Research program
Applications of Methods of Diagnostics Electromagnetic (AMDE)
University Co-ordinator
Università degli Studi di CASSINO -
AUTOMAZIONE, ELETTROMAGNETISMO, INGEGNERIA DELL'INFORMAZIONE E MATEMATICA INDUSTRIALE - CASSINO(FR)
Research Unit Leader
Antonello TAMBURRINO
Description
This research programme follows on from the results achieved during the two-year PRIN research project "Methods and Applications of Electromagnetic Nondestructive Evaluation (MADEND)" granted in 2001. Specifically, this research programme is concerned with the methodological and numerical issues, as well as some experimental issues, in the nondestructive reconstruction of the resistivity profiles of a conducting material using techniques based on stationary and quasi-stationary electromagnetic fields.The research will be aimed at applications in the field of nondestructive testing of aircraft. Specifically, the research will focus on the reconstruction of the spatial behaviour of resistivity in a given volume and on crack detection and identification through eddy current testing (ECT) and injected currents (electrical resistance tomography, ERT) techniques.The aim of this research is to consolidate the aspects related to methodologies and applications, in order to properly design the measurement system, to develop fast and robust procedures for numerical simulation and resistivity retrieval and to investigate new imaging methodologies.The research program is made up of three activities related to the study, consolidation and development of methods and algorithms for the solution of the inverse and the direct problem, and the experimental validation of some of the proposed algorithms. Specifically, it will move from the results obtained during the MADEND project with the aim of reducing the gap between "ideal" configurations, for which the methods for electromagnetic non-destructive evaluation are developed, and real world configurations. It is worth noting that in most of the scientific literature, quantitative methods have been developed with reference to ideal configurations, i.e., where the geometry is known, the defect is described by using simple geometrical shapes and simple electromagnetic properties, etc. In practical applications, problem aspects such as uncertain geometry and sensor position, low inspection time, easily understandable results not requiring trained personnel, electromagnetic noisy environment, etc. are all relevant factors. Therefore, the present research programme, while not aiming to solve all the extremely complex problems, refers to these real world problems in order to guide the research activities.ACTIVITY 1. Consolidating and developing methods and procedures for the solution of inverse problems. This activity regards the study of critical aspects of the inverse problem, such as developing inversion methods and reconstruction algorithms for the application of interest for the project. Activity 1 is structured as follows:a. Analysis and development of non-iterative reconstruction methods and algorithms;b. Analysis and development of Q-transform based reconstruction methods and algorithms.ACTIVITY 2. Consolidating procedures for the solution of the forward electromagnetic problem and their adaptation to the requirements of the inverse problem. This activity regards the consolidation of numerical models and methods used for computing the fields as required for the forward problem. Particular attention is dedicated to those aspects that are critical to the solution of the inverse problem such as computational cost and accuracy. Activity 2 is structured as follows:a. Development and optimization of fast finite elements numerical solvers for field calculation;b. Development of simulation models for measurements sensors.ACTIVITY 3. Developing techniques and systems of electromagnetic non-destructive imaging for aeronautical applications. This activity regards the setting up of an experimental system for the non-destructive identification of defects for aeronautical applications. Activity 3 is structured as follows:a. Development of methodologies and sensors for measuring quantities of interest for the applications under study;b. Characterisation and optimisation of sensors for the applications under consideration.The first activity will be developed in close co-operation with the units of Reggio Calabria and Seconda Università di Napoli. Activity 1.a will focus on non-iterative inversion methods for both ERT and ECT techniques. These methods are attractive because they require the solution of a number of direct problems that increase only linearly with the number of pixels (voxels) representing the unknown.These methods, which will be studied in the framework of ECT and ERT, are based on a monotonicity property of the resistivity-data operator. The monotonicity, proved during the MADEND project [9-12] for particular configurations (volumetric defects, etc.) will be further investigated and extended with reference to the configurations of interest for the project (zero-thickness defect, magnetic materials, etc.), and numerical methods will be developed in order to estimate the maximum achievable resolution and regularization methods to integrate in the monotonicity based imaging algorithm. With reference to the latter issue (regularization) it is fitting that we mention that the imaging algorithm is based on a full non linear data-unknown model and that the imaging algorithm is not equivalent to a formulation of the inverse problem as a minimization of a proper functional.Activity 1.b focuses on the development of Q transform based inversion methods for ECT. Specifically, the activity will aim to consolidate and develop inversion methods based on the measurements of quantities that can be associated to the time-of-flight for a proper wave-propagation problem. Dually, it will develop inversion methods based on the measurement of quantities that can be regarded as phasors for an associated wave-propagation problem. Therefore, under suitable conditions, it will be possible for reconstruction methods originally developed for hyperbolic inverse problems to be used for parabolic inverse problems. During the MADEND project the problem has been addressed with reference to the localization of small defects in a homogeneous material infinitely extended in the three directions [15, 17]. Activity 1.b will deal with more realistic situations related to defects in confined conductors, such as plates and cylinders. The presence of air-conductor interfaces gives rise to a considerably more difficult problem. Activity 1.b is part of the more general framework of data fusion of data from different sensors. The completion of activity 1.b will allow, although this is not one of the project targets, data fusion to be achieved between diffusive measurements (ECT) and wave propagation measurements (ultrasonic testing).The second activity, which will be developed in close co-operation with the units of Reggio Calabria and Udine, consists of the development and consolidation of numerical models to solve the forward problem, which are optimised for the inversion methods and algorithms proposed in activity 1. Specifically, issues such as the accuracy and computational cost of the solution of the direct problem will be carefully considered. Activity 2.a concerns the development and/or consolidation of fast iterative numerical methods based on Fast Multipole Methods or Precorrected FFT to solve the forward problem. These methods will then be integrated with other methods developed during the MADEND project [21], [27-33], [41] which exploit the spatial localization of the defects to significantly reduce the computational cost. This will make it possible to develop an extremely efficient numerical method for treating small defects in conductors of arbitrary shape. Another issue that will be considered during activity 2.a is the development of numerical methods to treat real defects as opposed to artificial defects which can typically be schematised as an impenetrable barrier to the current density. Activity 2.b is focused on the numerical simulation of the sensors that will be used during activity 3. Specifically, we plan to use a sensor consisting of an array of coils on magnetic cores. This sensor is specifically intended to provide measurements that can be processed by using the imaging method of section 1.a. Summing up, the goal of activities 2 is the development of fast and accurate numerical models for measurement prediction. These models will be partly validated during activity 3.The third activity, which will be developed in close co-operation with the units of Reggio Calabria and Perugia, consists of the creation of innovative experimental systems based on the inversion techniques proposed with reference to crack identification for aeronautical applications. It will be developed experimental benchmark and prototypes to prove the feasibility of the proposed techniques. The methods, techniques and systems will be shared with the other research units.We will focus mainly but not exclusively on the prototypal tomographic probe (array of coils) proposed during the MADEND project [74]. Activity 3.a will be dedicated to the development and improvement of measurement methods and signal processing techniques to increase the signal to noise ratio, sensitivity, repeatability and also to reduce the acquisition and processing time. Activity 3.b will be dedicated to optimizing the probe design and characterizing the probe to increase its diagnostic capability. Specifically, the optimization of the probe design will be achieved both experimentally and by using the numerical tools (in part available from the MADEND project) that will be developed during activity 2. The new probe design requires the use of proper magnetic circuits to increase the defect-sensor interaction, the optimization of the geometrical configuration, operating frequency etc. Finally, taking into account that in a real application environment several sources of undesired electromagnetic noise are present, hardware and software will be developed to guarantee correct sensor operation in the presence of reasonable levels of electromagnetic interference. Finally, the probes and algorithms will then be integrated into an automated measurement station controlled by means of software environments for virtual instrumentation running on a personal computer.We highlight that all the research activities are complementary and strictly interconnected. These complementarities and interconnections have already been successfully experimented in the past, as can be seem from the bibliographic references. We wish to stress again that the research activities will be characterized according to the specific applications under consideration and taking into account the typical non "ideal" conditions of usage of non-destructive electromagnetic systems.The research activities will be able to benefit from using the resources available in the laboratories of the Cassino research unit, consisting of parallel computers, field probes and measurement instruments, acquired in part during the MADEND project.In addition, we would like to point out that all the various activities and phases will be finalized towards the geometries of interest for the applications under study. In particular, the canonical problem of aeronautical interest consists of identifying defects measuring less than a few millimetres, in plane and layered conductors in contact with air and presenting holes and rivets.With reference to the aeronautical field, the unit's researchers from the Alenia company will contribute to defining typical test cases, including the analysis of geometrical bounds and accessibility to the structures to be examined.
|
