TL;DR
This paper introduces the Mobile Acoustic Subsurface Sensing (MASS) system, a novel, non-contact approach for rapidly assessing subsurface road and bridge deck conditions. By leveraging air-coupled acoustic sensors and advanced signal processing, MASS enables continuous, walking-speed data collection, significantly improving efficiency and reducing traffic disruption compared to traditional methods.
What Problem Does This Paper Address?
Maintaining the health of our nation's infrastructure, particularly roadways and bridge decks, is crucial for public safety and economic stability. Traditional methods for inspecting subsurface pavement conditions, such as those relying on ground-contact sensors or visual surface defect detection, have significant drawbacks. These include being labor-intensive, requiring traffic disruption for sensor placement, and often failing to detect subsurface deterioration before it becomes severe. While techniques like Spectral Analysis of Surface Waves (SASW) are effective for estimating subsurface properties, their application in the field is often slow and cumbersome. This paper addresses the need for a faster, more efficient, and less disruptive method for subsurface infrastructure assessment.
Methods
The core of the MASS system is an adaptation of the Spectral Analysis of Surface Waves (SASW) method, reimagined for non-contact, mobile operation. Instead of traditional accelerometers, MASS utilizes an array of microphones to detect "radiating surface waves" – acoustic waves generated by the propagation of surface waves (Rayleigh waves) through the pavement.
Key methodological innovations include:
- Air-Coupled SASW: Utilizing microphones to capture acoustic signals radiated from surface waves, eliminating the need for direct contact with the pavement.
- Mobile Platform: Integrating the sensing equipment onto a stable, 3-wheeled cart, allowing for continuous data acquisition at walking speed.
- Electromagnetic Impact Source: Employing a controlled electromagnetic hammer to generate repeatable impact loads for exciting surface waves.
- Automatic Signal Extraction: Developing a pre-processing program that automatically identifies and windows the radiating surface wave signal, distinguishing it from direct impact noise using arrival time differences and coherence criteria.
- Fast Inversion Algorithm: Implementing a computationally efficient inversion algorithm that directly relates the dispersion curve to the layered shear velocity profile, minimizing the need for manual parameter tuning.
- De-Noising Hardware: Designing a sound barrier enclosure for the microphones to significantly improve the signal-to-noise ratio in noisy field environments.
Datasets
The paper does not explicitly mention training datasets. Instead, it focuses on the system's performance evaluation using data collected from field tests on:
- An asphalt pavement near Northeastern University.
- A concrete slab in a laboratory setting.
- A concrete bridge deck on the Pingree Bridge.
These field tests serve as the primary evaluation benchmarks, comparing the MASS system's estimated subsurface profiles against known or visually inspected ground truth (e.g., actual pavement thickness, rebar locations).
Key Insights and Ideas
- Acoustic Radiation as a Proxy: The realization that surface waves generate detectable acoustic waves in the air (radiating surface waves) is fundamental. This insight allows for non-contact sensing, a major leap in efficiency.
- Real-time, Walking-Speed Assessment: The integration of hardware and software enables continuous data collection and near real-time analysis, transforming infrastructure inspection from a point-based, time-consuming task to a rapid, mobile survey.
- Robustness in Noisy Environments: The development of both software-based (time windowing, coherence checks) and hardware-based (sound enclosure) de-noising techniques is critical for practical fieldwork. The sound enclosure, in particular, dramatically improves signal clarity.
- Sensitivity to Defects: The system demonstrated the ability to detect anomalies, such as a slab joint crack on the bridge deck, which manifested as localized, abnormally low modulus readings. This highlights its potential for identifying localized structural issues.
Key Results
The MASS system demonstrated successful estimation of subsurface profiles for asphalt pavement, concrete slabs, and bridge decks.
- Asphalt Pavement: Estimated an asphalt layer thickness of 0.1 m (3.94 inches) with an elastic modulus of 6.5 GPa (942.7 ksi), consistent with typical roadway design.
- Concrete Slab: Accurately estimated the total thickness of 20.32 cm (8 inches) and an average elastic modulus of 11.9 GPa (1,720 ksi). The system also showed sensitivity to the underlying wood beam foundation.
- Concrete Bridge Deck: Estimated a thickness of 0.23 m (0.75 ft), close to the designed thickness. The average modulus was 17.7 GPa (2580 ksi), with the system sensitively identifying a slab joint crack through localized low modulus readings.
- Signal-to-Noise Ratio (SNR): Achieved an SNR of around 20 dB in field tests, indicating good signal quality for analysis.
Limitations and Future Work
The authors acknowledge several areas for improvement and future research:
- Impact Source Refinement: Developing a better impact source with adjustable force and more robust contact shapes for various test targets is underway.
- Enhanced De-Noising: Further research into more efficient de-noising devices is needed to improve signal quality in extremely noisy environments (e.g., heavy traffic).
- Higher Modes and Multi-Channel Sensing: While the current two-channel system provides acceptable estimations, using multi-channel sensing could resolve higher modes for more detailed subsurface profiling.
- Program Modifications: The system, originally designed for pavements, showed potential for bridge decks with program modifications, suggesting further adaptation for diverse structures.
- Broader Field Testing: More extensive field tests are planned to further enhance the MASS system's practical capabilities and validate its performance across a wider range of conditions.
Summary updated