Projects per year
Abstract
solutions as inaccurate gradients, especially in high-dimensional simulations, can compromise optimization reliability, causing convergence issues. Obtaining exact derivatives analytically is often infeasible due to complexity of meanline models and equations of state, while numerical differentiation techniques like finite difference methods introduce inaccuracies due to truncation and round-off errors. To address these challenges, this paper introduces the first application of automatic differentiation in turbomachinery meanline optimization. Exact gradients were computed by differentiating an existing meanline model using the JAX library, and the performance of various gradient-based optimization solvers was compared by evaluating convergence with exact derivatives from automatic differentiation versus approximate derivatives from finite differences. Results for turbines involving 1-, 2-, and 3-stage configurations indicate that using exact gradients obtained using automatic differentiation significantly improves computational efficiency by reducing model evaluations by 25–50 % with respect to finite difference approximations, depending on turbine configuration and finite difference step size. additionally, the computational cost of gradient evaluations with automatic differentiation is significantly lower, as JAX optimizes code execution using Accelerated Linear Algebra. These findings demonstrate that the meanline model using automatic differentiation for gradient calculations leads to faster and more reliable convergence, paving the way for its use in complex flow problems like reversible and two-phase turbomachinery.
Original language | English |
---|---|
Title of host publication | Proceedings of the ASME Turbo Expo 2025 |
Number of pages | 12 |
Publisher | The American Society of Mechanical Engineers (ASME) |
Publication status | Accepted/In press - 2025 |
Event | ASME Turbo Expo 2025: Turbomachinery Technical Conference and Exposition - Memphis, United States Duration: 16 Jun 2025 → 20 Jun 2025 |
Conference
Conference | ASME Turbo Expo 2025 |
---|---|
Country/Territory | United States |
City | Memphis |
Period | 16/06/2025 → 20/06/2025 |
Keywords
- Design optimization
- Gradient-based
- JAX
- Jacobian
- Axial turbine
- Equation-oriented
Fingerprint
Dive into the research topics of 'Advancing Turbomachinery Meanline Modeling and Optimization with Automatic Differentiation'. Together they form a unique fingerprint.Projects
- 2 Active
-
Advanced mean-line models for two-phase turbines in partial-evaporation organic Rankine cycle systems
Diwanji, S. P. (PhD Student), Haglind, F. (Main Supervisor), Agromayor, R. (Supervisor), Desai, N. B. (Supervisor) & Bhairapurada, K. (Supervisor)
01/09/2024 → 31/08/2027
Project: PhD
-
Training42Phase: Next generation turbomachinery with two-phase flow
Haglind, F. (PI), Desai, N. B. (Project Manager), Persico, G. B. A. (Project Participant), Barros, A. (Project Participant), Bassi, F. (Project Participant), Schwingshackl, C. (Project Participant), Nowell, D. (Project Participant), Bergamin, A. (PhD Student), Cioffi, A. (PhD Student), Prakash Diwanji, S. (PhD Student) & He, J. (PhD Student)
01/12/2023 → 30/11/2027
Project: Research