IEEE Best Paper 2025

Enabling Equitable EV Charger Deployment

A Multi-Objective Geospatial Optimization Framework for Berlin

The Problem

Berlin's EV infrastructure is heavily concentrated in central districts (Mitte), leaving outer residential areas (Treptow-Köpenick) with "charging deserts" and high user congestion (32 BEVs/Charger).

The Method

We developed a Multi-Objective Optimization (NSGA-II) model. It explicitly trades off Utility (coverage) vs. Equity (Gini coefficient) while accounting for uncertain demand.

The Result

A "Robust Portfolio" of strategies. Our Equitable Strategy reduces the city-wide Gini coefficient by 13% and boosts access in underserved areas by 21%.

I. Diagnosis: The State of Berlin (Q2 2025)

Explore the current infrastructure gap. Central districts are well-supplied, while outer districts face severe congestion.
Click on a district to view detailed metrics.

Select a District Red = High Congestion

Select a District

Click the grid on the left to analyze specific infrastructure data.

-

Total Chargers

-

Registered BEVs

-

Congestion (BEVs / Charger)

II. The Optimization: Choosing a Strategy

We modeled the deployment of 100 New Chargers. Compare how different optimization priorities shift resources across the city.

Max Equity Strategy

Prioritizes "charging deserts." It aggressively allocates resources to underserved districts like Treptow-Köpenick to lower the Gini coefficient.

Key Outcome (K=100)

  • Gini Coefficient: 0.258 (-13%)
  • Focus Area: Peripheral Districts

Charger Allocation by District (K=100)

Optimization Gain

vs. Deterministic Planning

+15% Resilience

III. The Trade-off (Pareto Front)

There is no single "perfect" solution. The chart below shows the optimal trade-offs found by the NSGA-II algorithm. Solutions to the top-left are better (Higher Utility, Higher Equity).

X: Equity (1 - Gini) | Y: Utility (Coverage)

IV. Why "Fuzzy" Matters

Standard plans are fragile. Our Robust (Fuzzy) model guarantees higher performance even when EV adoption fluctuates unexpectedly (Worst-Case Scenarios).

Comparison of Minimum Satisfaction levels in worst-case scenarios

V. Future Work — Grid Integration

Planned extensions to improve robustness, scalability and operational intelligence for large-scale EV integration. Focus areas include spatially-aware power flow, dynamic loading scenarios and smart charging control.

Geospatial Power Flow

Couple geospatial locations with grid topology to detect localized voltage constraints and prioritize reinforcement.

Planned Methods PowerFactory + GIS

Dynamic Loading & Reinforcement

Simulate peak EV penetration events and time-varying dispatch to size reinforcements and control strategies.

Key Deliverable Scenario Suite & Metrics

Smart Charging & V2G

Design demand-response and V2G control to flatten peaks and support distribution-level services.

Integration Grid Signals & Market