Wattr — 100 MW Data Centre Power, Cost & Emissions Simulator

Wattr command centre 100 MW live energy simulation Dashboard first. Tune the inputs when you need to.

Deal-room simulator Show whether a 100 MW data centre can operate as a managed energy asset — not just a grid burden.

Built for investors, project sponsors, grid/planning stakeholders and data-centre energy teams. Guided mode explains the case in plain English; Expert mode keeps the assumptions, dispatch and evidence available for diligence.

Investors Grid & planning Energy teams

Saved scenarios: Scenarios live in memory; use Export JSON for a permanent copy.

Wattr explains — geography & today's weather

24-Hour Power Balance Timeline (MW)

Values are illustrative. Drag sliders to adjust and see instant updates. Hatched region = shortfall vs the 100 MW load line.

Source Classification

Storage Rules of Thumb

Storage Sizing Guide Storage needed = shortfall (MW) × duration (hours)

Why the Local Grid Matters

Connection capacity limits total power you can draw.
Local congestion can constrain imports and raise costs.
Marginal emissions of the local grid affect your carbon footprint.
Resilience to outages and extreme weather varies by location.
Community heat / power opportunities can improve overall efficiency and ESG impact.

The MIT study finds the key lever is when a data centre uses power relative to grid peaks — flexible timing and modest on-site storage can cut peak-driven system costs without new generation.

Storage State of Charge (24h)

Hourly Dispatch Table (MW)

Shortfall Coverage Calculator

Modular Starting Specifications (illustrative)

Default sizes are illustrative modular engineering assumptions for a concept tool; real project values vary by site, contracts, and weather.

Cooling & Weather Module

AI Simulation Chat local fallback

AI chat is routed through api/chat.php when available. If the backend is not configured, the local advisor still works.

AI Backend Status

Production Backend Pattern

Never call the LLM provider directly from the browser. API keys belong server-side only.
Frontend: sliders, charts, scenario state and local fallback advisor.
Backend: api/chat.php receives simulator context, adds the system prompt, calls the selected AI provider and returns plain JSON.
Security: same-origin endpoint, environment/config key, basic rate limiting and no secret in HTML.
Deployment: upload index.html and the api/ folder to the same web root.

Backend Request Contract

POST api/chat.php

{
  "message": "Simulate a hot day at 16:00 with low wind",
  "mode": "guided",
  "context": "current simulator state summary",
  "history": [{"role":"user","content":"..."}]
}

Response

{
  "ok": true,
  "provider": "anthropic|openai|offline",
  "reply": "Recommended dispatch plan..."
}

Sources used for defaults and context

MIT News, 26 Jun 2026 — How data centers can better manage energy use. Key findings: flexibility in the timing of consumption matters most; U.S. data centres run above ~80% of peak for only tens of hours a year; ~3–7% flexible load reduction during peak hours could cut system costs materially; modest storage or workload shifting covers most peak events; low-wind winter mornings and early evenings are the critical windows.
EirGrid & SONI — All-Island Resource Adequacy Assessment 2026–2035 (context for grid constraints and interconnector availability).
CRU, 2025 — Large Energy User Connection Policy Decision Paper (context for connection capacity limits and flexibility obligations).
Locality feeds for Market mode — Ireland: EirGrid/Smart Grid Dashboard CO₂ and SEMOpx price context; France: RTE/ODRÉ éCO2mix real-time mix; Sweden: Svenska kraftnät/ENTSO-E bidding-zone feeds; Texas: ERCOT real-time system and market displays. Wattr falls back to labelled assumptions when a feed or token is unavailable.

Illustrative engineering defaults for concept simulation; real project values vary by site, contracts, and weather. Emissions factors are indicative lifecycle/grid-average values, not settlement data.