Modeling and reliability analysis of argon content degradation data in immersion test of insulating glass units based on Wiener process

Insulating glass units (IGUs) rely on argon-filled cavities to limit heat transfer; hence argon leakage signals seal deterioration and functional failure. To obtain degradation data under accelerated yet controlled conditions, we conducted an immersion test in which cold-bent IGUs were mounted on engineering-consistent fixtures, fully submerged, and inspected every 5 days for watertightness and argon content. The measured degradation series were modeled with a Wiener process with drift, and parameters were estimated via an expectation–maximization (EM) algorithm. The model accurately reproduced the observed trajectories, while bootstrap predictive bands with 95% confidence intervals (CIs) quantified fitting uncertainty. Based on the fitted parameters, we computed the reliability R ( t ) and the first-passage time to a practical decision threshold ( R ( t ) = 0.6), yielding specimen-specific maintenance timelines. For comparison, a local linear trend state-space benchmark was also fitted; it achieved similar curve-fit accuracy, whereas the Wiener framework directly links the fitted dynamics to reliability and time-to-threshold predictions. The results demonstrate that immersion testing, coupled with stochastic degradation modeling, provides a rigorous basis for diagnosing seal performance and planning maintenance, offering actionable guidance for IGU quality control, warranty screening, and life-cycle management in curtain-wall systems.