VESSEL RUPTURE IN THE HOT-OIL HEAT MEDIUM SYSTEM OF A CHEMICAL PLANT

junvy

Introduction

     An accident has been reported involving the inadvertent introduction of water into a hot-oil heat medium system during the commissioning of a reboiler. The resulting vaporization of the water subjected the hot- oil surge vessel to such a high pressure that it ruptured violently. The hot oil, which was thrown out as a mist-like cloud, ignited almost immediately. Three employees in the immediate area were severely burned, two of whom died as a result of their burns. Severe equipment damage was limited to the area immediately surrounding the surge vessel.

     Description of the incident

     1. The hot-oil heat medium system had a capacity of 30,000 tcal/h and supplied hot oil at a  temperature of 315*C to three reboilers and hot oil at a temperature of 290*C to four reboilers. The hot-oil surge vessel in the system had a capacity of 50 m3. The vessel pressure was controlled via
self-contained regulators using natural gas. The safety relief valve was set to open at 3.5 kg/cm2 g, discharging into a typical flare system.

     2. Part of the plant was shut down for a maintenance turnaround. As one reboiler remained in operation for reboiling a distillation column, the furnace and circulation system continued operating throughout the shutdown and the other reboilers were isolated from the hot-oil circulation.

     3. One of these reboilers was, after inspection, pressure tested with water to ensure that it was free of leaks. This reboiler was of the vertical thermosyphon type with a down-flow heat medium on the shell side.

     4. Subsequent to the test, water was drained from a drain valve on the outlet piping on the shell 21 cm above the tube plate. The available drain plug on the tube sheet had not been removed.

     5. During the commissioning the reboiler was taken back into operation with some water still on the heat-medium circulation side.

     6. Water intrusion into the heat-medium circulation piping followed and subsequent vaporization of the water by the hot oil subjected the surge vessel to an extremely high pressure, causing it to rupture violently.

     7. The hot oil escaped as a fine mist-like cloud of hydrocarbon droplets which ignited immediately and a fireball estimated at 120 metres in diameter resulted.

     8. The explosive rupture of the vessel also broke a number of process lines in the immediate area, causing the contents of lines and equipment to spill and contribute to the fire.

     Conclusions

     1. Subsequent to the fire the presence of water in the bottom of the reboiler shell side was confirmed.

     2. Calculations have indicated that the surge vessel was about 90% full of oil at the time of  commissioning the reboiler.

     3. On the basis of these data the following explanation for the incident is consistent with the metallurgical examination of the vessel and the pressure generation necessary to create the vessel failure and the vapour cloud formation.

     During the recommissioning of the reboiler flashing water in the heat-medium return system first forced liquid into the surge vessel and the gas cap was expelled through the pressure relief valve to the flare. The liquid-filled vessel was then subjected to a pressure as high as 130 atm (theoretical vessel rupture pressure 30 atm.). This phenomenon is commonly known as "liquid hammer".

     4. Fracture and tearing of the vessel progressed at such a speed that the vessel appeared to be exploding. The top head parted from the shell almost in one piece and was propelled straight upwards accompanied by the formation of a hydrocarbon vapour cloud. The shell "unwrapped" and
fractured into approximately twenty pieces, the majority of which were thrown considerable distances (up to 400 metres).

     Recommendations

     1. To prevent incidents of this nature, it has to be ensured that water intrusion into hot-oil heat medium systems does not occur, particularly when the furnace and the heat medium circulation system remain
in operation throughout the shutdown of a heat exchanger and its return to operation. This implies that adequate facilities for the complete drainage of water from equipment should be available. This entails not only the correct positioning of the drain but also the presence of a vent valve to
guarantee drainage at atmospheric pressure.

     2. To eliminate the hazards involved in commissioning a heat exchanger under the circumstances described above, the use of oil instead of water for pressure-testing should be considered, if this is convenient and/or practicable.

     3. In operating instructions and training programmes adequate attention should be paid to the hazards involved in operating hot-oil heat medium systems, emphasizing the precautions to be taken to prevent water intrusion.

     4. Surge vessels in hot-oil heat medium service should havesufficient ullage for any hot oil which might be forced out of the return system by the flashing of small quantities of water out of recommissioned
equipment.
     
     96.III: Tube bursts in Hot Oil systems

     The text of this Safety Newsletter is reprinted from an SIOP Luboil Manufacturing Newsletter, because of the importance of the subject, and the relevance to locations which operate Hot Oil or similar systems not associated with Luboil manufacture.

     Two recent incidents in Group-operated luboil complexes have highlighted that there is a need to improve awareness of the potential problems of hot oil systems and how to deal with the consequences of such problems.

     The first location concerned uses a waxy distillate stream, which contains about 2% sulphur, as their hot oil medium. The hot oil furnace has a convection bank. Three years ago a change was made from dual fuel firing to 100% gas firing. At the time that this change was being considered it was not fully appreciated that the flue gas radiant cell outlet temperature is higher for 100% gas firing than for other modes of firing. The consequence is that the top section of the convection bank (including the so-called
"shock tubes") is exposed to higher temperatures. This will have resulted in increased pipe wall temperatures (especially at the points where the tube spacers are connected to the tubes) with an increase in film temperatures within the tubes. The convection bank tubes were constructed of normal
carbon steel which is vulnerable to sulphur corrosion. The severity of sulphur corrosion of carbon steel depends upon the film temperature, but rises rapidly over the range 300-370°C.

junvy

The location suffered a tube perforation in the convection bank due to sulphur corrosion. The combination of the use of carbon steel with the use of a sulphur containing hot oil medium and the fact that the hot oil system was being operated at high furnace outlet temperatures was responsible for the corrosion. The increase in the temperature of the flue gas entering the convection bank will have increased the severity of the corrosion, leading to the perforation occurring earlier than would
Otherwise have been the case.

     The second location suffered tube failures in the first layer of finned convection bank tubes, where apparently the heat flux (based on bare tube area) was higher than on the bare shock tubes. Higher heat flux results in higher film temperatures inside the tube.

     Both these incidents followed an incident two years earlier which resulted in a change in the furnace DEP (DEP.31.24.00.30-Gen October 1995) to a recommended maximum film temperature in hot oil systems furnace/convection bank tubes (previously a maximum bulk outlet temperature had been specified). In both the latest and earlier incidents, fires occurred in the convection bank during the shutdown process.

     The learning points are as follows:-

     When considering a change to 100% gas firing remember to take into consideration the resultant increased temperature of the flue gas entering the convection bank. This is especially important if the convection bank is fabricated from carbon steel and/or the hot oil fluid contains sulphur. For carbon steel tubes, the film temperature in the convection bank should be kept below 330°C. Recognising that most locations do not measure film temperature (or even shock tube skin temperature) it is recommended that
where there are carbon steel tubes in the upper section of the convection bank, the outlet temperature of every furnace coil should be kept below 320 deg.C. Monitoring of the hot oil fluid flash point is a means to identify degradation of hot oil fluid - this can be taken as an indicator of excessive temperatures within the furnace. It should also be noted that the presence of light material can mean that vaporisation takes place within the furnace tubes which can lead to tube rupture due to localised overheating.

Consideration should be given to the use of 5Cr½Mo tubes in place of carbon steel in the upper sections of the convection bank, if the hot oil fluid contains sulphur, when replacing tubes or when considering a change in furnace firing mode. All incidents occurred on relatively old furnaces. After a long period of operation some more detailed inspection may be required to ensure that locations where slow corrosion rates are occurring are checked. In addition to the shock tubes, inspection should then focus on places with surface extensions, such as spacer strips and the first row of finned tubes.

     Provided that the perforation of the convection bank is not major, the first sign of a problem is usually a change in the colour of the stack. This becomes yellow/brown due to the presence of uncombusted material. In the event of several furnaces sharing a stack it is not always easy to identify the culprit. One technique which may be useful is to see if the convection bank outlet temperature is significantly different from normal - it will be higher in the event of combustion in the convection bank.

     Perforation of a radiant cell or convection bank tube will necessitate shutting down of the hot oil system in order to effect a repair. The recommendation from the location that suffered the latest incident is to shutdown immediately after a leak is identified.

     The shutdown procedure is not without risks - one location burnt down their furnace as a result of a flawed shutdown procedure which resulted in a severe fire within the convection bank. It is not possible to give a single standard procedure since four cases have to be considered; small radiant cell leak, large radiant cell leak, small convection bank leak and large convection bank leak. Furthermore the presence of other furnaces using the same stack can affect the choice of procedures. Thus each refinery should, if they have not already, develop procedures (and practice them!) for dealing with leaks in the hot oil furnace.

     The following points will help in the development of such procedure:

     With the exception of very minor leakages, care should be taken to keep oxygen levels as low as possible to minimise the risk of combustion or explosion in the convection bank or stack system. In the case of minor leakages the air flows should be increased to maximum once the furnace is stopped. In this way the air/hydrocarbon mixture in the convection bank is kept below the lower explosion limit. In case of doubt about the size of leak, ingress of air should be minimised (by stopping blowers and introducing smothering steam) in order to stay above the upper explosion limit during the cool down period. Where several furnaces share a common stack, all the furnaces connected to the stack should be operated at
Minimum oxygen levels. If possible hydrocarbons should be evacuated from the affected coil by injecting steam, whilst the fires are still lit in the furnace. It may be necessary to use the maintenance override to prevent the low coil flow trip from shutting off the fuel supply. Introducing steam into the coil shall be done gently, to prevent pressure surges which could lead to tube ruptures. Never steam out a coil in a hot furnace unless either air flow is stopped or burners are lit - this recommendation is made to avoid
excessive oxygen being present when there is a risk of leaking material igniting.
     ------------------------------------
     88.I: FURNACE TUBE BURST

     Summary

     During a statutory shutdown of a chemical plant, a heating system operating on santotherm heating fluid, and consisting of a circulation grid over various units and a heating furnace, was partly drained off. Upon restart, cracks developed in three of the twelve vertical heating coils in the radiant section. This resulted in santotherm leaking and burning in this section. While measures were being taken to remedy the situation, a complete tube burst of a fourth coil occurred. The suddenly increased outflow of
santotherm was partly transferred to the 100 m high furnace stack. For about 10 minutes an approximately 30 m high flame emerged from the top of the stack, accompanied by a large cloud of soot. This soot was carried away and deposited over populated areas several kilometres away from the site.
Public and press reacted with irritation to the event. Repair work on the furnace postponed the start-up by approximately three weeks.

     The furnace

     A circulation pump feeds the heating liquid to the furnace. In the radiant section the santotherm flows through 12 vertical radiant coils, of 12.30 m length. The coils have 6 vertical hairpins. Near the furnace roof the outlets are combined to 1 header. Each coil is provided with a temperature measuring point on the outlet side which is scanned every three minutes. Measured values are stored. Burners are mounted in the furnace floor.

     The radiant coils form a completely open parallel system. There are no individual block valves on the coils, nor flow indicators. During normal operation 1250 tons/hr of santotherm is circulated and heated up to approximately 340*c. Inlet pressure is approximately 12 barg and the pressure drop under full flow conditions is approximately 2 bar; design pressure drop over the radiant coils only is 0.6 bar.

     Some characteristics of santotherm are : flash point 178*C, auto-ignition at 390*C, boiling point over 450* C. It starts cracking with carbon formation above 400*C. Design skin temperature of the radiant coils is below400*C. Carbonization with creep and cracking starts at 500*C.

junvy

The cause
     Vertically-mounted radiant coils in a furnace of this type cannot be drained completely by opening drain and vent valves. This means that the partly filled legs of each coil form a series of liquid seals. Upon
refilling, the resistance against allowing full flow will depend on the combined height of the sealing liquid columns in each individual coil. Once
a situation has arisen that a number of coils has obtained full liquid flow (pressure drop 0.6 bar), the resistance of the liquid seals in the remaining ones will prevent flow.

     Analysis of coil outlet temperatures showed that, after start-up, 3 out of 12 lagged behind. Apparently because there was no flow. Normally, all these temperatures lie within a range of 2*C. Following an increasing furnace temperature the same three showed wide variations, with peak values far in excess of the ones showing normal behaviour, and in the range where carbon deposit would occur. This could be explained by santotherm vapour lifting slugs of overheated liquid across the measuring point. Finally, the
situation stabilized with the three temperatures mentioned showing values close to, but still lagging somewhat behind the other ones. Full flow had been re-established. However, coke deposits inside the corresponding coils caused the temperature of the piping material to rise to such an extent
that creep and carbonization could occur.

   The temperature behaviour of the tube that finally burst had been normal. It was concluded that three coils failed mechanically, developing cracks, by overheating. This, in turn, was caused by the absence of
santotherm flow, followed by carbon deposit. The actual coil that burst wide open (one hour after the furnace had been shutdown and was being emptied) was locally overheated by the still leaking and burning santotherm from the cracks.

     Measures taken to prevent recurrence

     Before refilling the partly drained coils, the whole coil system is
to be evacuated. Then refilling is to take place at a restricted flow rate.
This method was tried out and proved to be successful.

     The TI system in the coil outlets was fitted with an alarm. If the difference between any of the outlet coil temperatures and the common outlet header temperature is more than 10*C, this alarm is set off.

     The procedure to bring a tube burst situation under control was adapted: it now aims at the fastest possible inertization of the radiant cells, minimization of the leaking amount, and fast cooling down of the
system. In this respect, also flushing steam connections to the cells were increased and enlarged.

     Note: The back-pressure phenomenon described is not necessarily restricted to loop systems in furnaces. The same type of problem could happen with any sequential loop systems, e.g. in vent systems.

     Recommendations

     Check procedures for emptying and refilling.

     Ensure process information provides adequate and timely warning of overheating (no flow).

     Check inertization system and procedures for emergencies.

xiaoke127

hhstb

thanks