Modeling sprinkler performance
By Samuel S. Dannaway, PE,
President, S.S. Dannaway Associates, Inc., Honolulu
Fire protection engineers involved with performance-based design rely very heavily on fire models to determine the effectiveness of a trial design. Fire protection engineers often incorporate fire sprinklers into their designs. So let us look at how sprinkler system performance is addressed in the models.
Modeling sprinkler performance started with the first perforated pipe fire sprinkler systems. Someone determined the size, orientation and spacing of the perforations to optimize the system. The pipe had 1/8-inch diameter holes located 60 degrees off top vertical and spaced every 3 to 10 inches.
Following the development of the fusible sprinkler and the closed-head sprinkler system came the development of the next sprinkler performance model, the pipe schedule. The first schedule appeared in the first NFPA’s fire sprinkler standard in 1896. It is interesting to note that the 1896 pipe schedule did not distinguish occupancy hazard classifications and permitted the use of 3/4-inch steel pipe. The pipe schedule model was modified in 1940, eliminating the use of 3/4-inch pipe and introducing schedules by occupancy hazard. Soon, the closed-head pipe schedule will follow the way of the perforated pipe schedule.
The next major sprinkler performance model came with the advent of hydraulically calculated systems in the 1970s. Of course, pipe schedules and hydraulic calculation software, though they help us design sprinkler systems, really did not model sprinkler performance. They do not help us understand the impact of fire sprinklers in a given fire scenario. They do not tell us how many sprinklers would operate, their effect on heat release rate, available safe egress time or level of property damage.
Fire models address sprinkler performance in two primary ways: time to sprinkler activation and attenuation of the fire environment. Here are some fire models that deal with sprinkler performance.
Simple computer algorithms
Most notable in this group is DETACT-QS, available at www.nist.gov. DETACT is an empirical model that predicts time to detector or sprinkler activation. It uses algorithms based on correlations derived from experimental fire test data. Required model input is the height of the ceiling above the fuel, the distance of the sprinkler from the axis of the fire, the initial room temperature, the sprinkler actuation temperature, the sprinkler response time index, and the total heat release rate as a function of time for a given fire. The model assumes a large room with an unobstructed, flat ceiling not confined by walls. The model only considers convective heat transfer. A version that uses t-square fires is also available.
Zone models simulate room(s) fire conditions by dividing the room into two zones, an upper hot layer and a lower cool layer. The most notable U.S. zone fire model is the Consolidated Fire Model of Fire Growth and Smoke Transport (CFAST), developed by NIST. The model uses correlations to model sprinkler activation and suppression. Sprinkler activation uses a correlation to determine the time of the sprinkler activation element based on the ceiling jet temperature, ceiling jet velocity and the sprinkler RTI. Sprinkler suppression is modeled by a correlation that takes into account the sprinkler discharge density and reduces the heat release rate over time. The model can only handle the suppression effect of the first activated sprinkler. The model also assumes that the sprinkler is not overwhelmed by the fire, that is, it will cause the heat release rate to decrease.
With the increase of power in today’s desktop computers, 2-zone fire models have largely been replaced by field models (CFD models) as the tool of choice for the fire modeler. NIST’s Fire Dynamics Simulatoris the most prevalent U.S. CFD model in use. FDS is in version 5.5. FDS predicts the thermal environment resulting from a fire, using the underlying physics, but it relies on various empirical models that describe the activation of various fire detection devices. Sprinkler activation is modeled using a correlation developed by Hekestad at FM and adjusts it for link cooling resulting from adjacent water spray. Suppression by sprinklers is modeled using a combination of models using first principles and algorithms. The suppression model takes into account sprinkler droplet size and distribution, heat transfer to airborne droplets, the reduction of the rate of pyrolysis from droplets that reach solid burning surfaces, and the cooling from droplets reaching non-burning surfaces.
What all of these models have in common, with respect to sprinkler activation and suppression, is that they rely very heavily on correlations. A suppression model that accounts for detection, droplet size and distribution, and fire suppression based on first principles does not yet exist.
Samuel S. Dannaway, PE, is a registered fire protection engineer and mechanical engineer with bachelor’s and master’s degrees from the University of Maryland Department of Fire Protection Engineering. He is past president and a Fellow of the Society of Fire Protection Engineers. He is president of S. S. Dannaway Associates Inc., a 15-person fire protection engineering firm with offices in Honolulu and Guam. He can be reached via email at SDannaway@ssdafire.com.