5th Progress report

WP2: Techno-Economic Evaluation



Two scenarios have been modelled in EeFarm:

- the NSTG with wind power and transnational trade integrated in a single DC grid

- wind farms connected by AC cables to the national grids and a separate DC transnational trade grid


The load flow and economic performance of these scenarios can either be calculated by an integral or a partitioned EeFarm model. In the integral model, all components are interconnected in a meshed grid and components controlling the power flow have to be added. In the partitioned model, separate power flows are defined and the connections for these power flows are modelled separately, i.e. not meshed. For instance, in the integral model all wind farms connect to the transnational grid and are evaluated as a single interconnected system. In the partitioned model, each wind farm only connects to its own national grid connection point and is evaluated as such. If the control of the power flow and the component redundancy values (depending on the number of parallel connections) in both approaches are chosen the same, the results of both models are the same as well. Both models have been built and examined. The results showed that the partitioned model was much faster (as could be expected), easier to use and to verify and it requires less post-processing effort. Therefore, the partitioned model has been chosen throughout WP2.


In the final phase of the NSTG, a total of 48 wind farms of 1200 MW are connected to the national grids of six countries. The scenarios have be evaluated by calculating investment costs, transported power, losses, not produced power due to component failure and transport costs per kWh averaged over the life time of the system (levelised transport costs LPC). This has been executed for each wind farm and each transnational trade grid section.


For the NSTG integrated system, increasing the wind power creates parallel connections and this will have an effect on the results, since it increases the reliability of the system. For the AC connected wind farms, this effect is not present since no parallel connections are created by increasing the number of wind farms. However, the evaluation showed that by far the most significant effect is the difference in investment cost of the AC and the DC scenario’s, resulting in approximately the same relative differences in LPC.


The results are based on the 2009 component manufacturer investment data available in the EeFarm database. There are indications that, in the meantime, investment cost may have changed considerably. This will be investigated in collaboration with WP7.


The results will be reported in the next Advisory Group meeting scheduled in September 2012.

WP3: Multi-Terminal Converter


ECN: A Multi-terminal DC (MTDC) network topology has been defined, which represents a part of the NSTG, including three aggregated wind farms and three AC-grid connection points, as shown in Figure 1.


For this network a number of cases have been defined that could be used to evaluate the developed MTDC control strategy with different control options, which is described below.

1.     MTDC start-up

2.     Wind Farm black-start

3.     Normal Operation

a.     TSO priority to 1 country

b.     TSO proportional sharing of the power between countries

c.     TSO orders to switches the direction of the power, e.g. for trading

4.     Wind Farm curtailment (e.g. congestion management)

a.     Medium wind

b.     High wind

5.     Single AC-grid fault

a.     Without curtailment (N-1 secure)

b.     With curtailment (not N-1 secure)


For each case a time series for the active power has been defined for each wind farm and AC-grid node in the network. The wind farm time series have been produced by combining wind turbine time series under different operating conditions and with different time delays in order to obtain a realistic spatial smoothing.

TUD-EPP: 1) A new DC voltage control strategy for the control and operation of DC grids regardless of their size has been developed. The method is called Distributed Voltage Control Strategy (DVC). The DVC control strategy is much more suitable for the control strategy of large multi-terminal DC networks since it assigns each DC-voltage-controlling VSC terminal with a specific voltage set-point. The voltage reference (set-point) is obtained from the solution of an optimal power problem, solved with the previously developed DC load flow algorithm and in conjunction with either a steepest gradient optimization algorithm or with the genetic algorithm developed in WP5. In this way any predefined load flow scenario can be achieved while no single converter is left alone with the responsibility of balancing the power inside the transmission system, i.e. the control of the DC voltage is distributed between several nodes inside the MTDC network. This novel method of controlling the DC voltage inside MTDC grids is, for the time being, suitable only for VSC-MTDC networks, i.e. DC grids that apply only the VSC-HVDC transmission technology.


Through the optimization algorithm, the proposed control strategy is capable of operating MTDC networks with an arbitrary number of nodes while optimizing the system’s functionalities such as system losses, total generation cost, operational limits and network security. In recently submitted paper the steepest descent method was used to optimize the transmission losses in a 7-node MTDC network while guaranteeing the system was N-1 secure. Load flow results and dynamic simulations have been provided to support the findings and to demonstrate that the distributed voltage control strategy can successfully be applied for the control of large multi-terminal DC networks.


2) Through the developed MTDC dynamic models a study was carried out to see whether VSC-HVDC stations which are part of a MTDC network can comply with most common requirement from Grid Codes. The study was largely based on the Grid code requirements from the German TSO, E.ON Netz and the Spanish TSO.


Due to their flexibility and control capabilities, VSC-HVDC terminals can comply with the most common TSO requirements, even when sitting inside MTDC networks. However, while it is true that VSCs can control their active and reactive power independently, they can only do so within their rated capability. If the reactive power requirements are too strict, they will only be met in detriment of the converter active power and vice-versa. The study concluded that the impact of the grid code requirements, especially the supply of reactive current during voltage dips, on the operation of a MTDC network during high wind scenarios can be substantial. Due to their physical location, offshore wind farms will tend to be connected to the AC networks on peripheral locations. Therefore, there is the need for more detailed AC models in conjunction with DC models, in order to analyze the true efficacy of these requirements and to see whether they can be diminished, thus facilitating the control and operation of the MTDC network.


WP4: Multi-Terminal Converter testing


TUD-EPP: The three-small AC/DC converters, acquired from Belgium company Triphase®, have been commissioned at the EPP laboratory at TU Delft on February 7th. On February 8th, a short course on how to operate the converter was provided. The commissioning and operation course were followed both by TU Delft and JRC personnel. A picture showing two of the three cabinets can be seen below.



The models developed in WP3 have already started being ported into the AC/DC cabinets for testing purposes and model validation. The original phase-lock loop block and DC voltage controller provided by Triphase® have already been substituted by versions developed in TU Delft. The next blocks which need to be ported onto the cabinets are the active and reactive power controllers and the inner-current controller. This work is currently being performed.

The next steps in the laboratorial setup include the connection of two cabinets in a back-to-back fashion for control testing and model validation. After the successful connection between two cabinets has been accomplished, the third one will also be included forming a small MTDC network in the EPP laboratory. Studies are being carried out in order to validate what are the possibilities for simulating an external DC grid more similar to the one which will take place in the North Sea. The idea is to simulate the behavior of long DC cables in the laboratory. Preliminary results showed that lumped components might provide a good compromise between accuracy and practical implementation.


After all three cabinets have been successful connected and jointly operated, the scaled-down MTDC grid will be connected to a Real time digital simulator (OPAL-RT) at the Joint Research Centre (JRC-IET) in Petten. The idea is to integrate the WFs models into the scaled-down setup via the digital simulator, providing a hardware-in-the-loop type of setup in which all the main components inside a MTDC network for the integration of offshore energy in the North Sea are present.


WP5: Optimization.


ECN: Matlab steady state component models have been derived from the EeFarm steady state electrical and economic model (Simulink). These models will be used in optimization process to determine the electrical losses, not produced power due to component failure, electrical system investment costs and contribution of the electrical system costs to the price of the produced wind power. The Matlab component models use the EeFarm component parameter database.


TUD-EPP: The first step of the WP5 was the development of a multi-objective genetic optimization algorithm (MOOA). With such algorithm it is possible to optimize, through the maximization or minimization, of several objective functions simultaneously. In this way, an optimal trade-off between the objectives that are being considered is possible to be obtained.


The first application of the developed MOOA was the comparison of offshore power transmission technologies. The comparison was performed in a multi-objective basis and the objective functions used were the total investment costs and the energy lost in the transmission system itself. Both objectives were set to be minimized in the MOOA. The transmission systems considered were the HVAC technology and the HVDC VSC-based.

Firstly, the total investment costs and the energy losses were minimized in a single-objective basis for both transmission systems when different distances between the offshore wind farm (OWF) to shore were considered (50 to 150 km). After, trade-offs between the total investments costs and the energy losses, for different distances to shore, were obtained for both transmission technologies.

The components and their respective models used in this work were stored in a database that was provided by the Energy Research Centre of the Netherlands (ECN).

This work was presented in a conference in TUDelft: Young Researchers Symposium 2012.


Another case study was performed. Once again the developed MOOA was employed. The objective functions used were the transmission power losses and the social welfare. Now, instead of only a transmission system between an OWF and shore, a meshed-parallel multi-terminal DC (MTDC) network with 11 nodes and 11 DC cables was considered. Five OWFs and three onshore grids are connected to the grid. The onshore grids were considered to be countries, the United Kingdom, the Netherlands and Norway. Two of the OWFs belong to the UK, other two to the Netherlands, and the last one to Norway. Market clearing prices for electricity in the countries were used.

Optimal trade-offs between the social welfare and the transmission losses were obtained. Additional constraints to the problem were added in order to evaluate their influence on the optimal trade-off. Moreover, system security was taken into consideration. In this way, all the operating points obtained through the MOOA are guaranteed to be N-1 secure. Such feature makes sure that, in case of an outage in one of the DC voltage controlling station, the stability inside the MTDC grid is assured. This work resulted in a paper that will be presented in a conference: Energy Conversion Congress & Exposition (ECCE) Asia 2012.


A case study were the VSC-HVDC dynamic models of the WP3 and the MOOA are used is being developed.  The main purpose is to obtain optimal power flows, were the transmission losses are to be minimized, inside a MTDC network. The DC voltage reference values obtained with the MOOA will be employed in the VSC-HVDC dynamic models in order to validate the results.



WP6: Grid


Work Package 6.1: TUD-EPS

A Master's student (Line Bergfjord) has preliminarily investigated the sensitivities with regard to operational costs for several offshore grid topologies [1]. Though interesting determining parameters were found, it was concluded that more realistic market modeling including  more regions and an adequate treatment of hydro dispatch is a requisite.


Another Master's student (Aymeric Buatois) has completed the first part of his thesis to identify ramping events in offshore production and also confirm that corrections made to the wind speed data from the recently purchased meso-scale model were reasonable.


Parallel to this, the market model and the grid model for performing the round-the-year security analysis were developed. To that purpose a scenario of load and generation for the year 2030 was developed and detailed wind and solar power time series were created. After a market-based RES integration study performed for the North Sea area, without considering any NSTG [3], the base case for the upcoming NSTG simulations has been adjusted. The next step is to run market simulations for various NSTG topologies and then to run the subsequent security analysis. The market model consists of 8 countries, while the onshore grid model that will be used is smaller (due to data confidentiality issues) and is focused mainly on the Dutch transmission grid, with simplified representation of its neighbors.


Work Package 6.2: TUD-EPS

The dynamic model made available by TenneT has been checked for the following:

1) Coverage of geographical area;

2) level of detail and its applicability for the North-Sea region; and

3) applicability for small-signal stability analysis.


It turned out that the grids of Western Denmark and the UK were not included. Therefore, assumptions had to be taken for these countries for transient stability studies. Data from literature formed a basis for ongoing amendments. The remainder of the dynamic model consists of a detailed model of the Dutch transmission system as well as the surrounding countries (Belgium/Germany). Other continental European countries are modeled with less detail. As this is the only available model, it was agreed (with the NSTG project partner ECN) that no conclusions will be drawn about the impact of the transnational grid on the stability of countries other than the Netherlands. The non-harmonious degree of modeling detail has led to the conclusion that the TenneT model will not give realistic results for inter-area oscillations. Consequently, the small-signal stability research is now being focused on methods to damp oscillations with multiple VSC in-feed points and on the efficacy of NSTG landing points for damping of modes in the other two synchronous systems, UK and the Nordic system.


Grid models of the Great Britain and Scandinavian system have been entered based on publicly available literature and tested in PSS/E to complement the TenneT grid model. Several decisions about how to couple the detailed simulations have been made, but more work remains in this challenging task.


In the area of transient stability analysis, a Master's student (Paulo Chainho) has developed a general multi-terminal VSC-HVdc model for PSS/E [2]. Currently ongoing work with a new student (Mario Ndreko) is the implementation of power management methods developed for WP3. Next steps include the interaction between these control systems and emergency controls like fault-ride through and  power system stabilization.


Regional availability of ancillary services from wind turbines (such as power system stabilization and inertial response) were first roughly assessed based on TU Delft's meso-scale weather model output for a relevant North Sea wind power development scenario. A finer assessment has also been made by applying assumptions at the farm level and turbine level. Information about the diversity of wind speeds within a farm has been extracted from available wind farm data and also compared with empirical results in the literature.  Methods for computing the maximum available power system damping energy at different frequencies from a single wind turbine have been developed and applied.


Another Master's student (Trent Ratzlaff) has developed time series of synchronous inertial energy present in each of the three synchronous systems around the North Sea: Great Britain, Scandinavia and Continental Europe.  This work has been based on the market simulation output of Work Package 6.1. Hours with low inertia, low import, and high import have been noted for subsequent study with more detailed simulations. 


WP7: Costs, benefits, regulatory and policy aspects


ECN: Voor WP7 zijn door ECN voorbereidende werkzaamheden uitgevoerd om geschikte inputs te genereren voor de economische analyses met COMPETES wat betreft wind opbrengsten, elektriciteitsvraag en aanbod in 2030. Verder zijn er aanpassingen van het COMPETES model voorgesteld om nodes op zee mogelijk te maken.


WP8: IEA Annex 25 collaboration


TU Delft contributes to IEA Task 25 activities.