Acoustic emission monitoring of pipes; combining finite element simulation and experiment for advanced source location and identification
Abstract
Impact is a common source of damage in pipes and pipeline systems, detecting the
location and nature of damage is vital for reliability and safety of these systems. This work
sets out to assess the capacity of Acoustic Emission (AE) to monitor pipes and pipelines
for externally applied mechanical damage. AE is a non-destructive testing and monitoring
technique that relies on the propagation of elastic (stress) waves generated by impulsive
events such as particle impingement, cracking or fluid flow. These waves are recorded at
one or more sensors mounted on the surface of the object to be monitored. The key
scientific question was to determine the extent to which the structure of a non-impulsive
event could be reconstructed using sensors located on the external surface of a pipe.
The aim was to combine Finite Element simulations with a series of experiments in order
that the relationship between the generating event (source) and the resulting stress-time
history at a given point on the surface could be elucidated. Experiments and simulations
were carried out with impulsive sources (pencil-lead breaks) and dropped objects, the
latter being used to represent a non-impulsive event with a reproducible structure lasting
around one second. The AE resulting from these sources was recorded over a period of
around 2 seconds for both experiments and simulations. Two test objects, a solid
cylindrical steel block of diameter 307mm and length 166mm and various lengths of pipe
of diameter 100mm and wall thickness 10mm were used, the former to provide a relatively
simple and well-studied platform to examine a number of essential principles.
The work on the solid cylinder first validated the simulation of the stress wave from an
impulsive source and identified the main modes present, by comparing with analytical
solutions. Then it was possible to identify the part of the experimental time series record
at a given sensor which is uncontaminated by reflections from the edges and surfaces of
the cylinder. The dropped object measurements on the solid cylinder provided clear
records of the first and subsequent impacts as the dropped steel balls recoiled and returned
back to the surface. There was a clear relationship between the measured AE energy and
the estimated incident energy of the dropped objects at a range of timescales irrespective
of contamination by reflections.
The work on the pipe sections formed the main series of systematic experiments. First it
was established that an unloading time in the simulations of around 10-8
seconds gave a reasonable representation of the frequency structure of experimentally observed stress
waves. It was also observed from both experiments and simulations that a low amplitude
wave travelling at around 5500ms-1 was the first to arrive at any surface sensor. The
structure thereafter was complex, probably involving reflections from the inner wall of
the cylinder and geometric interference as the wave spreads around the circumference of
the pipe. The key finding of this aspect of the work is that the AE line structure of an
impulsive source can be reproduced by simulation for short times, for longer times, the
damping associated with reflections would require to be measured and introduced into the
simulations in order to fully represent the real practical simulation. The degree of damping
is important in making a cumulative assessment of multiple impulsive sources.
The dropped objects on the pipe confirmed that a mechanical disturbance which is
extended in time can be identified from its energy-time imprint carried on the stress wave.
The analysis was carried out at three different timescales; short (initial interactions free of
reflections), medium (first contact including recoil) and long (involving several bounces).
Generally, for medium and short timescales, the AE energy varied with drop height and
mass consistently with existing models for balls on plate. For multiple bounces, the
behaviour was more erratic probably due to the imprecise control of ball contact point.
The simulations of AE worked well at medium and long timescales, providing an idealised
framework unto which could be added effects of restitution and damping. At the short
timescale, the twin challenges of time and spatial resolution meant that a solution could
not be obtained within the limitations of the computing power available.
It is generally concluded that AE monitoring can be used to identify the nature of a
mechanical disturbance on the surface of a pipe. Suggestions for future work include
improvements to the simulations to include attenuation and to better simulate the
dynamics of mechanical interactions at the surface, and extensions to the experiments to
cover the effect of internal and external pipe environment and the use of mechanical
sources which involve actual pipe damage.