[modula id=”26980″]

The Hidden Threat From Rust

The overwhelming majority of our ultrasonic investigations into the condition of fire protection systems have been prompted by a prior leak or failure.  Very few investigations we perform are due to inspection requirements, and almost none are prompted by a concern for hidden internal rust deposits.

In fact, so many of our ultrasonic investigations have suggested an internal rust condition based upon high pipe wall losses that we have totally revised and enhanced our ultrasonic reporting format so that we may provide our clients with an estimate to the potential volume of rust contained.

A review of our Internet site Technical Bulletins, handouts, as well as our Photo Gallery relating to fire protection systems, should provide ample evidence to our concern about unquestionably the No. 1 corrosion related threat.

While a statistical calculation of potential rust product provides important new information to consider, the actual consequences from such rust product existing within a fire protection system is less defined.  Inspection guidelines state that no internal rust or debris can exist, although for the overwhelming majority of our ultrasonic investigations of carbon steel fabricated dry or pre-action fire systems, moderate to high wall losses have been identified along the bottom where water remains.

Given that pipe wall loss due to corrosion is directly proportional to a much larger volume of less dense iron oxide deposits, ultrasonic test results will provide a reliable indication to whether a potential life safety problem exists.

For that reason, and in order to visually express the threat present, CorrView International, LLC constructed a stand alone replica of a dry fire piping system in order to carry out rust volume vs. sprinkler head flow testing with the apparatus as shown below.

• Test Apparatus

The dry fire test platform consisted of all 1 in. carbon steel schedule 40 threaded pipe simulating the final branch line common to most dry and pre-action fire sprinkler systems.

• Individual Components

Shown at left, city water at approximately 65 psi is delivered to the dry fire system from below and held back by a normally closed 1 in. solenoid valve operated by 24 vac.

Following the solenoid valve is a tee to a normally closed pressure sensor and then a tee to a compressed air connection having a supply of air at 65 psi.

Isolation valves allow control of the air and water supplied during the set-up of the experiment, as well as to permit draining.

The dry fire line transitions horizontally where it then reaches a removable high pressure 1 in I.D. Lexan sight glass where rust product of known weight and volume can be added.  Unions at either side allow its easy removal and the addition of the rust sample.  This now represents the volume of rust moving down the final branch pipe section and into the sprinkler head.

The branch line then turns vertically downward where it is reduced from 1 in. diameter to the 1/2 in. NPT thread of the fire sprinkler head.  Following that, a new standard fire sprinkler head is threaded into the reducing adapter.

A trustworthy wood barbecue grill was positioned below the sprinkler head having a distance of approximately 18 in. between a small pile of wood kindling and the sprinkler head directly above.

Although there are many obvious differences between this experimental set-up and an actual dry fire sprinkler system, its method of operation remains basically the same.

• Preparation

Our first task was to precisely measure the amount of sample rust “contained” upstream of the fire sprinkler head.  Given that most final branch lines are 100 or more feet removed from the fire valve, and are preceded with mostly larger diameter pipe having greater surface area from which to generate more rust product, we anticipated that substantially larger volumes of rust would likely exist inside most such systems.

Our experiment was not conceived in order to prove that rust product could stop the flow of water into a fire sprinkler head, since that fact was obvious, and had already been tragically proven in prior actual fire events, some of which CorrView International was directly involved.

Instead, our interest was to attempt to define the minimum volume of rust product capable of stopping water flow, since to our knowledge, this criteria has never been established.

In one prior ultrasonic fire system investigation, CorrView removed 1/4 cup of rust product from a single 18 in. length of 1 in. horizontal dry fire sprinkler branch piping.  In another example, CorrView removed close to 2 full cups of rust product by just turning a 4 ft. length of 4 in. schedule 40 pre-action main on its end and tapping it on the ground.

Such examples illustrated the large volume of rust created not due to exceptional circumstances, but under conditions which we frequently document.  This information is provided in Technical Bulletin: FP-06.

In the photo below left we show actual rust product collected from multiple fire systems; that rust now having thoroughly dried.  The rust has a varying consistency ranging from a powdery mixture to larger flakes.  At center, a 1/4 cup measuring spoon, previously tared to 0.0 oz. has been filled with a representative mixture of the dry rust product and weighted.

This slightly less than 1/4 cup volume of rust product weighs 2.26 oz.  At right, the 1/4 cup volume of rust product is poured into the sight glass.

The filled sight glass is then assembled into the branch line.

• Set-Up Details

A full flow of water under a city pressure of approximately 65 psi is supplied up to the normally closed and de-energised solenoid valve.  With the rust filled sight glass installed and sprinkler head in place, the branch line between solenoid valve and sprinkler head is air tight.

Compressed air at approximately 65 psi is introduced into the line thereby raising line pressure and the system checked for leaks.  The stepdown transformer is energized sending power to one side of the pressure sensor switch, which at the line pressure of 65 psi, remains open.

At this time, no power is supplied to the water supply solenoid valve which remains closed.  With various minor differences, obviously, the experimental system now represents any dry fire sprinkler system in its charged and ready condition.

This experiment is also relative to pre-action fire sprinkler systems, with the only major difference being the method of activation.  The impact of the same volume of rust product created by the same corrosion conditions within the same type and size pipe, and against the same fire sprinkler head – would be exactly the same.

• Test Results

Lighter fluid and a match starts the fire, which quickly reaches and consumes the fire sprinkler head.  After approximately 25 seconds, the bulb bursts to release the charge of compressed air within the branch pipe.  The release of compressed air pushes forward some of the loose rust product into and through the sprinkler head in a blast of dust and rust debris.  The drop in air pressure is instantly sensed by the air pressure switch which in turn closes to send 25 vac to the water supply solenoid valve.

The solenoid valve instantly opens to flood the branch pipe and picks up the remaining rust product in the sight glass.  This rust product is then forced into the 1 in. to 1/2 in. reducer and finally into the sprinkler head body.

The end result, as the following photographs and videos illustrate, are somewhat surprising given the relatively small amount of rust product.

After a first quick release of air and dry rust through the sprinkler head, water carries the remainder of the rust into the sprinkler head to totally stop all water flow.  A few droplets of water and rust remain.

The fire is temporarily suppressed downward due to the force of the blast of air and rust, and then quickly envelops the dry fire sprinkler head.  A full supply of water capable of easily extinguishing the fire is stopped completely behind far less than 1/4 cup of rust product, given that a sizeable volume of rust having been blown through the sprinkler head during its early release.

The experiment was repeated multiple times using the same 1/4 cup volume of rust product and produced the same result.

• Real Time Video

  • • At left:   Slow motion shows the flames activating the fire sprinkler head, whereupon an initial blast of finer rust product is released.  Rust clogs the sprinkler head with no release of water, and the flames again continue unimpacted.

• • • Center:  A second run of the experiment shows a short release of water once the sprinkler head activates.  Slow motion video again shows the rust quickly sealing the sprinkler head, with a few final water droplets released before the flames continue.

• • • At right:  Real time video shows a lengthy interval before the sprinkler head releases.  Rust particulates are blasted out as the sprinkler head discharges, but then seal the sprinkler head orifice.  At the left side of the screen, the original rust product within the sight glass is seen being quickly forced downstream by the release of water through the solenoid valve, with water then filling the entire sight glass.  Although some unknown percentage of the 1/4 cup volume of rust has been forced through the sprinkler head orifice initially to produce the dust cloud, a lesser volume has remained, and proves enough to completely stop water flow.  A few small droplets of water trickle out as the fire continues.

Each video can be expanded to a full screen view.

• Failure Autopsy

Following the experiment, the fire sprinkler head was removed as well as the 1 in. to 1/2 in. reducer to the the sprinkler head.  Close inspection showed both totally filled with rust product.

In the below left photo we show the threaded body of the sprinkler head completely filled with rust debris of varying size.  At center, some of the rust product initially released after the bulb ruptured, rests on its deflector.  Below right we found the full volume of the 1 in. by 1/2 in. reducer also packed with rust, which means that less than 1/4 cup of rust was capable of filling not only the sprinkler head body, but the entire reducing adapter as well – a substantial blockage!

• Possible Influences

Very clearly, there are many variables which would have influenced this basic experiment, some of which include:

• • • Rust Composition
      

Of all factors, the widely variable consistency of rust would influence its migration downstream as well as its ability to clog a fire sprinkler head. Predicting its impact, however would be almost impossible.

Most rust product soaking in water laden pipe would be more liquid than the rust used in this experiment, which means it is less likely to be blown through the sprinkler head as a dust cloud as our demonstration revealed.  Additional experiments are planned using wet rust.

Past examinations of removed dry and pre-action fire sprinkler pipe have shown a wide variation in the size of the deposits produced by corrosion, but with generally larger size chunks of rust from the larger mains far more likely to clog the small diameter orifice of a typical fire sprinkler head.  In that respect, the finer composition of the rust used in this experiment would likely favor not producing a clog, yet it did.

An extension of that issue is the greater probability of larger rust deposits to produce a clog, and that an even lesser volume of larger size rust product would have produced the same failure event.

• • • Length Of Travel

The rust in our experiment moved through approximately 4 ft. of smooth new pristine 1 in. pipe before reaching the sprinkler head.  In the real world, however, water may travel hundreds of feet beginning at the valve, through various mains and distribution lines, changes in direction, and then finally flow into the branch piping.  This would produce turbulence which in turn would scour additional rust product from the pipe bottom and likely produce a heavier slurry or water / rust mixture.

• • • Sprinkler Head Location

The location of the fire would also be an issue, with a fire activating the first positioned sprinkler head closest to the valve receiving less rust product than the very last sprinkler head receiving the most.  From some of the investigations CorrView International has performed, some showing internal rust product covering the lower 40% of a 4 in. main, the possibility of any water ever reaching a remote downstream fire would be zero.

• • • Flow Velocity

The source of water for this experiment began with a standard 3/4 in. garden hose connection increased to 1 in. schedule 40 threaded pipe representing a branch line.  In reality, most such fire protection systems are be fed by a larger 4 in. main and 4 in. dry valve capable of much greater pressure, water flow, and forward directed force.

On one hand, this greater water pressure would have a higher potential to force rust through the sprinkler head orifice.  The likely greater result, however, would be for the higher flow rate and velocity to release and then push a larger volume of rust downstream.

The combination of a longer pipe run of larger diameter pipe having a larger volume of rust product being forced by a greater velocity of turbulent water would unquestionably force far more rust into a fire sprinkler head than used in this experiment.

Taking all factors into account, we can suggest that a greater volume of rust product would be loosened and pushed downstream as the size of the fire protection system increases.

• Conclusions Of The Experiment

There is no intent to suggest that any other dry fire protection system will fail under similar conditions although the possibility is clearly obvious.  With potentially hundreds of feet of fire pipe of larger diameter present for an actual system, thereby allowing the production of far greater rust product, substantially more rust will be moved downstream.

In past ultrasonic investigations, we have statistically estimated greater than 1/4 cup of rust product per linear foot of 4 in. main piping, and have confirmed that estimate following visual inspection.

Our initial interest was to perform the experiment beginning with a very small amount of rust product; fully anticipating such contamination to produce no impact.  After that, we would gradually increase the volume of rust product.

As we quickly learned, this extremely small volume of rust at approximately 1/4 cup and 2.26 oz. was more than enough to prevent the sprinkler head from functioning.  The experiment was repeated multiple times with the same result.

Reducing the charge of rust to 2 tablespoons, or half the prior amount, did not stop the flow of water completely, but greatly reduced its volume. 

You can view and download our two page handout on this subject below.

© Copyright 2023 – William P. Duncan, CorrView International, LLC