GCPS 2010 __________________________________________________________________________
Condensation Induced Water Hammer: Principles and Consequences
Carlos A. Barrera Exponent 149 Commonwealth Dr. Menlo Park, CA 94025 (650) 688-7051
[email protected] Abid Kemal Exponent 149 Commonwealth Dr. Menlo Park, CA 94025 (650) 688-7181
[email protected]
Prepared for Presentation at American Institute of Chemical Engineers 2010 Spring Meeting 6th Global Congress on Process Safety San Antonio, Texas March 22-24, 2010 UNPUBLISHED
AIChE shall not be responsible for statements or opinions contained in papers or printed in its publications
GCPS 2010 __________________________________________________________________________
Condensation Induced Water Hammer: Principles and Consequences Carlos A. Barrera Exponent 149 Commonwealth Dr. Menlo Park, CA 94025 (650) 688-7051
[email protected] Abid Kemal Exponent Keywords: water hammer, condensate, steam, condensation, CIWH
Abstract An explosion occurred in the boiler room of a food processing plant as the steam generation system was coming online after a maintenance shutdown. The explosion shattered the case of the condensate return pump fatally injuring an employee, while another employee was burned by the escaping steam. The low-pressure saturated steam needed for the food processing plant was generated in a heat exchanger that used highpressure exhaust steam from the turbine of a power plant. The resulting condensate was collected in a tank and pumped back to the power plant for reuse. The conclusion of the incident investigation was that the pump was destroyed by a condensation induced water hammer (CIWH) generated in the condensate return line; in this phenomenon, a volume of hot steam can get trapped by subcooled liquid, generating high-pressure shock waves when it rapidly collapses due to condensation. A CIWH can be devastating and, in this case, shattered the cast iron case of the pump and resulted in a workplace fatality. Understanding the physical principles behind CIWH is fundamental in preventing its occurrence and the losses associated with it. These principles are reviewed as part of the explosion’s root-cause investigation, and recommendations are provided for the prevention of CIWH.
1. Introduction Condensation induced water hammer (CIWH) is a transient two-phase flow phenomenon well known in the nuclear power and refrigeration industries. In the chemical and process industries, CIWH does not appear to be as well documented and understood as compared to water hammers caused by the sudden closure of valves. Steam and condensate handling systems are at risk of experiencing a CIWH; the high-pressure shockwaves generated during such an event can rupture pipes and fittings, check valves, blow seals or damage equipment, as well as break structural supports and anchor points.
GCPS 2010 __________________________________________________________________________
The consequences of CIWH in process steam generation and distribution systems are often destructive and can be fatal. Care must be taken to prevent its occurrence by following sound design guidelines in new projects, and, in existing facilities, any CIWH event must be identified, investigated and corrected. This paper describes the root-cause investigation of the explosion of a condensate return pump at a food processing plant. The basic principles describing the CIWH phenomenon are presented and applied to the specific incident scenario, leading to the conclusion that CIWH was the cause of the explosion, with the ultimate root cause being a combination of design and operational issues.
2. The steam generation system During normal operation, the food processing plant generated all of its low pressure (150 psig) utility steam on site in a kettle-type heat exchanger (boiler). The exchanger used high pressure exhaust steam from the turbine of the power plant located more than 2,000 feet away. The condensate generated in the boiler was collected by gravity in an adjacent horizontal tank (condensate tank), and sent back to a condenser, which operated under vacuum, in the turbine building at the power plant. Although the condensate could be returned by using the pressure differential between the condensate tank and the condenser, a 100 hp condensate return centrifugal pump (condensate pump) was used to boost the flow rate when necessary. The feed water for low-pressure steam was preheated by the condensate inside the condensate tank. The plant had a backup gas-fired boiler that was used to generate the low-pressure steam when the power plant was out of service. Both the boiler and the condensate tank were thermally insulated and located outdoors. The centrifugal pump was located inside the boiler building. See Figure 1 for a simplified schematic of the equipment involved. The steam generation system was controlled by a programmable logic controller (PLC). The variables measured and controlled included: pressure of the incoming and outgoing steam, flow of low-pressure steam to the food processing plant, water level in the boiler, liquid level in the condensate tank, and status of the condensate pump operation, in addition to some water quality parameters. The flow of high-pressure steam and condensate were controlled by split-range control valves located in the power plant and were connected to the PLC via an optical fiber link. The mass flow rates of high-pressure steam and condensate return were measured and recorded at the power plant, and were used for costing purposes; the food plant PLC did not have real-time access to these two critical variables.
3. Incident description On the day of the incident the power plant was bringing its turbine back online after an extended maintenance shutdown. During the power plant turbine shutdown, the food processing plant had been using its backup gas-fired boiler to meet its low-pressure steam needs.
GCPS 2010 __________________________________________________________________________
Figure 1.
Simplified steam plant configuration
The ambient temperature had been below freezing for several weeks. The boiler and condensate tank were cold and the condensate pump had not been used during the shutdown. As the steam pressure increased in the transfer lines, personnel at the food processing plant repeatedly noticed that the condensate tank’s pressure relief valve was opening and relieving water. This situation continued during the rest of the morning. Just after lunch, an explosion shattered the cast iron case of the condensate pump and one of the ensuing fragments struck an operator, killing him instantly; another employee was severely injured by the escaping steam. The steam generation and distribution systems were shut down and the boiler building sealed until an investigation was conducted.
4. CIWH background and principles The earliest reports regarding accidents caused by CIWH date back to the end of the 19th century in England [1] where steam boilers were in widespread use, but their behavior was not well understood and boiler explosions were common. In modern times, CIWH continues to be a safety issue in two-phase flows such as steam-condensate systems [2]. Such issues are well understood in the nuclear power industry [3] where the consequences of an accident can be catastrophic and the physical conditions inside the reactor vessel and steam generators are very demanding. Another sector that has historically paid attention to CIWH is the refrigeration industry; in particular, large industrial ammonia-based refrigeration plants [4]. In this case, CIWH is a misnomer,
GCPS 2010 __________________________________________________________________________
since no water is involved and condensation induced shock (CIS) is the preferred term. District heating or steam distribution systems are another area where the frequency of incidents involving CIWH is high, for instance, the rupture of a 120 psig 6-inch cast iron valve when an operator was opening a main isolation valve located in a confined space at the U.S. Department of Energy’s Hanford Site on June 1993 [5, 6]. Although the operator escaped the confined space, he later died due to burns and lung damage caused by the steam. In the chemical and process industries, a recent example is the rupture of an 18-inch medium pressure steam main at the BP Grangemouth refinery in Scotland on June 2000 [7] after steam traps were blocked for maintenance of the line. No injuries were reported in this incident. A review of literature indicates that condensation-induced water hammer occurrence is correlated with pipe geometry, the filling or draining rate and sequence, and the temperature difference between the steam and the condensate. CIWH can occur in vertical or horizontal pipes that are full of condensate and steam is being admitted from one end, or in pipes initially full of steam in which condensate is introduced. The necessary condition for the occurrence of CIWH is the presence of sub-cooled condensate in contact with saturated steam. Based on these parameters, CIWH events can be classified into three groups: 1. Condensate driven by steam: the condensate accumulates in a pipe or fitting initially filled with steam, causing additional condensation and ultimately a slug of liquid. 2. Condensate drawn by vacuum: steam is admitted into a space in the presence of condensate; the steam condenses and draws the condensate towards the collapsed space. 3. Flash steam: a pressure reduction of super-heated liquid generates flash steam that can quickly condense, generating a vacuum and drawing more liquid into the collapsed space. The onset of CIWH in the above-mentioned scenarios is directly related to transitions in two-phase flow regimes, which go from stable to unstable. For horizontal pipes, the initial regime can go from fully liquid to stratified to slug flow; these transitions can occur while the pipe is being filled or drained. In both instances, the steam will start flowing over the condensate and, depending on the relative velocity, it can start generating ripples on the liquid’s surface. These ripples can grow to occupy the whole pipe, generating a slug that can be pushed by the high-pressure steam, or, in the case of sub-cooled condensate, can trap a bubble of steam, condensing it rapidly and generating a vacuum due to the large difference in specific volumes of steam and condensate (Figure 2 illustrates this process). For vertical pipes, the collapse of a steam bubble can occur while the bulk liquid is traveling upwards with the steam bubble in the middle and, due to heat transfer, the bubble collapses, or it can also occur when the vertical pipe is filled with steam and
GCPS 2010 __________________________________________________________________________
liquid is admitted from the bottom, rapidly cooling the steam and creating a vacuum in the top portion that sucks in the incoming liquid. Both of these scenarios can occur in steam distribution or condensate return lines, or at the inlet/outlet of process equipment. In both cases, the common denominator is saturated or super-heated steam coming into contact with a sub-cooled condensate, resulting in a rapid condensation and subsequent collapse of the steam pocket. The subcooled condensate can accumulate in low points or dead ends of the pipe network, or downstream of valves or equipment that have been closed or shut down for some period of time. If the two phases are maintained at the same temperature, there is no possibility for sudden condensation to occur, although a water hammer can still occur when high pressure steam drives a slug of condensate at high speed down a pipe and hits a fitting or closed valve.
Figure 2.
CIWH generation process in a horizontal pipe
The transition between stratified and slug flow regimes has been extensively studied [8,9], and it has been determined that the minimum condensate flow rate associated with slug formation corresponds to a Froude (FR) number of 0.5. Froude number is a ratio between the inertial forces and the gravitational forces acting on the fluid; in its simplest form it is given by FR
V gD
[eq. 1]
GCPS 2010 __________________________________________________________________________
where V is the average condensate velocity, D is the pipe diameter and g is the gravitational constant. The same studies found that flow rates higher than those corresponding to a FR of one do not generate condensate induced water hammers regardless of the pipe size. This implies that, for a 4-inch diameter pipe, liquid flow velocities higher than 200 ft/min typically would not result in a CIWH event. A more detailed measure of the stability of the two-phase flow in horizontal or nearly horizontal circular pipes is given by the Taitel-Dukler (NTD) parameter [10], which is a function of FR and the geometry of the two-phase system, and is given by
N TD
φFR* dL 1 D
2
[eq. 2]
with the Froude number now taking into account the actual interface geometry between steam and condensate and given by 2
F * R
m L Si 2
ρ L gA L
[eq.3]
3
and is a dimensionless group that accounts for the relative velocity of the two phases and the void fraction () given by 2 1 α ρ s VS
α
α
ρ L VL
AS πD 2 4
2
[eq. 4]
[eq. 5]
where dL is the liquid depth, mL is the liquid mass flow rate, Si is the steam-liquid interface perimeter, L is the liquid density, AL is the liquid flow area, AS is the steam flow area, S is the steam density, VS and VL are the steam and liquid velocities, respectively. For partially filled pipes, the transition between stratified and slug regimes will occur if NTD is equal or larger than one. This condition is bounded by the liquid flow rate that causes the pipe to run full, which can be calculated using
GCPS 2010 __________________________________________________________________________
16m L
2
2
π 2 ρ L gD 5
0.25
[eq. 6]
For horizontal pipes, the inclination and length of straight sections have been shown [9, 11] to be factor in the occurrence of CIWH. The maximum pipe inclination able to generate a CIWH event is 2.4°; this is related to transition between stratified and slug flow. The occurrence of CIWH was also found to be related with the pipe length to diameter ratio (L/D); with events occurring only for L/D values between 24 and 48. The condensate temperature is an important factor because it determines the condensation rate of the steam. CIWH requires rapid condensation and it has been shown that at least 36°F of sub-cooling are necessary to generate a water hammer event [11]. The pressure spike (P) caused by a CIWH event can be initially calculated using the well known Joukowsky equation [9, 12] given by
P kaV
[eq. 7]
with the sonic or pressure wave velocity (a) given by
a
β ρL
[eq. 8]
β D 1 E b
where is the liquid bulk modulus of elasticity, E is the elastic modulus of pipe wall, b is the pipe wall thickness, and k is a parameter that takes into account the geometry of the surface where the wave impact occurs (1.0 for a closed surface like a valve or closed pipe end, 0.5 for another liquid column). In practice, eq. 7 does not give an accurate estimate for CIWH events because it assumes a solid liquid column. Taking into account the void fraction in the pipe () and the driving pressure, eq. 7 can be rewritten as P 0.707 a
P0 Pv L 0
10
[eq. 9]
where P0 is the system pressure (steam) and Pv is the pressure inside the void, which can be taken as the saturation pressure of the surrounding condensate. The maximum pressure spike will occur at the highest void fraction that can still generate a slug flow regime.
GCPS 2010 __________________________________________________________________________
Most CIWH incidents occur during transients associated with bringing equipment online after repairs or shutdowns. During these events, the steam and condensate lines are cold and usually full of liquid. Once steam is readmitted into the lines, the steam traps might not be able to handle the high volume of condensate produced and accumulation of liquid in the low points of the line can induce a CIWH event. During steady-state operations, CIWH can occur when steam traps or valves fail or are closed by accident, or when additional condensation in steam lines is produced by external factors such as damage in the thermal insulation or the lines being submerged underwater due to a flood or leak.
5. Incident investigation A root cause investigation was conducted that included the review of design documents, P&IDs, computer process data trends, set point and alarm data, and visual inspection of the facilities and remains of the pump. The investigation determined that the pump casing was destroyed by a single overpressure event and that such an overpressure could only have been caused by a CIWH in the line connecting the condensate tank to the pump. The investigation also showed that three factors combined to allow the high temperature steam to come in contact with the sub-cooled condensate in the condensate return line to cause the event: 1. Malfunction of the level transducer on the condensate tank. Review of the process data trends revealed that, during the morning of the day of the incident, the condensate tank’s level transmitter was sending erratic readings. The transducer could have been frozen at a medium level due to the cold weather predominant in the area at the time. Witnesses saw water coming out of the pressure relief valve of the condensate tank. The excess water came from cold condensate accumulated in the steam transmission lines as these were being warmed up, and from steam condensation in the lines and the boiler. This condensate was not being sent back to the condenser fast enough since, for the PLC, the condensate tank had a normal level and it did not turn on the condensate pump. The condensate tank High-High level trip for the steam valve was never activated. At some point before the explosion, the level transducer partially thawed, indicating a high level and signaling the PLC to start the condensate pump. 2. Extremely low value for the Low-Low condensate tank level trip of the condensate pump. Review of the level set-points valid at the time of the incident revealed that the condensate tank Low-Low level trip for the pump was less than 1%. Due to the negative pressure differential between the condensate tank and the turbine condenser at this extremely low level, the condensate tank can become empty
GCPS 2010 __________________________________________________________________________
even after the PLC has shut down the pump, allowing hot steam to partially fill the suction line. 3. Lack of safety interlock between the condensate tank’s Low-Low level and the high pressure steam supply valve. In case the condensate tank runs dry, high temperature steam could be admitted into the condensate return line because the valve that controls the steam supply to the boiler remained open. The only interlock present between the condensate tank and the steam supply valve was for High-High level trip. Since the food processing plan was not demanding any low-pressure steam from the boiler at the time, the condensation rate was low and the condensate tank would have been filled with high-pressure steam that entered the condensate return line filled with cold condensate. Inspection of the 4-inch piping connecting the condensate tank to the inlet of the condensate pump revealed that its horizontal segments had almost no declination and that one segment was more than ten feet long with an L/D ratio of 32, which falls between the CIWH occurrence limits of 24 and 48. The temperature of the condensate in the condensate return line at the time of the incident, although not monitored, is expected to have been well below saturation since the steam lines had been offline for several weeks; this ensures enough sub-cooling of the liquid to trigger a CIWH. Other factors not directly related with the CIWH generation but that could have prevented the accident are: 1. Lack of comparator alarm between steam and condensate mass flow rates. Review of the trend data revealed an inconsistency between the high-pressure steam and condensate return mass flow rates; more condensate was being returned to the plant than steam was being admitted into the boiler. This could have been an indication that the condensate tank was running full even though the level transducer was reporting a normal level. A comparator alarm in the PLC would have alerted the control room personnel about this discrepancy and possibly have stopped the warm-up operation until it was remedied. 2. Lack of field switch for the condensate return pump. The condensate pump could only be turned ON or OFF from the control room via the PLC. A field switch to manually turn ON the pump would have been useful for the boiler operator to force down the liquid level in the condensate tank. 3. High-pressure steam line drip leg was blocked.
GCPS 2010 __________________________________________________________________________
Due to the cold weather, at least one steam line drip leg was frozen shut, preventing condensate draining and forcing more condensate into the boiler and the condensate tank. 4. Not venting the steam and condensate during line warm up. The high pressure line should have been brought back online by slowly flowing steam and venting it at the food processing plant instead of attempting to run it in a closed loop, returning the condensate to the turbine condenser. In summary, the condensate tank was running full during the steam line warm-up operations due to erratic signals from a frozen level transducer. When this transducer partially thawed, the PLC started the condensate pump, which rapidly lowered the level to the low-level mark. At this point, high-pressure steam was still being sent to the boiler and the condensate tank, and, due to the pressure differential between the condensate tank and the turbine condenser, the tank emptied, allowing saturated steam into the condensate return line, triggering a CIWH in one of the horizontal sections, which in turn caused a single overpressure event that destroyed the cast-iron case of the condensate pump.
6. Recommendations for CIWH prevention The following best practices [13, 14] should be applied during normal operation, maintenance shutdowns, and the design of new facilities in order to avoid any instance of CIWH.
Long pipe runs should have a declination of 0.5 in/ft or greater in the direction of steam flow. Pipe layouts with straight horizontal sections that have an L/D ratio between 24 and 48 should be avoided when possible. Steam and condensate lines should be completely drained before returning them to service. Complete line draining during shut-down periods should be accomplished by steam traps and opening supplemental drain valves. Steam lines should be brought back online slowly, with the steam and condensate running though the system and into a blowdown tank until the system is completely heated. Steam main valves should not be cracked open to warm up the line; a smaller bypass valve should be initially used until the pressure has equalized. A drip leg pocket should be installed upstream of any main steam isolation valve. Dead-legs or low points where condensate can accumulate should be avoided. Globe valves and strainers are known to accumulate condensate; provision should be made to drain any accumulated liquid, such as mounting the valve with its spindle horizontally. Condensate line branches should be connected to the main line from above.
GCPS 2010 __________________________________________________________________________
Material compatibility and strength should be checked during the design phase and when components are replaced. Pipes, fittings and other steam/condensate system components should be rated to withstand not only the working pressure but also some pressure transients. “Weak links” in the systems should be avoided by using components (valves, pump casing) with appropriate pressure ratings. For existing facilities, all steam traps should be reviewed to verify correct sizing and working order, particularly before a startup. For new facilities or during improvement projects, conduct a process hazard analysis that takes into account CIWH.
7. Conclusion The explosion that destroyed the condensate return pump and killed an employee was caused by a CIWH in the suction line of the pump. This accident could have been prevented and is a reminder of the possible magnitude of such events. The condensation induced water hammer could have been avoided through a combination of design and procedural best practices, including the calculation of several numerical parameters aimed at diagnosing and avoiding the simultaneous presence of saturated steam and subcooled condensate in a pipe, performing a process hazard analysis, and selecting appropriate trip and alarm levels in the control system.
8. References [1]
Rimmer, E. J. “Boiler Explosions, Collapses and Mishaps,” Constable & Company LTD London 1912.
[2]
Woodruff, E. B, Lammers, H. B., Lammers, T. F. “Steam-Plant Operation,” Sixth Edition. McGraw-Hill, Inc. New York 1992.
[3]
Chou, Y., Griffith, P. “Avoiding Steam-bubble-collapse-induced water hammers in piping systems,” Interim report NP-6447 EPRI, Palo Alto, October 1989.
[4]
Shelton, J. C. and Jacobi, A. M. “Fundamental study of refrigerant-line transients: Part 1 – description and survey of relevant literature,” ASHRAE Transactions, Volume 103, No. 1, 1997 p. 65-87.
[5]
“Averting water hammers and other steam/condensate system incidents,” Environmental, Safety & Health Safety Bulleting Issue No. 95-1 U.S. Department of Energy Washington D.C. June 1995 www.hss.energy.gov/publications/esh_bulletins/BULL0099.html.
[6]
Debban, H. L. and Eyre, L. E. “Condensate Induce Water Hammer in a Steam Distribution System Results in Fatality,” U.S. Department of Energy KH-SA3026-FP February 1996.
GCPS 2010 __________________________________________________________________________
[7]
“Major incident investigation report BP Grangemouth Scotland: 29th May - 10th June 2000,” Health and Safety Executive Bootle, UK. www.hse.gov.uk/comah/bpgrange/execsumm/descript.htm.
[8]
Wallis, G., Crowley, C. J., Hagi, Y. “Conditions for a pipe to run full when discharging liquid into a space filled with gas,” ASME Journal of Fluids Engineering, Volume 99 1977 p. 405-413.
[9]
Bjorge, R. W. and Griffith, P. “Initiation of waterhammer in horizontal and nearly horizontal pipes containing steam and subcooled water,” Journal of Heat Transfer Volume 106 November 1984 p. 835-840.
[10]
Esselman, T. C., Semprucci, L. B., Van Duyne, D. A. “Test data and comparison to analysis for condensation induced waterhammer,” Proceedings of the 11th Annual Topical Meeting on Nuclear Reactor Thermal-Hydraulics (NURETH-11) October 2005 Paper 068.
[11]
Griffith, P. “Screening Reactor Steam/Water Systems for Waterhammer,” NUREG 6519, U.S. Nuclear Regulatory Commission. 1997.
[12]
Wylie, E.B., Streeter, V. L., Suo, L. “Fluid Transients in Systems,” Prentice Hall Upper Saddle River 1993.
[13]
Mannan, S. “Lee’s Loss Prevention in the Process Industries,” Third Edition Volume 1. Elsevier Butterworth-Heinemann Burlington 2005.
[14]
Tholey, A. R. D. “Fluid Transients in Pipeline Systems,” Second Edition ASME Press New York 2004.
