Sprinkler skipping

Sprinkler skipping is an unpredictable delay in the operation of sprinklers near a fire, caused by water droplets from sprinklers that have already opened. The fire plume carries some of these droplets up to the ceiling, where the hot ceiling jet takes them past the sprinklers that should open next. Droplets that strike the heat-sensitive link or glass bulb of such a sprinkler cool it by evaporation, so it heats more slowly and discharges later. Smoke vents that open on fusible links suffer from the same effect, which Cooper calls vent skipping.

Conventional sprinkler design accounts for skipping only empirically, through the experimental basis of its design rules. A sprinkler system has many heads and only a few are likely to be skipped, which makes such systems naturally tolerant of the effect. Vent systems have far fewer units, so skipping matters more where sprinklers and vents are combined. The Fire Dynamics Simulator (FDS) adds a droplet-cooling term to the equation that predicts when a sprinkler link reaches its operating temperature.

What skipping is

In the SFPE Handbook of Fire Protection Engineering, Cooper defines sprinkler skipping as the unpredictable and harmful delay in sprinkler discharge that results when droplets carried by the ceiling jet strike a link or bulb. The term therefore describes a delay in operation, not a defect of the sprinkler itself. A sprinkler spray consists of droplets of many sizes and velocities, and it varies even between nominally identical sprinkler heads. A full description of such a spray is beyond current technology.

The droplets of the first sprinklers to open meet the vigorous upward flow of the fire plume. Droplets that leave a sprinkler slowly are accelerated by the plume until they move with it, and the larger of them follow the plume well before they evaporate. The plume then turns into the ceiling jet, which carries random numbers of droplets past the sprinklers further away. This is the same flow that should heat those sprinklers until their links fuse or their bulbs break.

How skipping arises: small droplets from the first operating sprinkler are carried up by the plume and along the ceiling jet, where they strike and cool the link of the next sprinkler. Based on Cooper (2002b), p. 3-239, and Iqbal and Salley (2004), p. 10-11.

Iqbal and Salley explain why small droplets behave in this way. Water from an operating sprinkler must penetrate the rising plume to reach the burning fuel, so its droplets need enough size and velocity. Droplets that are too small evaporate or are carried upward by the plume. For fast-growing fires with a high heat release rate, large drop sprinklers are therefore sometimes needed.

How a sprinkler is triggered

An automatic sprinkler serves two functions: it is a heat detector as well as a nozzle that distributes water. Experiments show that convective heating by hot gases is the most important factor in its activation. The hot gases rise from the fire as a plume, strike the ceiling and spread out as a ceiling jet, a relatively fast and shallow flow driven by buoyancy. This jet is about 5 to 12 percent as thick as the height of the ceiling above the fire, with its highest temperature and velocity close to the ceiling.

The sensitivity of a sprinkler is expressed by its response time index (RTI), a concept developed by Factory Mutual Research Corporation. The index is measured in plunge tests with a gas flow of constant temperature and velocity, and it assumes that heat conduction to the sprinkler frame is negligible. It depends on the mass and surface area of the heat-sensitive element, and elements with little mass per unit area respond quickly. Typical values are 130 (m·s)^1/2 for a standard response link and 34 (m·s)^1/2 for a quick response link. Sprinklers with an ordinary temperature rating operate at 57 to 77 °C.

For a steady fire, the time needed to heat the element from room temperature to its operating temperature grows with the RTI and falls with the square root of the gas velocity. It also grows as the operating temperature approaches the temperature of the ceiling jet. A low RTI and a hot, fast ceiling jet therefore give early operation.

$$t_{\mathrm{a}\mathrm{c}\mathrm{t}}=\frac{\mathrm{R}\mathrm{T}\mathrm{I}}{\sqrt{u_{\mathrm{j}\mathrm{e}\mathrm{t}}}}\ln \left(\frac{T_{\mathrm{j}\mathrm{e}\mathrm{t}}-T_a}{T_{\mathrm{j}\mathrm{e}\mathrm{t}}-T_{\mathrm{a}\mathrm{c}\mathrm{t}}}\right)$$

Symbol	Description	Unit
$t_{\mathrm{a}\mathrm{c}\mathrm{t}}$	activation time of the sprinkler	s
$\mathrm{R}\mathrm{T}\mathrm{I}$	response time index of the sprinkler	(m·s)1/2
$u_{\mathrm{j}\mathrm{e}\mathrm{t}}$	velocity of the ceiling jet at the sprinkler	m/s
$T_{\mathrm{j}\mathrm{e}\mathrm{t}}$	temperature of the ceiling jet at the sprinkler	°C
$T_a$	ambient temperature	°C
$T_{\mathrm{a}\mathrm{c}\mathrm{t}}$	operating temperature of the sprinkler	°C

The temperature and velocity of the ceiling jet come from correlations based on large-scale fires of 668 kW to 98,000 kW. Separate expressions apply where the plume strikes the ceiling and where the flow has become horizontal further out. The method assumes a smooth, flat and unconfined ceiling, considers only convective heating and ignores radiation from the flames. It does not apply to sprinklers installed far below the ceiling on heat collectors, flat shields above the sprinkler, because most of the heat then rises past them and they may not activate.

Worked example

Iqbal and Salley consider a steady fire of 1,000 kW, 3.0 m below a ceiling, with a standard response link sprinkler 3.0 m from the axis of the plume. The sprinkler has an RTI of 130 (m·s)^1/2 and an operating temperature of 74 °C. Their spreadsheet uses an ambient temperature of 20 °C and a convective heat release rate of 700 kW for the jet temperature. The ceiling jet at the sprinkler then reaches 88 °C at 1.35 m/s, and the sprinkler operates after about 2.9 min. Under the same conditions, a quick response link with an RTI of 34 (m·s)^1/2 would operate after about 45 s.

How water droplets delay operation

The Fire Dynamics Simulator estimates the link temperature from an equation by Heskestad and Bill, with an added term for cooling by water droplets from previously activated sprinklers. In words, the link gains heat from the hot gas in proportion to the square root of the gas velocity and loses some heat by conduction to its mount. It loses further heat in proportion to the amount of liquid water carried in the gas. The constant of the droplet term comes from experiments by DiMarzo and co-workers and is relatively constant for different types of sprinkler.

$$\frac{\mathrm{d}{T}_l}{\mathrm{d}t}=\frac{\sqrt{|u|}}{\mathrm{R}\mathrm{T}\mathrm{I}}(T_g-T_l)-\frac{C}{\mathrm{R}\mathrm{T}\mathrm{I}}(T_l-T_m)-\frac{C_2}{\mathrm{R}\mathrm{T}\mathrm{I}}(T_g-T_l)\beta |u|$$

Symbol	Description	Unit
$T_l$	temperature of the sprinkler link	°C
$T_g$	gas temperature near the link	°C
$T_m$	temperature of the sprinkler mount (ambient)	°C
$u$	gas velocity at the link	m/s
$C$	conduction factor (C-factor) of the sprinkler	(m/s)1/2
$\beta$	volume fraction of liquid water in the gas	none
$C_2$	droplet cooling constant, 2.23 × 105 (the source prints the unit as (ms)−1/2; the equation is consistent only with (m/s)−1/2)	(m/s)−1/2

The source prints the unit of the constant as (ms)^−1/2, but the equation is dimensionally consistent only with (m/s)^−1/2, which is used here. If conduction to the mount is ignored, as in the activation time equation above, it follows from the equation that the droplet term reduces the heating of the link by the fraction $C_2\beta \sqrt{|u|}$. When this fraction reaches one, the link stops heating at all, however hot the gas. Because the constant is large, the equation implies that very small amounts of water can slow a link.

Worked example, continued

At the jet velocity of 1.35 m/s from the example above, the link stops heating at a liquid volume fraction of only 3.9 × 10⁻⁶. That is about 4 cm³ of water in each cubic metre of gas. For an assumed volume fraction of 1 × 10⁻⁶, the heating slows by 26 percent and operation is delayed from 2.9 to 3.9 min. For an assumed 2 × 10⁻⁶, the heating falls by 52 percent, and the sprinkler operates only after about 6.0 min.

Calculated temperature of a standard response sprinkler link in a 1,000 kW fire, without droplets and with liquid volume fractions of 1 and 2 millionths in the ceiling jet. Based on Iqbal and Salley (2004), pp. 10-15 to 10-18, and McGrattan et al. (2026), p. 107.

The heating term and the droplet term are both divided by the RTI, so it follows from the equation that a quick response link is slowed by the same fraction. At 1 × 10⁻⁶ its operation moves from about 45 s to 61 s. The two volume fractions are assumed values chosen for illustration, not measurements, so the results show how sensitive a link is to wetting rather than what happens at a real sprinkler. The volume of water that reaches a particular link remains unknown in practice. The FDS model also ignores thermal radiation to the link.

Calculated activation time against the volume fraction of liquid water in the ceiling jet, for a standard (RTI 130) and a quick response (RTI 34) link in the 1,000 kW example. Based on Iqbal and Salley (2004), pp. 10-15 to 10-18, and McGrattan et al. (2026), p. 107.

Skipping in sprinkler design

Sprinklers are designed to control a fire. If the first sprinkler cannot control it, a second one opens and adds cooling, and the process continues until the fire is under control. Design areas usually include 10 to 15 sprinklers, while incident data show that more than half of all fires are controlled or extinguished by one or two. Cooper notes that traditional sprinkler design accounts for skipping empirically, through the experimental basis of the design rules. The large number of heads in a typical installation, compared with the few sprinklers likely to be skipped, makes sprinkler systems naturally tolerant of the effect.

Many factors influence how sprinklers respond. They include the temperature rating, spacing and depth of the sprinklers below the ceiling, and the ceiling height. Openings, joists and air currents at ceiling level also change the flow of hot gases. Fleming adds that heat conduction to the sprinkler frame and the distance of the element below the ceiling have a significant effect on response. In one of his examples, a computer model shows that quick response sprinklers 0.46 m below a basement ceiling keep a sensitivity comparable to standard sprinklers at the usual maximum of 0.30 m. For the chosen fire, the standard sprinkler operates after 200 s and the quick response sprinkler after 172 s.

An ESFR sprinkler (early suppression fast response sprinkler) is designed to extinguish high-challenge fires, not only to control them. It combines a fast response with large droplets and a high droplet velocity. Its element has an RTI of 50 (m·s)^1/2 or less, against 80 (m·s)^1/2 or more for standard sprinklers. Large drop ESFR sprinklers protect high-piled storage in wet pipe systems and were developed by Factory Mutual Research Corporation in the late 1970s and early 1980s. Their design criteria come from full-scale testing, and very strict design and installation rules apply to them.

Skipping in performance-based analysis

Performance-based analyses can use sprinkler activation times to estimate the size of a controlled fire. The data sheets of the Australian Building Codes Board are an example from Australia. They take the maximum heat release rate of a controlled fire as the rate at the activation of the last sprinkler within the assumed area of operation. They assume that the fire activates no more than the design number of sprinklers for the hazard, or the number in the room if that is smaller. In a growing fire, any delay in the activation of these sprinklers therefore raises such an estimate.

CFAST (Consolidated Fire and Smoke Transport) is a two-zone fire model. Its suppression model records the activation of every sprinkler but ignores the effect of a second sprinkler on the fire. Fleming warns that the use of such calculations for sprinkler actuation times has not been fully established. He considers them useful mainly on a comparative basis.

Vent skipping and combined systems

Smoke vents that open when a fusible link melts work on the same principle as sprinklers, so they suffer from the same random delays. Cooper calls this vent skipping, and it is a central problem of sprinkler-vent interaction. Unlike sprinkler design, the design of automatic vent systems does not account for it. Far fewer vents than sprinklers are installed per unit floor area, so Cooper suggests increasing the total vent area of a compartment by a safety factor.

A more reliable alternative is to open all vents of the fire compartment together, triggered by smoke detectors or by a signal that a sprinkler has discharged. One strategy opens every vent of the compartment immediately after the first sprinkler discharges. A refined strategy opens at once all vents beyond the first or second ring of sprinklers around the first one. The other vents then open when the sprinklers have controlled the fire, and extra vent area compensates for the vents that stay closed at first.

Vent ganging after the first sprinkler discharge: vents beyond the second ring of sprinklers open at once, the vents closer to the fire after the sprinklers have controlled it. Based on Cooper (2002b), p. 3-239.

Sprinkler sprays also stir the smoke layer, and this mixing can lead to smoke logging, in which the smoke is mixed through the whole space. Cooper identifies six configurations of spray and smoke layer, of which four favour a combined design and two do not. Which one occurs depends mainly on the thickness and temperature of the layer and on the momentum, spread angle and droplet size of the spray. Shortly after rapid-response sprinklers discharge, the layer is still thin and not too hot, and smoke logging is then never observed.

Limits and practical points

Skipping cannot be predicted for a particular sprinkler. The droplet paths depend on the fire plume, which varies with the location and growth of the fire. Even a perfect knowledge of the spray as it leaves the sprinkler would therefore be of little use in practice. Activation models add their own simplifications: the method of Iqbal and Salley treats the element as a single lumped mass and ignores radiation.

Three practical points follow. First, a skipped sprinkler has not failed: Cooper describes a delay in discharge caused by cooling, not a fault in the sprinkler. Second, quick response sprinklers are not protected against skipping. As derived above from the link equation of FDS, droplets slow a quick response link by the same fraction as a standard link when conduction to the mount is ignored. Third, where the first sprinkler cannot control a fire, control depends on further sprinklers opening in turn, and these are the ones exposed to droplets from the first.

See also

• Wet pipe sprinkler system
• Sprinkler design density
• Hydraulic calculation
• Critical water application rate

References

• Australian Building Codes Board. (2020a). Data sheets: Fire Safety Verification Method (Handbook annex, Print version 1.1).
• Cooper, L. Y. (2002b). Smoke and heat venting. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 3-219–3-242). National Fire Protection Association; Society of Fire Protection Engineers.
• Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
• Hopkins, M., Schoenrock, J., & Budnick, E. (2005). Fire suppression systems (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
• Iqbal, N., & Salley, M. H. (2004). Fire Dynamics Tools (FDTs): Quantitative fire hazard analysis methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program (NUREG-1805). U.S. Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation.
• McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
• Peacock, R. D., McGrattan, K. B., Forney, G. P., & Reneke, P. A. (2026). CFAST – Consolidated Fire and Smoke Transport: Vol. 1. Technical reference guide (Version 7, NIST Technical Note 1889v1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.TN.1889v1