AgenticGrid.ai: Essential energy case study

Chief investigators

• Markus Wagner, Monash University (Lead)
• Phil Beadle, GridGuru Solutions

Purpose of project

• Remote and edge-of-grid feeders like Tibooburra and Ivanhoe carry real voltage, reliability and cost-to-serve pressure, yet are the hardest parts of the network to model — conventional planning studies are slow and expensive to run for small, isolated townships.
• This project proves that AI-assisted digital-twin modelling can build a feeder-level network model of these townships from DNSP-supplied topology and consumption data, and produce indicative sizing for battery storage, solar PV and back-up generation — faster and at lower cost than a conventional study.
• It rides on the success of the earlier AgenticGrid.ai project, “Breaking the barriers with AI-assisted energy modelling”, extending the method from proof-of-concept to a named, real-network case study inside Essential Energy’s distribution area.
• A limited set of seasonal and stress scenarios is run to demonstrate feasibility and sensitivity — this is an indicative proof of concept, not a bankable engineering design or a full RIT-D.

Impact of project

• Demonstrates a repeatable, low-cost method to build feeder digital twins directly from a DNSP’s own topology + interval-consumption data.
• De-risks investment decisions for off-grid / edge-of-grid communities by quantifying the BESS / solar-PV / back-up-generation trade-offs under real load.
• Gives Essential Energy a transferable template that can be re-pointed at the many other remote feeders across its network.
• Strengthens the regulatory and cost-benefit case for non-network solutions, aligned to AER CBA Guidelines — a stepping stone toward RIT-D-grade assessments.
• Further identifies grid-optimisation opportunities at the distribution level.

Project partners