Novel hybrid modeling approach for utilizing simple linear regression models to solve multi-input nonlinear problems of indoor humidity modeling

Abstract

Investigating indoor humidity is important because abnormal moisture levels can damage building structures and result in poor indoor air quality. Outdoor humidity, ventilation rate, and internal moisture load are the three dominant factors affecting indoor humidity. State-of-the-art methods, particularly full-scale field studies for determining these three factors and indoor humidity can be both time consuming and labor intensive. This study proposes a radically new methodology to effectively model the influence of these three factors on indoor humidity for a mechanically ventilated building/space. The methodology starts with a simple linear regression (SLR) constructed by measuring indoor and outdoor humidity and then hybrids a novel analytical approach that accurately predicts the impact of ventilation rate and internal moisture load on indoor humidity. The proposed upgraded SLR model was successfully validated with high accuracy by both experiments and numerical simulations using TRNSYS commercial software. The results demonstrate the ability of the developed SLR to accurately model indoor humidity and account for the moisture exchange between indoor air and building structures/furnishings. Inferring these relationships and their influences on indoor humidity presents a challenging task. The developed model is generic and unique and supports fast and inexpensive field studies by ensuring that the measurements of indoor and outdoor humidity are sufficient to derive field tests of the impacts of outdoor humidity, ventilation rate, and internal moisture loads on indoor humidity. The proposed model can be further developed as a standardized moisture assessment tool for benchmarking building performance.