Learning Objectives
By the end of this section, students will be able to:
- Describe basic nondestructive testing (NDT) technologies.
- Describe process control, process monitoring, and in-situ NDT.
- Understand how NDT methods relate to certification and qualification.
- Understand nondestructive testing methods and how they relate to certification.
Most of the parts and materials used in the world are made using robust, repeatable processes with minimal post-fabrication inspections. In industries where the consequences of failure can be high (aerospace, turbomachinery, nuclear, medical, oil and gas, high-value costly to replace automotive), parts are often subjected to nondestructive testing, which is process of inspecting components or assemblies for discontinuities, or differences in characteristics without destroying the serviceability of the parts, to see if they contain any defects that would cause failure while in service. Thus, the use of NDT to ensure freedom from defects that could cause in-service failure is integral to the certification of many systems.
While more detailed definitions exist, for the purposes of this book, a defect is a discontinuity that may compromise the ability of a part to meet design intent. While it can be expected that the above industries will continue to use NDT for AM parts, the lack of long-term service history may cause other industries, such as automotive, to consider the use of NDT until manufacturing and service history provides the data needed to eliminate it. One unique aspect of AM is that since the parts are built in layers, the opportunity presents itself to inspect the part as it is built, known as In-Situ AM. Related to this is the ability to gather and analyze significant amounts of data while a part is being built, which provides unprecedented ability to perform Process Monitoring or Process Control while the part is being built.
Post-build NDT, often referred to as conventional NDT, is divided into volumetric and surface methods. Volumetric methods look for discontinuities in the interior of a part or material, while surface methods look for discontinuities on the surface of a part. Each of these methods have their own attributes, with surface methods generally considered more critical. Surface discontinuities tend to have a greater impact on performance than subsurface ones due to higher stress concentrations and the potential for crevice corrosion. It can be imagined one day, that in-situ NDT of AM may address both volumetric and surface inspections. In the following paragraphs, each of the inspection technologies will be briefly described with relation to AM, followed by an example.
Radiographic Inspection
This original method of NDT is almost exclusively used for volumetric inspections and dates back to the days of Roentgen. It can be used for the full range of AM materials, and works by passing X-rays (although some systems use Gamma rays or neutrons) through a part and onto a detector. Cracks and pores are found because they have less matter to absorb the X-rays so more hit the detector. Inclusions (undesirable foreign material) are found because they either absorb more X-rays (higher density) or fewer X-rays (lower density). General practice is that when more X-rays hit the detector, the image is darker, and when fewer hit the detector, the image is brighter. It can be inferred that for a discontinuity to be detected, it must be oriented so that it has some thickness in the direction of the X-ray beam. This means that planar discontinuities, such as cracks, may be difficult to detect, unless one knows their preferred orientation.
While photographic film was the preferred detector until the 1980s, more and more industries are now using digital detectors to capture the image. These offer significant benefits in terms of time, sensitivity, and ability to analyze the image. The most sophisticated, and expensive, radiographic method is known as computed tomography (CT). In this method, the part is rotated while being continually exposed to the X-rays and the detector continually collects the data. The data is then processed to generate a 3D reconstruction of the part. In addition to being used to detect and characterize discontinuities, CT can also generate a solid model for comparison with the actual design. This extra capability can be quite useful in AM to determine if a complex part with internal passages meets dimensional requirements where conventional metrology would not be able to inspect.
For the purposes of this discussion, we will say the discontinuities of interest are pores, inclusions, and cracks. Additionally, it is necessary to know whether internal passages are the correct distance from the surface of the part. A series of radiographic shots, as shown below, could be used to look for pores and cracks, with a typical sensitivity being 2% of material thickness (thus if the part is 50mm thick, pores above 1mm across could be detected), and to provide some information on the cooling channels. The use of CT, however, would provide the ability to determine the actual geometry of the cooling channels and allow a digital comparison with the original model the part was built to. One aspect of AM where complexity can be a hindrance is in performing radiographic inspections. Parts with a large number for small features can have lots of noise around the edges of the features, making it difficult to interpret the radiographic results.
Ultrasonic Inspection
One version of ultrasonic inspection, called surface wave inspection, is used as a surface or near-surface inspection. This entails introducing a surface wave (called a Raleigh wave) into a part and then listening for it to reflect back off of a discontinuity. Because surface wave inspection is relatively rare in industry, and would be so for AM, we will focus on the use of ultrasonic as a volumetric inspection method. Ultrasonic inspection is effective in materials that transmit sound, such as metals and composites. It is less effective in porous ceramics, neat polymers, or elastomers. Volumetric ultrasonic inspection works much like sonar on a ship. A pulse, generally between 1MHz and 20MHz, is sent out from a transducer.
- In the most common mode, called pulse-echo, the transducer, then listens for the pulse. If it reflects off an internal discontinuity, the size and location can then be inferred by looking at the signal.
- Another approach, called pitch-catch, uses a second transducer that listens for the signal reflecting off of the discontinuity. This can be a benefit over the first for some geometries.
- A third approach, called through-transmission, uses a second transducer that records the volume (amplitude) of the signal transmitted through the part.
While the first is used in metals and composites, and the second mainly in metals (especially welds), through-transmission is most commonly used for the inspection of composites. Inspection of flat shapes will use general service probes, while more complicated shapes may use customized probes other devices (called shoes) to introduce the sound into the part. The most complicated shapes often require building custom phased array ultrasonic probes, which contain an array of transducers specifically designed for the inspected geometry. All of these methods are best for finding discontinuities that are oriented perpendicular to the direction of sound. Thus, radiography (good at finding discontinuities parallel to the beam direction) and ultrasonic inspection can be very complimentary, which is why they are often used together in the inspection of welds.
The primary challenges in performing ultrasonic inspection of AM parts are due to geometric complexity and surface roughness. Because the sound needs to be introduced into and received out of the part, the geometry needs to be relatively simple to allow both access for the transducer and the ability to have some area of contact. Additionally, a predictable back surface is needed. Finally, the wave needs room to propagate, so thin or narrow features are difficult to inspect. In conventional manufacturing, most ultrasonic inspections are performed on simple shapes, such as rectangular plates or round bars. The rough or irregular surfaces of many AM parts also make inspection difficult, as a smooth, consistent surface is needed to introduce the sound into the part.
Eddy-Current Inspection
Eddy-current inspection is primarily a surface inspection method, although discontinuities within 1mm of the surface can be detected with the correct set-up. Eddy-current uses an electromagnetic coil (generally between 10Hz and 4Mhz, depending on material and application) to induce eddy-currents in the part. Thus, only conductive materials can be inspected. The presence of a physical discontinuity will disturb the eddy-currents, which is picked up the test equipment. Eddy-current testing requires a reasonably smooth surface, so the rough surfaces of AM parts can be a significant hinderance. It can also be very labor-intensive. As a result, it is often relegated to inspecting small areas, where a production discontinuity or fatigue crack is suspected.
Magnetic Particle Inspection
Like Eddy-current, magnetic particle inspection can detect surface and near-surface (also up to ~1mm deep) discontinuities. A prerequisite for this inspection method is that the material in question be reasonably ferromagnetic (alloy steels or PH stainless steels). Magnetic particle inspection works by inducing a magnetic field in the part and applying magnetic particulate matter to highlight areas where the magnetic field is disturbed. These are areas of potential cracks, laps, etc. Needless to say, the rough surfaces and undercuts in many AM parts would generate many false positive indications, so magnetic particle inspection would be limited to smooth surfaces.
Penetrant Inspection
Due to the ability to use a common inspection procedure, applicability to all nonporous materials, and quick examination time, penetrant inspection is the most common surface inspection method in industry. A discontinuity must be open to the surface in order for penetrant inspection to work, as the method entails using capillary action to draw the penetrant into the discontinuity, which then bleeds back out after the excess penetrant is removed and a developer is applied. Like magnetic particle and eddy-current inspection, linear discontinuities around 1.5mm and above are typically detected with high-sensitivity fluorescent penetrant materials. Also, like magnetic particle and eddy-current inspection, the ability to inspect is significantly compromised by rough surfaces. Additionally, surfaces must be free of grease and oils. Finally, for the high-sensitivity penetrants, machines parts with yield strengths under 1500MPa require a pre-penetrant etch to remove any smeared metal. The combination of these often means penetrant inspection requires cleaning and acid or alkaline etch tanks, which is why the inspection is often performed by specialist chemical processing contractors.
Resonance Inspection
This method treats each part like a tuning fork, although in many cases the sound is outside of the audible range. The part is loosely held and tapped in a controlled method. The sound coming off the part will have a characteristic resonance frequency that is detected with a specialized microphone and analyzed. Changes in the resonance frequency may be the result of either discontinuities or discrepant geometry. Because it is quick to perform, resonance inspection can be used as the primary inspection to screen parts as acceptable or questionable, with the questionable part inspected using another method for final disposition.
Thermographic Inspection
Thermographic inspection involves imaging an area with an infrared camera to look for differences in temperature. The inspection can be either passive (using background thermal energy to look for hot or cold spots) or active (applying heat via flash lamps or the like and looking for hot or cold spots). Outside of AM, thermographic inspection is often used to inspect areas where a coating may be bonded onto a surface to check for unbonded areas. This also takes advantage of one of its advantages, which is being noncontact and able to inspect large areas quickly. Within AM, it is being developed and implemented for process monitoring and in-situ NDT.
Process Monitoring, Process Control, and In-Situ NDT
Because of the difficulty of inspecting AM parts post-build, and the opportunity to inspect parts while they are being built, process monitoring, process control, and In-Situ NDT are being implemented and developed. We will start with defining the three terms:
- Process Monitoring – Monitoring the inputs and responses of a process while it is being performed to see that it is operating within limits.
- Process Control – Using data from process monitoring to change the inputs to a process to keep it within desired limits.
- In-Situ NDT - Inspections performed while a part is being made using AM.
Because AM is an inherently digital process, and because the machines produce large amounts of data, Process Monitoring is the most widely applied of the three. Ideally, Process Monitoring is performed while the process is active, although it can be applied post-built as part of an accept/evaluate/reject decision tree. A wide range of process parameters can be monitored. These include input parameters (laser input power, recoater voltage, arc current, gas flow rate, etc.) or output parameters (melt puddle size, bed surface temperature, laser energy, etc.). The key is reliably determining how the parameters correlate with either an acceptable part, a rejectable part, or a suspect part that requires further evaluation, which is made more challenging in that the definition may be geometry dependent. Making this determination is being helped by the use of a range of data analytics and machine learning tools that are now available.
Process Control will often use the same sensors and data as Process Monitoring. The difference is that the AM machine uses the results of the sensing and analysis to determine if a process is going out of the desired limits and makes changes to the input parameters to keep it within its desired limits. Needless to say, this must be done while the process is active so the changes can be made as part of a closed loop. While not as widely used as Process Monitoring, Process Control is beginning to appear on newer-generation AM machines with increasingly sophisticated sensors and analysis methods.
Making parts without discontinuities is always the goal. Depending on the industry and the part requirements such as, the criticality of the part and the level of confidence in the process, NDT may still be required. In AM, the best time to perform this may be while the part is being built, either to achieve desired level of sensitivity and reliability, minimize NDT cost, or to avoid continuing to add value to a part that is not acceptable. In-Situ NDT may look very much like Process Monitoring and use the same sensors and analysis methods. The difference, however, is that the output of the analysis is a determination that a region of a part contains or does not contain an unacceptable discontinuity and its approximate size.
The decision to use Process Monitoring, Process Control, In-Situ NDT, or Post-Build NDT or some combination of the four, and the methods to be used requires the input of a combination of disciplines and needs to be integrated into the overall design and build of the system as part of the overall certification of that system.
Integration of NDT with Design and Analysis
Key to the effective use of NDT is knowing the size of discontinuities that can be detected. Ideally, the NDT methods used to inspect a part can detect discontinuities that are smaller than those that will cause a part to not meet design intent; that is, smaller than the applicable defect size. One measure of this is knowing the smallest possible discontinuity a method can detect on a part, often referred to as the minimum detectable defect size, Because of the complexities of part geometry, discontinuity geometry, operator skill, etc., however, having an absolute size for what can be detected in a part is rarely possible. Thus, the term Probability of Detection (PoD) is used to refer to the size of discontinuity that can be detected with a good chance of success. A typical definition of success is being able to find 90% of the discontinuities above a certain size with 95% confidence. Depending on the material, part geometry, type of discontinuity and NDT method, parts with either actual or simulated discontinuities in the locations of interest and of a range of sizes will undergo NDT multiple times to statistically determine that value.
The final stage in integrating NDT with design and analysis is determining the effects of defects (EoD). That is, how do different discontinuities of different sizes impact the performance of a material. This type of testing is often done when a new material or process is introduced, or it is used for the first time in a new application, especially one that is loaded in fatigue or where damage tolerance is needed.
EoD testing often requires the following:
- Building test parts with intentional discontinuities
- Performing NDT on the parts to see if the discontinuities are detectable and to characterize their apparent size
- Testing the parts to failure to see the impact on either the static strength, fatigue life, or fracture stress
- Correlating the size estimated by NDT, actual size, and impact on performance
When this is completed successfully, the results are analyzed and the size of discontinuity that needs to be detected is determined. Ideally, this size is larger than the size that can be detected with confidence. If not, changes to the design and/or materials and processes need to be made. In the case of AM, many industries are just beginning studies to determine both the detectable sizes and the effects of defects.