Fire Protection Sprinkler Problems

The events of September 11, in which the twin towers of the World Trade Center were ultimately brought down by the fires which raged after the upper floors of the buildings were flooded with thousands of gallons of flammable jet fuel, point out the vulnerability of large structures to uncontrolled fire. Throughout history, civilization has been plagued by out-of-control fires, from ancient Rome in Nero's time, to Chicago in 1871, to San Francisco in 1906, to name just a few incidents.

Fortunately, catastrophic fires such as these are becoming less common, thanks in large part to strict fire codes and the fire fighting capabilities of automatic sprinkler systems in buildings.

In fact, you might think of an automatic sprinkler system as having a whole team of firefighters hiding in your ceiling, ready to spring into action at a moments notice.

One of the challenges a fire sprinkler system designer faces is to devise a system that will last the entire life of a building, hopefully without ever being called upon to perform, yet remain in perfect working order such that it can function instantly when called into service.

Fortunately, these systems perform properly over 95 percent of the time. However, sometimes a system fails to deliver water during a fire, and other times the opposite happens—a system delivers water when there isn't a fire.

How widespread are problems with these systems? In October 1998, the Consumer Product Safety Commission (CPSC) recalled 8.4 million O-ring sprinkler heads. In June 2001, CPSC recalled another 33 million O-ring wet-pipe sprinkler heads and 2 million O-ring dry-pipe sprinkler heads.

Adding all these up yields about 43.4 million sprinkler heads! This sounds like a lot, and it is, but consider that about 40 million sprinkler heads are sold each year, and about 800 million heads are in service in the US and Canada; thus, while it represents an entire year's worth of industry sales, all these recalls affected less than six percent of the heads currently in service.

And, yes, as a matter of fact, we did check out the sprinkler heads in our own building after reading about the CPSC recalls, and yes, they were included in one of the recalls.

So what does an automatic fire sprinkler system consist of? Primarily, it consists of water supply piping and sprinkler heads.

A sprinkler head has a heat sensor that's supposed to actuate a mechanism to open and deliver water if and only if enough heat is present.

Two types of heat sensors are common: One is a fusible alloy (i.e., a solder-type material) that melts when it gets too hot. The other type is a liquid filled glass bulb that breaks when it gets too hot.

The left photo at the top of the page shows an example of the fusible alloy type of head. This head has a lever-type mechanism in which the water pressure on the valve is resisted by tension in the fusible link. If the link melts, the water pressure in the pipe forces the valve open.

For both types of heads, once the valve opens, water will be delivered—usually in prodigious quantities—until the supply valve is manually closed. Although there is also another type of sprinkler head which is capable of shutting off automatically, they are more expensive and consequently, not too common.

So much for the heads; let's consider the systems now for a moment. There are two main types of sprinkler systems: wet pipe and dry pipe. What are they? Just what they sound like. Wet pipe systems are full of water all the time, whereas dry pipe systems are filled only when they're activated by a fire.

Wet pipe systems are used in the vast majority of locations where freezing is not expected; for example, indoors in office buildings and schools.

Where freezing temperatures are expected inside a building, a dry pipe system must be used. This type of system has a special valve that supplies water when it senses an air pressure drop caused by the opening of a sprinkler head.

Although in concept a sprinkler system is fairly simple, a number of different problems can occur. Some problems, such as mechanical damage and corrosion occur year-round. Other problems are seasonal.

For example, in summer, the fusible link type of actuators can fail by creep, without a fire or actual melting of the link. How? Sprinkler head links for ordinary temperatures are usually made with "Wood's Metal" (an alloy of 50% bismuth, 25% lead, 12.5% tin, and 12.5% cadmium), which has a melting point of only 158^ºF. These heads are intended for use where ceiling temperatures do not exceed 100^ºF. (Other higher-temperature heads can be provided). Prolonged exposure of a head to higher-than-recommended temperatures can result in creep failure whereby the solder joint can fracture due to time-dependent permanent deformation (creep) of the fusible alloy.

Now, let's get back to the CPSC recalls. The recall was prompted when it was discovered that certain types of heads were sometimes failing to actuate during fires. Tests on representative samples of the heads revealed they did not consistently actuate at the rated pressure of 7 psi.

Interestingly, one of the locations that was studied was the NASA Johnson Space Center near Houston. The investigators found that in a sampling of 92 sprinkler heads, 60 required pressures greater than specified, with some needing as much as 55 psi to actuate.

So, what are some of the reasons for these problems? Well, for starters, consider the two basic ways to seal the opening in a sprinkler head. One way is by a cap, sort of like what you find on soft drinks or beer bottles but without the crimping. The head shown in the left photo on page 1 uses this type of seal.

Another way is with a plug, like the cork in a wine bottle. The recalled head shown in the right photo at the top of the page uses this type of seal. But instead of a cork, there is a synthetic rubber O-ring that sits inside a groove cut in the side of the metal plug, as shown in cross section in the photo at right.

If all the dimensions are correct, then the O-ring is squeezed just the right amount to seal properly. If the ring is squeezed too little, then it will leak; if the ring is squeezed too much, then the, "cork" gets stuck and the sprinkler will not open reliably when the head attempts to activate.

One way the O-ring can get stuck, of course, is if the dimensions aren't machined just right during manufacture. But another way is if the O-ring swells up while sitting in the head, patiently awaiting its "call to duty." What can cause it to swell? Contact with hydrocarbons is probably the most common cause.

So where would hydrocarbons come from in a sprinkler system, when the nominal service environment should be just plain old tap water? The answer is it usually comes from the process of installing the system.

In many installations, only the sprinkler heads show, and the piping is hidden above them by a suspended ceiling. The pipes which drop down to the heads are sometimes cut-to-fit by the installer after the ceiling is put in. If the installation is using steel pipe, the pipes will be threaded using an oil-based lubricant.

Unless the pipes are scrupulously cleaned, whatever is inside can drip down and wind up in contact with the heads, and therefore the O-rings. And some of these compounds have a deleterious effect on the O-rings, unless they're made of special materials.

Failure of sprinkler heads to open during a fire has pretty obvious consequences. But what happens at the other end of the spectrum, when heads open when they're not supposed to? The damage from a catastrophic opening, such as when fusible links fail by creep, is pretty easy to envision, but a more common problem is minor leakage.

In some environments, minor leakage can be tolerated to a degree, but not in clean rooms in the semiconductor industry. Recently, we were asked by a high-tech client to investigate why the O-ring seals in the brass heads of his clean room sprinkler system were leaking.

We found that the brass frame of the head was corroding in a crevice area where it contacted the O-ring (photo at left). Although we also found some shrinkage porosity from the casting process, it was not particularly severe and was not the cause of the leakage.

Investigating further, we examined the corroded area at higher magnification in the scanning electron microscope (photo at right). Energy-dispersive X-ray analysis revealed it to be predominantly copper and copper oxides rather than the original copper-zinc alloy.

Where did the zinc go? This form of corrosion is called selective leaching or de-alloying corrosion. It occurs when one element in a metal alloy is selectively leached from the matrix, leaving the other element(s) behind.

The result is a weakening of the remaining material. Interestingly, the remaining material generally will appear to be completely normal, with no thinning. The only external clue is a slight change in color as the alloy composition gradually changes.

When this form of corrosion occurs in brass it's called de-zincification. The corrosion probably was due to a nonscale-forming water environment containing chlorides and sulfates. Also, the crevice around the O-ring probably had a much higher chloride content than the surrounding area.

In other projects, we have identified sprinkler system failures with aggressive water characteristics and underdeposit corrosion from inadequate system cleaning after installation. We have also occasionally encountered leaks caused by microbial-induced corrosion, even in stainless steel fire sprinkler tubing.

Another problem sprinkler designers face is the installation environment of the system. For example, sprinkler heads on a gym ceiling obviously should be guarded from objects like flying basketballs. But what about the heads in a college dorm? Should they be guarded too, since to some students any indoor area may be used for a makeshift game of basketball?

What can be learned from all these problems? Quite a bit actually. Like all designs, (except mousetraps?) sprinklers have improved with industry experience. Probably the primary lesson, however, is that it's not easy to design a product that can sit around for fifty years and then function in an emergency. Both the interior and exterior environments have hazards that should be either designed out, guarded against, or warned about.

And the standards? In July 2001 Underwriters Laboratory (UL) announced that it was revising the UL standard for sprinklers, prohibiting the use of O-ring seals after January 2003. Other revisions to the standard include requiring a de-zincification test for brass parts, and an additional deposit loading test for dry sprinklers.

Corrosion Photo Gallery

Scanning electron micrograph of crystalline zinc deposits in unfused weld lap defect of tie-down chain link. The deposits formed during zinc electroplating of the chain assembly, following electric-resistance welding of the link. Although aesthetically pleasing, these deposits shouldn't be here; their presence confirmed that the defect responsible for the chain assembly failure had occurred during the manufacturing process. (magnification: 5,000)

MEI-C People

Holly M. Lundgren, assistant Manager of the American Product Safety Co. division of MEI-C, and Tim Howard were married on June 21, 2002 during a vacation in Las Vegas. Congratulations, Holly and Tim!

D.G. Chakrapani, president of MEI-C, has been invited to be on the panel of judges for the 2002 International Metallographic Contest in Quebec City, Canada, sponsored by ASM-IMS. Dr. Chakrapani was the 2001 winner of the Best in Show Jacquet-Lucas Award.