EXPERIMENTAL INVESTIGATION AND NUMERICAL SIMULATIONS ON STEEL ELEMENTS PROTECTED WITH INTUMESCENT COATING

Intumescent coatings (IC) have long been used as effective methods for longterm passive fire protection for buildings. However, much is still unknown about these coatings. IC react under the influence of fire and swell to many times their original thickness, producing an insulating char that protects the substrate from the effects of the fire (damage or excessive deformation). Its role is also to provide a smooth, aesthetically pleasing finish, which is durable and easy to maintain. In the current international framework for designing of structures in case of fire, the performance based approach is provided (Fire Safety Engineering,FSE). The performance based approach consists of detailed analysis of the fire,considering natural fire curves, which combine more sophisticated calculation(advanced methods) for structural models. In order to perform rigorous and realistic analyses on structures protected with IC, thermal properties of IC should be known. Nevertheless, the thermal characterization of these systems is not available. Hence, experimental tests at high temperature should be performed. The aim of this work is to provide a thermal characterization of reactive protective (IC), which can be used in advanced calculations. This aspect is important to predict the behaviour of IC , to optimize the experimental tests and to increase reliability. Moreover the thermal characterization of IC is useful to design an intervention of IC fire protection with performance-based approach and not only with prescriptive-based approach. So, in order to investigate the different fire phenomena that can affect the IC performance and their behaviour under different fire conditions, two sets of experiments representing different types of heating exposure were conducted for different water based IC. In the first set of experiments, in furnace, steel plates were exposed to ISO834 and Smouldering fire curves with different initial heating rates. The steel specimens were steel plates with three different section factors protected by 500 μm, 1000 μm, 1500 μm and 2000 μm of dry film thickness (dIC) of IC. In the second set of experiments, steel plate samples protected by 3 different IC and by two different dICs (1000 μm, 1500 μm) were tested in a cone calorimeter. The steel specimens were expose to different heat fluxes: 50 and 30 kW/m2. Moreover, the IC performance was quantitatively assessed according to two different parameters: the thermal conductivity based on the Eurocode formula for insulated steel sections and the IC swelling (directly measured during the test using the digital image correlation technique). The results underlined that many IC characteristics, such as the IC expansion and the equivalent thermal conductivity, are dependent on the section factor and on dIC; these two parameters depend also on the type of heating (e.g. furnace and cone calorimeter). However, other aspects like the paint activation temperature or the temperature at which the minimum value of thermal conductivity is reached, are intrinsic characteristics and they seems to be independent of the fire conditions. One of the main goals of this work was to find a thermal conductivity law of the IC, based on a series of experimental data, which can also be applied to cases of real structures, in order to model them. In particular, starting from the typical development of the IC equivalent conductivity, calculated according the Eurocode formula, a standard segmented multivariate linear regression analysis was applied to the data gathered in the previous phase at significant temperatures, depending on the two factors that have been seen to have a greater influence on IC behavior: the section factor and initial thickness of IC. In order to validate the calibrated regression laws of the equivalent IC conductivity, several real scale tests were also simulated. In particular, starting from experimental data (on the same IC tested in small scale), that are easily accessible by the current state-of-the-art testing procedures, several section of different type and protected with different IC thickness were modeled: hollow circular section, H shape sections and I shape section were considered. In all the cases the numerical/analytical results are in good agreement with the experimental temperatures.