In the development of the Strategic Research Agenda, the "Grid monitoring and data analytics" theme has been divided into two sub-themes to accommodate the wide range of research topics expected to be significant in the future. As a result, the EMN SEG strategy is here presented separately for "Grid monitoring and secure timing" and "Modelling and data analytics."

Grid monitoring and secure timing

Modelling and data analytics

Context - Grid monitoring and secure timing

Monitoring of transmission grids

The stability of the grid requires sophisticated interconnected control loops for steady-state and dynamic system monitoring. Monitoring the parameters such as frequency, voltage, and phase is therefore key in avoiding shutdowns – e.g., if production exceeds consumption, the frequency increases, and vice versa. A fine balance must thus be struck for active power, in order for the frequency to remain constant, and having sufficient system inertia greatly helps to maintain this balance. The reactive power must be similarly well balanced to keep the operating voltage constant, which is referred to as the ‘system strength’ or ‘short-circuit capacity’. Measuring the system inertia and the short-circuit capacity are both complicated tasks. The first commercial instruments have been developed and brought to market for monitoring these two parameters, but the reliability of these measurements is far from clear.

Grid control relies on a Supervisory Control And Data Acquisition (SCADA) system which measures power and voltage at key locations of the grid. Thanks to this infrastructure, energy flow in all parts of the grid can be monitored. The deployment of Global Navigation Satellite Systems (GNSS) made it possible to achieve high synchronisation of the measurements. Phasor Measurement Units (PMUs) enable an accurate measurement of voltage and current amplitude, phase difference at different nodes relative to UTC (Universal Time Coordinated), frequency, and rate-of-change-of-frequency (ROCOF). PMUs yield better state estimations with the possibility to observe long-distance oscillations with high refresh rates (up to 50 times per second compared to once every 15 minutes for SCADA). PMUs can be used to provide more information on a denser time scale, such as ROCOF, grid inertia from frequency and ROCOF, synthetic inertia, detection of sub-synchronous oscillations, fault location identification, dynamic thermal rating of overhead lines and cables, and remote instrument transformer calibration. PMUs have been deployed widely across North America, following spectacular grid blackouts [13], and now are becoming more generally introduced in the European transmission grid as well. However, verifying their measurement accuracy under all actual grid conditions is a challenging task.

Monitoring of distribution grids

The advent of renewable energy sources increases the variability and information density of distribution grids. One of the motivations of the large-scale deployment of smart meters is to provide fine geographical and time-resolved grid information to the utilities. This explosion in available data requires big data analytics in order to convert data to actionable information.

Sensors are used to measure network current, voltage and frequency, with this data aggregated to form a view of the overall network state. However, distributed generation will need distributed sensor deployments. Methods for optimising sensor placement, the use of phasor measurement units (PMUs), and state estimation using aggregated smart meter data are just a few examples that could improve network management beyond the present Supervisory Control And Data Acquisition (SCADA) systems. As an example, a 50 kV distribution network in the south-west part of the Netherlands provided data for comparing state estimation algorithms applied to PMU data with SCADA data. The grid was equipped with a SCADA system already, whereas PMUs were installed as additional monitoring devices. To what degree and to which scale the granularity of PMU installation is useful in distribution grids is the subject of further research. For shorter monitored scales, observing amplitude and phase differences between different locations requires improved voltage and current accuracy as well as a higher degree of synchronisation.

Reduced grid inertia, caused by the increased integration of grid-following renewable energy sources, requires faster responses to system instabilities. The related shorter time constants and time scales in grid control demand higher reporting rates. Dynamic phasors might not be adequate to determine the varying low inertia in the presence of disturbances such as phase steps, frequency steps, phase modulations, and so other techniques - like Hilbert Transforms - need to be investigated. Potentially, other parameters might also be necessary to guarantee system stability, such as power stability rather than inertia.

Secure timing

Conventional power grids, typically equipped with centralised Supervisory Control And Data Acquisition (SCADA) systems, receive state information from many substations only once every few seconds. Obviously, accurate timing for such measurements is not very crucial. However, future electrical grids will require real-time capable control and monitoring systems on a substation level for example exploiting PMUs to ensure stability under increasingly complex and challenging conditions. These PMUs and other digital measurement systems such as digital high-voltage sensors, stand-alone merging units, digital metering systems and digital intelligent electrical devices (IEDs) must be managed through accurate, secure and reliable time synchronisation across a wide area, both within and between substations. This has been achieved using Global Navigation Satellite Systems (GNSS) where state-of-the-art IT security technologies, supported by independent back-up systems based on other technologies like precision time protocol (PTP) and white rabbit, are necessary to protect against jamming and spoofing.

Measurement challenges

  • Monitoring the propagation of transient and disturbance phenomena
  • Novel definition of parameters in transient conditions
  • Definition of grid inertia measurement from power, frequency, and rate-of-change-of-frequency  including characterisation of dedicated equipment
  • Investigate parameters other than grid inertia to monitor and control the grid
  • Investigate the relation between short-circuit capacity and grid inertia including characterisation of dedicated equipment
  • Characterisation of the frequency response of power inverters
  • Detection of sub-synchronous oscillation
  • Secure timing protocols protecting against jamming and spoofing
  • Optimisation of sensor networks for grid monitoring
  • Development of state estimation in distribution grids and storage systems
  • Dynamic thermal rating of overhead lines and underground cables
  • Integration of power quality and phasor measurement unit (PMU) reporting rates and data stream
  • Assessing health and state of charge of storage systems


Context - Modelling and data analytics

Measurement equipment deployed by grid operators provide significant potential for the real-time monitoring of abnormal grid dynamics and post-mortem fault analysis. This results in very significant data volumes, especially for high monitoring rates up to 50 readings/s. For grid improvement and maintenance to avoid repetition of issues, there is a need for appropriate visualisation and big data analytics to convert data into actionable information.

Network operators are interested in the detection of abnormal events in response to faults or changes to system dynamics. Data analytics techniques can be used to detect anomalies and atypical behaviour in power system operation and facilitate new alarm metrics for control room staff and protection systems. An example of grid instability is the build-up of oscillations in power systems due to the increasing difficulty of convertors locking onto a stable grid frequency and their intrinsic sensitivity to abnormal events.

Hence, the need for early warning indicators based on fast data analytics. Another example is the change in grid inertia due to the relative increase of distributed generation with respect to traditional generation. Measurement and control of grid inertia is one of the most important issues facing system operators in future energy scenarios.

Measurement data from different origins, such as PMUs or other monitoring devices and even the sophisticated use of smart meter data, could therefore also be used to dynamically manage power flow in networks. As high levels of renewable energy sources and electrical vehicles are installed, parts of the grid are overloaded for short periods of supply and demand. Investing in the development of data analytics to manage power flow and rating management and consumption could result in lower investments in hardware related to rating over-dimensioning or substation reinforcement, and consequently enhance penetration of renewable energy sources.

Modelling, for instance of virtual power plants for prediction before integration in a smart grid, plays an increasing role in planning phases. Measurement data can be used to validate or improve the models predicting the influence of installation on grid stability and power quality.

Conventional grid models typically assume a constant frequency for the whole network, whereas PMU measurements indicate that the frequency differs from place to place. Furthermore, the integration of new grid monitoring equipment with higher reporting rates will require models that can deal with different reporting rates. Measurement data is crucial to validate or improve new grid models, with respect to local frequency deviations as well as varied reporting rates.

Measurement challenges in data analytics

  • Development of big data analytics and visualisation platforms with adequate evaluation of measurement uncertainty
  • Local aggregation of measurement data streams coming from different sources and instruments
  • Application-based data compression
  • Edge-computing applications to be carried out at a preliminary aggregation level (e.g., at substation level, before being sent to control room)
  • Development of machine learning algorithms for short-term load forecasting
  • Validation of new grid models characterized by reduced inertia
  • Modelling of virtual power plants interacting with smart grids
  • Definition of suitable control and quality flags for measurement data streams to be aggregated
  • Modelling of the broadband response of instrument transformer
  • Definition of a model of the frequency response of energy storage systems and batteries
  • Development of reference model for cable joints and possible faulty conditions for a prompt fault detection and location