Design and parameters optimization of an ultrasonic gas flow meter using simulation-assisted full factorial methodology

Tools

Gani, Adel and O'Leary, Richard and Galbraith, Walter (2026) Design and parameters optimization of an ultrasonic gas flow meter using simulation-assisted full factorial methodology. IEEE Sensors Journal. pp. 1-9. ISSN 1530-437X (https://doi.org/10.1109/JSEN.2026.3665814)

[thumbnail of Gani-etal-2026-Design-and-parameters-optimization-of-an-ultrasonic-gas-flow-meter]
Preview
 Text. Filename: Gani-etal-2026-Design-and-parameters-optimization-of-an-ultrasonic-gas-flow-meter.pdf
Accepted Author Manuscript
License: Creative Commons Attribution 4.0 logo
Download (899kB)| Preview

Abstract

Ultrasonic transducers are widely used in gas flow metering due to their high accuracy and absence of moving parts. However, designing high-performance transducers often requires extensive prototyping and time-consuming trial-and-error. This study presents a simulation-driven design approach that combines a full factorial Design of Experiments (DOE) with finite element analysis to streamline and optimise transducer development. A total of 208 transducer configurations were simulated using finite element software (OnScale), varying three key parameters: matching layer material, matching layer thickness, and the presence of a light backing layer. The active element was an 8 mm diameter, 1 mm thick PIC255 piezoelectric disc operating in radial mode at 238 kHz. The design objectives were to maximise acoustic pressure while minimising pulse duration, both of which are critical for gas flow measurement resolution. The optimal configuration identified was 1 mm grey matching layer (ρ = 681.6 kg/m³) combined with a 5 mm RTV 664 backing layer. Multivariate analysis showed that matching layer thickness was the most influential parameter for both amplitude (58.49%) and pulse duration (59.44%). Its interaction with material type and backing configuration also had a significant impact on performance. Experimental validation confirmed the simulation trends, yielding a measured peak voltage of 5.87 × 10⁻⁴ V and a pulse duration of 2.61 × 10⁻⁴ s, compared to simulated values of 6.54 × 10⁻⁴ V and 3.31 × 10⁻⁴ s. this corresponds to errors of 10.24% in amplitude and 26.82% in pulse duration. Overall, the proposed approach reduces reliance on physical prototyping while providing reliable performance prediction, offering a practical and scalable method for ultrasonic transducer optimisation in gas sensing applications.

ORCID iDs

Gani, Adel ORCID logoORCID: https://orcid.org/0000-0003-0558-3666, O'Leary, Richard ORCID logoORCID: https://orcid.org/0000-0002-4092-2101 and Galbraith, Walter;
CORE (COnnecting REpositories)