RICERCA ITALIANA     

Research program

Integrated Methods and Algorithms for NonDestructive Evaluations of architectural heritage

University Co-ordinator

Università degli Studi di ROMA "La Sapienza" - SCIENZA E TECNICA DELL'INFORMAZIONE E DELLA COMUNICAZIONE (INFOCOM) - ROMA(RM)

Research Unit Leader

Raffaele PARISI

Description

There are five major factors that need to be considered in the design of a non destructive testing (NDT) system:
1. the required depth of penetration into the structure;
2. the vertical and lateral resolution required for the anticipated targets;
3. the contrast in physical properties between the target and its surroundings;
4. signal to noise ratio for the physical property measured at the structure under investigation;
5. historical information concerning the methods used in the construction of the structure.
Careful application of all the above factors may result in a NDT system able to provide the desired information about the structure of interest, or may recommend an alternative and more adequate approach.
The application of NDT techniques to the monitoring of structures in civil engineering has sometimes been disappointing. This has arisen from either using a method which lacked the precision required in a particular structural investigation or by specifying a method that is inappropriate to the problem under consideration. In other cases, the physical condition of the structure is far more complex than expected and hence interpretation of data does not yield the information desired.
Researchers in this Unit through the years have acquired a significant experience both in sensor array processing and in antenna design and construction. Proper application of array processing to the specific problem of NDT of architectural structures would allow to overcome the limitations of currently used techniques. The goal of this project is to exploit the knowledge acquired in the array processing field and to integrate this approach with those concurrently developed by the other Units.
In particular, this Unit will focus on methodologies referring to mechanical (sonic and ultrasonic) and electromagnetic waves.
More specifically, sonic and ultrasonic methods refer to the transmission and reflection of mechanical waves through a medium, at sonic and ultrasonic frequencies. Direct transmission involves the passing of a compressional wave at frequencies between 100 Hz and 10 kHz through the wall (or the structure) under investigation. Transmission of the wave is initiated on one side of the structure by the impact of a hammer, while reception is performed by an accelerometer whose position depends on the type of measurement desired (direct, semidirect, indirect). The resulting wave velocity is an average of the local velocity along the path and it is not possible to establish the position and the extent of any possible inhomogeneity. This format allows an approximated evaluation of the relative condition of the masonry. This method has been successfully used to evaluate material uniformity, detect the presence of voids, estimate the depth of surface crack, and calculate an average compressive strength for the structure or the material. The detection of flaws is possible due to the fact that sonic waves cannot transmit across an air gap, which could be due to a crack, void or delamination at the interface between brick or stone and mortar.
Sonic tomography represents an improvement in the sonic transmission test method because tests are performed not only in the direct mode but also along paths which are not perpendicular to the wall surfaces. This technique allows a 3D reconstruction of the velocity distribution across the structure or selected cross-section, so that local variations in velocity can be identified and correlated with zones of weakness or flaws in the internal fabric of the structure. It is usual to assume a linear structural response in the application of the tomographic method, so that any variation from the expected velocity is attributed to in-homogeneity in the structure or damage occurred.
A recent development of sonic/ultrasonic methods is the impact–echo test method, which was developed to measure concrete thickness and integrity. By proper transforms into the frequency domain, this method allows to estimate the transfer function and reflections or echoes of the compressional wave energy as pronounced resonant frequency peaks in the transfer function or frequency spectrum. These peaks correspond to the thickness or flaw depth resonant frequencies and knowing the compressional wave velocity in concrete or any other construction material the depth to the corresponding flaw can be calculated. Defects can be identified provided that a sufficiently high frequency is used (half wavelength with respect to the defect). The main limitation of this method is the frequent ambiguity of the measurement, due to the dispersion of the wave through the concrete for the presence of aggregate and other inhomogeneities, possible reduction in frequency of the impact–echo signal due to crumbling of the concrete surface, possible lack of sensitivity of the transducer.
Ultrasonic reflection is another common NDT method. Ultrasonic waves are generated by a piezoelectric transducer at frequencies above 20 kHz and propagate with a wavelength around 50–100 mm. In case of concrete and masonry, which have much higher attenuation characteristics, lower frequency signals are required to obtain a reasonable penetration.
Concerning electromagnetic methods, a very common technique is the impulse radar, that usually deploys higher frequency antenna (above 1 GHz) to obtain the resolution required. In some instances greater penetration of the electromagnetic energy will be required and lower frequency antenna in the range 100–500 MHz will be used. It is likely that the method will undergo significant development in this area over the next few years, due to the high number of possible applications.
Among electromagnetic methods, conductivity measures yield another tool for analysis. Conductivity depends on the electrical properties of materials and their content in water. Electromagnetic fields are propagated into the structure and variations are monitored and recorded. These provide geometrical and electrical information on the materials investigated and their degree of saturation. Problems can arise when reinforcing rods are present within a concrete structure, since they substantially alter the electromagnetic field in the interior.
GOALS
The goal of the present project is to develop new NDT techniques able to overcome the limitations of currently available approaches. In order to perform this task, measurement systems based on proper sets of multiple transducers (sources and sensors) will be designed according to array processing concepts. In particular, accurate and efficient modelling procedures of masonry and concrete structures will be developed and analyzed. Modelling will be performed by exploiting the availability of multiple measurements, that will allow to build matrices of transfer functions, depending on the positions of sources and sensors. Array processing algorithms will be applied in order to determine locations and properties of internal flaws with the desired accuracy. Specific aspects related with signal design, efficient calibration of the source-sensor system and minimum resolution required will be considered in detail. In order to precisely localize the discontinuities, high resolution subspace-based methods will be employed. Finally, obtained results will be compared to those of the other Units in terms of computational complexity and accuracy (benchmarking phase), in order to devise the best possible strategy for the data fusion phase.
RESEARCH PLAN
In order to accomplish the objectives, the research plan of this Unit will consist of the following phases, to be developed in parallel for most complex tasks:
FIRST YEAR
1. Initial phase of coordination among the Units.
The objective is to establish the requirements of the common platform. (2 months)
2 Methodological phase
This phase will be directed to the development of most proper methodologies for data modelling, acquisition and processing and will consist of the following steps:
a) structure modelling.
The goal of this step will be the determination of a proper model of the structure of interest. This phase will be developed in cooperation with Units IV and V for what concerns the study of mechanical and electromagnetic properties respectively. In particular mechanical and electromagnetic propagation in inhomogeneous materials will be analyzed, in the presence of unknown geometries. In addition, a common problem is the integration of historical information about the structure with the results of the non destructive test. Before the practical measurement, it is necessary to build a precise model on the basis of any additional information about the structure. All these aspects will be considered during the modelling phase. In particular, the cooperation with the Unit V will make available versatile and open finite elements codes, that will make it possible to analyze highly inhomogeneous media. On the other side, the experience acquired by this Unit in the field of acoustic modelling will be applied to effectively model the acoustic propagation in the structure. The availability of multiple transducers will allow to estimate the matrices of the transfer functions of every source-sensor pair. (5 months)
b) Analysis of data processing algorithms
In this phase the most appropriate data processing algorithms will be established and carefully studied, depending on the array geometry and the requirements specified in the modelling phase. In particular, the best array processing techniques for the estimation of signal propagation velocities (and of mechanical and electrical properties as a consequence) will be determined, under the assumption of known directions of arrival. The possibility of using high resolution techniques will be carefully investigated. These techniques should ensure better robustness with respect to model uncertainties and/or errors and the presence of multipath, due to reflections originated by discontinuities in the structure. (3 months).
c) Requirements of signal transmission and reception systems.
The requirements of the transmitter and the receiver will be specified, in order to ensure the best performance of the NDT system. Aspects to be considered will include waveform design, power requirements and radiation pattern in transmission, frequency band, sensitivity and directivity in reception. (2 months)
d) Determination of the optimal geometry of sources (sonic and electromagnetic) and sensors (antennas and microphones), performed by comparing the performance of all possible solutions. (2 months)
SECOND YEAR
3. Experimental phase
The goal of this phase will be the realization of an experimental testbed and will consist of the following steps:
a) acquisition of devices.
Based on the requirements specified during the second phase, the NDT system will be either realized or acquired. The performance of commercially available devices will be evaluated. In the past this Unit has built several types of antennas, whose adherence to the requirements has been extensively tested and verified. This project will consider both currently available antennas and different kinds (like microstrip or leaky wave antennas), that might be particularly feasible for the specific task of interest. The final choice will be taken after an accurate performance comparison. (3 months)
b) Data acquisition
The data acquisition system will be specified and set up. Data will be properly organized and made available to other Units through the net. (2 months)
c) Experimental tests and system tuning.
First tests will be performed in the department lab in order to assess the system performance on different kind of structures and in the presence of different kinds of known flaws. (2 months)
d) On-field measurements.
On-field measurements will be performed in agreement and cooperation with the other Units. (2 months)
4. Data fusion
During this phase methodologies developed by the Units participating in the project will be integrated. In particular, the use of multiple heterogeneous sensors will improve system capabilities with respect to more conventional solutions based on a single type of sensors. In the simplest case the use of multiple sensors will improve the reliability of the system, since the redundancy will allow to deal with the failure of one or more sensors. In a more general case, acquisition performed by a set of multiple heterogeneous sensors will be properly exploited to compensate for inaccuracies and the limited operating capabilities of individual sensors. Multisensor data fusion (MDF) will be the final step of the system. The various methodologies will have to properly fuse data and features of different nature, at different levels, by combining the information coming from sensors and comparing them with other available data (e.g. on database). (4 months)
5. Dissemination of results.
The project will include the development of a common software and hardware platform, able to validate the proposed methodologies on synthetic and real data. Main results of the research will be published on specialized journals, presented at conferences and made available on a dedicated website.
In conclusion, the verifiable objectives of the proposed research can be summarized in the following points:
- Development and analysis of multi-point models of the structures of interest (matrices of transfer functions).
- Application of array processing high resolution techniques to the accurate localization of flaws.
- Development of a software and hardware demonstrator.
- Dissemination of results by publication of papers on journals, talks at conferences, workshop organization and seminars.