Document ZBgY6JwyyxRjy7G0QMwOEBOX8

A 5305 RESIDUAL VCM REDUCTION ABERDEEN PVC PLANT March 31, 1990 '41 Work by: 'x/c-^r R. C. Tebo Process Engineer Engineering Department 2? (. &Aui(tA and by: . v jetr -+**4?n D. E. Knittig Process Engineer Engineering Department Approved by: P. C. Schirber Senior Process Engineer Engineering Department -- 'IHnW IfI n PI *411 Hi VAB.0001097471 .TI-*rFT* To: R. R. Smith A Interoffice Communication From: Date: R. C. April o, PED, D. E. Knittig, 1990 Subj ect: 5305 RESIDUAL VCM REDUCTION During the last four months, many extensive test runs and experimental designs were done to evaluate further residual VCM in 5305 to 250 ppb. The results look very promising. Some of the major conclusions are listed below in order of priority. By addressing these conclusions, Vista will be able to produce 250 ppb RVCM 5305 for American Mirrex without lot Conclusions 1. Aberdeen's wet air system reliability should be improved. Large air pressure swings cause increased variability of blend tank and silo RVCM. Improved aeration would help RVCM removal in the blend tanks and silos and significantly reduce RVCM variability. 2. Aberdeen should evaluate the effect of changing the 5305 particle size specification. RVCM can be significantly reduced by reducing the resin particle size. By targeting a 120 micron average particle size, RVCM can be reduced by about 30%. 3. An air heater should be installed to heat the silo air to 150F. Heating the aeration air going to the silos also has a beneficial effect of reducing RVCM by 25%. 4. A system should be installed to steam heat the blend tanks to maintain the temperature at 150"F. By keeping the blend tank temperature at 150F, blend tank RVCM can be reduced by up to 50%. For immediate benefit, operators should try to keep the blend tank temperature high by rinsing the 5305 reactors with hot water. 5. Aberdeen needs to investigate ways to improve small and large blend tank agitation. Currently, a resin heel forms on the bottom of the blend tank which does not get aerated. Improved agitation would result in more uniform blend tank aeration. Blend tank RVCM would be reduced by over 20%. 6. Rotary dryer controls need to be improved. Loose control of the rotary dryers causes wide RVCM variability out of the VAB.0001097472 ripiHM i-iMirfclLNIIIII i dj Iftlri wmi m Ml These are believed to be the best ways to improve 5305 RVCM. The detailed test run report with a complete list of recommendations is attached. Recommendations A meeting will be held on April 10, 1990 at 2:00 p.m. in Aberdeen to develop a project action plan. Interested parties from PED, Aberdeen, R&D, and marketing should plan to attend. Rich Tebo Process Engineer Doug Knittig Process Engineer cc Aberdeen Houston Saddle Brook Austin: RWS \ JDO \ RBN \ JME \ SCH \ JMV EJM \ PCS \ RDM \ MCM MAE HGC File: ABD.0.900402 i|>^n n- HMHtMl i iiMlniUMi VAB.0001097473 A RESIDUAL VCM REDUCTION ABERDEEN PVC PLANT Table of Contents I. Introduction .................................................................................................................... 1 II. Recommendations ............................................................................................................... A. Major Effect ................................................................................................ B. Moderate Effect ........................................................................................ C. Unknown Effect ........................................................................................... 2 2 4 5 II. Observations/Conclusions ................................................................................... 6 IV. Discussion ............................................................................................................................. 8 A. Plant Air Reliability ........................................................................ 8 B. Resin Particle Size................ 9 C. Recipe Optimization ............................................................................. 10 l ii-T 'k D. Heated Silo Air........................................................................................ 11 E. Small Blend TankHeating..................................................................... 12 F. Blend Tank Agitation/Aeration...................................... 13 41IL1!<* G. Dryer Controls ........................................................................................... 14 H. Silo Operation Improvement .......................................................... 16 I. Silo Aeration Improvements........................................................ 16 J. Bench Scale Silo ..................................................................................... 17 K. Reactor Stripping ................................................................................... 17 L. Nested Experimental Designs ....................................................... 18 1. Blend Tank Nesting.................................................................. 19 2. Silo Nesting.................................................................................. 20 V. Summary...................................................................................................................................... 2 0 -fa *.*-* VI. Appendices A. Plant Wet Air Survey B. Resin Particle Size and Its Effect on 5305 C. Heated Silo Air D. Heated Blend Tanks E. Blend Tank Agitation/Aeration F. Dryer Controls IS G. Silo Aeration H. Reactor Stripping I. Nested Experimental Designs Hat Vk. n*t i*t-i VAB.0001097474 iurncwM A RESIDUAL VCM REDUCTION ABERDEEN PVC PLANT I. Introduction A competitor has guaranteed American Mirrex that 80% of its resin will have less than 250 ppb RVCM. Prior work has shown that Vista can produce 250 ppb or less RVCM resin only one third of the time during the winter months. The short term goal of this project is to provide American Mirrex with 250 ppb RVCM 5305 resin. The longer term goal is to consistently produce 5265 and 5305 resin with RVCM less than 250 ppb. Furthermore, the business area believes that eventually RVCM levels will be targeted down to 10 ppb in 5265 and 5305, and 100 ppb in 5385 and 5415. Historically, RVCM in Vista's resin is highly variable. I Wv There is a seasonal trend for RVCM, with RVCM being the highest during the winter months and lowest during the mid summer. Aberdeen can consistently meet the 250 ppb RVCM target during the summer months, but cannot during the winter and spring. To better illustrate, historical RVCM data is shown below. Table I Historical RVCM Data Control Limits *** -n Resin________ Avq Upper Lower Summer 1989 5265 70 ppb 120 ppb 20 ppb 5305 100 ppb 180 ppb 20 ppb Winter 1988 5265 5305 440 ppb 340 ppb 650 ppb 550 ppb 220 ppb 130 ppb For the last four months, PED and Aberdeen have been conducting extensive research into further RVCM reduction. Three major test runs were completed between November 1989 -ftiif and February 1990. Two experimental designs were done to determine the major factors of 5305 RVCM removal. This report is the culmination of that work. ............. " -'Iml, -'^"^1 iNiftii mu* if. ..............................-pril l HI i in-nrw^i^m| f i n iw^jiii^nwi -i-#hdii i IW IP^P VAB.0001097475 turn To: Disribution Interoffice Communication From: Date: Rich Tebo, Houston PED May 10, 1990 Subj ect: Status of RVCM Reduction Work The attached table shows the status of the projects that were identified during the April 10, 1990 RVCM reduction meeting in Aberdeen. The project status table will be updated every month to chart progress towards wintertime 250 ppb 5305. Rich Tebo Process Engineer Engineering Department Attachment Distribution: Aberdeen: RWS \ JDO \ RBN \ JME \ Houston: RRS \ MCM \ EJM \ PCS Saddle Brook: MAE Austin: HGC File ABD.0.900510 JMV \ DEK VAB.0001097476 American Mirrex 250 ppb 5305 Project Statius Update May 1990 1. Heated Silo Air 2. Hot Water Rinse Abd PED 3. Plant Air System Abd/PED 4. Blend Tank Agitation PED Stripping Abd + i- p' -i> M i -up m Qtr 4 90 Qtr 4 90 Qtr 3 91 Qtr 4 90 Qtr 3 91 Qtr 4 90 Project has been submitted as a fourth quarter project. Project is estimated at $10M Minor valving adjustments are required. Work will be installed under a work order or misc. PED is investigating a low pressure air system for silo and/ or blend tank aeration. Aberdeen is looking into near term means to improve reliability. Aberdeen is submitting a fourth quarter project to a) add blend tank spargers to blend tank T-407, and b) install air flow meters on all the 5305 blend tanks. PED is investigating alternate ways to improve blend tank agitation. Work is underway to study jet mixing and mechanical agitation. Aberdeen has conducted a number of test runs to improve reactor stripping. The project is budgeted for the fourth quarter. VAB.0001097477 m 6. Prepare Estimates for RVCM Reduction PED 7. Recipe Optimization R&D May 1 May 1 Sept 1 Statistical work has been done to estimate what RVCM levels will be after improvements are made. Work is complete. R&D is to contact BASF about recipe optimizations. No report yet. R&D to complete autoclave work and and to conduct plant trials for recipe optimization. No report yet. VA$t)001097478 Vista Chemical Company Pork 80 Plaza East i Saddle Brook, New Jersey 07662 Phone (201) 845-3800 A April 19, 1990 AMERICAN MIRREX CORPORATION 1389 School House Road New Castle, DE 19720 Attention : Dave Pullin -- Director, Materials Mngnt. John Blum -- Technical Services Mngr. Carmine Covias - Quality Assurance Mngr. Gen tlemen: I am sending you a list of topics that Harold Coleman and I will be prepared to discuss during our meeting scheduled for Wednesday, April 25th. We will arrive at the plant in time to begin the meeting at 10:00am and I would like to invite all the meeting attendees to join us for lunch. ur primary objective for this meeting is to receive, and provide, up-to-date feedback on a number "technical issues" relating to the quality of Vista's rigid calendering PVC resins. The feedback we receive from American Mirrex will allow us to prioritize and appropriately allocate resources to the various quality improvement projects underway at the Aberdeen plant. We hope that this meeting can also be the initiation of a "continuous feedback program" that will help us evaluate and monitor the benefits of quality improvement efforts. Harold and I are looking forward to our visit to the wonderful state of Delaware. If you have any questions or comments concerning the attached list, please give me a ca11. Very truly yours, a. Mark A. Eramo Technical Sales Representative Attachments bcc: CJM,MCM,RRS,RBN,JME * I--mi nTwum h i jw WS,EJM,HGC t VAB.0001097479 A ATTACHMENT AMERICAN MIRREX / VISTA POLYMERS TECHNICAL MEETING LIST OF TOPICS FOR DISCUSSION; I. 5305/5265 PHYSICAL PROPERTIES; A) CONTAMINATION - QC levels - Critical thin, clear film B) STREAMERS & 'VELCRO'1 C) RVCM D) COLOR E) OTHER ??? II. 5305/5265 PROCESSING CHARACTERISTICS; A) CLARITY B) GELS & "FISHEYES" C) PLATE-OUT D) GASSING E) GENERAL PROCESSING ( Fusion ) F) OTHER ??? Ill- "SPECIAL" 5305 PRODUCTION TRIAL; A) PHYSICAL PROPERTIES B) PROCESSING IV. 6015 PROJECT STATUS V. CONTINUOUS FEEDBACK PROGRAM VI. MISCELLANIOUS l VAB.0001097480 RECOMMENDATIONS j * * *1 * : , , fi ii VAB. OOO1097481 A II. Pqr;owfflUndatlons following are recommended ways tc reduce RVCM in 5305. se recommendations are classified A) those with a major effect on RVCM, B) with a RVCM, and C) those with an undetermined effect on RVCM. A. Major Effect The following is a list of recommendations which are believed to have a major effect on RVCM removal in 5305. Some of these recommendations, however, do not have an additive RVCM removal effect with others and discretion should be used in determining what recommendations are j 1. Plant Air Reliability - Aberdeen's wet air system reliability needs to be improved. Blend tank and silo aeration account for nearly all of the post reactor RVCM removal in the process. With the current systems, Aberdeen's aeration air is in very tight supply. The blend tanks, silos, and activators do not consistently get the required amount of air needed for their purposes. This results in higher final product RVCM. i 2. Resin Particle Size - Aberdeen should evaluate the effect of changing the 5305 particle size specification. RVCM has a strong correlation with resin particle size. A 20 micron decrease in targeted particle size (from 140 to 120 micron) will result in a 35% decrease in product RVCM. Customer quality specifications need to be reevaluated to determine if Aberdeen can shift its 1 4 ** fe A 3. Recipe Optimization - A detailed experimental design should be conducted in Aberdeen to determine optimum conditions for 5305. Variables should include Polivic, F-50, GH-20, initiator and kill charge amounts, variables should be evaluated to optimize CPAs, CoVs, RVCM, color and glassies. Morphology characteristics could also be studied by R&D by using their microbalance. 4. Heated Silo Air - An air heater should be installed to heat silo aeration air to 150F. Heating the silo air helps reduce residual RVCM by up to 25%, and does not have a detrimental effect on resin color or contamination. .! -.4. ><--.-*>- w eiM--i VAB.0001097482 A 5. Blend Tank Heating - Aberdeen should rinse the 5305 reactors with hot water. Increasing the small blend tank temperature improves RVCM removal in the slurry. By rinsing the reactor with hot water, blend tank temperatures could be raised by 10 to 15F. Hot water rinsing to the blend tanks can reduce slurry residual VCM by 20%. To further reduce blend tank RVCM, Aberdeen should install a blend tank steaming system. Continuously steaming the blend tanks to maintain temperature above 150"F during the 12 hour aeration period also helped reduce slurry RVCM. Steaming the blend tanks reduced slurry RVCM out of the blend tanks by 50% during January test runs. However, the effects of blend tank steaming and hot water rinsing are not additive. Steaming and hot water rinsing will not reduce RVCM beyond the 50% claimed by steaming alone. 6. Blend Tank Agitation - Aberdeen needs to investigate ways to improve small and large blend tank agitation. With the current systems, a large resin heel forms in the tanks which does not get aerated. Improved blend tank agitation and/or aeration would improve slurry blend tank RVCM. Increased agitation/aeration by increasing air rates by 50% resulted in a 50% reduction in blend tank RVCM. Further improvement should be expected if aeration air flow is increased for the large slurry blend tank T-407 7. Dryer Controls - Rotary dryer controls should be improved. The 5305 dryers (dryers 6 and 8) currently have the most difficulty controlling temperature. Improved dryer controls would reduce temperature swings in the dryers. Tight temperature control would cause RVCM variability out of the dryers to decrease. Reduced RVCM variability out of the dryer would significantly improve resin RVCM variability in the silo and consistency of the final product. i^* i t-m-tfppj-MHtf 4f HH>- IMS* m\ ihmip KHj VAB.0001097483 A B. Moderate Effect The following recommendations are believed to have a moderate effect on reducing product residual VCM. 1. Reactor Stripping - Temporary systems have been set up to allow a small flow of cooling water to the condenser during batch stripping. By supplying some cooling to the system, steam flow to the reactor will increase to maintain temperature setpoint. In reactors D-700 and D-745, there has been a 50% reduction in reactor RVCM with the new stripping procedures. However, reactor RVCM has only a moderate effect on product RVCM due to an RVCM dampening effect in the blend tanks. 2. F-50 Consistency - Aberdeen should investigate ways to decrease F-50 variability. In order to make consistently low RVCM, tight particle size control is essential. Inconsistent F-50 activity compounds the problem. 3. Aeration Indication - Blend tank sparger pluggage must be minimized. Installation of air flow meters on the large and small 5305 blend tanks would give early detection of sparger pluggage. Also, tank air flow meters would reduce RVCM variability by improving air distribution to all of the tanks. 4. silo Aeration Improvements - RVCM can be reduced by improving silo operation. Automatic air flow control is needed on every 5305 silo, and yard operators need to follow 5305 railcar unloading procedures carefully. Silo air valve indication lights need to be repaired on the dryer control panel also. 5. Silo Air Piping Changes - Piping changes to the aeration air lines at the silos would result in more consistent aeration to each silo. Because of the current piping setup to the silos, some silos get better aeration than others. Piping changes would allow a more uniform air distribution to all the silos. wpiwhpwipliw VAB.0001097484 A 6. Silo Air Pressure Control - Air pressure to the silos is now controlled by a manual valve. The manual valve does a poor job controlling air pressure to the silos. Replacing the existing inoperative pressure control valve would improve silo aeration and lower RVCM. 7. Silo Air Distribution - RVCM levels and variability can be reduced by improving the air distribution through the silos. With the present system, heavy air channeling occurs and can be witnessed by observing the large air "perks" at the top of the silo. The perks are all in the center of the silo. The silo's cone and outer edge do not get aerated as fully as the middle. C. Undetermined Effect The following recommendations have an undetermined effect on RVCM. However, it is believed that they do have merit and good potential for improving residuals. 1. Bench Scale Silo - Based on experience from the last series of test runs, it is believed that there is an optimum air flow rate. By testing a bench scale silo, many things can be learned and applied, such as optimum air flow, air temperature, and air distribution. 2. Increased Aeration of First Batch - RVCM seems to always be at its highest at the beginning of a campaign. One reason for this may be that the first batch of slurry gets a bare minimum amount of aeration in the large blend tank. It is recommended that if possible, the first batch of 5305 be held in the large blend tank for four to eight hours before the dryers are started. 3. Ethanol Based Kill - There is a theory claiming that ethanol, put in the reactor, oxidizes to form an aldehyde which is hard to strip and masks the VCM peak on the gas chromatograph. In December, ethanol was being used in the Isonox kill system for 5305. At this time, Mirrex noted that they saw a second peak on their gas chromatograph which was partially masking VCM. No evidence was found during the test runs to support or oppose the theory. However, just to be safe, it is recommended that ethanol not be used in the kill system. VAB.0001097485 CONCLUSIONS VAB. OOO1097486 A 4. Continuous Silo Baghouse Pulsing - Some of the 5305 silos had baghouse problems resulting in continuous air backflow to the filter. When the silo hatch was opened, it was noted that these silos were under a small amount of pressure. These silos also could not obtain adequate aeration flow rates and consequently had higher RVCM values. 5. Blend Tank Fume Exhausters - One observation during the RVCM test runs was that the fume exhauster was pulling air out of some blend tanks better than others. The west row of tanks seemed to have a higher vacuum on their exhaust line. It was also noted that the west row of tanks cooled off quicker than the east side tanks. These temperature differences could contribute to blend tank RVCM variability. Details about some of these recommendations can be found in the discussion portion of this text. III. Observations/Conclusions Most of the major conclusions were addressed in the recommendations section of this . Listed below are some further conclusions and observations pertinent to RVCM removal. 1. Reactor residuals have a positive effect on small blend tank pumpover residuals. To achieve the 250 ppb target, reactor RVCM should be consistently below 120 ppm. 2. Reactor particle size and RVCM do not vary during the reactor dump cycle. 3. RVCM removal in the blend tanks fits a non-Fickean diffusion curve. R&D microbalance results also show non-Fickean behavior for 5305. 5385 and 5415 show Fickean diffusion behavior. 4. For most efficient RVCM removal in the blend tanks, the small blend tanks should be aerated for at least 12 hours. Any further aeration beyond 16 hours in a cold blend tank does not significantly reduce RVCM. m "I )r . Hiiw^i Mi kxiarv m*- ****** VAB.0001097487 A 5. Of the total RVCM variation in the blend tanks, one third of it is attributable to sampling and lab testing. The remaining two thirds of the overall variation comes from the process itself. 6. Blend tank residuals have a positive effect on dryer RVCM. Lower blend tank RVCM reduces dryer and product RVCM. Large blend tank RVCM should be targeted at close to one ppm as it is fed to the centrifuge. 7. Small blend tank RVCM varied during pumpover to the large blend tank. Pumpover RVCM was at its highest during the very start and end of the pumpover. There was no particle size variation during pumpover. 8. Large blend tank T-407 has only four spargers in it. This allows for dead spots on the bottom of the tank. If a sparger were to plug, the dead spot would become very large. 9. All the blend tanks use only a fraction of the amount of air recommended for complete agitation. 10. CPAs were run on dryer particle size screen cuts to see if there were any relationship. The results gave a parabolic curve with a minimum CPA at 100 mesh. The overall composite, however, had an even lower CPA than the 100 mesh cut. This is illustrated further in Appendix B. 11. Average particle size seems to be related to the coefficient of variance. Usually there was a positive correlation. However, the correlation is not always positive. During July 1989, there was a strong negative correlation. All correlations had greater than 95% confidence limits. 12. During the test runs, there was a positive correlation between dryer moistures and dryer RVCM. Improved dryer controls giving more consistent moistures would reduce RVCM variability. 13. RVCM removal in the silos does not seem to follow a smooth removal curve. This may be explained by a. severe air channeling in the silos b. high dryer variability and mixing in the silos c. frontal chromatography IHwWtTT'IWllthi -wn^ip ^ippfcl ji.wr.rn WJlWMJJI m :Wl*1 : VAB.0001097488 DISCUSSION T VAB. OOO1097489 14. There is a significant amount of drying for 5305 in the silos. After 24 hours aeration, the resin moistures are reduced by over 50%. 15. Silo aeration removes at least 30% of the remaining RVCM out of the dryers. Silos should aerate at least 24 hours. However, no additional RVCM removal is seen with cold air after 30 hours aeration. 16. RVCM measured by the Aberdeen lab often matches the lab result reported by American Mirrex. Differences in Aberdeen's and Mirrex RVCM numbers are due to weathering and sample location. Aberdeen samples from the bottom of the car, while Mirrex often samples from the top of the car. 17. Lab RVCM testing for dry resin samples had very little variability. Slurry lab RVCM testing had a very large variability. Discussion A. Plant Air Reliability With the current systems, Aberdeen's aeration air is in very tight supply. Frequently the air demand is high enough that air flow to the silos, blend tanks, and activators is insufficient. When this occurs, RVCM greatly increases. Silos and blend tanks get minimal aeration resulting in higher RVCM. High air demand results in lower system air pressure. Low air pressure limits the efficiency of the dryer activators. Low air pressure can also lead to slower resin transfer, plugged lines, and increased sifter overflow. Furthermore, with low air pressure, the blend tank air spargers often plug. In order to adequately aerate the silos to achieve the 250 ppb target, air pressure to the silos must be maintained above 25 psig. Silo aeration air pressure is currently below 25 psig over 40% of the time. Low air pressure causes Aeration in the blend tanks is also inadequate to achieve the 250 ppb target. Blend tank air flow rates are up to 50% lower than recommended for complete agitation in RQIP. Furthermore, air flow to each of the blend tanks is variable. The blend tank air flow is controlled by throttling a manual plug valve at each individual sparger. Air distribution to each sparger is unequal, causing RVCM variability and possible sparger pluggage. I* WMI VAB.0001097490 n* * A Because the agitation is insufficient, a resin heel forms in the blend tank (especially in the large blend tank T-407). The resin in the heel does not get aerated and tends to yellow and degrade. Improved blend tank aeration as was provided for in the Blend Tank Mixing Improvement portion of RQIP would help reduce 5305 RVCM, b-color and contamination. A recently completed wet air survey for Aberdeen is attached in Appendix A. B. Resin Particle size The current target for average particle size of Vista's 5305 resin is 140 micron. However, the average 5305 railcar particle size ranges between 120 and 165 micron. Particle size distributions typically tend to get wider with coarser resin produced. Tight particle size control is important in making low RVCM 5305 resin. Lab results show a strong correlation between RVCM and average particle size in dryer and railcar samples for both summer and winter. In all cases, coarse resin tends to have higher RVCM. In order to test the theory that RVCM is higher in larger particles, several dryer samples were taken and screened into particle size cuts. Each screen cut was then tested for RVCM. In all cases, resin on the 60 and 80 mesh screens had RVCM over ten times higher than 200 mesh and pan resin. Results were duplicated using lab dried reactor dump samples. Table II shows the results of the lab tests for each particle size cut. Table II Dryer RVCM Particle Size Relationship Particle Size Cut_____________________RVCM Composite l.il-'VrTVH* 60 Mesh 4.7 80 Mesh 2.5 100 Mesh 1.4 120 Mesh 0.6 140 Mesh 0.4 200 Mesh 0.3 'P T Pan o. 1 1*. , I [PHI Jl "*" -iVWMffMWP m VAB.0001097491 -SV. . II A Using the information on RVCM content for each particle size cut, the improvement in RVCM can be predicted for any particle size distribution. Table III below shows the expected improvement for different particle size distributions. Since product RVCM is so variable, it is extremely difficult to assign absolute numbers to each particle size cut. Therefore, for the calculated RVCM, each particle size distribution is compared with the target particle size. The target particle size (140 micron) is assigned an arbitrary RVCM value of 1. Table III Particle Size Effects on Dryer RVCM Target-140 Actual Ava Micron 120 130 140 0.284 150 160 0 180 CoV 0.288 0.272 0.189 11.4% 0.281 298 23 0.357 % on 60/ Calculated 80 mesh DPI RVCM ppm RVCM 63..86%% 0.64 0.84 0.366 0.505 5.5% 0.90 1.00 0. 608 17.1% 1.20 0.811 7% 1.41 1. 487 31.4% 1.87 1.721 Attachments in appendix B also illustrate the particle size effect on RVCM. Table III shows that if the target particle size is shifted to the finer side, a significant reduction in RVCM can be expected. The table also shows that RVCM can be reduced by reducing the particle size distribution (measured by the coefficient of variance - CoV). Both lowering the target particle size and tightening the size distribution are in effect reducing the number of large, high RVCM particles. C. Recipe Optimization Since the RQIP temporary systems have been installed, there has been a lot of work done to determine how the recipe effects resin properties. However, the work done up to now concentrated on what was required to make an acceptable new technology performance resin as soon as possible. For example, Polivic charge amounts have not been changed or investigated since Aberdeen started making acceptable 5305. ku. ' 44#i-rt+4-. "IT kqr-H1 .-H hi.i!niilwp*ii i--biwppi i VAB.0001097492 p m 11 p --- -- " H A Now that Aberdeen has established a baseline quality for its performance resins, Vista has the opportunity to improve resin quality by optimizing reactor charge recipes. Recipe optimization probably would consist of a detailed experimental design to determine the effects of suspending agents, initiators, and kill agents. The 5305 resin recipe optimization should focus on improving resin color, CPA, RVCM, glassies, and CoV. Morphology characteristics should also be investigated by R&D by using their microbalance. Once 5305 recipe is optimized, should focus on optimizing 5415. D. Silo Air Heating In 1986, when Aberdeen was trying to reduce RVCM to below 2 ppm for Nibco, R&D built a bench scale model of a resin silo and performed aeration tests. One test showed that heating silo aeration air helped reduce RVCM in old technology 5265. However, since the increased reduction was minor (10%), no further work was done. One major conclusion to the bench scale tests was that the rate of diffusion was dependent on the resin temperature. Cold aeration air would cool the resin and slow the rate of diffusion. Results of the bench scale test are attached in Appendix C. During the February test runs, a 23 experimental design was completed which studied the effects of 1) silo aeration air temperature, 2) aeration time, and 3) Isonox kill amount. Results showed that air temperature and aeration time significantly reduced resin RVCM. Isonox kill amount (varied from 75 to 125 lb) had no effect on RVCM. In early January 1990, Aberdeen installed a small spare electric heater. The heater was used to heat the silo aeration air up to 100F. It was found that RVCM could be reduced by up to 25% in the silos by heating the air. To ensure 5305 quality, during the heated air tests, the lab monitored resin color, contamination, and moisture. Heated sir had no effect on resin color or contamination using heated air up to 130F. Resin moisture levels in the silo, however, were reduced. INv>l4^ m 11 VAB.0001097493 A A larger air heater should be installed to heat all the silo aeration air to 150F. The existing heater is too small and at full aeration rates can only heat the air to 80F in the winter. A larger heater with consistent temperature control would be able to reduce RVCM by at least 25% from the baseline RVCM, and also reduce final product variability. Another finding during the silo aeration tests is that silo aeration time has a significant effect on RVCM removal. For best RVCM removal, the silos should be aerated at least 24 hours. No significant RVCM removal is seen after 30 hours of aeration using cold aeration air. Results of the experimental design and aeration studies done by R&D are attached in Appendix C. E. Small Blend Tank Heating One of the experimental designs to lower product RVCM was heating the small blend tanks during the 12 hour aeration period. The 23 experimental design was done using blend tank steam heating, hot water reactor rinsing, and blend tank aeration time as the three variables for study. Steam heating the small blend tanks reduced small blend tank RVCM by 50%. Rinsing the reactors with hot water instead of cold water reduced the blend tank RVCM by 20%. As should be expected, hot water rinsing in combination with blend tank steam heating did not reduce blend tank RVCM beyond the 50% realized with steam heating alone. The benefits of steam heating and hot water rinsing are not additive. Currently, reactor D-700 and D-745 are dumped and rinsed with cold water to the small blend tanks. The slurry is then aerated for 12 hours prior to being pumped over to the large blend tank T-407 to be dried. The small blend tank temperature is about 150F after rinsing. In the winter, after aeration the tanks can cool down to as low as 100F (1356F in the summer). The cooling of the small blend tanks slows the RVCM removal. During the test runs, blend tank heating was supposed to keep the blend tanks from cooling off. By rinsing the reactor with hot water instead of cold water, the blend tank temperature after rinsing was 165F and the tank cooled down to 145*F after 12 hours aeration. Steam heating the blend tanks kept the tank temperature constant throughout the 12 hour aeration period. Steam heating kept the temperature above 160F when rinsing with hot water, and 150F when rinsing with cold water. m VMM *P VAB.0001097494 A During the blend tank heating tests, resin color and contamination were closely monitored. No increase in contamination or b~color were seen during the test. Due to practicality, only one set of blend tanks could be used for the test. The slurry was blended, so no attempt was made to determine the effect steam heating had on dryer RVCM. The small blend tank steam heating test also studied the effect of aeration time on blend tank RVCM removal. It was discovered that RVCM variability was higher with no blend tank steam heating. It was also discovered that there was effective RVCM removal in the blend tank for up to 12 hours aeration. Further aeration beyond 16 hours gave very little RVCM removal. The steam heated blend tanks had a much lower RVCM variability and a somewhat different RVCM removal curve. When the blend tanks were steamed, RVCM removal followed a log-normal curve very similar to non-Fickean diffusion curves developed by R&D. The R&D microbalance studies have shown that diffusion 5305 is non-Fickean in nature. 5415 and 5385 follow a Fickean diffusion curve. All this information implies that when the blend tanks are steamed, the rate limiting for RVCM removal in the blend tanks is VCM diffusion out of the PVC particle. When the blend tanks are not in any way, VCM stripping out of the water may be limiting the rate of RVCM removal. Appendix D contains the results of the experimental design, as well as results from the extensive sampling program done to determine blend tank RVCM removal e F. Blend Tank Agitation/Aeration With the present blend tank aeration system, air distribution is such that a large resin heel forms in the tanks. This heel does not get aerated and therefore contains high RVCM levels. The heel can degrade and cause high b-color and contamination. Improved blend tank agitation/aeration would potentially improve blend tank RVCM, b-color, and contamination. 13 VAB.0001097495 A Testing in November 1989 showed that increased blend tank air flow improves RVCM removal in the slurry. By increasing air flow rates to the blend tank by 50% , a 50% reduction in blend tank RVCM can be expected. Even further slurry RVCM improvement can be expected if air flow is increased for the large blend tank T-407 also. Aeration in all the blend tanks is inadequate. Blend tank air flow rates currently are up to 50% lower than recommended for adequate agitation in RQIP. Furthermore, air flow to each of the blend tanks is variable. The blend tank air flow is controlled by throttling a manual plug valve at each individual sparger. Air distribution to each sparger is unequal, causing RVCM variability and possible sparger pluggage. Blend tank agitation is the worst in the large blend tank T407. T-407 only has four air spargers, although according to the RQIP designs, it is supposed to have eight. Air flow rates to T-407 are well below the recommended rate for complete agitation. Improved blend tank aeration as was provided for in the Blend Tank Mixing Improvement portion of RQIP would help reduce 5305 RVCM and b-color. One of the biggest causes of high RVCM is blend tank sparger pluggage. Blend tank air spargers plug fairly frequently. However, there is no way to readily know when a air sparger is plugged. Air flow meters on the blend tank aeration line would not only show when a sparger is plugged, but would also help provide consistent air flow to all the tanks. Further information about blend tank agitation/aeration and its effect on RVCM is attached in Appendix E. G. Dryer Controls During the months of November 1989, and February 1990 extensive sampling was done to determine the how the process affects RVCM. A major finding was that the magnitude of the process variability of the dryers was one of the largest contributors to overall product variability. It was always known that dryer RVCM from day to day was highly variable. What was discovered was that dryer RVCM was highly variable from hour to hour. In fact, dryer RVCM could change up to 25% within one dryer activator cycle. Since the test variability, as determined from the silo nested experimental designs, is small, a great deal of process variability is being introduced in the dryers. 14 VAB.0001097496 mm*- HI Him anII Ml A There are some other factors that are believed to have a major effect on dryer RVCM, but have not been proven in the field. These include: a. Large Blend Tank Level b. Drying Efficiency c. Individual Dryer Performance d. Dryer Temperature It was shown during previous test runs that the slurry solids concentration changes significantly with blend tank level. Maintaining a consistent blend tank level or improving blend tank agitation would ensure a more consistent feed rate to the dryers. In the event of an external stripping column for 5305, the continuous nature of the process would help keep a consistent blend tank level. In the meantime, efforts should be focused on improving blend tank aeration. It is also believed that individual dryers have different VCM removal efficiencies. This theory could not be verified because of the high variability of the process and limited number of data points. Presently, dryers 6 and 8 are used to dry 5305. From looking at the dryer temperature strip charts, it can be concluded that dryers 6 and 8 are the least tightly controlled dryers. The outlet temperature for dryers 6 and 8 have a cyclic pattern while dryers 4, 5, and 7 have much tighter temperature control. Appendix F contains temperature strip charts for each of these dryers. Oscillating dryer temperature may explain why dryer RVCM is highly variable between each activator cycle. There are many differences between Aberdeen's rotary dryers. Dryer 4 seems to contaminate resin more rapidly than the other dryers and therefore cannot be used for 5305. Dryer 6 is the slowest dryer in Aberdeen. Often dryer 6 runs with somewhat higher moistures than the other dryers. This is probably because operators try to push as much resin through the dryer as possible to keep up with production. During the February test run, a correlation between dryer resin moistures and RVCM was revealed. Dryer resin had less RVCM than moist resin. This is not a surprise, since water removal and VCM removal in the dryers are similar processes. The attached figures in Appendix F help illustrate the dryer controls situation. VAB.0001097497 4 A H. Silo Operation improvements Silos need to always be aerated properly in order to make 250 ppb 5305. Most of the operational problems which result in high RVCM happen at the silos. There is a potential for RVCM improvement by improving silo operations. Suggested operating improvements are listed below. 1. Automatic air flow control is needed at the silos. This can be achieved by installing a 1" bypass line around the silo air control valve (identical to the set up on silos 678-680, and 682). The yard operators need to be aware of what these lines are for. 2. Silo aeration valve indication is needed in the control room so that the dryer operator can tell if the air is on. Currently, very few of the lights on the air valve on the dryer control panel are operable. 3. In order to ensure adequate 24 hour aeration, yard operators must unload the silos properly. When a silo is unloaded into a railcar, that silo needs to be completely empty before the railcar is topped off with another silo. 4. Improved communication between the dryer panel operator and the yard operator is needed. The yard operator needs to call the dryer panel operator to turn the air on or off when unloading a silo. After the yard operator is through, he needs to inform the dryer panel operator so that the air valve can be turned back on or Complete operating instructions are attached appendix G. I. Silo Aeration Improvements Silo aeration can be improved in a number of ways including: 1. Silo Air Pressure Control 2. Improved Air Distribution among the silos 3. Improved Air Distribution in the silos The items above are ranked in order of best effect on silo RVCM removal. Silo air pressure control valve would keep a constant air pressure out to the silos. Keeping a constant silo air pressure helps ensure that every silo gets adequately aerated. *r* IWIPM-THHIIIH VAB.0001097498 Currently, there is a problem with air distribution among the 5305 silos. The silos which are the farthest down on the air header (silos 678, 679, and 680) do not get as much air flow as the silos farthest upstream on the header (683 & 684). Consequently, silos 683 and 684 remove RVCM better than silos 678, 679 and 680. All silos experience severe air channeling during aeration. The center of the silo aerates much more than the outer edges or the cone. Improved air distribution would give a more uniform RVCM resin. Appendix G contains some charts which further illustrate the silo aeration problems. J. Bench Scale Silo Silo aeration currently is highly variable. Large pressure swings on the air header cause silo air flows to fluctuate widely. Because of this, it is practically impossible to conduct aeration tests in the plant. In 1986, R&D built a small bench scale silo to test aeration on old technology 5265. A similar apparatus would be helpful in determining optimum air flow rates and temperatures for 5305 RVCM removal. Results of the bench scale studies in 1986 are attached in Appendix C. During the move to Austin, R&D dismantled the silo and threw most of it away. However, according to R&D it would be very inexpensive to build another bench model. It would be desirable to construct another bench scale silo. K. Reactor Stripping In early November, it was decided to try to improve reactor stripping for 5305 in an effort to further reduce product RVCM. To do this, a system was set up that would open the condenser cooling water valve 10% when a) the reactor was in the recovery mode, and b) the slurry temperature was at least 175-F. By supplying some cooling during stripping, steam flow through the slurry would increase, resulting in lower reactor dump RVCM. A similar manual setup was tested on reactor D-744 with 5415. The results of that test showed a 42% reduction in slurry RVCM to 68 ppm. Testing lasted two months, and condenser fouling on D--744 was comparable to the other reactors. m* PMH VAB.0001097499 An automatic system was setup for reactors D-700 and D-745 in mid November, 1989. 5305 reactor residuals were reduced by 50% (185 ppm to 92 ppm) in late November through January 1990. The improved stripping procedure also significantly reduced slurry RVCM. Appendix H has some figures that better illustrate the RVCM improvement during the test. During the test, product quality was closely monitored. No resin quality deterioration was found during the test. Resin b-color did not change, and contamination was low. L. Nested Experimental Designs variability of a process to different areas. The area with the highest percentage of the total variability would be the area worked to improve. The actual percentage is not as important as the relative value in comparison to the other Two nested designs were completed in January. Both designs determined the percent variability due to the process, sampling and testing. Process variability refers to the actual variability in the process including equipment and operating variation. Sampling refers to the variation between samples and from sampler to sampler. Testing includes variation of the lab test and its technicians. In order to nest a process, samples are taken in "layers". Each time a data point is needed, the sampler takes two samples. Each of these samples is divided so that each sample is tested in the lab twice. This forms three "layers" of variability. Figure 1 shown below further illustrates this layering. Figure 1 Nested Design Layers i- Sample o o oo oo oo oo ihiibhhl IJ'rff- MP VMMB# ihmmi VAB.0001097500 1. Blend Tank Nesting Based on the blend tank nested design it was determined that the process must be improved before the test and sampling method are improved. The blend tanks were nested to discover the process, sampling, and testing variability. Two blend tank samples were taken and these samples were split into two. Four data points were obtained per sample so that the three layers can be formed. Eleven sets of blend tanks were sampled for this nesting design. The sets of two were divided into individual tanks for evaluation. The twelve hour aeration average RVCM for both modules was 1.54 ppm for all the tanks. Old module blend tanks averaged 2.16 ppm while the new module averaged 0.8 ppm. Listed below in Table IV are the assigned percent variations due to the process, sampling, and testing. Average RVCM and overall standard deviations are also included in Table IV Table IV Blend Tank Nested Design Overall Old Module New Module % Variance - Process 66.9 35.4 71.3 % Variance - Sampling % Variance - Testing 13.9 19.2 28.7 35.9 5.6 23.1 Average RVCM, ppm 1.54 2.16 0.80 Standard Deviation 1.04 0.97 0.51 In the overall and new module cases the process is the best area to concentrate the efforts to improve. The old module percentages are practically equal, so process concerns are as important as the other two areas. The process area becomes a larger contributor to the total variance as the RVCM value falls. The importance of the process is therefore magnified. There is a definite difference between old and new module blend tanks. The difference is probably caused by variations in the new reactor stripping procedures discussed earlier in this report. During the nesting experiment, the improved reactor stripping system was operating on D-745, but not on D--700. The detailed results of the blend tank nested experimental design are included in Appendix I. VAB.0001097501 WPIII SUMMARY 1 k VAB.0001097502 2. Silo Nesting Based on the silo nested design, the process needs to be improved before the lab testing and sampling methods are improved. The silos were nested to discover the process, sampling, and testing variability. Two silo samples were taken and the samples were split into two. Four data points are obtained per sample taken so that the three layers can be evaluated. Ten silos were chosen for the nested design data. The silos were sampled from the top. The average RVCM value was 0.37 ppm. Listed below in Table V are the assigned percent variations due to the process, sampling, and lab testing. Also included in Table V is the average RVCM value, and the overall standard deviation. Table V Silo Nested Design ItemPercent % Variance - Process 86.6 % Variance - Sampling 1.8 % Variance - Test 11.6 Average RVCM, ppm Standard Deviation 0.37 0.30 The percent variance in the process is the obvious area to concentrate efforts to improve. This data verifies past lab studies which showed testing to be only 10% of the overall variability of dried resin samples. The detailed results of the silo nested experimental design are included in Appendix I. V. Summary This report is the culmination of four months of intensive test run work at Aberdeen in an effort to further reduce RVCM. The recommendations presented in this report should allow Vista to produce 250 ppb 5305 resin consistently. To get to the long term target of 10 ppb RVCM, Vista will probably have to make some fundamental changes in the way it produces PVC. An experimental design for recipe optimization and a potential future external steam stripping columns are some fundamental changes that may contribute toward the long term target. 20 VAB.0001097503 *!*. I- ... t. MM 1, tv r> I ./ Art 1`iV^.tA ** fc*1-i- 1 h ' ji/m v 1-k 'v kl. k n i *<# J To* Dlerlbution From: Rich Tebo, Houston no Date: April 3, 1990 Subject: Agenda for Avarloan Mirrex 250 ppb ftVCM Hasting Attandaas: RMS, JDO, RBtf, RES, HCH, MAS# HOC, JMB, BJM, 8CH, PCS, DEK, RCT Purpose: To raviav tha atatua of tha projaot and identify notion steps Masting Dates Meeting Placet Startina Tines Sliding Tine i April 10, 1990 Aberdeen 2:00 p,n, 3:30 p.m. Review Test Run Results end Reaosnendations 4 Discussion About Test Run Findings Discuss project priority & Timing 1* Project justification 2. Project Schedule 3. Product Inventory Assign Action Steps 1 Marketing 2 RAD 3. PSD 4 Aberdeen Review Action Steps DB^/RCT All NAS/All HAS/All All RCT 30 15 0 VAB.0001097504 A JL ON MtM'FI' ^TT:r^, I'i !' 4Vl* tW- 3r ...i-h* - I I 1 Ifr nrppfnp |r--| VAB.0001097505 % Table II Coat/Benefit of RVCM Reduction Projects Item Hot Water ftlnaea 2. Improved Blend Tank Agitation 3. Blend Tank Steaming 4. Inoreaaed Blend Tank Aeration 5. Heated Slip Aeration Cost 3$ 290 100 8 % RVCM Coat/ Removal % Removed 14% 30% 24% 26% 262 $/% 973 |/% 412 |/% 32 $/% 't Table III Cost/Benefit for Combinations of Projects Combinad Cases Cimi and 8 Case 3 and 6 Gate 3 and 4 Caae2and4 Gate 2,4, and 6 Caae 3,4, and 6 Oaae 1.2. and 6 Caae 2.3.4, and Case 1,2,4, and Coet 298 390 139 144 998 64 434 164 to RVCM 46% 46% 30% 43% 91% 36% 63% 43% Coat/ to FVamowd 3/ % 648 9/ % 669 9/% 463 6/% 336 $/% 666 9/% 164 6/% 662 $/% 324 9/% A 60M reduction if required to achieve 260 ppb consistently without lot eeleotlng A 60* reduction la required If the plant air systom a improved VAB.0001097506 . 4**^ | ' 4-T ' !f ^ J ** -f - fc ' * p1, ' * .. .4 ** 4 ' -V* I fF !* 4yS Hi J.n tk -*< - -.J.. r. , f 4 * --` t ~ C . ' .. -. . . ' f . .' I . l- t. 1. .P . , ^ A *-3r i. * Table IV Projected Winter RVCM Vaiuee Base Case - Winter 1990 Railcar RVCM 0.80 Hot Rinat 1 Improve BT Steam InorMM Hot Rinat MtntfTankt Mrsnon 0.44 041 aas 035 Heat Wfi|mL|f MA La oss 048 081 046 0.14 Incrtaaa BT Aeration asi oas 0.28 oas 0.35 oss 038 0.38 080 Note: AH numbers ors sstimotss of final product ppm RVCM for 5305 in winter All estimates are +/- 25% k t* . i. * j, % *, i fi * *. i * *> 1L <44 v wm.[..ri:..aMi _ mi kiln , ; ^! Hr. fm-Kf.p 44<n ^........... jiweenr+MIrf-aa I| .ifiipixJr^w; .4^ -----------------------|- -! n'-m-H mniHawi. u*., t4#t8W|l!li <4ffP wnwiJU'iiumvn.nian iaii .n rpm*<ia-a*ti VAB.0001097507 I ... t1! Ij ^ f> . . . *i` I ' * i>K I'. . p $ fa *. t * * ' ' H fi i " r T i . 'i . * A Table VT Projected Winter RVCM Values Assuming Heated BHo Air and Hot Blend Tank Rinsing art Dona Improve BT Bass Casa 0.35________ Agitation improve BT Agttatfon 0.32 Seam Bland Tanka Inorsass BT Aeration 0.22 * ftM Steam Bland Tanks and BTAotttdon ass Btaam BMndTMcs 0,22 ass 020 ass increase BT Aeration ass n 020 ass air MM Nats; All numbsrs ars sstlmatss of final product ppm RVCM for 5305 In winter AH estimates are+/-25% Bass Cats - Wintsr 1990 RaUoar rvcm m. r !; f .7.- v A PH'kflrtIlifi4(ir4iH:ail.j ^1-, |_j ^laiinf1 June i-fm-h *'1^^^.^nH-Mni.ip^iNijw^M.MenejiP^eMiiii i*e*--!) isiimuipci-ihe* VAB.0001097508 T > O' VA * .. A '~A.1-7'\/\>j oJV" I V AV Vo *.*AC <2_v/ ,vTo: $ R. R. Smith k 1/ A - t > -4 . T1 r^ ft f EJ^ A A VAi> <1 A* vVJ From: R. C. Tebo, Houston, D. E. Knittig, Aberdeen March 5. 1990 ct Subject: 5305 RESIDUAL VCM REDUCTION During the four months, many done to Some of the results looked very promising, recommendations are listed below. runs and VCM in 5305. Some of the resulting Recommendations /A /ip- wet variability which flow swings reliability needs to be improved, blend tank and RVCM. Blend tank and silo consistently below recommended alp RVCM removal in the blend tanks and iduce RVCM variability should size shifting the micron average particle the effect of changing the 5305^ RVCM can be significantly reduced by^* Size Smaller. RVCM By ta ropf* i nn aa 11 20 S-- 4<7 should to 150F. Heating the going to the also has a RVCM bv 25%. LX In 4. A system should be installed to steam heat the blend tanks to PeO+j/0^maintain the temperature at 150"F. By keeping the blend tank p temperature at 150*F, blend tank RVCM can d CFor immediate benefit, operators should try 20% # 5. Aberdeen needs to investigate ways to improve small and large blend tank agitation. Currently, a heel forms on the bottom of!Re blend tank which does not get aerated. Improved agitation would in more uniform blend tank aeration. ^ 6. Rotary dryer controls need to be improved. Loose of the rotary wide RVCM variability out of the dryer run at a later -the best ways to improve 5305 RVCM. A ^ort with furfch Rich Tebo Process Engineer Doug Knittig Process Engineer cc: RWS \ JDO \ EJM \ RBN \ SCH \ PCS \ JME i: di --liq |i "Mtti'fr P..T. `*****%^F'` sf-nr--Wfgttl't*!'' VAB.0001097509