Low-voltage grids are insufficiently monitored.
The low-voltage grid was designed for one-way power flow, but rapid deployment of rooftop solar, EVs, and heat pumps has turned it into an active, bidirectional system. The monitoring infrastructure was never built to observe these dynamics, so DSOs have almost no real-time visibility below the MV/LV transformer. AMI polls at 15-minute resolution: too slow to catch voltage violations, reactive power imbalances, or oscillatory behavior before they propagate upstream.
The April 2025 Iberian blackout made the cost impossible to ignore: cascading failure within seconds, with ENTSO-E identifying limited LV observability as a compounding factor. With a congestion horizon extending well into the next decade, DSOs need continuous, high-resolution data at the LV level. That data does not yet exist at scale.
- AMI polls at 15-minute resolution, too slow for dynamic events
- No real-time visibility below the MV/LV transformer
- Flex limits applied without continuous voltage and current data
Edge-AI sensors that enable real-time visibility in the low-voltage grid.
Griddle combines sparse measurement hardware with a physics and machine-learning state estimation. Hardware is installed on a few feeders to infer voltage, current, and power quality across a large part of the low-voltage network in real time.
Beyond 15-minute AMI limits
Measure at 8 kSps and stream processed data continuously. Live detection of voltage violations and oscillatory instability before they propagate upstream.
Edge-Native State Estimation
Install hardware on only a few feeders. Each node runs local analytics, allowing the cloud layer to reconstruct the full grid state even where sensors are sparse.
From Reactive to Predictive
Stop waiting for failures or relying on episodic manual measurements. Issue Dynamic Operating Envelopes based on actual, live network state rather than conservative static limits.
Built for the grid, not the cloud.
Griddle runs inference at the edge. No cloud round-trips, no latency, no dependency on connectivity for critical calculations.
- Sampling Rate
- 8,000 sps
- Harmonic Coverage
- Up to 80th harmonic
- Connectivity
- 4G / LTE / Ethernet / Custom
- State Estimation
- PINN with KVL/KCL hard constraints
- Detection
- Sub-Hz oscillation (0.2–1 Hz)
- Inference
- Edge-native (on-device)
More insight, fewer sensors.
How Griddle compares to conventional monitoring approaches DSOs are evaluating today.
- Provide data only for the exact point of installation
- No built-in state estimation to fill the gaps
- High latency; heavily reliant on cloud processing
- 8 kSps sampling with actionable outputs
- Reconstruct grid state from sparse nodes
- Live anomaly detection on the edge, ready for dynamic flex control
- 15-minute polling resolution
- Too slow for dynamic events or sub-Hz oscillations
- Data is too aggregated to act upon in real time
Built for the challenges DSOs face today.
Dynamic Operating Envelopes
Calculate per-asset DOEs from actual measured conditions, not conservative static limits. Unlock second-by-second load intervention as an operational capability.
Continuous LV Network Visibility
Move from episodic, manually-triggered measurements to a permanent, continuous sensor layer below the MV/LV transformer.
Cable Reinforcement Planning
Base grid expansions on actual measured load profiles and congestion signatures. Direct capital and scarce technicians where they're needed most.
Real-Time Power Quality Monitoring
Detect oscillatory instability and reactive-power issues within seconds. The same class of triggers behind the April 2025 Iberian cascade.
Ready to see your grid?
Get in touch to schedule a demo or discuss how Griddle fits your network.