Architectural fabric properties : determination, representation & prediction
Coated woven fabrics are used in state-of-the-art structures and yet broad assumptions are made in both material testing and behaviour. Current design practice makes little reference to the complex non-linear behaviour of architectural fabrics. A better understanding of fabric response should enable reduction of safety factors, use of more economic materials and allow more architectural freedom in the forms that can be achieved. With limited availability of test equipment and no European standards on biaxial testing, a biaxial test rig has been designed and built for this work and a new test protocol has been developed. Application of prestress followed by cyclic loading conditions the fabric and enables medium to long term properties to be measured which are appropriate for structural design. Thorough sampling of all feasible design stress states fully quantifies the fabric response. Testing has been carried out on a wide range of PVC coated polyester and PTFE coated glass-fibre fabrics. Fabric test data is commonly manipulated to fit within a plane stress framework. It is shown in this work that plane stress theory is inappropriate for representing the complex deformation mechanisms of coated woven fabrics. It is proposed that the test data is used directly in finite element structural analysis by interpolation between values in a database of test results, with no limiting plane stress assumptions. 'Feasible strain plots' provide a new tool for quantifying fabric behaviour. A predictive fabric model based on force equilibrium in the fabric 'unit cell' has been developed. The model aims to be easily accessible to the design engineer, with all parameters derived from standard tests. Whilst avoiding unnecessary complexity, the model realistically models key fabric deformation mechanisms. The model provides a more accurate representation of fabric behaviour than current industry best practice (i.e. use of elastic constants based on biaxial test data), but without the need for specialist testing.