Degree project in
Implementation of DC Supervisory Control
Optimal Power Flow Calculator
MUHAMMAD HASSAN FIDAI
Stockholm, Sweden 2014
ICS
Master Thesis
Implementation of DC Supervisory Control (Optimal Power Flow Calculator)
by
Muhammad Hassan Fidai
A thesis submitted to the School of Electrical Engineering
in partial fulfilment of the requirements for the degree of
Master of Science in Systems Controls and Robotics
Department of Industrial Information and Control System KTH
October 2014
Abstract
Integration of renewable resources such as remote solar or wind farms and electric power trading between neighbouring countries lead to new requirements on the development of the transmission grids. Since AC grid expansion is limited by e.g. legislations issues, High Voltage Direct Current (HVDC) technology with its diverse benefits compared to AC is being considered as appropriate alternative solution. The developed HVDC grid can be either embedded inside one AC grid or connects several AC areas. In both architectures, the separate DC supervisory control can be proposed to control the HVDC grids using the interfacing information from AC Supervisory Control And Data Acquisition (SCADA). The supervisory control is supposed to calculate the optimal power flow (OPF) in order to run the system in the most optimal situation. Based on the architecture, the required information, boundary of the system and also objective function can vary.The aim of the thesis is to present the findings of a feasibility study to implement a supervisory control for bipolar Voltage Source Converter (VSC) HVDC grids in possible real time platforms. DC supervisory control has a network topology manager to identify the grid configuration and employs an OPF calculator based on interior point optimization method to determine the set-point values for all HVDC stations in a grid. OPF calculator takes into account the DC voltage, converter and DC line constraints.
Acknowledgements
First of all my deepest gratitude to my supervisor Davood Babazadeh. It is an honor and pleasure for me to have him as my supervisor. I am grateful to him for putting his confidence in me and helping me in securing this thesis. His patience and support helped me through the research work to implementation and even in the writing of the final report. Without his guidance, this project cannot reach the same level as it is today.
I want to express my gratitude to Prof. Lars Nordstrom not only for his contribution as the examiner of this thesis but also to provide the direction to follow during the entire period of the thesis. It is a privilege for me to have him as my professor. Surely he is one of the best lecturers I had during my entire academic life.
I am also great full to ABB for providing me with the opportunity and funding to carry out this thesis. My warm gratitude to my manager Tomas X. Larsson for entrusting me with this opportu- nity and providing all the logistical support necessary for the thesis. I am also thankful to him for arranging technical trainings which did not only help me with my thesis but also helped me with my personal career development. My deepest gratitude to my supervisor at ABB Jonathan Hanning despite his full agenda he always found time to help me out and arranging the needed resources.
I would also like to thank Mats Larsson from ABB, Switzerland for sharing his work and helping me out during the different phases of the project.
I would extend my gratitude to Matus Korman who helped me extensively when it came to the computer science aspect of the thesis. My thanks to Jens M ˙alare who always found time to respond to my emails and helping me out during the hardest part of the project.
To my colleague student Arvind Muthukrishann for providing his support in collecting results during the final phases of the project.
Finally I would like to thank Swedish Institute for funding my master studies and my stay in Sweden. Without their scholarship I would have not been able to study in one of the most prestigious engineering institute.
Muhammad Hassan Fidai Stockholm, Sweden October 2014.
Abbreviations
HVDC High Voltage Direct Current LCC Line Commutated Converter VSC Voltage Source Converter PWM Pulse Width Modulation
SCADA Supervisory Control And Data Acquisition PCC Point of Common Coupling
OPF Optimal Power Flow MKL Math Kernel Libraries
Contents
Abstract ii
Acknowledgements iii
Abbreviation iv
List of Tables vii
List of Figures viii
1 Introduction 1
2 Background 3
2.1 HVDC Technology . . . 3
2.1.1 Line Commutated Converter HVDC . . . 4
2.1.2 Voltage Sources Converter HVDC . . . 5
2.2 Voltage Source Converter HVDC . . . 6
2.2.1 Equipment in a Converter Station . . . 6
2.2.2 Control System . . . 7
2.2.2.1 Inner Control Loop . . . 7
2.2.2.2 Outer Control Loop . . . 7
2.2.3 VSC Modelling Approaches . . . 9
2.3 VSC-HVDC Grid . . . 10
2.3.1 Grid Topology . . . 10
2.3.2 Protection . . . 10
2.3.3 HVDC Grid Control . . . 11
2.4 OPNET Modeller . . . 12
2.5 DC Supervisory Control . . . 12
3 Results and Discussion 14 3.1 VSC-HVDC Grid Model . . . 14
3.1.1 Station Model . . . 14
3.2 Scenarios . . . 16
3.2.1 Variable Generation . . . 16
3.2.2 Station Disconnection . . . 19 3.2.3 Islanding . . . 20 3.2.4 Line Disconnection . . . 23
4 Conclusions 26
List of Tables
3.1 VSC terminals power ratings and base case control modes [38] . . . 14 3.2 HVDC GRID line parameters [38] . . . 15
List of Figures
2.1 Four-quadrant diagram with the voltage reference [14] . . . 4
2.2 Structure of VSC Station [16] . . . 6
2.3 Block digram of complete inner controller with vector current control scheme [21] . . 8
2.4 Block digram of complete control system for VSC station [21]. . . 8
2.5 Proposed Control Hierarchy for DC grids . . . 13
3.1 7-Terminal VSC-HVDC model simulated in OPAL-RT [38] . . . 15
3.2 Model of station simulated in OPAL-RT . . . 16
3.3 DC grid model for variable wind farm generation . . . 17
3.4 Results for scenario 1, Variable generation from wind farm) . . . 18
3.5 DC grid model for station disconnection . . . 19
3.6 Results for scenario 2, Station disconnection . . . 20
3.7 DC grid model for islanding . . . 21
3.8 Results for scenario 3, Islanding . . . 23
3.9 DC grid model for line disconnection . . . 24
3.10 Results for scenario 4, Line disconnection . . . 25
Chapter 1
Introduction
The electrical power markets are currently facing several challenges. The ever increasing demand of power with more focus being put towards the integration of renewable energy sources have presented with several technological issues. European union’s target to generate 20% of the power from renew- able energy sources by 2020 [1] has resulted in initialization of several projects, such as harvesting power from the off-shore wind farms in the north sea. Moreover there have been project proposals to exploit the solar power potential of Africa and its integration within the European grids. The integration of these large scale renewable energy sources, the will to boost transnational electricity trade and the desire of security of supply requires a novel approach to expand the current power systems.
High Voltage Direct Current (HVDC) technology is being used to transmit power over long distances and to connect different AC systems for past six decades. Line Commutated Converter (LCC) and Voltage Source converter (VSC) are the two HVDC technologies currently available. The later is a newer technology and has significant advantages over LCC when it comes to integration of off-shore wind energy [2] or to provide reactive power support to the connecting AC system. Unlike LCC, VSC technology also provides black start capability and also removes the need of strong AC networks. A multi-terminal VSC-HVDC grid has been proposed in the literature [3–6] to overcome the aforementioned challenges. Such HVDC grid will be augmented or integrated within a single or different AC systems.
A transmission in an AC system is supervised and control by SCADA. Where as local controllers are used to control the current HVDC converter stations used for point to point connection. The control of a future proposed DC grid which is going to be a bridge between SCADA and DC local controller is currently a critical research topic and several techniques have been proposed [7–12].
This thesis provides a feasibility study on the implementation of such a DC supervisory control on different platforms. Moreover the proposed DC grid supervisory controller is implemented on a Linux machine and is integrated in a simulation platform.
The implemented DC supervisory control has a network topology manager to identify the grid configuration and employs an OPF calculator based on interior point optimization method to de- termine the set-point values for all HVDC stations in a grid. OPF calculator takes into account the DC voltage, converter and DC line constraints. The work carried out in the thesis can be extended
to include more features for the future DC supervisory controller.
The master thesis has been carried out in collaboration with ABB. The control algorithm and the platform used for implementation is proprietary hence this report only presents a brief background study and the results of the master thesis.
Chapter 2
Background
2.1 HVDC Technology
Electric power sector can be characterized into three main activities generation, transmission and consumption. Electric power is generated by the transformation of various form of energies into electrical energy at the generation centers. This power is then transferred to the end users via electrical transmission and distribution network. The evolution of electrical systems started with the Edison’s DC generators. Due to non-availability of technology for DC voltage conversion, power was generated and had to be transmitted at low DC voltage potentials. This voltage potential was primarily set by the consumer load devices like an electric bulb. Hence it could not be transmitted over larger distances due to significant losses and voltage drops in the transmission network.
These drawbacks of the DC transmission systems proved to be the driving force for the develop- ment of AC systems. Transformers allowed to raise the voltage potential of the generated AC power which then can be transferred over long distances with acceptable power losses. Hence AC systems were adopted world wide as the standard electrical power systems.
AC power systems came with their own challenges such as requirement for the synchronized op- eration of the entire system and the reactive power losses. As the power demand increased and grids were pushed to their limits these challenges became more prominent. Moreover for long distances the reactive power losses are magnified to the extent that active power transfer is no longer viable.
The need for HVDC transmission was felt quite soon after the evolution of AC systems. How- ever there were many technological barriers in the way towards HVDC systems. The world’s first HVDC link based on mercury arc valve technology was commissioned between main land Sweden and Gotland in 1954. Ever since the accumulated installed power of HVDC transmission has in- creased steadily. HVDC links provided several advantages over AC systems which include but are not limited to, no requirement for the synchronization of the systems being linked, no technical limit to the length of over head or submarine cables and immunity from impedance, phase angle, frequency and voltage variations. HVDC links can also be used to improve the stability of the connecting AC systems therefore increasing their power carrying capacity. Currently most of the HVDC links installed world wide are only point to point connections.
The ever growing energy demands and the environmental considerations have increased the ef-
Figure 2.1: Four-quadrant diagram with the voltage reference [14]
forts towards the integration of large scale renewable energy sources. Such energy sources are located at great distances from the load centres. The need of transmission of bulk amount of power from such sources and its distribution among different AC systems when the AC grids are being already operated at the limits posses a great challenge for power companies. Hence there has been consider- able attention towards the application of HVDC grid on top or in compliment with already existing AC grids to overcome these challenges.
2.1.1 Line Commutated Converter HVDC
After the advent of HVDC technology in 1954 the next big technological step was the invention of thyristors based converters for HVDC stations. Thyristors based HVDC converters are called Line Commutated Converters (LCC) HVDC. Thyristor is a semi controllable, uni-directional solid state device which can be turned on by applying a positive pulse to its gate. However for turning off, it depends on the voltage across its terminals to become negative. Since AC system commutates the voltage of the converter hence the name Line Commutates Converters. Over the years the power ratings of thyristors have increased which has driven the increase in the maximum amount of transferable power via LCC-HVDC systems. The LCC-HVDC technology has been quite matured and still is the first choice for bulk power transfers over long distances.
A LCC-HVDC station can operate in the lower two quadrants of the complex diagram shown in figure 2.1. It allows the bi-directional flow of active power but always consumes reactive power from the AC system. Due to uni-directional current flow of the thyristor the DC current always flows in
forward direction but the voltage can be reversed.
LCC-HVDC stations provides full controllability of active power at the expense of varying de- mand for reactive power. This varying demand of reactive power has to be compensated by the switching of filters and extra capacitors via circuit breakers to keep the power factor close to unity.
It also requires for the connecting AC system to have certain minimum short-circuit level to with- stand the voltage fluctuations [14].
Despite the limited controllability and challenges of LCC-HVDC technology, it is still given prece- dence over newer VSC-HVDC technology (discussed in next section) for long distance bulk power transfers.
2.1.2 Voltage Sources Converter HVDC
In recent years the production of high power rated IGBTs have enabled the development of more flexible HVDC system called Voltage Source Converter (VSC) HVDC. An insulated gate bipolar transistor (IGBT) unlike thyristor is a fully controllable switch and hence can be turned off regardless of the voltage potential across its terminals. Hence in contrast to LCC, VSC-HVDC stations are self commutating.
VSC-HVDC transmission permits the control of active power as well as the reactive power in either direction independent of each other. Hence VSC-HVDC can operate in all four quadrants shown in figure 2.1.
Self commutating characteristics of VSC-HVDC allows its stable operation irrespective of the AC system’s short circuit capacity. Moreover the harmonics generated by VSC are considerably lower than the LCC and hence the filters size is reduced to absorb only higher order harmonics.
VSC technology has proven to be promising for future HVDC projects because of the following reasons [15].
• Active and reactive power can be controlled independent of each other, hence VSC can also be used to provide reactive power support to the connecting AC system.
• Flow of active power can be reversed without changing the voltage polarity.
• Self commutation of VSC reduces the risk of commutation failure.
• VSC technology has black start capability since AC voltage can be generated from the DC side.
In this application, inverter controls the frequency and the voltage of the receiving system.
• No minimum DC power flow restriction unlike LCC-HVDC that requires certain level of DC power flow for successful commutation.
• VSC-HVDC allows the possibility to inject power from an offshore wind farm to an onshore AC system.
Figure 2.2: Structure of VSC Station [16]
2.2 Voltage Source Converter HVDC
2.2.1 Equipment in a Converter Station
VSC converter as the name indicates has a constant voltage source on the DC side. This voltage source maintains the required voltage potential irrespective of the magnitude or polarity of the current through it. The basic structure of the VSC station is shown in figure 2.2 [16].
DC Capacitor The DC side of the VSC has a stiff voltage and hence is extremely capacitive.
However the switching of the valve produces harmonic currents on the DC side. These currents due, to the DC side impedance give rise to a voltage ripple. A capacitor on the DC side is used to filter out this voltage ripple [17]. While the DC capacitor improves the steady state response of the station, a too large capacitor can effect its dynamic response. Hence both steady state and dynamic responses have to be considered while selecting the capacitor [18].
Transformer HVDC converter is connected to the AC system via transformer. The main purpose of the transformer is to convert the AC system voltage to the voltage level suitable for the converter.
This transformer has to be specially designed to handle the DC stress exerted by the converter. Other than the voltage transformation, its functions also include the reduction of harmonics, especially the 5th and 7th harmonics, to act as a galvanic barrier between the AC and DC system and to provide reactive impedance in the AC system [19].
Phase Reactors Though some of the series reactance is provided by the transformers, but in order to provide the necessary reactive impedance to the AC system to reduce the short circuit currents, phasor reactors have to be used. They also control the rate of rise in valve current during commutation.
AC Filters In addition to the series connected inductances, AC filters are used to eliminate the voltage harmonics entering the AC system.
2.2.2 Control System
VSC generates a fundamental frequency AC voltage from the DC voltage. The control of this AC voltage is the primary function of VSC. The fundamental control of VSC is through Pulse Width Modulation (PWM) control of its valve [14]. The operational PWM frequency can vary greatly depending on the valve design. But ultimate goal is to control the magnitude and the phase angle (φ) of the generated AC voltage.
The magnitude of phase angle φ defines the amount of active power flowing to or from the VSC.
Whereas the sign of φ decides the direction of power flow i.e whether the VSC works as an inverter or a rectifier.
The flow of reactive power is controlled by the magnitude of the voltage generated by VSC.
When the converter voltage is greater than the AC system voltage the reactive power is injected by the converter to the AC system and when the converter voltage is less than the AC system voltage, reactive power is absorbed by the converter. The control structure of the VSC station can be divided into two levels, inner control loop and outer control loop.
2.2.2.1 Inner Control Loop
The inner control loop of the VSC controls the magnitude and the phase angle of the voltage generated by it. Typically there are two type of schemes used for the inner control loop [7].
m−φ Direct Control In this control scheme the magnitude and the phase angle φ of the generated voltage are directly controlled. m is known as the modulation index and is equal to the ratio of generated converter voltage and the AC system voltage. Active power of the VSC is more sensitive to φ as compared to m whereas reactive power is more effected by m rather than φ. Hence φ is primarily responsible for controlling the active power and m controls the reactive power of the VSC.
d − q Vector Current Control This control approach is based upon representing the three-phase AC quantities by an equivalent set of two-phase quantities resulting in identical resultant space vector as the original three-phase space-time phasor representation [20]. The d − q vector control scheme originated from electrical machines and drives area and is now extensively used for VSC control. The major advantage of this control scheme is that it enables a fully decoupled linear control of active and reactive power of VSC. The block diagram of the inner control loop with vector current control is shown in figure 2.3 [21].
2.2.2.2 Outer Control Loop
The references of the inner control loop are provided by the outer controller. The outer controller consists of two control loops one for active and the other for reactive power. The block diagram of the VSC-HVDC station control system showing inner controller and most commonly used outer controller control schemes, is presented in figure 2.4 [21].
The junction point where VSC unit is connected to the AC grid is known as point of common coupling (PCC). For reactive power control loop VSC-HVDC stations are commonly operated in either constant reactive power control mode or in constant AC voltage control mode. In former the
Figure 2.3: Block digram of complete inner controller with vector current control scheme [21]
.
Figure 2.4: Block digram of complete control system for VSC station [21].
station is controlled such that the reactive power at PCC remains constant, whereas for later the aim is to keep the AC voltage constant.
There are different control modes available for active power control. Most commonly used active power control modes are further discussed below.
Power Control Mode A station operating in active power control mode always ensures that the active power at PCC always remains equal to a certain reference value. Hence it keeps the active power at a constant as long as the reference value has not changed.
Constant Voltage Control Mode VSC-HVDC station operating in voltage control always fol- lows the DC voltage reference of the bus it is connected to.
Droop Control Mode If the aforementioned two type of control modes are combined we get droop control mode also known as power droop against local DC voltage. It is inspired by the schemes applied for primary frequency control in AC systems. In this mode active power of the station changes linearly with the DC voltage. The droop or droop constant Dpdefines the sensitivity of the active power to the voltage error.
Droop Control With Deadband This control mode is similar to the classical droop control mentioned above however it has an additional deadband function active on the DC voltage such that for small voltage deviation there is no change in power.
Frequency Control Mode The full controllability of the power being injected in the AC grid from the VSC-HVDC station allows it to be considered as a virtual synchronous machine. This makes it easy to employ frequency droop control mode for the VSC-HVDC station.
2.2.3 VSC Modelling Approaches
VSCs can either be modelled in detail i.e. including all semiconductor components or by time- average approach. In detail modelling of VSC the electrical model of each semiconductor device is used as a single unit in the entire model. The type and the number of voltage levels of the VSC are clearly shown in the detailed model. Such VSC models are usually used for analysing pulse width modulation (PWM) techniques, studying different converter topologies and for carrying out high order harmonics analysis for precise loss calculation [7].
In the time averaged VSC models there is no distinction between the modulation techniques, converter topology or the voltage levels. However such models are satisfactory for the study of phenomena involving the fundamental frequency voltage and current components. Time average model consists of the controllable AC voltage sources connected to the AC side and controllable current sources connected to the DC side.
2.3 VSC-HVDC Grid
The history of HVDC systems go back to around 60 years and the technology has evolved expo- nentially in past couple of decades. However so far HVDC systems have only been used for point to point connection. In the past DC grid was technologically inconceivable due to the lack of one important component, DC breaker. In the absence of a DC breaker a fault on the DC line cannot be isolated and the whole DC system has to be shut down to remove the fault. With the invention of the DC breaker all the main technological components are available and DC grids are the way forward towards the smart grids.
2.3.1 Grid Topology
The topology of a DC grid has vital impact on its control architecture and protection system. More- over the cost of development and operation of a grid also has a direct correlation with its topology.
The simplest grid topology is to have a radial configuration i.e. the main generation source is located at the middle and the network branches originate from it distributing power to various ter- minals. However the major application of the DC grids is the integration of large renewable energy sources in the existing AC system and such energy sources are located at great distances from the load centres. It might also be desirable that the power from such resources is distributed among different AC systems. In such scenarios it is of great interest that if required the VSC grid can also be used to transfer power between different AC systems. Radial grid is not a viable option for such VSC HVDC grid and hence most of the proposed future grid are assumed to have meshed configuration.
Meshed grid also have a major advantage over radial grid when it comes to handling of contin- gencies. If there is a fault in the element of a DC grid, it can be localized and isolated. The power then can be rerouted through the non faulty elements to maximize the power flow in the remaining system. Fault detection for VSC-HVDC grid is a challenging topic and [22] can be reviewed for more details.
Meshed grid has several advantages over other grid configurations however its control is much more challenging. DC voltage in HVDC is an important parameter analogous to frequency in an AC system [23]. Its variation indicates the power unbalance in a DC system. Moreover the dynamics of a DC system are extremely fast compared to the AC system and stable operation of a DC system demand fast actions. Hence automatic measures must take place in order to keep the DC system stable.
Meshed DC grids are also prone to circulating current in the sections of the grid forming a loop.
Slightly unbalanced power can give rise to such circulating currents which can account for major losses in a DC system. Hence a fast and sensitive power flow control is necessary for the operation of meshed DC grid.
2.3.2 Protection
Protection of the DC grid is one of the most important operational concerns. In DC grid a short circuit between line to ground or between line to line can cause over currents. In LCC-HVDC systems the magnitude of these over currents is not expected to be too large due to the large DC
smoothing reactance. However the discharge of DC link capacitor in VSC-HVDC can cause huge short circuit currents. Moreover an open circuit in a DC grid or loss of a converter due to a fault can cause over voltages.
In point to point DC connection, traditionally when a fault occurs, the circuit breakers on the AC side are used to shut down the entire system. Schemes have also been developed and presented in the literature to clear the faults in a DC grid using AC breakers [24]. However in all such schemes the entire DC system has to be shut down until the removal of the fault. This is not a viable option for future smart DC grid. It is desirable that in the case of a fault in HVDC grid, that section is isolated to ensure the stable operation of the rest of the grid. In other words a DC breaker is required for the realization of a DC grid.
The major challenge with the DC breaker is the non existence of zero crossing in the DC system, as in AC system. [25] presents a resonant circuits to achieve zero crossing in a DC system.
2.3.3 HVDC Grid Control
As described in section 2.3.1 the dynamics of DC grid are fast and an efficient and accurate control based on power flow calculations of the grid is required for the stable and optimal operation of the entire DC system. Such power flow calculations and control of the grid can be provided by the design of a supervisory control for the grid.
The supervisory control is going to be responsible for monitoring all the terminals and lines in the grid, calculate the OPF and on the bases of the OPF results, assign the set points of the local controllers for all the terminals in the grid.
In recent years the OPF of a DC grid has been quite popular research topic. Most of the work has been carried out on the combined AC/DC load flow and can be subdivided into unified and sequential methods. In the unified approach both AC and DC system equations are solved simul- taneously [8], whereas in sequential method, first AC system equations are solved and then the DC system equations [9]. [10] argues that a sequential approach is more convenient since it can be imple- mented as an addition to the existing AC power flow programs. It continues and presents a detailed general, steady state VSC-HVDC model for sequential AC/DC power flow. However [11] advocates for unified approach and argues that solving the AC/DC systems of equations one after the other introduces high number of iterative loops making the algorithm computationally expensive and less reliable. It presents a unified approach of AC/DC combined power flow while taking the converter losses under consideration.
The fundamental component of any optimal power flow scheme is the optimization solver. Dif- ferent optimization approaches have been presented in the literature for DC OPF. [12] presents the model of VSC-HVDC suitable for optimal power flow solution using Newton Raphon’s algorithm.
Where, [26] presents a novel approach to use genetic algorithm to obtain optimal and controllable power flow for a DC grid. A second order cone programming formulation of the AC/DC power flow problem has been presented by [27] which is solved using interior point optimization method. [28]
also present the use of interior point optimization method for OPF. Different optimization solvers for combined AC/DC power flow are tested by [29] and IPOPT has been declared to provide the best results. IPOPT solver uses an interior point line search filter method and is commonly used in solving large-scale nonlinear optimization problems.
The DC supervisory control implemented in this thesis only considers the DC system. The op- timal power flow problem is formulated for the VSC-HVDC grid and IPOPT solver is used to solve the optimization problem.
2.4 OPNET Modeller
OPNET modeller is a discrete event communication simulator with built in models for LTE, WIMAX, UMTS, ZigBee, Wi-Fi, etc [32]. It is capable of performing fine-grained and detail simulations of the communication network incorporating terrain, mobility and path-loss characteristics. It provides convenient high level graphical user interface with an access to a diverse library of blocks based on C and C++ code. OPNET also provides a system-in-loop module (SITL) through which a simulation model can be connected to live network hardware by providing interfaces or gateways. Each SITL module inside the simulation environment is assigned to a specific network adapter which can be a real interface or a virtual one.
Communication networks play a very crucial in the future smart grids. OPNET provides the means of studying the most efficient topology of the communication network, physical media and protocols etc. required to conceive the future smart grid. Hence it is being extensively used with real time power system simulators to provide a co-simulation platforms to study the communication challenges of the future power systems.
2.5 DC Supervisory Control
The DC supervisory control provides the control strategy for the coordination of multiple VSC- HVDC terminals connected in a grid configuration. The control hierarchy of the DC grid is shown in figure 2.5. Each station is equipped with the fast modulation control schemes similar to that of point to point VSC-HVDC station. There is also a station control scheme that aims at tracking local or global reference values based on the active and reactive power control mode of the station.
In case of contingencies this control layer is first to respond to ensure stability of the DC voltages.
The DC supervisory control responds to the contingencies on the AC and DC side and periodically recompute the set point references for the station control systems based on the measurement and line and converter statuses from the grid. DC supervisory control also optimizes the the post contingencies set points to reduce the loses in the system. In this way supervisory control also helps to minimize the effect of the DC side contingencies on to the neighbouring AC system.
The DC supervisory control bridges the gap between the power schedule which is usually received from the SCADA/EMS and has a time scale of tens of minutes and the time constants of the DC grid.
The dynamics of the DC grid usually settle on the order of hundreds of milliseconds. To coordinate and optimize the power from several DC terminals in the grid it is necessary to periodically compute and update the set points of each station in real time. This becomes even more necessary in the event of a fault which causes a change in grid topology.
DC supervisory control tracks the schedule and computes the set points which ensure the grid function within the operational limitations of the grid and the stations.
Station Control
VSC1 Station Control
VSC2 Station Control VSC3
Station Control VSC4
Station Control Station Control VSC5
VSC6 Station Control
VSC7
L46 L12
L24 L23
L35
L47
L57
DC Supervisory Control
SCADA/EMS
Figure 2.5: Proposed Control Hierarchy for DC grids
Chapter 3
Results and Discussion
3.1 VSC-HVDC Grid Model
The 7-terminal VSC-HVDC Grid model presented in [38] has been simulated in OPAL-RT simulator and is shown in figure 3.1. The simulated grid is uni-polar whereas the DC supervisory control is designed for bi-polar VSC grid. Moreover the simulated model uses averaged model for the VSC as described in section 2.2.3. Hence the DC grid model has been mapped to the bi-polar detailed HVDC model which can be handled by the supervisory control.
In order to simulate the DC line fault scenarios leading to line disconnection, ideal switches were installed on the DC lines in the model. The parameters for VSC-HVDC terminals and the and DC lines are shown in table 3.1 and 3.2 respectively.
Terminal Power Rating
[MW] Control Mode
VSC1 200 Constant PCC active power VSC2 300 Constant PCC active power VSC3 150 Constant PCC active power VSC4 200 Constant PCC active power
VSC5 300 Constant Voltage
VSC6 100 Constant PCC active power
VSC7 50 Constant PCC active power
Table 3.1: VSC terminals power ratings and base case control modes [38]
3.1.1 Station Model
The VSC station model used in OPAL-RT is shown in figure 3.2. Various parameters of the station are provided below
• Transformer Resistance, R = 0.0025p.u.
T1 VSC 1
T2 VSC 2
VSC 5 T4
VSC4
Grid
Grid VSC 6
T7 VSC7
Grid
T3 VSC 3
Grid L46
L12 L24
L35 L23 L47
L57 T6
T5
Figure 3.1: 7-Terminal VSC-HVDC model simulated in OPAL-RT [38]
Lines Distance [km]
Resistance [Ohm]
Inductance [mH]
Maximum Current [kA]
L12 413 5 43.6 1
L23 248 3 26.2 0.5
L24 207 2.5 21.9 1
L35 331 4 35.0 1
L45 83 1 8.76 0.5
L46 207 2.5 21.9 0.5
L47 289 3.5 30.5 0.5
L57 165 2 17.4 0.5
Table 3.2: HVDC GRID line parameters [38]
VSC Converter
Transformer Phasor Reactor
AC Filters
27th Harmonic
54th Harmonic
Figure 3.2: Model of station simulated in OPAL-RT
• Transformer Reactance, Xt= 0.075p.u.
• Phasor Reactor Resistance, Rr= 0.0015p.u.
• Phasor Reactor Resistance, Xr= 0.15p.u.
• Shunt Capacitance, C = 6.84µF
• DC side admittance modelling switching losses, Yshloss= 0.0017p.u.
3.2 Scenarios
Simulated DC grid model described in section 3.1 is tested with the DC supervisory control for various scenarios. For all cases, VSC stations are operated to be at a maximum of 80% of their rated power. Following are the scenarios DC supervisory control has been tested for
• Variable Generation
• Station Disconnection
• Islanding
• Line Disconnection
3.2.1 Variable Generation
In future smart transmission grids wind power will hold considerable amount of share. One of the major challenges in power handling from windfarm is that it changes continuously. The DC supervisory control should be capable and fast enough to utilize the maximum amount of power being generated by wind farms.
In order to test the capability of DC supervisory control for variable power generation, VSC1 and VSC6 are considered to be connected with a windfarm as shown in figure 3.3. VSC5 is considered to be in constant voltage control mode where as all the rest are in constant active power control mode. Since VSC5 has Pcost= 0 hence it will be the first one to take all the toll of change in power generation from VSC1 and VSC6.
The simulation is run for a total of 35 sec. Following are the events that occur during this time
• At T = 10sec power generation from VSC1 is reduced by 50%
• At T = 15sec power generation from VSC1 is reduced to zero
• At T = 20sec power generation from VSC6 is reduced by 50%
• At T = 25sec power generation from VSC6 is reduced to zero
T1 VSC 1
T2 VSC 2
VSC 5 T4
VSC4
Grid
Grid VSC 6
T7 VSC7
Grid
T3 VSC 3
Grid L46
L12 L24
L35 L23 L47
L57
T6
T5
Wind Farms
VSC 5 in Voltage Control Mode All others are in Contant Power
Control Mode
Figure 3.3: DC grid model for variable wind farm generation
Discussion Results of the simulation are shown in figure 3.4. The negative sign corresponds to the injection of power into the DC system where as the positive sign corresponds to the injection of power into an AC system. At T = 10secs the power being generated by VSC1 is reduced from 160M W to 80M W this difference in power is compensated by generation of more power by VSC5, since it is in voltage control mode and has least active power priority. Similarly when the power generated by VSC1 is reduced to 0M W at T = 15sec and power from VSC6 is reduced from 80M W to 40M W at T = 20sec, VSC5 generates even more power to match the power imbalance. At T = 25sec when the power from VSC6 is completely cut off, the maximum toll of the power difference is yet again
taken by VSC5 but since its limit has been hit (80% of its installed capacity), and all the other stations have same active power priority there is a slight change of power in all of them.
0 5 10 15 20 25 30 35
−200
−100 0
Station 1
0 5 10 15 20 25 30 35
190 200 210
Station 2
0 5 10 15 20 25 30 35
−130
−120
Station 3
0 5 10 15 20 25 30 35
140 160 180
Station 4
PCC Active Power(MW)
0 5 10 15 20 25 30 35
−300
−200
−100 0
Station 5
0 5 10 15 20 25 30 35
−100 0 100
Station 6
0 5 10 15 20 25 30 35
35 40 45
Station 7
Time(s)
Figure 3.4: Results for scenario 1, Variable generation from wind farm)
3.2.2 Station Disconnection
A fault in the VSC station can cause its outage. In case of such an event DC supervisory control should be capable to recalculate the new set points to account imbalance of power. DC supervisory control computes new set points for stations on the bases of their Pcost. Supervisory control will follow the schedule more strictly for stations with higher Pcost. Hence in case of a disturbance which leads to a power imbalance, the station with lower Pcost will take the most toll.
The DC grid model for this station disconnection scenario is shown in figure 3.5. VSC5 is in voltage control mode, hence it has default Pcost= 0 where as all the other stations are in PCC active power control mode. The simulation is run twice. First with all stations having a same Pcost and then, with VSC3 having a Pcost= 0. In the former case all the imbalance of power will be accounted for by VSC5 whereas for the later it will be shared between VSC5 and VSC3. The simulation is run for 35sec and VSC1 is disconnected at T = 10sec.
T1 VSC 1
T2 VSC 2
VSC 5 T4
VSC4
Grid
Grid VSC 6
T7
VSC7
Grid
T3 VSC 3
Grid L46
L12 L24
L35 L23 L47
L57
T6
T5
VSC 1 is Lost
VSC 5 in Voltage Control Mode All others are in Contant Power
Control Mode
Figure 3.5: DC grid model for station disconnection
Discussion The results are shown in figure 3.6. The red graphs show results when all the stations in active power control mode (VSC1 to VSC4 and VSC6, VSC7) have same Pcost. As expected VSC5 generates more power to cater for all the reduction of power caused by the disconnection of VSC1 at T = 10sec.
The blue graph shows the system response for the same disturbance i.e. disconnection of VSC1, however in this case VSC3 has a Pcost = 0. Hence now when VSC1 is lost both VSC3 and VSC5 share the toll of power difference and generate more power.
0 5 10 15 20 25 30 35
−200
−100 0
Station 1
0 5 10 15 20 25 30 35
180 200 220
Station 2
0 5 10 15 20 25 30 35
−200
−100 0
Station 3
Same Pcost Pcost
3=0
0 5 10 15 20 25 30 35
140 160 180
Station 4
PCC Active Power(MW)
0 5 10 15 20 25 30 35
−300
−200
−100 0
Station 5
0 5 10 15 20 25 30 35
−100
−80
−60
Station 6
0 5 10 15 20 25 30 35
35 40 45
Station 7
Time(s)
Figure 3.6: Results for scenario 2, Station disconnection
3.2.3 Islanding
Line faults in a DC grid at times can lead to creation of multiple subsystems. This phenomenon is known as islanding. In the case of islanding DC supervisory control should be able to operate the subsystem under stable conditions. Although the current version of supervisory control cannot identify the individual sub systems but it can still operate them under stable conditions by running OPF for the entire system. For the stable operation of the sub systems it is mandatory for each of
them to have at least one station to be in droop or voltage control mode to act as a slack bus.
The available HVDC grid model is tested with supervisory control for islanding. For this scenario, in addition to VSC5, VSC2 also operates in voltage control mode. Faults causing the disconnection of the lines L34and L24lead to the creation of two subsystems as shown in figure 3.7. Once there are two subsystems, disturbance in one system should not have any effect on the other. The simulation is run for 50sec and following are the events occurring in this time
• At T = 10sec faults in line L34 and L24 leads to islanding
• At T = 20sec generation from VSC1 is reduced by 50%
• At T = 25sec generation from VSC1 is reduced to zero
• At T = 30sec generation from VSC6 is reduced by 50%
• At T = 35sec generation from VSC6 is reduced to zero
T1 VSC 1
T2 VSC 2
VSC 5 T4
VSC4
Grid
Grid VSC 6
T7
VSC7
Grid
T3 VSC 3
Grid L46
L12 L24
L45 L23 L47
L57
T6
T5
System A VSC 2 and 5 in Voltage Control Mode
All others are in Contant Power Control Mode
L34 System B
Figure 3.7: DC grid model for islanding
Discussion The results of the scenario are shown in figure 3.8. At T = 10sec islanding takes place and two subsystems are formed. Lets call them subsystem A and B. Subsystem A consists of VSC1 to
VSC3, whereas subsystem B comprises of VSC4 to VSC7. Before islanding occurs VSC1 and VSC3 are producing 280M W of power and part of this power has been consumed by stations in subsystem B. Hence once the systems are isolated due to faults, VSC2 starts to consume the excessive amount of power being generated by VSC1 and VSC3. Whereas in subsystem B the imbalance of power is accounted for by VSC5 which now generates more power.
After the islanding, changes of power in one system should not have any effect on the other.
Hence when the power being generated by VSC1 is reduced to 80M W and 0M W at T = 20sec and T = 25sec respectively, VSC2 starts to consume less power and there is no effect on subsystem B. Similarly in subsystem B, when power being generated by VSC6 is reduced at T = 30sec and T = 35sec there is no effect on system A and VSC5 generates more power to maintain power balance.
0 5 10 15 20 25 30 35 40 45 50
−200
−100 0
Station 1
0 5 10 15 20 25 30 35 40 45 50
100 200 300
Station 2
0 5 10 15 20 25 30 35 40 45 50
−140
−120
−100
Station 3
0 5 10 15 20 25 30 35 40 45 50
140 160 180
Station 4
Time(s)
PCC Active Power(MW)
0 5 10 15 20 25 30 35 40 45 50
−200
−100 0
Station 5
0 5 10 15 20 25 30 35 40 45 50
−100
−50 0 50
Station 6
0 5 10 15 20 25 30 35 40 45 50
20 40 60
Station 7
Time(s)
Figure 3.8: Results for scenario 3, Islanding
3.2.4 Line Disconnection
Fault in DC lines leading to disconnection at times can cause over current in other lines in the system. DC supervisory control ensures the operation of the grid within operational limits of the lines. The DC grid model is tested with the supervisory control for such an event. To pronounce the effect of line disconnection on the system, current though line L23 is limited to 150A. A fault in line L35 leads to its disconnection as shown in figure 3.9. Since the amount of power line L23 can
Figure 3.9: DC grid model for line disconnection
transmit is constrained, the power profile of the grid is going to change.
Discussion The simulation results are shown in figure 3.10. At T = 10sec line L35is disconnected.
Since the maximum power that can be transmitted by L23 which connects VSC3 with rest of the system is restricted, the generation from VSC3 is reduced. This power imbalance is compensated by VSC5 which generates more power.
0 5 10 15 20 25 30 35 40 45 50
−160
−140
Station 1
0 5 10 15 20 25 30 35 40 45 50
180 200 220
Station 2
0 5 10 15 20 25 30 35 40 45 50
−120
−110
−100
−90
Station 3
0 5 10 15 20 25 30 35 40 45 50
140 160 180
Station 4
PCC Active Power(MW)
0 5 10 15 20 25 30 35 40 45 50
−100
−50
Station 5
0 5 10 15 20 25 30 35 40 45 50
−100
−80
−60
Station 6
0 5 10 15 20 25 30 35 40 45 50
30 40 50
Station 7
Time(s)
Figure 3.10: Results for scenario 4, Line disconnection
Chapter 4
Conclusions
The presented and implemented version of DC supervisory control has shown satisfactory results during its testing and validation within the co-simulation platform. The result show that the DC supervisory control can improve the power extraction from wind farms by updating the set-points following any change in the system. Considering the variety of test cases, the original OPF calculator could have been modified to deal with transient conditions in more robust way. Although in the case of islanding, the current algorithm is not designed to identify separate islands and reassign required control modes such as new slack bus, but the station set-points necessary for the stable operation of individual islands can be calculated.
References
[1] European Commission, The 2020 climate and energy package,
ec.europa.eu.[Online].Available:http://ec.europa.eu/clima/policies/package/indexen.htm [Ac- cessed: May 22, 2014]
[2] N. Flourentzou, V. G. Agelidis, and G. D. Demetriades, VSC-based HVDC power transmission systems: An overview, IEEE Trans. Power Electron., vol. 24, no. 3, pp. 592602, Mar. 2009.
[3] VSC transmission Cigre Working Group B4.37-269, Tech. Rep., 2005
[4] G. Asplund, K. Eriksson, and H. Jiang, DC transmission based on voltage source converters in Cigre Colloquium on HVDC and FACTS, South Africa, 1997
[5] S. Johansson, G. Asplund, E. Jansson, and R. Rudervall, Power system stability benefits with VSC DC-transmission systems in Cigre Conference B4-204, Paris, France, 2004
[6] L. Zhang, L. Harnefors, and P. Rey, Power system reliability and transfer capability improvement by VSC-HVDC (HVDC light) in Cigre Regional Meeting, Tallin, Estonia, 2007.
[7] T.M. Haileselassie Control, Dynamics and Operation of Multi-terminal VSC-HVDC Transmis- sion Systems PhD thesis, December 2012, NTNU.
[8] M. El-Marsafawy and R. Mathur, A new, fast technique for load-flow solution of integrated multi-terminal AC/DC systems, IEEE Trans. Power App. Syst., vol. PAS-99, no. 1, pp. 246255, Jan./Feb. 1980
[9] M. E. El-Hawary and S. T. Ibrahim, A new approach to AC-DC load flow analysis, Elect. Power Syst. Res., vol. 33, no. 3, pp. 193200, 1995.
[10] Jef Beerten, Stijn Cole, and Ronnie Belmans Generalized Steady-State VSC MTDC Model for Sequential AC/DC Power Flow Algorithms IEEE transaction on power systems, vol. 27, No. 2, May 2012
[11] Mohamadreza Baradar, and Mehrdad Ghandhari A Multi-Option Unified Power Flow Approach for Hybrid AC/DC Grids Incorporating Multi-Terminal VSC-HVDC IEEE transaction on power systems, vol. 28, No. 3, August 2013
[12] Alejandro Pizano-Martinez, Claudio R. Fuerte-Esquivel, H. Ambriz-Perez, and Enrique Acha, Modeling of VSC-Based HVDC Systems for a Newton-Raphson OPF Algorithm IEEE transaction on power systems, vol. 22, No. 4, November 2007
[13] Leandro Bertoni, Mkcio Ferraz Arahjo, Wagner Alexandre da Silva and Luiz Felipe Araujo Feio Nobre MACH2 - Modular Advanced Control 2nd Edition 2004 IEEUPES Transmission &
Distribution Conference & Exposition: Latin America.
[14] Jos Arrillaga, Y.H. Liu, Neville R. Watson Flexible Power Transmission The HVDC Option 2007 John Wiley and Sons, Ltd.
[15] Cuiqing Du, VSC-HVDC for Industrial Power Systems Phd Thesis, Department of Energy and Environment, Chalmers University of Technology, Gteborg, Sweden, 2007
[16] CIGRE SC-B4 WG 37 (2004), VSC transmission, Final document, Paris.
[17] T.M. Haileselassie Control of Multi-terminal VSC-HVDC Systems, master thesis, 2008. NTNU [18] Jan Machowski, Janusz W. Bialek, Jams R. Bumby, Power System Dynamics and Stability;
John Wiley & Sons, West Sussex pp.327, 1997
[19] ˙Ake Carlson Specific requirements on HVDC converter transformers ABB Transformers AB, Ludvika Sweden
[20] N. Mohan. Electric Drives: An Integrative Approach. Chapter 9, 2003
[21] Lars Nordstrm, Davood Babazadeh Ch. Cyber physical Approach to HVDC grid control, book Cyber Physical Systems Approach to Smart Electric Power Grid submitted to Spinger 2014 [22] Lianxiang Tang, and Boon-Teck Ooi, Locating and Isolating DC Faults in Multi-terminal DC
Systems, IEEE Transaction Delivery, Vol. 22, No 3, July, 2007.
[23] G.P. Ccorimanya Operation of HVDC Grids in Parallel with AC Grids Master thesis, Chalmers university of technology, 2010.
[24] L. Tang and B. Ooi, Locating and isolating dc faults in multi-terminal dc systems, IEEE Trans- actions on power delivery, vol. 22, no. 3, pp. 18771884, 2007.
[25] W. F. McNichol, J. Reeve, J. R. McNichol, R. E. Harrison, J. P. Bowles, Considerations for Implementing Multiterminal DC Systems IEEE Transaction on Power Apparatus and Systems, Vol PAS-104, September 1985.
[26] S. Rodrigues, R. Teixeira Pinto, P. Bauer, E. Wiggelinkhuizen, J. Pierik, Optimal Power Flow of VSC-based Multi-Terminal DC Networks using Genetic Algorithm Optimization 978-1-4673- 0803-8/12, IEEE, 2012
[27] Mohamadreza Baradar, Mohammad Reza Hesamzadeh and Mehrdad Ghandhari, Second-Order Cone Programming for Optimal Power Flow in VSC-Type AC-DC Grids IEEE transaction on power systems, vol. 28, No. 4, November 2013
[28] J. A. Momoh, M. E. El-Hawari, and R. Adapa, A review of selected optimal power flow literature to 1993 part II: Newton, linear program- ming and interior point methods IEEE Trans. Power Syst., vol. 14, no. 1, pp. 105111, Feb. 1999.
[29] Wang Feng, Student Member, IEEE, Le Anh Tuan, Member, IEEE, Lina Bertling Tjernberg, Anders Mannikoff and Anders Bergman, A New Approach for Benefit Evaluation of Multiterminal VSCHVDC Using A Proposed Mixed AC/DC Optimal Power Flow IEEE transaction on power systems, vol. 29, No. 1, February 2014
[30] http://www.tenasys.com/tenasys-products/intime-rtos-family/overview-rtos [Accessed: April 20, 2014]
[31] http : //www.tenasys.com/index.php [Accessed: August 10, 2014]
[32] Kevin Mets, Juan Aparicio Ojea, and Chris Develder Combining Power and Communication Network Simulation for Cost-Effective Smart Grid Analysis, IEEE communications and surveys
& tutorials.
[33] http : //www.tenasys.com/overview − if w [Retrieved: August 1, 2014]
[34] D. Glover and M. Sarma Power System Analysis and Design. PWS Publishing Company, 1994.
[35] Conrad Sanderson. Armadillo c++ linear algebra library, Retrieved on 2013-02-03.
http://arma.sourceforge.net/.
[36] Introduction to Ipopt: A tutorial for downloading, installing, and using Ipopt, Revision: 2474, April 8, 2014
[37] A.Wchter. An Interior Point Algorithm for Large-Scale Nonlinear Optimization with Ap- plications in Process Engineering. PhD thesis, Carnegie Mellon University, Pittsburgh, PA, USA, January 2002. available at http://researcher.watson.ibm.com/researcher/files/us- andreasw/thesis.pdf
[38] E.Velleux and B.T.Ooi Multi-Terminal HVDC Grid with Power Flow Controllability B4-301, Cigre 2012.
[39] http : //mumps.enseeiht.f r/ [Accessed: August 10, 2014]
[40] http : //www.netlib.org/blas/ [Accessed: August 10, 2014]
[41] http : //www.netlib.org/lapack/ [Accessed: August 10, 2014]
[42] http : //www.pardiso − project.org/ [Accessed: August 10, 2014]
