What Went Wrong Part 9 pptx

It is therefore better to protect equipment that cannot withstand vacuum by means of a pressure control valve that admits nitrogen or, if nitrogen is not available, another gas such as f

1 Identification numbers stamped on springs, thus weakening them

2 The sides of springs ground down so that they fit

3 Corroded springs

4 A small spring put inside a corroded spring to maintain its strength Sometimes the second spring was wound the same way as the first spring so that the two interlocked (Figure 10-6)

(Figure 10-5)

5 Use of washers to maintain spring strength

6 Welding of springs to end caps (Figure 10-7)

7 Deliberate bending of the spindle to gag the valve (Figure 10-8)

8 Too many coils allowing little, if any, lift at set pressure (Figure 10-9)

Figure 10-5 Identification marks on body coils could lead to spring failure

Figure 10-6 Use of additional inner springs of unknown quality in an atten

to obtain set pressure

Figure 10-7 End caps welded to spring Failure occurred at weld

Figure 10-8 Deliberate bending of the spindle to gag the valve

Other Equipment 215

Figure 10-9 Example of too many coils

Do not assume that such things could not happen in your company (unless you have spent some time in the relief-valve workshop) All relief valves should be tested and inspected regularly Reference 3 describes

model equipment and procedures When a large petroleum company introduced a test program, it was shocked by the results: out of 187

valves sent for testing, 23 could not be tested because they were leaking

or because the springs were broken, and 74 failed to open within 10% of the set pressure-that is, more than half of them could not operate as required [4]

Testing, of course, must be thorough The following incident is described in the form of a conversation between an inspector investigat- ing a boiler explosion and the maintenance foreman [ 121

Inspector: “When was the relief valve last checked?’

Foreman: “After the last overhaul.”

Inspector: “How was it checked?’

Foreman: “I set it myself, using the boiler’s own pressure gauge.”

Inspector: “Why didn’t you use a master gauge?’

Foreman: “I didn’t need to The gauge had been checked and found

Inspector: “Who checked it?,

Foreman: “Mr X, one of my fitters He has often done so in the past.”

accurate only two weeks before.”

The inspector then spoke to Mr X

Inspector: “I understand you checked the pressure gauge two weeks

Mr X: “Yes, the foreman told me to do so.”

before the explosion.”

Inspector-: “When was your master gauge last calibrated?”

M K X : “I didn’t use one.”

Inspector: “You didn’t use a master gauge? Then how did you check it?’

MI: X : ”I checked it against the relief valve I knew it was correct

because the foreman told me he had adjusted it himself.”

This incident occurred in the 19th century when boiler explosions were much more frequent than they are today But are you sure some- thing similar could not occur today? Read Sections 10.7.2 (b) and (c) before you decide

Similar incidents have occurred in technical reports A writes some- thing in a book or paper B copies it without acknowledgment A then repeats it in another report, citing B as the source and thus giving it an authenticity it lacked in its first publication

10.4.6 Disposal of Relief Discharges

Material discharged from relief valves and rupture discs should not be discharged to atmosphere unless:

9 It will have no harmful effects, for example, steam, compressed air,

or nitrogen

It is a gas at a pressure high enough to disperse by jet mixing It is necessary to use a pilot-operated relief valve that is either open or shut and is not a type that will hover Although it is safe to discharge gases such as ethylene and propylene in this way, there may be objections on environmental grounds

The amount released is negligible, for example, the relief valves that protect a pipeline that has been isolated

A system of trips or interlocks makes the probability that the relief valve will lift very low, say, less than once in 1,000 years for flam- mable liquids and less than once in 100 years for flammable gases The relief valve will lift only after prolonged exposure of the equip- ment to fire and will discharge within the fire area so that the dis- charge will ignite

Here are some examples of the results of letting relief valves discharge

to atmosphere:

Other Equipment 21 7

0 A 6-in (150-mm) relief valve on a petrochemical plant discharged ben- zene vapor to atmosphere It was ignited by a furnace and exploded, rupturing piping, which released more than 100 tons of various flam- mable liquids One man was killed, and damage was extensive [5]

1 At Seveso in Italy in 1976, a runaway reaction led to the discharge of the reactor contents, including dioxin, a toxic chemical, through a rupture disc direct to the atmosphere Although no one was killed, many people developed chloracne, an unpleasant skin disease, and an area of about 17 k m 2 was made uninhabitable A catchpot after the relief device would have prevented the reactor contents from reach- ing the atmosphere No catchpot was installed as the designers did not foresee that a runaway might occur, although similar runaways had occurred on other plants (see Section 21.2.5) [6]

e Naphtha vapor from a relief valve on a town gas plant in the UK was ignited by a flare stack The flame impinged on the naphtha line,

which burst, starting a secondary fire [7]

A relief valve sprayed liquid into the face of a passing operator with such force that it ?docked his goggles off

0 A reaction involving concentrated sulfuric acid was carried out at atmospheric pressure in a vessel with an opening to the atmosphere

at the top When a runaway occurred, acid was ejected over the sur- rounding area [ 131

The rupture discs on some water compressors were allowed to dis- charge inside a building as the water was clean However, by the time it had drained down through several floors to the basement of the building, it had dissolved some solid material that had been spilt

on one of the intervening floors and became hazardous Discs had failed on several occasions, for unknown reasons Possible causes were vibration, hammer pressure, and low-cycle fatigue

If2 despite my advice, you let relief devices discharge to atmosphere, make sure that if the discharge ignites, the flame will not impinge on other equipment and that no one will be in the line of fire

10.4.7 Vacuum Relied Valves

Some large equipment, though designed to withstand pressure, cannot withstand vacuum and has to be fitted with vacuum relief valves These

usually admit air from the atmosphere If the equipment contains a flam- mable gas or vapor, then an explosion could occur with results more seri- ous than collapse of the vessel Experience shows that a source of igni- tion may be present even though we have tried to remove all possible sources (see Section 5.4) It is therefore better to protect equipment that cannot withstand vacuum by means of a pressure control valve that admits nitrogen or, if nitrogen is not available, another gas such as fuel gas Very large amounts may be necessary For example, if the heat input

to a large refinery distillation column stopped but condensation contin- ued, 8,000 m3/hr of gas, the entire consumption of the refinery, would be required Instead, a much smaller amount was supplied to the inlet of the condenser, thus blanketing it and stopping heat transfer [ 141 The sim- plest solution, of course, is to design equipment to withstand vacuum Protection of storage tanks against vacuum is discussed in Sections 5.3 and 5.4

10.5 HEAT EXCHANGERS

10.5.1 Leaks into Steam and Water Lines

Hydrocarbons can leak through heat exchangers into steam or conden- sate systems and appear in unexpected places Some hydrocarbon gas leaked into a steam line that supplied a heater in the basement of a con- trol building The gas came out of a steam trap and exploded, killing two men The operators in the control building had smelled gas but thought it had entered via the ventilation system, so they had switched off the fan The control gear was ordinary industrial equipment, not suitable for use

in a flammable atmosphere, and the sparking ignited the gas It was for- tunate that more people were not killed, as the building housed adminis- trative staff as well as operators

A leak in another heat exchanger allowed flammable gas to enter a cooling-water return line The gas was ignited by welding, which was being carried out on the cooling tower The atmosphere had been tested before work started, five hours earlier (see Section 1.3.2)

10.5.2 Leaks Due to Evaporative Cooling

If the pressure on a liquefied gas is reduced, some of the liquid evapo- rates, and the rest gets colder All refrigeration plants, domestic and

Other Eqoipmenf 21 9

industrial, make use of this principle This cooling can affect equipment

in two ways: it can make it so cold that the metal becomes brittle and cracks as discussed in Sections 8.3 and 10.5.2 (g) and it can cause water,

or even steam on the other side of a heat exchanger to freeze and rupture

a tube or tubes The leak that caused the explosion in a control building (see last section) started this way

Figure 10-10 illustrates another incident When a plant was shutting down, the flow o f cooling water to the tubes o f a heat exchanger was iso- lated The propylene on the shell side got colder as its pressure fell The water in the tubes froze, breaking seven bolts The operators saw ice forming on the outside o f the cooler but did not realize that this was haz- ardous and took no action When the plant stai-ted up again, propylene entered the cooling-water system, and the pressure blew out a section of the 16-in (400-mm) line The gas was ignited by a furnace 40 m away and the fire caused serious damage

Reduction of pressure here will cause liquefied gas to evaporate and cool and may freeze the water in the tubes

L

Waterto tubes

Figure 10-10 Evaporative cooling

The cooling water should have been kept flowing while the plant was depressured This would have prevented the water from freezing, provid-

ed that depressuring took more than ten minutes

Water hammer (hydraulic shock) in pipelines is discussed in Section 9.1.5 It can also damage heat exchangers, and Figure 10- 11 illustrates such an incident

A Impingement plate

f Figure 10-11 Condensate in the steam-the result of too few steam traps- knocked off the impingement plate and damaged the calandria tubes

The steam supplied to the shell of a distillation column reboiler was very wet, as there was only one steam trap on the supply line although at least three were needed In addition condensate in the reboiler drained away only slowly because the level in the drum into which it drained was only 1.4 m below the level in the reboiler

An impingement plate was fitted to the reboiler to protect the tubes, but it fell off, probably as a result of repeated blows by slugs of conden- sate The condensate then impinged on the tubes and squashed or broke

30 of them

The impingement plate had fallen off several times before and was merely put back with stronger attachments When something comes apart, we should ask why, not just make it stronger (see Section 1.5.5) Buildup of condensate in a heat exchanger can cause operating prob- lems as well as water hammer If the steam supply is controlled by a motor valve and the valve is not fully open, the steam pressure may be too low to expel the condensate, and its level will rise This will reduce heat transfer, and ultimately the steam supply valve will open fully and expel the condensate The cycle will then start again This temperature cycling is bad for the heat exchanger and the plant and may be accompa-

Other Equipment 226

nied by water hammer and corrosion Proprietary devices are available for overcoming the problem [SI

10.5.4 An Accident During Maintenance

The tube bundle was being withdrawn from a horizontal shell and tube heat exchanger It was pulled out a few inches and then became stuck The mechanics decided that the cause was sludge, and to soften it they reconnected the steam supply to the shell The tube bundle was blown out with some force, causing serious injuries [9]

10.6 COOLING TOWERS

These are involved in a surprisingly large number of incidents; one is described in Section 10.5.1 Wooden packing after it dries out, is veiy easily ignited and many cooling towers have caught fire while they were shut down For example, the support of a force draft fan had to be repaired by welding An iron sheet was put underneath to catch the sparks, but it was not big enough, and some of the sparks fell into the tower and set the packing on fire

Corrosion of metal reinforcement bars has caused concrete to fall off the corners of cooling towers

A large natural-draft cooling tower collapsed in a 70-mph (1 lo-kmihr) wind, probably due to imperfections in the shape of the tower which led

to stresses greater than those it was designed to take and caused bending collapse [ lo]

An explosion in a pyrolysis gas plant in Rumania demolished a cooling tower It fell on the administration block, killing 162 people Many people who would not build offices close to an operating plant would consider it s.afe to build them close to a cooling tower It is doubtful if this is wise

10.7 FURNACES

10.7.1 Explosions While Lighting a Furnace

Many explosions have occurred while furnaces were being lit The two incidents described below occurred some years ago on furnaces with simple manual ignition systems, but they illustrate the principles to be followed when lighting a furnace, whether this is carried out manually or automatically

(a) A foreman tested the atmosphere inside a furnace (Figure 10-12) with a combustible gas detector No gas was detected, so the slip- plate was removed, and two minutes later, a lighted poker was inserted An explosion occurred The foreman and another man were hit by flying bricks, and the brickwork was badly damaged The inlet valve was leaking, and during the two minutes that elapsed after the slip-plate was removed, enough fuel gas for an explosion leaked into the furnace (Suppose the leak was equiva- lent to a 1.6-mm [X/;h-in.] diameter hole, and the gauge pressure of the fuel gas was 0.34 bar [5 psi] The calculation shows that 80 L

[3 ft’] of gas entered the furnace in two minutes If this burned in

0.01 second, the power output of the explosion was 100 MW.) The correct way to light a furnace (hot or cold) that bums gas or burns light oil is to start with a positive isolation, such as a slip- plate, in the fuel line Other positive isolations are disconnected hoses, lutes filled with water (if the fuel is gas at low pressure), and double block and bleed valves; closed valves without a bleed are not sufficient Then:

1 Test the atmosphere inside the furnace

2 If no gas is detected, light and then insert the poker (or switch on the electric igniter)

3 Remove the slip-plate (or connect the hose, drain the lute, or change over the double block and bleed valves) If the isolation valve is leaking, the leaking fuel will be ignited by the poker or igniter before it forms an explosive mixture (The solenoid valve shown in Figure 10-12 should open automatically when the poker

is inserted or the igniter is switched on If it does not, it should be held open until the main burner is lit.)

4 Open the fuel-gas isolation valve

The furnace had been lit in an incorrect way for many years before the isolation valve started to leak and an explosion occurred Never say, “It must be safe because we have been doing it this way for years and have never had an accident.”

On furnaces with more than one burner, it may be possible to light a burner from another one if the two are close to each other If they are not, the full procedure just described should be followed Explosions have occurred on multiburner furnaces because opera- tors assumed that one burner could always be lit from the next one

Figure 10-12 Lighting a furnace heated by fuel gas

(b) A reduction in fuel oil pressure caused the burner in an oil-fired furnace to go out and the flame failure device closed the solenoid valve in the fuel oil line (Figure 10-13) The operator closed the two hand-isolation valves and opened the bleed between them (The group of three valves is equivalent to the slip-plate shown in Figure 10-12) When the fuel oil pressure was restored, the fore- man tested the atmosphere in the furnace with a combustible gas detector No gas was detected, so he inserted a lighted poker The fuel oil supply was still positively isolated, but nevertheless an explosion occurred and the foreman was injured, fortunately not seriously

Fuel oil

-

Figure 10-13 Lighting a furnace heated by heavy fuel oil

When the burner went out, the solenoid valve took a few sec- onds to close, and during this time some oil entered the furnace In addition, the line between the last valve and the furnace may have drained into the furnace The flash point of the fuel oil was 65"C, too high for the oil to be detected by the combustible gas detector Even though the oil was vaporized by the hot furnace, it would have condensed in the sample tube of the gas detector or on the sin- tered metal that surrounds the detector head

Before relighting a hot furnace that burns fuel oil with a flash point above ambient temperature sweep it out for a period of time, long enough to make sure that any unburnt oil has evaporated If this causes too much delay, then pilot burners supplied by an alter- native supply should be kept alight at all times

To keep the sweeping-out or purge time as short as possible, the solenoid valve should be close to the burner, and it should close quickly In addition, the line between the solenoid valve and the

burner should not drain into the furnace As in the previous inci-

dent, the furnace had been lit incorrectly for many years before an explosion occurred

To calculate the purge time:

burner

1 Calculate the amount of oil between the solenoid valve and the

Other Equipment 225

2 Assume it a14 drains into the furnace and evaporates Calculate the volume of the flammable mixture, assuming it is at the lower flammable limit, probably about 0.5% v/v (If it forms a richer mixture, the volume will be less.)

3 Multiply by four to give a safety margin, and sweep out the fur- nace with this volume of air [ l l]

10.7.2 Furnace Tube Ruptures

(a) A heat transfer oil was heated in a furnace A tube in the convection section ruptured along 8 in (200 mm) of its length, and Ihe ensuing

fire damaged three other furnaces No one could stop the flow of heat transfer oil into the fire, as the valves in the line and the pump switch were too near the furnace The fire continued until all the oil was burned

The tube failure was due to prolonged, though not necessarily continuous, overheating of the furnace tubes at times when maxi- mum output was wanted This led to a creep failure There were not enough instruments on the furnace to measure the temperature of the tubes, always a difficult problem, and the operators did not understand the way furnace tubes behave They are usually designed to last for ten years, but if they get too hot, they will not last so long For instance, if the tubes are designed to operate at 5QO"C, then:

If they are kept at 506°C they will last 6 years

Hf they are kept at 550°C they will last 3 months

If they are kept at 635°C they will last 20 hours

If we let the tubes get too hot, however carefully we treat them afterward, they will never be the same again If we heat them to

550°C say, for six weeks, we will have used up half their creep life,

and they will fail after about five years at design temperature If we find that our pumps, heat exchangers, and distillation columns will handle a greater throughput than design, we can use it If we try the same with our furnaces, we may be in trouble in the future

The following recommendations were made after the fire They apply to all furnaces

Provide good viewing ports (Although this failure occurred in the convection section of the furnace where the tubes are heated by

hot flue gas, most failures occur in the radiant section as the result

of flames impinging on the tubes.)

Provide tube temperature measurements (but we can never be sure that we are measuring the temperature at the hottest point)

Train operators in the principles of furnace operation

Look for signs of overheating during overhauls

Provide remotely operated emergency isolation valves (see Sec- Provide remote Stop buttons for circulation pumps well away

9 Lay out new plants so that circulation pumps are well away from

Do not take a furnace above design output without advice from a Examine contractors' proposals critically

(b) A heat transfer oil was heated in a furnace used only during start-

ups to bring reactors up to operating temperature Startup is always a busy time, and the operator lit the furnace and forgot to open the valves leading from the furnace to the reactors (an exam- ple of the sort of lapse of attention we all make from time to time, especially when we are under stress: see Chapter 3) Within 20-30 minutes, a furnace tube ruptured, and there was a large fire with flames 15 m tall

The furnace was fitted with interlocks that should have isolated the fuel supply if the tube wall temperature or the pressure of the heat transfer oil got too high Neither interlock worked and neither had been tested or maintained The set-point of the high tube wall interlock had been raised far above its original set-point, from 433°C to 870°C, a simple way of putting it out of action [15] Changing the set-point of an interlock is a modification and should

be allowed only when the equipment is capable of withstanding the new conditions (see Chapter 2)

A similar incident occurred on another furnace when the heat transfer oil froze inside the furnace during unusually cold weather Outside the furnace, the lines were steam-traced The operating team decided to thaw the frozen oil by lighting one of the burners

tion 7.2.1)

from the furnace

furnaces

materials engineer

Other Equipment 227

in the furnace at a low rate Later on someone increased the flow

of fuel About an hour after the furnace was lit a tube ruptured There were no instructions on the action to be taken nhen the oil froze Lighting a burner had been used before, successfully on that occasion [16]

Cc) A feed water pump supplied two boilers The backup pump also supplied another unit, which was under repair, so the operator on this unit blocked it in He did not tell the boiler operator what he had done (as in the incident described in Section 17.4)

The on-line feed water pump tripped but the operator ignored the alarm signal, presumably because he thought the backup pump would start up automatically

The smaller of the two boilers became short of water first, and the low water level trip shut it down The operator was so busy try- ing to get it back on line that he ignored the low water level and other alarms that were sounding on the other boiler Unfortunately the trips on this boiler did not work, as it had been rewired (incor- rectly) since it was last checked Fifteen to 30 minutes later, some- one saw flames corning out of the boiler stack The boiler was then shut down manually By this time most of the tubes had melted After the furnace had been allowed to cool the operating team not realizing the extent of the damage, restarted the flow of feed mater They stopped it when they saw water running out of the fire- box It is fortunate they did not start the water flow earlier, or it would have caused explosive vaporization of the water [17] As stated in Section 9.3.2 (e), equipment that has been taken outside its design or test range should not be used again until it has been examined

(d) Another tube failure had an unusual cause A pipe, sent to an out- side workshop for bending, was returned plugged with sand and was welded into the exit line from a furnace Not surprisingly the furnace tubes overheated and failed during startup

The pipe was returned to the plant with a warning that it might contain some sand The plant staff took this to mean that a few grains might be stuck to the walls not that the pipe might be full of sand Section 14.2.3 describes another failure