This project will review the current methods of calculating system stability needs and implement automation and machine learning to calculate system stability needs for the GB network at a granular level. This project will:
- Review the current methods of calculating system stability needs and identify areas of improvement.
- Perform the analysis on a sufficiently granular representation of the active and passive network components in the GB system.
- Apply automation and other necessary methods (machine learning) to manage additional computational burden of using detailed network representation.
Benefits
The project will directly facilitate the energy transition as it will improve the calculation of system requirements in terms of stability; this an essential parameter to allow the integration of renewable generation onto the GB network. By understanding more accurately stability needs, it is possible to anticipate the technical solutions that will guarantee the system operates in a reliable and safe way, facilitating the transition to a zero carbon energy system while keeping the lights on.
Learnings
Outcomes
The WP1 report provides a summary of the current methods for calculating inertia, short circuit, and voltage management requirements in the GB system. To extend these methods to a scenario-based year-round analysis, a review of different indices is included in the report. These indices are available in literature for different applications, and not every index is appropriate for the purpose of the project. To aide in the selection of a subset of indices that are most suitable, a qualitative comparison is included in the report along with suggestions for further implementation in WP2. Furthermore, the report covers the fundamentals of unsupervised machine learning methods such as different clustering algorithms. These algorithms can be used in an innovative way to resolve the problem of non-convergence of load flow solutions and is pivotal to the success of this project.
WP2 report will cover the development of the power factory automation framework to interface with the detailed GB system model. This framework will allow ESO to study half-hourly scenarios using data from the market dispatch simulator for system needs assessment. The details of the framework along with the machine learning approach for load flow convergence, the different indices for system needs calculation and a scoring system for final index selection will be included in this report.
Some of the key deliverables through WP1 & WP2 are a comprehensive literature review of indices for inertia and short circuit calculation, a DIgSILENT power factory automation framework, a machine learning approach to resolve network convergence issues, voltage profiling algorithm, and system needs assessment for half-hourly scenarios.
A paper on the initial work on the machine learning algorithm for network convergence is accepted for presentation in CIGRE Paris 2024.
Lessons Learnt
Data sharing especially confidential or sensitive data sharing between NGESO and NIA project partner would be paid more attention for proper arrangement and timeline plan.
Set a clearer and more practical procedure and policy to facilitate data sharing between NGESO and external project partner including security measurement & check, data sharing/storage facility and relative NDA, data protection policy etc.
Work to improve appropriate data classification to allow more useful data could be shared with project partner.
