Reliability-based calibration of partial factors for the design of membrane structures

Membrane structures are widely used because of their lightweight properties and expressive architectural shapes. However, there does not yet exist a unified design approach for membrane structures as is available for conventional buildings. Because of their flexibility, membrane structures exhibit a different structural behaviour than conventional buildings. Membrane structures are doubly curved, prestressed and the prestress contributes to the stiffness. The structural response of these structures is non-linear, and the hanging and arching directions interact.

A probabilistic framework for tensioned membrane structures is developed herein. Using this tool, partial factors are calibrated for a representative tensioned hypar membrane structure made of PVC-coated polyester fabric. Reliability analyses are performed using a First Order Reliability Method in combination with Latin Hypercube Sampling. The approach is illustrated for two load cases, i.e. considering snow load and wind uplift load, respectively. Three hypars are investigated, each with a different rise to span ratio.

The partial factors are determined through a minimisation process. As a result of the performed process, the recommended partial factor for prestress is 1.0, for snow load is 2.0 and for wind uplift load is 2.0 for the investigated membrane structure. In this study, the hanging direction under snow load proves to be decisive for the calibration of the partial factors. The obtained partial factors are not in line with the partial factors found in the Eurocode (1.5 for variable loads). Alternatively, when considering a partial factor for snow load and wind uplift load of 1.5, partial factor for prestress should be increased to 1.2 for the structural type which have been investigated.

The application of Latin Hypercube Sampling in combination with the structural analysis of membrane structures enables to perform the reliability analysis calculations in a more computationally efficient way. Moreover, the used optimisation procedure for the calibration of the partial factors makes it possible to interpret the obtained results with respect to the complex non-linear behaviour of the investigated tensile membrane structures.