A suite of models were integrated to predict the potential of a large liquid hydrocarbon storage tank fire escalating and involving neighbouring tanks, as a result of thermal loading. A steady state pool fire radiant heat model was combined with a further model, in order to predict the distribution of thermal loading over the surface of an adjacent tank, and another model was incorporated to predict the thermal response of the contents of the adjacent tank. In order to predict if, or when, an adjacent tank will ignite, the radiant heat from the fire received by the adjacent tank must be quantified. There are a range of mathematical models available in the literature to calculate the radiant heat flux to a specified target and each of these models is based on assumptions about the fire. The performance of three of these models, which vary in complication, was analysed (the single point source model, the solid flame model and the fire dynamics simulator computational fluid dynamics model) and, in order to determine the performance of each model, the predictions made by each of the models were compared with actual experimental measurements of radiant heat flux. Experiments were undertaken involving different liquid fuels and under a range of weather conditions and, upon comparing the predictions of the models with the experimental measurements, the solid flame model was found to be the one most appropriate for safety assessment work. Thus, the solid flame model was incorporated into the thermal loading model, in order to predict the distribution of radiant heat flux falling onto an adjacent tank wall and roof. A model was developed to predict the thermal response of the contents of an adjacent tank, in order to predict variations in the liquid and vapour temperature, any increase in the vapour space pressure and the evolution of the vapours within the given time and the distribution of thermal loading over the surface of the tank as predicted by previous models; of particular importance was the identification of the possibility of forming a flammable vapour/air mixture outside the adjacent tank. To assess the performance of the response model, experiments were undertaken at both laboratory and field scale. The laboratory experiments were conducted in the Chemical Engineering Laboratory at Loughborough University and required the design and construction of an experimental facility representing a small-scale storage tank exposed to an adjacent fire. The field scale experiments were undertaken at Centro Jovellanos, Asturias, Spain. An experimental vessel was designed and fabricated specifically to conduct the laboratory tests and to measure the response of a tank containing hydrocarbon liquids to an external heat load. The vessel was instrumented with a network of thermocouples and pressure transmitter and gauge, in order to monitor the internal pressure and distribution in temperature throughout the liquid and its variation with time. The model predicting the thermal response of an adjacent tank was shown to produce predictions that correlated with the experimental results, particularly in terms of the vapour space pressure and liquid surface temperature. The vapour space pressure is important in predicting the time when the vacuum/pressure valve opens, while the liquid surface temperature is important as it governs the rate of evaporation. Combining the three models (the Pool Fire model, the Thermal Loading model and the Response model) forms the basis of the storage tanks spacing international codes and presents a number of innovative features, in terms of assessing the response to an adjacent tank fire: such features include predicting the distribution of thermal load on tanks adjacent to the tank on fire and thermal load on the ground. These models can predict the time required for the opening of the pressure vacuum relief valve on adjacent tanks and the release of the flammable vapour/air mixture into the atmosphere. A wide range of design and fire protection alternatives, such as the water cooling system and the minimum separation distance between storage tanks, can be assessed using these models. The subsequent results will help to identify any recommended improvements in the design of facilities and management systems (inspection and maintenance), in addition to the fire fighting response to such fires.

Yes / In this study, a Fuzzy Bayesian network (FBN) approach was proposed to analyze the domino effects of pool fire in storage tanks. Failure probabilities were calculated using triangular fuzzy numbers, the combined Center of area (CoA)/Sum-Product method, and the BN approach. Consequence modeling, probit equations, and Leaky-Noisy OR (L-NOR) gates were used to analyze the domino effects, and modify conditional probability tables (CPTs). Methanol storage tanks were selected to confirm the practical feasibility of the suggested method. Then the domino probability using bow-tie analysis (BTA), and FBN in the first and second levels was compared, and the Ratio of Variation (RoV) was used for sensitivity analysis. The probability of the domino effect in the first and second levels (FBN) was 0.0071472631 and 0.0090630640, respectively. The results confirm that this method is a suitable tool for analyzing the domino effects and using FBN and L-NOR gate is a good way for assessing the reliability of tanks. / National Petrochemical Company (NPC) of Iran