GRID ARCHITECTURE ARTICLE 3: Laminar Coordination An Important Concept For The Energy IndustryPosted to GridIntellect, LLC – A Veteran-owned Company
image credit: Stuart McCafferty
- Aug 31, 2020 1:27 pm GMTAug 31, 2020 2:44 pm GMT
- 805 views
Laminar Coordination An Important Concept For The Energy Industry
In this third GRID ARCHITECTURE ARTICLE series, we introduce the idea of “layering” architectural structures and components. The DOE calls this “laminar decomposition” and it is fundamental in how to manage distributed hierarchical control. The EnergyIoT article series also discusses this in great detail as part of the “green cloud” services abstraction, the Digital Twin Agent, and physical OT asset distributed control.
As the energy industry transitions from a centralized control to a distributed grid, how will the grid maintain reliability? In a distributed system, how will cooperation occur to solve the problem of delivering electricity when and where it is needed? No doubt there is still some need for centralized control, but in a decentralized system, how the collaboration will occur becomes a significant concern.
The common new problem in electricity is that there needs to be systemic operational alignment of utility and non-utility owned assets to provide electricity delivery. In yesterday’s world, most if not all assets that needed to be controlled were utility-owned. Because we have so many new energy devices entering the ecosystem from third parties and customers in-front and behind-the-meter, the amount of communication and coordination is vastly different. Utility and non-utility owned assets need to work together, even though there are different priorities, different communication protocols, and different parties in charge of controlling assets. The centralized control paradigm of yesterday no longer works.
This has become a big issue outside of the electric power industry, too. Control and distributed control are a general problem that involves decision processes, data aggregation, information access and how to coordinate changes. Intelligent “Internet of Things (IoT)” assets proliferated "distributed computing" and is rapidly migrating industries away from "mainframes" to much more distributed systems that can operate on a global scale. In IT architecture terminology, we have two “patterns”; one that is a coordinated/orchestrated approach and the other that is choreographed, where we rely on the nodes to collaborate with one another.
Figure 6: Traditional vs Grid Architecture Approach Differences
In classic architectural approaches, the focus is on system of systems, grid circuits, system integration, and components. Structures are ill-defined, improvised, or are inherited from legacy systems. In PNNL’s Grid Architecture approach, the concept of grid structures is used to show that different systems have natural dependencies and interactions. PNNL calls this layered decomposition and the idea “laminar” coordination is how the different layers interoperate with one another. A good coordination framework provides us with a good basis to generate an architecture. PNNL laminar coordination model was developed to help the industry solve a broad set of problems. This framework was developed to help visualize distributed control structures, distributed grid assets and DER, transactive energy, and the coordination/interaction between those structures.
The Grid Architecture methodology for laminar decomposition is to divide problems into sub problems. A structure will emerge from this decomposition where we have a problem layer and secondary sub problem components. This implies communication within AND between other problem layers in the decomposition models. This allows you to break the Grid Architecture into "layers" of coordination that can be further broken down into sub boundaries. The coordination nodes become an abstraction of all the details within the node. The challenge here is to think about not only one type of coordination. We need to distinguish the types of coordination in different time periods such as daily, hourly fifteen minutes or even 5 minutes. There is a limit to coordination based on the time involved but certainly this approach with modern communications could work all the way to the sub five minute level as we look at getting very high levels of penetration of DER.
Another important architectural issue is that there may be hidden structures and potential redundancies or coupling that are not obvious. This can lead to disastrous results if, for instance, there are multiple controllers trying to control the same asset – an asset that is coupled to more than one control authority. The physical grid and the control structures should be viewed together to see this more clearly. Without considering this we can miss hidden coupling issues, and this hidden coupling is more common than one may think.
It may be difficult to implement some of these concepts unless we can predict and align behavior appropriately. Therefore, there is a need to align incentives as much as possible to align with good behavior. This is one of the “softer” and sometimes missed points in building a successful laminar coordination framework - we need to ensure we have proper incentives for laminar models and Grid Architecture designs, in general, to behave the way they were expected to behave.
According to PNNL some key properties of Laminar Coordination Frameworks are:
- Extensibility – the composable nature of laminar coordination domains means that a framework can be made to fit an existing grid structure, can be built out incrementally, and can be extended incrementally when grid structure changes.
- Boundary deference – the decomposition method and composability of coordination domains enables the creation of an interface wherever one is needed to accommodate a system or organizational boundary.
- Local objective support (selfish optimization) – by introducing additional objective terms at any particular coordinator node, local objectives can be integrated into the overall solution. This is a form of goal decomposition.
- Constraint fusion – by adding in constraints as needed at any coordinator node, local constraints can be accommodated in a distributed fashion.
- Scalability – since coordination signals do not need to aggregate up or down the coordination chain, no communication scalability issues arises due to depth of the coordination chain. Layered decomposition can be used to create new layers as needed if the southbound fan-out for any particular node becomes too large, thus providing structural scalability.
- Securability – the inherent form of the coordination framework and consequent coordination signal flows provides a degree of regularity that supports signature and traffic analytic security measures much more so than arbitrary networking for Transactive Energy nodes and other unstructured coordination schemes.
 PNNL, Grid Architecture Training, Dr. Jeff Taft, 2019