Corrosion evaluation by thermal image processing and 3D modelling

Ermanno Grinzato, Vladimir Vavilov

Research output: Contribution to journalReview articlepeer-review

34 Citations (Scopus)


Quantitative transient IR thermography has been applied to the characterization of hidden corrosion in metals. A dedicated 3D numerical model of heat transfer has been used to solve the direct thermal problem and to simulate the test. Theoretical modelling allows the verification of limits of the 1D solution and the derivation of coefficients which take heat diffusion into account. An analysis of inversion accuracy was carried out. A simple algorithm based on a surface temperature time-derivative is proposed for detecting thickness variations. Then, material loss in an area of arbitrary shape is evaluated applying a modified algorithm, originally developed for a 1D thermal model. The potential of dedicated image processing in enhancing a signal-to-noise ratio is explored. The feasibility of corrosion quantification by the proposed inversion algorithm is demonstrated with experimental results. Detection and evaluation of hidden material loss within a boiler section, typically used at a power plant station, has been performed. The external surface was heated with flash lamps and temperature response was analyzed both in time and space domains. The masking effect due to the noisy inspected surface (not painted and affected by a long time service) were substantially removed before evaluating corrosion. Obtained results have been compared with measurements produced by the ultrasonic method.

Original languageEnglish
Pages (from-to)669-679
Number of pages11
JournalRevue Generale de Thermique
Issue number8
Publication statusPublished - 1 Jan 1998


  • Heat exchanger
  • Metal corrosion
  • Quantitative infrared thermography
  • Thermal modelling

ASJC Scopus subject areas

  • Chemical Engineering(all)

Fingerprint Dive into the research topics of 'Corrosion evaluation by thermal image processing and 3D modelling'. Together they form a unique fingerprint.

Cite this