Summary of Disturbance

The Union for the Coordination of Transmission of Electricity (UCTE) grid suffered a severe disturbance on 4 November 2006. Immediate action by TSOs prevented a Europe-wide blackout. This article summarizes the findings (http://www.ucte.org) from the subsequent UCTE investigation.

E.ON Netz was requested to disconnect the double circuit 380kV line Diele-Conneforde in Northern Germany on 5 November at 01:00. E.ON Netz's empirical analysis showed no violation of the N-1 (The N-1 security rule aims at avoiding cascading effects and requires that a single incident should not jeopardize the secure operation of the interconnected network) criterion from the proposed switching in its network. On the 27 October E.ON Netz provisionally approved the switching request and informed neighboring TenneT and RWE TSO. Results of their N-1 analysis confirmed that the grid would be highly loaded but secure.

The manual switching of the Diele-Conneforde line on the 4 November caused increased power flow on the Landesbergen-Wehrendorf line (interconnection between E.ON Netz and RWE TSO). Action taken to remedy the situation actually deteriorated line loading eventually causing it to trip. This triggered cascade tripping splitting the UCTE synchronously interconnected grid (which covers 23 countries across Europe) into 3 islands (West, North East and South East - see figure 1) causing significant power imbalances in each island. Initial power imbalances in the West induced a severe frequency drop interrupting power supply to more than 15 million West European households. Sufficient generation reserves in the under-frequency areas (West and South East) allowed restoration of normal frequency in a relatively short time.

Lack of control over generation units (quick reduction of schedules and automatic reconnection of wind generation) in the over-frequency area (North East) contributed to a deterioration of system conditions in this area (long lasting over-frequency and severe transmission line overloading) and complicated the process of restoring normal system conditions.

Full resynchronization of the 3 islands was completed 38 minutes after the system split and normal conditions were restored in all European countries in less than 2 hours.

Sequence of Events

Evolution of system conditions before system splitting

Protection settings on both sides of the Landesergen-Wehrendorf line are different (see Table 1). Despite previous settings data exchange among TSOs, E.ON Netz claimed they were unaware of the differences and did not take the actual Wehrendorf (RWE TSO substation) values into account.

Table 1: Current limit values on the line Landesbergen-Wehrendorf

E.ON Netz's empirical assessment of corrective switching measures suggested that coupling the busbars in Landesbergen substation would reduce current. They initiated this switching at 22:10 without coordination with RWE TSO. Contrary to expectations the line current increased. Wehrendorf substation (RWE TSO) protection tripped and triggered cascade tripping all over the UCTE area.

UCTE Separation
The UCTE system split into 3 areas at 22:10:28. The interconnection between Morocco and Spain tripped on low frequency at 22:10:32. Figure 2 shows the frequency in the 3 areas from 22:10:06 to 22:10:30. Time stamps of key events are indicated.

Figure 3 shows the frequency for the 3 islands from 22:09:30 to 22:20:00 are shown. The split caused a power imbalance by cutting 9,500MW power transfer from the East to the Western area which induced a frequency drop to 49Hz in the West. The corresponding power surplus in the North East increased the frequency to 51.4Hz. The split caused an 800MW power deficit in the South East area inducing under-frequency of 49.7Hz.

Resynchronization

A number of resynchronization attempts took place however the first successful reconnection between the West and North-East areas was completed at 22:47:11 on the 380kV Bechterdissen-Elsen line. The frequency and phase angle differences prior to connection were 180mHz and less than 10.

The first successful reconnection between the already reconnected West and North-East areas and the South-East was at 22:49. The resynchronization was completed when the last 400kV line between Croatia and Hungary was switched in at 23:57.

Results of Investigations

Analysis indicates 2 root causes and 6 critical factors.

Root Causes

• N-1 criterion security rule
  Switching off the 380kV Diele - Conneforde line resulted in non N-1 secure conditions in the E.ON Netz grid and on some tie-lines to neighboring TSOs. E.ON Netz relied on an empirical evaluation and did not perform a numerical analysis of N-1 criteria.

E.ON Netz coupled the busbars in the Landesbergen substation in an effort to reduce the loading on the Landesbergen-Wehrendorf line. This was done without numerical analysis of the impact of this change on the power flow. The result was contrary to E.ON Netz's empirical assessment. It caused the line to trip triggering the cascade tripping that split the grid.

• Inter TSO co-ordination
  Relevant neighboring TSOs checked N-1 compliance with the proposed 380kV Conneforde Diele line switching during the planning stage but a late switching change was not communicated in a timely manner by E.ON Netz to other TSOs.  Despite the fact that the network was highly loaded, E.ON Netz did not prepare efficient remedial action to keep a minimum safety margin and prevent possible power flow increases.

Prior to the triggering event at the Landesbergen substation E.ON Netz did not fully coordinated with TSOs. No account was taken of different protection settings on either side of the Landesbergen-Wehrendorf line (i.e. on E.ON Netz and RWE TSO systems).

Critical Factors

• Generation related issues
  A significant amount of generation (wind power, CHP) that tripped was connected to the distribution grid. Most TSOs do not have the real time data for these units. Frequency restoration requires sufficient means for rescheduling generation in individual control areas. Lack of control over generation units during the disturbance complicated the process of re-establishing normal conditions and was key to long lasting frequency deviations in the North-East and for re-synchronization failures.

• Range of possible actions for the dispatchers to handle grid congestions
  A limited number of measures are available to TSOs to manage emergency situations while also ensuring compliance with national regulations, market requirements and internal procedures. The adequacy of these measures needs further scrutiny.

• Defense and restoration plans
  In some control areas, the DSOs started the re-energization of customers without proper knowledge of the overall situation in the UCTE system and without coordination with TSOs thereby hampering attempts to restore normal system conditions.

• Resynchronization
  Actions taken by TSOs during the resynchronization process were not sufficiently coordinated. There were several unsuccessful attempts to resynchronize with only a partial view of the whole grid. Protection devices performed correctly and prevented further negative consequences.

• Training of dispatchers
  The accuracy and completeness of dispatchers training was considered particularly in relation to (i) procedures and tools; and (ii) inter-TSO co-ordination under normal and emergency conditions. It was noted that dispatchers are not always trained on incidents that originate in external networks but affect their TSO network. It is not common practice to have joint simulation training with neighboring TSOs although there are discussions on procedures and exchanges of experience.

• Communication among TSOs
  The communication among UCTE TSOs was as expected in an emergency situation given the current framework of co-operation, the scope of responsibilities of individual TSOs and on-line data available at individual control centers. However information about the actual state of the UCTE system after splitting was not quickly available to all TSOs. The lack of information made it difficult for dispatchers to quickly and efficiently identify the source of the problem.