Analysis of MMC Energy-Based Control Structures for VSC-HVDC Links

Abstract

An analysis of control structures for Modular Multilevel Converters (MMC) used in High-Voltage Direct Current (HVDC) applications is addressed. In particular, the study focuses on the case of a point-to-point link with masterslave control, considering an energy-based scheme (also known as closed-loop or energy-controlled) for the MMC, meaning that the internal energy of the converter is explicitly controlled. With such an approach, the MMC internal energy can be controlled independently from the energy of the HVDC link, and whereas the internal capacitance of the MMC depends on the converter's rating, the capacitance at the DC terminal depends on the cable length. Therefore, several possibilities regarding the outer control structure (internal energy and DC voltage) arise, affecting the overall dynamics differently. Whereas for a long link the classic control structures should perform well, for shorter links the transient performance might not be acceptable and other alternatives shall be used instead. Different control structures are presented and evaluated in this paper through small-signal and frequency-domain analysis, and validated through time-domain simulation with Matlab R Simscape Power Systems TM .

Optimal Multivariable MMC Energy-Based Control for DC Voltage Regulation in HVDC Applications

“…where ξ is the damping factor, ω n is the natural frequency, and C dc denotes the equivalent DC capacitance. However, the MMC does not contain a large physical DC capacitance, resulting in unsatisfactory DC voltage dynamic performance under DC power changes when this tuning rule is used [18]. While this tuning results in acceptable performance for HVDC links of several hundreds of kilometers, it is not obvious how to choose adequate gains for medium and short links.…”

“…As detailed in [9], [18], a coordinates transformation from upper − lower to dif f − sum frame is useful to represent the system in state-space and to better identify the magnitudes to be controlled. The following variable change is defined…”

“…A feed-forward of the AC power is usually added after the output of the total energy PI controller [9], [11], [18], otherwise the gains of that PI controller have to be increased notably to obtain a stable system. • Cross control [17], [18] without diagonal coupling, K 1 = K 4 = 0: the outputs of the outer controllers are connected to the opposite current reference inputs as in the classic control structure. When a DC voltage disturbance occurs, the first reaction will be through the DC current.…”

“…This structure improves the decoupling between the AC and the DC sides, and appears to provide more consistent performance. • Weighted control [18] for full (non-zero) coupling: this structure combines the two previous structures by connecting both PI outputs to both current reference inputs through four weighting factors, affecting the two current references at the same time. In this approach, the control system is provided with two extra degrees of freedom, obtaining six degrees of freedom overall (the k p − k i ratio for both PI, and the four scaling factors).…”

“…Compared to the standard classic structure, the performance can be improved with a cross structure, which connects the outer controllers to the opposite inner controllers (i.e., opposing the standard interconnection), thereby resulting in a better decoupling between the AC and DC sides [17], [18]. Furthermore, a weighted structure connecting both PI outputs to both current references at the same time was presented in [18], combining the effect of the classic and the cross structures. However, no insights were given with respect to the selection of the weights and tuning of the gains.…”