RICERCA ITALIANA     

Integrated Methods and Algorithms for NonDestructive Evaluations of architectural heritage

Politecnico di Bari

Abstract

The objective of this project is the realization of a system for the non-destructive diagnosis of important masonries, able to combine traditional and innovative diagnosis techniques with cooperative methodologies. The development of an integrated system for data acquisition and for information and knowledge management able to highlight the specified characteristics of the masonry will be developed; this automatic system will give objective and repeatable information about the state of the masonry and it will cluster and measure the typologies of the defects.
The main result of the project will be a software package which allows the user to evaluate the acquired data by a multisensor system.
The proposed project is constituted by three sections with strong interaction between all the 5 units. Each unit will process theoretical and experimental results which will be shared with the other units and integrated in a final software.
Optimization of traditional non-destructive diagnosis techniques. Some units will utilize investigation methodologies in which they are already experience, e.g. infra-red thermo-graphy (UR-1), sonic techniques (UR-4), ultrasonic techniques (UR-2) and they will do theoretical and experimental analyses, in order to individuate the testing procedures to supply the most efficient and effective information to individuate the state of an architectural work.
The expected results will allow the identification of typical defects of the masonries and the arrangement of a test masonry to be individuated. Moreover, a phase of pre-processing and feature extraction will allow data alignment and multisensor data collection.
Development of innovative ND methodologies.
Units 3 and 5 will develop modelling procedures for the masonry and for scalar and vector sensors to acquire radio-frequency data and they will design antennas and array for RF and sonic/ultrasonic applications.
The objective will be the definition and the implementation of a procedure for the optimized design of the transmitter/receiver system, in order to individuate the characteristics and the parameters that guarantee optimal performances related to reliability in individuating defects in the structure.
Numerical models realized by the units 3 and 5 will be validated by means of real data obtained by the previous investigations performed by units 1, 2 and 4.
Expected results: development of sensors and at the same time simulation based on numerical models, optimization of sonic/ultrasonic, scalar and/or vector sensors (array) to analyze masonries and implementation and validation of numerical models.
Integration and evaluation of the information. Structure diagnosis.
Soft Computing will be used to process together the information obtained by the multisensor system, realized by means of the heterogeneous sensors. The objective will consist in the realization of a procedure of Data Fusion and Decision Fusion for the diagnosis characterization of the object under test.
The results to be obtained at the end of this step will be the following: clustering of multi-dimension data to extract classes and to update the pre-processing techniques utilized by each unit, attuning of the results to realize the neural networks training, defects detection and classification based on cooperative clustering techniques and Data Fusion and Decision Fusion techniques. <<<

Principal Investigator

Giuseppe ACCIANI Politecnico di BARI

Research Objectives

The evaluation of both the preservation state and the vulnerability of the artistic and architectural heritage with respect to the structural risks, particularly the seismical one, is a most important issue in our country. Well-established methods to evaluate mechanical and structural characteristics cannot be used to analyze masonries, because in our country there are a lot of different masonries from a typological point of view as well as from a constructive one. Then, dedicated non-destructive techniques have to be used in order to evaluate the risks of both static instabilities and collapses due to a seism. For historical buildings a first investigation requires the knowledge of the materials and of the preservation state.
The methodologies of non-destructive diagnosis and control for masonries arise from many different fields of application and are based on various techniques, able to supply some needed information about the state of the structure under test.
The evaluation of the structure state is often made on the basis of incomplete or less reliability information; in these cases the expertise and the experience has an important role in giving a different weight to the available indexes in order to obtain final information.
Sometimes these techniques are not correctly applied because of the poor resolution and precision of the investigation results. These drawbacks depend on un-appropriated mathematical tools, on the subjectivity of the interpretation by the user and on the unavailability of the multi-sensor data.
The objective of this project is the realization of a system for the non-destructive diagnosis of important masonries, able to combine traditional and innovative diagnosis techniques with cooperative methodologies. The development of an integrated system for data acquisition and for information and knowledge management able to highlight the specified characteristics of the masonry will be developed; this automatic system will give information about the state of the masonry and it will cluster and measure the typologies of the defects.
Project activities are complementary and require a very strong integration and cooperation between the skills of each unit in order to complete the intermediate stages and reach the final targets. Experimental activity is the main project object. Each unit will develop its activity by means of different tools to provide a complementary and essential contribute for the whole project.
The data obtained by each unit will be shared with other units after the alignment of the multisensor data.
Moreover, the availability of all data deriving from several sensors will require a study step to classify the typical defects for each investigation diagnosis, in order to test the capability of the proposed system to overcome the limitations of each investigation technique.
The tests will require the realization of a devoted structure and the data acquisition with all the available sensors in prefixed conditions, as specified by each unit.
The main result of the project will be a software package which allows the user to evaluate the acquired data by a multisensor system.
The proposed project is constituted by three sections:
a) Optimization of traditional non-destructive diagnosis techniques. Some units will utilize investigation methodologies in which they are already experience, e.g. infra-red thermo-graphy and sonic/ultrasonic techniques. The objective is to individuate the testing procedures to supply the most efficient and effective information (feature) to individuate the state of an architectural work.
The results to be obtained at the end of this step will be the following:
− identification of typical defects of the masonries;
− arrangement of a test masonry;
− data acquisition by means of IR thermo-camera and sonic/ultrasonic tools;
− simulation based on numerical models;
− pre-processing and feature extraction of each measure typology;
− data alignment and multisensor data collection.
b) Development of innovative ND methodologies.
Some units will develop modelling procedures for the masonry and for scalar and vector sensors to acquire radio-frequency data and they will design antennas and array for RF and sonic/ultrasonic applications. The objective will be the definition and the implementation of a procedure for the optimized design of the transmitter/receiver system, in order to individuate the characteristics and the parameters that guarantee optimal performances related to reliability in individuating defects in the structure.
Expected results:
− development of scalar and/or vector (array) sensors to analyze masonries;
− designing of a finite element code suited in electromagnetic area as well as in ultrasonic one;
− simulation of ultrasonic and microwave analysis techniques applied to a masonry in order to individuate the most suitable frequencies to be utilized;
− design and develop of antennas and arrays with both RF and sonic/ultrasonic applications;
− Use of high resolution array processing techniques to locate carefully the defects;
− Implementation of an hardware/software prototype.
c) Integration and evaluation of the information. Structure diagnosis.
Soft Computing will be used to process together the information obtained by the multisensor system, realized by means of the heterogeneous sensors. The objective will consist in the realization of a procedure of Data Fusion and Decision Fusion for the diagnosis characterization of the object under test.
The results to be obtained at the end of this step will be the following:
− clustering of multi-dimension data to extract classes and to update the pre-processing techniques utilized by each unit;
− attuning of the results to realize the neural networks training;
− defects detection and classification based on cooperative clustering techniques;
− use of Soft Computing and Data Fusion techniques to detect and to classify the defects (unit 4);
− realization of decision platform. <<<

Timescale

24 months

National and international background

To preserve historical and artistic buildings several diagnosis systems have to be employed to provide information about the construction methods, the conservation state, the alteration and degradation phenomena that might have taken place. The information collection must be organized in a systematic and scientific way, to allow an accurate diagnosis of the present damages, of state of the components and materials in the architectonic structure. Structure defects must be detected, highlighted and measured so that further damages can be prevented.
If the diagnosis is based on sure data and in-depth research, it is possible to carry into effect precautionary measures and to restrict action on the mansory. The present trend employs the most effort to prevent the damages, therefore this approach must be able to monitor the decay. Moreover the monitoring should use as possible non-destructive techniques.
The data acquisition is carried out by different methods which include the human observation, geometrical relief and experimental method. The experimental methods allow to verify and deepen the acquired information by means of the first two investigations, in this way the data are analyzed beyond the surface. They can measure and discriminate: the masonry structure, the discontinuities, the layers, the materials, the chemical elements.
The basis for many non-destructive test (NDT) procedures arises from the medical, aerospace, and geophysical fields.
Useful methods for differentiating between regions of varying quality include sonic and ultrasonic techniques, Electromagnetic waves and infrared thermography analysis [1].
The sonic methods' studies have been centred on medical or material's engineering applications for laboratory testing [2,3]. The valuable handmade analysis has begun in the first of '90 [4-8] and nowadays it is performed using considerable approximations to the detriment of result's precision. The inaccurate results are also caused by the subjective methodology often used for interpretations. This contributed to form the common opinion that the use of such methodologies to analyze valuable masonry does not give reliable results. Conversely, the aforesaid technique could show notable diagnostic properties providing it with appropriate devices. In effect, the sounding surveys give information directly related to structure's elastic parameters, crucial for inspections of structures' stability
and lastingness. Main advantages of this method are recognised in fastness of acquisition and in high penetration depth, tanks to frequency range involved (1-20 kHz).
In several diagnostic techniques, such as sonic surveys, the shape of acquired wave can not be directly used for structure analysis. Therefore an Inverse Modelling is needed to extract the exploitable knowledge from measured data. The unit n.4 worked many years about these problems [9-14].
Moreover the Ultrasonic tests are very interesting with a frequency range between 20 kHz and 1000 MHz. Since gaseous media does not transmit such waves, they are used to identify micro-cracks that are able to reflect the wave's front; on the contrary, the ultrasonic signal is highly attenuated because of its small wavelength in comparison to the dimensions of the masonry components [15-17].
However, recently, the techniques based on ultrasounds have been introduced also in the non destructive analysis in different engineering fields. In particular, in the case of plants with systems of pipes, ultrasonic waves, guided by the walls of the pipes themselves, have been used. In this field the research unit 2 will make use of its experience. In the past years, the unit 2 has set up a two port equivalent model of guided waves systems for the NDT (Non Destructive Test) of not accessible pipes [18].
Concerning electromagnetic techniques, the wavefield reflected by discontinuities can provide additional useful information. By varying the frequency of the transmitted wave (usually between 100 MHz and 3 GHz) it is possible to explore the structure at different depths. Critical choices concern the frequency band of employed antennas (usually one in transmission and one in reception [19][20]), their positioning (usually manual, close to points of interest) [21], data interpretation and processing [22]. All these techniques are usually based on single sensors, while multichannel settings have been used only in a limited number of cases, mainly in reception [23]. For this reason the research unit 3 will investigate the array signal processing emerged in the last two decades with a number of significant applications [24].
Generally speaking, the goal of array processing is the estimation of the significant parameters of a physical phenomenon by integration of spatial and temporal information, by proper sampling of a wavefield. The wavefield is assumed to be generated by a finite number of emitters and contains information about the parameters of emitted signals. In this framework, a priori information about the data acquisition system (e.g. array geometry, sensor properties, etc.) can be properly exploited to improve the performance. Developed methods proved particularly effective in many practical applications, e.g. radar and sonar. As a matter of fact, array processing concepts and methodologies could be fruitfully applied to non destructive testing of architectural structures.
Infrared thermographic imaging provides a visible representation of infrared energy radiated by an object. Scanning with infrared cameras is a truly ‘global' approach, permitting rapid evaluation of large regions without requiring direct access to the wall. In a state of heat flux, differences in surface temperature will exist in the vicinity of materials with different densities, heat capacities, and/or thermal conductivities; these variations in surface temperature are measured with special cameras sensitive in the infrared range 0.76–30mm. Originally developed by the military in the 1960s, in recent years infrared thermography has seen wide application to evaluating features of masonry building envelopes [25-30], including:
subsurface anomalies such as voids, near-surfacecracks, or incipient spalls;
variations in wall construction;
missing or displaced wall insulation;
moisture rise by capillary action;
air leakage and variations in moisture content;
features hidden by surface plaster or frescoes, such as blocked openings or previous repairs;
internal cavities such as flues, ducts, or chimneys;
the presence of grouted cells in reinforced masonry construction;
thermal bridging of mortar obstructions in wall drainage cavities.
Infrared scans may be conducted by either an active or passive approach. Active thermography relies on homogeneous forced heating of the wall using an external heat source such as sunlight or a bank of heat radiators. Imaging during heating or cooling (after removal of the heat source) provides information on near-surface anomalies [31]. Passive investigations are more useful for locating defects deeper within the wall section, relying on a temperature differential across the wall section to develop steady-state heat transfer through the wall section. Through-wall temperature differentials of the order of 10 °C or more will generally provide a readily recognizable thermal pattern.
Interpretation of infrared images relies on interpretation by the expert user to determine the meaning of temperature anomalies. Operator experience is essential as well as an understanding of the physics behind heat transfer processes and the performance of the wall.. Under different heating and cooling conditions, for example, sections containing internal voids may show as either warmer or cooler regions. Temperature variations may also arise due to differences in material moisture content, surface texture, material emissivity, or reflections from nearby heat sources.
The IR thermography can result into a powerful mean for "in situ" and laboratory inspections. This technique has been recently strongly used in industry manufacturing, in masonry inspection systems, especially for bridges and civil building testing[32-35]. It offers several advantages if applied to historical and architectural relevant masonry structures as it results almost completely non invasive and simple in "in situ" application. It can result helpful in the evaluation of the static masonry behaviour, as it can offer immediately comprehensible images highlighting the areas of maintenance interventions, even if hidden under layers of surface refinement materials [36].
Almost all methods previously mentioned cannot be carried out with only experimental data due to the large number of variables that could yield an overwhelming number of data to be collected. In general numerical modelling can be used as a flexible tool for examining the generation, propagation and interaction of elastic waves in solid materials for nondestructive evaluation.
In this context it is important the activity in the last few years of the research unit 5 that has been mainly concerned with the theoretical and experimental study of numerical methods for the computation of electromagnetic fields [37-40].
The unit 5 intends to develop a finite-element code for the computation of both the electromagnetic field (EM) produced by the scattering of an EM wave generated by a transmitter antenna (generally working in the microwave range), placed in the proximity of a building structure, and the displacement field of an ultrasonic wave created by a suitable generator tool.
It's important to underline that no technique alone is able to highlight all the possible defects a general complex and composite masonry structure might experience. The need for integration of several measures and the knowledge's experts is obvious.
The research goal is the construction of an "integrated data management environment" whose objective is the architectonic constructions state monitoring.
Thus, the proposed integrated system will stress the redundancies of the information coming from several
sensors of different typologies to enhance the detection sensibility, and together will be able to use each peculiarity in different sensor data acquired to make the system both automatic and as general as possible.
In order to reach the target will be developed Soft Computing, Decision Fusion and Knowledge-Based Clustering techniques for the processing of data recorded by various ND systems. This research stage will take advantage by the expertise of research units 1 and 4 about fuzzy and neural data processing. Unit n.4 has obtained remarkable experience in neural techniques, both in learning algorithms [41, 42] and in diagnostic employment [43-48], obtaining founding for great number of projects by Public and Private institutes. Unit 1 will tackle the problem to analyze the data coming from different sensors by means of the cooperative clustering technique. <<<