Case Study

The 2021 Caldor Fire Event

The Caldor Fire ignited August 14, 2021 in El Dorado County, ultimately burning 221,835 acres. PG&E initiated PSPS shutoffs across the Sierra foothills as the fire spread. AQI peaked at 318 (Hazardous) on August 16—precisely when affected communities most needed backup power. Standard solar-forward microgrids that failed to model smoke derate exhausted their stored energy within 72 hours.

221,835
Acres Burned
318
Peak AQI (Hazardous)
5
Day Islanding Window Modeled
213
Peak PM2.5 (µg/m³)

AQI & Solar Capacity — Aug 14–18, 2021

Left axis: Air Quality Index. Right axis: PV output as a percentage of nameplate capacity. As smoke thickens, solar generation collapses while community demand remains unchanged.

Key insight: Peak smoke on Aug 16 reduced PV output to just 38% of rated capacity—while the community's energy demand remained unchanged due to the active PSPS shutoff.
GHI vs. DNI: Wildfire smoke preferentially scatters direct-normal irradiance (DNI) while allowing diffuse GHI to partially transmit. Monocrystalline silicon panels optimized for direct beam suffer a disproportionately larger derate than bifacial or thin-film alternatives—a spectral shift that standard PVWatts models do not capture. The Scipionic model incorporates an empirical smoke derate coefficient derived from PM2.5 sensor fusion.
Simulation

Death Curve vs. Survival Curve

Modeling a 15 kW PV + 100 kWh battery system serving a 10-home cluster (avg. 75 kWh/day load). Both systems start at 100% SoC at PSPS onset. The critical difference: Scipionic's RaaS software monitors SoC in real time and sheds non-critical loads when reserves drop below 35%.

Battery SoC (%) — Death Curve vs. Survival Curve

+3/8/12/15% cumulative
Both systems track identically through Day 3. On Day 4 the Scipionic load-shedding algorithm engages—cutting non-critical loads (HVAC, EV chargers, pool pumps) by 40%—while the standard system continues drawing full load until failure.

Scipionic RaaS — Day 4 Actions

  • Medical devices — ON
  • Well water pumps — ON
  • Refrigeration — ON
  • Emergency lighting — ON
  • HVAC — SHED
  • EV chargers — SHED
  • Pool pumps — SHED

Standard System — Day 4 Status

28%
SoC entering Day 4
Critical threshold

No intelligent load management. System continues drawing 75 kWh/day from a 28 kWh remaining reserve. Complete discharge by midday.

⚠ SYSTEM FAILURE — DAY 4
Operations

RaaS Control Plane — Community Fleet

Operator view of a 50-node cluster in Nevada City, CA. Use the controls below to simulate a utility PSPS shutoff and a low-battery load shedding response—the two critical events in a multi-day wildfire islanding scenario.

Event Type:
Grid-Tied — Normal Operations
Grid-Tied Islanded (PSPS) Critical (always on) Load Shed
Tier 1 — Critical
5 Nodes
Medical clinic, fire station, shelter, well pump, comms hub
● POWERED
Tier 2 — Priority Residential
15 Nodes
Elderly residents, medical device users, low-income households
● POWERED
Tier 3 — Standard Residential
30 Nodes
General residential, non-critical loads
● POWERED
Justice40 Calculator

Financial & Equity Comparison

Adjust the community parameters below to compare traditional utility undergrounding costs against a RaaS microgrid deployment with stacked federal and state incentives. This is the financial model behind the grant application.

Apply:
Loading...
Cost of Standard System Failure
$37,500
50 homes × 5 days × $150/day (food spoilage $50 + lost wages $70 + emergency shelter $30)
This is what deploying a standard system costs the community.
Service vs. Asset — Legal Structure

The RaaS entity owns the hardware. The resident pays a monthly service fee. No lien, no capital improvement, no Prop 13 reassessment trigger.

Positioned identically to a utility bill, not a home improvement loan.
Grant Narrative

The SRE Advantage

Site Reliability Engineering disciplines—originally developed to manage planetary-scale distributed systems—translate directly to microgrid operations. Scipionic's control plane is built on these principles, bringing enterprise-grade reliability to community energy infrastructure.

Chaos Engineering

Smoke Derate Simulation = SRE GameDay

This simulation engine is a GameDay exercise—deliberately injecting a wildfire event to verify that the system degrades gracefully. Just as SREs inject failures into production to validate runbooks, Scipionic validates islanding behavior before deployment.

Observability

RaaS Control Plane = Distributed Telemetry

Every node emits SoC, load, and PV yield metrics in real time. The control plane aggregates fleet-wide telemetry with sub-second latency—the same distributed tracing and metrics pipeline used in cloud infrastructure, adapted for energy systems.

Incident Response

Load Shedding Tiers = Runbook-Driven Automation

The three-tier load shedding protocol is a codified runbook. When SoC drops below threshold, automated actions execute in priority order—no human decision required. This is incident response at machine speed.

SLO / SLA

Contractual Power Guarantee = Energy SLA

The RaaS service agreement defines a baseline power availability target for critical loads—an energy SLA backed by the control plane's automated enforcement. Communities get the same contractual reliability guarantees as enterprise cloud customers.