Sensitivity analysis of a predictive model for the fire behaviour of a sandwich panel Antonio Galgano a , Colomba Di Blasi a, * , Eva Milella b a Dipartimento di Ingegneria Chimica, Università degli Studi di Napoli “Federico II”, P.le V. Tecchio, 80125 Napoli, Italy b IMAST SCarl, P.le E. Fermi, Granatello, 80055 Portici, Italy article info Article history: Received 2 November 2009 Received in revised form 10 June 2010 Accepted 18 August 2010 Available online 24 August 2010 Keywords: Glass-reinforced plastics Sandwich panels Mathematical modelling Fire response Sensitivity analysis abstract A sensitivity analysis to assumptions and input variables is carried out for a predictive model previously developed [1] for the fire response of a glass-fibre/polyester panel and a glass-fibre/polyestereVermiculux sandwich. It is an unsteady, one-dimensional model using the porous medium approximation and a constant gas pressure with two-step, finite rate kinetics for the thermal decomposition and combustion of the polymeric resin, moisture evaporation described by an Arrhenius rate law, heat and mass transfer by convection, heat conduction and radiation described by effective thermal conductivities, variation of the volumetric fractions of the polymeric resin and the moisture with the conversion degree, effective specific heats, external heat transfer resistances and surface ablation. The strongest impact on the model predictions is exerted by the imposed external heat flux with variations on the characteristic process times between 49 and 774%. An important role in sample heating/conversion is also played by surface ablation and/or external heat transfer resistance with variations up to 30e72% or, when ablation is disregarded, with temperatures along the core layer well below those of the degrading skin. These are also significantly affected by surface heat losses, with the assumption of adiabatic bottom surface leading to heterogeneous ignition of the lower skin, and evaporation of moisture with variations in the characteristic times up to 35%. The model for the effective thermal conductivity of the fibre-reinforced skin (the Parallel, the Maxwell eEucken and the Effective Medium Theory models versus the Series model) is also important resulting in characteristic time variations up to 35%. The absence of local thermal equilibrium between the condensed and the gas/vapour phase and the kinetic details of the polymer reactions are comparatively less important (maximum diminution in the characteristic times of 16%). Moreover, although over-pressures, modelled by the Darcy law, become quite high especially during the moisture evaporation stage (up to ten times the atmospheric value), their effects on the thermal response of the structure are completely negligible when structural changes are not modelled. Finally, a sensitivity analysis is also carried out to input parameters. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction The thermal response of composite materials exposed to fire conditions is the result of strong interactions between chemical and physical processes under highly dynamic conditions [2e9]. The application of multi-layered structures, such as sandwich panels, in spacecraft, aircraft, automotive, marine and building industries, where it is important to obtain high strength-to-weight and stiff- ness-to-weight ratios [10], introduce further complications, due to the different chemical and physical properties between the inner (core) and external (skin) layers and the large influences exerted by the core on the thermal response [1]. The sandwich composite used in marine applications is usually made with thin face skins of fibre- reinforced polymer laminate enchasing a thick core of ultra-light material [11]. The skins are made using a wide range of fibres and resins, including glass, carbon and Kevlar fibres and polyesters, vinyl esters and epoxies whereas the most common materials for the core layer are poly(vinyl chloride) foam, polyurethane foam and balsa. In particular, a glass-reinforced plastic (GRP) sandwich panel, con- sisting of GRP outer skins with calcium silicate material (Vermic- ulux) sandwiched between, was proposed [12] and modelled [1,12]. The mathematical modelling of the thermal response of simple or multi-layered composite structures to fire is needed to interpret the results of standard tests for material qualification and to develop multi-disciplinary design of innovative appliances, thus reducing the experimental efforts and related costs. The simplest approach, * Corresponding author. Tel.: þ39 081 7682232; fax: þ39 081 2391800. E-mail address: diblasi@unina.it (C. Di Blasi). Contents lists available at ScienceDirect Polymer Degradation and Stability journal homepage: www.elsevier.com/locate/polydegstab 0141-3910/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.polymdegradstab.2010.08.012 Polymer Degradation and Stability 95 (2010) 2430e2444