Improve the Energy Sustainability of AI Data Centers in Future Energy System

As AI and data centers expand, their energy demands are becoming a significant sustainability issue, with AI projected to consume 85-134 TWh of electricity by 2027—comparable to a small country's annual usage. The growing adoption of AI models like OpenAI’s ChatGPT and Google’s Bard underscores the need to reduce data centers' carbon footprints. Future AI data centers will focus on low-carbon strategies to mitigate their environmental impact.

This cross-disciplinary PhD program in electrical engineering and computer science aims to develop real-time strategies for minimizing the carbon footprint of data centers. The research will optimize the jobs scheduling at large scale data center, addressing the challenge of managing high volumes of computational tasks required by large AI models. Advanced machine learning, optimization, and data analytics techniques will be explored to enable a more proactive response to the electricity supply profile, ensuring that data center operations are effectively integrated with the broader energy network.

A key component of the program is exploring how AI data centers can support energy networks, providing grid services like frequency and demand response during periods of low-carbon electricity or when the grid requires ancillary services. This research seeks to enhance the interaction between data centers and the electricity grid, promoting better coordination and contributing to grid stability and efficiency.

The solution will focus on developing practical, low-carbon data center operations for real-world deployment. By improving demand response services, data centers can operate more sustainably while helping stabilize the grid. This research is vital for addressing the energy demands of AI systems and ensuring that future AI developments align with global sustainability goals.

To apply for this position, It will be an advantage if applicants have relevant industry or research experience, or good programming skills using Python, Julia or matlab. Strong knowledge in one or more of the following areas is highly desirable:

• Power network modelling and control

• Optimization, data science, operational research and mathematical programming.

• Machine Learning and Reinforcement Learning

• Computer science with understanding of data center structure and operation

**NOTE: There is no closing date for this position, which will remain open until filled.  Early contact is highly recommended.

Reference:

   ‘Generative AI’s Energy Problem Today Is Foundational’ https://spectrum.ieee.org/ai-energy-consumption   Misaghian, M. Saeed, et al. "Assessment of carbon-aware flexibility measures from data centres using machine learning." IEEE Transactions on Industry Applications 59.1 (2022): 70-80.   Sarkar, Soumyendu, et al. "Real-time Carbon Footprint Minimization in Sustainable Data Centers with Reinforcement Learning." NeurIPS 2023 Workshop on Tackling Climate Change with Machine Learning. 2023.

The University of Edinburgh is committed to equality of opportunity for all its staff and students, and promotes a culture of inclusivity. Please see details here: https://www.ed.ac.uk/equality-diversity

Interested applicants are welcome to contact Dr Wei Sun by email for pre-application enquiries (W.Sun@ed.ac.uk).

Minimum entry qualification - an Honours degree at 2:1 or above (or International equivalent) in Electrical Engineering, Computer Science, Mathematics, or related areas. Possibly supported by an MSc Degree.

Further information on English language requirements for EU/Overseas applicants.

Applications are welcomed from self-funded students, or students who are applying for scholarships from the University of Edinburgh or elsewhere.

Further information and other funding options.