Acoustic emission (AE) is the term used for dynamic stress waves that are created within a material due to the application of a force.  Some examples are the sound of fibers breaking when a piece of wood is bent, high-frequency stress waves created when a crack grows in a metal structure undergoing mechanical fatigue, and the pulse of stress waves emanating from the impact site of a meteorite colliding with a spaceship hull.  AE differs from most of the other NDI methods in that no directed energy is put into the test object.  Whole-body forces create the localized stress waves that propagate through the test object to AE sensors.

AE NDI is done by placing multiple acoustic sensors on the object being inspected and then recording and correlating the signals generated when stress waves reach the sensors.  The sensors typically are responsive to acoustic frequencies between 50 kHz and 1 MHz.  The lower limit is important in order to limit acoustic noise, although it should be noted that common objects such as jingling car keys or grinding wheels produce acoustic energy above 100 kHz.  The upper limit is strongly dependent on the bandwidth of the AE sensor.  Occasionally, AE tests utilize sensors with the upper limit extending into the 2-3 MHz range.  The sensors are connected to AE instruments that amplify, filter, store, and process the signals produced by the sensors.  Typical results from AE tests are the number of AE “events” recorded; the energy, time, and duration of each event; and the location of the event within the test object.

Some advantages of AE NDI are: 1) the method is sensitive to stress waves emanating from anywhere within the test object; the sensors do not have to be focused or scanned across the object; 2) triangulation of the time of detection of the stress wave at different sensors allows identification of the location of the emission, and 3) sensors can be placed on objects with very limited access.

Disadvantages of AE NDI are: 1) the instrumentation is expensive, 2) appropriate signal processing to eliminate unimportant signals can be complicated, 3) large amounts of data often are generated, creating data storage problems.