Kenneth O. Lapp
Westech Industrial Ltd.
Calgary
Correct flame-arrester design is important for safely destroying vapors from sour production facilities.
Experience with flame arresters in Alberta illustrates efficient and safe oil field battery vapor-control systems.
Because of increasingly stringent requirements for emissions control, oil producers in the U.S. and elsewhere are facing a problem Canadian operators have worked with for more than 15 years.
Westech Industrial Ltd. Calgary, in conjunction with various government agencies and industry groups, investigated the causes for failures in Alberta. The resulting report1 examined 10 separate flame-arrester failures in battery vapor-control operations.
These incidents, from 1979 to 1989, represent only a small fraction of the flame-arrester failures that take place every year in Alberta.
None of these incidents involved human injury, but losses averaged more than $100,000/incident. Fig. 1 shows tanks damaged as a result of flashback from a flare stack through an end-of-line arrester.
EXAMINED INCIDENTS
The 10 flame-arrester failures occurred in battery vapor-control systems (VCS) where the tank vapors were being flared because of H2S.
In a typical battery (Fig. 2), oil is pumped to a separator that removes most of the vapor. The oil is then routed to a storage tank. From there, vapors go through a storage-tank header to the flare stack. Nine of the incidents were explosions in storage tanks. The tenth was a flashback to a knockout tank.
Eight occurred during start-up, one took place during emergency shutdown, and the tenth occurred when the system was being shut down for repairs.
Flare stacks ranged from 40 to 100 ft in height and 2 to 4 in. in diameter. The arrester was placed at the base of the flare stack in nine of the installations. The tenth site had the arrester 74 ft from the base of a 40-ft high flare stack.
The storage tanks ranged from 400 to 1,000 bbl.
All of the cases involved flame propagation within a system that contained an explosive air/gas mixture. The air entered the systems because of improper operating procedures.
In eight cases, the systems were improperly purged during start-up, one system contained air because of faulty gauge-hatch seals, and the tenth system pulled air in through a gauge hatch left open during operations.
ARRESTER FAILURES
Three likely ignition sources for a battery VCS are the flare stack, static electricity, and exothermic reaction of pyrophoric iron sulfide with oxygen.
In all 10 cases studied, static electricity was ruled out as an ignition source because all tanks and pipes were properly grounded. Also, flow velocities and mist content were too low to generate a static discharge.
Ignition from the oxidation of iron sulfide was also ruled out because, in every case, the tanks contained nearly 100% air for significant periods of time before the incidents. The air would have neutralized any sulfide coatings in the system.
The conclusion is that all 10 incidents occurred as a result of flashback from the flare stack.
The failures can be divided into two major categories, failure of certified units and uncertified units.
Certified flame arresters had been tested to and accepted by a recognized test standard such as Underwriters Laboratories.2 The uncertified arresters had not been tested to or recognized by any official standards organization.
Seven of the cases (Table 1) involved uncertified, crimped-ribbon-style flame arresters; the remaining three arresters were certified (UL525) formed-plate arresters.
Of the 10 arresters, 5 had passed flames with no physical damage to the arrester. The other 5 had been damaged, compromising their quenching ability. This damage was caused by the impact of solids within the line, corrosion, or faulty initial assembly.
PREVENTING FAILURES
In the initial stages, a flame travels as a deflagration at relatively slow speed, about 12 fps in quiescent propane-air mixtures. As the flame travels down the line, the speed and pressure associated with the deflagration front increase.
Bends, elbows, and tees in the flame path create turbulence which can greatly accelerate the flame. Eventually the pressure and speed of the flame build to a point of detonation. At this point, the flame is traveling at supersonic speeds (up to Mach 7) with pressures as high as 3,000 psi.
Fig. 3 illustrates a typical flame profile for a 3-in. line, along with the limitations of the UL525 test. In all failures of certified arresters, the arresters were installed outside of the limitations established during testing.
The UL525 standard is intended for testing of end-of-line arresters, such as those installed beneath pressure-relief vents or installed at the end of a vent line.
The UL525 test requires the testing of an arrester to a specific length of flame run up, typically 5-15 ft. The flame is initiated from the open end of the pipe.
An arrester is outside of the performance capability established with the UL525 test if:
- Installed with more than this short length of pipe between the ignition source and the arrester.
- Used with a vapor more sensitive than that used for testing, such as vapors with an H2S content of 20% or more.
A flare with excessive height (Fig. 4) allows flashback speeds and pressures much greater than UL525 requirements.
Because of maintenance requirements for the arrester, it is impractical to install an arrester within 5-15 ft of the tip of the flare stack. As a result, virtually all arresters are installed at the base of the stack, resulting in conditions far exceeding those of the UL525 acceptance.
In addition, studies have shown that lightning discharge at the tip of the flare stack can generate a boosted flame front, or detonation within a few feet of the tip of the stack.
Currently, use of end-of-line flame arresters for in-line applications is still widespread. End-of-line flame arresters installed in these applications are being used well outside of their capability, resulting in arrester failure and flashback to the storage tank.
Recently, standards have been developed by the U.S. Coast Guard (USCG) and the Canadian Standards Association (CSA) which establish a detailed testing procedure to prove the capability of an in-line detonation flame arrester. While many companies are manufacturing "in-line arresters," very few have successfully met these standards.
Fig. 5 shows in-line detonation arresters located at the base of a Marathon Oil Co. flare near Tyler, Tex.
The standards require testing throughout the entire flame-propagation spectrum as well as an endurance burn test that establishes an arrester's ability to withstand a stabilized flame.
Although the USCG and CSA standards represent a major step forward in the area of flame arrestment, recent developments have shown shortcomings to these standards. Work continues to identify changes for updated standards.
For battery-site VCS applications, an arrester that has at a minimum been accepted by the USCG should be used because the installation conditions fall well outside those for which an end-of-line device has been designed.
REFERENCES
- "A Summary of Investigations from Ten In-Line Flame Arrester Failures," presented to the API and the Canadian Gas Processors Association, 1989.
- Underwriters Laboratories, "Standard for Flame Arresters for Use on Vents of Storage Tanks for Petroleum Oil and Gasoline," 1984.
Copyright 1993 Oil & Gas Journal. All Rights Reserved.
