📋 Case Study

San Francisco Municipal Utility District (SFMUD) Office Tower DR Pilot

Limited rooftop space for generation; required 20% peak load reduction during CAISO evening ramps without occupant disruption

🏗️ Project Overview

12-story municipal office building in downtown SF with 1.2 MW peak load

🎯 Challenge

Limited rooftop space for generation; required 20% peak load reduction during CAISO evening ramps without occupant disruption

🔧 Design Approach

AI-driven HVAC pre-cooling + smart plug load management + battery-assisted load shift using OpenADR 2.0b signals

📐 Design Diagram

AI-generated project design illustration

📐 Key Calculations

Pre-cooling Thermal Lag Offset

ΔT × C × t

Result: 3.2°C·kWh/hr

Determines optimal start time for pre-cooling

Plug Load Shed Margin

Baseline Avg − 95th Percentile

Result: 185 kW

Safe shed capacity without impacting operations

📊 Results

22.6% peak reduction achieved across 142 events; $189K annual demand charge savings; zero occupant complaints

💡 Lessons Learned

• •Occupant comfort algorithms must be calibrated to local thermal preferences
• •OpenADR signal latency requires 5-min lookahead buffer

✅ Key Takeaways

• 1Occupant comfort algorithms must be calibrated to local thermal preferences
• 2OpenADR signal latency requires 5-min lookahead buffer