Flow reconstruction accuracy and cost vary by method and turbulence type
FlowReconBench: Accuracy, calibration and cost of learned and classical flow reconstruction from sparse sensors
Machine Learning
Summary
Measuring the full pattern of flowing fluids like air or water is hard when sensors are few and far between. This paper shows that different computer methods work better depending on how complex the flow is and how many sensors there are. Classical techniques are good when sensors are many and the flow is simple, while modern learning-based methods shine when sensors are scarce or the flow is complex. The authors also explore how reliable and costly these methods are, finding trade-offs between speed and prediction quality.
What this means in practice
- •For engineering simulation teams: Improve fluid flow predictions from limited sensor data across different turbulence scenarios, choosing the best method to balance accuracy and cost.
- •For environmental monitoring professionals: Reconstruct incomplete flow fields around environmental sensors to better understand air or water movement when direct measurements are limited.
Authors
Nicolas Tricard, Linzheng Wang, Jason Chen, Sili Deng
Abstract
Reconstructing full flow fields from sparse sensors requires prior assumptions about unmeasured regions and physical variables. Yet it remains unclear when learned models outperform classical estimators and which aspects of their predictions can be trusted. FlowReconBench evaluates ten learned and five classical methods on three regimes: a laminar cylinder wake across Reynolds numbers, two-dimensional Kolmogorov turbulence and a large-scale three-dimensional (3D) isotropic turbulence. Methods share leakage-controlled splits and sensor draws and are scored on per-channel error, ensemble spread, calibration, spectra and cost. The leading method changes for each regime. Classical methods perform strongly in low-order and dense-sensing regimes. On the cylinder wake, linear stochastic estimation and gappy proper orthogonal decomposition lead, including for unobserved pressure. In 3D, inverse-distance weighting matches the best learned methods on observed channels. Data-driven methods become superior at decreasing sensor density or poor sensor observability. On Kolmogorov flow, continuous ranked probability score and error yield different method rankings: generative models offer broader but underdispersed ensembles, while deterministic models prioritize pointwise error. Generative models balance this tradeoff through their sampler budget, with a larger budget improving ensemble quality at higher computational cost. In 3D, scaling tests demonstrate that queryable methods are flat in memory, while full-grid construction methods run out of memory. Observing only 2 of 4 fields leaves the problem kinematically underconstrained, yet a latent model can reproduce structural similarities to the ground truth.