[av_textblock size=” av-medium-font-size=” av-small-font-size=” av-mini-font-size=” font_color=” color=” id=” custom_class=” av_uid=’av-koimleaj’ admin_preview_bg=”]

NOTE TO BLOG READERS


OPAL-RT hosted a live webinar entitled ‘Digital Twin: Ensuring the Stability and Reliability of Tomorrow’s Power Grid’.

Because this content was so well-received, we have chosen to run it as a multi-part blog series.

This is part 1 of 3.



The featured host and guests were:

• Etienne Leduc, Host, Offering manager/Energy market, OPAL-RT

• Jean-Nicolas Paquin, Division manager, AXES, OPAL-RT

• Dr. Ryan Quint, Senior Manager, BPS Security and Grid Transformation | North American Electric Reliability Corporation (NERC)

• Miguel Angel Cova Acosta, Specialist Grid Modelling & Analysis | Vestas

• Tim Cervenjak, Network Modeling & Information Manager | Australian Energy Market Operator (AEMO)

• Jean Belanger, CEO/CTO, OPAL-RT



OPAL-RT’s Digital Twin landing page is here.

Jean-Nicolas Paquin, Speaker | AXES @ OPAL-RT, Context for the Adoption of Digital Twins and Related Challenges 

Evolution of Power Systems & Simulation Tools to Assess Dynamic Security 

We can confidently say that simulation technologies have evolved alongside computing technologies. And there has always been a need for simulation tools to evolve alongside power system technologies as well.

On a simplified timeline such as the one I am showing here, we can see the evolution and purposes of EMT simulation, HIL testing and Phasor simulation: 

• EMT offline simulation for fast transient analysis–and nowadays, more and more for system studies with complex control systems
• Digital Real-time EMT simulation has emerged for HIL testing of complex control and protection testing of FACTS and HVDC control replicas
• Still today, Phasor type simulation is widely used for dynamic security assessment and planning studies, but experts are starting to see limitations and are looking towards EMT simulation to overcome these issues.

And part of this is because phasor simulation assumes the grid has large inertia with mostly conventional rotating generators. It also assumes the system is perfectly balanced if it is a positive sequence simulation tool. These assumptions may no longer apply with higher penetration of Inverter Based Resources (or IBRs) to assess power system stability. 

There definitely are opportunities to adopt new ways of analyzing grid security, and Digital Twins may be a solution.

The Complexity of the Power System with Increased Communications 

In modern power systems, we see vastly more interactions between (faster) local control and protection systems and coordination with wide-area control and protections through complex communication systems. 

We also see more IBRs within centralized power plants at the transmission level–but they are also found throughout the distribution system, and all these sources will decrease total inertia as a result. Reaction times of the connected sources will be reduced. Since they require faster control and protection functions that will interact with the rest of the system, it can seriously affect stability. 

So here is a definition of a Digital Twin that could help us to analyze the complex modern power system: 

A Digital Twin is a virtual representation of the system to better understand and predict its behavior. 

Ideal/Key Attributes of a Power System Digital Twin 

So, what might be some of the ideal attributes or features of a power system digital twin? Here we suggest three main categories which are Real-Time Adaptability, Observability, and Predictability. 

• Real-Time adaptability implies the DT is synchronized and connected to the grid. It should be able to adapt based on current operating conditions in real-time or near real-time. It should or could adjust its internal models based on parameter estimation methods. 
• The Observability attribute is to say that having a digital replica of the real system can allow better assessment of all system states that monitoring alone cannot offer. The observed variables on the digital twins can be analyzed in greater detail, with higher sampling rates, and with less of the traffic and latency limitations of the monitoring and communication infrastructure. 
• Predictability, because a DT should continuously assess the most critical potential fault scenarios and help prevent system collapse. As an example, it could assess multiple scenarios based on actual operating conditions every 5 to 10 minutes.

Example for Power System Digital Twin Components

And here is an example of a power system Digital Twin with these attributes that I discuss above. 

We can see it has real-time adaptability, and since it is synchronized and connected to the power system, it adapts to the changing conditions and includes parameter tuning. 

Here, the Observability and Predictability would be ensured mainly thanks to real-time OR faster than real-time simulation, on the cloud or on a high-performance computer. The DT provides additional data to the operator through multiple scenario assessments, which will help in decision-making. 

This may seem difficult to realize, especially for a digital twin that would use detailed EMT models, but the reality is that processing and simulation technologies are now available to deploy this concept.

Challenges Related to Modeling Tools and Model Data Management

But model fidelity remains a very important challenge. 

Some utilities have limited amounts of data from the OEMs. The current practice is to use generic models in the phasor domain, and reliability councils such as NERC are enforcing rules that require model validation through on-site testing. These rules may have significantly improved dynamic assessment with conventional generator models, but the proposed generic models in phasor mode may not necessarily represent power electronics controls and protections properly when it comes to IBRs. 

In certain cases, manufacturers are requested by utilities to provide black box models using the real code of their controllers, but these models are currently not interoperable from one simulation tool to another. 

So, there may be a need to propose new rules regarding OEM models and even adopt new standards for accepted and interoperable models that may be used for grid security assessment. 

Moreover, these models may come in various versions with varying sets of parameters. There is no guarantee that the settings of the model are always the same as those of the power plant at all times. There can also be different departments of the same utility working on system studies, and they need to use the same models in their adequate contexts. So, the use of centralized model databases within utilities should be considered. 

Challenges Related to Dynamic Security Assessment 

Another important challenge I mentioned earlier is the questionability of current simulation techniques used for security assessment. Phasor simulation alone has major limitations when it comes to simulating systems with high penetration of IBRs, and EMT is gaining much attention. 

So, it is important to consider a change towards EMT simulation or a combination of phasor and EMT simulations. Using real-time simulation technology which allows both simulation domains seem like a promising solution. 

There are also challenges related to testing coverage and contingency analysis, which I will discuss using the following illustration.

Concept of AI-Based Dynamic Security Assessment using Power System Digital Twins

When it comes to the simulation of hundreds or thousands of scenarios, especially when adding EMT simulation in the mix, HPCs and cloud computing may help in deploying parallel simulation runs, for instance, but if there is no intelligent selection of the scenarios, we are still looking at a compromise between test execution speed and available computing resources. 

So there clearly is a major opportunity here to start looking at Artificial Intelligence to optimize the simulation runs and identify the most critical cases.

[TO BE CONTINUED]