The capability and versatility of protection hardware platforms has significantly increased over recent years opening opportunities for more centralisation and virtualisation of functions. This has been recognised and preparations for future standards, requirements are currently in preparation as part of CIGRE working groups and current innovation projects.
The proposed solution aims to take the next step in this development by delivering a demonstrator for a centralised substation platform. The solution will consist of Merging Units, process bus network and centralised PAC platform.
The project will include an assessment of current HW capability, SW architectures, development of specifications and requirements for scalable, open, cyber secure transmission substation solution, test methodology and a demonstrator for laboratory-based evaluation
Benefits
As the current TRL level for centralised PAC systems is low, benefits have been estimated based on the experience as part of the recent evolution of fully digital secondary systems. The assumptions for the benefit estimation include the following: The technology development continues, is successful and leads to the availability of centralised PAC systems from 2026 onwards. Should this be the case, a conservative estimate would provide a cost saving over the period of 2026 – 2031 of around £7M (NPV).
Learnings
Outcomes
Year 2022/2023
- Using a typical network load of 64 IEC61850 SV streams at 60 HZ, the SEAPATH configuration on the computer achieved an average packet processing latency of 200 µs and a maximum of 1300 µs.
- SEAPATH was able to schedule ten real-time applications scheduled every 1 ms each, with a maximum total jitter of 250 µs.
Year 2023/2024
- Achieving Virtualised Protection using SEAPATH as of our studies and investigations. The prototype deployed in the lab demonstrates this. The prototype included a fully functioning deployment of the SEAPATH architecture with 4 merging units connected to the process bus and capability for further digital injection of SMV and Generic Object-Oriented Substation Event (GOOSE)
The prototype has been proven to deliver deterministic tripping for overcurrent protection with a round trip time of 10ms.
Year 2024/2025:
Major outcomes of this project include the scaling up of Boundary Conditions to up to 32 streams per Virtual Machine, the Deterministic Low Latency Performance resulting in round trip times of 5ms, and the demonstration of remote management, deployment and even injection testing of virtualised protection applications.
This project also contributed to the official release of LFE SEAPATH hypervisor and reference architecture.
Recommendations for further work
Year 2024/2025:
- There are two very important steps that should be taken in future projects. The deployment and testing of a full suite of transmission protection applications and thereafter the field-based implementation of these Virtualised Protections. In the first instance, this may be carried out Open Loop, i.e. without sending trip messages and in parallel with conventional IEDs whose functionality is fixed to hardware. Next generation chipsets such as the 64 Core Sierra Forest may also be tested in future projects which may also wish to test AI applications on the edge. The execution of containers directly on a container engine may well be an interesting test case, especially to see how this moves the boundary conditions. The application of cybersecurity to the VPAC system would be a critical part of future projects.
Lessons Learnt
Year 2022/2023
From the tests carried out so far, the following lessons have been learnt:
- Optimal SEAPATH configuration is to use KVM/QEMU Virtual Machines (VMs). This provides a high level of network security.
- Real time performance can be achieved through Interrupt Request (IRQ) and application priority optimisation, as well as CPU core assignments.
- Average Packet Processing latencies are 200µs with a maximum of 1250µs.
- Each Virtual Machine was able to schedule one application per CPU Core every 100µs.
- For the network stack, Peripheral Component Interconnect (PCI) passthrough provides the best latency. If more VMs are required than available Network Interface Cards (NIC), then Single Root I/O Virtualization (SR-IOV) technology can be applied to share once physical NIC amongst multiple VMs.
- Border Peer Routing Control Filtering (BPRC) filtering can be used to reduce maximum latencies by half. Normally, all frames including non-sampled value frames are sent from the driver to the subscriber software. However, BPFC filtering is a solution offered by LINUX so that only the relevant Sampled Values message are received in the SV Subscriber.
Year 2023/2024
- Deterministic and dependable operation of an overcurrent protection function has been proven as part of a full analogue secondary injection test via merging units.
- This is the result of extensive optimisation work on the Sample Measured Values (SMV) publisher, the SMV client receiving and analyzing the SMV as well as the SMV dispatcher between SMV client and Straton protection function.
- Performance of 10ms from fault inception to MU trip output was validated.
- Deployment of other protection functions has been proven as well as a development kit for the deployment of further protection functions.
- In general, as with many innovation projects some flexibility with regards to the effort and duration for some of the tasks should be accounted for due to the innovative nature of the work.
Year 2024/2025:
- The principal lessons learnt within this project that can be carried forward to other projects are that Virtualisation Technology, using an Open Source Hypervisor, can provide the Low latency deterministic performance, scalability and remote management and testing necessary for mission critical Power System Protection with real time requirements. Secondary lessons focussed on how this can be achieved, i.e. what software architecture can guarantee the required performance. This required not just the use of a pre-emptive Linux kernel, but also the fixed pre-assignment of the optimum Central Processing Unit (CPU) and Memory resources to each Virtual IED, the acceleration of network packet communication with PCI passthrough and the real time tuning of the Virtual Machine itself (IRQ offloading, CPU isolation) and prioritisation of very important processes.
Dissemination
Year 2024/2025
Our learnings and outputs from this innovation project were shared at the following events:
- System Architectures for Virtualisation and Hardware Consolidation
CIGRE Paris 2024
- Demonstration of Real Time Low Latency
LFE Energy Brussels 2024. Public Demonstration and Presentation
Distributech 2025 Dallas Tx
CIGRE Paris 2024
- It is expected that there will a Case Study on the project published by the Linux Foundation Energy on their website and presented at their conference in September 2025.