Analysis on powering Data Centers with islanded renewable micro-grids

Abdullah Alsayegh

This projects aims to design a network that meets the power of AI Data Centers exclusively with off-grid PV, HAWT and LFP. The thesis exploits AI training’s ability to extend it’s training duration if power is lacking as a form of flexibility. The analysis thus uses 45 years of load profiles from renewables.ninja to model the system’s resilience in renewable uptime and training duration. Microgrid component uptime is also considered and redundancy is added to meet Tier I availability. The sizing for each year is done with a MILP Gurobi solver that sizes the system to minimize the NPC. Battery degradation is also modeled after each annual dispatch cycle using Rain-flow counting.

Supervisor(s)

  • Dr Fei Teng (Electrical and Electronic Engineering)
  • Ying Yu (Electrical and Electronic Engineering)

Optimising Data Centre Operations with Small Modular Reactors

Alexandre Chekroun

In the context of accelerating digital transformation and urgent net-zero targets, data centres face unprecedented energy demands. Driven by exponential data growth and AI’s rising computational intensity, they now consume over 1 % of global electricity, straining grids. This project investigates the technical and economic viability of integrating Small Modular Reactors (SMRs)—50–300 MWe advanced nuclear modules—as dedicated, low-carbon, dispatchable power sources for high-capacity data centres.

Supervisor(s)

  • Dr Gbemi Oluleye (The Grantham Institute)

Life Cycle Assessment of Data Centre Cooling Systems: A Case Study in Malaysia

Ashraf bin Ab Llah

This project conducts a Life Cycle Assessment (LCA) of data center cooling systems with a case study for Malaysia, comparing conventional air cooling, cold plate liquid cooling, immersion cooling, and a novel LNG-based approach. It evaluates both embodied and operational emissions across various climate scenarios. Thermodynamic modeling and component-level inventory data are used to quantify environmental impact and guide low-carbon cooling strategies for tropical data centers.

Supervisor(s)

  • Dr Fei Teng (Electrical and Electronic Engineering)

Techno-Economic and Business Model Assessment of SMR-powered Data Centres in the UK vs. Ireland

Emma Panerai

As data centres become significant electricity consumers, interest is growing in using Small Modular Reactors (SMRs) to meet their increasing need for reliable, low-carbon power. In 2023, data centres accounted for 21% of Ireland's total electricity consumption, whereas the UK share could rise to 6% by 2030. Thus, this project investigates the feasibility of powering data centres with SMRs in both countries by assessing their technical suitability and economic performance. It also examines site suitability for SMR-powered data centres and evaluates viable business models. This includes an analysis of ownership structures and stakeholder dynamics within the UK and Ireland.

Supervisor(s)

  • Dr Mark Wenman (Centre for Nuclear Engineering)
  • Dr Michael Bluck (Centre for Nuclear Engineering)
  • Dr Fei Teng (Electrical and Electronic Engineering)

Investing in Data Centres: Robust development strategies in possible electricity constrained futures

Sukhjit Singh

AI’s rapid growth is accelerating global data centre deployment. By 2030, data centres could consume 3% of global electricity. However, limited global power availability and the transition to net-zero are already contributing to grid connection delays. This project uses a Robust Decision Making framework to examine key uncertainties faced by private developers and their strategies against them, particularly delays from transmission operators that threaten timely deployment and investment returns. Through exploratory modelling, it identifies powerful strategies that remain robust under deeply uncertain delay scenarios.

Supervisor(s)

  • Dr Mark Workman (Centre for Environmental Policy)

Harnessing Data Centre Flexibility: Optimising Active IT Load Management for Energy Markets

Thomas Potgieter

As data centres fuel growing energy demand, this project explores how they can become grid stability assets. This research develops an optimisation algorithm and bidding policy that drives climate impact through improved operational profitability, enabling operators to participate in energy markets whilst maintaining service quality. Using real-world ML workload data, it proposes a profit-maximising framework for active IT load management.

Supervisor(s)

  • Dr Fei Teng (Electrical and Electronic Engineering)
  • Dr Yifu Ding (MIT)
  • Ying Yu (Electrical and Electronic Engineering)