OPTIMIZED STEAM TURBINE GOVERNOR CONTROLLING SINGLE OR MULTIPLE GRID FAULTS

Transcription

1 Proceedings of ASME Turbo Expo 2012 GT2012 June 11-15, 2012, Copenhagen, Denmark GT OPTIMIZED STEAM TURBINE GOVERNOR CONTROLLING SINGLE OR MULTIPLE GRID FAULTS Martin Bennauer SIEMENS AG Energy Sector Fossil Power Generation Division Mülheim an der Ruhr, Germany Christian Kreischer TU Dortmund University Department of Electrical Engineering and Information Technology Institute of Electrical Drives and Mechatronics Dortmund, Germany Jens Rosendahl SIEMENS AG Energy Sector Fossil Power Generation Division Mülheim an der Ruhr, Germany Heribert Werthes PowerDyn GmbH Essen, Germany ABSTRACT This paper uses simulation results to figure out how SIEMENS handles a single grid fault of a three phase short circuit close to the power plant and a multiple grid fault of a load rejection due to a three phase short circuit close to the power plant. The correct response of the steam turbine governor to the single and the multiple grid fault can increase power plant availability, prevent steam turbine trips and reduce overspeed. Therefore the power supply to the electrical grid can still be maintained, and in case of a de-stabilized electrical grid a black-out could be prevented. If the steam turbine trips caused by the single or the multiple grid fault the steam turbine runs down and has to be restarted. This could take several hours as worst case. During this time the power plant is not available to support and stabilize the electrical grid. If the steam turbine governor is able to handle the single grid fault, the power plant is still connected to the grid. This will not further stress the de-stabilized electrical grid. If the steam turbine governor is able to handle the multiple grid fault, the house load of the power plant will still be provided by the steam turbine. The operator could resynchronize the power plant with the de-stabilized electrical grid as fast as possible. In both cases for the single and the multiple grid fault the power plant remains available by the optimized steam turbine governor. INTRODUCTION Many European Grid Codes stipulate that a power plant must withstand short circuits in the power transmission system during a specified fault clearing time without damage to support voltage and frequency stability on the grid. The Nordic Grid Code [1] for example demands stability of the generator for a single grid fault like a three phase short circuit close to the power plant for a minimum fault clearing time of 250 ms. This special customer demand exists for small power generation units connected to the medium-voltage power grid as well as for large fossil power plants. The examples discussed in the following work deal with a large steam turbine power plant. Grid Codes do not necessarily assume that power plants must also be able to control another fault occurring a few seconds later. Since load rejection to house load can certainly occur following a short circuit, e.g. during extreme weather conditions or environmental disasters, it would be of great advantage to enhance the power plant availability if a multiple grid fault of this kind could also be controlled. For the multiple grid fault it is assumed that the power plant described here is connected to the grid via two transmission lines. During storm a three-phase short circuit (for example a tree cuts all three 1 Copyright Siemens Energy, Inc All rights reserved.

2 phases) occurs in one of the two transmission lines which is cleared after 250 ms. Thereafter a two-phase short circuit (for example two phases contact each other) occurs in the remaining second transmission line which result in a load rejection to house load for the power plant. The rejection of electrical load shortly after a short circuit can give rise to very high overspeeds however and these will cause tripping by the dedicated overspeed protection system. An overspeed trip results in shutdown of the turbine generator and the extended downtimes associated with it. The main benefit of load rejection to house load as compared with the shutdown by a protection system is the immediate re-establishment of availability, i.e. the power plant could be synchronized with the grid only a few seconds later. A SIEMENS steam turbine generally has a combined speed/load controller with PI structure. The setpoints of position for the valves are determined by the setpoint of the load via proportional and integral components as long as the generator breaker is closed and the speed and grid frequency are constant. By the occurrence of a sudden change to stationary operation which leads to greater deviations from the nominal speed, automatic switchover to the speed controller is performed. Than the output of the speed controller throttles the valves and controls indirectly the mechanical torque of the turbine that corresponds to the active power output of the generator. The simulations are based on a previously validated NETOMAC (Network Torsion Machine Control, [2]) model of a steam turbine power plant on the grid with a rated power of 1000 MVA. The model has been improved constantly. The investigation of interdependencies between turbine control, generator and grid was topic of several interdisciplinary research projects. SINGLE GRID FAULT In the event of a single grid fault, a generator must supply active and reactive power for as long as possible so as to support the stability of frequency and voltage. If, however, the generator slips out of synchronism with the grid or if the shaft train is endangered due to overspeed, for example, the turbine generator must immediately be disconnected from the grid and switched to house load operation. The controller monitors the electrical power on the generator terminals. If specified limits (load band around zero) are exceeded, fast travel mode is triggered for the valves and the mechanical load is decreased as quickly as possible. The valves remain closed for a hold time of approx. 1.5 s, after which time they are slowly reopened. It is impossible to distinguish between load rejection to house load and a grid short circuit on the basis of the active power of the terminals at the beginning of the event. Unlike load rejection, electrical active power returns after a grid short circuit once the fault has been cleared and the valves must then reopen. The generator then must continue to supply active and reactive power to the grid, so as to stabilize frequency and voltage. This means that if the short circuit is active for 250 ms or less, no slip must occur on the shaft train and the mechanical power input must not be decreased completely [3]. Since decreasing the turbine momentum has a stabilizing effect [4], immediate fast closure of the valves enhances the dynamic behavior of the shaft train in the case of the single grid fault like the three-phase short circuit. Because the electrical power continues to oscillate through zero several times, the controller interprets it as another load drop when the lower limit values (load band around zero) are violated. The load rejection identification function integrated in steam turbine controllers makes it possible to differentiate between a short circuit and load rejection. This controller module supplies a signal which is set only once for a short time following a short circuit and which is subsequently blocked for several seconds, i.e. until the after-effects of the grid fault have abated for the most part. Figure 1 shows the results of the simulation. The mechanical momentum is restored quickly and electrical power reaches the steady-state condition after about 8 s. Stable voltage and frequency accompany this steady-state electrical power. Figure 1: Single grid fault with a fault clearing time of 250 ms It can also be seen that the IP and HP valves close quickly and that they reopen completely after a short time. The speed oscillates around the rated value within a deviation of a few percent, i.e. the machine is stable. The power on the generator terminals exhibits the familiar behavior of a steam power plant in the event of a three-phase grid short circuit with the associated fault clearing times [5]. From the point of view of stability, significantly longer fault clearing times or connections to weak power grids can generally only be controlled by means of special control measures, such as Fast Valving [4] or Early Valve Actuation. Serious grid faults like a three-phase short circuits cannot be tested on real applications economically and therefore investigations have to rely on adequate simulation models. The simulation results of the model have to be validated somehow with available measurements of existing power plants. For example a good agreement with measurements of a load rejection to house load is appropriate to show that the dynamic model is adequate for the following investigations. The results of the simulation of load rejection shown in figure 2 illustrate the behavior of the steam turbine power plant 2 Copyright Siemens Energy, Inc All rights reserved.

3 following a decrease in the power on the terminals to house load. Fast closure of valves is also necessary during load rejection to house load so as to protect against overspeed. Since the same criteria are fulfilled as during a grid short circuit, fast travel mode is automatically actuated once. The valves must remain closed thereafter, however, until such time as the active power of the turbine generator has reached the house load level. The signal for load rejection is first delayed by several 100 ms for this purpose, until it has been ensured that a grid short circuit has not occurred. If the power stays close to zero for significantly longer than the fault clearing time, the control can be switched over to house load. rejection identification function immediately triggers fast travel mode for the IP and HP valves (second diagram in figure 3) by issuing the transient grid disturbance signal. As shown in the third diagram the fast travel mode decreases the mechanical load as fast as possible and the valves stay closed for a preset time of approximately 1.5 s before reopening again. For all following simulations results the fault is active for a fault clearing time (FKZ) of 250 ms which corresponds to the worst case. During the fault the generator cannot deliver active power to the grid. The remaining mechanical power of the steam turbine therefore initially results in an acceleration of the shaft and an small increase of the shaft speed (third diagram in figure 3). Figure 2: Load rejection to house load Following fast closure of the valves, the mechanical momentum decreases slowly due to the high moment of inertia. This causes the shaft train to accelerate, thus permitting it to absorb the excess turbine power which can no longer be transferred to the grid. The speed increases up to about 6 % above the rated value, as can be seen from the bottom graph in figure 2. After that the speed controller largely determines the control signal for valve opening. This causes the valves to remain closed and the driving torque of the turbine is reduced to zero as specified until the speed falls below the setpoint. MULTIPLE GRID FAULT The behavior of the current control system and the turbine generator on the grid in response to a multiple grid fault in the form of load rejection to house load occurring shortly after a three phase grid short circuit close to the power plant is described below. Special attention is paid to the level of overspeed relative to the speed at rated operation. If the specified requirements are to be fulfilled, the protection system must not trip due to high overspeeds ( 110 % of the rated speed). MULTIPLE GRID FAULT: LOAD REJECTION DURING VALVE HOLD TIME The first diagram in figure 3 shows the active power of the Generator over time related to the nominal apparent power. After one second of simulation time the three phase grid short circuit is initiated close to the high voltage side of the transformer. The power on the generator terminals decreases instantly due to the drop in voltage to almost zero. The load Figure 3: Multiple grid fault with 1 s delay prior to load Although the generator power is immediately restored following the fault clearing, it exhibits major oscillations around the rated value until steady state operation resumes a few seconds later, assuming no further fault has occurred. In the case illustrated in figure 3, however, load rejection to house load occurs 1 s after fault clearing, i.e. while the power oscillations resulting from the grid short circuit are still active. Because load rejection occurs during the valve hold time, the control system can generally handle it without any problem. However it is important to mention that a delay time - labeled TLAW - in the diagram remains between the detected start of the load rejection and the second actuation of fast travel mode. The load rejection signal which triggers fast travel mode the second time is delayed by time TLAW in order to differentiate between the grid short circuit and a load rejection. Since both faults exhibit the same behavior initially and the same measures are required in both cases, fast travel mode is first triggered by the transient grid disturbance signal and is later confirmed by the load rejection signal. Hence, TLAW is an individual preset value of the load rejection identification function. However, because the transient grid disturbance signal is inhibited for some time, the delayed load rejection signal must trigger fast travel mode in the second instance. As we can see from figure 3, this correlation does not cause any problems in the event of a 1 s delay between the two events, as the valves are still closed when load rejection occurs and are only partly reopened after the time TLAW has elapsed. The turbine momentum is accordingly low in this condition and 3 Copyright Siemens Energy, Inc All rights reserved.

4 the shaft train is accelerated only slightly. The final diagram shows that the maximum overspeed is only approximately 2 %. MULTIPLE GRID FAULT: LOAD REJECTION FOLLOWING VALVE REOPENING Based on the above considerations we can conclude that load rejection can only result in significant overspeeds once the valves have reopened and the turbine momentum has increased to higher levels. Figure 4 illustrates such a case for the power plant as an example. At the time when load rejection occurs, that is to say 2.6 s after fault clearing, the valves are already partly open and the turbine momentum is accordingly high (refer to second and third diagram in figure 2). The delay time TLAW is problematic here, as the valves remain open during this time in the absence of a load. Because the momentum increases during this time, the shaft train is accelerated for a longer period than it would be in the event of load rejection to house load without a prior grid fault. The overspeed of approximately 8 % which occur as a result is in the range not yet monitored by the speed protection system. The time of 2.6 s between the two faults results in maximum overspeed determined by the model used in this study. Figure 4: Multiple grid fault with 2.6 s delay prior to load For this reason a value as low as possible should be set for TLAW, so that overspeeds are limited to avoid an unnecessary speed protection trip in the event of the multiple grid fault. MULTIPLE GRID FAULT: LOAD REJECTION AFTER INHIBIT TIME FOR TRANSIENT GRID DISTURBANCE HAS ELAPSED The inhibit time for the transient grid disturbance signal ensures that load rejection identification is reset to the basic state once this time has elapsed. This means that the control system will treat any subsequent fault in exactly the same way as the initial fault. It is assumed that once the inhibit time has elapsed the effects of the first fault will have abated to such an extent that they will have no effect on a subsequent fault. In the event of load rejection to house load following by the fault clearing time for the grid short circuit plus the inhibit time, it follows that the unit will behave as if two independent faults have occurred. This behavior is shown in figure 5. The time between the end of the three-phase short circuit and the event of the load rejection to house load is 7.5 s in the considered simulation. The overspeed after the load rejection to house load, which can be treated as a single event, exceeds the rated speed about 6 % and therefore is not critical regarding the protection system (refer to load rejection in figure 2). Figure 5: Multiple grid fault with 7.5 s delay prior to load SUMMARY Large-scale steam power plants must be able to withstand a single grid fault like a three-phase grid short circuit close to the power plant for a specified fault clearing time without damage and must support voltage and frequency stability on the grid, e.g. the Nordic Grid Code [1] demands a minimum time of 250 ms. The control concept of SIEMENS steam power plants meets these requirements and is also capable of controlling further load rejection to house load occurring a few seconds later. Load rejection to house load increases availability of the power plant as compared with shutdown by a protection system, as the plant can be synchronized with the grid only a few seconds after this event. Although such multiple grid fault occurs infrequently, their handling might be of high interest in case of stability problems in larger grid areas, e.g. during extreme weather conditions or environmental disasters. The load rejection identification function in the control system of SIEMENS steam turbines is capable of handling the single and the multiple grid fault. The simulations performed on the power plant model using NETOMAC [2] show that overspeeds which trigger protection trips do not occur when load rejection occurs subsequent to a grid short circuit. The delay time preset by the parameter TLAW must be set to a low value which still allows a sure distinction between load rejections to house load and three-phase short circuits. NOMENCLATURE A Ampere FKZ Fault clearing time HP High pressure IP Intermediate pressure M 10 6 NETOMAC Digital simulation software (Network Torsion Machine Control) p.u. Per unit Registered trademark 4 Copyright Siemens Energy, Inc All rights reserved.

5 s TLAW V Second Delay time for load rejection identification Voltage Permission for Use: The content of this paper is copyrighted by Siemens Energy, Inc. and is licensed to ASME for publication and distribution only. Any inquiries regarding permission to use the content of this paper, in whole or in part, for any purpose must be addressed to Siemens Energy, Inc. directly. REFERENCES [1] Nordic Grid Code 2007 (Nordic collection of rules). [2] Kulicke, B., Digitalprogramm NETOMAC zur Simulation elektromagnetischer und magnetischer Ausgleichsvorgänge in Drehstromnetzen. Elektrizitätswirtschaft, H78, S [3] Kulig, T. S., Auswirkungen von Störfällen in elektrischen Energieübertragungsnetzen auf Kraftwerksturbosätze. Habilitationsschrift, Fernuniversität Hagen. [4] Canay, M., Bloch, H., Fast Valving, ein Mittel zur Netzstabilität. Brown Boveri Mitt. 6-79, S [5] Kindermann, W., Fork, K., Verhalten der Dampfturbinen-Leistungsregelung großer Kraftwerksblöcke bei Netzkurzschlüssen. VGB Kraftwerkstechnik 59, Heft 6, S Copyright Siemens Energy, Inc All rights reserved.