MSXVI Strategy System

Race strategy optimization for Midnight Sun solar racing team competing in Formula Sun Grand Prix (FSGP) and American Solar Challenge (ASC).

FSGP (Qualifier): Maximize distance over fixed time period. Scoring based on total distance with elapsed time as tiebreaker.

ASC (Main Event):

1. Complete base route without trailering
2. Maximize official miles (1st tiebreaker)
3. Minimize elapsed time (2nd tiebreaker)

Successfully achieving these objectives hinges on two factors:

• Designing an efficient solar car (less power out)
• Following an optimal race strategy

Optimal race strategy boils down to a single question:

What speed should we drive at to maximize distance within race constraints?

We address this question by leveraging the route, weather, and solar irradiance data.

Architecture

Database (db/)

PostgreSQL database with dual-deployment support:

• Local instance: Offline operation during race
• Cloud instance: Remote data storage and team access
• Synchronization: Bidirectional sync via db/sync.py

Tables:

• route_model: Waypoints, elevation, gradients, and road characteristics from GPX files
• irradiance_archive / irradiance: Historical/live solar irradiance (GHI), latitude/longitude coordinates, timestamps, etc.

Key Files:

• db/connect.py - Database connection and initialization
• db/load.py - Query execution and data loading to memory
• db/sync.py - Cloud-to-local synchronization
• db/setup/route_model/ - GPX parsing and route table generation
• db/setup/irradiance/irradiance.py - Live solar irradiance data processing from Solcast (not tested)
• db/setup/irradiance/irradiance_archive.py - Historical solar irradiance data processing

Simulator (src/simulation.py)

Physics-based energy model computing instantaneous power flows:

Power Consumption:

• Rolling resistance: P_rr = (M·g·C_r1 + 4·C_r2·v)·v
• Aerodynamic drag: P_drag = 0.5·ρ·C_d·A·v³
• Gradient resistance: P_grad = M·g·sin(θ)·v

Power Generation:

• Solar array: P_solar = A_solar·GHI·η_solar

Vehicle Parameters:

• Mass: 300 kg
• Drag coefficient: 0.13
• Frontal area: 1.357 m²
• Solar array: 4.0 m² @ 16% efficiency
• Battery capacity: 5.23 kWh (40s, 3.63V, 36Ah cells)

Integrates power flows over time to track battery State of Charge (SOC) for any velocity profile.

Optimization (src/optimize.py)

SLSQP (Sequential Least Squares Programming) to find optimal velocity profile.

Note: SLSQP finds the gradients numerically (requires tuning).

Objective: Maximize distance traveled within race constraints.

Constraints:

• Velocity bounds: 10-20 m/s
• Battery SOC ≥ 20% at all timesteps

Algorithm: Iteratively adjusts velocity profile until convergence on maximum-distance solution.

Visualization (src/overview.py)

• Elevation profiles: Distance vs. elevation for individual stages or full route
• Irradiance profiles: GHI over time for specific locations

Installation

# Clone repository
git clone <repository-url>
cd strategy_msxvi

# Install dependencies
uv sync

# Configure environment variables
cp .env.example .env
# Edit .env with:
#   - Local PostgreSQL database credentials
#   - Cloud database credentials (optional)
#   - Solcast API key (optional, needed for live irradiance data)

Usage

Run Simulation (with or without Optimization)

Edit src/main.py to configure:

• OPTIMIZE: Enable/disable optimization (default: False)
• SIMULATION_DURATION_SEC: Race duration
• TIMESTEP_SEC: Simulation timestep
• MIN_SPEED_MS, MAX_SPEED_MS: Velocity bounds

uv run -m src.main

Database Setup

Option 1: Import Database Dump (Recommended)

If a database dump is available, this is the fastest way to get started:

# Import the database dump (creates database if needed)
python -m db.import data/database_dump.sql

# Or manually with psql
psql -U postgres -d strategy_msxvi -f data/database_dump.sql

Option 2: Build Database from Scratch

Build the database from GPX files and irradiance data:

# 1. Create route_model table from GPX files
python -m db.setup.route_model.route_model

# 2a. Create irradiance_archive table with historical data (requires SOLCAST_API_KEY)
python -m db.setup.irradiance.irradiance_archive

# 2b. Or create irradiance table with live/forecast data
# MODE 1: Live forecast (requires API key)
# MODE 2: Historical API (requires API key)
# MODE 3: Synthetic data (no API key needed)
python -m db.setup.irradiance.irradiance

# 3. Sync cloud database to local (optional, not tested)
python db/sync.py

Export Database (for maintainers):

# Export local database to share with others
python -m db.export

# Or export cloud database
python -m db.export --cloud

# Manually with pg_dump
pg_dump -U postgres -d strategy_msxvi -f data/database_dump.sql

Note: Database dumps are excluded from git due to size. Share the dump file via Google Drive, Dropbox, or team file server.

Data

Route Files: ASC 2024 GPX files in data/asc_24/

• Full base route and individual stage files
• Format: <stage>_<StartCity>To<EndCity>.gpx and <stage>L_<City>Loop.gpx

Irradiance Data: Historical GHI data from Solcast in data/asc_24/historical_irradiance.csv

Project Structure

strategy_msxvi/
├── src/
│   ├── main.py           # Main simulation runner
│   ├── simulation.py     # Physics-based energy model
│   ├── optimize.py       # SLSQP velocity optimization
│   ├── overview.py       # Route and irradiance visualization
│   ├── plot.py           # Energy flow plotting utilities
│   └── utils.py          # Data loading and mapping functions
├── db/
│   ├── connect.py        # PostgreSQL connection management
│   ├── load.py           # Database query and data loading
│   ├── sync.py           # Cloud-local database synchronization
│   ├── export.py         # Export database to SQL dump
│   ├── import.py         # Import database from SQL dump
│   └── setup/
│       ├── route_model/  # GPX parsing scripts
│       └── irradiance/   # Solar irradiance processing scripts
├── data/
│   └── asc_24/           # Route GPX files and irradiance data
└── docs/
    └── architecture.PNG  # System architecture diagram

Dependencies

• Core: Python 3.12+, NumPy, SciPy, Pandas
• Database: psycopg2-binary, python-dotenv
• Visualization: Matplotlib
• Route parsing: gpxpy
• Progress: tqdm

Technical Notes

• Database: PostgreSQL with environment-based configuration (local/cloud)
• Optimization: SciPy SLSQP with nonlinear constraints
• Data caching: Lazy loading of route and irradiance data to memory (pandas)
• Coordinate mapping: Distance-based route lookups, spatial discretization for irradiance

Future Development

• Model Predictive Control (MPC) implementation for real-time receding horizon optimization
• Telemetry integration for actual vs. predicted comparison
• Multi-day race strategy planning

---

Email: [email protected] if you have questions, concerns, or want to contribute.