Resumen de Power and frequency control of an offshore wind farm connected to grid through an HVDC link with LCC-based rectifier
Offshore wind farms (OWFs) present certain advantages over onshore wind farms. Offshore sites present wind velocities which are more uniform and reach sufficient values at lower heights if they are compared to onshore locations. Moreover, best onshore locations are already in use while free proper locations can be still found in the sea, what also increases the interest in OWFs.
OWFs are connected to the onshore grid electrical system by means of a transmission system in order to evacuate the energy which is generated in the offshore sites. This transmission system has traditionally been accomplished by the use of high voltage alternating current technology. However, the cost makes high voltage direct current (HVDC) transmissions be the most suitable solution if the distance is over some tens of kilometers. This is the reason why there are HVDC transmissions for OWFs operating nowadays. The existing HVDC transmissions use voltage-source converter (VSC) technology in the HVDC rectifier station, because this converter is able to startup and provide the offshore alternating current (AC) grid that the conventional wind turbine generator systems (WTGSs) need to be able to operate and produce energy. However, line-commutated converter (LCC) is another technology which could be used in the HVDC rectifier. LCCs present better features when they are compared to VSCs in terms of converter size, efficiency, reliability and cost. Although LCC advantages become significantly important for offshore sites, this rectifier converter solution is not used in any of the existing HVDC-connected OWFs nowadays. This is due to the fact that LCCs also present some drawbacks. The offshore AC grid needs the voltage and frequency to be controlled in order to allow the operation of both the WTGSs and the LCC rectifier. LCC rectifiers are not able to startup the OWF and provide a controlled voltage and frequency, while the VSCs inherently do it. However, some means could be used in order to make the connection of OWFs with LCC-rectifier-based HVDC links feasible and then take advantage of the benefits of LCC rectifiers compared to VSC rectifiers. Therefore, this Thesis proposes control strategies which overcome the aforementioned drawbacks of LCC rectifiers when they are used for the HVDC connection of OWFs.
The first chapter of the Thesis includes the analysis of the existing literature about the control of the HVDC connection of OWFs using LCC rectifiers, what has motivated this work together with the information given in the previous paragraphs. The state of the art of the control solutions is given by considering three different categories: conventional centralized controls, hybrid centralized controls and distributed controls. Then, the Thesis objectives are addressed. They mainly consist of proposing a centralized and decentralized control for the connection of OWFs through LCC-rectifier-based HVDC links. Not modifying the active power control channel of the WTGSs is a specification for both control proposals. Besides, it is required to develop models that allow the derivation and justification of the control proposals. Finally, the stability and the performance of the OWF must be evaluated when the proposed control strategies are applied.
Chapter 2 is about system description and modeling due to the fact that it is used in order to derive and justify the control strategies which are proposed. For this purpose, average-value models (AVMs) of the power electronics components are addressed because they reduce the computing time and simplifies the extraction of their small-signal characteristics. Therefore, this chapter contains a brief state of the art of LCC rectifier average-value modeling, where the lack of a proper model containing the dynamics of the capacitors placed in the AC side of the rectifier is stated. These capacitors represent both the AC harmonic currents filtering required by the LCC rectifiers at fundamental frequency together with the reactive power compensation of the LCC rectifier and its transformer.
First, the models of the HVDC link cables and the onshore inverter are presented. The T-model of the HVDC Benchmark model is used for the cables and a DC voltage source is used to represent the onshore inverter. This is due to the fact that the onshore inverter must operate in constant DC voltage mode.
Then, the proposed LCC rectifier AVM is derived and presented. First, the classic AVM of the LCC rectifier and its transformer is addressed because the equations of the conversion between AC and DC variables will be used. Moreover, the small-signal model of the classic AVM is also presented, because it will be used for later studies. Next, the enhanced LCC rectifier station AVM is derived in a synchronous reference frame and finally transformed into polar coordinates. After its derivation, the small-signal response of the proposed AVM is used for its validation, where the model is connected to the Thévenin equivalent of an AC grid in the AC side and the already defined HVDC link model in the DC side. The small-signal models are first derived and then the response of the transfer function between the rectifier DC current and the Thévenin voltage magnitude is evaluated. As it is expected, it fits the response of the detailed switching model (DSM) implemented in PSIM for frequencies below the switching frequency.
Finally, the WTGSs are represented by a power injection in most of the OWF equivalent models used in this Thesis. This is due to the fact that active and reactive powers are the WTGS output control variables in the conventional voltage oriented control which would be used in the centralized control. However, the decentralized control uses the reactive power control channel of the WTGS, so a model of the type-4 WTGS is addressed for these studies. This model has the DC power injected by the back-end converter as input. The DC link has both a crowbar protection and a capacitor. In the AC side of the front-end converter, the output filter is composed of a series inductor and a shunt capacitor.
The proposed LCC rectifier station AVM has several advantages which will be addressed following. In addition, it allows the study of the dynamic response of the HVDC link and the devices which are connected to the OWF AC grid by a static consideration of the complete topology of this grid. Such a study needs an iterative process where the rectifier AC bus is considered as the slack bus of the AC grid load flow. Then, the slack bus active and reactive powers which result from the load flow are the inputs to the dynamic AVM which will in turn provide the new slack bus voltage magnitude and angle to be considered in the AC grid load flow. Next, the process is repeated in this iterative solution.
Once the modeling of the elements in the system under study has been defined, Chapter 3 presents the centralized control proposal of this Thesis, which can be used with both types of LCC rectifier: the controlled thyristor rectifier and the uncontrolled diode rectifier. First, the system under study is addressed. Specifically, a single-aggregated WTGS equivalent model of the OWF is used in the centralized control studies, together with the proposed LCC rectifier station model and with the HVDC link model.
Then, the proposed AVM equations allow the derivation of the centralized control strategies. This AVM has the rectifier AC side voltage magnitude and angle as state variables. Therefore, the model explicitly has the state variables which must be controlled to perform the voltage and frequency control which is required in the OWF AC grid. Moreover, the proposed AVM dynamic equations demonstrate that the reactive power balance at the rectifier station AC bus, i.e. the point of common coupling (PCC), drives the offshore AC grid frequency variations. Likewise, the active power balance at the PCC bus drives the PCC bus voltage magnitude variations. Then, taking advantage of the information provided by the proposed AVM, the centralized voltage and frequency control strategies are presented. The only reactive power which could be controlled in the system without modifying the active power generated and transferred is the one produced by the WTGSs. However, communications should be used to perform the control and the response would not be as fast as required for the frequency control. Therefore, it is proposed to inject reactive power with an additional electrical mean in order to control the frequency, what it is accomplished by keeping a zero q component (zero voltage angle) of the PCC voltage vector. This is due to the fact that the synchronous reference frame used in the proposed control rotates at the desired reference frequency, thus constant angle implies constant frequency. Regarding the voltage control, the active power balance at the PCC bus can be affected by modifying the thyristor firing angle, as it is proposed. It is worth noting that although the voltage control cannot be performed when the diode rectifier is used, it is demonstrated that the PCC voltage magnitude varies between acceptable values.
Therefore, there are two proposals in this chapter depending on the use of a thyristor rectifier or a diode rectifier. The small-signal models of the systems before applying the control are obtained and used to design the controller parameters. Moreover, the small-signal models with the control strategies embedded are then used to check the stability of the controlled systems. After the controllers design, the AVM is used to get simulation results in Matlab/Simulink against active and reactive power changes at different initial active power levels. The controls performance is appropriate and the results show that the increase of the active power generated increases the reactive power to be injected to control the frequency and the PCC voltage magnitude if the diode rectifier is used. In the case of the thyristor rectifier, the active power increment produces a firing angle decrement in order to maintain the PCC voltage magnitude value. Besides, these results verify that the system response is more damped at higher active power levels, what it had been previously stated in the stability analysis. The centralized frequency control follows an internal reference frequency which is not subject to any measurement noise or grid disturbance being an important advantage of the proposed control. Moreover, this reference frequency could even be a variable frequency if required.
Given that the derivation, the design and the simulation results evaluation of the centralized control strategies are performed by using the proposed AVM, the model with the embedded control is validated against a DSM. Both, the small-signal frequency-response and the large-signal time-response are validated. On the one hand, the small-signal results, which are evaluated by means of the transfer function between the rectifier DC current and the active power generated by the WTGS, show that the AVM response fits the DSM response for frequencies below the switching frequency. On the other hand, the large-signal simulation results also fit the DSM results and demonstrate that the control proposal would achieve the objectives in a real implementation. Besides, the fault response of the controlled system is also simulated and evaluated at the end of this chapter.
After the centralized control proposal, Chapter 4 presents the decentralized frequency control proposal for an OWF which is connected through an HVDC link with diode rectifier. This control is derived based on two control fundamentals which are first presented: the direct frequency control (DFC) and the reactive power sharing strategy (QSS). In order to present the DFC, a single-aggregated WTGS equivalent model of the OWF is also used. The concept is based on the one which has been used in the proposed centralized frequency control because the reactive power is used to maintain the voltage vector oriented to the synchronous reference frame, and thus control the frequency to the one of the reference frame. However, there are two differences between the control concept here and in the proposed centralized frequency control. First, the DFC is applied at a WTGS level in the decentralized control, so the reactive power produced by the WTGS can be used for the control and no additional source is needed. Second, the reference frequency in the decentralized frequency control is not stiff. Each of the WTGSs will follow a variable reference frequency which will allow the synchronous operation of them all. Besides, an analytical small-signal study is presented to demonstrate that the voltage magnitude of the PCC bus is clamped between acceptable values and the active power is naturally evacuated by the diode rectifier. It has also been demonstrated that the direct frequency controller proportional and integral constants provide equivalent damping and synchronizing components if the studied system is compared to the swing equation of a synchronous generator. Concerning the reactive power strategy, a two-aggregated WTGS equivalent model of the OWF is used to justify the chosen strategy. For this purpose, a steady-state study is presented and it is demonstrated that the reactive power of one of the WTGSs must be determined by any additional strategy. Therefore, different power factor strategies are implemented in addition to the QSS, which implies setting the same reactive power level in all the WTGSs. Then, it is demonstrated how the WTGSs would exceed their reactive power limits if the QSS is not used, what would be even worse if a model with a high number of WTGSs is used and one of them is responsible of closing the reactive power balance. Actually, the QSS is more suitable because all the WTGSs help close the reactive power balance, but not only one.
Then, the decentralized frequency control strategy is detailed using the type-4 WTGS model addressed in Chapter 2. The d-component channel tries to maintain the DC voltage magnitude in order to inject all the incoming power by means of the front-end converter. The DFC strategy is performed in the q-component channel of the WTGS in order to align the WTGS output voltage vector to the internal reference frame of the WTGS. Besides, the synchronous operation of the WTGSs is achieved by implementing a droop between reactive power and frequency which modifies the internal frequency of each of the WTGSs depending on the reactive power that they are generating. This control strategy simultaneously leads to an equal reactive power level at the WTGSs without the need of communications.
This decentralized frequency control leads to small frequency deviations which are acceptable if the droop gain is properly chosen. Despite these acceptable frequency deviations, a secondary frequency control is proposed if a stiff frequency control is required. However, communications among the WTGSs of the OWF are then required to send a signal from the master WTGS to the other WTGSs.
Then, the stability of the decentralized frequency control is checked by the small-signal model of the two-aggregated WTGS equivalent OWF model with the control embedded. This model considers the static model of the grid, the classic model of the diode rectifier and its transformer together with the HVDC link and the small-signal model of the type-4 WTGS, which is derived in this chapter.
Both the decentralized frequency control and the decentralized frequency control with secondary regulation are validated in simulation, performing proofs of concept in Matlab/Simulink. However, once the stability is checked, simulation results are obtained in PSIM by using DSMs and a six-aggregated WTGS equivalent model of the OWF. Startup, operation with active power changes and fault response are checked in the simulations which have been performed. Results show an appropriate performance of both versions of the decentralized control proposal.
Finally, general conclusions are addressed, specifying the original contributions, the publications yielded from this Thesis studies and the funding.
