Laminate analysis is used to model a loaded composite plate under one-sided heat flux. The input to the laminate analysis comes from a thermal/ablative model, which predicts the temperature evolution through the thickness. It also gives the profile of residual resin content, which reflects the extent of thermal damage. Relationships are proposed to enable the computation of the elastic constants and other mechanical properties as functions of temperature and resin content. The model was applied to a 12 mm thick woven glass/polyester laminate exposed to a heat flux of 75 kW m 2. The laminate A, B, and D matrices were modeled, along with the variation of failure loads in compression and tension. The predictions agreed well with experimental values for compression of a constrained plate. Both the local buckling load, which is proportional to √D1 D2, and the compressive failure load fall rapidly on exposure to heat flux. The bending/tensile coupling matrix, B, which is zero initially, becomes finite due to the asymmetric thermal profile, then declines as the thermal front progresses. For tensile loading, the residual properties after fire were accurately modeled, but the fall in tensile failure load was somewhat over-predicted.

This paper reports on changes to the mechanical properties of woven glass laminates with polyester, vinyl ester and phenolic resins during fire exposure. Two sets of experiments were carried out. First, unstressed laminates were exposed to a constant one-sided heat flux (50 kW m 2) for various times, and the residual post-fire strength at room temperature was reported. In a second series of experiments, laminates were tested under load. The times corresponding to a given loss of properties were 2-3 times shorter than in the previous case. It was found in both cases that modes of loading involving compressive stress were more adversely affected by fire exposure than those involving tension. A simple ‘two-layer’ model is proposed, in which the laminate is assumed to comprise (i) an unaffected layer with virgin properties and (ii) a heat-affected layer with zero properties. For residual properties after fire, the ‘effective’ thickness of undamaged laminate was calculated using this model and compared with measured values. A thermal model was employed to predict the temperature and the residual resin profile through the laminate versus time. Comparing the model predictions with the measured values of effective laminate thickness enabled simple criteria to be developed for determining the position of the ‘boundary’ between heat-affected and undamaged material. For post-fire integrity of unloaded laminates, this boundary corresponds to a Residual Resin Content (RRC) of 80%, a criterion that applies to all the resin types tested. For polyester laminate under load in fire, the boundary in compressive loading (buckling failure) appears to correspond to the point where the resin reaches 170 C. In tensile loading, significant strength is retained, because of the residual strength of the glass reinforcement. The model was used to produce predictions for ‘generic’ composite laminates in fire.

No abstract

A thermomechanical modelling approach is proposed for estimating the residual properties of bre reinforced polymer composites damaged by re. The modelling was carried out in two parts: (i) prediction of the extent of thermal decomposition (or charring) using a thermal model; and (ii) prediction of the post-re behaviour using a two layer model that combines the properties of the undamaged laminate and the residual char. Fire experiments were performed on glass-polyester, vinyl ester, and phenolic laminates using a cone calorimeter operated at heat uxes in the range 25-100 kW m -2, for times up to 30 min. After cooling to room temperature the thickness of the thermal damage layer was determined, along with values of the residual tensile, compressive and exural properties. For the 'two layer' model it was found that the eVective boundary between char material and undamaged laminate corresponded to the point where the residual resin content (RRC) of the laminate was 80%. Surprisingly, this value was found to hold for all three resin types tested. Using this RRC value, excellent agreement was found between the measured and predicted post-re char thickness and the residual mechanical properties. The approach presented is the rst reliable method for accurately predicting the residual properties of composites after re.

No abstract