Reliably Detected Flaw Size

An NDE technique is limited by the minimum size of flaws that the technique can reliably detect, and as such is a measure of the capability of the technique. The reliably detected flaw size is a statistical measure of the smallest flaw can be reliably detected by a NDE technique. This flaw size is typically denoted as a90/95, which means that there is a 95% confidence that the probability of detection (POD) of the flaw is 90%. Since this means that only ~9 out of 10 flaws of this size may be reliably detected, in terms of the 10-9 probability of catastrophic failure per flight hour required by the FAA for commercial aircraft, a single NDE inspection may only contribute 1 order of magnitude to this probability. However, it is important to remember that NDE inspection is only one part of the process.

The reliably detected flaw size is determined though the statistical process of a POD study. Flaws smaller than the reliably detected flaw size are less likely to properly show indications during a test. The use of POD assumes that the system should be tolerant of flaws smaller than the a90/95. As a result, for fracture mechanics analysis in damage tolerant systems, it is assumed that there are flaws of the reliably detectable flaw size present, even if there are no indications from NDE tests.

Many factors influence the reliably detected flaw size, these are usually physical limitations of the method, or limitations of the reading device. For example, in eddy current NDE, the inspection frequency is correlated with the size of flaw that can be detected. In fluorescent penetrant inspections the flaw must have a large enough volume to hold enough penetrant to be visible to the eye after development. The POD also assumes that flaws larger than a90/95 should be easy to detect. However, this may not always be the case. For example, if the flaw is too large may expel the penetrant during development. Due to the many factors effecting the sensitivity of a test, the reliably detected flaw size must be determined for the specific testing parameters, materials system, and laboratory process. If these parameters for an NDE test are not all the same as used for determining the POD, the respective a90/95 is not applicable to that test.

NASA provides a guide for minimum detectable crack sizes based on their testing requirements in the document 5009B for a few specific cases. It is even acknowledged in the standard that "applying the crack sizes to complex geometries, other materials, material forms, material processes, and nonstandard NDE applications should be done with caution" and for cases of significant deviation from the conditions specified, minimum detectable flaw sizes need to be verified with documented evidence or additional tests. The table below serves as an illustration of some of the values that they consider for open surface partly through cracks for a flat fatigue cracked plate (Compiled from Tables 1 and 2, from 5009B pg.29-30).

Minimum Detectable (Surface) Crack Lengths for NASA standard NDE procedures in aluminum plates.1

NDE Technique

Part Thickness, 't'

Surface Half-Crack Length

with Minimum Center Depth

(in) (mm) (in) (mm) (in) (mm)
Eddy Current > 0.050 > 1.27 0.050 1.27 0.050 1.27
Penetrant > 0.075 > 1.91 0.075 1.91 0.075 1.91
Mag Particle > 0.075 > 1.91 0.125 3.18 0.075 1.91
Radiography > 0.107 > 2.72 0.7t 0.7t 0.7t 0.7t
Ultrasound > 0.100 > 2.54 0.065 1.65 0.065 1.65

1. Data compiled from Tables 1 and 2 of NASA-STD-5009B "Nondestructive Evaluation Requirements for Fracture-Critical Metallic Components". The values in the above table consider partially through surface cracks in flat, fatigue-cracked, 2219-T87 aluminum panels from early in the Space Shuttle Program and meet the a90/95 capability. These values should be interpreted as plausible and not as definite. They are representative of specific NASA procedures performed on NASA equipment.

References and Resources

  1. NASA-STD-5009B, Nondestructive Evaluation Requirements for Fracture-Critical Metallic Components, 05/08/2019,