Urban Water Network Design via Multi-Objective Reinforcement Learning

During this internship, you will apply Multi-Objective Reinforcement Learning (MORL) to model urban water network design trade-offs (such as balancing cost, pressure reliability, energy use, and emissions) and generate strategies that support utilities in making better long-term decisions.

Problem Description

Water distribution networks are dynamic infrastructures: demand and climate conditions evolve, assets deteriorate, and construction interventions are required to maintain reliable service over a lifecycle that spans several decades. Because the future is highly uncertain, water utilities need planning approaches that can adapt to new information and remain flexible across many possible scenarios. Deep Reinforcement Learning has recently shown promise in tackling such planning problems. However, in practice, utilities do not optimize for a single goal; they must balance multiple objectives such as cost, reliability, pressure, energy use, greenhouse gas emissions, and water quality.

In this project, you will apply Multi-Objective Reinforcement Learning (MORL) to model design trade-offs (such as balancing cost, pressure reliability, energy use, and emissions) and generate strategies that support utilities in making better long-term decisions.

What You’ll Do

As an intern, you will:

• Model the urban water network as a MORL environment using the mo-gym API.
• Train existing MORL algorithms on this environment and analyse their performance.
• Document your findings and contribute to research on applying AI to real-world sustainability challenges.
• Gain valuable work and research experience at KWR and TU Delft.

Requirements

We are looking for a motivated student with the following skills and interests:

• Strong Python programming skills.
• Basic knowledge of reinforcement learning and its implementations.
• Enrollment in a Master’s program in Computer Science, Mathematics, or Engineering (or a related field).
• Interest in water management and sustainability applications.

What You’ll Gain

• Hands-on experience applying state-of-the-art AI techniques to a real-world engineering challenge.
• Opportunity to work with an interdisciplinary team at the intersection of AI, sustainability, and water systems.
• Insights into multi-objective reinforcement learning research and its practical applications.
• Potential for contributions to academic publications.

Duration & Location

• Duration: 3 months
• Location: Hybrid
• Start date: Flexible

How to Apply

Please send your CV and a short motivation letter to:

• Zuzanna Osika — z.osika@tudelft.nl
• Lydia Tsiami — lydia.tsiami@kwrwater.nl