Detecting wide-area oscillation modes in electrical power transmission

Following several serious power blackout incidents in 2003, policy makers in the European Union and the USA have highlighted the importance of emerging measurement-based technology towards achieving more stable operation of a.c. transmission grids.

This project used the emerging measurement technologies of high-bandwidth SCADA systems and the wide area measurement systems (WAMS) which are based on time-synchronized phasor measurement units.

The university researchers have completed secondments with National Grid and ABB.

Results to date have shown:

  • Spectral Independent Component Analysis (ICA) is a promising method for detection of the presence of wide area modes, even when they are present at a low level.
  • An ICA significance index localizes the modes and shows where in the network an oscillation is strongest.
  • The Random Decrement method for impulse response estimation provides good estimates for the damping of the system.

Involved departments at Imperial

Chemical Engineering, Electrical Engineering.

Project sponsor

EPSRC. Research partners: ABB, National Grid, Fingrid, Helsinki University of Technology.

Publications

Final project report

The research conducted in the project has addressed the enhancement of grid stability and security in a.c. power transmission systems. In this context, grid security means uninterrupted power transmission. The work is of significance because of challenges posed by renewable generation from new wind farms and projected changes in the grid infrastructure in the UK.

The idea was to use the emerging measurement technology of wide area measurement systems (WAMS) based on time-synchronized phasor measurement units (PMUs). The study used methods for near real-time analysis. It included multivariate methods whereby time trends of the measurements from different locations were analysed together to give information about the geographical distribution of emerging stability problems as well as their evolution over time.

The outcome was a method for early detection of a type of instability called a wide-area oscillation using WAMS data from ambient operation (i.e. during everyday routine operation). It finds the frequency and damping ratio of the oscillation. If the damping becomes too small then grid operators have to take action, hence early warning is of benefit. The PhD student engineered a software tool and interface which displays the time evolution of damping and frequency of the oscillating mode, and locates where in the grid the mode is most strongly present.

The project adapted methods of multivariate signal analysis from the process sector (oil, gas and chemicals) where measurement-based analysis has been long established. Methods from structural engineering for detecting vibrations in buildings were also useful.

Estimation of system damping during ambient operation is a significant advance. Previously, damping and mode frequency have had to wait for the onset of a system fault or event which would initiate a clear transient oscillatory response called a ring-down.

The problem with waiting for ring-downs is that they are rare and may only occur after a grid problem has already arisen. Alternatively, oscillation frequency and damping have been estimated off-line from models, however models are becoming less accurate because of uncertainties in renewable generation now linked to the grid.

The PhD project student undertook a three month internship with National Grid as part of his PhD training and also tested his methods with measurements from the Finnish section of the Scandinavian transmission grid. Therefore his academic research training has been complemented by industrial experience.

The results of the project have been disseminated via journal publications and a PhD thesis. An international workshop was held at Imperial in April 2010 organized by Dr Pal.

New collaborations have developed. Aalto University in Finland and Fingrid, the Finnish grid transmission operator, have joined Imperial, National Grid and ABB to publish a study of the performance of the developed method using data from the Finnish part of the Scandinavian grid and also from simulation (where the damping was known). Further contacts include Statnett in Norway who operate the Norwegian part of the Scandinavian transmission grid. The expanded group made a successful bid for an FP7 project called REAL-SMART which is coordinated by Imperial College London (Thornhill and Pal).

The project is a €1.1M FP7 Marie Curie project which started in September 2010. The partners are Aalto University, ABB, Fingrid, GE Research, National Grid, Imperial, Statnett and the Technical University of Graz. The new project concerns mo nitoring and management of hi gh vo ltage transmission grids.

The investigators thank EPSRC for funding the WAMS project which made possible all the above activity.