Preparation for Maintenance 33 closed, and a drain valve in between opened. A flow through the drain valve showed that one of the isolating valves was leaking, so the drain valve was closed and a message left for the employees workmg the next shift, telling them to open the drain valve before work started. Nothing was written on the permit-to-work. The mes- sage was not passed on: the drain valve was not opened. and the fitter broke the joint the wrong way, removing all the bolts. The joint blew apart, and the fitter received head injuries from which he will never fully recover [37]. Cc) It is not only flamiable oils that cause accidents. In another inci- dent two workers were badly scalded when removing the cover from a large valve on a hot water line, although the gauge pressure was only 9 in. of water (0.33 psi or 0.023 bar). They removed all the nuts, attached the cover to a chain block, and tried to lift it. To release the cover they tried to rock it. The cover suddenly released itself. and hot water flowed out onto the workers' legs. se of Excessive Force A joint on an 8-in. line containing a hot solvent had to be remade. The two sides were %-in. out of line. There was a crane in the plant at the time, SO it was decided to use it to lift one of the lines slightly. The lifting strap pulled against a %-in. branch and broke it off (Figure 1-11). It was not a good idea to use a crane for a job like this on a line full of process material. Fortunately the leaking vapor did not ignite, although 8" Pipe Full of Hot Solvent Joint to be Remade 3/4" Branch Figure 1-11. A branch broke when a crane was used to move a live line. 34 What Went Wrong? nearby water was being pumped out of an excavation. At one time a diesel pump would have been used, but the use of diesel pumps had been banned only a few months before the incident. Section 2.11.1 describes an explosion caused by the failure of nuts that had been tightened with excessive force. 1.5.3 Ignorance of Material Strength (a) When a plant came back on line after a long shutdown, some of the flanges had been secured with stud bolts and nuts instead of ordi- nary bolts and nuts. And some of the stud bolts were located so that more protruded on one side than on the other. On some flanges, one of the nuts was secured by only two or three threads (Figure 1-12). Nobody knows why this had been done. Probably one nut was tighter than the other, and in attempting to tighten this nut, the whole stud was screwed through the second nut. Whatever the rea- son, it produced a dangerous situation because the pressure on dif- ferent parts of the flange was not the same. In addition, stud bolts should not be indiscriminately mixed with ordinary bolts or used in their place. They are often made of differ- ent grades of steel and produce a different tension. In the plant concerned, for the eight-bolt joints the bolts were changed one bolt at a time. Four-bolt joints were secured with clamps until the next shutdown. nn Figure 1-12. Nuts fitted incorrectly to studs. Preparation for Maintenance 35 (b) There was a leak on a large fuel-gas system operating at gasholder pressure. To avoid a shutdown, a wooden box was built around the leak and filled with concrete. It was intended as a temporary job but was so successfbl that it lasted for many years. On other occasions, leaks have been successfully boxed in or encased in concrete. But the operation can only be done at low pressures, and expert advice is needed, as shown by the following incident. There was a bad steam leak from the bonnet gasket of a 3-in. steam valve at a gauge pressure of 300 psi (20 bar). An attempt to clamp the bonnet was unsuccessful, so the shift crew decided to encase the valve in a box. Crew members made one 36 in. long, 24 in. wide, and 14 in. deep out of %-in. steel plate. Plate of this thick- ness is strong, but the shape of the box was unsuitable for pressure and could hardly have held a gauge pressure of more than 50 psi (3 bar), even if the welds had been full penetration, which they were not (Figure 1-13). The box was fitted with a vent and valve. When the valve was closed, the box started to swell, and the valve was quickly opened. A piece of 2-in. by 2-in. angle iron was then welded around the box to strengthen it. The vent valve was closed. A few minutes later the box exploded. Fortunately the mechanic-if he deserves the title-had moved away. This did not happen in a back-street firm but in a major interna- tional company. These incidents show the need for continual vigilance. We can- not assume that because we employ qualified craftsmen and gradu- ate engineers they will never carry out repairs in a foolish or unsafe manner. 1.5.4 Failure to Understand How Things Work or How They Are Constructed (a) Several spillages have occurred from power-operated valves while the actuators were being removed because the bolts holding the valve bonnets in position were removed in error. Figures 1-14 and 1-15 show how two such incidents occurred. The second system is particularly vulnerable because in trying to unscrew the nuts that 36 What Went Wrong? Figure 1-13. This steel box was quite incapable of containing a leak of steam at a gauge pressure of 300 psi (20 bar). Preparation for Maintenance 37 Valve These bolts should have been removed Motor mounting Plug i3 packing bracket retainer Figure 1-14. Wrong nuts undone to remove valve actuator. Actuator mountinq bracket Cover , Seal adjustment Figure 1-15. Wrong nuts undone to remove valve actuator. hold the actuator mounting bracket in place, the stud may unscrew out of the lower nuts. This incident could be classified as due to poor design [ 101. The first incident resulted in the release of 70-100 tons of vinyl chloride. There was little wind, and the cloud of vapor and mist drifted SIOWIY backward and forward. After an hour, when the cloud was about 240 m across and 1.5 m deep, it ignited. Some of 38 What Went Wrong? the vinyl chloride had entered buildings, and it exploded, destroy- ing the buildings. The rest burned outside and caused several vinyl chloride tanks to burst, adding further fuel to the fire. Remarkably, only one man was killed. The injured included spectators who arrived to watch the fire [30]. (b) A similar accident occurred on a common type of ball valve. Two workers were asked to fit a drain line below the valve. There was not much room. So they decided to remove what they thought was a distance piece or adaptor below the valve but which was in fact the lower part of the valve body (Figure 1-16). When they had removed three bolts and loosened the fourth, it got dark, and they left the job to the next day. The valve was the drain valve on a small tank containing lique- fied petroleum gas (LPG). The 5 tons of LPG that were in the tank escaped over two to three hours but fortunately did not catch fire. However, 2,000 people who lived near the plant were evacuated from their homes [ 1 11. Flow from Tank PTFE Seat Ring 1 Body 1 Ad apt o r Figure 1-16. Valve dismantled in error. Preparation for Maintenance 39 (c)In canned pumps the moving part of the electric motor-the rotor-is immersed in the process liquid; there is no gland, and gland leaks cannot occur. The fixed part of the electric motor-the stator-is not immersed in the process liquid and is separated from the rotor by a stainless steel can (Figure 1-17). If there is a hole in the can, process liquid can get into the stator compartment. A pressure relief plug is therefore fitted to the com- partment and should be used before the compartment is opened for work on the stator. Warning plates, reminding us to do this, are often fitted to the pumps. The stator compartment of a pump was opened up without the pressure relief plug being used. There was a hole in the can. This had caused a pressure buildup in the stator compartment. When the cover was unbolted, it was blown off and hit a scaffold pole 2 m above. On the way up, it hit a man on the knee, and the escaping process vapor caused eye irritation. Persons working on the pump did not know the purpose of the plug, and the warning notice was missing. Cover I A Pressure Relief L ' Plug Pump Casing Stainless Steel Can Figure 1-17. Canned pumps. 40 What Went Wrong? For a more detailed diagram and description of a canned pump, see Reference 12. (d) On several occasions. fitters have removed thermowells without realizing that this would result in a leak. They did not realize that the thermowell-the pocket into which a thermocouple or other temperature measuring device sits-is in direct contact with the process fluid. A serious fire that started this way is described in Reference 13. (e) A high-pressure reciprocating ammonia pump (known as an injec- tor) had run for 23 years without serious problems when the crank- shaft suddenly fractured, due to fatigue, and the plungers came out of the cylinders. Tko men were killed by the ammonia. No one real- ized that a failure of the motion work would produce a massive release of ammonia. If people had realized this. they would have installed remotely operated emergency isolation valves (see Section 7.2.1). These would have greatly reduced the size of the leak but would not have acted quickly enough to prevent the fatalities [31]. 1.5.5 Treating the Symptoms Instead of the Disease The following incidents and Section 10.5.3 show what can happen if we go on repairing faults but never ask why so many faults occur. (a) A cylinder lining on a high-pressure compressor was changed 27 times in nine years. On 11 occasions it was found to be cracked, and on the other 16 occasions it showed signs of wear. No one asked why it had to be changed so often. Everyone just went on changing it. Finally. a bit of the lining got caught between the pis- ton and the cylinder head and split the cylinder. (b) While a man was unbolting some %in. bolts, one of them sheared. The sudden jerk caused a back strain and absence from work. Dur- ing the investigation of the accident, seven bolts that had been sim- ilarly sheared on previous occasions were found nearby. It was clear that the bolts sheared frequently. If, instead of simply replac- ing them and carrying on, the workers had reported the failures, then a more suitable bolt material could have been found. Why did they not report the failures? If they had reported them would anything have been done? The accident would not have occurred if the foreman or the engineer, on their plant tours, had noticed the broken bolts and asked why there were so many. Preparation for Maintenance 41 (c) A line frequently choked. As a result of attempts to clear the chokes. the line was hammered almost flat in several places. It would have been better to have replaced the line with a larger one or with a line that had a greater fall, more gentle bends. or rodding points. 1.5.6 Flameproof Electrical Equipment On many occasions. detailed inspections of flameprooF electrical equipment have shown that many items were faulty. For example, at one plant a first look around indicated that nothing much was wrang. A more tlhorough inspection, paying particular attention to equipment not readily accessible and that could be examined only from a ladder, showed that out of 121 items examined, 33 needed repair. The faults included missing and loose screws, gaps too large, broken glasses, and incorrect glands. Not all the faults would have made the equipment a source of ignition. hut many would have done so. 'Why were there so many faults'? Before this inspection, there had been no regular inspections. Many electricians did not understand why flame- proof equipment was used and what would happen if it was badly main- tained. Spare screws and screwdrivers of the special types used were not in stock, so there was no way of replacing those lost. Regular inspections were set up. Electricians were trained in the rea- sons why flameproof equipment is used, and spares were stocked. In addition, it was found that in many cases flameproof equipment was not really necessary. Division (Zone) 2 equipment-cheaper to buy and easi- er to maintain-could be used instead. 1.5.7 Botching Section 1.5.3 (a) described a botched job. Here are two more. (a) A pressure vessel was fitted with a quick-opening lid, 10 in. diame- ter, secured by four eye-bolts (Figure 1-18). They had to be replaced, as the threads were corroded. Instead of replacing the whole eye-bolt, a well-meaning person decided to save time by simply cutting the eyes off the bolts and welding new studs onto them. As soon as the vessel was pressurized (with compressed air) the new studs, which had been made brittle by the welding, failed, and the lid flew off. Fortunately a short length of chain restrained it, and it did not fly very far [38]. (See Sections 13.5 and 17.1 for the hazards of quick-opening lids.) 42 What Went Wrong? Figure 1-18. Instead of replacing the eye-bolts, new studs were welded in place of the threaded portions. They were made brittle by the heat and failed in use. Fortunately the chain prevented the lid from going into orbit. (b)A screwdriver was left in the steering column of a truck after the truck was serviced. The truck and semitrailer crashed, and the ser- vicing company had to pay $250,000 in damages. To quote from the report, “Workplaces need to be as rigorous as the aviation and medical industries in ensuring that all tools are accounted for when servicing is completed” [39]. [...]... p 91 93 Petroleurn Revieu; Oct 1981, p 21 14 J H Christiansen and L E Jorgensen, Proceedirzgs o the Fourth f Iiitematioizul Syrnposium oil Loss Prevention arid Safeh Promotion in the Process Iridustries (Symposium held in Harrogate, UK, Sept 19 83) Institution of Chemical Engineers, Rugby UK, p L9 15 L G Britton and J A Smith, PlaidOperations Progress, Vol 7 , No 1 Jan 1988, p 53 46 What Went Wrong? ... No 6, U.S Dept of Energy, Washington, D.C., June 19 93, p 6 28 B A Prine, “Analysis of TitaniudCarbon Steel Heat Exchanger Fire,” Paper presented at AIChE Loss Prevention Symposium, Aug 1991 29 Occupational Health and Safety Obseniel; Vol 2, No 1 U.S Dept of Energy, Washington, D.C., Sept 19 93, p 3 30 Loss Prevention Bulletin, No 100, Aug 1991, p 35 31 P J Nightingale, “Major Incident Following the Failure... Symposium, Aug 1990 32 Operating E-xperience Weekly Summary, No 96-50, Office of Nuclear and Facility Safety, U.S Dept of Energy, Washington, D.C., 1996, p 9 33 Loss Prevention Bulletin, No 107, Oct 1992, p 17 Preparation for Maintenance 47 34 Operating Experierice Weekly Siimrnary, No 96-44, Office of Nuclear and Facility Safety, U.S Dept of Energy, Washington, D.C., 1996, p 2 35 Occupatiorzal Snfeg... Health Obseiver; Vol 3 Nos 7 and 8, U S Dept of Energy, Washington, D.C., July-Aug 1994 p 1 36 Underground Utilities Detectiorz arid Excavation, Safety Notice No DOEEH-0541, Office of Nuclear and Facility Safety, U.S Dept of Energy, Washington D.C., 1996 37 Efective Shift Haridoiler-A Literature Revien: Report No OTO 96 0 03, Health and Safety Executive, Sheffield, UK, 1996, p 7 38 R E Sanders and W... procedure before filling the 52 What Went Wrong? tank The air was not getting through, and the operator suspected a choke in the pipeline In fact, the vent on the tank was choked The gauge air pressure (75 psi or 5 bar) was sufficient to burst the tank (design gauge pressure 5 psi or 0 .3 bar) Originally the tank had a 6-in.-diameter vent But at some time this was blanked off, and a 3- in.-diameter dip branch... (c) A “carbon copy” plant was built with the floors 3 m (10 ft) apart instead of 2.4 m (8 ft) apart, as 3 m was the company standard The increased height was too much for convection flow, and efficiency was lost (dj When construction of a nuclear power station was well under way, an advisory committee suggested that a zirconium liner should be added to part of the cooling circuit The operating company... plant not an advisory committee however eminent, is responsible for the safety of the plant and should not follow advice that it believes to be wrong just to save costs or avoid arguments Another sanctioned modification is described in Section 13. 4.6 60 What Went Wrong? 2.6 PROCESS MODIFICATIONS So far we have discussed modifications to the plant equipment Accidents can also occur because changes to process... heated to 90°C and kept at this temperature for 30 minutes; it was then cooled After a year's operation someone decided to let the batch cool by heat loss to the surroundings with no one in attendance, as soon as the temperature reached 90°C An explosion occurred; the building u'as wrecked, and parts of the reactor were found 75 m away [7] 62 What Went Wrong? (e) The heating in a building had to be... attention to it, and they did not check that the modification had been made correctly [ 131 Modifications should always be marked on a line diagram before they are approved and the person who authorizes them should always inspect the finished modification to make sure that his or her intentions have been followed 54 What Went Wrong? (d) A group of three rooms in a control center were pressurized to prevent... the main instead of the top A slug of liquid filled up the catchpot and extinguished the burners I Compressed Air , I 1 I , Automatic Drain Figure 2-4 Original arrangement of air lines - 56 What Went Wrong? Some hot (37 0°C) pipework was supported by spring hangers to minimize stress as it was heated and cooled The atmosphere was corrosive, and the spring hangers became impaired They were removed, and . such incidents occurred. The second system is particularly vulnerable because in trying to unscrew the nuts that 36 What Went Wrong? Figure 1- 13. This steel box was quite incapable of containing. Hot Solvent Joint to be Remade 3/ 4" Branch Figure 1-11. A branch broke when a crane was used to move a live line. 34 What Went Wrong? nearby water was being pumped. chain restrained it, and it did not fly very far [38 ]. (See Sections 13. 5 and 17.1 for the hazards of quick-opening lids.) 42 What Went Wrong? Figure 1-18. Instead of replacing the eye-bolts,