VELA is an early decision-intelligence product for virtual power plant operators. This first build is a serious product skeleton: an operational console plus a typed local decision model for scoring dispatch, bidding, reserve, curtailment, and load-shifting choices across an aggregate DER portfolio.
VELA is positioned as an intelligence layer above existing DERMS, EMS, storage controls, market bidding systems, and asset management tools. It should help operators answer: should we sell, store, shift, reserve, or curtail right now, and which assets should participate?
The interface prioritizes patterns common in VPP and energy operations software:
- Portfolio-wide asset visibility before individual device detail
- Market/program readiness and obligation coverage
- Explicit operator recommendations with rationale and constraints
- Telemetry/source health because dispatch trust depends on data freshness
- Multi-objective tradeoffs rather than pure revenue maximization
The current TypeScript model in src/decisionEngine.ts scores candidate decisions using:
- Market price and forecast confidence
- Asset readiness and telemetry quality
- Obligation fit against required MW
- Degradation cost
- Customer flexibility and comfort constraints
- Risk from market/program uncertainty
The next backend step is to move this into a service that can support model predictive control, stochastic scenarios, constraint programming, and auditable human approvals.
Early product assumptions were informed by public materials from Tesla Autobidder, Fluence Mosaic, Wartsila GEMS, EnergyHub Mercury DERMS, Leap, KrakenFlex, and Stem Athena/PowerTrack. Common themes: automated bidding, price/load/generation forecasting, DER aggregation, market access, secure APIs, fleet visibility, asset health, and human-configurable operating strategy.
npm install
npm run dev