TL;DR
This paper introduces a novel ultrasonic sensor array method for precisely locating continuous gas leaks in pressurized vessels with integral stiffeners. By analyzing the acoustic characteristics of the stiffener and employing frequency compensation with a time-domain superposition technique, the method achieves high accuracy, with a mean location error of just 13.51 mm on a 1m² plate.
What Problem Does This Paper Address?
Detecting and locating gas leaks in pressurized vessels, especially those used in critical applications like spacecraft, is crucial for safety and operational integrity. Integral stiffeners, while adding structural rigidity, complicate the propagation of acoustic signals, making traditional leak detection methods less effective. Existing methods often struggle with continuous broadband noise generated by leaks, leading to significant location errors or requiring complex, time-consuming manual scanning. This paper aims to bridge this gap by developing a more accurate and efficient method specifically for leak localization on integrally stiffened structures.
Methods
The core of the proposed method lies in leveraging ultrasonic signals generated by the leak. A piezoelectric ultrasonic sensor array is used to capture these signals. The key innovations include:
- Space-Time Correlation Analysis: Analyzing the correlation of signals across the sensor array to pinpoint the source direction.
- Frequency Compensation: Accounting for how the stiffener's geometry filters ultrasonic signals at different frequencies. This involves characterizing the stiffener's transmission coefficients, denoted as H(f), through experimental measurements.
- Superposition in Time Domain (SITD): A technique to reduce the influence of spurious peaks (fake-peaks) in the signal that can arise from reflections and the complex acoustic environment introduced by the stiffener. This helps stabilize orientation results.
The method combines these techniques to achieve a high-accuracy location on the stiffener wall.
Datasets
The paper describes experimental tests conducted on a custom-built test plate. This plate is a 1000 × 1000 × 2.5 mm magnesium-aluminum alloy sheet with integral stiffeners (4 mm width, 20 mm height, 197 mm spacing). Simulated leakage holes of varying apertures (1-2 mm diameter) were created on the plate. The experimental setup involved a signal generator to create ultrasonic signals, a vacuum pump to simulate leakage pressure, and a Polytec PSV-500 laser vibrometer for precise measurement of surface particle vibrations. Acoustic data was acquired using a 16-channel digital recorder. No pre-existing datasets were used; the experiments generated the necessary data for validation.
Key Insights and Ideas
- Stiffeners as Frequency Filters: The stiffener significantly influences ultrasonic wave propagation, acting as a geometry-dependent filter. This means certain frequencies are transmitted better than others, and this characteristic can be exploited for improved localization.
- Reflection Challenges: The presence of stiffeners leads to significant reflected waves, which can mask or interfere with the direct signal from the leak, posing a major challenge for accurate localization.
- SITD for Stability: The SITD technique is crucial for improving the stability of orientation results by mitigating the impact of transient reflections and noise, which can otherwise lead to false positives or inaccurate direction estimations.
- Combined Approach is Key: Neither frequency compensation nor SITD alone is sufficient. The synergistic combination of both is necessary to achieve the desired accuracy and stability in the presence of stiffeners.
Key Results
The proposed method demonstrated significant improvements in leak localization accuracy:
- Mean Absolute Error: 13.51 mm on a 1m² plate with integral stiffeners.
- Maximum Location Absolute Error: Generally within ±25 mm.
- Orientation Accuracy: When using both frequency compensation and SITD, the average absolute orientation error was reduced to 2°, with a variance of 1.7°. This is a substantial improvement compared to methods without these techniques. For instance, without H(f) and SITD, the average error was 10.9° with a variance of 211.1°.
- Probability of High Accuracy: The method achieved a probability of 92.5% for orientation error within ±5°, which translates to a maximum location error of less than 20 cm in the worst-case geometric scenario.
Limitations and Future Work
- Complex Acoustic Characteristics: The paper acknowledges that the acoustic characteristics of the stiffener are complex, and while the proposed method improves accuracy, some results still exhibit large errors. This suggests that a deeper understanding and more sophisticated modeling of these characteristics might be needed.
- Further Improvement Needed: The authors explicitly state that "a way to greatly improve the accuracy of orientation needs to be further researched." This indicates ongoing work to refine the method and address the remaining error sources.
- Comparison to Non-Stiffened Plates: While the method is effective for stiffened plates, the paper notes that the maximum location error (25 mm) is larger than that obtained on plates without stiffeners (which was around 10 mm in previous work by the same authors). This highlights the inherent difficulty introduced by stiffeners.
Summary updated