Document QJDZZR0rE8ZwyLEeJ2bJn5j56

Summary: VCM RELEASE WHICH OCCURRED FEBRUARY 2, 1979 FROM REACTOR D-300 . r ? -VA. !j ' j V. V* 1 r-*^ *w i Is !5 * -f rv 11 < J This release occurred due to an uncontrolled pressure build-up in reactor D-300. Corrective actions to reduce the possibility of reoccurance of this type of event include revised kill procedures, installation of an effective emergency AMS kill system and installation of emergency cooling and emergency instrument air systems. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSES(S). Response: Descriptions giving the nature and cause of the discharge, including primary and/or related cause(s) have been submitted to the EPA. These are listed below. A 1. The initial discharge report was sent to EPA-Region IV in a letter dated February 9, 1979 which is attached as reference A-l, Appendix A. 2. Mr. Leon Folsom, EPA-Region IV, visited the Aberdeen Plant on February 21, 1979 to obtain information on this and other releases. As a result of his visit, requested additional information was supplied in a letter dated March 5, 1979 which is attached as reference A-2, Appendix A. 3. On August 21, 1979, Wayne Aronson of EPA Region IV and an EPA contractor (PEDCo Environmental, Inc.) conducted an inspection of the plant. Records were searched and relevant information was copied. As a result of this inspection, a report dated August 31, 1979 was issued by PEDCo. which contains details on the February 2, 1979 discharge. (Reference 1) 4. On October 15, 1980, Mr. Curry L. Miller, Plant Manager, responded to items in the PEDCo inspection report, correcting misleading and incorrect statements in this report. This correspondence, sent to Mr. Wayne Aronson, is attached as reference A-3, Appendix A. QUESTION 2; FOR EACH DISCHARGE DESCRIBED IN RESPONSE TO III.A.1. A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VC TO THE ATMOSPHERE INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VCM EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. VAB.0001113158 ri Question 2: (Continued) CONFIDENTIAL Response: In addition to the correspondence mentioned in question 1 above, the following information has been supplied to the EPA Atlanta office relevant to corrective action taken to prevent or minimize the release. 1. Pursuant to the April 11, 1979 show cause conference, Mr. Curry L. Miller,Plant Manager, submitted a list of measures to reduce the potential for relief valve discharges at Aberdeen. The correspondence was sent to Mr.James T. Wilburn in a document dated April 23, 1979 which is attached as reference A-4, Appendix A. The following additional information is submitted relevant to corrective action taken before and after February 2, 1979 to prevent and/or minimize the release of VCM due to uncontrolled pressure build-up in reactors. I. OPERATING PHILOSOPHY It has been the plant's operating philosophy that the polymerization reaction in a PVC reactor must be stopped during an upset condition in order to prevent uncontrolled pressure buildup in the reactor and subse quent relief discharge of VCM. Various systems could be employed to handle specific upset conditions, e.g.back up power generation in case of power failure. However, these systems will not be necessary if the polymerization reaction is stopped because there can be no pressure buildup in the absence of a polymerization reaction. Additionally, backup power would be of no use in certain upsets, e.g. loss of agita tion. The following material will show that the plant took appropriate action both before and after the release of February 2, 1979 to minimize the possibility of a relief valve discharge occurring from uncontrolled reactor pressure buildup. This is evidenced by the fact that since February 2, 1979 in which the Nitric Oxide (NO) Kill System tubing failed unpredictably, there has not been another relief valve discharge for this reason. This information will also show that the plant has been constantly improving procedures and adding equipment and instrumen tation as new ideas and process development has become available to further insure that there will be no uncontrolled pressure buildup in the reactors resulting in the release of vinyl chloride. Following is information which describes the equipment, instrumentation and procedures which were in use in the plant prior to the February 2, 1979 release. Since the release, the following improvements have been made to further reduce the likelihood of a relief valve discharge due to over-pressurization of a polymerization reactor: A VAB.0001113159 . Additional reactors instrumentation 2 Additional reactor interlocks 3. Procedures have been improved 4. An improved emergency kill system has been installed 5. An emergency reactor cooling system to supply cooling water . in case of a power failure has been installed 6 An emergency diesel generator has been installed to supply power to operate two of the plant's instrument air compressors in case of a power failure 7. A chain transfer agent was developed and is now used in the . production of low molecular weight resins 8 The expansion reactors were designed with a pressure rating of 25 pounds per square inch greater than the previously installed These improvements are described in detail in the following sections. The procedures and equipment are designed to control the reactors during the polymerization stage, and to insure that the kill system will be effective 100% of the time. II. REACTOR INSTRUMENTATION TO DETECT UPSET CONDITIONS A. Prior To February 2, 1979 Prior to February 2, 1979, the following instrumentation was avail able to detect upset conditions in the reactor. 1. Reactor temperature and pressure readouts were available in the control room. The temperature and pressure were also con tinuously recorded on the control panel recorder for each reactor. 2. High reactor pressure and temperature alarms that sound in the control room were available in the control room. These alarms must be acknowledged by the control panel operator to silence the alarms. 3. Cooling water flow to the jacket and the condenser was indicated and recorded on the control panel. 4. Local instrumentation (pressure gauges, flowmeters, etc.) were available to the outside operator in detecting upset conditions or malfunctions that could lead to upset conditions. 5. The outside operator was equipped with a two-way radio to keep in constant communication with the panel operator and the shift supervisor. 6. Reactor instrumentation in the control room was supplied with 24 volt D.C. backup power from batteries. 7. Numerous interlocks were installed that prevent an operator from making errors or opening valves that would result in a runaway reaction or other upset condition. (See Reactor Interlock Section, page 14) 8. Reactor agitator amperage indicators were available in the con trol room for each reactor. -3- VAB.0001113160 B. Since February 2, 1979 The following instrumentation has been added since February 2, 1979 in order to increase the capability to detect upset conditions in the reactor. 1. Reactor agitator amperage recorders have been added in the control room. 2. A severe weather radio is available in the control room to warn of approaching storms that could interfere with normal opera tions. III. REACTOR KILL SYSTEMS A. General Description On February 2, 1979, the plant employed two reactor kill systems-- the normal kill system and the emergency Nitric Oxide (NO) kill system. The normal kill system uses a small pot dedicated to each reactor. Alpha methyl styrene (AMS) is poured into the pot and the contents of the pot are then injected into the reactor using high pressure service water. This system is used to terminate the polymerization reaction at the end of each batch. This is typically done with the addition of a small amount of AMS (1*5-2 gallons). This system can be used to inject larger amounts of AMS during non-power-failure emergency situations to terminate reactions. As a backup system to the normal kill system, an emergency nitric oxide (NO) kill system was available on each reactor. This system was activated by opening valves on high pressure NO cylinders connected to the reactors. Through efforts to further reduce the possibility of a discharge of VCM due to uncontrolled reactor pressure buildup, new reactor kill system technology was developed. This new technology also utilizes AMS as a kill agent. This superior kill system is currently utilized at both the Oklahoma City PVC Plant and the Aberdeen Plant. B. Emergency NO Kill System 1. General Description of Emergency NO Kill System The emergency Nitric Oxide (NO) kill system consisted of several pressurized cylinders containing NO connected to each reactor through valves and high pressure tubing. Each reactor had its own bank of NO cylinders. The cylinders were equipped with individual valves as supplied by the vendor. High pressure tubing was connected to these valves and extended to the reactor lip seal flush inlet. At all times, pressurized water flows into the reactor through the seal flush line. When the Emergen cy NO kill system was to be activated, a valve at the reactor agitator lip seal flush was opened and the individual valves at the NO cylinders were opened. This allowed NO to flow into the reactor and terminate the polymerization reaction. VAB.0001113161 i CONFIDENTIAL 2. Training and Instructions On Emergency NO Kill Systems and Power Failure Procedures a. Prior To February 2, 1979 Plant personnel were trained in the use of the emergency NO kill system and were trained in emergency operating proce dures. Note that records are not required to be kept for this time period. Training records and documentation of the exact instructions are not complete for this time period since it was so long ago. However, it is strongly believed that operators were thoroughly trained in responding to emergency situations as evidenced by the records that have been kept. It is certain that much more training took place than the documentation shows. Plant records show formal training of operators in emergency procedures took place on 4/22/77, 7/22/77, 3/20/78 and 6/12/78. These records are attached as references A-5 thru A-8, Appendix A. In addition to this, on the job training was also received by plant operating personnel. b. Since February 2, 1979 1. On February 8, 1979, revisions to the operating manuals were issued which included pressure and temperature limits for killing runaway reactions and wearing of Scott Air Paks, which are positive pressure full face respirators, during use of the Emergency NO kill system. Copies of these procedures were posted on Bulletin Boards and covered with all appropriate operating personnel in February and March, 1979. The revised Operating Manual section is attached as reference A-9, Appendix A. Plant records showing formal training of operators in emergency procedures are attached as references A-10 thru A-13, Appendix A. 2. On March 20, 1979, further instructions were issued regarding emergency kill procedures, checking of the emergency NO kill system, inspection of the reactor condensers after a coarse batch, and several other items relative to relief valve discharge prevention. This is attached as reference A-14, Appendix A. Plant records showing formal training of the operators in the revised emergency procedures are attached as references A-15 through A-18, Appendix A. A coarse batch can occur as a result of a suspension system failure. When a coarse batch occurs, large chunks of polymer can form. These chunks can reduce the effectiveness of the reactor heat transfer surfaces on subsequent batches. On April 10, 1979, these instructions were expanded to include reducing catalyst loading to the reactor on the batch after a coarse batch. In effect, this slows reaction rate which reduces the possibility of uncontrolled pressure buildup in a reactor. These instructions were given to the shift supervisors who give instruction to operators on catalyst loading. The expanded procedure is attached as reference A-19, Appendix A. VAB.0001113162 n 4 ir b. Since February 2, 1979 (Continued) CONFIDENTIAL 3. Other training to the operators in use of the Emergency NO kill system was given on 7/7/79 and 7/14/79. Training meeting records are attached as references A-20 through A-23, Appendix A. 4. On October 25, 1979 and October 30, 1979, the emergency pro cedures were revised to include injection of large amounts of AMS at specified temperature and pressure increases. Also, specific instructions for power failure operations were given. These are attached as references A-24 and A-25, Appendix A. Plant records showing formal training of operators in the revised procedures are attached as re ferences A-26 through A-30, Appendix A. 5. Another specific memo regarding emergency NO kill procedures was issued March 4, 1981. This is attached as reference A-31, Appendix A. Additional specific training on emergency kill procedures was given on June 29, 1980, July 15, 1980 and April 9, 1981. These are attached as references A-32 through A-34, Appendix A. 3. Emergency NO kill system inspections a. Prior To February 2, 1979 Prior to February 2, 1979, the NO kill systems were checked on a routine basis by operating personnel who daily monitor equipment and especially emergency equipment. There is no documentation of this daily monitoring, but the monitoring occurred just as all other equipment is monitored to keep it operational. Formal checks of the NO kill systems took place on a less frequent basis and documentation of such inspections is sketchy. One such inspection took place on March 13, 1978, in which all the NO kill systems were checked. This inspection was noted in the the supervisors log book. The item is entry number 5 of the 1500-2300 log dated 3/13/78 which is attached as reference A-35, Appendix A. These checks of the NO systems consisted of manual inspection of the systems to assure valves were in the correct open and closed positions and nitrogen was injected into the reactor to assure there was no blockage of the tubing. Since NO is an extremely toxic gas, pressurization of the tubing which was used to inject the NO into the reactor was not done with NO as this could re sult in employee exposure to NO. There was no reason to expect that a pressurized check of the tubing was necessary as the pressure rating of the tubing was in excess of 7000 pounds per square inch and the pressure in the NO cylinders was approxi mately only 500 pounds per square inch. Failure of the tubing was such an unlikely event that it could not be anticipated. A i `T i VAB.0001113163 3. Emergency NO kill system inspections (Continued) a. Prior to February 2, 1979 (Continued) The NO entered the reactor through the water lip seal flush line to the reactor agitator seal. The lip seal flush must be maintained at all times to prevent damage to the agitator seal. A flow meter is provided on each lip seal flush line and this flow is monitored at least once per batch to assure proper flow. Thus, since the NO injection point into the reactor is through a port that always has water flowing through it, the port is always open and NO can be injected into the reactor when needed. Using this port prevents the possibility of trying to inject NO into a reactor through a seldom used port which would likely be plugged with polymer when an emergency exists. b. Since February 2, 1979 After the February 2, 1979 incident, the plant instituted a formal procedure to check the NO kill systems every six months. In addition to the visual check to assure proper valve alignment and the addition of N into the reactor to assure no pluggage which were previously done, the systems were also pressured with N to check the physical integrity of the system. This pressure cneck of the system was done to prevent a mechanical failure of the system as occurred in the February 2, 1979 incident if the systems were to be needed in an emergency. Records of the six-month inspections of the NO kill systems were kept on large check-off sheets by the Vinyl Operations Depart ment. Copies of these sheets for 1980 and 1981 are available at the plant for inspection. The dates of the inspections of the NO kill systems during this period are shown in the attached two tables. This six-month check of the NO kill systems was continued until the systems were removed in mid 1984. The large check-off sheets for the period 1982 to 1984 have been discarded and there fore a listing of the dates the systems were checked during this period cannot be provided. The Vinyl Operations supervisors have stated such inspections and records were kept during this period but the check-off sheets were not saved. A VAB.0001113164 1 - .++*' Iij ' Dd'r -lUHr - : CO i 'T -8- VAB.0001113165 -9- r i * ? VAB.0001113166 4. Removal of Emergency NO Kill System CONFIDENTIAL Initially, the plant's NO emergency kill system was installed to stop the polymerization reaction in a PVC reactor during an emer gency in order to reduce the possibility of an uncontrolled pressure buildup in a reactor. However, since these original installations, the plant installed a superior emergency kill system that uses AMS as the kill agent. The installation of these kill systems was complete in 1982. A description of this system is given later. Since the installation of this system, the plant has successfully used the emergency AMS kill system several times and had not had occasion to use the NO kill system. Because of our experience with this system and our concern for the health and safety of our employees, a decision was made to remove the emergency NO kill sys tem. Notification of this decision was given to the Mississippi Bureau of Pollution Control in a letter dated June 11, 1984. This is attached as reference A-36, Appendix A. On July 3, 1984, the Bureau responded that it concurred with the decision . The response is attached as reference A-37, Appendix A. The emergency NO kill system was removed in mid 1984. C. Emergency AMS Kill System As an added measure to assure that polymerization reactions could be terminated in emergency situations, a superior kill system was developed by Conoco which utilizes AMS as the kill agent. Funds to install this type kill system at Aberdeen were approved in November of 1980 for the existing eight reactors. The design for the two expansion reactors, one of which started up in late 1981 and the other in early 1982, included this type AMS kill system for each reactor. Installation of the AMS emergency kill system was complete on the existing eight reactors in mid 1982. Following is a descrip tion of the development of this system and a description of the sys tem as it exists today. Large Reactor Pilot Plant AMS Iniection Tests In order to confirm that AMS was effective in killing the polymeri zation reaction in an emergency situation, test runs were made at ConocoTs large reactor pilot plant in Oklahoma City, Oklahoma. The 1600-gallon pilot plant reactor with its condenser, agitator and general configuration is similar to that of the plant's commercial reactors. More than 1000 test batches of PVC made in this pilot plant reactor show that it does simulate the plant's commercial * The pilot plant AMS test runs were carefully planned to simulate concurrent loss of reactor agitation and normal cooling water flow as could occur during a power failure. Every effort was made to simulate a realistic, worst case situation. In each case the simulated power failure (shutdown of agitator and stop cooling water flow) was started three hours into the polymerization to insure that the polymerization rate was near maximum. Also the was allowed to increase from 120 psig, the normal run pressure, to 140 psig before Injecting AMS to the reactor. The seven to eleven minutes required to reach 140 psig provided sufficient time for the swirling inside the reactor to stop thereby minimizing the AMS -10- VAB.0001113167 T J " C. Emergency AMS Kill System (Cent.) Large Reactor Pilot Plant AMS Injection Tests (Cont.) mixing that could occur due to residual swirling. This was done to test the ability of the AMS to deactivate radical polymerization initiators with minimum mixing. AMS ble in VCM. The results of the first test are displayed graphically as Curve A on the graph attached as reference A-38, Appendix A. After AMS injection the reactor pressure increased 6 psi to a maximum of 146 psi and then decreased. Sixty minutes into the test, the reactor pressure had fallen to 139 psig. At that point maximum cooling water flow was started through the reactor1s condenser. This caused the reactor pressure to drop quickly and is a good demonstration of the pressure control capabilities of a system combining AMS injection with back-up cooling water flow. While this test did show that AMS injection would stop the polymerization, it appears that there must have been some cooling water flow leaking through the reactorrs condenser and jacket during the test since the reactor pressure decreased from 146 psig to 139 psig while the cooling water was supposedly shut off. This may have masked a slow reactor rate if the AMS was not 100% effective. Therefore, the test was repeated with the cooling water valve and main supply header valve blocked in and the condenser and jacket drain plugs removed at three hours into the polymerization run. The results of this test are shown by Curve B on the reference A-38, Appendix A. After AMS injection, reactor pressure increased from 140 psig to a maximum of 150 psig. It then slowly lowered to 147 psig seventy minutes into the test. This gradual pressure drop was probably due to atmospheric heat losses from the reactor and condenser. In past work, Conoco*s research group evaluated nine other polymeri zation stopping chemicals or chemical mixtures and has found none to be better than AMS. System Description The emergency AMS kill system consists of a pressurized pot filled with AMS for each reactor with piping and valves to inject the AMS into the reactor. The pressurized pot is maintained at above 250 PSIG with bottled nitrogen. Piping runs from the bottom of the AMS kill pot to the top of the reactors. AMS can be injected into the reactor by either of two nozzles. One nozzle has remote operated valves that can be operated from the vinyl control room panel. The other nozzle has manual valves that can be operated manually from the platform at the top of the reactor. Both nozzles inject the AMS into the reactor vapor space. Bottled nitrogen is used to pressu rize the injection pot. The pressure in the injection pot and in -11- A VAB.0001113168 C. Emergency AMS Kill System (Continued) System Description (Continued) the nitrogen bottle is checked before each reactor charge. If the pressure in either the bottle or the pot is low, the problems must be corrected before the reactor is charged. The remote operated valves are also operated with bottled nitrogen. This not only assures that these valves can be activated during a power failure. but also eliminates possible freezing problems that might occur with instrument air. The pressure in this nitrogen bottle is also checked before each charge to insure enough pressure is available to operate the valves. If the pressure in the nitrogen cylinder is too low, the cylinder must be changed before the reactor is charged. The remote operated valve switches also are connected to the plant's emergency power to insure their operability during a power failure. Each AMS injection pot is equipped with a local level gauge to moni tor the amount of AMS available for stopping polymerization. This level is checked before each charge. Each pot also has a low level alarm that will sound in the vinyl control room and light a beacon on top of the reactor if the level in the pot is below a certain point. This level alarm system serves several functions. If the vinyl panel alarm is activated before charge, AMS must be added to the pot before the reactor can be charged. During AMS injection, the low level switch will close the automatic valves. If AMS is injected into the reactor with the manual valves, the beacon on the reactor will alert the operator to close the valves so that nitrogen will not be injected into the reactor. In either the manual or automatic AMS addition, the level alarms are evidence that AMS has actually been injected into the reactor. There are also several other items that insure the operability of the emergency AMS kill system in almost any conceivable situation. 1) Each injection pot is equipped with a low pressure alarm that is activated when the pressure in the pot is below 250 PSIG. If the alarm should sound, the low pressure problem must be immediately corrected. 2) There is also a low pressure alarm on the nitrogen supply to the automatic valve actuators. This alarm will sound if the pressure in the cylinder is below 600 PSIG and the cylinder must then be changed. 3) All the AMS kill pots in each reactor module are connected with a flexible hose in such a manner as to allow AMS to be injected manually to any reactor from any injection pot. For example, AMS can be injected into new module reactor D-741 with AMS from the injection pots for reactors D-742, D-743, D-744, or D-745, 4) The AMS injection nozzles on each reactor are checked weekly to assure they are open. This frequency adequately ensures that AMS can be added to the reactor when required. 5) When the ambient temperature is below 20F AMS is injected into each reactor once per shift to assure the AMS injection lines are not frozen. 6) In addition to the checks made by the operator before each reactor batch, an independent audit of the emergency AMS kill system is conducted weekly. Among the items checked weekly are AMS pot , nitrogen bottle pressure, AMS level, and valve position. A r.,, i -L -Un. b- ' VAB.0001113169 Successful Emergency AMS Utilization fi i i -l !* J i j UL11 The emergency AMS kill system has been used successfully on numerous occasions since its installation. Listed below are a number of occasions on which it has been successfully used. It should be noted that this list is not meant to be all inclusive, rather representative of the effectiveness of this system. It should also be pointed out that the emergency AMS kill system has never failed to fulfill its intended function of killing a reactor in emergency situation, thereby preventing releases of vinyl chloride Date Number Of Reactors Emergency Type 9/21/83 1 Agitator Failure 11/2/83 1 Cooling Water Valve Failure 11/14/83 3 Power Failure 7/17/84 3 Power Failure 8/28/84 7 Power Failure 9/28/84 2 Power Failure 1/2/85 6/4/85 1 7 Agitator Failure Power Failure Changes to Standard Operating Procedure Mr. Curry L. Miller's letter to Mr. James T. Wilburn, dated April 23, 1979 is attached as reference A-4, Appendix A. The letter listed measures which would be taken at the plant to reduce the potential of relief valve discharges. Two of the items listed are no longer part of the plant standard operating procedure. These items are: "6. The standard operating procedure has been changed so that after every coarse batch, the reactor condenser will be inspected to determine if it is plugged. If it is found to be plugged, it will be cleaned prior to charging the next batch in that reactor. 7. The standard operating procedure has been changed to require a reduction in initiator loading for any batch following a coarse batch in order to slow down the reaction rate." VAB.0001113170 Changes to Standard Operating Procedure (Continued) CONFIDENTIAL A These practices have been discontinued because the reactor condensers have not been found to be plugged after coarse batches and temperature control has not been a problem after coarse batches. With the proven effectiveness of the emergency AMS kill system and the lack of fouling due to coarse batches these two items are no longer necessary and are therefore no longer part of plant practice. Item 9 of this same letter pertains to operator training. The current operator training procedures will be detailed in the compliance plan which will be submitted at a later date as required by Section III.B. of the Consent Decree. IV. REACTOR INTERLOCKS To reduce the possibility of a reactor panel operator error causing an emergency situation that could result in the discharge of vinyl chloride, the plant uses an extensive reactor interlock system. A pro grammable controller is utilized to accomplish this function. The main interlocks prevent certain valves from opening in specific reactor operating modes. These modes of operation are those that are required to produce a reactor batch and are listed below. Evacuation Charge Polymerization Recovery/Steam Stripping Dump Clean/Rinse Swirl A typical reactor interlock listing is attached as reference A-39, Appendix A. A "N" indicates that a valve cannot be opened in a specific mode. A "y" indicates that the valve can be opened in the specific mode. For example, the steam stripping valve (valve 712B) cannot be opened in the polymerization mode. There are also other specific interlocks which preclude operators from opening certain valves or doing certain operations when permissive con ditions are not met. There are numerous other interlocks that are provided to prevent certain combinations of valves from being opened which would result in undesirable operating conditions. There are also interlocks provided specifically to reduce the possibility of relief valve discharges. These are listed below. 1. A reactor cannot be charged unless the agitator is running and the amperage indicates that the reactor is empty. VAB.0001113171 1 ' 2. The quantities necessary to charge the expansion reactors ?, i pu cannot be charged to the other reactors. The quantities charged* in the expansion reactors are much larger than those charged to the other reactors in the module. If this interlock were not present, reactors could be overfilled with the wrong charge size. 3. There are precharge interlocks that prevent more than a total quantity of ingredients to be charged to the reactor to prevent overfilling the reactor due to overcharging. V. EMERGENCY REACTOR COOLING AND EMERGENCY INSTRUMENT AIR SYSTEMS In order to further increase capabilities to control emergency situations dealing with power outages, the plant has installed an emergency reactor cooling system and an emergency instrument air system. The cooling can aid the effectiveness of the emergency AMS kill system by creating some mixing in the reactor. The mixing occurs due to the boiling action resulting from condensing vapors in the reactors with the reflux condensers on each reactor. The emergency cooling water system provides cooling water to the reactor condensers from the plant fire water loop. The water supply to the fire water loop is provided by diesel driven pumps. This emergency cooling system provides 2,000 gallons per minute of cooling to each reactor module. With this cooling water available, water can be directed from reactor condenser to reactor condenser to control the The emergency instrument air system uses a diesel driven electric genera tor to supply power for two plant reciprocating air compressors that are dedicated to supplying air to both reactor modules. This system provides air to insure that valves on reactors will be operational during a power failure emergency. The reactor valve switches located in the control room are powered by emergency power (24 volts DC). VI. DEVELOPMENT OF A CHAIN TRANSFER AGENT Production of resin grades at the highest operating pressures use a chain transfer agent (CTA) to lower the pressure. These resins have low molecular weights and the CTA enables the same grade of resin to be pro duced at lower polymerization pressure. This technology was developed to lower the production pressure of these grades of resin. Lowering of the production pressure allows more time to respond to an upset condition and provides more of a gap between the run pressure and the set point pressure of the reactor relief valves. VII.HIGHER PRESSURE RATING OF EXPANSION REACTORS The pressure rating of the reactors installed in the 1982 expansion is 25 pounds per square inch higher than the other reactors. This allows more time to respond to an upset condition and provides more gap between the run pressure and the set pressure of the reactor relief valves. VIII. OTHER RELATED ITEMS Other items or procedures that have been installed to reduce the likeli hood of a discharge to the atmosphere from an uncontrolled pressure buildup or reduce the magnitude of the release are listed below. -15- VAB.0001113172 VIII. Other Related Items (Continued) 1. The double rupture disc assemblies on the reactors have all been replaced with rupture disc/relief valve assemblies. This was done to reduce the amount of a release from a discharge that might occur on a reactor as the relief valve should reseat whereas the double rupture disc assemblies would continue to discharge as long as there is pressure in the reactor. 2. Often, during an emergency situation, other reactors in the same module are in operating modes that would allow them to accept material from another reactor in the same module. During an emergency situation, this can be used to reduce the possibility of discharges of vinyl chloride. 3. A severe weather radio is located in the Shift Supervisors office in the control room and is monitored by the control room personnel. The Shift Supervisor, at his discretion, may have reactors that are in the polymerization portion of the cycle killed if he believes the weather is severe enough that a power failure may occur. Also it is plant practice not to charge reactors during severe thunder storms. QUESTION 4: FOR EACH VC DISCHARGE IDENTIFIED IN III.A.l. AS BEING CAUSED, IN WHOLE OR IN PART BY EQUIPMENT MALFUNCTION OR DEFECT, THE FOLLOWING INFORMATION: a) DESCRIPTION IN DETAIL OF THE CAUSE OF THE MALFUNCTION OR DEFECT WHICH RESULTED IN THE VC DISCHARGE. b) DESCRIPTION OF THE APPLICABLE INSPECTION AND MAIN TENANCE PROCEDURES FOR THE EQUIPMENT THAT MALFUNCTIONED AND IDENTIFICATION OF THE FREQUENCY OF INSPECTION RE QUIRED BY SUCH PROCEDURE. IF THE ACTUAL INSPECTION/ MAINTENANCE OF THE AFFECTED EQUIPMENT IS NOT CONSIS TENT WITH THE REQUIRED PROCEDURE, DESCRIBE THE INCON SISTENCY AND EXPLAIN WHY. Response: The February 2, 1984 release was in part due to the unpredicted failure of the NO kill system tubing. From February 2, 1979 to July of 1984, when it was removed, the NO kill system was checked for integrity every six months. VAB.0001113173 T r CO m co \n oo i VAB.0001113174 VCM RELEASES FROM THE OLD MODULE BLOWDOWN TANK SAFETY RELIEF VALVE ON APRIL 3, 1980 AND APRIL 7, 1980 Summary: These releases occurred because the old module blowdown tank was rated for only 66 psig whereas some of the systems feeding the blowdown tank operate normally at 70 to 80 psig. In this situation, any malfunction in the emission recovery system which removed vapors from the blowdown tank or an excessive VCM feed rate could quickly lead to overpressuring the tank. This tank has now been replaced with two new vessels with design pressures of 150 psig so that this type of problem can no longer occur. Details on the corrective steps which were taken to prevent any more VCM releases from this source are included in the answer to Question 2. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSE(S). Response: Descriptions of these incidents are provided in the documents which are referenced below: April 3, 1980 1. This incident was reported In a letter to Rebecca Hanmer, Region IV EPA dated April 8, 1980 which is attached as reference D-l, Appendix D. April 7, 1980 1. This incident was reported to Rebecca Hanmer, Region IV, EPA in a letter dated April 16, 1980 which is attached as reference D-2, Appendix D. QUESTION 2: x FOR EACH DISCHARGE DESCRIBED IN RESPONSE TO III.A.l., A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VC TO THE ATMOSPHERE, INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VC EMISSION VENTS TO A COLLECTION VESSEL OR INCINERA TOR. * VAB.0001113175 Response; The relevant corrective steps taken before and during the April 3, 1980 and April 7, 1980 releases are discussed in the references provided under Question 1. The information provided below and documented in the listed references is additional information relevant to corrective steps that were taken before and after these incidents. ADDITIONAL DETAILS ON CORRECTIVE STEPS THAT WERE TAKEN BEFORE THE APRIL 3, 1980 RELEASE Operating procedures were written and issued to provide the operators with guidelines to prevent operating problems. A copy of the operating procedure dated April 27, 1979 is attached as reference D-3, Appendix D. ADDITIONAL DETAILS ON CORRECTIVE STEPS THAT WERE TAKEN AFTER THE APRIL 3, 1980 RELEASE The shift operators were trained in the blowdown tank operating procedures. A record of the training of one of the shifts dated 4/6/80 is attached as reference D-4, Appendix D. Normal plant practice would have called for the operators on the other shifts to have also been trained. However, records of this training have not been located. ADDITIONAL DETAILS ON CORRECTIVE STEPS THAT WERE TAKEN AFTER THE APRIL 7, 1980 RELEASE A new batch waste water stripping system was installed and put into operation in July 1981. Notification of the startup of this stripper was given to Rebecca Hanmer, EPA Region IV, in a letter dated June 1, 1981 which is attached as reference D-5, Appendix D. This stripping system replaces the old blowdown drum which was taken out of service. Since the new stripper has a design pressure of 150 psig it is not subject to the type of problem which caused these releases. VAB.0001113176 > LT> cmo VD H \J1 VAB.0001113177 A VCM RELEASES WHICH OCCURRED SEPT. 16, 1981 AND NOV. 5, 1981 FROM THE VCM CHARGE FILTER IN REACTOR MODULE NO. 1 Summary: These two releases occurred from the thermal relief valve on the north fresh VCM charge filter in reactor module No. 1. The filterpak was overpressured in both cases but the cause for the overpressuring was different. These thermal relief valves have now been tied into the batch water strippers. Since no further have occurred this change is believed to have solved the problem. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSE(S). Response: Descriptions of each of these incidents have previously been provided to the Region IV EPA in the documents which are referenced below: September 16, 1981 1. This incident was reported in a letter to Charles R. Jeter, Region IV EPA, in a letter dated September 23, 1981, which is attached as reference E-l, Appendix E. 2. This incident was also reported to the National Response Center on September 16, 1981. November 5, 1981 1. This incident was reported in a letter to Charles R. Jeter, Region IV EPA, dated November 13, 1981, which is attached as reference E-2, Appendix E. 2. A .report was also made on this incident to the National Response Center on November 5, 1981. VAB.0001113178 -* QUESTION 2; FOR EACH DISCHARGE DESCRIBED IN RESPONSE TO III.A.l., A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VC TO THE ATMOSPHERE, INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VC EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. Response: The relevant corrective steps which were taken before and during each discharge to prevent and/or minimize the discharge are discussed in the references provided under Question 1. The information provided below and documented in the appropriate references is additional information relevant to corrective actions taken before, between, and after the September 16, 1981 and November 5, 1981 incidents to prevent VCM releases caused by overpressuring of the VCM filter pak. FILTER PAK VAPOR EQUALIZATION LINE Prior to September 16, 1981, this filter was vented to the fresh VCM receiver with a vapor equalization line. The line was designed to be open at all times that the filter was in service so that the filter could not be overpressured. The rupture disc and relief valve were provided only as a safety device to guard against the remote possibility that the equalization line could become plugged or otherwise closed off. The September 16, 1981 incident occurred when this equalization line became plugged with PVC polymer. Subsequent to the September 16, 1981 release, the standard operating procedure was changed to include a daily check of the filter pak pressures. This item was added to the reactor "A" checklist on September 23, 1981. Equalization of the pressure between the VCM filter pak and the VCM receiver after the charge is complete indicates that the equalization line must be open. This pressure check procedure was discontinued after the relief valve discharge was piped to a containment system. It should be noted that the equalization line was open when the November 5, 1981 release occurred. It is believed that the November 5 release occurred when the VCM charge pump ran blocked in for a short period of time. The equalization line was too small to relieve all the pressure in the system. A VAB.0001113179 'T 4 - ! n s' FILTER PAK RELIEF VALVE CONTAINMENT A plan was developed after the September 16, 1981 release to tie the discharge of the filter pak relief valve in to the new batch water stripper which normally operate at a pressure below 10 psig. The materials necessary to make this tie-in were ordered on October 15th. Because of the time required to procure the necessary relief valves and other material required to make this tie-in, the installation had not been done when the November 5, 1981 discharge occurred. On November 6, 1981 a temporary tie-in was made from the filter pak relief valve discharge to the recovered VCM receiver. This tie-in ensured that any future overpressuring of the filter would not result in an atmospheric VCM release from the relief valve. The permanent blowdown system described above to contain the filter pak relief valve discharges was completed as soon as practicable after the valves and other material were received. It should be noted that this tie-in required a unit shutdown. Also because of problems that can occur during freezing weather, the plant operating philosophy is to not take planned unit shutdowns during the cold weather months. This system was operational in February of 1982. This should provide a good solution to the problem since the batch water stripper normally operates at a pressure below 10 psig. The tie-in line should also not be subject to plugging since it does not normally have a high VCM concentration. This installation is believed to have solved the problem which led to these discharges. QUESTION 4: FOR EACH VC DISCHARGE IDENTIFIED IN III.A.l. AS BEING CAUSED, IN WHOLE OR IN PART BY EQUIPMENT MALFUNCTION OR DEFECT, THE FOLLOWING INFORMATION: a) DESCRIPTION IN DETAIL OF THE CAUSE OF THE MALFUNCTION OR DEFECT WHICH RESULTED IN THE VCM DISCHARGE. b) DESCRIPTION OF THE APPLICABLE INSPECTION AND MAINTENANCE PROCEDURES FOR THE EQUIPMENT THAT MALFUNCTIONED AND IDENTIFICATION OF THE FREQUENCY OF INSPECTION REQUIRED BY SUCH PROCEDURE. IF THE ACTUAL INSPECTION/MAINTENANCE OF THE AFFECTED EQUIPMENT IS NOT CONSISTENT WITH THE REQUIRED PROCEDURE, DESCRIBE THE INCONSISTENCY AND EXPLAIN WHY. A 'T H k I . a a; i -v VAB.0001113180 h Response: 9/16/81 Release There is no indication that any equipment malfunction or defect contributed to this release. 11/5/81 Release The direct cause of this release was the overpressuring of the filter pak during VCM charge which resulted in bursting the rupture disc and lifting of the relief valve which discharged to the atmosphere. Mechanical malfunctions lead to the overpressuring of the filter pak. Overpressuring of the filter pak is believed to have been the result of a sequence of events which was not foreseeable when the system was installed. The VCM charge was controlled by a MODICON programmable controller which was programmed to start the charge pump when the pump discharge valve was opened. In this case, the charging operation proceeded normally until charge was complete and the signal was given to close the pump discharge valve and shutdown the pump. It appears that the problem started when the pump discharge valve did not close properly. A signal would normally be provided to the programmable controller at that point indicating that the valve was closed. In this case, a closed signal from the valve was apparently not received by the programmable controller. The programmable controller then restarted the pump since its information indicated that the valve was open. Since the other valves in the charge line were still closed, the pump started up dead headed and overpressured the filter. Overpressuring of the filter was only momentary since as soon as the pump started up the operator shut it down. Thus the conclusion is that the problem was caused by a malfunction of the valve or of the switch that indicates that the valve is closed. Immediately after the incident, additional interlocks were programmed into the controller to prevent the VCM pump from starting up if any of the valves in the pump discharge line are closed. A <1 1 H| I t VAB.0001113181 DM iii i I I h i i i r i i i4 j i. -tah >t1 1I- j >THT- If i*t i i * VAB.OOOl 113182 RELEASE 1 0 /1 6 /8 1 VCM RELEASE WHICH OCCURRED OCTOBER 16, 1981 FROM THE TANK FARM SURGE DRUM Summary: On October 16, 1981, the relief valve on the tank farm surge drum, T-201, discharged. The discharge occurred due to the unloading compressors being run in a "blocked-in" condition from operator error. The compressors then relieved into the surge drum, T-201, which became overpressured. The corrective action taken as a result of this discharge includes operator training, revised operating procedures and installation of appropriate instrumentation. These actions are considered to be adequate because no further releases have occurred. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATD CAUSES(S). Response: A description of the nature and cause of the discharge was sent to Region IV of the EPA in a letter dated October 26, 1981 which is attached as reference F-l, Appendix F. QUESTION 2: FOR EACH DISCHARGE DESCRIBED IN RESPONSE TO III.A.1. A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VC TO THE ATMOSPHERE INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VCM EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. Response: Relevant corrective action taken to reduce the possibility of this type incident recurring is discussed below: 1. Operators were reinstructed on the proper valves to have open and closed at the tank farm during the unloading operation. The operators were also instructed in the proper use of a checklist to be followed at each unloading. The checklist shows the proper valve arrangement that the operators must have in order to not run the compressors in a blocked-in situation. The training record and the checklist are attached to the letter from P. E. Markey dated November 17, 1981, which documents corrective action taken. This letter is attached as reference F-2, Appendix F. 2. A two-day disciplinary lay-off for "inattention to operations" was given to the yard operator who improperly lined up the valving at the tank farm. The letter of suspension is attached to the corrective action step letter from P. E. Markey dated November 17, 1981. (reference F-2, Appendix F) October 16, 1981 CONFIDENTIAL QUESTION 2: Continued 3. Since the cause of the release was the compressors running in a "blocked-in" condition which caused the compressor relief valves to relieve to the surge vessel, T-201, a pressure switch was in stalled in T--201. When the pressure in T-201 reaches 100 psig, the pressure switch will shut down the compressors. This pressure is well below the T--201 relief valve set pressure of 150 psig. This action was completed and tested by November 13, 1981 as docu mented by the attached letter from P. E. Markey dated November 17, 1981.(Reference F-2, Appendix F) 4. As an additional measure, high temperature shutdown switches were installed on each compressor. If a compressor runs in a "blockedin" condition, the temperature of the gas and the compressor will increase greatly. These temperature switches with associated in strumentation were installed in late 1981. The sensors were attached to the compressor with an epoxy compound. Due to vibra tion, this proved to be an unsatisfactory long term solution as the sensors came loose after a period of time. The compressors were then machined with a special slot for the sensors. This work was completed in late 1983. VAB.0001113184 ii pa m co m N> . i VAB.0001113185 VCM RELEASE FROM THE RECOVERED VCM CONDENSER RELIEF VALVE, REACTOR MODULE MO, 2 ON JANUARY 11, 1982 CONFIDENTIAL Summary: This release was caused by overpressuring of the recovered VCM condenser during an attempt to thaw it out with steam. Operating practices were changed to reduce the possibility of reoccurance. QUESTION Is FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSE(S). Response; EPA Region IV has previously been provided with a full description of the January 11, 1982 incident. This information has been provided on the following dates in the documents noted below. 1. The incident was reported to the National Response Center on Janu ary 12, 1982. 2. A report on the discharge was filed with the Region IV EPA in a letter to Charles R. Jeter dated January 14, 1982 which is attached as reference G-l, Appendix G. QUESTION 2; FOR EACH DISCHARGE DESCRIBED [IN RESPONSE TO III.A.l.] A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VCM TO THE ATMOSPHERE INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VC EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. Response; The documents noted in response to Question 1 above describe measures taken before or during the incident to prevent or minimize the release of VCM to the atmosphere. No further information is available. The measures taken at that time to prevent future incidents of this type are believed to be effective since no further incidents of this type have occurred. VAB.0001113186 .,11 ll ................IIhiII mmm mm I i i 4: j: I T: 1 I J 4I I ! ii r r Ii i i i* i 1 l i 4 i r l t l i I I ' Lh I * It I I j: hr i; i i> i i- * i: j. . i t r j r i VAB.OOOl 113187 CONFIDENTIAL VCM RELEASE FROM THE RECOVERED VCH CONDENSER RELIEF VALVES REACTOR MODULE HO. 2 ON JUNE 10, 1982 Simwu>ry i This release was caused by a plugged liquid VCM discharge line. Instru mentation has now been installed which will alert the operator to a high pressure In the VCM recovery system before a VCM release from these relief valves can occur. This Instrumentation is believed to be the best means to prevent a repetition of this Incident. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VCM") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSE(S). Response; The Mississippi Department of Natural Resources has previously been pro vided with a full description of the June 10, 1982 incident. This informa tion has been provided on the following dates in the documents noted below. 1. The incident was reported to the National Response Center on June 10, 1982. 2. A report on the discharge was filed with the Mississippi Department of Natural Resources In a letter to Jerry B. Banks, Bureau of Pollution Control, dated June 18, 1982 which is attached as reference H-l, Appendix H. A QUESTION 2: FOR EACH DISCHARGE DESCRIBED [IN RESPONSE TO III.A.l.] A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VCM TO THE ATMOSPHERE INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VC EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. r VAB.0001113188 H+HJ-l I-.......... rmn il*IIIMIITWHIilHlwnnfHiiiwwii:>HI|>NllliiMilll*lT Tm^HFfflWTMHeTH*lwwH""n-lll|rn*l*ii|Pl>....... .^friFi+M.- UII ri II .............................................................................................................................................. mi II1111II IIWHHip^WHWiwrwwiiPl ........................... :j i rrmniTi-PT-m H 1,1 VpI 1 Question 2; (Continued) Response: The documents noted in response to Question 1 above describe measures taken before or during the incident to prevent or minimize the release of VCM to the atmosphere. The information provided below is additional information relevant to corrective actions taken before and after the June 10, 1982 Incident to prevent any VCM release caused by a plugged recovered VCM line. VCM POLYMERIZATION INHIBITOR The polymerization of VCM in the recovered VCM system is minimized by adding a Hydroquinone-water solution to the VCM compressor sealwater. This system was in place before the June 10, 1982 incident. Experience has shown that this Inhibitor is effective in reducing the polymerization rate and that plugged VCM lines are much less frequent than they were before Hydroqulnone injection was practiced. The practice of hydroquinone in jection continues to date. INSTRUMENTATION TO DETECT HIGH PRESSURE IN THE VCM RECOVERY SYSTEM Prior to the June 10, 1982 Incident, the pressure instrumentation in the recovered VCM system for both reactor modules consisted of local pressure gauges. The purpose of these gauges is to permit troubleshooting of the recovery unit when problems develop. Polymer buildup in the liquid VCM line between the condenser and receiver can be detected by reading the pressures at the two ends of the line at a time when the VCM flow is high. A greater than normal difference between these pressures is an Indication that buildup is probably occurring. These pressure checks were added to the "A" operators checklist after the June 10, 1982 incident to provide advance warning of a developing problem. Plans were also developed at the same time to install a pressure recorder and pressure alarm on the discharge of the VCM compressors. These Instruments were operational in December 1983. The new Instruments allow the panel operator to monitor the condenser pressure and provide a record of any pressure increase. Cleaning of the line will be scheduled before the blockage is complete whether the increased pressure drop was detected by the pressure gauge readings or the new pressure recorder. However, since June 1982, Increased pressure drop in this part of the system has not occurred. e VAB.0001113189 RELEASES 1 /1 9 /8 0 , 1 /1 3 /8 3 r 1. i i ( 5 1 S i i i i j r VAB.OOOl 113190 VCM RELEASES WHICH OCCURRED JANUARY 19, 1980 AND JANUARY 13, 1983 DUE TO HYDROSTATICALLY OVERFILLED REACTOR CONDITIONS Summary! On January 19, 1980 and January 13, 1983, the plant experienced emergency relief valve discharges on PVC reactors due to hydrostatically overfilled conditions* Response to relevant questions Is provided on both discharges In this section since the Incidents are closely interrelated. The corrective action taken since these discharges Include changes In procedures, operator training. Installation of additional Instrumentation and additional reactor charge Interlocks. These actions are believed to be adequate as no further releases have occured. QUESTION 1: FOR EACH RELIEF VALVE OR MANUAL VENT VALVE DISCHARGE OF VINYL CHLORIDE ("VC") WHICH OCCURRED FOR THE PERIOD FROM JANUARY 1, 1979 TO JANUARY 1, 1985, A DESCRIPTION OF THE NATURE AND CAUSE OF THE DISCHARGE, INCLUDING PRIMARY AND/OR RELATED CAUSES(S). Response: Descriptions giving the nature and cause of the discharges have been submitted and are listed below: 1. The Initial discharge report for the January 19, 1980 release was submitted to EPA Region IV in a letter dated January 24, 1980 which is attached as reference C-l, Appendix C. 2. On March 4, 1980 a copy of the reactor strip chart for the January 19, 1980 release was sent to Mr. Wayne Aronson, EPA Region IV, per his phone request which is attached as reference C-2, Appendix C. 3. The initial discharge report for the January 13, 1983 release was submitted to the Mississippi Bureau of Pollution Control in a letter dated January 21, 1983 and is attached as reference C-3, Appendix C. Subsequent to the initial discharge report submitted to EPA on January 24, 1980 concerning the January 19, 1980 release, it was determined that the agitator amperage indicator on D-400 reactor, the reactor on which the release occurred, was not functioning properly at the time of the release. A more detailed description of the January 19, 1980 release and the effect of the agitator amperage Indicator not working properly follows. All reactors were equipped with agitator amperage indicators at the time of the . These amperage Indicators show different amperages for different levels of liquid in the reactors. Although the differences in amperage cannot be used to accurately measure the level Inside the reactor. It is a good indication of the presence or absence of material In the reactor. When a reactor contains material, amperage will be higher than when it Is empty and it should return to approximately the same amperage after the reactor is empty again. However, the amperage indicator for reactor D-400 was not functioning correctly at the time of the VAB.0001113191 CONFIDENTIAL The reactor batch previous to the one on which the release occurred took a little longer than usual to dump. This was a sign to the panel operator that there could be a small amount of plugging in the dump lines, so he proceeded cautiously with the pre-charge procedures. Upon finishing reactor rinsing after dumping the slurry, the panel operator radioed to the outside operator and asked him to check the bullseye on the line to the sewer. The bullseye is used to check flow from the reactor to determine if it is empty or if there is pluggage in the dump manifold. If the reactor is empty and there is no pluggage, there will be a large flow Indicated at the bullseye when the reactor is rinsed and this flow will almost stop when the reactor drains empty. The outside operator said that "nothing was coming out of the reactor except the usual small amount of water from reactor flushes". Two reactor flushes normally add about 0.8 gallons per minute to the reactor. The outside operator saw an amount of water flowing through the bullseye and it looked to be like the normal reactor flush water which would indicate that the reactor was empty. If there had been more or less water flowing, the outside operator would have suspected pluggage problems. The amount of water flowing through the bullseye appeared normal. The panel operator then checked the agitator amperage indicator and it was showing a little less than 100 amps. Since this was the normal amperage for this reactor at the time, the panel opertor thought the reactor was empty. The panel operator then proceeded to evacuate the reactor for the next charge. Another Indicator of liquid in the reactor is usually inability to obtain a vacuum quickly. However, evacuation proceeded normally so there was no warning of a problem. When the required vacuum was reached in the reactor, the outside operator, who reads the mercury manometer on the reactor, informed the panel operator to proceed with the charge. The first step in the charge sequence is to "break vacuum". This procedure opens up one vinyl chloride charge valve, allowing a small amount of vinyl chloride in the charge manifold to enter the reactor and pressurize the reactor to about 30 pounds per square inch gauge. The outside operator can then check the reactor for leaks and ensure rupture discs are in tact. When the VCM charge valve was opened to break vacuum, both the outside operator and the chief lead operator, who was near the reactor, noticed unusual vibration. The chief lead operator is normally the most experienced person on the shift besides the shift supervisor. The chief lead operator suspected something was wrong and went to the control room to check the amperage Indicator. The amperage indicator showed less than 100 amps. Because this was normal for this reactor and because of the other checks done previously, the chief lead operator thought everything was normal. The charge then proceeded as normal until the reactor charge was about 87% complete. The panel operator noticed that the pressure in the reactor was 110 pslg and increasing rapidly. He Immediately shut off the VCM and water charge pumps. Before the panel operator could put the reactor on the recovery system, the pressure in the reactor reached 182 pslg as indicated by the reactor strip chart. At this time, there was a momentary release through the relief valve. * VAB.0001113192 Description - continued After the release, a new amperage meter was ordered and installed on reactor D-400 agitator. On January 22, 1980 all reactor batch sheets were modified to include a space for reactor agitator pre-charge amperage. The operator must record this amperage before each batch is charged. The problem with the D-400 amperage indicator is not fully understood. However, to reduce the possibility of obtaining misleading information from these indicators, reactor empty pre-charge amperages have been established. If the panel operator does not get a reading equal to the established empty amperage, the- reactor is not to be charged until the problem is resolved. QUESTION 2 FOR EACH DISCHARGE DESCRIBED IN RESPONSE TO III.A.1 A DETAILED DESCRIPTION OF ALL RELEVANT CORRECTIVE STEPS TAKEN BEFORE AND DURING EACH DISCHARGE TO PREVENT AND/OR MINIMIZE THE RELEASE OF VC TO THE ATMOSPHERE INCLUDING, BUT NOT LIMITED TO, SUCH STEPS AS REROUTING VCM EMISSION VENTS TO A COLLECTION VESSEL OR INCINERATOR. Response: The correspondence mentioned in Question 1 above contains information on relevant preventative steps taken before and after the January 20, 1980 and the January 13, 1983 releases to minimize the release of VC to the atmosphere. The following additional Information is submitted. PREVENTIVE MEASURES I. Introduction The causes of hydrostatically overpressuring a reactor have been studied and preventive measures have been taken. Both of the incidents which have resulted in releases due to hydrostatically overfilling a reactor and other causes or incidents which have not resulted in releases but were potential causes for a release have been analyzed. The following Information shows proper preventive measures have been taken to reduce the possibility of hydrostatically overpressuring a reactor. II. Possible Hydrostatic Overpressure Conditions A. Reactor Not Completely Drained If a reactor is not completely drained, a discharge can occur be cause of hydrostatically overpressuring the reactor during the charge of the next batch. This condition resulted in releases on January 19, 1980 and January 13, 1983. Listed below are the pre ventive measures taken to reduce the possibility of a similar occurrence VAB.0001113193 1. Prior to January 19, 1980 CONFIDENTIAL Prior to January 19, 1980, the following procedures and instru- mentation were available to reduce the possibility of a vinyl chloride release due to charging a reactor that was not completely drained. a. The reactors were equipped with agitator amperage indicators. A no-load agitator amperage is an indication that the reactor is empty. The operators had been trained to check this parameter before charging. Although documentation of this specific training has not been located, the fact that checks were performed immediately prior to the January 19, 1980 release show that the operators had been taught and understood the significance of the no-load amperage ^ b. The reactor sewer line was equipped with a bullseye for watch ing flow from the reactor to the sewer. A flow larger than the normal 0.8 gallons per minute flow would be an indication that the piping is partially plugged. A flow lower than normal or no flow would indicate that piping is likely plugged. 2. Between January 19, 1980 and January 13, 1983 a. Reactor emptying procedures were issued on January 21, 1980. These are attached as references C-4 and C-5, Appendix C. These procedures specify checking the reactor amperage before charge, posting no-load amperage for each reactor and rinsing through the reactor with flow that is not hindered by pluggage in the lines. Specific training on these procedures was given to the operators in January, 1980. Plant records showing that formal training was given are attached as references C-6 and C-7, Appendix C. b. The reactor batch sheets were revised on January 22, 1980 to include the pre-charge agitator amperage. The no-load agitator amperage was established for each reactor. Opera tors were instructed not to charge reactors unless there is a no-load amperage showing for the reactor agitator. c. The amperage indicator that read incorrect no-load amperage * on reactor D-400 was replaced. d. Reactor agitator amperage recorders were installed for all reactors. This recording of amperage helps operators see trends in the agitator amperage and spot possible problems. e. Remote controlled cameras were installed to enable the panel operator to also inspect the bullseye to check that the reactor is empty according to the procedure. f. Reactor emptying procedures were again reemphasized to vinyl employees on January 31, 1981. Plant records are attached as references C--8 and C--9, Appendix C. The importance of following procedures to make sure reactors are empty was VAB.0001113194 m Since Janua 1983 a. An Interlock was added to the reactors that will not allow automatic reactor charge if the agitator amperage is above a certain setting or below a certain setting. This inter lock reduces the possibility of charging the reactor with high amperage that would Indicate liquid in the reactor or charging the reactor with the agitator off. b. Robert E. (Bob) Morgan was the panel operator whose failure to follow established procedures caused the January 13, 1983 relief valve discharge. The specific errors that were made are as follows: 1) The low agitator amperage panel alarm was ignored during rinse of the reactor. 2) The proper procedure for checking for an empty reactor was not followed. According to the reactor emptying procedures issued January 21, 1980, the operator should have checked for water being able to flow through the reactor. The operator checked the manifold below the reactor for pluggage. This procedure did not check for pluggage in the reactor or in the sewer piping. 3) The panel operator recorded Incorrect information on the reactor batch sheet. Instead of checking the amperage on the reactor before recording the amperage, the panel operator recorded what the pre-charge amperage should have been. A copy of the batch sheet and the amperage recording are attached as references C-10 and C-ll, Appendix C. On the amperage strip chart, the reading on the right is the dumping of the previous batch. Note that the amperage starts at 54-59 amps (5.4 x 10, etc.) and drops as the batch is dumped to about 32 amps. Then the amperage drops to zero indi cating that the agitator is off. When the agitator is again turned on, the reading was 53-58 amps. The panel operator turned on the agitator during evacuation before the charge started, but failed to look at the indicator or the recorder. As a result of these errors. Bob Morgan, was given a two-day disciplinary suspension. Documentation of this suspension is attached as reference C-12, Appendix C. It should also be noted that Bob Morgan had been trained in the proper emptying procedures. Bob Morgan signed his name on the training sheet for training given January 27, 1980 which is attached as reference C-7, Appendix C. A VAB.0001113195 CONFIDENTIAL A It should also be noted that recording of Incorrect Information on batch sheets receives appropriate attention. One such Incident Involving recording erroneous amperage Is attached as reference C-13, Appendix C. A written warning to panel operator J. D. Sims was given In a letter dated July 13, 1981. c. On January 25, 1983, the empty reactor checkout procedures were again reemphasized to all vinyl operators. A copy of this memo Is attached as reference 014, Appendix C. B. Catalyst Injection Water Overpressure Since the catalyst Injection water could hydrostatically overfill a reactor, measures have been taken to reduce the possibility of a re lief valve discharge for this reason. These measures are listed be low: 1. Training has been given to operators stressing the Importance of only adding the proper amount of catalyst water to the reactor. Operators have also been Instructed they must be In the Immediate area when catalyst flush water Is being added to the reactor. This procedure cannot take place unattended. If for some reason the operator must leave the area, the flow of flush water must be stopped. Formal and documented training of this fact has been given on at least seven different occasions. Training records are attached as references C-15 through C-21. More training was no doubt given, but these are the only documented specific training sessions for which records are still available. However, In extensive training given to outside vinyl operators In 1982, this fact was stressed as noted on Question 2 of the attached operator quiz as referenced C-22, Appendix C. The note at the top of the July 8, 1982 training session shows the type of attention given by management to environmental programs. The note from the plant manager, John Friend, asks the vinyl super intendent, Fete Markey, If this training has already been given. The reply by Pete Markey is that the shift supervisors are going through the training again to emphasize its importance. 2. When the catalyst injection bombs are pressurized, a light is illuminated in the control room so that the panel operator can tell that there is water entering the reactor. If the catalyst injec tion bomb remains pressurized due to water flowing Into the reactor for a specific length of time, an alarm will sound In the control room. The panel operator can then communicate with the vinyl outside operator to stop the flow of water into the reactor. VAB.0001113196 C. Excessive Flush Water Rates There are two small constant water flushes that enter the reactor. The agitator lip seal flush water and the spray nozzle flush. These com bined flows amount to about 0.8 gallons per minute. Because of the small volume of these streams and the large capacity of the reactor vapor space, it is unlikely that these flushes would overpressure a reactor. To reduce the possibility of these flushes causing problems, the following measures are taken: 1) If the reactor must contain liquid or slurry for any long period of time, the flow rates to these flushes are reduced to a minimum. 2) Under normal circumstances, the flush water is checked every batch to make sure they are at the proper settings. There is a flow meter available at each flush to facilitate monitoring. D. Reactor Overcharge If the reactor was overfilled due to excessive charge ingredients, a relief valve discharge could occur. In order to reduce the possibility of this occurrence, the following measures are taken: Since the product quality is highly dependent on following a specific formula for each product, careful attention is given to the addition of material to the reactor. The addition of major volume ingredients is controlled by a programmable con troller which starts and stops pumps and opens and closes valves. Dual flow meters exist for the water and the vinyl chloride mono mer so that if one meter were to malfuntion, another meter would be available for use. These flow meters have readouts in the control room. During charge, the panel operator constantly monitor the progress of the charge since addition of the proper type and and overfilling 2) There is an interlock which will not allow the charge of a larger reactor volume in a smaller reactor. The larger reactor formula cannot be entered in the charge mode for a smaller reactor. 3) There is also an Interlock that limits the total charge quantity to each reactor. The programmable controller will prevent charge to a specific reactor if it exceeds a preset volume. VAB.0001113197 t r VAB.0001113198 A QUESTION 3; A DATED COPY OF ALL PVC PLANT DOCUMENTS REGARDING (1) STANDARD PROCESS OPERATING AND MAINTENANCE (O&M) PROCEDURES, (2) EQUIPMENT TESTING AND INSTALLATION PRO CEDURES, AND (3) TRAINING MANUALS (OR OTHER INSTRUCTIONS) THAT HAVE BEEN MADE AND/OR EMPLOYED AT THE PVC PLANT SINCE OCTOBER 21, 1976, AND WHICH RELATE TO PREVENTION OF RELIEF VALVE OR MANUAL VENT VALVE DISCHARGES OF VC. Response(1): The standard operating procedures are attached In Appendix I. The reference identification for each procedure is as listed. REFERENCE NUMBER PROCEDURE 1-2 Operating Manual for the Old Module. This manual was prepared by the vinyl operations engineer in early 1982, but was not dated at that time. This manual is an update of one Issued March 16, 1979. 1-3 Operating Manual for 745 Reactors dated October 1981 1-4 Operating Procedure for the Refrigeration System on the RVCM Receivers dated August 1, 1976 1-5 Incinerator Training Manual dated June 21, 1978 1-6 Operating Procedure for the Deltech Instrument Air Dryer dated March 17, 1980 1-7 Operating Manual for the Batch Water Stripper dated June 19, 1981 1-8 Operating Manual for the AMS Emergency Kill System dated September 1, 1981 1-9 Operating Manual for the VCM Tank Farm dated June 3, 1983 1-10 Operating Procedures for the Initiator Injection Pot Replacement dated March 21, 1983 -44- VAB.0001113199 Question 3: (Continued) ffiNFIDENTIAL 1-11 Vinyl Winterization and Operations Program. 1983-1984 1-12 Operating Manual for Emergency Cooling to the Reactors dated February 23, 1984 1-13 BME Addition System Operating Manual dated May 15, 1984 1-14 Revised Operating Procedure for the Emission Recovery System dated July 25, 1984 1-15 Operating Manual for the Pall Instrument Air Dryer dated October 18, 1984 1-16 Power Failure Procedure dated April 30, 1985 The maintenance procedures are attached in Appendix B and I* The reference identification for each procedure is as listed. REFERENCE NUMBER PROCEDURE B-27 Rupture Disc/Relief Valve Practices Manual dated September 26, 1983 1-17 Checklist for Boiler Interlock & Instrument Test Log 1-18 Checklist for Emergency AMS Kill System 1-19 Checklist for Incinerator Interlock & Instrument Test Log 1-20 Checklist for Reactor Calibration 1-21 Standard Operating Procedure for Safety Disc Change and Testing dated August 20, 1983 1-22 Standard Operating Procedure for Propane Vaporizer Start-Up dated March 30, 1981 Question 3:(Continued) Response (2); The equipment testing and installation procedures are attached as separate documents. These documents are as follows: LIST OF DOCUMENTS Conoco Engineering Standards, Volume 1 Conoco Engineering Standards, Volume 2 Conoco Piping Specifications and Valve Catalog Conoco Engineering Specifications Conoco Standard Reference Material These attached standards and specifications are updated versions of slmillar documents which were used for engineering, installations and testing of equipment at the Aberdeen PVC plant. These documents are included to show that it is Conoco*s and Vista's policy to use good engineering practice. Exceptions to the standards are allowed only following approval of the Engineering Department. VAB.0001113201 L Response (3): The prevention of relief valve discharges is accomplished through the proper design, operation and malntainence of the systems in vinyl chloride service. Training manuals specific to the prevention of relief valve discharges are not used for this purpose. Rather, the operators must have an overall understanding of the total process in order to prevent discharges. This under standing is effected by the training program described below. A new hire enters the plant in the General Helper classification. In this classification the employee may work in any of the production areas depending on overall manpower requirements. The Aberdeen plant consists of four major production areas which are polyvinyl chloride, dry blend, compound and plasticizer. The' NESHAP for vinyl chloride covers only the operation of the polyvinyl chloride production area. Normal progression in the polyvinyl chloride production area is illustrated below: Chief Lead Operator A Lead Operator (Control Panel Operator) A A-Operator (Outside Operator) A General Helper A Outside Hire -47- VAB.0001113202 Response (3): Page 2 CONFIDENTIAL A A General Helper working In the polyvinyl chloride production area does not perform operator functions until he starts his training to become an A-Operator. Prior to starting the training program the General Helper does such tasks as bagging PVC resin, cleaning dryers, changing dust collector socks, pumping pond resin and being a fire watch. The current training program in the polyvinyl chloride area consists of on-the-job training, training by supervision and testing. It is the plants desire to have operators trained such that they are qualified to perform the duties of the next higher position. This is done to assure qualified personnel are available to provide coverage for vacations, sickness. National Guard Duty etc. When an employee is selected to train up, the first step is an orientation introduction conducted by the Operations Supervisor and the Shift Supervisor. In this session the functions of the position are discussed and the importance of knowing the unit and understanding the process is stressed* Safety and environmental responsibilities are also covered in this first meeting. The trainee then starts on-the-job training with a qualified operator. The trainee works alongside the qualified operator to observe and learn the routine duties and responsibilities required to satisfactorily perform the job. The on-the-job training is conducted for approximately 300 hours. During the on-the-job training period the progress of the trainee is monitored by the Chief Lead Operator and the Shift Supervisor. Also during the on-the-job training period the trainee is given a list of questions which he studies and must answer verbally and in writing to successfully complete the training. Copies of the current questions used for the A-Operator and Lead Operator positions are attached as reference 1-23 and 1-24, Appendix I. The employee must also demonstrate to the Shift Supervisor that he/she has learned all the duties of the new position. This is accomplished with the use of check lists. A copy of the current check list for the A-Operator position is attached as reference 1-25, Appendix I. Upon approval by the Shift Supervisor and the Operations Supervisor the employee is then officially deemed to be qualified for the job and at this time is then allowed to perform the job on his/her own. In case there is an open position which needs to be filled and the person selected is not completely trained, the training program described above is used to qualify the person for the new position. The normal progression of a General Helper is to A-Operator to Lead Operator and then to Chief Lead Operator. The Shift Supervisors observe and train the personnel on their shifts on a continuing basis. The normal make up of a shift is the Shift Supervisor, Chief Lead Operator, three Lead Operators and four A-Operators. Since the number of operators on a shift is quite small, the Shift Supervisor personally checks their performance and corrects any problems if they should occur. -48- VAB.0001113203 Response (3): Several possibilities exist during the training program. If the operator In training decides for some reason he does not want to continue the training for the next level, he can return to his previous position. If after the 300 hours of on-the-job training the Shift Supervisor and the Operations Supervisor consider the employee does not have the ability to master the new position, the employee returns to his/her previous position. If the Shift Supervisor and the Operations Supervisor consider the trainee is making progress and learning the new position but is not fully capable to handle the new job after the 300 hours of on-the-job training, more on-the-job training is continued until the person is fully trained to do the new position on his/her own. If an employee who has qualified for a new position and is working the new job on his own, feels the job is too difficult for him to handle in a satisfactory fashion the employee may with the Company's consent bid down to a lower rated job. Also the Company uses disciplinary action in unsatisfactory employee performance. The disciplinary result in disqualification or termination if the performance does not return to standard. cases of action may employee's A training program consisting of on-the-job training and training by supervision has been in use in the polyvinyl chloride plant since the early 1970's. As a result of a plant expansion in 1982 four additional Lead Operators and four additional A-Operators were required to fill increased staffing requirements. The training consisted of classroom training, on-the-job training and training by supervision. Seven people started the training to be Lead Operators and two choose to remain A-Operators because the Lead Operator position was too difficult for them to master. Five people started the A-Operator training. One of these was terminated by the Company for violating plant rules before his training was complete. Since this large group was trained, three additional A-Operators and two Lead Operators have been trained with a similar program which now includes a written test. When major modifications or new equipment are added to the process, operator training is conducted before the equipment is placed in service. The training includes classroom instructions in which the operating manual for the new system including drawings is covered in detail. Also field training on the new unit is conducted and a quiz is normally given at the conclusion of the training. Engineers then assist in the actual start up of the new equipment and provide the operators with on-the-job training. A VAB.0001113204 Response (3)t Page 4 In 1981 a new replacement batch water stripper system was installed in the plant. The handling of this project and the operator training are discussed below as a typical example of how projects are executed at the plant. The first step of the project was to prepare a process design for a batch water stripping system that would remove vinyl chloride from inprocess waste water to continue to meet the requirements of the NESHAP for vinyl chloride. The next step was to prepare the mechanical design for the system. Both the process design and the mechanical design were prepared in accordance with Conoco Engineering standards which are attached. After completion of the mechanical design, equipment was ordered and upon arrival the system was installed. Prior to completion of the installation an operating manual for the batch water stripping operation was prepared. Two hour training sessions were held with the vinyl operators and shift supervisors. A total of 34 people Including operators, supervisors and engineers attended the training sessions. The first portion of the training session was classroom type in which the operating manual was discussed in detail. The next part of the training session was conducted in the field where the operation of the valves, equipment, and instruments was explained. A quiz was then given to each trainee. The training session was conducted by a Senior Process Engineer. The Senior Process Engineer and Process Engineers from the plant assisted in the actual start up of the system. A copy of the operating manual and the quiz are Included as references 1-7 and 1-26, Appendix I. Extensive changes to Plant operating procedures were made between 1974 and 1977 due to efforts to reduce VCM emissions and exposure. Operators and supervisors were provided training on the new procedures and new equipment Installed to reduce VCM emissions. The operating manual for the vinyl production unit was updated in March of 1979 to include the new procedures and operation of new equipment. Training records from this time period were not required to be kept and are incomplete and therefore the actual amount of training completed can not be documented. A review of the training records, which are incomplete. Indicates there have been 127 classroom training sessions comprising 1271 manhours conducted at the plant during the period 1977 through 1984 pertaining to relief valve dlschage prevention or reduction of vinyl chloride emissions. Topics covered Include emergency procedures, power failures, VCM railcar unloading operation, reactor kill procedures, catalyst injection procedures, assuring reactor empty before the next charge, winterization, and proper handling and installation of rupture discs. Records of these training session are attached as in reference 1-27, Appendix I. VAB.0001113205