As the number of large wind and solar connections increases, any potential interaction, due to the differences in their converter control system, will be an important consideration during planning and design studies.
It will be increasingly important to understand the impact of any new connection in terms of unacceptable oscillatory behaviour considering the possible sources of uncertainty (e.g., forecast errors, parameter errors) and variability (e.g., wind speed) that can affect the network condition.
This project will explore, develop, and test a combination of novel frequency domain methodologies and machine learning techniques to identify potential system operating conditions which can lead to Sub-Synchronous Oscillations (SSOs) and implement an automated control interaction studies framework.
Benefits
The project will enable a significantly improved characterisation and management of complex dynamics of the evolving GB electricity transmission system. The project will address and enhance computation time for EMT type studies and the ability to scan a wider pool of scenarios. A significant impediment to an exhaustive search of potential scenarios of concern regarding the SSO phenomenon is the inherent computational burden of EMT simulations.
In addition to this, the huge volume of data generated from the scenarios is difficult to process, navigate and analyse without an automated framework. This project will precisely provide solutions to these challenges by using advanced frequency-domain techniques. Another big challenge in root cause analysis of SSO events is proprietary controllers' 'black box' model. The project would look into techniques such as 'impedance participation factors' to represent such models as 'grey box', a technique recently developed in scientific research.
This project will build on the learnings that have already been gathered from other innovation projects in the area of EMT modelling and taking this forward into improved and automated analysis capabilities. This will allow for many scenarios and uncertainties to be captured while performing EMT types of analysis.
Learnings
Outcomes
The project has delivered a python-based tool that can automate screening of potential SSO related scenarios in system planning studies.
Key outputs include a delivery of frequency scan component that can be used with PSCAD for obtaining impedance profile of apparatus and grid and python scripts. The python scripts automate the impedance scanning process, has different assessment criteria to evaluate critical sub-synchronous oscillations (SSO) scenarios. The python script also has grey box algorithm incorporated which provides a detailed insight into the SSO issues, allows system modes identification without developing the system ‘A’ matrix analytically. This is particularly useful for practical studies as User models for converters are always ‘Blackbox’ to protect trade secrets. Currently, this algorithm is only modelled and available in this tool. Additionally, the python script has a module for automatic detection of sub-synchronous oscillations in the offline simulation results using machine learning techniques. Instead of manually sifting through hundreds of results, the machine learning model can quickly identify SSO in system variables such as voltage, currents, and power.
The python scripts have been scanned and approved by our IT security team to be able to use it within our local machines. The frequency scan component and the python scripts were validated by NESO in different environments including a wider EMT GB model with Vendor black-boxed models. A workshop was organized in March 2024 to a wider audience within NESO for demonstration of results and feedback. The workshop was a successful event in terms of the positive response.
These scripts can be particularly useful to our offline modelling team, especially the EMT team in assessing the impact of new connections in terms of their interaction with the GB network. Currently, enhancements have been identified, and the development of the tool as BAU is being discussed with DD&T team.
Lessons Learnt
The lessons learnt are:
As per the original scope of the project, a graphical user interface (GUI) was developed in WP3 phase. However, no guidelines were communicated to the partner covering the guideline with which the GUI was to be developed. Consequently, the interface was developed based on JAVA script (with web interface) which turned out to be unacceptable to NESO’s IT security team. This necessitates the requirement of IT Security to be involved in earlier stages of projects of this nature in future.
Sharing of User Models or the GB network model turned out to be extremely difficult considering NDA concerns. For future projects, such requirements can be anticipated early to minimize delay.
There was difficulty in sharing the required online measurement data or the possibility of generating synthetic data that can accurately reflect the field measurements. Consequently, it was not possible to validate the performance of the tool for its ability to automatically detect SSO scenario from the PMU measurement data.
The availability of the test systems in the required tools and format was a challenge. At the time of development and testing of the tool, there were no standard test systems available in the target software platform (PSCAD) to validate the performance of the tool. The test system that was available was on another software platform (MATLAB). Considerable effort was made to convert the test system from MATLAB to PSCAD. In future in similar projects, development of the test system in PSCAD should be considered in earlier stages.
The testing of the beta version of the tool in the NESO systems while being developed allowed for the early identification of IT and security requirements issues. This facilitated the final delivery and handover of the tools by the end of the project.
