Issue 78

C. Vendittozzi et alii, Fracture and Structural Integrity, 78 (2026) 519-540; DOI: 10.3221/IGF-ESIS.78.30

objective was to evaluate whether field-acquired FBG records, including realistic acquisition imperfections, could be screened and reduced to compact time–frequency descriptors of landing-gear response.

Figure 1: The Erickson (Sikorsky) S-64 Skycrane used for testing.

Dataset selection and scope of validation The present work analyzes two representative field recordings selected from the measurement campaign. Dataset A has a duration of approximately 130 s, whereas Dataset B has a duration of approximately 70 s. These records were selected because they contain clearly observable transient responses, different levels of post-event persistence, and acquisition imperfections representative of field operation, including short discontinuities, dropouts, and saturation artifacts. The purpose of using these two datasets is not to establish statistically validated repeatability over a population of landings, nor to train or validate a hard-landing classifier. Rather, the datasets are used to demonstrate the feasibility of the processing workflow and to examine whether comparable frequency families can be recovered from independent operational recordings acquired with the same sensor installation. The descriptors reported in the Results section should therefore be interpreted as case-study evidence of recurrent behavior within the analyzed recordings, not as population-level statistics of landing-gear response. A broader validation of repeatability, robustness, and event severity would require a larger set of flights, independently labeled landing events, and synchronized operational references. Sensor configuration, installation, and acquisition setup To obtain a quasi-distributed view of strain, a set of four FBGs was bonded to the structural members of the main landing gear. The sensors are identified as follows: S1 at 1530 nm, S2 at 1550 nm, S3 at 1540 nm, and S4 at 1570 nm. The four FBGs were arranged in series in a single monitoring chain (sensor array) on a single-mode 9/125 µm polyimide-jacketed fiber. Locations were selected to capture bending and local load transfer while remaining compatible with clearance constraints, landing-gear articulation, and routine maintenance access, Fig. 2(a). Prior to bonding, the substrate surface was prepared by local abrasion followed by solvent cleaning to promote adhesion and strain transfer. Each grating segment was bonded by means of a fast curing cyanoacrylate structural adhesive suitable for strain transfer and protected using a thin overlayer to reduce vulnerability to handling, splash exposure, and incidental contact. Fiber routing followed existing harness paths where feasible and included strain relief at transitions to limit micro-bending and connector loading. The installation was designed to remain external and non-intrusive, avoiding any interaction with flight-critical systems or moving landing gear components. Because the installation was performed under field-compatible constraints, no independent strain-transfer calibration was carried out for each bonded grating. Therefore, absolute strain amplitudes should be interpreted as local sensor responses rather than as fully calibrated structural strain fields. Comparisons among sensors are used primarily to assess relative event response, timing, and spectral content, while amplitude differences may also reflect local bonding quality, adhesive-layer effects, sensor orientation, and strain-transfer conditions. The FBG array was interrogated using a multiplexed optical interrogator (Tecnobis Aerogator). Acquisition was performed at a fixed sampling rate of 1.0 kHz, selected to resolve transient landing-gear response components within the frequency range of interest. Wavelength outputs were logged to file together with time stamps generated by the interrogator. The acquisition system was connected to a dedicated laptop for real-time data saving and verification and was installed onboard near the rear seat (Fig. 2b). No additional avionics channels, accelerometers, video recordings, weight-on-wheels signals, or braking-command signals were synchronized with the FBG records. Consequently, event segmentation was derived from the FBG response itself and

522