Photo credit: www.sciencedaily.com
Wood, from the ancient Japanese cypress to the resilient ponderosa pine, has been a staple in construction for thousands of years. Despite the dominance of steel and concrete in modern architecture, wood is regaining prominence, especially in public and multi-story buildings, due to its environmental advantages.
Historically, wood has often been neglected in favor of other materials because it is prone to damage from sunlight and moisture when used externally. Although wood coatings have been developed to safeguard these surfaces, the deterioration typically begins before it is visible. By the time such damage becomes apparent, it is often too late for effective remediation.
A research team at Kyoto University is addressing this challenge by developing a straightforward yet effective method to diagnose this hidden degradation before it reaches a critical stage.
“If we can ‘see’ what the eye cannot, we can extend the life of wooden structures and enhance sustainability in the construction sector,” states Yoshikuni Teramoto, the corresponding author of the study.
The group is integrating data-driven tools into traditional wood maintenance practices by utilizing mid-infrared spectroscopy in conjunction with machine learning techniques. To kick off their project, they have experimented with artificially weathered wood coatings and those enhanced with cellulose nanofiber, a natural additive that can bolster the coatings’ durability.
Their machine learning algorithm incorporates a technique known as partial least square, which helps in constructing a model capable of predicting the level of deterioration. Additionally, a genetic algorithm was employed to isolate the most informative infrared signals, thereby enhancing both the accuracy and interpretability of their findings.
“We were astonished to discover that very subtle chemical changes — too minor to be detected visually — could be identified through infrared spectroscopy and predicted by our model,” Teramoto notes.
This innovative framework allows the detection of minute chemical alterations and provides precise estimates of deterioration levels. By facilitating quick and non-invasive diagnosis of early coating failures, this method could also lower the necessity for expensive visual inspections by identifying warning signs of deterioration early enough to prevent significant decay.
In their study, the researchers showcase the synergy between chemical analysis and data-driven modeling approaches as a means to foster more intelligent maintenance of sustainable buildings. “We aspire for this technology to connect traditional craftsmanship with modern data science,” Teramoto adds.
The research team is currently testing their methods on actual wooden structures and aims to refine their model for use in new paint and coating product development.
Beyond wood applications, their method holds potential for use with materials such as concrete or metal, paving the way for new diagnostic techniques that could enhance sustainability across various sectors and industries.
Source
www.sciencedaily.com