Finned tubes are widely used in tightly-packed bundles in heat exchangers and chemical processing facilities. They are preferable to bare tubes because of their enhanced heat transfer characteristics. Heat transfer can be adjusted by altering the fin height, thickness, density, and/or material. The fins are spiral wound on to the tube, either by brazing or welding, and may be solid or serrated.
Current NDE technologies for finned tube are limited to time-consuming, invasive methods which require direct access to the internal or external surface of the tube. For example, visual inspection, remote field eddy current, and ultrasonic internal rotating inspection systems (IRIS).
Guided Wave Testing (GWT) is a low-frequency ultrasonic technique that has been utilized extensively for screening above ground and buried pipe. GWT facilitates inspection of tens to hundreds of feet of pipe from a single location, thereby enabling inspection of pipe with limited access. For example, inspecting buried segments of pipe from within an excavation, inspecting insulated segments of pipe by removing a few feet of insulation, or inspecting buried segments of pipe through a wall penetration. It is important to note that GWT is a qualitative screening tool and that GWT inspection results provide an axial location, approximate circumferential location, and relative severity of any wall loss indications that are identified. The relative severity is typically referenced to the magnitude of one or more girth weld indications. The sensitivity of GWT can vary, but is typically around five percent cross-sectional area change for most applications. In most GWT applications, the torsional wave mode, which is characterized by material displacements in the circumferential direction of the pipe, is used at frequencies ranging from approximately 20 kHz to 80 kHz.
Finned tubes add an additional layer of complexity to GWT due to the presence of periodically spaced fins along the tube axis. Most GWT field experts and engineers would likely conclude that the presence of mechanically-coupled structures on the OD surface would successively reflect a portion of the incident energy, leading to rapid attenuation and practically no penetration. This conclusion is not unfounded, as GWT training courses typically teach inspectors that anything attached to the OD surface of a pipe will produce attenuation and indications in the data. However, such courses overlook a class of solutions to the guided wave problem known as Bloch wave functions. Bloch wave functions account for structural periodicity in the problem (i.e. they assume the presence of the periodically-spaced fins on the external surface of the tube). Physically, this results in frequency pass bands and frequency stop bands which depend on dimensions and material properties of the tube and the fins.
To validate these theoretical conclusions, finite element models of guided wave propagation in a finned tube were generated. Figure 1 shows an image of the model geometry of the finned tube. The modeled structure was 10’ long with 6” of bare tube on either side to simulate the bare area that would be needed for transducer placement. Figure 2 shows a zoomed image of the finite element mesh that was generated. Incident guided waves were generated over the range of frequencies typical of GWT inspections, using both torsional and longitudinal mode excitation.
As a baseline for comparison, the same frequencies were tested on a bare tube having the same dimensions and material properties. Figure 3 shows the results of the frequency sweep on the bare tube. The vertical strip of black/red pixels near the center of the image are indications from the tube end across the frequency range. One can see that the indication from the tube end is consistent across the entire frequency range.
Figure 4 shows the frequency sweep results on the finned tube. Notice that, in contrast to the bare tube, there are specific frequency ranges at which there are indications from the tube end and specific frequency ranges where there is no indication from the tube end. The regions where there are reflections from the tube end are the frequency pass bands. Conversely, were there are no indications from the tube ends, there are frequency stop bands.
The finite element results corroborate the theoretical hypothesis and provide a method for determining the frequency pass bands associated with the specific dimensions and material properties of the finned tube. This implies that, utilizing finite element models, the frequency pass bands of any finned tube configuration can be predicted; thereby facilitating the design of ultrasonic transducers that can operate within the frequency pass band ranges. Utilizing this process, GWT can effectively be used to rapidly screen finned tubes for damage in the tube wall.
Finally, experimental testing was carried out to corroborate the finite element models. Experimental testing was carried out on a section of finned tube that had been removed from service. Figure 5 shows the results of the frequency sweep conducted on the experimental testing component. Similar to the finite element models, the experimental results show frequency pass bands and stop bands. Within the frequency pass bands, there is a discernable indication from the cut end of the finned tube.
Figure 6 shows a schematic, including photographs, of the experimental flaw detection test setup. Fins have been removed from three sections of the finned tube to create access points for transducer placement. The access points were selected such that a saw cut flaw was located 10’, 15’ and 25’ from the transducer. The depth of the saw cut flaw was varied incrementally to represent 7.5%, 10%, 12.5%, and 13.3% cross-sectional area changes.
Figure 7 show A-scan waveforms acquired on the test finned tube. Each graph includes the waveform from the pristine finned tube (black trace) for comparison with the waveform from the flawed finned tube (red trace). Moving from top to bottom, the red trace represents increasingly severe cross-sectional area loss, from 7.5% to 13.3%, respectively. The annotations in each graph indicate the echo from the defect, as well as the echo from the cut end of the tube in the data.
Figure 7 demonstrates that GWT has excellent sensitivity to the sawcut flaw. It is important to consider the flaw geometry in assessing the capabilities of GWT in screening finned tubes as the technique is, in general, more sensitive to sharp changes in cross-sectional area (e.g. a sawcut flaw, localized wall thinning) than to gradual changes in cross-sectional area (e.g. gradual wall thinning from erosion). Additional testing is required to validate the detection capabilities of GWT in finned tubes with other flaw geometries.
As shown in the waveform graphs, the amplitude of the response from the flaw grows in proportional to the flaw cross-sectional area. Figure 8 shows the correlation between cross-sectional area loss and amplitude.
The results of this theoretical, numerical, and experimental investigation demonstrate that GWT clearly has potential to be used in a screening capacity for identifying flaws in finned tubes. The pass band frequencies for any combination of material properties, tube dimensions, and fin dimensions can be determined numerically via finite element analysis. Flaw sensitivity study results show that GWT is sensitive to the sharp change in cross-sectional area of the sawcut flaw geometry. Additional tests are necessary to establish the sensitivity of GWT to flaw geometries characterized by gradual wall thinning. This work has shown that GWT has the potential to be applied as a rapid and non-intrusive screening method for these previously difficult to inspect components which have otherwise required direct and intrusive access for inspections.