Work

Data-driven surrogate models for forecasting experimentally measured fluid flows

Year: 2026

Type: article

Abstract: Data-driven modeling shows significant promise for faster-than-real-time forecasting of fluid flows. For real-world engineering applications (e.g., flow control), models must contend with limited, imperfect, and incomplete experimental measurements. However, most data-driven surrogates have been developed and demonstrated entirely within the computational domain. In this work, we present an analysis of data-driven surrogate models trained to forecast the time evolution of experimentally measured cylinder wakes. Using a dataset of two-dimensional, two-component particle image velocimetry measurements, we train fully convolutional neural networks, U-Nets, Fourier neural operators, and dynamic mode decomposition-based models to forecast the development of experimentally measured velocity fields. To characterize data-driven approaches contending with transient flow features and limited, imperfect observations, the development of predictions over extended forecast horizons is examined at a fixed Reynolds number (Re=590). Next, models are trained at a range of Reynolds numbers (Re=230–2920) to investigate the impact of increasingly turbulent and three-dimensional flow phenomena on forecast quality. Through a triple decomposition of measured and forecast velocity fields, we find that error is dominated by the fluctuating velocity component, with all surrogates exhibiting a decay in estimated planar turbulent kinetic energy. The characteristics of this decay are architecture-dependent. Regardless, we find that the experimentally trained surrogate models can provide meaningful predictions over short time horizons, propagate low-frequency dynamics over longer forecast periods, and achieve faster-than-real-time evaluation in the considered contexts. These results highlight the opportunities and challenges that remain for data-driven fluid dynamics in real-world, real-time engineering applications.

Source: Physics of Fluids

DOI: 10.1063/5.0354356

Authors: Peter I. Renn, Emily H. Palmer, Cong Wang, Morteza Gharib

Institutions: California Institute of Technology, Cooper Union, University of Iowa

Language: English

Topic: Model Reduction and Neural Networks

Subfield: Statistical and Nonlinear Physics

Field: Physics and Astronomy

Domain: Physical Sciences

Open Access status: hybrid

Funders: Center for Autonomous Systems and Technologies, National Science Foundation Graduate Research Fellowship Program

Award: G500318789