Probabilistic Models for Spatially Varying Mechanical Properties of In-Service GFRP Cladding Panels

The physical uncertainty associated with fiber-reinforced polymer composites has to be quantified and dealt with for their widespread use to be reliable. Developing probabilistic models based on experimental studies form an important part of this task. In the present paper, such models are developed for glass fiber-reinforced polymer (GFRP) composites based on an experimental study on panels obtained from Mondial House, a 32 year old building demolished in 2006 in London. Having an average size of 1.5 mx1.7 m and made of chopped strand mat composites, these panels have been exposed to varying ambient conditions, protected only by a fire retardant gel coat for self-cleaning. Tensile and compressive tests are performed to quantify the variability in stiffness and strength properties of these panels. Intra- and interpanel effects and correlations between random variables are studied using statistical methods. A range of probability distributions is tested and suggestions are made with regard to their suitability for modeling different mechanical and geometric properties.