Page 438 - Pipeline Pigging Technology

Interpretation of pig survey results

will have been processed digitally in real time, securely coded against errors,
and organized in a particular format for acceptance by the on-board tape
recorder. Clearly, early validation of the data, to confirm the successful
operation of the system, is essential. This is a complex task in view of the huge
quantities of data involved, and has demanded major developments in
microcomputer-based test equipment for its completion. Following the
confirmation of a successful survey run, the magnetic tape, containing the
inspection data, is returned to the British Gas Computer Centre in England for
detailed analysis and interpretation.

Interpretation of inspection data

At the On-Line Inspection Centre, the data recorded on tape during the
inspection run is replayed via a process-control type of computer on to
standard computer tapes, which can then be analysed using one of the
Centre's five main computers. These machines reformat and reorganize the
data so that information from the various types of sensor is properly aligned
and correlated with positional data.
The next process is to reject signals from normal, defect-free pipeline
fittings such as welds and bends. Each fitting gives a particular shape of signal
which can be identified, checked and then eliminated. If existing pipeline
maps resulting from previous inspection runs are available, these are also used
to verify and reject data. Significant sensor data is then presented on an
electrostatic plotter, and interpreted by trained operators. This form of
output allows many parallel sensor traces to be plotted and quickly analysed.
Finally, a mathematical sizing model, used in conjunction with a computer
graphics terminal, is employed to obtain a direct estimate of the size and shape
of defects. This system is complemented by a comparative sizing technique
based on an automatic search through a large library of known signals.
Inspection data must be preserved for comparison with subsequent
inspection logs and as a historical record. The scale and frequency of
inspection operations demand that data analysis must be a highly-automated
process. The keys to rapid and reliable data analysis are defect sizing
capability, and the ability to recognize and classify automatically the signals
which characterize particular pipeline fittings. When such a signal is identi-
fied, it is necessary to check that the fitting is not faulty in some way, for
example to check that a weld between sections of pipe has not become
corroded. The integrity of each fitting must be verified, but the obvious
approach of comparing new signals with standard examples works only in a
limited number of cases.

419