Master's thesis on scalable, gradient-based aiming point optimization for commercial solar tower power plants
Deutsches Zentrum für Luft- und Raumfahrt e.V.
ThesisGerman needed: not said
Still open on 3 NovKey facts
- Pay
- Paid by a fixed federal public service pay scale
What you'll do
- Analyze how the runtime of ARTIST-based aiming point optimization changes as the number of heliostats grows
- Develop and evaluate strategies to reduce runtime: heliostat clustering, gradient calculation in batches and hierarchical optimization methods
- Enable optimization runs for complete commercial fields within computing times suitable for plant operations
- Investigate clustering and grouping strategies and measure their effects on the quality of the aiming point solution
- Investigate how shading and blocking effects can be integrated into the aiming point strategy and actively offset
What you need
- Very good Python skills, particularly experience with PyTorch
- Strong interest in physical modeling and numerical optimization
- Enjoyment of collaborative, structured software development
- An advantage: A preceding required internship at DLR or willingness to undertake one
- An advantage: Experience with high-dimensional optimization problems or ray tracing
What you get
- Work with ARTIST, a fully physical, differentiable ray tracer validated at the Jülich solar tower
- Take part in a coordinated research project with two parallel master's theses and clearly defined research questions
- Shared repository, coordinated interfaces, mutual code reviews and presentations of interim results directly to the ARTIST development team
- Your work contributes to an internationally recognized research tool
- Option to complete a preceding required internship to adapt the tower, receiver and field geometry
- Research-level PyTorch: automatic differentiation, GPU optimization and scaling scientific simulations to real-world problem sizes
- Learn how to approach high-dimensional, physically coupled optimization problems, a core skill useful well beyond solar technology
- Collaborative scientific software development: code reviews, shared interfaces and structured version control, as in a real development team
- Diversity, respect and equal treatment for everyone
- Independent work and individual personal and professional development
- Numerous training and further education opportunities
- Equal opportunities and support for women in science and leadership
- Preference for applicants with severe disabilities who meet the professional requirements