• DTDHandbook
• Contact
• Contributors
• Sections
• Examples

# Section 3.1.1.9. NDI Methods Summary

Figure 3.1.1 summarizes and compares attributes of the five principal non-visual NDI methods that are in widespread use.  This subjective comparison describes the types of defects that can be characterized, the structural applications, the advantages, and limitations of each of the methods.  For damage tolerance considerations, the key characteristic of an NDI system is the size of the flaws that can be missed when the system is applied in the field.  Quantifying inspection capability in terms of flaw size is referred to as inspection or NDI reliability.  Because of the many differences in material and geometry of structural details and the many approaches to the application of any of the methods, there is no single characterization of capability in terms of a reliably-detected crack size for any of the methods.  Further, because of the difficulty and cost of quantifying NDI reliability, relatively few capability demonstrations have been conducted.  Only very general statements can be made comparing the NDI reliability of the five methods.

Because of the random nature of inspection response to flaws of ostensibly the same size, NDI capability is characterized in probabilistic terms and estimated using statistical methods.  In particular, NDI reliability is quantified in terms of the probability of detection as a function of flaw size, POD(a).  There is no practical flaw size for which there is 100 percent assured detection.  For damage tolerance applications in the aircraft industry, it has become customary to characterize inspection capability in terms of the crack size for which there is a 90 percent probability of detection, the a90 crack size.  To reflect the statistical uncertainty in the estimate of a90, a 95 percent confidence bound can be calculated yielding the a90/95 crack size characterization of capability.  There is 95 percent confidence that at least 90 percent of all cracks of size a90/95 will be detected. The reliably detected crack size for a system is usually taken to be either a90 or a90/95.  Note that cracks smaller than a90/95 are readily detected by the NDI systems since POD(a) functions for production inspections increase over a relatively large crack size region.  Typically, the 50 percent detectable crack size is less than half the a90 crack size for a non-automated inspection.  Subsection 3.1.2 describes in considerable detail the approach to demonstrating NDI reliability for an application.

 Method Measures or Defects Applications Advantages Limitations Magnetic Particles Surface and slightly subsurface defects; cracks, seams, porosity, inclusions Permeability variations Extremely sensitive for locating small, tight cracks Ferromagnetic materials, bar, forgings, weldments, extrusions, etc. Advantage over penetrant in that it indicates subsurface defects, particularly inclusions Relatively fast and low costMay be portable Alignment of magnetic field is critical Demagnetization of parts required after tests Parts must be cleaned before and after inspection Masking by surface coatings Liquid Penetrant Defects open to surface of parts; cracks, porosity, seams, laps, etc. Through-wall leaks All parts with non-absorbing surfaces (forgings, weldments, castings, etc.)  Note:  Bleed-out from porous surfaces can mask indications of defects Low cost Portable Indications may be further examined visually Results easily interpreted Surface films, such as coatings, scale, and smeared metal may prevent detection of defects Parts must be cleaned before and after inspection Defect must be open to surface Ultrasonic (0.125 MHz) Internal defects and variations, cracks, lack of fusion, porosity, inclusions, delaminations, lack of bond, texturing Thickness or velocity Poisson’s ratio, elastic modulus Wrought metals Welds Brazed joints Adhesive-bonded joints Nonmetallics In-service parts Most sensitive to cracks Test results known immediately Automating and permanent recording capability Portable High penetration Couplant required Small, thin, complex parts may be difficult to check Reference standards required Trained operators for manual inspection Special probes Eddy Current (200 Hz to 6 MHz) Surface and subsurface cracks and seams Alloy content Heat treatment variations Wall thickness, coating thickness Crack depth Conductivity Permeability Tubing Wire Ball bearings “Spot checks” on all types of surfaces Proximity gage Metal detector Metal sorting Measure conductivity in % IACS No special  operator skills required High speed, low cost Automation possible for symmetrical parts Permanent record capability for symmetrical parts No couplant or probe contact required Conductive materials Shallow depth of penetration (thin walls only) Masked or false indications caused by sensitivity to variation, such as part geometry, lift-off Reference standards required Permeability variations Radiography (X-rays-film) Internal defects and variations; porosity, inclusions; cracks; lack of fusion; geometry variations; corrosion thinning Density variations Thickness, gap and position Misassembly Misalignment Castings Electrical assemblies Weldments Small, thin, complex wrought products Nonmetallics Solid propellant rocket motors Composites Permanent records; film Adjustable energy levels (5 kv-25 mev) High sensitivity to density changes No couplant required Geometry variations do not effect directions of X-ray beam High initial costs Orientation of linear defects in part may not be favorable Radiation hazard Depth of defect not indicated Sensitivity decreases with increase in scattered radiation

Figure 3.1.1.  Summary and Comparison of Principal Nondestructive Testing Methods [Walker, et al., 1979]

Table 3.1.1 presents approximate lower limits of reliably-detected crack sizes for the NDI methods in common use in the aircraft industry.  These limits are achievable on some structures by well-trained inspectors working in a good production environment.  Because the crack sizes of Table 3.1.1 represent the limits of the methods, such capabilities must be demonstrated before use in a damage tolerance based inspection schedule.  Note that most routine inspections are not designed for these target crack sizes.

Table 3.1.1.  Approximate Limits of Reliably Detectable Crack Sizes

 Method Location Dimension Size (in.) Eddy Current Manual Near Surface Length 0.030-0.040 Semi-Automated Near Surface Length 0.020-0.030 Automated Near Surface Length 0.005-0.010 Ultrasonic Manual Subsurface FBH* 0.032-0.064 Automated Subsurface FBH* 0.016-0.032 Fluorpenetrant Manual Surface Length 0.075-0.100 Automated Surface Length 0.060-0.075 Magnetic Particle Manual Near Surface Length 0.010-0.020

*FBH – capability based on flat bottom holes

There have been a number of demonstrations of NDI reliability for different structures and NDI methods.  An early compilation of such results can be found in Yee, et al. [1974], but the analysis methods for POD data were still evolving at that time and the quoted a90/95 values in this report are not compatible with those of more recent vintage. A major study sponsored by the United States Air Force was that of a program known as “Have Cracks, Will Travel” [Lewis, et al., 1978].  This study evaluated inspection capability at Air Force facilities and demonstrated the need for improving NDI reliability.  More recently, Rummel and Matzkanin [1997] have produced a data book that lists POD results for aluminum and titanium flat plates and panels and steel turbine engine bolt holes.  Among others, this data book contains the results of NDI demonstrations produced by the Aging Aircraft NDI Development and Demonstration Center at Sandia National Laboratories (see for example Spencer & Schurman [1995] and those of an AGARD round robin [Fahr, et al., 1995]).  A number of POD evaluations have been performed on the Retirement for Cause Eddy Current Inspection System (RFC/ECIS) for the inspection of turbine engine components but the results of these evaluations have not been released.

Another quantitative comparison of the various NDI methods is represented by the default reliably detected crack sizes that can be used in structural design. See, for example, NASA/FLAGRO Version 2.03, in which such default crack sizes are listed for 24 different crack types and the five common NDI methods. As an example of such default reliably detected crack sizes, Figure 3.1.2, from Rummel & Matzkanin [1997] and NASA/FLAGRO Version 2.03, presents one of the crack types and the corresponding default crack sizes.

Figure 3.1.2.  Standard NDE Flaw Sizes for STS Payloads – Edge Corner Cracks [Rummel & Matzkanin, 1997]