✈️ Gate Assignment Optimization (GAP)

This repository implements a Mixed-Integer Programming (MIP) model for solving an airport gate assignment problem using operational flight data and historic gate usage patterns.

The project is designed to be:

• cleanly structured
• easy to reproduce
• suitable for GitHub portfolios

---

📁 Repository Structure

GAP/
|
|-- gap_main.py                # Main optimization model (MIP)
|-- historic_analysis.py       # Historic data analysis & plots
|-- optimized_analysis.py      # Optimized solution analysis & plots
|
|-- flight_data.xlsx           # Input data (one-day + historic)
|
|-- csv_outputs/               # Generated CSV outputs
|-- plot_outputs/              # Generated plots
|
|-- requirements.txt
|-- README.md

---

📊 Input Data

The model reads data from flight_data.xlsx containing two sheets.

1️⃣ one_day_departures (Optimization Input)

Required columns:

• flight_code – Unique flight identifier
• boardingtime – Boarding start time (HH:MM)
• departure_time – Scheduled departure time (HH:MM)
• passengers – Passenger count
• aircraft_wing_span – Aircraft wingspan (meters)

2️⃣ historic_gate_allocation (Historic Reference)

Used to derive gate capacity distributions and benchmark performance.

Required columns:

• departure_gate – Gate / stand number
• passengers – Passenger count
• boarding_time or flight_time – Time reference
• coaches (optional) – Number of coaches used

---

🧮 Optimization Model

Sets

• F : set of flights
• G : set of gates
• T : set of discretized time slots
• C = {0,1} : assignment modes
  • 0 = contact gate
  • 1 = coached (remote stand)

Parameters

• P_f : passengers of flight f
• B_f : boarding start time of flight f (minutes)
• D_f : departure time of flight f (minutes)
• A_{f,t} : 1 if flight f occupies time slot t
• K_g : passenger capacity of gate g (derived from historic data)
• w_h(f) : hour-of-day weight for flight f

Decision Variables

• x_{f,g,c} ∈ {0,1}
  = 1 if flight f is assigned to gate g using mode c (0: direct, 1: coached)

• z_g ≥ 0
  = total passengers assigned to gate g

• z_max, z_min ≥ 0
  = maximum and minimum gate passenger loads

---

🎯 Objective Function

Minimize coached flights while balancing passenger load across gates:

minimize
    Σ_f Σ_g w_h(f) · x_f,g,1
    +
    λ · (z_max − z_min) / (Σ_f P_f)

• First term penalizes coached (remote) assignments, especially during peak hours
• Second term minimizes passenger load imbalance

---

📐 Constraints

1️⃣ Flight Assignment

Each flight must be assigned exactly once:

Σ_g Σ_c x_f,g,c = 1     ∀ f ∈ F

2️⃣ Gate Time Compatibility

At most one flight may occupy a gate at any time slot:

Σ_f A_f,t · Σ_c x_f,g,c ≤ 1     ∀ g ∈ G, t ∈ T

3️⃣ Gate Passenger Capacity

z_g = Σ_f P_f · Σ_c x_f,g,c     ∀ g ∈ G
z_g ≤ K_g                      ∀ g ∈ G

4️⃣ Load Spread Definition

z_g ≤ z_max
z_g ≥ z_min     ∀ g ∈ G

---

📈 Outputs

CSV Outputs

• solution_gap.csv – Optimized flight-to-gate assignments
• gate_capacity.csv – Computed gate capacities
• Historic KPI tables
• Optimized KPI tables

Plots

• Historic passenger distribution by gate
• Historic coach usage and PSL by hour
• Optimized gate timelines
• Optimized passenger load distribution

---

▶️ How to Run

1️⃣ Activate environment

2️⃣ Run optimization

python gap_main.py

3️⃣ Optional: Run analysis separately

import historic_analysis as ha
ha.run_historic_analysis(hist_df)

import optimized_analysis as oa
oa.run_optimized_analysis(solution_df)

---

📦 Requirements

requirements.txt:

numpy
pandas
matplotlib
openpyxl
gurobipy

Install with:

pip install -r requirements.txt

---

🧠 Notes

• The model is intentionally kept simple and transparent.
• Additional constraints (wingspan–gate compatibility, remote-only gates, buffer tuning) can be modularly added.

---

👤 Author

Huseyin Ugur Yildiz