ETD system

Electronic theses and dissertations repository


Tesi etd-05092011-233156

Thesis type
Tesi di laurea specialistica
email address
Modelling the magnetic response of dents in pipelines to improve the triaxial MFL inspection technique
Corso di studi
relatore Prof. Barmada, Sami
relatore Ing. Baccani, Roberto
Parole chiave
  • dents
  • pipelines
  • mfl
  • inspection
  • magnetic
  • response
  • flux
  • leakage
  • technique
  • technology
  • ili
Data inizio appello
Data di rilascio
Riassunto analitico
This thesis concerns the studies performed on the magnetic response of dents in pipes. The target is to define a method to characterise MFL signals from dents including effect of geometry and mechanical state.
This study is mainly based on FEA modelling.
A range of models were built and analysed with variations in geometry and magnetic properties. The correlation between magnetic properties and stress/strain was derived from previous studies. The models mentioned above were analyzed to investigate the correlation between dents geometry and magnetic response.
Correlations were found between geometry and modelled MFL signals.
However there were some discrepancies between the experimental data (MFL4 6” pull-throughs) and the modelled MFL signals. There are a number of possible causes for the discrepancies, the most relevant of which are listed below:
1 – Simplistic model assumption of sensor ride behaviour.
2 – The correlation mechanic vs. magnetic is not still well understood. This aspect can be ignored unless decoupling of signals is done.
3 – The experimental data showed asymmetry on the radial and severe noise on the transverse component, which could not be used at all in the analysis.
In order to partially remove discrepancies, the sensor ride should be studied through specific bench tests. Additionally, the magnetic properties due to strain/stress should be properly investigated by experimental tests.
The models do however qualitatively match the data available in literature [2]. Hence, the predicted signals should be compared with a more controlled bench-test, removing most of the possible causes of inconsistency between the model and pull-through data.