Power flow models in interconnection analysis
Applications, accuracy, and access
Introduction
The quality of any power flow study is determined by the inputs to the analysis. Most critical to these inputs is the power flow model, which captures a topological and parametric state of all or part(s) of a power grid. ISOs and utilities must maintain models of their respective systems, and often these models are combined for broader system-wide studies, including generator interconnection studies.
How are these models developed and used, who has access to them, and what are their limitations? In this article we’ll answer these questions and explore the past, present, and future of power flow models, and what considerations could be given when using these data sets.
Overview of power flow models for interconnection
Models for power flow-based analysis of electric grids have existed in some form for decades. System topology and device parameters were initially stored and leveraged via manual or (by comparison to today) simplistic computational processes. Over time, software tools have been developed to improve study accuracy and efficiency, as well as to make the process of maintaining, extending, and evaluating models more feasible.
As discussed in our previous article on power flow, current power flow models depict a state or operating condition of the grid at a single snapshot at some point in time, and often a particular state of the grid that reflects a user’s intended use cases. Seasonal models maintained by ISOs and utilities make a best guess at the future state of the system with various planned changes applied in accordance with their transmission expansion plans, including transmission facility build-out, load increases, topology changes, and approved changes in the generation mix. Further, these models are developed in an attempt to capture a hypothetical operational edge condition, or a worst-case seasonal scenario representing the system several years into the future. Interconnection studies commonly use models representing future-year (e.g., 1 year out, 5 years out, etc.) peak summer loading conditions, peak winter loading conditions, and light loading conditions.
In interconnection analysis, these models are extended further to include, and account for, generation projects seeking interconnection onto the grid. A user would take the ISO and utility models (i.e., seasonal base cases), apply changes representative of further generation and topology changes, solve the power flow models to establish these new post-generation states, and use the models for subsequent analysis. A prior Knowledge Center article gives an overview of the variety of analysis performed in an interconnection study.
Challenges in building accurate interconnection models
As mentioned above, interconnection studies are reliant on what is effectively a new base case operating point where changes to the generation mix and system topology have been made. These changes can be substantial, including upwards of several gigawatts of generation change being balanced out across a portion of the system, for example. Achieving this operating state is easier said than done, and the step presents a significant computational challenge, often warranting manual intervention by engineering teams to achieve a solved state. We’d also note that this effort is simply to establish a base reference point for analysis – actually running analysis can incur its own set of power flow solution challenges. Pearl Street develops software with robust power flow solving capabilities, and is deployed commercially for automating and significantly expediting workflows for this particular application. Linked here is a recent case study highlighting this capability.
It should be acknowledged that power system models have historically been extremely complex and difficult to maintain, and this complexity grows considerably when short-horizon studies (like interconnection studies) incorporate substantial changes to models. Numerical solutions aside, incorporation of prior-approved mitigations, the presence of data errors, and a lack of clarity on allowable changes in study contexts present a unique set of challenges in interconnection study model development. Studies require a considerable amount of data to be performed accurately so as to avoid unnecessary costs to generation developers. Care is taken by ISOs and utilities to avoid these issues, but they often exist and impact study results and timelines. Below we provide a listing of a few examples of modeling data concerns that frequently find their way into interconnection studies:
Studies necessitate large amounts of data to track and maintain
New projects, old projects that haven’t yet come on-line, mitigations that might (or might not) be built, etc.
Data accuracy for new projects may be lacking due to a project life cycle
Developers may not have a detailed design if the project is far away from coming on-line (see our previous article on project development), but accurate inverter characteristics and transformer and lead line impedances are critical in modeling a viable project
Lingering data errors might exist in grid planning models for years
Could cause solution issues under particular modeling assumptions
Coordination with external parties to develop models is a manual process
Transmission owners in the footprint, neighboring ISOs, etc.
In rare instances, issues in other parts of the system that are not relevant to the study may influence convergence
Transmission owner permissions and historical information often passed to external parties via word-of-mouth or email
Uncertainty in what projects/mitigations from prior studies are “real” because dropouts may still happen
We’d like to mention here that the operational edges of models mentioned above are not necessarily the true limits of stressed conditions that would still allow for an operational and reliable grid, or that reflect actual states the grid has demonstrated. The scenarios depicted by base case power flow models are determined based on projections and judgment by planning teams according to accepted criteria. Obtaining a true operational edge case requires robust evaluation by power flow software and additional criteria development by ISOs and utilities.
Model limitations
British statistician George Box once said, “All models are wrong, but some are useful.” This quote can apply to nearly any mathematical model, since it is commonly practical to exchange some degrees of mathematical and physical accuracy to get a “good-enough” depiction of reality. This is certainly true in power system modeling intended to capture the behavior of a complex system like an electric grid.
Simplifications in power system modeling generally apply for any system state, including in those models developed for interconnection studies. Simplifications in power system analysis are also not unique to power flow models. Various time horizons require different types of models and assumptions, and complexity reductions are largely motivated by computational complexity. Examples of time horizons that are evaluated include everything from power flow (i.e., no dynamic or time-domain consideration) to influences of lightning (e.g., time horizons that could capture wave propagation caused by a lightning strike). A variety of software tools exist to study phenomena that might be “simplified away” in a power flow study or model.
While we will not get into the details of the nuances of power system analysis software, a few examples of these simplifications include exclusion of details in distribution networks and device simplifications (e.g., load composition and behavior, generator component actions, variation in line and transformer operation and parameters). These simplifications improve computational speed while giving generally accurate results for the purposes of the time horizon of power flow, but they may mask grid reliability issues that are only observed when more modeling details are included (such as the transient characteristics of inverters).
Data access
Naturally, not just anyone can access power system models and data. This data is considered extremely sensitive, and with good reason. In the wrong hands, the data could potentially be leveraged to cause substantial damage to critical infrastructure by domestic or foreign attackers. Access to the data is controlled via processes implemented and maintained by ISOs, utilities, and regional and government entities in accordance with FERC rules on Critical Energy Infrastructure Information (CEII). Processes to access data often vary by entity and also the arrangement of work being done, such as studies performed by a third-party consultant. Data is usually provided directly by the ISO or utility via secure file transfer. Access to the data requires signing non-disclosure agreements (NDAs), and may also require monetary payments or background checks. Further, the types of data made available for particular study purposes and access to certain data sets can vary still. Anyone can explore these model request processes online via ISO and utility websites.
Takeaways
Today, building good power flow models is a complex, manual process
All the more reason why ISOs and utilities should strive to make the process formulaic, repeatable, and automated
Models are generally good, but have limitations
Can be either imposed by computational complexities or criteria
Access and accuracy of this data has limits
It is important to understand these bounds for analysis purposes
Questions?
If you enjoyed this article and want to learn more about some of the things we’re working on at Pearl Street, reach out to us at hello@pearlstreettechnologies.com. Thanks for reading!