| 19 Mar 2026
Aerodynamic evaluation of wind speed sensor placement on UAVs for meteorological applications
Abstract. Vertical takeoff and landing Unmanned Aerial Vehicles (VTOL-UAVs) can provide accurate, highly resolved, and repeatable atmosphere measurements, especially in scenarios where conventional measurement techniques are inadequate, or impractical. Wind estimation using UAV-mounted sensors is significantly influenced and contaminated by rotor-induced flows, making its placement a critical design consideration for meteorological measurements. This work investigates optimal locations for wind speed sensor placement on various UAV systems by studying the rotor-induced flow field using a free-vortex wake model (FVM) across a single-rotor, quad-rotor, and hexa-rotor configurations under hover, axial descent, and forward-descent flight conditions. This model is validated with an in-house experimental setup for velocity measurements. Care is taken to ensure these results are applicable across a wide range of practical UAV operating conditions through the dimensional disk loading (DL) parameter. The rotor-induced velocity fields are evaluated on multiple planes perpendicular to the rotor disk, and "quiet-zones" for sensor placement are identified based on a threshold of 1% rotor-tip speed. Results reveal that the location and extent of the quiet-zones are strongly dependent on the flight condition. For single-rotor in hover, a well-defined quiet-zone exists above the rotor disk, while viable sensor placement locations are substantially reduced in axial descent. Forward descent introduces asymmetric wake skew, limiting quiet-zones to the upstream, at smaller axial distances. For multi-rotor configurations, the system geometric center is particularly suitable for sensor placement, compared to axial locations about the individual rotor hubs. Overall, by analyzing various rotor systems in different flight conditions, the present work provides a practical guidance for the design of accurate UAV-based atmospheric measurement systems.
Review of "Aerodynamic evaluation of wind speed sensor placement on UAVs for meteorological applications" by Pastay, Narayanan, and Govindarajan
Recommendation: Accept with Minor Revision
This paper addresses a practically important and under-explored problem in UAV-based atmospheric measurement: the identification of aerodynamically quiet regions around VTOL drone systems within which wind-speed sensors can be reliably mounted without contamination from rotor-induced flow. The topic is timely given the rapid proliferation of UAV platforms for boundary-layer meteorology and the acknowledged gap in publicly available, systematic guidance for sensor placement on commercial multi-rotor configurations. The authors make a genuine and well-executed contribution to filling this gap, and I recommend acceptance subject to a minor revision described at the end of this review.
The methodological choice to use a free-vortex wake model (FVM) rather than high-fidelity Computational Fluid Dynamics is well-motivated and carefully justified. The authors correctly identify that CFD, while accurate, is computationally prohibitive for the kind of broad parametric sweep — across rotor configurations, flight conditions, and disk loading values — that is needed to produce practically generalizable guidance. The FVM occupies an appropriate middle ground: it captures the blade-tip vortex dynamics and rotor-rotor wake interactions that are critical to quiet-zone identification, while remaining tractable enough to explore the full parameter space the study requires. The decision to exclude viscous effects and vortex breakdown is acknowledged, and the authors correctly note that this limits model accuracy below the rotor disk plane, where wake turbulence dominates. Since the region below the rotor is in any case unsuitable for sensor placement on practical multi-rotor platforms — due to fuselage obstruction — this limitation does not undermine the paper's central conclusions.
The validation strategy is sound. The FVM was previously validated against thrust and power data for the APC single-rotor in Narayanan and Govindarajan (2024), and the present work meaningfully extends this by validating predicted induced velocity distributions against in-house TriSonica ultrasonic anemometer measurements. The agreement above the rotor plane — precisely the region of interest for sensor placement — is good, with deviations of approximately 0.5% of tip speed, well within the 1% threshold used to define quiet zones. The authors also present the discrepancies that emerge below the rotor plane and provide a physically coherent explanation for them. The additional validation cases presented in Appendix B, spanning the full-scale Harrington coaxial rotor in hover, the APC propeller across RPM ranges, and the Carpenter-Fridovich unsteady pitch-rate experiment, collectively build a convincing case that the FVM implementation is reliable across a broad range of operating conditions and rotor scales.
The choice of disk loading as the scaling parameter for cross-vehicle generalization is well-argued, and the appendix justifying it over the conventional blade-loading coefficient for electrically-driven variable-RPM UAVs is a thoughtful addition that will be useful to readers from the meteorological community who may be less familiar with rotorcraft aerodynamics. The demonstration in Figure A1 that three rotors spanning dramatically different physical scales produce nearly coincident normalized induced velocity profiles at the same disk loading provides strong empirical support for the approach.
The results themselves are clearly presented and physically well-interpreted. The finding that hover represents the most restrictive flight condition for single-rotor quiet-zone extent — and is therefore the appropriate design case — is both intuitive and important. The characterization of axial descent as practically eliminating viable sensor locations, and of forward descent as introducing asymmetric wake skew that confines quiet zones to the upstream side, provides nuance that has not previously been captured in a systematic, scalable form. For multi-rotor configurations, the identification of the geometric centre as consistently preferable to positions above individual rotor hubs is a practically actionable result, particularly given that the difference in minimum quiet-zone height between the geometric centre plane and the most favorable inter-rotor plane is modest enough that structural and stability constraints are likely to dominate the final placement decision anyway. The paper's engagement with the existing literature is thorough: Wilson et al. (2022), Jin et al. (2024), Ghirardelli et al. (2023, 2025), and Thielicke et al. (2021) are accurately represented, and the paper correctly identifies the key limitation of prior studies — namely, their restriction to specific platforms without a generalized, scalable framework — as the gap the present work addresses.
The one minor revision I would suggest concerns the multi-rotor FVM predictions. While the single-rotor FVM results are validated against in-house experimental data, the multi-rotor quiet-zone predictions rest on the computational model alone. This may be okay at first sight since they have tested against the single rotor configuration. Wilson et al. (2022), which the authors already cite, provides experimental PIV and field measurements of rotor-induced velocities above a quadcopter in hover and reports a quiet-zone separation distance of approximately 5.3 rotor diameters — a figure that invites at least qualitative comparison with the present FVM predictions. I would encourage the authors either to include a brief discussion of how their multi-rotor results compare with Wilson et al. (2022), even acknowledging the differences in disturbance metric and platform, or, to briefly explain why such a comparison is not straightforward. This would further strengthen reader confidence in the multi-rotor results without requiring any additional computations or experiments.
In summary, this is a well-written, carefully reasoned, and practically valuable paper. The assumptions are appropriate and clearly stated, the methodology is well-suited to the problem, the results are physically consistent and usefully general, and the literature context is handled with accuracy. I recommend publication following the minor revision noted above.