Introduction
NESO have been pushing for the majority of developers to carry out Sub Synchronous Oscillation (SSO) studies for all new generation systems connected to the network. This has always been the case for HVDC links, but now includes, Storage, Wind and Solar PV. Note this only applies to sites that fall under the Grid Code – not ENA G99 generation.
The reason for this is that NESO (particularly in the Scottish area) have been experiencing an increase number of SSO events, that has lead to system disturbances and loss of generation. It is a present and growing problem, that unfortunately does need addressing.
One of the issues (that we have raised frequently) is that the requirements are not well defined, and can lead to a number of headaches and delays to a project of not handled correctly. The important part to note for the SSO requirement is that the simulations have to be carried out as an EMT (PSCAD) analysis, – and this can only be started once the main EMT model is completed and accepted (or nearly accepted) by NESO.
The requirements in the Grid Code are somewhat sketchy, and hidden away from the main sections, with a few single lines in section ECC 6.1.9 and ECC 6.1.10; these requirements are brief and would be missed unless you are looking for them. More commonly there is a requirement often detailed in Appendix F5 of any site specific requirements.
There is also the NESO document “System Oscillation Assessment of Inverter Based Resources (IBRs)” (Jan 2024). There is a new version of this document coming out in May / June 2025, as NESO have had a lot of push back on this document.
Some other documents well worth a read are:
- Stability definitions and characterization of dynamic behavior in systems with high penetration of power electronic interfaced technologies, IEEE TR-77 (2020)
- CIGRE Technical Brochure 909 Guidelines for Sub synchronous Oscillation Studies in Power Electronics Dominated Power Systems (June 2023)
- ESIG – Diagnosis and Mitigation of Observed Oscillations in IBR-Dominant Power Systems (August 2024)
SSO Analysis – Theory & Practice
The presence of SSOs on power system networks has been long studied in power systems analysis, and has been a relatively well understood field since the 1970s following a few major incidents. Analysis was typically carried out using a process called Eigenvalue analysis, which was used to solve all the differential calculations in a model and identify any poles and zeros. These were then calculated to determine any stable, marginally stable or unstable operating points.
In simple terms, the underpinning assumptions used in Eigenvalue analysis is that the state space models of all the control systems are defined and that the poles of the systems can be calculated. Eigenvalue analysis is a computationally simple process based on a linearized operating point, and relatively easy process to carry out, once the theory is understood. It is simple to perform on large networks and can be easily configured for different operating cases and conditions. However, recent system disturbances has led to a loss in confidence in a pure Eigenvalue approach, as they cannot be represented in state space format.
When IBR generation is a significant proportion of the network, and the IBR models are encrypted / black boxed, the Eigenvalue approach begins to break down as the technique cannot accurately identify all the possible modes on the system. The reasons for this are complex, but one of the main contributory factors are that
- Control systems are much faster and more complex than traditional synchronous machines (ok maybe not more complex)
- There are a lot more generation and plants generally
- RMS models are necessarily simplified compared to reality
- Most RMS models are encrypted rendering them unsuitable for Eigenvalue based techniques.
Consequently, it has been found that traditional Eigenvalue analysis was not working as well as hoped, and there has been a general move towards use of EMT type tools to carry out analysis. Such EMT approaches have been shown to be robust in many applications, but have inherent practical and logistical difficulties in scaling up in large networks, due to the computational time and complexity of modelling systems in EMT.
As the inverter model is ‘black boxed’ the only way to fully define its behavior is through a combination of system impedance scans to determine its impedance and network perturbations to determine its response. This is what NESO are trying to achieve with their methodology. The idea is that an EMT model is the most accurate model available of a system, and whilst it is likely to be encrypted, its response to SSOs can be understood by applying a series of perturbations to it, in order to determine its response to differing frequencies. Incidentally, this is what control engineers do on very complex systems, that cannot be reduced to a state space model.
The output is a series of tables / graphs that show the IBR response for the various perturbations. If the system is totally passive all the output signals should be less / not amplified by the input signals. Where an amplification is seen, this is indicative of a resonance inside the generator system. The perturbations are carried out for all main control modes (frequency response, voltage control and at various plant loadings and power factors).
A fundamental weakness of the approach by NESO is the assumption that any resonance from a system is unsatisfactory and must be tuned out. This is practically very difficult. Almost all systems, will have some form of resonance due to the RLC components within the power system. The resonance only becomes an issue when its resonant point coincides with a natural mode already present on the NESO system. Traditionally these modes were around 0.1 Hz to 7 Hz, but more modern SSOs have been seen up to 20 Hz. However, NESO do not currently make these results generally available.
A more serious general issues with the NESO approach, is that for an EMT based SSO analysis to work, you need to have a really good model of the whole network in EMT. Which is something NESO don’t have – this is why there is an ongoing push for all new sites to provide an EMT model (GC0141) and a retrospective push for existing sites to provide an EMT model (GC0168).
What Happens if an SSO is Encountered?
Conventional rotating machines used to deal with this problem using a Power System Stabilizer (PSS). There are currently no standard models of this kind of control system available for IBR. Therefore the only solution is to retune the control systems causing the oscillation to try and address the issue – in some cases this can be simple, in other cases less so, as the control system will have other performance requirements such as response time, overshoot and settling time.
Practically this means that some / all of the control systems will need to be re-tuned and possibly reconfigured. This is a time consuming and potentially tedious task – and once complete, it will mean that all the Grid Code studies (and EMT model), will need to be rerun and updated.
Summary
SSO studies are hear to stay, and are likely to be a significant delay and obstacle in achieving an ION for most developers. The problem with an SSO study, is that it can only be started once the main EMT model has been completed, and there is no way of predicting if a plant will cause a problem or not. If a plant does cause a problem, there is no easy solution other than to begin re-tuning the relevant controllers and then rerunning all the studies.
In simple terms this is going to be a headache for the foreseeable future, and is likely to lead to project delays in achieving an ION. Make sure you start your studies early and use a good consultant!