Document aJNLEZ0L5BmnYBRkLpZGJV49b

R&S 038656 DOW CHEMICAL U.S.A OYSTER CREEK DIVISION POST OFFICE BOX 63 FREEPOHT. TEXAS 77SAI November 3, 1983 Ken West Paul Rozas Peter Burns Reg Daniels Joe Hegyesi Anthony Ho Sid Brunson Attached are some notes I made at the Vinyl Safety Association meeting in New Orleans in October. Please route the report to folks in your plants. Also attached is a survey on valve position switches, types, dependability, reason for failure, etc. Your Instrument and Electrical Supervisor might find this interesting. pm Robert R. Oubre AN OPERATING UNIT OF THE DOW CHEMICAL. COMPANY R&S 038657 NOTES FROM VINYL CHLORIDE SAFETY ASSOCIATION MEETING IN NEW ORLEANS ON OCTOBER 20 and 21, 1983 1 * Old Case Histories a. John Barr related the Thompson Chemical explosion of 1964 in which there were 7 fatalities and 22 injured. A PVC kettle was taken out of service to repair a sight glass that was leaking. The glass put in was not designed for the service. It began leaking after start-up and when the operator tried to tighten the flange, the glass chipped and began leaking worse. The reactor building soon filled with vapor and an explosion occurred. Piping on other reactors broke and fed 150,000 lbs. VCM to the fireball. Nearby school buildings suffered extensive damage in addition to total destruction of the reactor building and contents. Errors made and lessons learned: 1) The old type sight glass was replaced with new high pressure type. 2) No hydro test was made after repair. 3) Operator attempted to tighten sight glass with vessel under full pressure. 4) Area was congested overall with a school building within 2000 feet of reactor building. 5) There was inadequate crowd control -- one person was found standing on a vinyl bullet watching the fire and smoking. 6) Building construction was inadequate -- concrete slab floors (multistory) which collapsed. b. Joe King reviewed the 1968 Hooker Burlington PVC reactor fire. The incident occurred during the start up of a new French process. The process was a bulk process (no water) and the first of its type in the U.S. The process involves a two-step reaction in which there is a prepolymerization and a post-polymerization reactor. The material was being transferred from pre poly to post-poly and a leak occurred at transfer valve. The material caught fire and could not be extinguished with deluge system and local fire depart ment. The firewater capacity was being depleted after R&S 038658 2 two hours so additional fire groups were called. Nine fire trucks fought the fire by hooking up hoses to the Delaware River a mile away. The next day, after the reactor vessels were emptied, the fire was extinguished. Errors made and lessons learned: 1) The water capacity at site was insufficient. 2) After the first batch the transfer valve had been removed, worked on and reassembled. The seats were installed wrong and somehow (?) this contributed to the leak. 3) The reactor system was hydroed but the transfer valve was closed during test and only prepoly reactor and one side of transfer valve were checked. The leak occurred on downstream side. 4) Procedures were lacking. c. Dan Langlois described the Goodyear PVC reactor fire of 1972. The incident involved a 30-inch diameter nozzle which apparently was an entry manhole. Instead of a bolted flange cover the nozzle was fitted with a locking ring cover, the rotation of which wedged it shut. Somehow after startup the flange rotated to the open position and the cover blew off. and caught, and the escaping VCM caught fire. Three people were burned seriously. Prevention -- A locking bolt was devised with a proximity probe to prevent the ring from moving while in operation. d. Ben Reynolds of PPG described an upset and fire in their Oxy-EDC in Lake Charles in 1977. A feed imbalance of the C>2 to ethylene rate allowed O2 to rise to 30%. The first explosion occurred in the vent scrubber from possibly static ignition. The ensuing fire destroyed one reactor train and damaged the other two. Prevention - Redundant instruments and analyzers were installed. e. Herm Waltmate of BFG reviewed the VCM sphere spill in Altona, Australia. The design was very poor with nine 3 separate nozzles entering the bottom of the sphere. Maintenance workers were in the process of changing out a section of pipe on one of the 4" sphere nozzles. The nozzle had an air operated ball valve. The air actuator was supported by U-bolt on the section of pipe which was to be replaced. The workers disconnected the U-bolt and left the air actuator hanging loose on the valve handle. During the work activity the weight of the actuator finally opened the valve, causing a large spill. All the contents of the sphere, 1.5 MM lbs., was lost but luckily no fire occurred. Lessons learned: 1) A new strict maintenance work order procedure was established. 2) A new standard was established for liquid VCM valves. 2. Current Case Histories a. Jim Fields of Formosa Plastics described the tank car fire incident of July 30, 1983 in Baton Rouge. Five cars had been loaded before the fire. Two more were in the process of being loaded. The operator had disconnected one T/C and partially disconnected the second. After the fire all three angle valves on the second car were found open. Excess flow check valves did not seat and after the fire they were found to be improperly installed and could not seat off. The loading operator was burned seriously and has not been able to recall what happened. Errors and lessons: 1) Loader apparently disconnected the load hose with T/C valve still open. or 2) In attempting to adjust hose fitting, it blew off. 3) The clamp lock fittings were replaced with hammer unions. 4) EBV's were installed on all lines between T/C and load rack. 5) Instituted a check list to be filled out and signed for any leak situation. R&S 038659 4 6) The Railroad has suggested that T/C's being loaded have the placard display as loaded, to help fire fighters - b. Joe King of Occidental related an accident which resulted in a fatality in one of their plants. A high pressure water cleaning room is used to unplug heat exchangers. The room consists of two areas -- one where the water tanks and high pressure pumps are located, and the second where the exchangers are cleaned. The water is piped to the cleaning area and into a "foot" valve on the floor. A 1/4" stainless flexible hose 18 feet long with a nozzle is attached to the "foot" valve. The nozzle is a "mole" design, that is, it has one hole pointing forward, and six back, the thrust of those six causing it to self-propel through the tubes. There is no lance. The procedure is as follows: The cleaning operator inserts the nozzle into a tube, presses the foot valve, then signals a second operator to turn the pump switch on. Since the second operator is behind the cleaner, it is impossible to watch the nozzle insertion. After the accident the second operator mentioned that the stainless hose had been bucking unusually hard. The cleaner had cleaned seven tubes and when the pump was turned on for the eighth tube, the cleaner jumped back and grabbed at his throat and fell forward. The second operator turned him over and blood was gushing out at the neck. Help was summoned and they found the nozzle had entered at the base of the neck and shoulder and had traveled 18" into his chest. He died in the hospital 40 minutes later. Corrective actions: 1) Occidental personnel will no longer do this. The work will be done by contractors who do this type work. 2) The most significant cause was the use of a high pressure nozzle without a stiff lance for control. 3) The job will require that a pre-determined depth be drilled so that safe insertion of the nozzle with lance can be accomplished. 038660 5 c. Jim Wallace of Esso Canada told of a PVC resin hopper car which split open. The aluminum car had been in a previous wreck and required some rewelding of seams. The work had not been properly done (no weld penetration) and as the car was being unloaded an air pad was used to facilitate the job. About 50 psig had been applied and the shell split open. No one was injured. d. Wallace also reported on a VCM line failure caused by galvanic corrosion. A monel weld neck had been welded to a 304 stainless line and the stainless line corroded and failed. e. Lamar White of PPG, Lake Charles, reported on a loading pump incident. A large spill occurred from a pump after which it was determined that some force had cracked the pump pedestal and sheared all the pump and motor base bolts. The cracked pedestal had allowed the seal plate to move backward, causing the spill. The pump was steel construction on wetted parts and cast iron for the pedestal or support head. It was theorized that either 1) the pump had been blocked in liquid full, or 2) a water hammer had occurred when a long line was closed. The plant intends to replace the cast iron parts with steel or wrought iron. f. Solvay reported on a VCM T/C rupture in Austria. Their plant was contacted by a VCM producer and asked to unload a certain VCM T/C which had some type problem. When the car arrived it did not have the proper papers and was left in the track yard. A few days later, the T/C ruptured and spilled all of its contents. At one time liquid was seen blowing out of the car to a height of thirty feet. A large vapor cloud enveloped the area 5 feet deep. The plant facilities were shut down quickly and no fire occurred. Within an hour, most of the cloud had dissipated. The tank car inspection revealed a crack of 5 feet long by 8 inches wide. The T/C was not equipped with a relief valve. Errors uncovered: 1) Tank car had been overfilled originally because of math error. R&S 038661 -6 2) Person who reviewed weight sheets did not detect problem. 3) Procedure called for second weighing before shi,. ment but second scale had been out of order for six years. 33 So CO 4) Custom inspector did not find error. 5) When tank car reached Italian border, error was discovered but not communicated. 6) Solvay was not advised what problem was but was asked to unload. Health - PVC Dust John Vaughan of ICI reviewed the work going on in his company with regard to eliminating PVC dust inhalation. Although not considered toxic, the dust does coat the lungs and cause respiratory problems. In order to pinpoint sources and concentrations of dust, they have developed a high intensity light method which clearly shows the dust in the atmosphere and source. NOTE: Maybe the high intensity light method would be useful within Dow, for example in detection and measurement of asbestos fibers. The rest of the meeting was PVC related or had no message that needed to be communicated. RRO/pm 11-3-83 1983 VSCA MEETING VALVE POSITION SWITCH SURVEY SUMMARY A. Survey Participation Total Number of Companies Responding Total Number of Plants Responding Total Number of Forms Returned B. Types of Switches Used Electromechanical - 15/26 Inductive Proximity - 6/26 Hall Effect Proximity - 5/26 58% 23% 19% C. Mean Time Between Failures (Months) Type Switch Electromechanical Inductive Proximity Hall Effect Proximity Min Avg Max 1 16 60 1 9 24 2 7 10 D. Cause of Failure Shifting Mechanical Failure Water Damage by Personnel Vibration Bent Indicator Arm Electrical Failure Total Electro mechanical 4 2 8 5 2 2 22 Inductive Proximity 1 5 1 1 8 Hall Effect Proximity 3 2 3 1 1 _2 12 E. Maintenance Performed Realign or Replace 33 Co CO too o0>0 05 CO 1083 VCSA MEETING VALVE POSITION SWITCH SURVEY ELECTROMECHANICAL LIHIT SWITCHES Manufacturer Proximity Microswitch Model Number V3L-3-DB V3L-139-08 RX21 and 3R021 No. of Switches 21 + 21 + 21 + 21+ Cycles per Day 10 20 2 4 Bettis Fisher Controls Jamesburg Microsvitch Microswitch Mlcroswltch Bettis Switch 3R 304 069-0072-00 EXAR and EXAR30 5IML1-7646 1CX42-7714 2CX3-7904 3R321AFC 3R02IAFC 21 + 21 + 21 + 21 + 21 + 21 + 21 + 6 8 4 4 8 8 8 YamatakeHoneywell ILX5001 VCX500I 21 + <20 Microswitch Bettis Microswitch LSXA - 21 + 6-20 21 + 2 5 3-30 Mean Time Between Failures Cause of Failure Maintenance Performed 3 Tears X Years 2 Years Shifting of switch Shifting of switch Mechanical failure Cam indicator shifting and armature failure 3 Years 10 Months 1 Year 6 Months 3 Months 10 Months 6 Months Broken apring (rare) Water in switch Shifting of switch Water in switch Mechanical damage by personnel Water in switch Bent indicator arm Water in switch Mechanical damage by personnel Dent indicator arm Water in switch Mechanical damage by personnel 5 Years Water in switch (2 or 3/Year) Mechanical damage (2 or 3/Year) by personnel 1 Month Water in switch Vibration 2 Months (w/o failure) 6 to 8 Water in switch Months Shifting due to vibration Alignment A11gnment Replacement Cam alignment and replacement Armature repair occasionally Alignment Alignment Replace switch Alignment Replacement of switch Alignment Replacement Replacement of feedback arm Alignment of cam;i end Indicator a rtn Replacement of sw11 ches Alignment or rep LacemenC Replacement -- Repair (replace internals) Realign Unusual Ambient Conditions Vibration Vibration No No Comments on Performance Excellent Very Good Fair Fair No Good No Cood No Good Mo Cood No Good No Good Nn Cood* most problems caused by water and personnel No Good Deluge Fair Vibration Installed for only two months Vibration Good. Some in service High Humidity for eight years with no failure ^99800 S9H 1983 VCSA MEETING VAEVE POSITION SWITCH SURVEY HALL EFFECT PROXIMITY SWITCH CKACNETIC SENS1NC) Manufacturer In house Wesco Wesco Wesco Model MS2A, MS2B MS2AAOO MS2-BB-00 No. of 21* 21 + Cycles per Day 2 16 21 + 5 ,tnax 15/day 21 + 12 In house 21+ 12 Mean Time Between Failures Cause of Failure Maintenance Performed -- 8 Months 2 to 3 Montha 10 Months 8 Honths Slipped magnet Mechanical damage Water in switch Slipped magnet or loss of magnet Slipped magnet Water in switch Vibration Mechanical damage 301 Electrical failure 701 Bent indicator arm LOX Mechanical damage AOS .otrical failure 301 Alignment or replacement Replacement of switch Alignment of magnet Alignment or replacement Replacement Alignment of magnet arms and magnets Unusual Ambient Conditions No No Comments on Performance Good on automatic valves Fair on manual valves Cood Vibration and deluge water Fair. Performance improved with new magnet bracket design. Average -- Fair performance, requires excessive maintenance S9980 S9b 1983 VCSA MEETING VAl.VE POSITION SWITCH SURVEY INDUCTIVE PROXIMITY SWITCH (METAL SENSING) Manufacturer Pepperl and Fuchs (Germany) Model Number N7-2-12-CK-N No. of Switches 21* Cycles per Day 2 General Equip- 211117 oent and Manufacturing^ Co. Inc. (CO1 ) Microswitch 6FRI-6 Ceneral Equipment and Manufacturing Co. Inc. Ceneral Equip ment and Manufacturing Co. Inc. Ceneral Equip ment and Manufacturing Co. Inc. Series 20 and AO 257116 and 13929 211536 21 + 2 21 + 16 21 + 5, max of 15/day B/year 21* 12 Kean Time Between Failures X Years +6 Months Cause of Failure Maintenance Performed Bent indicator arm (2/Year) Mechanical damage by personnel (1/Year) None yet, installed 6 months ago Alignment Unusual Ambient Conditions None None Comments on Performance Satis!led Good 3 Months 1 Month Mechanical damage by personnel Vibration Mechanical damage Allgnmcnt or replacement Alignment or replacement Only in service a few months 2 Years Mechanical damage by personnel Phyaical damage 951 Shifting of switch 52 Alignment Replace due to excessive dead band Replacement None Good Vibration and deluge water Fair Dead band has been a problem. Difficult t adjust position of th switches. Works well and Is the most reliable positto switch used In the plant. 3998S0