GridMind AI — AI-Managed Energy: a Modeled 15-30% Saving

The problem: manual energy management can't keep up

A factory with Solar + BESS in 2026 faces three layers of complexity:

1. Complex TOU tariff — Peak / Off-Peak / Holiday — changing by time and season
2. Volatile solar generation — clouds, rain, seasons — changing every 5 minutes
3. Unsteady load — depends on shift, production, weather

The result: a person managing by hand can't keep up — using a simple rule-based approach (charge off-peak, discharge peak) captures only 30-40% of the real potential.

---

What does GridMind AI do?

GridMind AI is an edge-native AI platform — running on hardware at the real site, not the cloud — latency <50ms — without needing constant internet.

8 core functions

1. AI Load Forecasting

• Forecasts load 24-48 hours ahead
• Uses historical data + weather + calendar (holidays, seasons)
• Accuracy: MAPE 5-10%

2. AI Solar Forecasting

• Forecasts solar generation every 5 minutes
• Uses sensor GHI + Open-Meteo weather API + cloud cover
• Accuracy: MAPE 15-20% (now-cast), 25% (+1h)

3. Smart Dispatch

• Charges/discharges the BESS automatically by real TOU tariff
• Optimizes long-term (not just now)
• Protects against battery degradation (DoD limit, C-rate limit)

4. Peak Shaving

• Cuts the Demand Charge — a factory's highest cost item
• Auto-detects the demand peak before it happens and trims it with the BESS
• Reduces the Demand Charge 15-25% in the modeled case

5. TOU Shifting

• Shifts load from Peak → Off-Peak using the BESS
• Saves the Off-Peak vs Peak rate difference (typically 4-5 THB/kWh apart)

6. Solar Clipping Recovery

• When solar exceeds inverter capacity (clipping) — normally this energy is wasted
• GridMind AI pulls it into the BESS via DC Coupling — gaining 3-5% more energy per year

7. Generator Fuel Optimizer

• For PV-Diesel hybrid microgrids
• The AI calculates when to run the generator vs use the battery
• Cuts diesel by 60-80% vs diesel-only

8. Automated Carbon Reporting

• Calculates CO₂ saved automatically
• Reports tCO2e on TGO/IEA emission factors, ready for CBAM and T-VER
• Ready to export PDF / CSV — usable in sustainability reporting immediately

---

What the impact can look like — an illustrative example

Food factory, 1 MWp PV + 500 kWh BESS (modeled scenario)

The figures below are a modeled example for a typical Thai food factory on a TOU tariff — not a specific installed site — to illustrate the kind of impact forecast-driven dispatch can have. Actual results vary by tariff, load profile, and system size:

---

GridMind AI vs a rule-based EMS

---

Does it work with other brands' equipment?

GridMind AI is vendor-agnostic — it works with:

• PCS / inverters from LEONICS, SMA, Sungrow, Huawei, ABB
• BMS from CATL, BYD, LG, Samsung, Panasonic
• Solar inverters from Huawei, SMA, Sungrow, GoodWe
• Smart meters from Schneider, ABB, Siemens

It installs as an edge controller on top of your existing system — no hardware replacement needed.

---

2 models of GridMind AI

Model A — Edge Controller (retrofit)

• Adds AI to your existing system
• Plug-in module + cloud subscription
• Setup 1-2 days
• Best for: customers who already have BESS / PCS

Model B — Full Turnkey

• A whole new system — GridMind + LEONICS PCS combined in one unit
• Pre-integrated, single SKU
• Setup per project (typically 1-3 months)
• Best for: new utility / C&I projects

---

Summary

GridMind AI isn't "a new EMS" — it's an AI platform with a cloud-trained, edge-run architecture, which means:

• Forecasts 3-5× more accurately than rule-based
• Adapts automatically when tariffs/patterns change
• Cuts the electricity bill 15-30% in the modeled case
• Carbon Report ready to use
• No vendor lock-in

---

Learn more on the GridMind AI page, or see it working inside a factory peak-shaving solution and a utility-scale energy trading project.

Interested in GridMind AI? — LEONICS offers a free assessment to check how much you could save — contact us or call 02-746-9500