Context

Data centers in Tunisia and North Africa face a specific energy challenge: an unstable grid, exceptional solar resource, available gas infrastructure, and rising sustainability requirements — yet most deployed power systems are simply copied from European templates. This project focuses on the energy and economic dimension: modelling, simulating, and comparing candidate power architectures to identify the optimal solution for the regional context, using exclusively open-source tools.

Objective

Model and simulate four candidate power architectures for a mid-size data center (1–10 MW IT load) and recommend the best-performing solution across energy efficiency, cost, carbon footprint, and grid independence — with cogeneration as a key candidate option.

Architectures Compared
- A — Classical Baseline: Grid + Diesel Genset + UPS
- B — Hybrid Renewable: Grid + PV + BESS + Genset backup
- C — Gas CHP: Onsite gas cogeneration as primary source + grid backup + UPS
- D — Hybrid CHP + Renewable: Gas CHP + PV + BESS, minimising grid dependency

Scope of Work
1. Local Context and Data Collection
- Define IT load profile and reference site (Tunis or Sfax)
- Collect grid quality data (outage frequency, voltage/frequency deviations, STEG tariffs)
- Extract solar irradiance and climate data
- Assess local gas availability and pricing (STEG gas tariffs)
2. Energy System Simulation Using custom Python scripts
- Hourly power flow over a full representative year
- CHP dispatch logic and fuel consumption modelled via custom Python scripts based on engine performance curves
- BESS charge/discharge cycles and state of charge profiles
- Grid import/export balance

3. Performance Evaluation Compute for each architecture:
- Power Usage Effectiveness — full breakdown (IT, cooling, conversion losses)
- Energy cost — 10-year OPEX based on STEG electricity and gas tariffs
- CAPEX estimate — from vendor catalog data and engineering benchmarks
- Carbon footprint — CO₂/kWh using Tunisian grid and gas emission factors
- Grid independence ratio — share of IT load served without grid draw
4. Multi-Criteria Optimisation
- Weighted MCDA (Multi-Criteria Decision Analysis) scoring matrix implemented in Python
- Pareto front visualisation (cost vs. carbon vs. independence)
- Sensitivity analysis: gas price variation, grid tariff evolution, solar irradiance scenarios
5. Results and Recommendation
- Comparative results dashboard
- Justified architecture recommendation with scenario robustness analysis

Deliverables
- Local context data report
- Simulation codebase (Python, fully documented)
- Comparative performance results for all four architectures
- Architecture recommendation report
- Final report

Required Profile
- Final-year student in Electrical Engineering, Energy Engineering, or Industrial Engineering
- Python programming
- Power systems and energy conversion fundamentals
- Interest in renewable energy and sustainable infrastructure
- Autonomous and rigorous

Duration
5 to 6 months