CFD investigation of balcony spill plumes in atria (part II)

This paper proposes an integrated method for using experimental data and CFD modeling to develop engineering correlations for atrium smoke management. Part I focused on the experimental program and validation of a CFD model of the experimental facility. Part II describes the extension of this model to a parametric study of balcony spill plumes. Smoke management in buildings during fire events often uses mechanical ventilation systems to maintain smoke layer elevation above a safe evacuation path. Design of these systems requires accurate correlations for the smoke production or mass flow rate of the buoyant fire plume. One design issue is the mass flow rate of fire plumes which spill out from a fire compartment, under a balcony and up through an atrium or other large volume. Current engineering correlations for these balcony spill plumes (BSPs) are based on a combination of one-tenth scale test data and theoretical analysis. The suitability of these correlations for real-scale designs has been questioned. A combined program of full-scale experimentation and CFD modeling is being conducted to analyze the accuracy of these correlations. A full-scale experimental facility was constructed with a 5 m by 5 m by 15 m fire compartment connected to a four-storey atrium. Propane fires in the compartment produced balcony spill plumes which formed steady-state smoke layers in the atrium. Experimental variables included fire size, compartment opening width, compartment fascia depth and draft curtain depth. A variable exhaust system was used to achieve various smoke layer heights for each of 100 experimental configurations. Temperatures were measured in the compartment, atrium and exhaust system. The experimental data was used to determine the atrium smoke layer elevation and balcony spill plume mass flow rate for each configuration and fire size. This data was compared against design correlations for atrium smoke management systems to evaluate their accuracy. This data set also provided validation data for a CFD model of the facility. A CFD model of the experimental facility was implemented using the Fire Dynamics Simulator software (Version 3). Large-eddy simulations were performed with a constant radiative fraction and an infinitely fast mixture fraction combustion model. Data from these simulations was compared to the experimental data. The CFD model was then extended to a 50 m high atrium to overcome limitations in the experimental data. Grid sizes on the order of 10-1 m were evaluated in a grid sensitivity analysis with smoke layer elevation as the comparison variable. A parametric study focusing on the variation of plume mass flow rate with elevation was conducted using the same variables as the experimental program. Results from the parametric study are being compared to existing engineering correlations. A new proposed correlation for the variation in balcony spill plume mass flow rate with elevation is under development. (author)