Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 93
PROCEEDINGS OF THE TENTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STRUCTURES TECHNOLOGY
Edited by: B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru and M.L. Romero
Paper 61

Computational Simulation of a Non-Destructive Testing Technique for Steel Bar Sizing in Concrete Structures using Eddy Current Testing and Neural Networks

N.P. de Alcantara Jr1, T. Marques1, D.C. Costa1, D. Guedes1, G.P. Sanchis2 and P.S. da Silva2

1Electrical Engineering Department, 2Civil Engineering Department,
São Paulo State University, Bauru/SP, Brazil

Full Bibliographic Reference for this paper
N.P. de Alcantara Jr, T. Marques, D.C. Costa, D. Guedes, G.P. Sanchis, P.S. da Silva, "Computational Simulation of a Non-Destructive Testing Technique for Steel Bar Sizing in Concrete Structures using Eddy Current Testing and Neural Networks", in B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru, M.L. Romero, (Editors), "Proceedings of the Tenth International Conference on Computational Structures Technology", Civil-Comp Press, Stirlingshire, UK, Paper 61, 2010. doi:10.4203/ccp.93.61
Keywords: non-destructive testing, eddy current test, artificial neural networks, finite element methods, reinforced concrete.

Summary
The inspection of the constitutive elements (steel bars) of reinforced concrete structures is a particularly difficult problem in non-destructive testing (NDT) area, because they are completely immersed in an opaque and solid media. In the past decades several NDT techniques have been used in the evaluation of the general conditions of reinforced concrete, in particular the eddy current testing (ECT) [1].

The paper is an investigative study in order to develop a methodology of NDT for reinforced concrete structures using ECT techniques and artificial neural networks. These steps were followed in this work:

  1. An electromagnetic device was proposed to generate and sense time varying electromagnetic fields of the concrete structure region.
  2. Test bodies were proposed for the tests, considering seven sizes of steel bars and four concrete classes.
  3. Computational simulations were done using the finite element program FEMM [2], considering fine variations of the bar position, the movement of the electromagnetic device on the concrete surface and seven frequencies for the excitation electric current.
  4. The simulations were automated by using the LUA scripting, an embedded language incorporated into the finite element program.
  5. The results obtained with the simulations were then used to generate training vectors for the perceptron multi-layer artificial neural networks. The input vectors were constructed with the real and imaginary components of the induced voltages at the sensor coils and the output vectors were constructed with the size bar number and the level position of the bar within the concrete.
  6. The networks were exhaustively trained using a Dell T3500 workstation. 280 training vectors were generated for each frequency. 245 vectors were used to train the ANN and the 35 remaining vectors were used in the validation of the networks. A strategy to accelerate the training process was adopted.
The results obtained were very encouraging. The success in the identification was within 90% for the bar size, and 85% of the level position of the bar, considering all the seven frequencies.

Electromagnetic devices have been produced and measurements with real concrete structures are in progress.

References
1
O. Yokota, "Study of Reinforced Bars Detection Buried in Concrete Structures Using Eddy Current Method", 15th WCNDT, URL, 2000.
2
D. Meeker, "Finite Element Magnetic Method - User Manual, Version 4.2", 2006.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description