Magnetic flux leakage (TFI or Transverse Field Inspection technology) is a magnetic method of nondestructive testing that is used to detect corrosion and pitting in steel structures, most commonly pipelines and storage tanks. The basic principle is that a powerful magnet is used to magnetize the steel. At areas where there is corrosion or missing metal, the magnetic field “leaks" from the steel. In an MFL (or Magnetic Flux Leakage) tool, a magnetic detector is placed between the poles of the magnet to detect the leakage field. Analysts interpret the chart recording of the leakage field to identify damaged areas and to estimate the depth of metal loss.
As a MFL tool navigates the pipeline a magnetic circuit is created between the pipewall and the tool. Brushes typically act as a transmitter of magnetic flux from the tool into the pipewall, and as the magnets are oriented in opposing directions, a flow of flux is created in an elliptical pattern. High Field MFL tools saturate the pipewall with magnetic flux until the pipewall can no longer hold any more flux. The remaining flux leaks out of the pipewall and strategically placed tri-axial Hall effect sensor heads can accurately measure the three-dimensional vector of the leakage field.
Given the fact that magnetic flux leakage is a vector quantity and that a hall sensor can only measure in one direction, three sensors must be oriented within a sensor head to accurately measure the axial, radial and circumferential components of an MFL signal. The axial component of the vector signal is measured by a sensor mounted orthogonal to the axis of the pipe, and the radial sensor is mounted to measure the strength of the flux that leaks out of the pipe. The circumferential component of the vector signal can be measured by mounting a sensor perpendicular to this field. Earlier MFL tools recorded only the axial component but high-resolution tools typically measure all three components. To determine if metal loss is occurring on the internal or external surface of a pipe, a separate eddy current sensor is utilized to indicate wall surface location of the anomaly. The unit of measure when sensing an MFL signal is the gauss or the tesla and generally speaking, the larger the change in the detected magnetic field, the larger the anomaly.
The primary purpose of a MFL tool is to detect corrosion in a pipeline. To more accurately predict the dimensions (length, width and depth) of a corrosion feature, extensive testing is performed before the tool enters an operational pipeline. Using a known collection of measured defects, tools can be trained and tested to accurately interpret MFL signals. Defects can be simulated using a variety of methods.
Creating and therefore knowing the actual dimensions of a feature makes it relatively easy to make simple correlations of signals to actual anomalies found in a pipeline. When signals in an actual pipeline inspection have similar characteristics to the signals found during testing it is logical to assume that the features would be similar. The algorithms and neural nets designed for calculating the dimensions of a corrosion feature are complicated and often they are closely guarded trade secrets. An anomaly is often reported in a simplified fashion as a cubic feature with an estimated length, width and depth. In this way, the effective area of metal loss can be calculated and used in acknowledged formulas to predict the estimated burst pressure of the pipe due to the detected anomaly.
Another important factor in the ongoing improvement of sizing algorithms is customer feedback to the ILI vendors. Every anomaly in a pipeline is unique and it is impossible to replicate in the shop what exists in all cases in the field. Open lines of communication usually exist between the inspection companies and the pipeline operators as to what was reported and what was actually visually observed in an excavation.
After an inspection, the collected data is downloaded and compiled so that an analyst is able to accurately interpret the collected signals. Most pipeline inspection companies have proprietary software designed to view their own tool's collected data. The three components of the MFL vector field are viewed independently and collectively to identify and classify corrosion features. Metal loss features have unique signals that analysts are trained to identify.