PSI - Issue 54

Francisco Afonso et al. / Procedia Structural Integrity 54 (2024) 545–552 Francisco Afonso / Structural Integrity Procedia 00 (2019) 000 – 000

550

6

implemented, with a supporting structure, to capture and process the bidimensional profile of the train wheel. The entire wheel must rotate to be fully analyzed, as the sensor continuously captures and processes the various profiles. Additionally, to allow flexibility when defining the defects, the software must also allow the technician to customize the tolerances of a different number of metrics, used to analyze the profiles. The chosen laser line sensor was the Zebra AltiZ AZ1D4LR model. It can acquire profiles between 160 mm and 545 mm of distance from its reference point, a depth resolution between 0.01 mm and 0.89 mm, depending on the distance of the object to the sensor, and 0.157 mm resolution along the laser line (Zebra Technologies, 2023) 4.1. Support structure The support structure was developed to be implemented on a train undercarriage, separated from the main carriage structure. That train component is sometimes separated from the main train in maintenance facilities, and they usually possess a planar surface with direct access to the top of the wheel. Additionally, its relatively small weight allows it to be pushed manually, rotating the wheel, as is required for the analysis. In Fig. 4 both the final version of the train wheel system ’s support and the area chosen to implement the system can be found.

Fig. 4. (a) wheel defect detection support structure; (b) fully assembled system in chosen area.

Once more, since no operations such as drilling or welding can be applied to the undercarriage, the support structure is attached to the plane surface displayed in Fig. 4 (b) through five magnets. It was designed in a T-shape composed of two profile bars and a group of connecting brackets, a different group of brackets also attaches the structure to the five magnets. This structure shape allows the attachment of the Zebra AltiZ sensor on one end, while on the opposite end a counterweight can be introduced to help reduce loads at the base of the structure. Finally, in Fig. 4 (b), the fully assembled system is placed on the train undercarriage’s chosen area, with direct access to the top of the wheel, where the sensor will be able to capture wheel profiles. 4.2. Software Custom software was also developed to process and display the data acquired from the sensor. The data processing algorithm is also based on a continuous comparison between the sensor’s profile acquisitions and a previously acquired template profile, containing no defects. In Fig. 5, a diagram of the logic behind the developed software is shown. Before the start of the analysis, a defect-free template of the object that will be analyzed must be taken, i.e. the template profile. The software stores a single template profile at a time. At this point it will be possible to observe two profiles in a chart, containing both the template profile and the currently acquired profile, which is updated live. Different metrics can now be used to define the limits of what constitutes a surface defect, including customizable tolerances. By initiating the comparison, the live profile’s points will be subtracted to the template ’s , resulting in the difference profile which is also constantly updated and can be observed in Fig. 5.