Fast Computation of Hydrodynamic Pressure in Lubricated Contacts: Which Lp Loss is Most Suitable for Physics-Informed Neural Networks Solving the Reynolds Equation?
DOI:
https://doi.org/10.13052/ijfp1439-9776.2542Keywords:
Tribology, physics-informed neural networks, elastohydrodynamic lubrication, pneumatic sealing, physics-informed machine learningAbstract
The frictional behavior of pneumatic seals significantly impacts the functionality of fluid power systems, particularly in fast-switching applications where precision and responsiveness are critical. However, the complex relationship between component properties and friction often makes experimental characterization infeasible or prohibitively expensive. To address this challenge, the Institute for Fluid Power Drives and Systems (ifas) developed the ifas Dynamic Seal Simulation (DDS), an iterative elastohydrodynamic lubrication (EHL) simulation capable of accurately solving the relevant partial differential equations (PDEs) [3]. Since iterative solvers are computationally intensive, neural networks have been explored as a more efficient alternative. While traditional neural networks offer computational advantages, they often lack the ability to understand the physical context of the systems they model, potentially limiting their accuracy and reliability. Physics-Informed Neural Networks (PINNs) have been introduced to overcome these limitations. PINNs integrate the governing physical laws directly into the training process, allowing them to grasp the system’s physical context. This approach opens up new possibilities, including more robust training and the ability to extrapolate beyond the training domain, thereby providing a more reliable and efficient tool for modeling the friction of seals in fluid power systems. In this paper, a previously validated hydrodynamic PINN framework [8, 9] is utilized to solve a variant of the averaged Reynolds equation across three scenarios: divergent, convergent, and curved gaps. The investigation focuses on four p-norm training metrics: L1, L2, L22, and L∞. The results indicate that the commonly used L22 metric is the most suitable for the scenarios examined.
Downloads
References
Andreas Almqvist. Fundamentals of physics-informed neural networks applied to solve the reynolds boundary value problem. Lubricants, 9(8):82, 2021.
Julian Angerhausen, Maik Woyciniuk, Hubertus Murrenhoff, and Katharina Schmitz. Simulation and experimental validation of translational hydraulic seal wear. Tribology International, 134:296–307, 2019.
Niklas Bauer, Matthias Baumann, Simon Feldmeth, Frank Bauer, and Katharina Schmitz. Elastohydrodynamic simulation of pneumatic sealing friction considering 3d surface topography. Chemical Engineering & Technology, 46(1):167–174, 2023.
Niklas Bauer, Andris Rambaks, Corinna Müller, Hubertus Murrenhoff, and Katharina Schmitz. Strategies for implementing the jakobsson-floberg-olsson cavitation model in ehl simulations of translational seals. International Journal of Fluid Power, 2021.
Niklas Bauer, Bekgulyan Sumbat, Simon Feldmeth, Frank Bauer, and Katharina Schmitz. Experimental determination and ehl simulation of transient friction of pneumatic seals in spool valves. Sealing technology - old school and cutting edge : International Sealing Conference : 21st ISC, pages 503–522, 2022.
Atilim Gunes Baydin, Barak A. Pearlmutter, Alexey Andreyevich Radul, and Jeffrey Mark Siskind. Automatic differentiation in machine learning: a survey. Atilim Gunes Baydin.
Rafael Bischof and Michael Kraus. Multi-objective loss balancing for physics-informed deep learning.
Faras Brumand-Poor, Niklas Bauer, Nils Plückhahn, and Katharina Schmitz. Fast computation of lubricated contacts: A physics-informed deep learning approach. International Journal of Fluid Power, 19:1–12, 2024.
Faras Brumand-Poor, Niklas Bauer, Nils Plückhahn, Matteo Thebelt, Silas Woyda, and Katharina Schmitz. Extrapolation of hydrodynamic pressure in lubricated contacts: A novel multi-case physics-informed neural network framework. Lubricants, 12(4):122, 2024.
Shengze Cai, Zhiping Mao, Zhicheng Wang, Minglang Yin, and George Em Karniadakis. Physics-informed neural networks (pinns) for fluid mechanics: a review. Acta Mechanica Sinica, 37(12):1727–1738, 2021.
R. Caruana. Multitask learning. Machine Learning, 1997.
Zhao Chen, Vijay Badrinarayanan, Chen-Yu Lee, and Andrew Rabinovich. Gradnorm: Gradient normalization for adaptive loss balancing in deep multitask networks. 2018.
Yiqian Cheng, Qiang He, Weifeng Huang, Ying Liu, Yanwen Li, and Decai Li. Hl-nets: Physics-informed neural networks for hydrodynamic lubrication with cavitation. Tribology International, 188:108871, 2023.
Salvatore Cuomo, Vincenzo Schiano Di Cola, Fabio Giampaolo, Gianluigi Rozza, Maziar Raissi, and Francesco Piccialli. Scientific machine learning through physics–informed neural networks: Where we are and what’s next. Journal of Scientific Computing, 92(3), 2022.
G. Cybenko. Approximation by superpositions of a sigmoidal function. Mathematics of Control, Signals, and Systems, 2(4):303–314, 1989.
Paul Escapil-Inchauspé and Gonzalo A. Ruz. Hyper-parameter tuning of physics-informed neural networks: Application to helmholtz problems.
Nathan Hess and Lizhi Shang. Development of a machine learning model for elastohydrodynamic pressure prediction in journal bearings. Journal of Tribology, 144(8), 2022.
A. Ali Heydari, Craig A. Thompson, and Asif Mehmood. Softadapt: Techniques for adaptive loss weighting of neural networks with multi-part loss functions.
Kurt Hornik, Maxwell Stinchcombe, and Halbert White. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5):359–366, 1989.
Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization.
Hyuk Lee and In Seok Kang. Neural algorithm for solving differential equations. Journal of Computational Physics, 91(1):110–131, 1990.
Liangliang Li, Yunzhu Li, Qiuwan Du, Tianyuan Liu, and Yonghui Xie. Ref-nets: Physics-informed neural network for reynolds equation of gas bearing. Computer Methods in Applied Mechanics and Engineering, 391:114524, 2022.
Max Marian and Stephan Tremmel. Current trends and applications of machine learning in tribology—a review. Lubricants, 9(9):86, 2021.
J. Močkus. On bayesian methods for seeking the extremum. In G. Goos, J. Hartmanis, P. Brinch Hansen, D. Gries, C. Moler, G. Seegmüller, N. Wirth, and G. I. Marchuk, editors, Optimization Techniques IFIP Technical Conference Novosibirsk, July 1–7, 1974, volume 27 of Lecture Notes in Computer Science, pages 400–404. Springer Berlin Heidelberg, Berlin, Heidelberg, 1975.
Houman Owhadi. Bayesian numerical homogenization.
Nadir Patir and H. S. Cheng. An average flow model for determining effects of three-dimensional roughness on partial hydrodynamic lubrication. Journal of Lubrication Technology, 100(1):12–17, 1978.
Nadir Patir and H. S. Cheng. Application of average flow model to lubrication between rough sliding surfaces. Journal of Lubrication Technology, 101(2):220–229, 1979.
Uma Maheshwera Reddy Paturi, Sai Teja Palakurthy, and N. S. Reddy. The role of machine learning in tribology: A systematic review. Archives of Computational Methods in Engineering, 30(2):1345–1397, 2023.
M. Raissi, P. Perdikaris, and G. E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019.
Maziar Raissi and George Em Karniadakis. Hidden physics models: Machine learning of nonlinear partial differential equations. Journal of Computational Physics, 357:125–141, 2018.
Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Inferring solutions of differential equations using noisy multi-fidelity data.
Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Machine learning of linear differential equations using gaussian processes.
Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Numerical gaussian processes for time-dependent and non-linear partial differential equations.
Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Physics informed deep learning (part i): Data-driven solutions of nonlinear partial differential equations.
Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Physics informed deep learning (part ii): Data-driven discovery of nonlinear partial differential equations.
Aravind Rajeswaran, Chelsea Finn, Sham M. Kakade, and Sergey Levine. Meta-learning with implicit gradients. CoRR, abs/1909.04630, 2019.
Mohamed Reyad, Amany M. Sarhan, and M. Arafa. A modified adam algorithm for deep neural network optimization. Neural Computing and Applications, 35(23):17095–17112, 2023.
Michael Rom. Physics-informed neural networks for the reynolds equation with cavitation modeling. Tribology International, 179:108141, 2023.
Robin M. Schmidt, Frank Schneider, and Philipp Hennig. Descending through a crowded valley - benchmarking deep learning optimizers.
Ozan Sener and Vladlen Koltun. Multi-task learning as multi-objective optimization. CoRR, abs/1810.04650, 2018.
Soumyendra. Singh, Dharminder Chaudhary, B. Yogiraj, Ram Narayan Prajapathi, and Saurabh Rana. Adam optimization of burger’s equation using physics-informed neural networks. In 2023 International Conference on Advancement in Computation & Computer Technologies (InCACCT), pages 129–133. IEEE, 2023.
Chuwei Wang, Shanda Li, Di He, and Liwei Wang. Is l2
physics-informed loss always suitable for training physics-informed neural network?, 2022.
Sifan Wang, Yujun Teng, and Paris Perdikaris. Understanding and mitigating gradient pathologies in physics-informed neural networks.
Yang Zhao, Liang Guo, and Patrick Pat Lam Wong. Application of physics-informed neural network in the analysis of hydrodynamic lubrication. Friction, 11(7):1253–1264, 2023.
