Step-Response Tests of a Unit at Atatürk Hydro Power Plant and Investigation of the Simple Representation of Unit Control System

Transcription

1 Step-Response Tests of a Unit at Atatürk Hydro Power Plant and Investigation of the Simple Representation of Unit Control System O.B.Tör, U. Karaağaç, and, E. Benlier Information Technology and Electronics Research Institute (TUBITAK BILTEN) METU, Ankara, Turkey Abstract--This paper presents the methodology, results and experience of the step-response tests performed at Ataturk Power Plant, the largest hydro power plant in Turkey with 2400 MW total rating, regarding primary frequency control performance of the coherent units. Sensitivity of the unit control system and unit response to ±200 mhz frequency deviation both at maximum and minimum output power were tested according to the test program of Union for the Coordination of Transmission of Electricity (UCTE) which was prepared specifically for the interconnection of Turkish grid to the UCTE synchronous network. Speed (frequency) signal was applied to the governor control system by a signal generator artificially during the tests. Main results are such that the sensitivity of the unit control system is smaller than ±5 mhz corresponding to a rated frequency 50 Hz, and complete primary control reserve was activated linearly within seconds roughly depending on loading of the unit. The representation of the unit control system with a simple linear model, IEEEG2, is also investigated in the paper to demonstrate the performance of linear models at ±200 mhz frequency deviation which is the amount of maximum frequency deviation allowed in the synchronous UCTE network. Keywords--Hydro power plant, speed-governor, step-response, primary frequency control, linear model, IEEEG2. I. INTRODUCTION The interconnection of Turkish electrical network with Union for the Coordination of Transmission of Electricity (UCTE) network is an ongoing project since the end of 90 s. The synchronisation of Turkish grid to UCTE through 380 kv network will enlarge the capacity of UCTE system by MW roughly when the project is realized. Preliminary works for this interconnection have been carried out by Turkish utility according to UCTE regulations. Primary frequency control performance test of the power plants planned for primary control is one of those works which were carried out at Ataturk Hydrolic Power Plant (HPP) by the delegations of Turkish Utility and TUBITAK-BILTEN within 2003 summer. Primary frequency-load control, decentralized on the generators, adapts automatically and rapidly the power generated to the demand within the synchronous area, and allows the exchange of mutual help between the interconnnected partners to cope with usual hazards [1]. Ataturk HPP is the largest hydrolic power plant in Turkey consisting of 300 MW rated 8 coherent francis type units connected to 380 kv network. Sensitivity of unit control system and unit response to ±200 mhz artificial frequency deviations both at maximum and minimum output power were tested according to the test procedure of UCTE. The test procedure was prepared specifically for the interconnection of Turkish network to the UCTE synchronous network [2]. This paper consists of two main parts. In the first part, some informations about unit control system at the plant, method of supplying artificial frequency signal, experience in performing tests and results of the tests are presented. The representation of the unit control system with a simple linear speed-governing model is investigated in the second part. II. REVIEW OF UNIT CONTROL SYSTEM AT ATATURK HPP The units of Ataturk HPP have two operating modes: Speed Control (SC) and Power Control (PC). SC is used for start-up, in disturbance states and in island operation. PC mode is used for primary frequency control while the units are supplying power to 380 kv Turkish network. Therefore the tests were carried out in this mode. Speed-governors compare actual turbine speed with the governor reference, 150 rpm, which corresponds to 50 Hz rated frequency, and outputs a control signal to the gate controller in accordance with speeddroop set value, while the units are operating in PC mode. Speed-droop is the change in active power output of the unit propotional to the frequency deviation (open-loop control) as illustrated in Equation 1: f / f n speed - droop (%) = 100 (1) P / Pn f = steady-state frequency deviation, P = change of active power generation caused by turbine governor as a result of the frequency deviation f, f n = rated frequency, P n = unit rated power. The power plant was put into operation in Governors of the units are electro-hydraulic type and the manufacturer is Sulzer Escher Wyss from Switzerland. Since the control systems are not computerized, the artificial speed signal required for the tests was applied to the governor by a signal generator as explained in the following section.

2 III. METHOD OF SUPPLYING ARTIFICIAL SPEED SIGNAL There are four tachometers located on the shaft of each turbine to measure the turbine speed. Their output signal frequency is proportional to turbine speed, such that rpm corresponds to Hz, 10 V peak (nearly) wave. Two tachometers provide the speed value to relays for protection against overspeed. The other two supply governor for gate control. The highest value of the measurements is considered in both control systems and therefore, loosing one of the tachometer output signal does not cause any problem for the operation of units. This is a method of providing reliability and utilized for supplying artificial speed signal to the governor during the tests as follows: While the unit was supplying the network at both maximum (max) and minimum (min) output power, one of the tachometers output signal was disconnected from the governor control system from its supply connector. Then, a digital signal generator supplying 1000 Hz, 10 V peak square wave signal, which corresponds to the rated grid frequency (50 Hz), was connected instead of the disconnected signal. Finally, the other tachometer output signal was disconnected from the circuit. The procedure is illustrated in Figure 1. The switches in the figure are numbered according to their operation order. f (p.u.) Fig. 2. ±5 mhz step-response of the unit. Table 1. Speed-droop card (RPV10) settings. P Calculated RPV10 Settings (p.u.) speed-droop Tuned Specified % 5% 8% Scale = 37.5 (X=10 Kp=3.75) Scale = 29 (X=10 Kp=2.90) Scale = 17.5 (X=10 Kp=1.75) Scale = 50 Scale = 40 Scale = 25 Fig. 1. Scheme of supplying artificial speed signal to the speed governor. IV. STEP-RESPONSE TESTS AND RESULTS Sensitivity of the unit control system was tested firstly. Dead-band was adjusted to zero from the potentiometer existing on the dead-band card (ETG10 in Figure 1) and small step deviations were applied in artificial frequency signal by the signal generator. Gate opening and active output power of the unit were recorded to a laptop through a multi-channel data acqusition device with 10 sample/sec rate. It was observed that the sensitivity is smaller than ±5 mhz at 50 Hz rated frequency. The response of the unit is given in Figure 2. The plot covers 4 minutes of data approximately. According to the requirements of UCTE, the entire primary control reserve of the units must be fully activated in response to a quasi-steady-state frequency deviation of ±200 mhz or more [3]. Regarding this criteria, responses of the unit to ±200 mhz step deviations were observed at almost max and min output power (initial loading or operating point). The potantiometers existing on the droop-card of speed-governor (RPV10 in Figure 1) were tuned firstly. The difference between the settings determined for 3 speed-droop values and those specified in data-sheet of the card is significant, as seen in Table 1. Responses of the unit for 5% speed-droop setting are given in Figure 3. The plots cover 2 minutes of data. The complete primary reserve (0.08 p.u. according to 5% speed-droop setting) was activated within 20 and 40 seconds roughly at min and max output power, respectively, as seen in the figure. The amount of gate opening deviation depends on the initial loading of the unit as illustrated in Table 2. Therefore, the settling times at max and min loading differ for the same amount of output power activation. This issue is covered in the following section. Fig. 3. Responses of the unit to ±200 mhz deviation (speed-droop = 5%).

3 Table 2. The amount of deviations in gate opening ( g). Speed-droop Output power f P g (initial loading) (p.u.) (p.u.) (p.u.) (p.u.) 5% V. SIMPLE REPRESENTATION OF THE UNIT CONTROL SYSTEM AT ATATURK HPP In this part of the paper, investigation of representing the unit control system with a simple linear model is presented. It is intended to see the performance of a linear model at ±200 mhz step-deviation which is the amount of maximum frequency deviation allowed in the synchronous UCTE network. A. Theory The dynamical system models enable to simulate the network failures such as, short circuits, line or unit tripping and significant load changing. The model level of the systems depends on the purpose, which they are used for. There are several standard models available for hydro-electric power plant control systems. Although they have similar governor models, they generally differ in hydraulic system representation. Linear and non-linear models are the most common representations. The other sophisticated models, such as travelling wave model, are sometimes used for detailed plant studies. The fundamental scheme of speed-governing system for hydro-electric power plants is given in Figure 4. and head and uses the water starting time (T w ) to represent the transient behaviour of the hydrolics. 1 Tw s G( s) = 1+ T 0.5s The water starting time, T w, represents the acceleration time of water in the penstock between the turbine and the forebay (or surge tank) and it is calculated by the following equation: T w L Q = (sec) g A H v Q = Flow through turbine at rated loading level (m 3 /s), L = Length of penstock (m), g v = Gravitational acceleration (m/s 2 ), A = Penstock cross-sectional area (m 2 ), H = Head at the rated loading level (m). The flow through turbine (Q) varies with load, therefore so does T w. Thus for simulation studies, the linearised hydrolic system model requires different values for T w to reflect changes in water rate and it is valid only as long as the unit s load does not deviate greatly from the initial load [6]. The effects of the governors and turbines on the damping of low requency inter-area modes can be modelled adequately by linearizing the non-linear turbine and governor models about the appropriate operating point. Linear models are also used for guidance in speed control tuning using linear control analysis techniques [7]. C. Linear representation of the tested ınit Representation of the unit control system by a linear model is investigated in this part. Standard speed governor model of IEEEG2 (Figure 5) is utilized for this intension. w (2) (3) Fig. 4. Speed-governing control system fundamental scheme for hydroelectric power plants. In reality, the dynamics of hydro turbines and the water flowing through them are quite complex and can be modeled in great detail, most of which is unnecessary for the purposes of large scale system stability studies. For most studies, that is, those in which system frequency does not change drastically (e.g., islanded or small systems), the components of the hydro governor model of primary importance are the portions with time constants which are represented well in linear models [4]. B. Linear approach The penstock/turbine combination (hydrolic system) is represented by the classical transfer function of Equation 2 in most of linear models relating changes in the mechanical power to changes in the guide vane opening [5]. This representation assumes an ideal turbine operated at rated flow Fig. 5. IEEEG2 speed-governing model. w = Speed deviation [p.u.], K = Governor gain (inverse of the governor speed droop), Po = Initial loading of the machine [p.u.], P MAX = Maximum output power of the machine [p.u.], P MIN = Minimum output power of the machine [p.u.], T 1 = Governor controller lag (sec), T 2 = Governor controller lead (sec), T 3 = Governor lag (sec), T 4 = Water starting time (sec). MATLAB/SIMULINK [8] software tool was utilized to simulate the model. The actual responses of the unit were compared against simulated ones determined using different model parameters. Step-responses of the model which

4 approximates the actual responses most both at max and min output power are shown in Figure 6. smaller than ±5 mhz corresponding to a rated frequency 50 Hz, and primary reserve of the unit activates linearly within seconds roughly depending on the initial loading of the unit, as illustrated in the paper. Representation of the unit control system with a linear model is also reviewed in the paper. Simple linear model IEEEG2 was utilized for this intension. It is demonstrated that, although the unit control system can not be identified by this simple linear model accurately due to dependence of unit transient behaviour on operating point mainly, if the time constants of the model are adjusted for the operating point, then the response of the model approximates the actual response quite tolerably at ±200 mhz frequency deviation which is the amount of maximum frequency deviation allowed in the synchronous UCTE network. Fig. 6. Comparison of measured and simulated response (IEEEG2 parameters: K= 20 (for 5% speed-droop), T 1 = 8, T 2 = 2.85, T 3 = 1, T 4 = 0.8). Dependency of the response time on initial loading is more clear in Figure 6. Since the penstock/turbine combination is linear, response of the model does not depend on loading of the unit (i.e. the simulated responses at both operating points are same). Also, output power oscillation due to travelling wave effects in the water column [6] is absent in the linearised model as seen in the figure. If an accurate representation of the governor control system is necessitated, then the models that take into account those nonlinear effects should be taken into consideration which require more detailed turbine data and additional tests for validation [7]. On the other hand, it is possible to adjust the parameters of the linear model in such a way that the model response approximates the actual response more closely at either loading of the unit. However, the approximation deteriorates at the other loading obviously (see Figure 6). The responses of the model which the parameters are adapted for 0.64 p.u. loading are illustrated in Figure 7. Fig. 7.a. IEEEG2 parameters adapted for 0.64 p.u. loading (-200 mhz stepresponse). VI. CONCLUSION This paper presents the experience and results of stepresponse tests performed at one unit of Ataturk Hydro Power Plant regarding primary frequency control performance of the coherent units. Sensitivity of the unit control system and unit response to ±200 mhz frequency deviation both at maximum and minimum output power were tested according to UCTE test program which was prepared specifically for the interconnection of Turkish network to the UCTE synchronous network. Frequency signal was applied to the governor control system artificially by a signal generator during the tests. Test results show that, the sensitivity of the unit control system is Fig. 7.b. IEEEG2 parameters adapted for 0.64 p.u. loading (+200 mhz stepresponse).

5 VII. REFERENCES [1] M. Trotignon, Load Frequency Control, Operational Studies for the Interconnection of the Electrical Power Systems of the EIJST Countries, EIJST Workshop, EDF, Dec [2] Program of Preliminary Tests of the Units Planned for Primary Control, Primary Frequency Control Issues for the Interconnection of Turkey to the UCTE Synchronous Area, UCTE. [3] Policy 1: Load-Frequency Control and Performance [D], UCTE Operation Handbook (Draft), UCTE, July Available: [4] L. Pereira, J. Undrill, D. Kosterev, D. Davies, S. Patterson, New Thermal Turbine Governor Modeling for the Wecc (Revised), WECC Modeling & Validation Work Group, October [5] S. Patterson, Hydro Governor Modelling, WECC governor modelling workshop presentation, Aug [6] Sa'ad Mansoor, Behaviour and Operation of Pumped Storage Hydro Plants, Ph.D dissertation, Dept. Elec. And Cont. Eng., University of Wales, July [7] Hydraulic Turbine and Turbine Control Models for System Dynamic Studies, IEEE Committee Report, IEEE Transactions on Power Systems, Vol. 7, No. 1, pp , Feb [8] VIII. BIOGRAPHIES Osman Bülent Tör graduated from electrical and electronics engineering department of Middle East Technical University (METU) in Ankara, Turkey with B.Sc. in 1998 and M.Sc. in He is studying towards completion of Ph.D. at METU. He has worked mainly on modeling and simulating power systems since He has taken different roles in several power system analysis projects performed for Turkish utility. He is working as a specialist R&D power engineer at Information Technology and Electronics Research Institute (BILTEN) of the Scientific and Technical Research Council of Turkey (TUBITAK). osman.tor@bilten.metu.edu.tr ph: Ulaş Karaağaç graduated from METU in Ankara, Turkey with B.Sc. (1999) and M.Sc. (2002) in electrical and electronics engineering. Since 1999 he has worked mainly on steady-state and time-domain simulations of Turkish power system. He is working as a specialist R&D power engineer at Information Technology and Electronics Research Institute (BILTEN) of the Scientific and Technical Research Council of Turkey (TUBITAK). He is studying towards completion of Ph.D. at METU. ulas.karaagac@bilten.metu.edu.tr Emre Benlier graduated from electrical and electronics engineering department of METU in Ankara, Turkey with B.Sc. in 1997 and M.Sc. in He is studying towards completion of Ph.D. at Ankara University in Ankara, Turkey. He has worked mainly in power electronics and control systems areas since He is specialized on electric distribution automation. He is working as a senior R&D engineer at Information Technology and Electronics Research Institute (BILTEN) of the Scientific and Technical Research Council of Turkey (TUBITAK). emre.benlier@bilten.metu.edu.tr