Multi-Mechanism Wearout Reliability Prediction for Long-Lifetime Optical FBGs SHM system

Abstract

Abstract

Structural Health Monitoring (SHM) systems, comprising complex assemblies of ICs, optical fibers, and packaging, generally rely on established standards such as MIL-HDBK-217, JEDEC, or FIDES for reliability modeling. However, the MTTF and FIT metrics derived from these standards are explicitly limited to random failures during the useful life period, ignoring the critical impact of aging. To ensure reliable operation over extended timescales (typically 20 to 30 years), this study argues that wear-out must be addressed through a distinct analytical approach. We propose a methodology based on the Matrix framework recently formalized by Bernstein et al. (“Reliability Prediction for Microelectronics”, Wiley series, 2024). Instead of relying on constant failure rates, this approach isolates individual degradation mechanisms—modeled via Lognormal or Weibull b > 1) distributions as appropriate to their specific physics—and intelligently combines them. This method provides a more accurate prediction of system lifetime by capturing the cumulative effect of concurrent, competing failure modes. To illustrate this, a proof-of-concept was conducted on a 45 nm FinFET FPGA (Xilinx technology). Testing three failure mechanisms across nine DOE configurations (1.2–3 V, 10 MHz–1 GHz, −60 °C–155 °C) yielded the comprehensive Matrix-based physics-of-failure model presented in Figure 1. This multi-mechanism paradigm, is extended to SHM systems. This empirical foundation justifies systematic mapping of Peck (humidity), Fick (diffusion), Arrhenius (thermal), Coffin-Manson (thermo-mechanical fatigue), and Paris (crack propagation) models—whose coupling effects are captured through experimentally calibrated linear combinations tailored to each material and assembly context. For SHM architectures, environmental stressors—temperature extremes and cycling, humidity and salinity (offshore wind turbines), and vibration spectra—are directly mapped to their respective physics-based models. Fiber Bragg grating (FBG) sensors embedded in the SHM chain provide a real-time degradation signature via spectral observables—wavelength shift, reflectivity, bandwidth, and line-shape—that directly encode the combined effects of thermally induced effect, residual stress accumulation, microstructural embrittlement, and interfacial degradation. This integration of spectral indicators into the multi-mechanism reliability framework translates underlying physics-of-failure models into actionable health metrics, enabling quantitative assessment of cumulative wearout over the sensor's full operational lifetime. Our approach leverages numerical modeling from the design phase to identify sensor weaknesses by analyzing frequencies, geometric features, and material interfaces. This enables prediction of degradation mechanisms under combined stresses and allows early anticipation of failures through Multiphysics simulations. We validate this methodology on a functional SHM demonstrator system exposed to various combined environments, monitoring performance signatures across a wide range of stress conditions to support extrapolation to realistic operational contexts. We also address the critical issue of ensuring that accelerated or extrapolated tests are truly representative. Reliable lifetime prediction using such data requires thorough calibration, validated acceleration laws, and a quantified assessment of projection uncertainty. By integrating experimentally calibrated, multi-stress wearout models mapped to sensor signatures, we can deliver not only MTTF and RUL estimates, but also real-time health indices for proactive SHM system management.