Transmission System Operator CYPRUS Modeling of the Dynamic Behavior of the Island Power System of Cyprus Andreas G. Petoussis Stavros Stavrinos Session 6: Modeling and Power Quality Friday, 24 September 2010, 17:00-18:30 DEMSEE 2010: 5th International Conference on Deregulated Electricity Market issues in South-Eastern Europe 23-24 September 2010, Sitia, Crete, Greece 1
Table of Contents The Power System of Cyprus Dynamic Behavior of Isolated Island Systems and the need for precise modeling Modeling of Power Plants: Generator Turbine-governor System Excitation System Model Development Model Validation Comparison of simulated and recorded system dynamic responses Conclusions, ongoing and future work 2
The Power System of Cyprus Isolated power system (50 Hz) Three power stations: 1438 MW installed Capacity Transmission system (132 k V and 66 kv) Distribution system (22 and 11 kv ) Installed/Available Generation Capacity: 1438/1272 MW Steam, Gas, ICE and CCGT technologies Fuel: mainly heavy fuel oil, diesel Natural Gas is to arrive in 2013-2014 Peak Load: 2008: 1010MW, 2009: 1093 MW, 2010: 1148 MW Installed generation capacity from RES-e: very small at present but increasing: The first Wind Farm in Cyprus was commissioned in Summer 2010 with a capacity of 82 MW 3
Generation Capacity in Cyprus Company/ Power Station Combined Cycle Plant Steam Turbines Gas Turbines Internal Combustion Engine (ICE) Nominal Nominal Nominal Nominal Capacity Capacity Capacity Capacity (MW) (MW) (MW) (MW) TOTAL GENERATION CAPACITY Installed Generation Nominal Capacity (MW) Available Generation Capacity (MW) EAC /Moni - 6 x 30 = 180 4 x 37,5 = 150-330 236 EAC / Dhekelia - 6 x 60 = 360-100 460 400 EAC/ Vassilikos 2 x 72,5 + 75 = 220 3 x 130 = 390 1 x 37,5 = 37,5 - - 648 636 TOTALS 220 930 188 100 1438 1272 S. Stavrinos, DEMSEE 2010 4
Dynamic Behaviour of Isolated Systems Large interconnected systems are able to share reserve capacities => can maintain reduced reserves as Frequency control is not a significant concern Frequency control is a significant concern in the operation of islanded power systems. Large frequency excursions caused by sudden changes in system operating conditions such as forced generation and/or transmission outages must be quickly compensated to preserve system integrity and facilitate survival of system Small isolated systems do not have robust dynamic response in the event of MW generation outage => under-frequency load shedding (UFLS) may operate even with large reserves maintained 5
The need for precise dynamic modelling Adequate amounts of fast-actingspinning reserve must be carried to provide rapid support of system frequency following such disturbances, in order to reduce the potential for under-frequency load shedding and help protect generating plant equipment from unnecessary and prolonged exposure to under-frequencies If the amount of primary spinning reserve carried at any time is kept at the minimum required for satisfying the security criteria, can lead to significant savings 6
Spinning Reserve vs Primary Operating Reserve Conventional thinking is that the entire spinning reserve in the system is available to support system frequency in the first few seconds following a disturbance. But, the amount of additional output from each generating unit in the seconds timeframe depends upon its primary response capability. Hence, operating a system with larger amount of spinning reserve, without considering primary response capability of the on-line units, may not prevent under-frequency load shedding. Furthermore, lack of knowledge can lead to carrying more reserve than is needed to satisfy the security criteria. This results in higher operating costs 7
Significance of Primary Operating Reserve Recorded System Behaviour With Large Spinning But Rather low POR
Windfarmbehaviour: a new challenge for System Operation Dr. Andreas Petoussis 9
Derivation ofpor Capability Curves The Primary Response Capability Curve for each Generating Unit was constructed by artificially subjecting the unit to a representative test-frequency curve at various MW loading levels:
POR Capability Curves
Real-time Calculation of System POR and display in SCADA HMI
Important Benefit Real-time monitoring of the POR offered by online generating Units, enabled the TSO to base the dispatch of the system and the scheduling of the system reserves on the POR instead of Spinning Reserve. This is considered to be a great accomplishment as system security of the small and isolated system of Cyprus has substantially improved. 13
Actions taken for dynamic model validation Designed and installed high speed loggers at each generating unit Collecting generator response system frequency data during Developed an interface between the SCADA/EMS and Dynamic Simulation Software Snapshots of the State Estimator solution are saved every 5 minutes Dr. Andreas Petoussis 14
Modeling of Power Plants Generator model Turbine-governor system model Excitation system model 15
Generator Model Modeling of the generator of a power plant requires accurate values for the following information: MW, MVA, kv, p.f. ratings Reactive power limits Stator resistance Zero/negative sequence resistance and reactance Leakage reactances Saturation parameters Mechanical damping Turbine-generator inertia constant (kw-sec/kva) Direct/quadrature-axis unsaturated/saturated synchronous/transient/subtransient reactances Direct/quadrature-axis transient/subtransienttime constants 16
Turbine-governor System Model Modeling of the governor system of a power plant: turbine droop characteristics various time constants turbine rated power other constants, factors and coefficients that represent the governor operation (depending on the governor type) Define the transfer function of the governor in relation to frequency deviations and setpointoperation using Laplace-domain control block diagrams in accordance with IEEE standard prime mover models for thermal units Incorporate other control functions, such as the nonlinear operation of frequency deadband 17
Excitation System Model Modeling of the automatic voltage regulators (AVR) of the excitation systems: various time constants and time delays saturation factors other constants and coefficients that represent the excitation system operation depending on AVR type AVR models can be described using Laplacedomain per-unit block diagrams in accordance with IEEE models 18
Dynamic Data Collection and Model Development Sources of data: Witness Test reports on the actual units Manufacturers design data Correspondence documents between owner and contractors Technical data provided by consultants Turbine-governor and AVR models were chosen/customisedto match the technical characteristics of each generating unit the associated parameters were adjusted according to practical requirements parameter fine-tuning was performed to enhance the accuracy of the simulated dynamic response 19
System Conditions Historical Archiving The model automatically retrieves system snapshots from the external SCADA/EMS system to be used as the initial power flow conditions for the dynamic simulation Real-time state estimator solution is available every five minutes to be readily imported in the model The model keeps historically the topology of the transmission/generation system in time-stamped versions of the network => ability to recall the system topology in the past and perform analysis of recorded events/disturbences 20
Model Validation Model validation against actual system responses recorded from the data acquisition system comparison of Recorded and Simulated Events Validation considers the frequency response of the system and the MW response of: individual units individual power stations the system as a whole MVArresponses and bus voltage fluctuations are also considered 21
Trip of 130MW Steam Unit (March 2006) Simulated minimum frequency matches the recorded 22
Trip of 130MW Steam Unit Simulated generator MW response per Power Plant is very close to the recorded 23
Trip of 130MW Steam Unit Simulated MW response of individual generating units match the recorded response with a deviation of only a few MW After 10 seconds, the simulated generation matches almost exactly the recorded 24
Trip of Vassilikos CCGT Steam #40 (March 2010) Frequency at the end of the primary reserve operation (20sec) matches the recorded 25
Trip of Vassilikos CCGT Steam #40 (March 2010) Simulated MW response of power stations overlaps or is very close to the recorded 26
Trip of Vassilikos CCGT Steam #40 (March 2010) (Hydraulic governor) Simulated MW response of individual generating units matches the recorded with a deviation of only a few MW After 20 seconds the two responses match exactly 27
Ongoing/Future Work Analyze as many past disturbances as possible for fine tuning the model Incorporate Windfarms in the dynamic model Models for Wind Turbines are requested by the TSO at the early stage of projects Calculate the minimum required Primary Operating Reserve in real time by iterative dynamic simulations and warn the Control Centre Engineers in cases where the security criteria are not satisfied 28
Conclusions A model has been successfully implemented to simulate the dynamic behavior of the island power system of Cyprus The model has been validated against past recorded frequency and MW responses The simulations have been proven to be reliable and able to predict accurately the dynamic behavior of the system following a disturbance 29
Conclusions (continued) The model can be utilized to: Predict the dynamic behavior of the system at different loading conditions in order to determine the minimum primary operating reserve requirements at any point of time optimize the UFLS settings investigate future generation expansion scenarios Thank you for your attention! 30