Modeling, Analysis, and Control of MMCBased HVDC Converters for Grid Services
MetadataShow full item record
- Institutt for elkraftteknikk 
The integration of large-scale, remote, renewable energy sources has led to the proliferation of HVDC converters in the power system. Following this proliferation, more and more ancillary services, such as Power Oscillation Damping (POD), that were traditionally provided by synchronous generators, are being demanded from the HVDC converters in order to support the power system. One challenge in this regard is that these services can require the manipulation of active power, which is taken from the dc side of the converter. This is undesirable because it propagates active power disturbances (oscillations) into other connected ac grids. With the dc side transitioning to meshed HVDC grids, which can interconnect several asynchronous ac grids, the propagations of such power disturbances become even more important. In addition to the propagation of the disturbance itself, a second problem is that the observability of an oscillation in another connected system can cause negative interaction among system-level controllers, such as Power System Stabilizers (PSSs), located in different systems. Both these problems can be avoided if the power needed for the service is obtained from a different source. The Modular Multilevel Converter (MMC), which is currently the most attractive topology in HVDC applications, has a potential to provide a temporary energy storage capability that can be used for ancillary services. In addition to the ac-dc MMC, dc-dc converters based on the MMC, such as the Front-to-Front (F2F) topology, also offer such a capability. This thesis focuses on developing and testing methods to utilize the energy storage capability of these converters to source the power needed for the ancillary services, so that the other connected ac grids are not disturbed. The thesis provides contributions on the modeling and control of the aforementioned converters. Modeling: MMC models of varying levels of detail are presented and compared. The levels of detail range from the most detailed models targeting fast transients studies to the simplified ones suitable for large-scale studies. Based on these models, new simplified models of the MMC-based Front-to-Front (F2F) dc-dc converter are also proposed in this thesis. The models aggregate the arm energy states of the converter in order to reduce the number of states, which is desired when performing large-scale studies. The proposed models were validated against detailedmodels, and the results showed that the simplified models can accurately represent the F2F converter in large-scale system studies. Control: The thesis proposes control methods on two different levels. The first is at the arm energy control level, where an improved arm energy controller is proposed. The other is at a higher level, where different implementations of methods to utilize the arm energy are proposed and analyzed. The arm energy controller is instrumental for the effective implementation of the high-level controllers. The proposed arm energy control is a complete approach for implementing a closed-loop compensated modulation based on estimated arm voltages without accurate knowledge of the system parameters. The arm capacitance and the time delay between the controller and the converter are the parameters which are not known accurately or are assumed to change throughout the lifetime of the converter. The proposed method has been tested using both simulation and experimental tests with a strong correlation between the two results. The experimental tests have shown that the method can work in cases that are not ideal where the number of modules is as low as 18. The tests have also demonstrated that the method works effectively in the presence of noise and asymmetrical capacitance changes. At a higher level, the thesis proposes two methods for utilizing the MMC energy for grid services, so that the power disturbance generated by the services is diverted into the submodule capacitors of the converter. These are the power cancellation method and Virtual Capacitance Support (VCS). The power cancellation method is a local implementation where active power disturbance due to the grid service is diverted into the arm energy. This method is local because it relies on the availability of the active power disturbance measurement. On the other hand, VCS requires measurement of the dc voltage variations which can be accessed by multiple converters. This enables a distributed implementation where the energy storage capability of multiple converters can be utilized. The methods have been shown to be effective in three applications: Power Oscillation Damping (POD), wind farm active power smoothening, and dc voltage dynamic support. A detailed analysis and validation of the VCS scheme in a POD application is also presented. The effectiveness of the scheme was studied using modal analysis, time-domain simulation, and Power Hardware In The Loop (PHIL) tests. There is a strong correlation between the results from the three studies, and they all show that the scheme can effectively divert the active power disturbance (oscillation) into the arm capacitors. It was shown that the VCS method only affects the observability of the mode in other grids, not the damping or frequency, as expected. Additionally, the PHIL tests show that the scheme is practically applicable since it was implemented using a scaled 18-level MMC prototype under realistic conditions. A requirement for the scheme towork properly is that the converter be designed with a slightly higher storage capacity compared to nominal design, for example 15% increase in sub-module voltage was used in this study. The amount of additional capacity can be chosen depending on the desired level of participation of the converters in providing the service.