EV Charging Simulation (Python)

Learn and practice EV charging simulation for a fleet of vehicles using hourly time steps. Current scope covers:

• Synthetic EV profile generation (Day 1)
• Uncontrolled charging simulation + fleet load aggregation (Day 2)
• Rule-based smart charging (peak avoidance) + feasibility root-cause analysis (Day 3)
• Time-of-use tariff and charging cost comparison (Day 4)

---

What’s implemented

Day 1 — Synthetic EV profile generation

Generates realistic EV charging profiles for a single day:

• arrival_hour, departure_hour
• battery_kwh
• initial_soc, target_soc
• derived: energy_needed_kwh, available_hours

Outputs:

• results/ev_profiles_day1.csv
• results/figures/arrival_hour_histogram.png
• results/figures/energy_needed_histogram.png

Day 2 — Uncontrolled charging simulation (baseline)

Implements uncontrolled charging:

• Charging starts immediately at arrival_hour
• Constant charging power (default: 7 kW)
• Stops when energy target is reached or when departure_hour is reached

Outputs:

• results/fleet_load_uncontrolled.csv (hourly aggregate fleet load, kW)
• results/ev_results_uncontrolled.csv (per-EV delivered energy, shortfall, completion)
• results/figures/aggregate_load_uncontrolled.png

Day 3 — Rule-based smart charging + feasibility analysis

Adds a rule-based smart charging strategy:

• Allocates charging to non-peak hours first within each EV’s available window
• Uses peak hours only if needed
• Purpose: peak shaving without changing total delivered energy

Also adds a feasibility/root-cause check:

• If energy_needed_kwh > available_hours * charging_power_kw, the EV cannot complete (infeasible)
• Confirms whether incomplete sessions are due to algorithm issues or time-window constraints

Outputs:

• results/fleet_load_smart_rule_based.csv
• results/ev_results_smart_rule_based.csv
• results/figures/fleet_load_comparison_uncontrolled_vs_smart.png

Day 4 — Time-of-use tariff and charging cost comparison

Adds a simple time-of-use tariff model and calculates total charging cost per scenario.

Tariff structure:

• Off-peak: 0.20 EUR/kWh (hours 0–6)
• Shoulder: 0.30 EUR/kWh (hours 7–15, 21–23)
• Peak: 0.45 EUR/kWh (hours 16–20)

New functionality:

• Calculates total fleet charging cost from hourly load profile
• Compares uncontrolled vs smart charging under the same tariff
• Visualizes fleet load against tariff profile

Outputs:

• results/tariff_schedule.csv
• results/metrics_comparison_day4.csv
• results/figures/fleet_load_and_tariff_comparison.png

---

Results (current run)

Configuration:

• Fleet size: 50 EVs
• Seed: 42
• Charging power: 7 kW
• Peak avoidance hours: [16, 17, 18]
• Time resolution: 1 hour (0–23)

Uncontrolled (baseline)

• Peak load: 82.67 kW
• Total energy delivered: 1073.94 kWh
• Total energy needed: 1330.32 kWh
• Completion rate: 68%
• Incomplete EVs: 16
• Avg shortfall (incomplete only): 16.024 kWh
• P95 shortfall (incomplete only): 37.2 kWh
• Total cost: 339.68 EUR

Smart (rule-based peak avoidance)

• Peak load: 74.64 kW (~9.7% reduction vs uncontrolled)
• Total energy delivered: 1073.94 kWh (unchanged)
• Completion rate: 68% (unchanged)
• Incomplete EVs: 16 (unchanged)- Total cost: 336.07 EUR
• Cost saving: 3.61 EUR (~1.06% reduction vs uncontrolled)

Root-cause insight (uncontrolled)

• Not completed but feasible: 0
• Interpretation: all incomplete EVs are infeasible given fixed charging power and limited available hours, not due to simulation logic.
• This means current service limitations come from physical/time constraints, not from simulation or scheduling errors.

---

Figures

EV arrival distribution

Energy needed distribution

Uncontrolled charging fleet load

Fleet load comparison (uncontrolled vs smart)

Fleet load and tariff comparison

---

Assumptions

• Single-day simulation with 24 hourly slots (0–23)
• departure_hour is not inclusive
  • Example: arrival=10, departure=13 → charging can occur at hours 10, 11, 12
• Constant charging power (no CC–CV tapering yet)
• No grid import limit, transformer constraint, or feeder capacity limit
• No V2H/V2G yet
• Smart charging is currently rule-based, not optimization-based

---

How to run (Windows CMD)

1. Create and activate virtual environment

python -m venv .venv
.venv\Scripts\activate

2. Install dependencies

pip install -r requirements.txt

3. Run the pipeline

python run.py

4. Outputs

Outputs are saved under results/ and figures under results/figures/.

---

Project structure

ev-charging-simulation/
├─ src/
│  ├─ profile_generator.py      # generate + validate synthetic EV profiles
│  ├─ charging_simulator.py     # uncontrolled + rule-based smart charging
│  └─ metrics.py                # KPIs (peak load, energy, completion, shortfall stats)
├─ results/
│  ├─ ev_profiles_day1.csv
│  ├─ fleet_load_uncontrolled.csv
│  ├─ ev_results_uncontrolled.csv
│  ├─ fleet_load_smart_rule_based.csv
│  ├─ ev_results_smart_rule_based.csv
│  └─ figures/
│     ├─ arrival_hour_histogram.png
│     ├─ energy_needed_histogram.png
│     ├─ aggregate_load_uncontrolled.png
│     └─ fleet_load_comparison_uncontrolled_vs_smart.png
├─ requirements.txt
└─ run.py