In our latest paper, we introduce a cost-effective offline method for optimizing sensor placement to reconstruct complex flow fields. This technique eliminates the need to insert physical probes during the preliminary phase, saving time, resources, and preserving flow integrity.
We propose a method that works by using row surrogate sensors derived from a single, non-time-resolved PIV dataset. Leveraging Taylor’s hypothesis, these surrogates accurately simulate sensor behavior before deploying real probes. We also apply a mask to the surrogate data, ensuring only high-correlation regions are used for maximum accuracy. The technique offers advantages:
- Offline Optimization: Find the ideal sensor positions using only one PIV dataset—no real probes required for planning.
- High Accuracy: The surrogate-based placement rivals and often outperforms traditional, online-optimized methods.
- Broad Application: The technique is effective across planar, stereoscopic, and volumetric PIV (including tomographic setups).
Precise sensor placement is critical for accurate data-driven flow reconstruction. Our method drastically cuts experimental overhead by allowing you to pre-plan optimal sensor layouts without expensive, trial-and-error physical experiments. This ensures optimal estimation quality from the start.
Read the open access publication, co-authored by Junwei Chen, Marco Raiola, and Stefano Discetti here. You can also access the data sets in Zenodo, as well as the code used on Github.
