This Digital Twin calculates grid hosting capacity in under 100ms. Not a guess. A deterministic solve backed by Kirchhoff's laws. Every click runs a real Newton-Raphson power flow.
+44%
Capacity Gain
<100ms
Solve Latency
100%
Physics Compliance
0
Training Data Required
The Key Result
One firmware setting change. cos φ 0.95 reactive absorption. +44% hosting capacity. Proven by physics, auditable for regulators.
Every click triggers a real Newton-Raphson power flow solve. No lookup tables. No approximations.
How much solar PV can each bus accept before hitting EN 50160 voltage limits? You get the answer live.
Newton-Raphson solver fires on every interaction. Real physics, under 100ms.
Smart-inverter Q(V) reactive absorption. A firmware toggle. No new cables.
The network sits on real Essen streets. Per-bus voltage labels. Color-coded by grid state.
The system finds the exact bottleneck. Bus 41, margin 0.0004 pu. No manual searching.
A full year of real PVGIS irradiance. The grid state is solved for every single hour.
What's running under the hood.
Solver
pandapower 3.4.0
Method
Newton-Raphson AC
Compute
Pure CPU
Latency
<100ms
Benchmark
CIGRE TF C6.04.02
Irradiance
PVGIS v5.2 (EU JRC)
Nano-Models
111,572
Verification
Z3 SMT Solver
E.ON runs world-class tools. ARYA doesn't replace them. It fills gaps they weren't built to address.
Each scenario is a separate run. 50 scenarios means 50x the wait.
Hours to days per study
The UI shows pre-computed results. You can't ask 'what if?' on the fly.
You decide what to simulate before knowing what to ask
PV x inverter x load x curtailment = millions of combinations. Most never get tested.
Critical edge cases missed
Finding which bus binds first means manually inspecting result tables.
Bottlenecks found late in the process
Planning tools run offline. You can't plug them into live operations.
Planning and operations stay disconnected
Results live in spreadsheets. Reconstructing the decision chain for regulators is manual work.
Weeks of documentation effort
Same data, same network. Completely different speed and capability.
Move a slider, see the grid respond. No queue. No waiting.
Why existing tools can't: Traditional tools optimize for large-network accuracy, not interactive speed.
Tests every PV level from 0 to 80 kWp in 2.5 kWp steps, across all inverter modes. Takes seconds.
Why existing tools can't: Doing this manually would mean hundreds of individual simulation runs.
Finds the binding constraint automatically. Bus 41, voltage limit. Q(V) relaxes it by +44%.
Why existing tools can't: In traditional tools, engineers compare results across scenarios by hand.
Every number traces back to its model, solver version, data source, and assumptions.
Why existing tools can't: Traditional tools produce result files. The audit trail is manual documentation.
Satisfies Kirchhoff's laws by construction. Same input, same output. Every time.
Why existing tools can't: LLMs hallucinate. ARYA computes. It physically cannot produce a wrong answer.
Works from physics alone. No historical data needed. No training. No model drift.
Why existing tools can't: ML tools need months of data collection and constant retraining.
Four real scenarios. Measurable impact.
Today
Customer applies for 50 kWp PV. Planner sets up a scenario, runs simulation (takes hours), reviews results, writes a report. Customer waits weeks.
With ARYA
Connection portal calls ARYA. Under 100ms later: 'Bus 41 can't accept 50 kWp (V=1.11 pu). With Q(V): accepted (V=1.07 pu). Alternative: 45 kWp without Q(V).' Customer gets an answer in seconds.
Weeks become seconds. No engineering bottleneck. Full audit trail included.
Today
E.ON knows smart inverters could defer cable upgrades. But proving it to regulators takes months of simulation studies and documentation.
With ARYA
Constraint-Breaker generates the proof: Baseline HC = 45 kWp. With Q(V) = 65 kWp (+44%). Transformer stays under 59%. Cable upgrades deferred 5+ years via firmware.
Regulatory approval timeline: months become weeks. Deferred CAPEX backed by physics.
Today
Planning uses worst-case assumptions. The grid usually has more headroom than worst case suggests, but nobody quantifies it.
With ARYA
Time-Series engine shows the grid hits voltage limits during only 127 hours/year (1.4%). For 98.6% of the time, additional capacity is sitting there unused.
20-30% more DER capacity unlocked through operational intelligence. Zero hardware spend.
Today
Running hosting capacity analysis for every feeder is impractical. Thousands of engineering-hours.
With ARYA
Same engine: 100ms per feeder. 500,000 feeders in hours (batch) or on-demand via API. Zero-shot means no per-feeder calibration needed.
Network-wide visibility for the first time. Upgrade priorities based on data, not guesswork.
Every answer ARYA produces satisfies the laws of physics by construction. This isn't a language model guessing. It's a first-principles solver computing the exact answer.
| LLM / Traditional AI | ARYA | |
|---|---|---|
| Method | Statistical pattern matching | First-principles physics |
| Guarantee | 70-90% on a good day | 100%. Deterministic by construction |
| Failure mode | Hallucination | Cannot violate physics |
| Auditability | Black box | Full lineage, every number traceable |
| Regulator-ready | No | Yes |
| Training data | Millions of examples | Zero. Works from physics alone |
Six modules. Each one runs real-time physics.
Tab 1
The executive view. Baseline vs. smart-inverter capacity, binding constraints, key findings at a glance.
Tab 2
Per-bus voltage table that updates on every parameter change. Move a slider, see the numbers move.
Tab 3
Interactive map on real Essen streets. Schematic topology. Hosting capacity heatmap. Full-year playback.
Tab 4
Complete parametric sweep across PV levels. Shows the entire feasibility boundary, not just one point.
Tab 5
Finds what binds, what relaxes it, and by how much. Automated. No manual comparison needed.
Tab 6
The full audit trail. Network model, data sources, assumptions. Everything a regulator would ask for.
ARYA doesn't replace your stack. It makes it faster.
envelio IGP
Workflow Layer
Asset management, GIS, workflow orchestration, connection processing
ARYA
Physics Engine
Real-time solve, deterministic answers, full provenance, sub-100ms latency
envelio handles the workflow. ARYA handles the physics. Together: instant, auditable, scalable grid decisions.
The demo runs a real physics solver on every interaction. Nothing is pre-computed. Click around. The physics holds.
Launch Live Demo