A DYNAMIC POWER FLOW CONTROLLER FOR POWER SYSTEM STABILITY IMPROVEMENT AND LOSS REDUCTION

Transcription

1 A DYNAMIC POWER FLOW CONTROLLER FOR POWER SYSTEM STABILITY IMPROVEMENT AND LOSS REDUCTION Nicklas Johansson 1 Lennart Ängquist Hans-Peter Nee Bertil Berggren Royal Institute of Technology KTH School of Electrical Engineering / EME 1 44, Stockholm, Sweden ABB AB, Corporate Research Västerås, Sweden Abstract In this paper, a novel FACTS device denoted Dynamic Power Flow Controller (DPFC) is described. The device consists of a Phase-Shifting Transformer (PST) connected in series with a set of thyristor-switched capacitive and/or inductive elements. When compared to a normal PST, this device has faster dynamic properties and in addition to the normal PST functions, it also allows for power oscillation damping, transient stability improvement, and voltage stability improvement in a power grid. The DPFC is believed to be less costly than other FACTS devices with the same functionalities. In this paper, the benefits and functions of the DPFC are discussed. Additionally, an adaptive controller for DPFC power oscillation damping, transient stability improvement, and power flow control is presented and verified by means of digital simulations. Keywords: Adaptive control, Dynamic Power Flow Controller, FACTS, Phase-Shifting Transformer, Power Oscillation Damping, Transient Stability, Power Flow Control 1 INTRODUCTION The deregulation of the electricity markets has led to a changing load flow pattern in the power systems of many countries. As a consequence of this, transmission networks are operated closer to their thermal and dynamical stability limits. Since the construction of new transmission lines to relieve the overloaded lines is both very expensive and often politically unacceptable, the market for alternative technologies is growing. Traditionally, power flow control has been performed by the use of Phase Shifting Transformers (PST). These are devices which can be used to control the active power flow, thereby reducing the power flow on overloaded lines while redirecting the power to flow on lines with free capacity. The disadvantages of PSTs are their reactive power consumption and their limited speed of control. Several Flexible AC Transmission System (FACTS) devices [1] with a fast dynamic range for power flow control and power system stability improvement have been discussed lately. One example is the Unified Power Flow Controller (UPFC) which is able to control reactive and active power on a line independently. In this paper, an approach which is believed to be less costly than the UPFC is described. This device, which was presented in [2] is denoted Dynamic Power Flow Controller (DPFC). It consists of a traditional PST combined with a number of thyristorswitched capacitive and/or inductive reactances connected in series with the PST. This configuration expands the dynamic control range of the PST, enabling fast power flow control, power oscillation damping, transient stability improvement and voltage stability improvement. Low-frequency electromechanical oscillations between groups of generators are inherent in power systems, and adequate damping of these is a prerequisite for secure operation of the power system. Even if the traditional method of installation of Power System Stabilizers (PSS) acting on the excitation systems of the generators is effective in most systems, adding supplementary damping by installation of FACTS devices like the DPFC may improve the system stability and transfer capacity of many systems. The design of an effective damping controller for the DPFC is complicated by the fact that the equations governing the oscillations in a power system are non-linear and that the power system parameters often change dramatically during the contingencies causing the power oscillations. Additionally, the tap setting of the PST affects the controllability of the critical power oscillation mode as well as the mode observability in the locally available signals. The thyristor-switched reactances in the DPFC configuration are by themselves a device belonging to the group of Controlled Series Compensators (CSC). Many approaches to power oscillation damping by means of CSC have been proposed [3], including traditional loop-shaping [4], non-linear and adaptive control [5]. This paper contains a presentation of the DPFC and its functions. In addition to this, an adaptive controller for power oscillation damping, transient stability improvement, and power flow control by means of the DPFC is presented and verified in digital simulations. The damping controller is based on a simple generic grid model and it has been designed to provide an alternative to controllers based on full scale system models in order to reduce the implementation effort. The generic system model proposed 1 Contact author. nicklas.johansson@ee.kth.se, tel: +46 () , fax: +46 () th PSCC, Glasgow, Scotland, July 14-18, 28 Page 1

2 here requires very few system parameters to be known initially and the controller utilizes only locally measured signals as inputs. The damping controller is limited to be used for power oscillation damping in systems which are subject to electro-mechanical inter-area oscillations with one dominating oscillation mode. 2 THE DYNAMIC POWER FLOW CONTROLLER (2) The DPFC allows for control of the reactive power flow on the line. This feature can be used for improvement of system voltage stability/recovery. For example, it can be used to give fast voltage support in an overload situation by engaging full capacitive compensation of the TSSC. Additionally, in some power grids, the reactive power consumption of a regular PST is unacceptable. By using a DPFC instead, the TSSC can be used to compensate for the tap-dependent series inductance of the PST, keeping the total series reactance of the device close to zero. By utilizing the full benefits of the DPFC, the transfer capacity of the power system may be increased and the transmission losses can be lowered while maintaining the system security margin. Figure 1: Schematic describing the DPFC topology. The topology of the DPFC is shown in Figure 1. Here, the switched series reactances are represented by two capacitive and two inductive elements. The number of Thyristor-Switched Series Capacitors (TSSC) and Thyristor-Switched Series Reactors (TSSR) may vary depending on the application. The Mechanically Switched shunt Capacitor (MSC) is optional, depending on the system reactive power requirements. In this configuration, the PST is used for the long-term power flow control while the TSSC/TSSR are used for fast power flow control, power system stability improvement, and for optimizing the reactive power demand of the DPFC. 4 CONTROL OF THE DPFC The control objectives of the DPFC are different depending on the characteristics of the power system where it is installed. For example: In a meshed grid, the device may be installed to control power flows and to provide voltage support. In a weakly interconnected grid, the device may be used to control power flows, damp power oscillations, and to improve the system transient stability. This paper mainly focuses on the application of DPFC in systems of the second type. A schematic picture of the proposed control system for the DPFC master controller is shown in Figure 2. 3 FUNCTIONAL ASPECTS OF THE DPFC The main advantages of the DPFC in relation to a common PST are: (1) The speed of the power flow control of the DPFC is much higher than that of the PST. This means that the DPFC can be used for power oscillation damping and transient stability improvement in the power system. Additionally, due to its limited speed of control, a PST is usually operated with preventive control, that is, with a preset tap changer value which does not change during or immediately after a fault [6]. This tap setting needs to satisfy the stability criteria of all (N-1) contingency situations and it is usually not the optimal setting if the system transmission losses are considered. The DPFC may in contrast to this be operated with corrective control, where the tap setting which gives the lowest power system losses at all times is used. In case of a contingency, the power flows in the system can be rapidly altered using the TSSR and TSSC to comply with the (N-1) requirements. Figure 2: Schematic of the DPFC master controller. The controller includes a Recursive Least Squares (RLS) algorithm for extraction of the oscillation mode frequency, phase, and amplitude from the locally measured line power. It also includes parts for Power Oscillation Damping (POD), First-Swing (transient) stability improvement (FSW), and active power flow control. Generally, a severe fault in the power system initially leads to a risk of transient instability. This leads to a triggering of the first-swing controller which is based on an open loop control scheme changing the line reactance in two steps during the first swing of the generator rotor angles. Subsequently, when the first-swing controller has performed its sequence, there is commonly a power oscillation in the system which is detected by the RLS algorithm. The RLS module then initiates the damping 16th PSCC, Glasgow, Scotland, July 14-18, 28 Page 2

3 controller which has a built-in power flow control feature for fast control of the power on the line after a fault towards a reference value (P xsp ) provided by the transient controller. Since this feature is only active when power oscillations are present, a separate slow PIcontroller is necessary for long-term power flow control. These controllers contribute with the terms POD, FSW and PI to give the desired reactance value for the set of switched reactances - TSSC/TSSR. All of the controller parts use the DPFC line active power ( ) as input signal. The transient controller also utilizes the DPFC line current (I line ) as an input. It is assumed that the Transmission System Operator (TSO) is in charge of the long-term control of the device, where remote system information is considered for optimization of system losses and security margins to yield suitable values for the power flow reference value (P ref ) and the tap setting of the PST. An aim for the TSO would be to control the setting of the PST such that it provides most of the longterm power flow control, saving the TSSC/TSSR capacity for emergency operation during contingencies. 4.1 Oscillation damping and fast power flow control In this paper, an adaptive model predictive control approach for power oscillation damping by means of the DPFC is described and verified using digital simulations. The controller uses a time-discrete approach based on a simple generic model of the power grid (see Figure 3). This model is based on the assumption that there is one dominant inter-area mode of oscillation in the system. When applied to real power systems, the model may represent two different grid areas and their interconnecting power lines where it is assumed that a DPFC is installed on one of the inter-tie lines. The load in each of these areas is modeled as a constant voltageindependent load. The model requires little prior knowledge of the power system parameters; the model parameters are adaptively set during the controller operation. The model utilizes three parameters which are continuously estimated by measuring the step responses in the DPFC line power when the TSSC/TSSR reactance is changed. The parameters are: a series reactance i, a parallel reactance eq, and the angular frequency of the critical power oscillation mode (). Additionally, another four parameters are assumed to be known from grid and device data: The voltage phase angle shift of the PST ( ), the TSSC/TSSR reactance value ( TSSC/TSSR ), the (tap-dependent) series reactance of the PST ( PST ), and the (constant) value of the DPFC line reactance ( line ) not including the DPFC itself. A similar model for control of CSC is described in [7]. Figure 3: Generic system model used to design damping- and fast power flow controllers The controller is based on a dead-beat approach designed to ideally stabilize the oscillation in two discrete reactance steps which are calculated based on model prediction. This objective is rarely met in reality since the estimated system parameters have errors and there are model errors between the reduced model and the actual system. Also, the control signal, in this case the variable series reactance, is limited. An implementation of a damping controller for CSC based on the same principles is described and verified in [8], [9] and [1]. The derivation of the control laws used in this paper is similar to the derivation of the control laws in [9] and [1] with the difference that a voltage angle shift is introduced in the system model in order to represent the quadrature voltage of the PST in the DPFC case. The developed controller is time-discrete in nature with a discretization determined by the inter-area mode oscillation frequency. According to this discretization, the controller is only allowed to act at the instants coinciding with high or low peaks in the power oscillation. An RLS algorithm is used in order to determine these instants along with the power oscillation amplitude and the mean value of power flow through the DPFC line during the oscillation. The algorithm utilizes an expected value of the oscillation angular frequency () when no oscillations are at hand. When the algorithm detects an oscillation, is adapted to its actual value using a PI-regulator. The damping controller is designed to include a feature for fast power flow control which is only active when power oscillations are present. This is done by combining the objectives of oscillation damping and power flow control in the model prediction performed at each time-step. This feature is used in order to allow the device to change the line power very fast after a contingency to improve the system transient stability. 4.2 Transient stability improvement A transient stability improvement feature which is based on an open-loop approach is included in the master controller. This is done in order to improve the system stability during the first swing of the synchronous motor angles following a severe contingency. To improve transient stability by means of CSC, different approaches have been discussed. Many of these are based on optimal control laws described as bang-bang strategies [11]. The approach to transient stability improvement used in this work is based on a near-optimal bang-bang control strategy described in [8]. When applied to the DPFC, this strategy yields a control law for the TSSC/TSSR during the first swing of the generator angles after a fault described in mathematical terms as: TSSC = min ; ( δs δr ) > TSSC = min ; ( δs δr ) <, ( δs δr ) > 9 (1) = c ; ( δ δ ) <, ( δ δ ) < 9, c [,1] TSSC min s r s r TSSR = Here, min is the minimum reactance value attainable by the set of switched capacitive units (that is the maxi- 16th PSCC, Glasgow, Scotland, July 14-18, 28 Page 3

4 mum capacitive compensation) and s and r are the voltage phase angles of the sending and the receiving grid areas in a Center Of Inertia (COI) reference frame. The constant in equation (1) is set to c=.5 in this work. Since the controller discussed here uses only locally measured signals, the instants when δ δ = and δ s δr = 9 are found from the RLS phase estimation of the power oscillation and the behavior of I line (t) during the first angular swing after the fault. The transient stability scheme is limited to switching the TSSC only once at the fault instant and once when the voltage phase angle difference is declining after the fault. After this scheme has completed, the transient controller modifies the set-point (P xsp ) of the fast power flow control feature in the damping controller to equal a value close to the short-term overload limit of the line. This is done in order to improve the transient stability during the damping of subsequent oscillations and to increase the damping performance by moving the system to a state with smaller voltage phase angle differences between areas where the errors in the linearity assumptions of the damping controller are smaller. The elevated power flow on the DPFC line is then lowered when the TSO redispatches the power system after the fault to comply with the new N-1 requirements. The transient controller may be triggered either using locally measured signals or using remote signals. A triggering based on a stability assessment utilizing remote measurements of voltage phasors and generator speeds at suitable locations in both areas like in [11] is attractive. However, if such signals are not available, it may be considered to trigger the scheme if the DPFC line power magnitude rises unexpectedly with a high time derivative, leading to a rapid and significant increase in the line power. This indicates that a major disturbance has occurred and that there is a risk for transient instability. Such a triggering procedure would possibly be excessively sensitive when compared to a more stringent approach using remote signals but it may still be useful due to its relative ease of implementation. 5 SIMULATION RESULTS s r 5.1 An example of DPFC operation An example of DPFC operation in an interconnected power system with one critical inter-area oscillation mode is provided below. To illustrate this case, timedomain simulations are performed. In order to simulate measurement noise, pseudo-random noise of 1% (standard deviation relative to the input signal value) was added to the controller input in all simulations. The power system is a four-machine system inspired by [12] described in [8] which is shown in Figure 4. A DPFC with three switched capacitive elements and no MSC is placed in the middle of one of the interconnecting lines N8-N9. The TSSC reactances are selected on a binary basis resulting in eight allowed reactance values which are equidistant between zero and the maximum capacitive compensation min (which is 2/3 of the inductive reactance of the DPFC line). The PST is a Quadrature Booster which enables voltage phase angle shifts in the region [-15, 15 ]. The generators are all equipped with PSS which were chosen to have a low gain to give a system with poor damping of the inter-area mode. In this paper, this system is studied in two loading cases: high power transfer and low power transfer. The loads used for the loading cases (high/low) are P L7 =967/1367 MW, Q L7 =1/2 MVAr and P L9 =1967/1367 MW, Q L9 =1/2 MVAr and the active power generation of generators 1 and 2 is 16 MW in total. All loads in the system are voltage dependent with constant current characteristics for the active power and constant impedance characteristics for the reactive power load. In this first example, the high power transfer case is used and the total active power transmitted on the interconnecting lines between N7 and N9 is 6 MW in steady state. Figure 4: Four-machine system used for digital simulations. To demonstrate the benefits of the DPFC in this simple system, it is assumed that the DPFC line has a high line resistance equal to twice the value for the parallel lines. This leads to a lower short-term thermal overload rating of this line compared to that of the parallel lines. This limit is assumed to be 25 MW while the thermal limit of the parallel lines is assumed to be 35 MW. In order to minimize the losses in the system, the PST is used to lower the transmitted power on the high resistive line during normal operation. Thus, the PST tap is set at -15 normally. Now, a three-phase short circuit at node 8 is applied to the system at t=1. s. The fault is cleared by disconnection of one line N8-N9 in parallel to the DPFC line at t=1.2 s (which is a worst case). The fault results in electro-mechanical oscillations and an overload of the remaining line connected in parallel to the DPFC if no action is taken. The proposed controller now operates with three objectives: to improve the transient stability, to damp the power oscillations, and to relieve the overloaded line by increasing the power flow on the DPFC line. The post-contingency set-point for the fast power flow controller controlling the DPFC line power immediately after the fault is set to 25 MW. No slow power flow controller is used here. The DPFC line power and the TSSC reactance are shown in Figure 5. It can be seen that the controller operation improves the system damping significantly. Between t=1.2 s and t=1.7 s, the first-swing stability improvement feature is active, maximizing the TSSC compensation during the first swing and reverting to half compensation at t=1.7 s according to the transient control scheme (1). Then, at 16th PSCC, Glasgow, Scotland, July 14-18, 28 Page 4

5 t=2.3 s, the damping controller is activated to damp the power oscillation and the fast power flow controller is operating towards the predefined set-point of 25 MW. The damping is completed at t=5.3 s and the power flow on the DPFC line is, at this point, close to the set-point. Active power (MW) Fault 2 with PST in buck mode -15 deg - controller engaged 1 - TSSC - controller engaged -.2 controller disengaged Figure 5: DPFC line active power ( ) including pseudorandom noise and connected series reactance ( TSSC ) for the system contingency with and without the TSSC controller engaged. The TSSC series reactance is limited to [-.1, ] p.u. The grid parameters i and eq in the system model have a large impact on the gain of the controller. Since an adaptive control approach is used, the parameters of the grid where the DPFC device is placed are estimated continuously by the controller according to the model of Figure 3. The controller is time-discrete in nature making it possible for the estimation routines to be developed based on the open loop step response of the reduced system in Figure 3 to changes in the TSSC/TSSR reactance in the same way as in [8]. This procedure simplifies the estimation since closed loop system identification which may introduce difficulties in determining unique estimates [13] is avoided. Since the estimation routines are dependent on step response data, the parameters cannot be estimated prior to the first reactance step in a damping sequence. Therefore, a starting guess for the parameters is necessary to determine the initial action of the controller. This starting guess is chosen as the set of parameters which corresponds to the grid (N-1) configuration case where the product of the controllability and the observability of the inter-area oscillation mode at the DPFC location is the largest. This approach leads to a controller which generally is sub-optimal initially in order for it to maintain stability in all system configurations. Once the first step in a damping sequence has been executed, step response data is collected and the actual parameters are estimated. The input to the estimation routine is both the instantaneous and the average change in line power resulting from the change in reactance. Since the estimation of the average power on the line requires time to stabilize, the parameters are not updated until slightly before the next step is taken by the controller. This procedure is repeated after each new step in the TSSC reactance. This approach Series reactance (p.u.) leads to a controller where the parameters i and eq are updated at every time step of the damping controller. The parameter is in contrast continuously updated by the PI-controller connected to the RLS estimator while the power oscillation amplitude exceeds a certain minimal value (5 MW in this simulation). The evolution of the system parameters is shown in Figure 6. To improve the controller robustness, a forgetting factor is used in the time-discrete parameter estimation, forming a weighted average of old and new parameter estimates. It can be seen from Figure 6 that the parameter eq starts from a high value and i starts from a low value corresponding to the starting guess. As the parameter estimation progresses, the damping performance of the controller is improved. If the system parameters are accurately set, the controller will ideally damp any power oscillation in two time-steps (assuming no control signal limits and no errors between the model and the actual system). Reactance eq (pu) Oscillation frequency (rad/s) Fault 2 with PST in buck mode -15 deg i eq Figure 6: Estimated parameters i, eq and. ω Controller verification In order to test the performance of the damping controller in different operating points, the four machine system in Figure 4 was studied in four contingencies at two different inter-area power transfer levels, 18 MW (low), and 6 MW (high). In this case, the line resistance per unit length was chosen equal for all transmission lines in the system and the thermal short-term overload limit was assumed to be in the range of 3-35 MW for the inter-area tie lines in the system. The postfault contingency set-point for the fast power flow controller was set to 3 MW. The studied cases were: 1. A 3-phase short circuit (SC) at node 8 cleared with no line disconnection after 2 ms. 2. A 3-phase SC at node 8 cleared by disconnection of one N8-N9 line in parallel to the DPFC after 2 ms. 3. A 3-phase SC at node 8 cleared by disconnecting one of the N7-N8 lines after 2 ms. 4. A disconnection of load at node 7. P=-25 MW Q=-5 MVAr in the high load case and P=-25 MW Q=-1 MVAr in the low load case. The results are summarized in Figure 7 and Table 1. It can be seen from Figure 7 that the damping of the system with no TSSC control is rather poor for all fault Reactance i (pu) 16th PSCC, Glasgow, Scotland, July 14-18, 28 Page 5

6 cases, especially in the high transfer case. With TSSC damping enabled, the damping of the inter-area mode is significantly improved. Every fault has been simulated with different tap settings of the PST. Boosting tap levels, characterized by their advance in voltage phase angle are denoted with positive (+) angles and buck mode phase shifts are conversely marked as negative (-) angles. A couple of things are worth noting in Figure 7: Damping exponent Damping exponent -.5 Damping of critical mode in different fault cases, 6 MW transfer Cntrl., -15 deg. Cntrl., deg Fault case # Damping of critical mode in different fault cases, 18 MW transfer -.5 Cntrl., +15 deg. No Cntrl., -15//+15 deg. Cntrl., +15 deg. No Cntrl., /+15 deg. Cntrl., deg Fault case # Figure 7: Damping exponents () for different faults and settings of the PST. No Cntrl.-values are the real part of the linearized system eigenvalues corresponding to the critical mode with the DPFC controller disabled. Cntrl.-values are calculated from time-domain simulations with the DPFC controller enabled by fitting a curve y(t)=ce t to the envelope of the total inter-area power oscillation. (1) There is a relation between the tap setting of the PST and the damping performance. In operating conditions characterized by large boosting (+) phase angles, the damping performance is generally degraded compared to buck (-) and zero () phase angle cases. One reason for this is that the observability of the oscillation mode in the DPFC line power signal decreases with increasing boosting phase shift leading to a reduced Signal-To-Noise (SNR) ratio of the controller input signal. Another reason is that the controllability of the mode from the TSSC increases with boost angle, amplifying the effect of the control errors introduced by the discretization of the inserted series reactances. A third reason for the degradation is that the assumptions of linearity used in the derivation of the control laws are less accurate at high boosting angles. The degradation issue may affect how the DPFC should be operated regardless of which type of (linear) controller that is used for power oscillation damping. If a certain power oscillation damping performance is required, certain tap/power transfer configurations may not be allowed due to this issue. It should be pointed out that a controller utilizing remote signals for oscillation damping would not suffer from this degradation problem. (2) In Fault 4 of the high power transfer case at =- 15, the DPFC does not improve the damping of the system. This is due to that only discrete values of series reactances are available for power oscillation damping. In this fault case, the power oscillation is initially fairly small, and the controller demands changes in reactance which are smaller than the lowest available value given by the discretization. As a result, the controller will not insert any series reactance in the circuit and the damping will not be improved from its initial value. Detailed information of each fault is given in Table 1. In this table, the power on the DPFC line before the fault is given as P (MW), the resulting maximum total transfer power oscillation amplitude (peak-to-peak) between the areas measured after the fault is denoted P osc (MW) and the time until the DPFC line power oscillation is damped to below 1 MW in amplitude (p-p) is given as T PST (s) with the TSSC controller disabled and T DPFC (s) with the controller enabled. The frequency of the inter-area oscillation mode is given as f (Hz). It can be seen that T DPFC is in the range of s in all cases except for the one discussed under (2) above. With no supplementary damping added, the damping time (T PST ) is in the range of 8-13 s in the high power transfer case and in the range of 9-26 s in the low power transfer case. Fault 1 Fault 2 Fault 3 Fault 4 6 MW MW +15 P osc=393 f =.66 T PST =13 T DPFC=8.4 P =175 P osc =396 f =.65 T PST =31 T DPFC=6.6 P =79 P osc =393 f =.65 T PST =2 T DPFC=1.2 P =275 P osc =261 f =.73 T PST =21 T DPFC=4.8 P =6 P osc =258 f =.72 T PST =18 T DPFC=5.5 P =176 P osc =253 f =.54 T PST =47 T DPFC=4.9 P =175 P osc =236 f =.52 T PST =41 T DPFC=5.4 P =79 P osc =24 f =.52 T PST =32 T DPFC=6.4 P =275 P osc =222 f =.68 T PST =26 T DPFC=5.2 P =6 P osc =225 f =.67 T PST =26 T DPFC=3.5 P =176 P osc =236 f =.55 T PST =32 T DPFC=6.2 P =175 P osc =216 f =.54 T PST =32 T DPFC=6.3 P =79 P osc =215 f =.53 T PST =18 T DPFC=7.7 P =275 P osc =236 f =.68 T PST =21 T DPFC=5.2 P =6 P osc =23 f =.68 T PST =2 T DPFC=5.1 P =176 P osc =87 f =.66 T PST =27 T DPFC=2.3 P =175 P osc =84 f =.65 T PST =31 T DPFC=31 P =79 P osc =84 f =.65 T PST =8. T DPFC=2.8 P =275 P osc =15 f =.74 T PST =1 T DPFC=2.7 P =6 P osc =99 f =.73 T PST =9. T DPFC=2.2 P =176 Table 1: Maximum power oscillation amplitude (P osc ), oscillation frequency (f), initial DPFC line power (P ) and damping times with the controller enabled (T DPFC ) and disabled (T PST ) for inter-area transfer of 6 MW (high) and 18 MW (low). Figure 8 further illustrates the performance degradation of the damping controller at high boosting angles. Here, fault 3 is studied with different tap settings of the PST. In the upper part of the figure, the power on the DPFC line after the fault is given when a 15 boosting tap is used with the TSSC controller enabled and disabled and in the lower part of the figure, the simulation results for the case with the tap set at 15 in buck mode are given. The fault initiates oscillations in the rotor 16th PSCC, Glasgow, Scotland, July 14-18, 28 Page 6

7 angles of the generators with close to the same amplitude regardless of the PST setting. Thus, it can be clearly seen that the inter-area mode has a larger observability in the DPFC line power signal in the case with a -15 phase shift. It can also be noted that the reactance step magnitudes required to stabilize the oscillation in the boost mode case are smaller than those required in buck mode, indicating a larger controllability in boost mode. As expected, the damping controller performs sub-optimally in the PST boost case while it functions flawlessly in the PST buck case. Active power (MW) Active power (MW) Fault 3 with PST in boost mode +15 deg., Cntrl, No Cntrl Fault 3 with PST in buck mode -15 deg. 3 TSSC, Cntrl 2, No Cntrl 1 TSSC Series reactance (p.u.) Figure 8: Comparison of the controller performance for fault 3 with different tap settings of the PST. 6 CONCLUSIONS In this paper, the topology and functions of the Dynamic Power Flow Controller (DPFC) have been presented. The device is seen as an alternative to traditional PSTs providing a faster speed of control and an extended functionality. An adaptive DPFC power oscillation damping controller with transient stability and power flow control features based on a simple generic system model has been proposed and verified by means of digital simulations. Despite the simple structure of the system model and the TSSC discretization, the DPFC damping controller performs well in several studied contingencies for different operating points in a model of a four-generator system of a type commonly used to study inter-area oscillations. A performance degradation of the damping controller at high boost phase shifts was noted. It is believed to be due to changes in the oscillation mode controllability and observability from the DPFC and increased errors in the linearizations used in the derivation of the control laws when the boosting phase shift is increased. This issue is believed to affect also other control approaches for power oscillation damping by means of a DPFC relying on locally measured signals as controller inputs. Future work on the DPFC controller includes studies on the applicability of the controller in larger power systems and voltage stability improvement strategies for the DPFC. Series reactance (p.u.) 7 REFERENCES [1] N. G. Hindigorani and L. Gyugyi, Understanding FACTS, New York, IEEE Press, 2, ISBN [2] C. Rehtanz, Dynamic Power Flow Controllers for Transmission Corridors, Proceedings of the 27 irep Symposium, Charleston, USA, August 27 [3]. Zhou and J. Liang, Overview of control schemes for TCSC to enhance the stability of power systems, IEE Proc.- Gener. Transm. Distrib. Vol. 146 No. 2, March 1999 [4] N. Yang, Q. Liu and J. D. McCalley, TCSC Controller Design for Damping Interarea Oscillations, IEEE Transactions of Power Systems, Vol. 13, No. 4, November 1998 [5] B. Chaudhuri, R. Majumder and B. C. Pal, Application of Multiple-Model Adaptive Control Strategy for Robust Damping of Interarea Oscillations in Power System, IEEE Transactions on Control Systems Technology, Vol. 12, No. 5, September 24 [6] P. Bresesti, M. Sforna, V. Allegranza, D. Canever and R. Vailati, Application of Phase Shifting Transformers for a secure and efficient operation of the interconnection corridors, Proc. of the IEEE Power Engineering Society General Meeting, June 24 [7] N. P. Johansson, H.-P. Nee and L. Ängquist, Estimation of Grid Parameters for The Control of Variable Series Reactance FACTS Devices, Proceedings of the IEEE PES General Meeting, June 26 [8] N. P. Johansson, H.-P. Nee and L. Ängquist, Adaptive Control of Controlled Series Compensators for Power System Stability improvement, Proceedings of IEEE PowerTech 27, July 27 [9] N. P. Johansson, H.-P. Nee and L. Ängquist, Discrete Open Loop Control for Power Oscillation damping utilizing Variable Series Reactance FACTS Devices, Proc. of Universities Power Engineering Conference, Newcastle, UK, September 26 [1] N. P. Johansson, H.-P. Nee and L. Ängquist, An Adaptive Model Predictive Approach to Power Oscillation Damping utilizing Variable Series Reactance FACTS Devices, Proc. of the Universities Power Engineering Conference, Newcastle, UK, September 26 [11] U. Gabrijel and R. Mihalic, Direct Methods for Transient Stability Assessment in Power Systems comprising Controllable Series Devices, IEEE Trans. on Power Systems, Vol. 17, No. 4, Nov. 22 [12] P. Kundur, Power system stability and control, McGraw-Hill, 1994, ISBN , pp and [13] J. K. Åström and B. Wittenmark, Adaptive Control, Addison-Wesley, 1995, ISBN , pp th PSCC, Glasgow, Scotland, July 14-18, 28 Page 7