Campus or Military Base Microgrid Operations Automation

Manual islanding during a utility outage is a high-risk, time-sensitive operation prone to human error, potentially causing equipment damage or extended blackouts. A custom workflow automates the entire sequence: detecting the grid disturbance, verifying safe separation points via SCADA, sequencing the disconnection of breakers, stabilizing local frequency with battery storage, and initiating prioritized load shedding. This ensures seamless transition to islanded mode in seconds, guaranteeing uptime for mission-critical loads like data centers, labs, or command posts.

Operating a microgrid with solar, CHP, and storage involves constant trade-offs between buying grid power, running generators, and using stored energy. A custom agent continuously ingests real-time utility tariffs, fuel costs, carbon intensity signals, and generation forecasts. It runs multi-objective optimization to dispatch assets, deciding minute-by-minute whether to import, export, store, or generate on-site. This autonomous economic dispatch typically captures 15-25% in annual energy cost savings and reduces carbon footprint by maximizing clean asset utilization.

A campus microgrid's aggregated storage and flexible load are valuable grid assets. A custom workflow integrates with utility or ISO market signals (e.g., for demand response, frequency regulation) to autonomously bid capacity and execute service provision. The system models battery state-of-charge and load flexibility constraints, submits offers, and upon acceptance, dispatches the assets without operator intervention. This creates a direct revenue stream, turning the microgrid from a cost center into a profit center, with full audit trails for settlement.

Managing a microgrid requires constant monitoring of dozens of data points and manual adjustment of setpoints. A custom orchestration layer built with frameworks like LangGraph or custom agents acts as a virtual operator. It handles routine monitoring, alarm triage, log keeping, report generation, and standard dispatch decisions. Engineering staff are escalated to only for true exceptions or strategic overrides, allowing a single operator to manage multiple microgrids or focus on higher-value planning and optimization tasks.

Operating in both grid-connected and islanded modes involves strict compliance with utility interconnection agreements (IEEE 1547), emissions regulations for backup generators, and internal energy transaction policies. A custom workflow embeds these rules directly into the decision logic. It automatically prevents non-compliant export, ensures generator run-time limits are not exceeded, generates audit-ready logs for emissions reporting, and validates all internal billing transactions between campus departments, eliminating compliance risk and administrative overhead.

A one-off, scripted automation solution is brittle. A custom workflow architecture is designed as a scalable, modular platform. It uses a digital twin for safe simulation and planning, an agentic orchestration layer for decision-making, and robust APIs to integrate with existing building management systems (BMS), SCADA, and financial software. This allows the microgrid to easily incorporate new assets (like EV fleets or additional storage), adapt to changing tariffs, and serve as a blueprint for replicating resilient, automated energy operations across multiple campuses or bases.

This core agent ingests second-by-second telemetry from solar arrays, combined heat and power (CHP) units, and battery storage. Using a reinforcement learning model, it calculates optimal setpoints to match load within the microgrid's electrical island, prioritizing either lowest cost or lowest carbon based on the facility's operating mode. It directly issues commands via the DER management system (DERMS) API.

Monitors utility grid voltage and frequency via protective relays. Upon detecting an outage beyond preset thresholds, this agent executes a sequenced islanding protocol: it commands the Point of Common Coupling (PCC) breaker open, stabilizes local frequency using battery inverters, and initiates a prioritized load-shedding routine to keep essential loads (e.g., command centers, hospitals) online. It manages the entire black-start and re-synchronization process.

Pulls hyper-local weather forecasts and wholesale energy market prices (e.g., from CAISO, PJM). It predicts on-site solar generation for the next 24-48 hours and models battery state-of-charge trajectories. Based on price signals and facility mode, it submits automated bids for demand response programs or decides when to import/export energy to the main grid, with all transactions logged for settlement.

Continuously profiles the facility's load, classifying circuits into tiers (critical, shiftable, deferrable). Using non-intrusive load monitoring (NILM) and building management system (BMS) data, it predicts load spikes. During islanded operation or high-price events, it autonomously sheds or defers non-critical loads (e.g., HVAC setbacks, deferred EV charging) and coordinates with the DER Dispatch Agent to maintain stability.

A guardian agent that runs continuous simulations in a digital twin to validate every autonomous action before execution. It enforces safety interlocks (e.g., prevents islanding if generator fuel is low), monitors for cyber anomalies, and maintains a full audit trail of decisions, setpoints, and market bids for regulatory and military compliance reviews. All actions require its runtime approval.

The single pane of glass for facility energy managers. This interface displays real-time system state, agent recommendations, and projected economics. It provides override controls, allows mode switching (e.g., from 'Cost Optimization' to 'Maximum Resilience'), and surfaces exception alerts requiring human judgment. All agent actions are executed under a 'supervised autonomy' model where critical changes require one-click operator approval.