Imbalance Cost Analyzer

Calculates specific net imbalance costs, forecast bias, and RMSE for renewable energy portfolios in the Baltic electricity market, following the methodology of the RTUCON2025 paper (pending indexing in IEEEXplore).

Quick Start

Setup

# Windows
setup.bat

# Linux/Mac
chmod +x setup.sh && ./setup.sh

Run

# Windows
run_analyzer.bat

# From Spyder: open main.py, set working directory to this folder, press F5

Calculation Modes

Mode 1: ENTSO-E Aggregate Analysis

Analyzes aggregate national wind/solar portfolios for Latvia, Lithuania, and Estonia using publicly available ENTSO-E Transparency Platform data.

Required input files:

• Actual Generation per Production Type CSV(s) — one or more per country
• Generation Forecasts for Wind and Solar CSV(s) — one or more per country
• Day-ahead price CSV — one per country (format: Country,ISO3 Code,Datetime (UTC),Datetime (Local),Price (EUR/MWhe))
• Imbalance price CSV — single file from Baltic Balancing Dashboard

Outputs: Tables I–III from the paper (specific costs, bias, RMSE) + visualizations.

Mode 2: Plant-Level Analysis

Analyzes individual wind/solar plants from standardized CSV files. Computes metrics per plant and for the aggregate portfolio.

Required input files:

• One CSV per plant in the standard template format (see templates/plant_template.csv)
• Day-ahead price CSV (same format as Mode 1)
• Imbalance price CSV (same as Mode 1)

Standard plant CSV format:

datetime_utc,actual_mwh,forecast_mwh,installed_capacity_mw,plant_type
2025-01-01 00:00:00+00:00,4.88,4.85,4.9,wind

Example files are provided in the templates/ directory.

Methodology

The core formulas (from the RTUCON2025 paper):

1. Hourly net imbalance cost: C_t = (P_actual - P_forecast) × (π_imb - π_da)
2. Specific imbalance cost: C_sp = ΣC_t / ΣP_actual (€/MWh)
3. Percentage Bias: PctBias = -Σ(P_actual - P_forecast) / ΣP_actual × 100%
   • Negative = under-forecasting, Positive = over-forecasting
4. RMSE: sqrt(Σ(P_actual - P_forecast)² / T), normalized vs max generation or installed capacity

When input files cover different time periods, calculations use only the overlapping timespan.

Project Structure

imbalance_cost_analyzer/
├── main.py                     # Entry point (launch GUI)
├── setup.bat / setup.sh        # Environment setup
├── requirements.txt
├── calculations/
│   └── metrics.py              # Core formulas
├── data_loaders/
│   ├── entso_loader.py         # ENTSO-E actual gen + forecast CSVs
│   ├── price_loader.py         # Day-ahead price CSVs
│   ├── imbalance_price_loader.py  # Baltic Balancing Dashboard CSV
│   └── plant_loader.py         # Standard plant CSV format
├── modes/
│   ├── entso_mode.py           # Mode 1 orchestration
│   └── plant_mode.py           # Mode 2 orchestration
├── gui/
│   └── app.py                  # Tkinter GUI
├── visualization/
│   └── plots.py                # All charts
├── templates/
│   ├── plant_template.csv      # CSV format specification
│   ├── example_wind_plant.csv  # Example data
│   └── example_solar_plant.csv # Example data
└── tests/
    └── test_core.py            # Test suite