(b) has a distinction in the MAMI. The distinct voltage is 95 mV (0.019 obtained from (21)) between 4.685 V(MAMI=0.937) at rectifier side and 4.590 (MAMI=0.918) at inverter side. Table 2 shows the verified parameters obtained from the proposed systematic design procedure without an iterated optimization process. As mentioned above, economic and reliability evaluations are required to obtain the final parameters for an optimized MMC-HVDC system. The remaining procedure is as follows:  • Economical evaluation for DC voltage that determines insulators and the number of the switching device through a process of the design step (Step 2) with Fig. 4 • Economical evaluation for DC current that determines the cost of long-distance DC cable and the number of switching devices • Economical and reliability evaluations for arm inductance that determines fault current rising rate relating to turn-off time delay of the switching device to protect the MMC-HVDC system and analysis of the reactive power loss • Economical evaluations for SM capacitance that determines the voltage ripple affecting the switching frequency in terms of power conversion loss with the k max through Fig. 10 based on [26].

C. REVIEW OF DETERMINED PARAMETERS
• Economical evaluations for IGBT current rating considering the cost of commercial products.

VI. CONCLUSION
This study analyzed the coupled relationship between HVDC system and MMC, and proposed a systematic design procedure for determining the parameters of the MMC-HVDC system. The inspection is conducted to discover the relationship between control algorithms and the parameters, and the validated results are presented through PSCAD/EMTDC time-domain simulations and RTDS based HILS testing. The optimization process to finalize the complete parameters for the MMC-HVDC system was not included. However, this research could provide researchers with a superior design approach to extend their high technology challenges, and expands the MMC-HVDC based future DC grid.