Design Optimization & Generative Engineering
The Design Optimization & Generative Engineering (DO&GE) Laboratory, founded by Prof. Luís Sá at the Department of Mechatronics and Mechanical Systems Engineering (PMR), Escola Politécnica da USP, is a research center dedicated to developing innovative solutions for complex engineering challenges through the integration of parametric optimization, topology optimization, and machine learning-based methods. Our focus is on multidisciplinary problems involving structural, thermal, and fluid flow analysis, aiming to maximize performance, efficiency, and sustainability.
The lab has produced over 30 peer-reviewed publications with 440+ citations, and collaborates with leading researchers at institutions including Imperial College London, Technical University of Denmark, and Laboratório Nacional de Computação Científica (LNCC).
Approach & Methodology
Combining classical optimization with AI-driven generative design to tackle multiphysics problems across structural, thermal, and fluid domains.
Classical Topology & Parametric Optimization
We apply state-of-the-art algorithms to redistribute material or tune design parameters, achieving highly efficient geometries that meet structural, thermal, or fluid dynamic requirements. Our methods include SIMP, TOBS (Topology Optimization of Binary Structures), topological derivatives, and level-set approaches.
Generative Design Powered by AI
Leveraging neural networks and deep learning, we explore vast design spaces beyond conventional intuition. By using limited information from past designs and focusing on boundary conditions, our generative methods produce novel, non-intuitive solutions that balance multiple objectives — such as stiffness vs. weight or pressure drop vs. flow rate.
Data-Driven Optimization
We develop surrogate models trained on historical design data, enabling rapid performance prediction and accelerating convergence to optimal solutions. This approach is especially effective for computationally intensive problems, such as heat exchanger or complex flow device optimization.
Research Lines
Topology Optimization of Fluid Flow Devices
Our flagship research line applies topology optimization to the design of fluid flow devices — from microfluidic channels to industrial-scale turbomachinery. We develop novel formulations using Integer Linear Programming (ILP), topological derivatives, and continuous boundary propagation models. Key contributions include:
- Topology optimization of fluid flow using ILP with binary design variables, eliminating grayscale issues common in density-based methods
- Continuous boundary condition propagation model for multi-physics topology optimization problems
- Geometry trimming algorithms that reduce seepage in optimized fluid devices
- Discrete adjoint sensitivity analysis for both incompressible and compressible flows
Fuel Cell Design (PEMFC & SOFC)
We apply topology optimization to redesign flow channels in Proton Exchange Membrane Fuel Cells (PEMFCs) and Solid Oxide Fuel Cells (SOFCs). Our work includes:
- Multilayered pseudo-3D optimization of two-phase nonisothermal PEM fuel cells, simultaneously optimizing anode and cathode channels
- Radial flow field optimization for circular PEMFCs with water management models
- SOFC channel layout design using a design variable propagation approach coupled with 3D multiphysics FEM modeling
Compressible & Turbulent Flow Optimization
We have pioneered topology optimization formulations for compressible subsonic flows and turbulent rotating flows:
- Subsonic compressible flow topology optimization using density-based material models coupled with compressible Navier-Stokes equations
- Turbulent flow optimization using the Spalart-Allmaras model adapted for rotating reference frames
- Applications to diffusers, nozzles, and aerodynamic components
Rotating Machinery & Labyrinth Seals
Design of rotors, Tesla-type pumps, and labyrinth seals using topology optimization:
- Rotor-stator device optimization with simultaneous multi-component design
- Tesla-type pump optimization based on 2D swirl flow models
- Labyrinth seal design considering subsonic compressible turbulent flow with experimental validation via 3D-printed prototypes
- Multi-objective formulations combining forward and backward flow optimization for seal performance (awarded best oral presentation at ETRI 2024)
Biomedical Devices
Application of optimization and generative design to medical devices:
- Design, optimization, manufacturing, and characterization of ventricular assist pumps (VADs)
- 3D-printed prototypes validated experimentally for flow and pressure head performance
- Patents for ventricular assist devices
Lab Numbers
| Metric | Value |
|---|---|
| Publications | 33+ |
| Citations | 440+ |
| Co-authors | 38 |
| Patents | Ventricular assist devices, CO₂ storage systems |
Join the Lab
We are always looking for motivated students and researchers to join our team. Current openings:
Undergraduate (Scientific Initiation & TCC)
We welcome students from Mechanical, Mechatronics, and related engineering programs interested in hands-on research. Available topics are listed on the Research page. No prior optimization experience is required — curiosity and commitment are what matter.
Requirements: enrolled at Poli-USP or partner institution, basic programming skills (Python or MATLAB), interest in numerical methods or machine learning.
How to apply: send your transcript and a brief statement of interest to Prof. Sá via the Contact page.
M.Sc. & Ph.D.
Graduate positions are available in all research lines: topology optimization of fluid devices, fuel cell design, compressible/turbulent flow optimization, labyrinth seals, and ML-driven generative design. Students are expected to develop original research, publish in international journals, and present at conferences.
Requirements: degree in Engineering, Physics, Applied Mathematics, or Computer Science. Strong background in at least one of: numerical methods / FEM, fluid mechanics, optimization, or machine learning. Programming experience (Python, C/C++, or MATLAB).
Funding: positions are typically funded through FAPESP, CNPq, or CAPES scholarships.
How to apply: email Prof. Sá with your CV, academic transcript, and a short research proposal aligned with one of the lab’s research lines.
Postdoctoral Researchers
Postdoc opportunities arise periodically, usually tied to funded projects (FAPESP, Shell/RCGI). Candidates with experience in topology optimization, computational fluid dynamics, or scientific machine learning are encouraged to reach out.
Supervision
Postdoctoral Supervision (Ongoing)
| Researcher | Start |
|---|---|
| Andre Luis Ferreira da Silva | 2024 |
Ph.D. Co-supervision (Ongoing)
| Student | Title | Start |
|---|---|---|
| Alberto Lemos Dúran | Topology optimization method applied to the design of compressor impellers for supercritical CO₂ considering resonance frequency and vorticity local constraints | 2025 |
| Gabriela Alves Barbosa | Simulation and Optimization of Solid Oxide Fuel Cells | 2024 |
M.Sc. Completed (Co-supervision)
| Student | Title | Year | Link |
|---|---|---|---|
| Lucas Neves Braga Soares Ribeiro | Experimental Study of Labyrinth Seal Designed by Topology Optimization | 2024 | USP Theses |
Examination Boards
| Type | Student | Title | Year |
|---|---|---|---|
| M.Sc. Defense | José Pereira Ramos Junior | Topology Optimization Based on Moving Morphable Components Applied to Passive Flow Control Devices | 2025 |
| Ph.D. Qualifying | Rômulo Luz Cortez | Numerical filters for topology optimization with binary design variables applied to rotormachinery | 2025 |
| Ph.D. Qualifying | Felipe Silva Maffei | A Methodology for Topology Optimization of Turbulent and Compressible Flows for Large-Scale Applications | 2024 |
| M.Sc. Qualifying | José Pereira Ramos Junior | Topology Optimization Based on Moving Morphable Components (...) | 2025 |