Hygrothermal assessment of timber frame walls using a convolutional neural network

A correct design of a timbre frame wall's composition is vital to avoid moisture damage. Unfortunately, currently, no general guidelines exist determine the most optimal wall in specific context. To develop such guidelines, comprehensive study required, taking into account inherent uncertainty and variability involved input parameters. Such probabilistic assessment typically carried out through Monte-Carlo approach, which easily becomes computationally inhibitive. This paper thus makes use metamodel, mimics complex hygrothermal model while being considerably faster. The authors previously developed convolutional neural network demonstrated its' capacity predict highly non-linear response massive masonry wall. In this paper, adapted for timber walls. hyper-parameter optimisation performed, leading rules-of-thumb on architecture. It shown that can accurately time series, it be employed with confidence estimate damage risks. Subsequently, used calculate 96 types, all influencing uncertainties. results indicated compositions should not recommend based solely sd-ratio between vapour wind barrier. lower limit appears good criterion mould growth, if climate cladding type. condensation, one ensure either insulation or barrier buffer excess moisture.