28
Recent Developments and Applications of PMU/WAMS in China In the development of smart grid, the measurement technology plays a fundamental role for many advanced power system analysis and control. Phasor measurement unit (PMU) and wide area measurement system (WAMS) are becoming the critical measurement infrastructures for the transmission and generation system. As of 2013, about 2400 PMU sets have been deployed in Chinese power grids, covering all 500kV substations and some important power plants and 220/110kV substations. In addition, more than 30 WAMS center stations are in service, providing important dynamic information about the power system operation. Most of the PMU devices were deployed after 2006, when the previous article introducing basic architectures and functions of PMU/WAMS in China was published in Power and Energy Magazine. In this paper, the recent and emerging development of PMU/WAMS, communication and synchronization network in China will be summarized briefly, and then some major advanced applications or novel demonstrating projects utilizing synchrophasor measurement technology will be emphasized and presented. 1. Recent Development and Applications of PMUs in China

essaystar.comessaystar.com/.../WordNameA/UP-201501190658reb-1.d…  · Web viewPower System Dynamics Monitoring and Analysis. ... have been studied and implemented as a key function

Embed Size (px)

Citation preview

Recent Developments and Applications of PMU/WAMS in China

In the development of smart grid, the measurement technology plays a fundamental role for many advanced power system analysis and control. Phasor measurement unit (PMU) and wide area measurement system (WAMS) are becoming the critical measurement infrastructures for the transmission and generation system. As of 2013, about 2400 PMU sets have been deployed in Chinese power grids, covering all 500kV substations and some important power plants and 220/110kV substations. In addition, more than 30 WAMS center stations are in service, providing important dynamic information about the power system operation. Most of the PMU devices were deployed after 2006, when the previous article introducing basic architectures and functions of PMU/WAMS in China was published in Power and Energy Magazine. In this paper, the recent and emerging development of PMU/WAMS, communication and synchronization network in China will be summarized briefly, and then some major advanced applications or novel demonstrating projects utilizing synchrophasor measurement technology will be emphasized and presented.

1. Recent Development and Applications of PMUs in China

Chinese power transmission grid is run by two Transmission System Operators (TSOs), i.e., State Grid Corporation of China and China Southern Power Grid. Both have deployed large numbers of PMUs as part of nationwide mandates that require all substations on the 500kV networks and above and generators larger than 100MW are required to be monitored by PMU. The deployment map of PMU is displayed in Figure 1.

Figure 1. Deployment of PMU in China (as of 2012)

By the end of 2013, the number of PMUs installed at the substations and power plants in State Grid of China is 2027. Sorted by the voltage level, the statistical installation of PMUs is shown in Table 1.

All of the running PMUs are made by 7 domestic manufactures. The top 4 manufactures in terms of amount of equipment are: Beijing Sifang Automation, NARI-Technology, China EPRI and NARI-Relays. The total amount of equipment manufactured by them are over 98%.

Table 1 The amount and proportion of PMUs in the generators and substations at over 110kV level

Voltage Level (kV)

1000

800

750

660

500

400

330

220

110

Amount

7

6

39

26

723

165

861

200

Proportion (%)

0.35%

0.30%

1.92%

1.28%

35.67%

8.14%

42.48%

9.87%

Currently, 234 PMUs have been under operation for less than 1 year, 1337 for more than 1 year but less than 6 years, 362 for more than 6 years but less than 10 years, and 15 for more than 15 years. The following diagraph shows the distribution of operation duration of PMUs. This figure shows that more than 80% PMUs were installed in the past 5 years, which means a very rapid growth of the deployment of PMU after launching the smart grid projects in 2008.

Figure 2. Distribution of operation duration of PMUs installed at over 110kV level

Furthermore, China Southern Power Grid has installed 373 PMUs, including 211 located in substations and 162 in power plants.

With the maturing of synchronized phasor measurement technology, current major practice for PMU development in China includes two fields. The first effort is made to support smart substation technique, and the other work is devoted to improve the measurement accuracy based on latest released standards on PMU.

Near 100 PMUs supporting IEC 61850 have been commissioned in recent years. With the fast development of smart substation technologies, over 3000 smart substations have been put into operation, covering voltage range from 110kV to 750kV. Most of these smart substations are located in low voltage levels, but about 100 high voltage substations, are required to possess the synchronized phasor measurement function. Phasor measurement and communication functions realized in smart substation structure are displayed in Figure 3.

In Fig.3, PMU takes inputs from Merging Unit (MU) and Switchgear Control Unit (SCU) via Process Buses, in the form of IEC 61850-9-2 SV and GOOSE protocols. After phasor calculation, the phasors can be transmitted, at slow rate, to monitoring server via Station Buses, in the form of Manufacturing Message Specification (MMS). At the same time, phasor can be transmitted, at fast rate up to 100 frames per second, to substation phasor data concentrator (SPDC) according to IEEE C37.118 standard, and then to remote dispatching center.

Figure 3. PMU functions realized in smart substation structure

At same time, Chinese standards about PMU are keeping updated to conform to the recently released IEEE standards C37.118.1 and C37.118.1a. Much efforts have been devoted to improve the measurement accuracy based on latest released PMU standards. From 2013, some tests about PMU measurement accuracy according to the new standards have been carried out by China EPRI. Further work, including tests for supporting IEC 61850, is also undertaking.

2. Data and Time Synchronization Networks in Chinese Power Grids

In consideration of power transfer security and dispatching automation, the applications based on State Grid Dispatching Digital Network (SPDnet) have a high demand for reliability, real-time and safety. Therefore, key performance requirements, such as network time delay, convergence time and safety, must be guaranteed. Currently, according to the principle of dual independent backbone networks and multiple layers integration, an exclusive dispatching data network based on exclusive channels (synchronous digital hierarchy + IP + optical fiber) has been put into service, which has the URL stratification structure and alternate route function. Highly reliable core equipments together with double master controls are adopted in this network. Three-level Border Gateway Protocol (BGP) route is used to realize a Multi-Protocol Label Switching Virtual Private Network (MPLS VPN), through which different types of network tasks can be isolated and the safety of dispatching tasks can be ensured. Quality of Service (QoS) strategy is employed to ensure the transmission bandwidth and real time demand of data flow for important tasks.

A three-layer structure is designed in SPDnet, which includes core layer, backbone layer and integration layer. The core layer includes state, backup, regional and some related provincial dispatching centers. The backbone layer contains most provincial dispatching center nodes, and integration layer includes directly dispatched power plants, substations and converter stations.

Several standards have been established for the purpose of power system real time data application, including DL/T 476-2012 for real time data communication among dispatching centers, DL/T 634/5 (104) for real time data communication between central station and substation, DL/T 860 7-2 (IEC 61850 compatible) for data communication within substation. In addition, IEEE C37.118 is used for WAMS/PMU dynamic data communication.

The time synchronization management system is another critical infrastructure for PMU and WAMS. In this system, Beidou Navigation Satellite (BNS) system is the main time clock source, with the accuracy of 20~100ns. Global Position System (GPS) is the auxiliary one, with the accuracy of 6~12ns and synchronous digital hierarchy (SDH) is the backup solution. A hierarchical management system is implemented for the monitoring of time synchronization status. Some Intelligent electronic devices (IEDs), such as power grid dispatch and control system, substation monitoring system and synchronized devices, run coordinately to realize the synchronization monitoring.

Different synchronization protocols have been adopted inside the substation. Network Time Protocol (NTP) should be used for the equipment in the station control level, while in bay level and process level, optical fiber pulse per second (PPS) or Inter Range Instrumentation Group B (IRIG-B) would be applied to set time. In the future, the IEEE 1588 v2 standard can also be used.

3. Recent Development and New Framework of WAMS in China

All of the national, inter-regional, provincial dispatching centers have been deployed with WAMS central stations. The total number is 39, until the end of 2013. In addition, some of the control centers in metropolitan power grids, such as Guangzhou and Shenzhen, also designed and are constructing their own WAMS.

Two comprehensive dispatching supporting platforms have been developed, i.e. D5000 in State Grid of China and OS2 in China Southern Power Grid, to integrate different systems deployed in control room, such as Energy Management System (EMS), Dynamic Security Assessment (DSA), Dispatcher Training System (DTS), etc. WAMS is one of the core sub-systems in these platforms as well. The framework of a typical WAMS is demonstrated in Figure 4. Besides the measurement terminals (PMU) and the communication network, the WAMS central station is composed by several servers, including front communication server, WEB server, historical data server, real time data server and advanced application server. Among these servers, the real time database and advanced application servers are the most key components.

The real time database techniques oriented to power system equipment have been developed based on Common Information Model (CIM) criterion, wide-area distributed storage and direct location methods. There is a substantial increase on the access efficiency than the previous system. The scope of data access is extended from local to remote area, which solves the problem of real-time data share in the entire network with high efficiency. Several techniques including fixed time duration, equal data time interval and Hash table index have been used to meet the requirements of storage and searching of massive data. The processing capability of the real-time database reaches the level of 5 million events per second.

Figure 4. The framework of WAMS

Many successful advanced applications based on PMU data or WAMS have been implemented in the D5000 or OS2 platform. Most of these applications can be categorized into two classes, i.e., model &parameter identification, real time dynamic monitoring and warning. In addition, wide area control and protection have also been researched and demonstrated based on PMU/WAMS. The major applications will be described in the following sections.

4. Power System Model and Parameter Identification and Validation

1) Load model and parameter identification

Many disturbance data recorded by PMU showed that there were significant differences between the simulation results and real system responses. Errors in the system models and parameters are the main causes, especially the load. In order to validate the simulation accuracy, four field tests of artificial three phase to ground short-circuit in 500kV substations were carried out in Northeast China Power Grid on March 25th, 2004 and March 29th, 2005. Based on the collected PMU measurement data, several frequently used load models were studied, including static load with different percentages of constant impedance, current and power (ZIP), different combinations of static load and induction motor with IEEE recommended parameters. Nevertheless, the measured curves could not be matched by using all of the above models. So, the Chinese Electric Power Research Institute (CEPRI) proposed a new synthesis load model (SLM) shown in Figure 5 (a), where the major improvement is the addition of a distribution impedance and a reactive power compensation capacitor. By using SLM, most of the field test results can be matched better, and some typical ones are shown in Figure 5 (b).

(a) Configuration of SLM (b) the active power of 500kV Yongyuan-Baojia line on March 29th, 2005

Figure 5. Comparisons between the recorded PMU data and simulations using different load models

Based on these studies, the traditional load model was replaced by SLM in the simulations and dispatching of Northeast China Power Grid after July 2006, and the power transfer limits between different provinces in this area were enhanced about 450MW without influencing the security of the real system. After 2007, SLM model and parameters were also validated by recorded PMU data after many disturbances in North China and Middle China regional power grids and some provincial grids. This model is also implemented in the Power System Analysis Software Package (PSASP), which is the most popular power system simulation software in China.

In addition, several load parameter identification and management system based on PMU and energy management system (EMS) have been put into service in Hebei, Henan, Guangdong and Fujian provincial power systems. Fujian system completed by Hohai University in 2008 is the typical one. Firstly, the loads in substations are classified according to the one-day load curve in EMS. Then the parameters for every load category are identified based on the measured data from PMU or digital fault recorder (DFR), and some parameters may also be summarized through the data collected by smart meters. Finally, the generalization of the load models and parameters will be validated and improved by using some new PMU data after contingencies.

2) Generator parameter identification

Generator parameters may vary greatly in different operating conditions. Some identification or parameter optimization techniques have been employed to realize generator parameter identification, such as generic algorithm (GA). The inputted PMU curves are the generator terminal voltage and excitation voltage, and the objective is to minimize the error between the simulated generator current and the current measured by PMU. The identified generator parameters include synchronous reactance, transient reactance, sub-transient reactance for direct axis and quadrature axis, and their corresponding time constants. The moment of inertia can also be identified.

5. Power System Dynamics Monitoring and Analysis

1) Substation state estimation

Traditional state estimation (SE) is implemented in electrical power control center on bus-oriented model using measurements remotely collected from Supervisory Control and Data Acquisition (SCADA). Identifying topology errors is a major challenge for traditional SE. Substation state estimation (SSE) provides an alternative effective way to prefilter bad data with detailed model and redundant local measurements. In SSE, it is proposed to handle the zero impedance subnetwork (with CBs as zero impedance branches) at each voltage level separately. In traditional SE, the topology has to be determined first in order to build nodal equations using Ohms Law. Instead, only the analog measurements from both PMU and SCADA are used to solve a local SE for each voltage level first using Kirchhoffs Current Law. The estimated power flows across each CB are then used to check for suspicious CB status bad data. In this process, the detection of analog and topology bad data are decoupled and less susceptible to CB status measurement errors. Topology bad data can be further identified with methods such as hypothesis test.

Due to the smaller problem size, the complete utilization of fast-sampled PMU measurements becomes possible, which could further benefit SSE in accuracy and reliability. The dynamic transformation of topology, power flow and phasor changes can be captured with PMU. Also, system three-phase imbalance can be monitored by way of substation three-phase modeling.

Through statistical analysis by way of Monte Carlo simulation, it is shown that the rates of analog and digital bad data could be significantly reduced by SSE, as shown in Figure 6. The local SSE results are then sent to control center and further enhance the reliability of global state estimation.

Figure 6. Comparison of Analog and Digital Bad Data Rate Before and After SSE

In China, this SSE method has been first implemented in four 500 kV experimental substations of the East China power system, and further applied to substations in other power companies. It is verified through field test that data quality could be greatly improved with SSE using hybrid measurements from WAMS/SCADA. The average computing time for SSE is 20ms, which is feasible for real time application.

2) Low frequency oscillation identification and assessment

Low frequency oscillation (LFO) is always a stability threat for the Chinese inter-connected power grids. A comprehensive solution for LFO stability monitoring and assessment have been deployed in some of the Chinese power dispatching centers. This system includes three functions, i.e., real-time warning of harmful oscillation based on ringdown signals identification, small oscillation early-warning by statistics, and oscillation modes and mode shapes identification based on ambient data of power systems. Thus, the weakly damped oscillation modes could be detected early, and the dispatchers could make better decisions based on the identified information.

Prony algorithm is the basic method for the oscillation detection and identification based on ringdown signals after disturbances, and has been implemented in most of the dispatching system. Some advanced algorithms, such as Hilbert-Huang transform (HHT) and estimation of signal parameters via rotational invariance techniques (ESPRIT), have also been tested in the Zhejiang and Sichuan provincial power grids.

Because disturbances in the real power systems are relative rare, oscillation identification based on ambient data is more practical for early warning and decision making. Some improved methods based on Auto-Regressive Moving Average (ARMA) model and the recursive algorithms have been studied and implemented as a key function in the data mining system in China Southern Power Grid (CSG) by Tsinghua University and Beijing Sifang Automation Company. Taking the LFO event happened on April 21st, 2008 as an example, the ambient and ringdown signals are shown in Figure 7 (a). Figure 7 (b) demonstrates the identification results of three inter-area oscillation modes using AMRA method, which can be validated by the Prony calculations based on the following ringdown data. This softwares interface is shown in Figure 8, where the oscillation frequencies and damping ratios of different oscillation modes were estimated for a time duration of one hour. With this system, 24x7 monitoring on the small signal stability of CSG can be realized.

(a)Power oscillation on 04/21/2008 in CSG (b)ID results using ARMA method

Figure 7. Power oscillation on 04/21/2008 in CSG and the ID results using ARMA method

Figure 8. LFO real-time monitoring interface in the CSG data mining system

3) Power grid disturbance identification

When disturbance occurs in power systems, the operators of control centers can received many warnings and sequence of events (SOE) information. However, these information usually only contains breaker trips and protection relay actions. The operators cannot directly know what kind of faults lead to these actions of breaker or protection relays. The action reasons usually can be obtained by analyzing the fault recording wave files after the fault events. It will be time consuming and are not benefit for fault process. Based on real-time dynamic data of PMUs, online identification methods for different kind of faults or disturbances have been designed and implemented. These fault or disturbance types that can be identified include short circuit faults, generator breaker trips, phase shift failures of DC converters, islanding and etc.

Sometimes the message of relay action cannot be transmitted to the dispatching center, due to some reasons, including communication traffic under system faults. But with the PMU measured dynamic response, the fault type and position can be derived. The dispatcher can quickly know what kind of fault has occurred on which part of the power grid. This information is quite important for the dispatcher to decide whether further measures should be taken to keep the safe operation.

4) Assessment on Generator Ancillary Services

To ensure generator safe and reliable operation, ancillary services such as speed control, excitation control are necessary. When the system frequency deviates from the nominal frequency over a threshold, the unit power output and frequency curve are analyzed to assess the contribution of primary frequency regulation. Based on active power and frequency measurement of PMU of a generator during a frequency disturbance, the primary frequency regulation performance parameters of the generator can be calculated. The parameters include delay time, response time, frequency dead zone, speed variation ratio, contribution energy and etc. Figure 9 shows the primary frequency regulation assessment result of a generator during a frequency disturbance. The monitoring of generator excitation control can also be carried out in a similar way.

Figure 9. Performance assessment on generator primary frequency regulation

Generator primary frequency regulation function is very important for power grid to maintain frequency within certain secure range while suffering some active power deficit, such as tripping of a large generator. During some tests on primary frequency regulation function, some generators that claimed this function had been switched on, were found to be switched off. This fact may deteriorate the system frequency response, and thus every generator with primary frequency regulation switched on should be monitored and assessed. In State Grid of China, all large generators are now assessed and the system frequency characteristics are improved compared with that before this assessment system is deployed.

5) Assessment on Wind Farm Performances

Nowadays some power grids in China are operating under high penetration of new energy resources, such as wind power. Thus, the integrated wind farm is required to have LVRT (low voltage ride through) ability. Generally, required duration of low voltage ride through ability is 0.625s, which is out of the time scope of SCADA capability.

With help of PMU data, due to high reporting rate, it is easier to see whether a wind farm satisfied the LVRT requirement by comparing the recorded voltage curve with the standard curve. Dispatcher in control center can easily check whether a wind farm can satisfy the LVRT requirement during a voltage disturbance, such as a short circuit near the step-up substation.

In China, many cascading tripping events occurred after large scale deployment of wind farms. For example, in 2011, at least 12 events of cascading tripping appeared, more than 500MW power was loss in each event averagely, which substantially affect the security of wind farm and connected power grid. Some cascading tripping of Doubly Fed Induction Generators (DFIG) were induced by low voltage caused by short circuit fault due to poor Low Voltage Ride Through (LVRT) capability. Thus these wind turbines are required to get upgraded. Under synchrophasor based LVRT monitoring, all the wind farms are now equipped with qualified LVRT function. Since then, no large scale cascading tripping events has ever occurred.

6) Intelligent Alarming on Cascading Tripping of Wind Turbines and Wind Farms

Wind turbine in China suffers cascading tripping induced by internal or external faults of a wind farm. When a group of wind farms locate electrically near to each other, it may develop into cascading tripping of wind farms. Utility companies are harassed by such event, however they have no access to monitoring data of wind turbines due to commercial secrecy. Using PMUs installed at the step-up substation, an intelligent alarming system can detect the occurrence of cascading tripping events, trace the entire process and assess the impact.

Cascading failures of wind farms can last for several seconds to 1 minute, accompanying with sudden voltage dips due to contingency, continuing losses of active powers and increases of voltages. On the other hand, the magnitude and speed of output power ramps of a wind farm even under the most violent fluctuation scenario, which can be 25% of the installed capacity within 5 minutes, are far below those under cascading scenarios, which can be 100% of the installed capacity within 5 seconds. The intelligent alarming system detects big voltage dips at PCC points of wind farms as start-up conditions, and identifies power ramps of wind farms on a minute-based time window. All the detected events are synthesized to get a whole picture of the cascading failures. One example is given in Figure 10-11.

Figure 10. Detection of voltage dips at the instant of faults

Figure 11. Detection of power ramps during cascading failures

6. Wide Area Control and Protection

1) Wide Area Damping Control (WADC)

In order to suppress the inter-area low frequency oscillations more efficiently, the global dynamics of different areas are required, and then PMU is becoming necessary for real time and continuous damping control. In the design of WADC system, three issues should be considered carefully, as the following:

selection of feedback signals and control sites;

controller structure, parameters and adaptiveness;

modelling and compensation of random time delay in communication system.

The WADC system commissioned in 2008 in CSG is a representative project operating in the real complicated power grid. Based on the calculations of observability and controllability, 6 PMUs located in three provinces spanning over 1000km and three HVDC links with the capacity more than 11GW were chosen to be included in this system. The controllers were coordinated to damp two dominant inter-area oscillation modes limiting the long distance transfer power. The systems configuration is shown in Figure 12.

Figure 12. Configuration of the HVDC WADC system in CSG

Some new oscillations with higher frequencies (about 5Hz) caused by time delays and the HVDC fast response characteristics were investigated, and a solution using low-pass filter was proposed. Based on the improved online Prony identification, the controller parameters can be adapted according to the changes of the oscillation frequency. This WADC system were implemented in a real time system and carefully tested in the CSG real time digital simulations (RTDS) platform, which is composed of more than 10 racks of RTDS, real HVDC control and protection cubicles. After field debugging and trial operations, this systems performances were validated through artificial block and de-block of three different HVDC links and tripping a 500kV AC tie-line in the years of 2008 and 2009. The field test results show that the commission of WCADC system can increase the damping ratio of the dominant modes from 5% to more than 15%, which means the transfer limitation enhancement of 650MW in CSG. The test results are shown in Figure 13.

Figure 13. Field test results of HVDC WADC system in CSG

In addition, the research and implementation work on wide-area generator damping control are being carried out, and the power system stabilizer (PSS) is used as the controller. This kind of Wide-area PSSs have been installed in Silin power plant in Guizhou Province and Ertan power plant in Sichuan Province respectively, both in 2013.

2) Wide Area Protection (WAP) Application

Existing protection system is a distributed control system, which mainly using local information to detect power system fault and abnormal state. To solve the coordination problem between different backup protection, wide area current differential backup protection is utilized, and fault can be correctly identified and cleared, and the time for fault clearance can be shortened. For the bus or the substation losing its power supply after the fault clearance, fast recovery can be fulfilled under WAP platform.

Duyun WAP project was implemented on seven 110kV substations in Duyun district of Guizhou province and was put into operation in Dec.2011. The communication structure takes the ring form. Architecture of Duyun WAP system is shown in Figure 14, which consists of double (one for redundancy) master and seven subsidiary control stations. Master station can analysis power grid operation mode and status, determine fault location and send corresponding commands to control stations. Subsidiary control station is responsible for acquisition of electrical and state quantities, and issues the protection and control functions that only require local-substation information.

Figure 14. Architecture of Duyun WAP system

The coordination among backup protections can be realized by using WAP system to isolate faulty component and accelerate fault clearance. The field disturbance recordings show that the fault clearance time for circuit breaker failure can be decreased from 1.0~2.0s to 0.2s with the help of WAP system. Based on this WAP platform, fast wide area backup power supply auto switching can be fulfilled.

7. Experiences and Outlook on PMU/WAMS applications in China

In the past decade, more and more PMUs and WAMSs have been put into service, and the synchrophasor technology development is focusing on the advanced applications for power grid management, stability enhancement and efficiency improvement. The application trends are heading from offline to online, and from monitoring to control. Some demonstration projects have been commissioned in China, and show great performances and potentials. However, the huge amount of data measured by PMU/WAMS is still far from full exploitation to meet the smart grid requirements. Combination with big data technology will be an important opportunity to further PMU/WAMS applications. Other further research directions include developing high-precision phasor measurement units and exploring the applications of PMU data for power distribution systems and some industrial power systems.

Acknowledgements

Further Readings

Biographies