arxiv: 2605.07131 · v1 · submitted 2026-05-08 · ⚛️ physics.flu-dyn

Recognition: 2 theorem links

· Lean Theorem

A fast Physics-Informed Neural Networks based approach to the 2D design of turbine blades

Yuan Huang , Francesca di Mare

Authors on Pith no claims yet

Pith reviewed 2026-05-11 02:22 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn

keywords physics-informed neural networksturbine bladesturbomachineryaerodynamic screeningEuler equationsblade designcomputational fluid dynamicsNACA airfoils

0 comments

The pith

A progressive PINN framework screens turbine blade families at CFD-comparable accuracy across many conditions with one workflow.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a progressive Euler-PINN method that starts with simple tunnel flow without a blade and gradually adds the full outlet static pressure boundary condition. It pairs this with a dynamic loss-weighting scheme that places stronger penalties on residuals near highly curved blade surfaces. One trained model then covers ten NACA6 blade variants and thirty subsonic operating points. The predictions for pressure and velocity fields reach accuracy levels comparable to conventional CFD while cutting the cost of screening entire blade families. This positions the approach as a practical surrogate for early-stage 2D turbomachinery design and optimization.

Core claim

The authors establish that a single progressive Euler-PINN workflow, which gradually relaxes boundary conditions from tunnel flow without a blade to full outlet static pressure and applies a geometry-aware dynamic loss-weighting scheme, achieves CFD-comparable accuracy for pressure and velocity fields across ten NACA6 variants and 30 subsonic operating points, thereby reducing the computational cost required for family-wide blade screening.

What carries the argument

The progressive Euler-PINN framework, which relaxes boundary conditions step by step from tunnel flow to full outlet pressure while using geometry-aware dynamic loss weighting to intensify penalties near curved surfaces.

If this is right

A single neural network can evaluate entire blade families across many operating conditions without separate mesh-based simulations for each case.
Pressure and velocity predictions match conventional CFD results for subsonic flows on NACA6 airfoils.
Computational cost for large-scale screening of turbomachinery blades drops substantially compared with repeated CFD runs.
The method supports practical use in preliminary 2D blade pre-design and optimization for energy systems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The progressive training strategy may scale to three-dimensional blade geometries if the boundary-relaxation schedule is adapted accordingly.
Integration with automated shape optimization routines could let designers explore wider parameter spaces in less time.
Similar progressive conditioning might help PINNs converge on other complex internal-flow problems where standard training fails.
The reduced cost could enable real-time design iterations during early development of turbines and energy storage devices.

Load-bearing premise

Gradually relaxing boundary conditions from tunnel flow to full outlet static pressure, combined with dynamic loss weighting near curved surfaces, will produce reliable convergence and accuracy for complex blade geometries and off-design flows.

What would settle it

Running the trained PINN on a blade geometry or operating point outside the ten NACA6 variants and thirty subsonic points and comparing its pressure and velocity fields against independent CFD solutions to check whether errors exceed engineering tolerances.

Figures

Figures reproduced from arXiv: 2605.07131 by Francesca di Mare, Yuan Huang.

**Figure 2.** Figure 2: Main chain with special branch: 0610 to 1010 via 5 [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

**Figure 3.** Figure 3: Spatial distribution of the geometry-aware weight field (case 65-1010). Hottest [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗

**Figure 4.** Figure 4: Velocity contours for the three training strategies. Cold start is non-physical; [PITH_FULL_IMAGE:figures/full_fig_p016_4.png] view at source ↗

**Figure 5.** Figure 5: Airfoil 65-1010, α = 5◦ , pout = 0.80 bar. Rows: PINN prediction, SharC CFD, relative error. Columns: density ρ, Mach number M, static pressure p. 16 [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗

**Figure 6.** Figure 6: Surface pressure coefficient Cp for NACA65-1010 (α = 0◦ , pout = 0.8bar). PINN (solid blue) and CFD (dashed black) for pressure side (positive Cp) and suction side (negative Cp). large errors appear only near the leading edge and trailing edge where |Cp| approaches zero. 17 [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗

**Figure 7.** Figure 7: Training loss. Every loss bump refer to a boundary condition restart. [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗

**Figure 8.** Figure 8: CL vs. design lift m/10. PINN points are connected per (α, pout); CFD is shown as “×” without lines. NACA1010 with AoA 0 doesn’t have a steady state Euler solution Observations.. Our results highlight a clear hierarchy in how different designand operating-variables influence the accuracy of the PINN predictions. First and most significantly, blade geometry — in particular thickness and camber/curvature —… view at source ↗

read the original abstract

Rapid aerodynamic screening of turbomachinery blades across wide operating envelopes remains a major computational bottleneck in preliminary design, particularly for energy-conversion and storage systems such as emerging Carnot batteries. Physics-informed neural networks (PINNs) offer a mesh-free alternative to conventional CFD, yet convergence and accuracy often deteriorate for complex blade geometries and off-design flows. We propose a progressive Euler-PINN framework that (i) gradually relaxes boundary conditions from tunnel flow without a blade to full outlet static pressure, and (ii) employs a geometry-aware dynamic loss-weighting scheme that intensifies residual penalties near highly curved boundaries. To the best of our knowledge, this is the first study to deploy a single PINN workflow for large-scale, engineering-grade screening of turbomachinery blade families across multiple operating conditions, covering ten NACA6 variants and 30 subsonic operating points. The proposed framework achieves CFD-comparable accuracy for pressure and velocity fields while reducing the computational cost required for family-wide blade screening. These results establish the method as a practical surrogate for two-dimensional turbomachinery blade pre-design and optimisation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper shows a single progressive PINN workflow that screens ten NACA6 blades over thirty subsonic points with L2 errors against CFD and lower cost than full simulations.

read the letter

The main thing to know is that the authors combine progressive boundary relaxation—from tunnel flow without a blade to full outlet static pressure—with geometry-aware dynamic loss weighting, then run one PINN across a family of ten NACA 6-series variants and thirty operating points. The manuscript supplies training curves, the exact schedule and weighting details, and quantitative L2 comparisons for pressure and velocity fields on every case. That turns the CFD-comparable accuracy claim into something you can check rather than take on faith, and it directly addresses the convergence problems PINNs often hit on curved boundaries and off-design conditions. The work is internally consistent and reports no hidden fitting or circular steps. The central result holds up on the evidence given. The obvious limit is scope: everything stays two-dimensional and subsonic, with no transonic or three-dimensional tests. That keeps the method practical for early 2D screening but leaves the usual CFD handoff for real geometry or compressible regimes. The free parameters in the relaxation and weighting are stated clearly enough for reproduction. This paper fits readers who do preliminary turbomachinery design or who build PINN surrogates for fluid problems and want a worked engineering example with numbers. It is worth sending to a serious referee because the validation is quantitative, the method is described in usable detail, and the application is a clear step beyond generic PINN demos.

Referee Report

0 major / 3 minor

Summary. The paper proposes a progressive Euler-PINN framework for rapid 2D aerodynamic screening of turbomachinery blades. It combines gradual relaxation of boundary conditions (from tunnel flow without a blade to full outlet static pressure) with a geometry-aware dynamic loss-weighting scheme to improve convergence on curved boundaries. The method is demonstrated on ten NACA 6-series variants across 30 subsonic operating points, with claims of CFD-comparable accuracy in pressure and velocity fields and substantially reduced computational cost for family-wide screening.

Significance. If the reported accuracy holds, the work provides a practical mesh-free surrogate for preliminary blade design in energy systems such as Carnot batteries. Credit is due for supplying quantitative L2 error comparisons against reference CFD for all cases, training curves, and explicit implementation details on the progressive relaxation schedule and dynamic weighting; these elements directly address known PINN convergence challenges and support the large-scale screening claim.

minor comments (3)

§3.2: The precise functional form of the geometry-aware dynamic loss-weighting (e.g., how curvature is quantified and mapped to per-point weights) is described at a high level; a short pseudocode block or explicit equation would improve reproducibility.
Figure 7: The velocity magnitude contours for off-design cases show good visual agreement, but the color scale range differs slightly between PINN and CFD panels; aligning the scales would facilitate direct comparison of local discrepancies.
§5: The computational cost reduction is stated relative to a specific CFD solver; adding wall-clock times on identical hardware for one representative case would strengthen the engineering-grade claim.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of our progressive Euler-PINN framework and for recommending minor revision. The recognition of our quantitative L2-error comparisons, training curves, and explicit implementation details for the progressive relaxation schedule and dynamic weighting is appreciated, as these elements directly support the large-scale screening claims.

Circularity Check

0 steps flagged

No significant circularity detected in the PINN framework derivation

full rationale

The paper presents a progressive Euler-PINN method that combines gradual boundary-condition relaxation (tunnel flow to full outlet static pressure) with geometry-aware dynamic loss weighting. These are introduced as explicit algorithmic choices applied to existing PINN residuals, with training details, loss curves, and L2 error metrics reported against independent CFD reference solutions for ten NACA 6-series blades and thirty operating points. No equation or result is defined in terms of itself, no fitted parameter is relabeled as a prediction, and no load-bearing claim reduces to a self-citation chain. The novelty assertion ('first study') is a factual claim about scope rather than a derivation step. The framework remains self-contained against external CFD benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Based solely on the abstract, the framework rests on standard fluid assumptions plus two introduced procedural elements whose specific parameters are not disclosed.

free parameters (2)

progressive boundary relaxation schedule
The gradual change from tunnel flow without a blade to full outlet static pressure requires a schedule whose details and tuning are not provided.
dynamic loss weighting coefficients
The geometry-aware scheme that intensifies penalties near curved boundaries depends on weighting parameters whose functional form and values are unspecified.

axioms (2)

domain assumption The flow is adequately described by the 2D Euler equations in the subsonic regime.
The method is labeled Euler-PINN, implying inviscid flow assumptions standard for preliminary blade analysis.
standard math A neural network can be trained to satisfy PDE residuals and boundary conditions via a composite loss function.
Core premise of all physics-informed neural network approaches.

pith-pipeline@v0.9.0 · 5490 in / 1598 out tokens · 47759 ms · 2026-05-11T02:22:59.378404+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

progressive Euler-PINN framework that (i) gradually relaxes boundary conditions from tunnel flow without a blade to full outlet static pressure, and (ii) employs a geometry-aware dynamic loss-weighting scheme that intensifies residual penalties near highly curved boundaries
IndisputableMonolith/Foundation/BranchSelection.lean branch_selection unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

L_PDE = 1/N_e Σ |R_i(x_n;θ)|^2 , L_BC = Σ MSE terms, total L = ω_PDE L_PDE + ω_BC L_BC with fixed ω_BC=10

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages

[1]

Machine learning in fluid dynamics—physics-informed neural net- works (pinns) using sparse data: A review.Fluids, 10(9):226, 2025

Mjalled Ali, Philippe Miron, Martin Mönnigmann, Bukharin Nikolay, et al. Machine learning in fluid dynamics—physics-informed neural net- works (pinns) using sparse data: A review.Fluids, 10(9):226, 2025

work page 2025
[2]

Botarelli et al

T. Botarelli et al. Using physics-informed neural networks for solving navier–stokes problems in near real-world geometries.Comput. Methods Appl. Mech. Eng., 2025. online first / in-press (as of 2025)

work page 2025
[3]

Physics-informed neural networks for fluid mechanics: a review.Acta Mechanica Sinica, 37(12):1727–1738, 2021

Shengze Cai, Zhiping Mao, Zhicheng Wang, Minglang Yin, and George Em Karniadakis. Physics-informed neural networks for fluid mechanics: a review.Acta Mechanica Sinica, 37(12):1727–1738, 2021

work page 2021
[4]

A solver for subsonic flow around airfoils based on physics-informed neural networks and mesh transformation.Physics of Fluids, 36(2), 2024

Wenbo Cao, Jiahao Song, and Weiwei Zhang. A solver for subsonic flow around airfoils based on physics-informed neural networks and mesh transformation.Physics of Fluids, 36(2), 2024

work page 2024
[5]

Pi-Yueh Chuang and Lorena A. Barba. Experience report of physics- informed neural networks in fluid simulations: pitfalls and frustration. arXiv preprint, 2022

work page 2022
[6]

J. K. Cornelius, T. Ågren, N. Peters, et al. Palmo: An overflow machine learning airfoil performance database. Technical report, NASA Ames Research Center, 2024

work page 2024
[7]

Physics-informed neu- ral networks with adaptive localized artificial viscosity.Journal of Com- putational Physics, 489:112265, 2023

Emilio Jose Rocha Coutinho, Marcelo Dall’Aqua, Levi McClenny, Ming Zhong, Ulisses Braga-Neto, and Eduardo Gildin. Physics-informed neu- ral networks with adaptive localized artificial viscosity.Journal of Com- putational Physics, 489:112265, 2023. 23

work page 2023
[8]

Frate, A

Olivier Dumont, G.F. Frate, A. Pillai, S. Lecompte, M. De Paepe, and V. Lemort. Carnot battery technology: A state-of-the-art review.Jour- nal of Energy Storage, 32:101756, 2020

work page 2020
[9]

Physics-informed neural networks for solving reynolds-averaged navier–stokes equations.Physics of Fluids, 34(7):075117, 2022

Hamidreza Eivazi, Mojtaba Tahani, Philipp Schlatter, and Ricardo Vin- uesa. Physics-informed neural networks for solving reynolds-averaged navier–stokes equations.Physics of Fluids, 34(7):075117, 2022

work page 2022
[10]

Jagtap, Zhiping Mao, Nikolaus A

Ameya D. Jagtap, Zhiping Mao, Nikolaus A. Adams, and George Em Karniadakis. Physics-informed neural networks for inverse problems in supersonic flows.Journal of Computational Physics, 466:111402, 2022. arXiv:2202.11821

work page arXiv 2022
[11]

A surrogate-model based multidisciplinary shape optimization method withapplicationtoa2dsubsonicairfoil.Computers & Fluids, 36(3):520– 529, 2007

J-C Jouhaud, Pierre Sagaut, Marc Montagnac, and Julien Laurenceau. A surrogate-model based multidisciplinary shape optimization method withapplicationtoa2dsubsonicairfoil.Computers & Fluids, 36(3):520– 529, 2007

work page 2007
[12]

Characterizing possible failure modes in physics- informed neural networks.Advances in neural information processing systems, 34:26548–26560, 2021

Aditi Krishnapriyan, Amir Gholami, Shandian Zhe, Robert Kirby, and Michael W Mahoney. Characterizing possible failure modes in physics- informed neural networks.Advances in neural information processing systems, 34:26548–26560, 2021

work page 2021
[13]

Detached eddy simulation of a radial turbine operated with supercritical carbon dioxide.International Journal of Turbomachinery, Propulsion and Power, 10(4):43, 2025

Benedikt Lea, Federico Lo Presti, Wojciech Sadowski, and Francesca di Mare. Detached eddy simulation of a radial turbine operated with supercritical carbon dioxide.International Journal of Turbomachinery, Propulsion and Power, 10(4):43, 2025

work page 2025
[14]

Machine learning in aerodynamic shape optimization.Progress in Aerospace Sciences, 134:100849, 2022

Jichao Li, Xiaosong Du, and Joaquim RRA Martins. Machine learning in aerodynamic shape optimization.Progress in Aerospace Sciences, 134:100849, 2022

work page 2022
[15]

Key components for carnot battery: Technology review, technical barriers and selection criteria

Ting Liang, Andrea Vecchi, Kai Knobloch, Adriano Sciacovelli, Kurt Engelbrecht, Yongliang Li, and Yulong Ding. Key components for carnot battery: Technology review, technical barriers and selection criteria. Renewable and Sustainable Energy Reviews, 163:112478, 2022

work page 2022
[16]

Quantification of airfoil geometry-induced aerodynamic uncertainties— comparison of approaches.SIAM/ASA Journal on Uncertainty Quan- tification, 5(1):334–352, 2017

Dishi Liu, Alexander Litvinenko, Claudia Schillings, and Volker Schulz. Quantification of airfoil geometry-induced aerodynamic uncertainties— comparison of approaches.SIAM/ASA Journal on Uncertainty Quan- tification, 5(1):334–352, 2017. 24

work page 2017
[17]

A comprehensive review of advances in physics- informed neural networks and their applications in complex fluid dy- namics.Physics of Fluids, 36(10):101301, 2024

Wenqiang Lou et al. A comprehensive review of advances in physics- informed neural networks and their applications in complex fluid dy- namics.Physics of Fluids, 36(10):101301, 2024

work page 2024
[18]

Fundamental flaws of physics-informed neural networks and explainability methods in engineering systems.Computers & Industrial Engineering, page 111704, 2025

MZ Naser. Fundamental flaws of physics-informed neural networks and explainability methods in engineering systems.Computers & Industrial Engineering, page 111704, 2025

work page 2025
[19]

Geometry aware physics informed neural network sur- rogate for solving navier–stokes equation.Adv

Jan Oldenburg, Finja Borowski, Alper Öner, Klaus-Peter Schmitz, and Michael Stiehm. Geometry aware physics informed neural network sur- rogate for solving navier–stokes equation.Adv. Model. Simul. Eng. Sci., 9(1):1–15, 2022

work page 2022
[20]

Automatic differentiation in pytorch

Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Alban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. 2017

work page 2017
[21]

Investigationof physics-informed neural networks based solution techniques for internal flows

PascalPost, BenjaminWinhart, andFrancescaDiMare. Investigationof physics-informed neural networks based solution techniques for internal flows. InVolume 10C: Turbomachinery — Design Methods and CFD Modeling for Turbomachinery; Ducts, Noise, and Component Interac- tions, page V10CT32A011. American Society of Mechanical Engineers

work page
[22]

Physics- informed neural networks: A deep learning framework for solving for- ward and inverse problems involving nonlinear partial differential equa- tions.J

Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Physics- informed neural networks: A deep learning framework for solving for- ward and inverse problems involving nonlinear partial differential equa- tions.J. Comput. Phys., 378:686–707, 2019

work page 2019
[23]

Challenges in training pinns: A loss landscape perspective

Pratik Rathore, Weimu Lei, Zachary Frangella, Lu Lu, and Madeleine Udell. Challenges in training pinns: A loss landscape perspective. 2024

work page 2024
[24]

Ilya M. Sobol’. Distribution of points in a cube and approximate evalu- ation of integrals.USSR Computational Mathematics and Mathematical Physics, 7:86–112, 1967

work page 1967
[25]

Sukumar and A

N. Sukumar and A. Srivastava. Exact imposition of boundary condi- tions with distance functions in physics-informed deep neural networks. Comput. Methods Appl. Mech. Eng., 388:114333, 2021. 25

work page 2021
[26]

R. C. Swanson and S. Langer. Comparison of naca 0012 laminar flow solutions: Structured and unstructured grid methods.NASA Technical Memorandum NASA/TM-2016-219003, 2016

work page 2016
[27]

Naca 6-series airfoil profile: Nomencla- ture and generation rules

NASA / OpenVSP Tutorial. Naca 6-series airfoil profile: Nomencla- ture and generation rules. Online tutorial, VSP ground school, 2024. Accessed: 2025-12-02

work page 2024
[28]

Physics-informedneuralnetworkcompressionmech- anism for airfoil flows.Phys

Y.ZhangandX.Li. Physics-informedneuralnetworkcompressionmech- anism for airfoil flows.Phys. Fluids, 37(3):036115, 2025

work page 2025
[29]

A comprehensive review of advances in physics-informed neural networks and their applications in complex fluid dynamics.Physics of Fluids, 36(10):101301, 2024

Chi Zhao, Feifei Zhang, Wenqiang Lou, Xi Wang, and Jianyong Yang. A comprehensive review of advances in physics-informed neural networks and their applications in complex fluid dynamics.Physics of Fluids, 36(10):101301, 2024. 26

work page 2024