Faculty Advisor or Committee Member

Robert G. Zalosh, Advisor


Erdem Ural, Mark W. Richman




An engineering method has been developed for calculating the blowdown of agent from a pressurized dry chemical fire suppression system supply cylinder, and the flow rate of agent through a piping delivery system. Its goal is to provide the means to determine the blowdown time and agent delivery capabilities of pre-engineered and simple engineered systems. The method is based on the treatment of the two-phase powder-gas flow as an equivalent fluid with thermodynamic properties that account for agent composition and the relative proportions of agent and gas propellant. The mixture is treated as compressible, and the expansion in the supply tank is assumed isentropic. A key assumption in the model is that the agent (powder) mass fraction remains constant, in both the tank and delivery system. Laboratory tests were conducted to examine the validity of the model and its assumptions. Simple systems were discharged to measure pressures in the cylinder and nozzle inlet during discharge, and the mass of agent discharged. A 0.43 cubic foot cylinder containing 0-25 lbm of either sodium bicarbonate or moammonium phosphate, pressurized at up to 300 psig of nitrogen, was discharged, either alone, or with an 8-foot length of piping and a single nozzle. For the cylinder by itself, gas alone pressurized to 300 psig discharged in 1.5 seconds, while 25 lbm of sodium bicarbonate agent pressurized to 300 psig discharged in 5.2 seconds with 0.10 lbm of agent remaining in the cylinder after discharge. There was no significant difference in the discharge times or residual masses in the cylinder after discharge between the sodium bicarbonate and monoammonium phosphate agents. For a cylinder-pipe-nozzle system, gas-alone discharges pressurized to 300 psig took 7 seconds, while 25 lbm of sodium bicarbonate agent pressurized to 300 psig discharged in 26 seconds with 0.64 lbm of residual agent in the cylinder after discharge. Predictions generated by the model were compared with test results. Cylinder alone gas-only discharge model predictions agreed well with test data for the full duration of tests using a discharge coefficient of 0.380 to characterize the gas flow through the dip tube / valve assembly; a simple isentropic analytical model gave a good prediction using a discharge coefficient of 0.430. Gas-solids predictions using a discharge coefficient of 0.500 agreed well with test data up to the observed inflection point near the end of discharge. This inflection point is caused by the agent in the cylinder reaching the bottom of the dip tube, resulting in reduced flow of agent from the cylinder, and thus reducing the mass fraction of the flow. Cylinder-pipe-nozzle model discharge predictions for gas-only discharges agreed well with test data for the full duration of tests using a discharge coefficient of 0.470 for the 0.173-inch diameter nozzle used in the testing. Model predictions agreed well with the gas-solids mixture test data up to the inflection point, using a discharge coefficient of 0.999. The constant mass fraction assumption results in residual agent mass predictions of 2.0 lbm or more after discharge. Test data shows 0.6 lbm or less of residual. This residual discrepancy, and the presence of the inflection point observed in solids-gas tests, suggests that the constant mass fraction assumption is not adequate to accurately model agent discharge from the cylinder. Using an appropriate discharge coefficient, the model can be used to determine approximate discharge times for simple systems.


Worcester Polytechnic Institute

Degree Name



Fire Protection Engineering

Project Type


Date Accepted





program, testing, modeling, test, model, blow down, discharge, dry chemical, Fire extinction, Chemical systems, Fire sprinklers, Testing, Dry sprinkler systems, Testing, Dry sprinkler systems, Mathematical models