)
ADMS Model Validation: Thinking Outside of the Black Box: Identifying model errors for 9,000 circuits in 24 hours
As utilities deploy and expand Advanced Distribution Management Systems, model quality becomes a critical success factor—but validating that model is often treated as an in-house, manual process, or a contractor-led “black box scripted" exercise. This presentation explores a different approach: using transparent, in-house automation to validate ADMS model data across source systems, transformation logic, and the final operational model.
Exelon’s ADMS challenge: Exelon, a utility with 4 operating companies and over 10 million customers, is undertaking a project to implement the AspenTech ADMS. With a new GIS as the source of the ADMS network model, Exelon needed to find a way to quickly, and accurately, validate the network model interface from GIS to ADMS. The validation needed to identify missing or de-energizing equipment or connectivity problms within the ADMS for its 9,000+ circuits.
Solution: A two-prong approach was identified where a subset of circuits would be manually reviewed while a set of scripts were written to validate the ADMS model against the GIS source data to identify missing or de-energized equipment. This presentation will focus on the approach, challenges, results, and successes of the scripted approach to validate the ADMS network model.
The session will describe how scripting and repeatable analytics were used to move beyond traditional “stare and compare” validation methods. By extracting and comparing GIS/datamart records, ETL configuration outputs, ADMS model exports, display symbology data, and circuit connectivity, the team developed automated checks for device existence, attribute accuracy, phase/state/voltage consistency, rating alignment, circuit hierarchy, source connectivity, and disconnected model components.
This approach emphasizes explainable logic, auditable outputs, and rapid iteration. The validation process produces detailed exception reports, high-level reports, and circuit-level summaries that help GIS and ADMS teams identify missing devices, incorrect attributes, source data issues, refresh errors, symbology inconsistencies, and connectivity gaps to bring the model to go-live readiness.
