Study of Full Scale Fire Test Results Versus BRANZFIRE Zone Model Output
Degree GrantorUniversity of Canterbury
During 1995 a series of full scale rig tests were carried out by the Building Research Establishment (BRE) in the United Kingdom to study the tenability conditions in the lounge and upstairs bedroom of a typical domestic residence in the event of a fire in the lounge. A typical foam cushion chair was set alight in the lounge of the house and the temperature, smoke density, gas concentrations and smoke alarm activation times were measured in the rooms of the house. From this data the time to untenable conditions in the lounge and upstairs bedroom were calculated for each of the tests that was carried out. The use of computational zone models for the prediction of conditions in enclosures has increased in the past two decades with the advent of cheap and powerful personal computers. One such zone model is BRANZFIRE which is a multi compartment two zone model that is based on a set of differential equations which are derived from the principles of conservation of energy and mass, and the ideal gas law. The experimental setup from the BRE full scale rig was entered into BRANZFIRE and a series of the test scenarios were simulated. The simulations included defining the hallway, stair and landing rooms in two different methods, one using two compartments to describe the three rooms and the other using three compartments to describe the three rooms. Two of the full scale rig tests with the lounge door closed were examined, and one of these tests was simulated with two arrangements for the door from the lounge to the hallway. One simulation used a small horizontal vent across the width of the door, the other used a narrow vertical vent the full height of the door. The simulations showed BRANZFIRE to overestimate the compartment temperatures in the room of fire origin and to predict accurate results or under predict the temperature in other compartments in these tests. The oxygen concentration was generally predicted to lower to a greater extent than was seen in the full scale testing and the carbon dioxide concentration was under estimated by the simulations. The optical density was over estimated by a significant factor in all of the rooms and this had an impact on the prediction of smoke alarm activation times with faster activation times predicted by the simulations than were seen in the full scale testing. The predicted time for an occupant ii to receive a fractional effective dose of 1.0 due to heat was predicted well when compared to the full scale rig testing. The time to receive an incapacitating fractional effective dose due to asphyxiant gases was under estimated by the simulations. The comparison between the simulation results and the full scale rig testing data highlighted the sensitivity of the simulation outputs to the information that is input into the simulation. Wherever possible a sensitivity analysis for both the inputs and the compartment geometries should be carried out.