Home--||--Steam Accident Investigations--||--HVAC Projects--||--Authored Articles--||--Steam Safety Training--||--The Firm--||--Contact Us
|The steam pipe started to vibrate and shake. Don yelled at Clyde--"Let's get the Hell out of here...this thing's going to blow!" Clyde stuck his head out from beneath the steam pipe from where he was removing insulation. He heard a loud roar rumbling down the steam line like a freight train coming from the direction of the C-4 Manhole. Don was already clamoring up the exitbehind Don. Don was trying to break through the Visqueen plastic sheet that covered the manhole. It was sealed tight to prevent asbestos fibers from escaping. A white steam cloud rolled down the utilidor fr ladder. Clyde slid from beneath the maze of pipes and scrambled up the ladder om the direction of C-4 and began to flood the manhole. Another worker fleeing the encroaching steam crawled up behind them. Together they desperately tore through the stubborn Visqueen seal until it finally gave way, shoved open the steel hatch above, and tumbled out into the fresh air above. The swelling heat from the utilidor rose around them. Up top there was pandemonium. Steam was billowing out the C-4 manhole as well as the manhole they'd just exited. Fire engines were arriving. Men were shouting trying to figure out who was still down in the utilidors. Bobby and Wayne had not gotten out.
Moments earlier, before the Accident, Bobby had opened the 10" gate valve at Manhole C-4 a second incremental turn. He thought "this is strange, the valve's handwheel spun freely". Just fifteen minutes earlier, he'd "cracked open" the 10" cast iron valve to admit steam into the 2,200 foot steam line to begin warming it up. For 3 weeks now, he'd been energizing the G-line at the end of the asbestos worker's shift and had never had the system warm up this quickly. It usually took from 30 to 45 minutes. When the handwheel spun freely, he understood the lack of friction to mean that steam pressure on either side of the valve had equalized, so the warm-up was complete and he could open the valve the rest of the way. This seemed too quick though. He'd better check with his supervisor before spinning the valve open the rest of the way.
Bobby nudged past his co-worker, Wayne, as he made his way over to the material passout and yelled up through the plastic flaps to his boss --"She's spinnin' freely, is it okay to open her up all the way?". The supervisor was puzzled too. "No", finally came the muffled response, "better continue to open her a little at a time like we were told to do". About a minute had elapsed since Bobby had opened the valve enough to lift it off its seat. As he turned back to the valve, a "pop" was heard. Then a moment later, "KABOOM"! Hot water and steam exploded from the 10" valve. A white cloud of flashing condensate and steam engulfed the utilidor with a palpable wave of heat. Wayne was knocked down and stunned by the scalding water spraying from the valve. Egress via the manhole exit was cut off by steam spraying from the valve. The only way out appeared to be through the material passouts constructed into the roof of the utilidor. Bobby clambered up on top of the pipes and jumped up catching his armpits above the opening. From there he was able to hoist himself through the plastic covered opening. He emerged with second and third degree burns but otherwise okay.
Wayne stumbled through the piping to the other material passout. His first jump was too weak and he fell back onto the piping which by now was becoming slippery with condensing steam. Air temperature in the utilidor was approaching 200 degreeF. Wayne desperately collected himself. He knew that this might be his final chance. He groped his way back up onto the slippery pipes, took a breath of the searing air, and leapt up again into the plastic covered opening. This time he was able to hook one elbow above the rim and, with his life on the line, kick up through the opening.
Clyde and Don saw Wayne crawl out thru the plastic flaps of the material passout. He rose to his feet and started screaming for help. His protective clothing was shredded, loose skin was sloughing off his exposed arms and legs. He was badly burned. Clyde yelled at passers-by to call an ambulance as they ushered Wayne away from the steaming manholes. Soldiers with a knowledge of first aid rushed him to a barracks across the street and started to apply cold packs to his burns and give him cold drinks. Wayne's throat was beginning to constrict. An ambulance arrived to rush he and Bobby to the Hospital. As the injured workers were being cared for, Clyde turned his fury on his supervisor -- "You stupid son-of-a-bitch, we told you this would happen."
For four weeks asbestos workers had been removing asbestos insulation from the 2,200 foot section of steam main known as the G-Line and the 120 foot H-Line. (See Figure 1) Like all steam mains at Fort Wainwright, Alaska, the G and H Lines ran underground in narrow utilidors filled with pipe. Originally, the contractor had tried to abate the steam main with the lines energized. This proved to be near impossible for the workers. Utilidor temperatures reached 160 degree F as insulation was removed from the 325 degree F pipe carrying 80 psig steam. Laborers who had to be suited-up and masked to work in the asbestos laden environment were dropping like flies from the heat and/or quitting. The contractor was forced to seek relief from the Owner. A compromise was negotiated after the first week-- steam would be de-energized at midnight before each workday, asbestos abators would start work at 4:00 a.m. and finish by noontime at which time steam would be restored. The asbestos removal contractor would be responsible for de-energizing and re-energizing the steam line daily. For the three weeks before the accident this was the procedure. By the beginning of the laborers' workday, temperatures in the utilidors were still around 120 degree F but, with frequent breaks to cool off and re-hydrate, conditions were tolerable.
Abatement began at Manhole G-1 and headed south toward C-4 at the rate of about 125 feet a day. As abatement proceeded down the G-Line, local traps serving the uninsulated portion of the line were overwhelmed with condensate during the period the lines were energized each day. In the first two weeks, however, this didn't cause a problem. Excess condensate merely rolled down to C-4 on the south end and G-1 on the North end. Traps on the south end still serving insulated portions of the line had adequate capacity to remove the excess condensate. On the north end, the steam valve was left closed so trouble was avoided. After two weeks of daily start-ups without serious incident, save some minor waterhammers, asbestos crew operators grew confident that start-up of the steam line was no big deal. By the beginning of the third week, insulation removal had reached Manhole G-9. Calculations show that at this point the rate of condensate being generated in the southern section of the G-Line began to exceed the net capacity of the traps to remove it.
Condensate accumulation during steam operation is potentially destructive, but even so, as long as condensate is religiously drained everyday before start-up, a catastrophic waterhammer accident might still be averted. The problem was--condensate wasn't being drained religiously. The asbestos workers given responsibility for energizing the steam main daily didn't fully appreciate the danger inherent in starting up a high pressure steam system with condensate in it. They did not routinely open drain valves to bleed the system of excess condensate either at night, when they shut the system down, or at noontime, when they re-admitted steam through the C-4 valve to re-energize the steam main. Their belief was that steam admitted through the C-4 valve would blow condensate to the far end of the main at G-1. Thus, in their view, only the drain at G-1 "really" needed to be opened at start-up. Accordingly, there was a tacit understanding that the bleeder valve at G-1 would be opened daily by the quality control supervisor for the prime contractor, and any condensate that wasn't drained at start-up, they apparently thought, would be mopped up by traps after start-up.
As the third week began, the severity and frequency of waterhammer began to accelerate. Residual condensate accumulated in the steam pipe at C-4 due not only to operation of the uninsulated steam main, but also due to condensate formed at start-up that went undrained. Early in the third week, heavy banging forced workers to evacuate the utilidor. Clyde, one of the more vocal evacuees, warned the abatement supervisor "this thing sounds like its ready to explode... What are you going to do about it?"
Condensate also accumulated each night in the double-elbow riser to the south of the C-4 valve. (See Figure 3 below). During the period after midnight when the C-4 valve was closed, steam would condense in the uninsulated double-elbow riser and come to rest against the south side of the closed valve. From midnight until noon the following day, enough condensate accumulated to almost fill the riser.
One other event shaped the conditions just prior to the accident. On the day of the accident, there was a change in the start-up procedure. In order to isolate another steam main for repair work, the contractor needed to energize the G Line early. The asbestos crew was instructed to start up the G Line an hour and fifteen minutes before their quitting time. In addition, but unbeknownst to the asbestos crew, the contractor's quality control supervisor decided to expedite warm up the G-Line by admitting steam through the G-1 valve at the far end of the G Line as well as through C-4. The G-1 valve was opened as much as 30 minutes before Bobby first cracked open the C-4 steam valve. This had the likely effect of sweeping undrained condensate residing against the G-1 valve on the north end of the line south to C-4, and completely filling the H-Line as explained in the sequence of figures a thru c in sidebar at
Subcooled condensate filled the steam line on both sides of the C-4 valve as well as completely filling the H Line. High pressure steam admitted through G-1 had pressurized the steam main and was sitting atop the condensate on the north side of C-4. The south side of the valve was also under steam pressure which, based on testimony, was likely slightly less that that on the north side.
#1. Is this Situation Dangerous? Some steam people would say "no, as long as there is no fast moving steam, there's no danger of waterhammer. Opening C-4 slowly and incrementally should prevent steam or condensate from moving quickly and thus prevent a waterhammer." This is wrong, dead wrong. High pressure steam in contact with subcooled condensate is dangerous. It's a recipe for Condensation Induced Waterhammer. The sidebar (See sidebar near bottom page) explains why this type event is 10 to 100 times more powerful that conventional "steam flow" driven waterhammer.
#2. What Would You have Done in Bobby's place? If your answer is, "I'd first open the C-4 bleeder valve to drain the condensate," you're toast. Although this is the answer most steam operators would give, it will trigger the accident. Neither the bleeder valve nor the steam valve can be opened without provoking this accident. To understand why, it's crucial for steam fitters and operators to understand the mechanism of Condensation Induced Waterhammer.
Condensation Induced Waterhammer
A condensation induced water hammer is a rapid condensation event. It could also be aptly termed a "rapid steam bubble collapse". It occurs when a steam pocket becomes totally entrapped in subcooled condensate. As the steam gives up its heat to the surrounding condensate and pipe walls, steam changes from a vapor to a liquid state. As a liquid, the volume formerly occupied by the steam shrinks by a factor of from several hundred to over a thousand, depending on the saturated steam pressure. Likewise, the pressure in the void drops to the saturated vapor pressure of the surrounding condensate. (For example, the saturated vapor pressure of condensate at ambient temperature is less than 1 psia). This leaves a low pressure void in the space formerly occupied by the steam that the surrounding condensate, under steam pressure itself, will rush in to fill. The resulting collision of condensate generates an overpressurization which reverberates throughout the condensate filled portion of the pipe. How severe is the over-pressurization? Remember that water is virtually incompressible. In a collision, it does not give. Think of the last time you did a belly flop off the low dive. The water felt pretty "stiff" didn't it?
The specific factors which influence the severity of a condensation induced waterhammer are: (1) the steam pressure, (2) the degree of condensate subcooling, (3) the presence of non-condensables left over in the void, and (4) the size of the void. If the steam pressure is high, the condensate is subcooled, non-condensables are absent, and the void is large enough for a slug to pick up some velocity, the overpressure resulting from an event can easily exceed 1000 psi. This is enough pressure to fracture a cast iron valve, blow out a steam gasket, or burst an accordion type expansion joint. And, in fact, failure of each of these components in separate condensation induced waterhammer accidents has resulted in operator fatalities.
One might ask at this point, "But wait, isn't it common for steam and condensate to come into contact in a steam system?" Good design and operating practice aim to avoid mixing high pressure steam and excess condensate by making sure steam mains are properly trapped and live steam is kept out of condensate return systems. Nevertheless, it does happen. Condensate lines, for instance, are often heard to pop and bang when steam squirts into them through traps. Why don't the collapsing steam bubbles destroy condensate pipes? They can over time. But the shock waves generated are not catastrophic because the pressure in a condensate system is generally low-- on the order of just a few psi, subcooling is not great, and the steam bubbles are small. Of course, high pressure steam can contact subcooled condensate in steam lines when something goes wrong-- for example when a trap assembly becomes plugged with scale causing a drip leg to fill with condensate. Why don't situations like this result in destructive condensation induced waterhammer? One reason is pipe geometry. A steam bubble must become entrapped for there to be a collapse. In a vertical pipe such as a drip leg where steam is above the condensate, it's difficult to entrap the steam because natural buoyancy tends to keep the two fluids separate . In fact, research experiments show that it's difficult to entrap a steam void in any pipe sloped downward in the direction of steam flow more than 1/2" in 1.0 foot . At slopes less than this, however, and in upwardly sloped pipes, it's a different story.
1. Steam residing over subcooled condensate losses heat to the condensate and the surrounding pipe which causes the steam to condense. The continual loss of steam, induces fresh steam to flow in to replace it. Steam flow over condensate will tend to draw up a wave in the condensate via the Bernoulli effect .
2. If the rate of heat transfer is rapid enough for a given condensate level, the induced steam velocity will draw up a wave high enough to seal the pipe.
3. The creation of a seal immediately isolates the downstream steam pocket from the upstream supply creating a steam pocket. Ongoing condensation in the isolated steam pocket drops the pressure causing a slug to accelerate into the void.
The formation of a condensate seal is a necessary condition for a rapid condensation event in a horizontal line. Often, however, heat transfer is not rapid enough to induce sufficient steam flow to seal the pipe and cause a rapid steam bubble collapse. In fact, generally, to initiate a condensation induced waterhammer in a horizontal line where neither steam or condensate is flowing through the pipe, a "trigger" is needed. That's because in a non-flowing situation, heat transfer between the steam and condensate is retarded by a stagnant layer of hot condensate and non-condensables laid down by steam as it condenses atop cooler condensate. This interphase boundary layer insulates the steam void. On the one hand, the layer prevents rapid condensation, but on the other, it allows a potential steam void to grow in magnitude and potential energy like an over expanded balloon. Often times there will be no rapid condensation event if the layer goes undisturbed. Steam will fill a pipe atop subcooled condensate without incident. Minor collapses may occur, but due to the lack of rapid heat transfer, they will be mild and go unnoticed. If, however, the insulating layer is disturbed in such a way that the layer is breached at some point, then the local intrusion of subcooled condensate can result in a chain reaction which shatters the entire insulating layer. In a millisecond, the rate of heat transfer can increase a thousand fold inducing a rapid steam influx which seals the pipe and sets off a rapid condensation event resulting in condensation induced waterhammer. The key, then, to whether or not an event is initiated depends on the occurrence of a trigger to cause interface shattering.
Back at the Accident
Now, return to Manhole C-4 fifteen minutes before the Accident. Bobby had opened the bleeder valve at C-4 for the first time per a special instruction from the QC supervisor. He then proceeded to crack open the C-4 steam valve. Both of these actions presumably resulted in condensate draining from the system on the north side of the C-4 valve. The pipe volume vacated by the draining condensate at C-4 drew in steam along the top of the pipe from the north to replace it (Figure 5). Fifteen minutes after the first crack, Bobby opened the C-4 valve again, this time lifting the disk 1/2" off its seat. This action further speeded the removal of condensate and the advance of steam along the top of the pipe towards C-4.
From this point on, I'll describe what we understand happened based on tests with this model and others used to understand the accident. (Fig 6)
As the tongue of steam reached down the nearly horizontal line toward C-4, it probably collapsed several times as seals developed, but the collapses were not violent enough to be termed waterhammer events. When the steam finally reached the vertical opening to the H Line, the steam licked up around the corner seeking to buoyantly flow up into the H Line riser. This was the trigger necessary to set off the event. The tip of the tongue distended, then detached, releasing a bubble containing steam and non-condensables which rose up into the vertical H Line while an equal volume of subcooled water spilled down into the G Line. The remainder of the steam tongue snapped back into the G Line after releasing the bubble. The release of the bubble and the exposure to the cool condensate assaulted the stability of the boundary layer. It caused a ripple to reverberate down the length of the steam-condensate interface perturbing it and quickening heat transfer. This could have been sufficient to trigger the event. It depends on how much air has seeped into the system during cooling to suppress a collapse. As condensate continued to drain, steam advanced toward the H Line opening a second time again peaking around the corner, and again releasing a bubble of steam and non-condensables. This time the interface shattered. The entrapped steam pocket collapsed hard drawing a slug of water from the north toward C-4 crashing into the collapsing void faster than the eye could follow. The collision of the slug with the condensate at C-4 created an overpressurization that rebounded through-out the water filled portion of the system including up the H Line where Clyde and Don would have been working.
At Fort Wainwright, we believe the overpressure caused the double-elbow riser at C-4 to compress as shown below.(Fig 7).
The pipe and valve flanges twisted in response to the deflection of the double-elbows riser. The twisting flange caused the cast iron valve body to crack at the flange neck causing first condensate, then steam to spray from the valve. The actual damage to the valve is shown in Photo 2.
Could this accident have been prevented?
Of course. Numerous procedural blunders should be obvious to experienced steam operators and their supervisors as they read this article, not the least of which is assigning responsibility for start up of a high pressure steam system to asbestos workers. But I'm most interested in putting this question to the guy who's in the last line of defense-- the steam operator standing in Bobby's shoes with his hands on the valve's handwheel just moments before the accident. Suppose from the feel of the valve's handwheel, he surmises that there must already be full steam pressure on the steam line, and he believes not only that the line contains subcooled condensate, but that it is FULL and resting against the valve he is about to open. The question is: Is it possible, given the circumstances with which he's confronted, to avoid this accident?
The answer is YES. But, there's only ONE WAY. Cut the steam off. Don't open the C-4 steam valve. Don't open the bleeder valve. You've got to exit the manhole and close the G-1 steam valve, then drain the lines to empty the condensate. This is what must be done to avoid a condensation induced waterhammer in the situation described. Trying to drain the condensate with high pressure steam atop the subcooled condensate will trigger a rapid condensation event.
Here's what I want steam fitters and operators to know:
1. High pressure steam in contact with subcooled condensate is an unstable and potentially explosive mixture.
2. Don't admit steam into a line filled with subcooled condensate. In fact, always be wary of admitting steam to any cold steam line if you cannot be absolutely certain that the line's been completely drained.
3. Allowing subcooled condensate to flow into a steam filled line is more dangerous than admitting steam into a line with subcooled condensate.
4. If you suspect a pressurized steam line is filled with subcooled condensate, don't attempt to drain the condensate. Shut the steam off first, then drain the condensate. If you do open a drain, and the line hammers, close it and get the steam off. The line may continue to hammer until you get the steam off.
5. A mixture of steam above subcooled condensate can sit dormant in an isolated steam line like a loaded gun awaiting a triggering event. Opening a valve to admit steam or opening a bleeder to drain condensate can trigger an event. Don't let yourself or those you supervise inadvertently pull that trigger without first making sure the gun is unloaded.
Condensation Induced Waterhammer
"Waterhammer" according to a major steam trap manufacturer's engineering guide is "the impact caused by a sudden stopping of a rapidly moving slug of water". The guide goes on to explain that
The waterhammer described above is but one type of waterhammer. I term it "steam flow driven waterhammer". It describes an impact event where a slug of fast moving water strikes a stationary object and gives up its momentum much like an ocean wave striking a sea wall. The formula for the maximum impact pressure over the target area is:
where "rho" is fluid density and "v" is slug velocity. For water at 60 #/c.f. and v = 100 mph, Pmax = 279 psi. Lab experiments indicate that peak pressures for actual events are typically less than the maximum theoretical value .
Condensation Induced Waterhammer is a different animal. The pressure pulse generated by a Condensation Induced Waterhammer is due to the compression of water by a piston formed by a moving plug of water. This is the same phenomena which generates waterhammer in a hydronic single phase system (i.e. plumbing). The formula to calculate the magnitude of the maximum pressure pulse is:
where "c'" represents the speed of sound in water--about 4,300 fps. Note that this formula is similar to the former except that "c" replaces one "v". What's the relevance of the speed of sound? The sonic speed squared, "c2", is in essence a shorthand notation for the ratio of the stiffness of the material, represented by Young's Modulous "E", divided by the density of the material; i.e.
Clearly, the magnitude of a pressure pulse reverberating through, say, a piece of steel compressed upon collision would in some measure be a function of the stiffness of the steel. Same for water. Hence, the dependence on "c" is really a dependence on "E".
At 4,300 fps, "c" is roughly two orders of magnitude larger than "v". Thus, the overpressurization generated by Condensation Induced Waterhammer can be 10 to 100 times greater than that caused by steam flow driven waterhammer.
Condensate Collection in Vertical Lines
Condensate will fill a vertical take-off like the H-Line against gravity if the horizontal line beneath it becomes filled or nearly filled with condensate. To illustrate this point, I have exaggerated the rise of the H-Line in the figures below. Fig. A shows steam flowing into all open portions of the steam line and condensing. The condensing steam causes a reduction in local pressure that induces steam movement to flow in to replace it.
If the horizontal section of steam pipe fills or be-comes nearly full, a condensate seal forms that isolates the steam downstream of the seal. Condensing steam in the pocket causes the pressure to fall. The falling pressure in the isolated steam pocket will then suck up condensate into the pocket to fill the void. The result is shown in Fig. C.
Home--||--Steam Accident Investigations--||--HVAC Projects--||--Authored Articles--||--Steam Safety Training--||--The Firm--||--Contact Us