Document KVaxNLqQ6XZbN8ZgdDjNxLa0

1Jd ^VW IVWJ r* c. ^ fc. Interoffice CoImJnunication File EV-85|7.6! To From Michael D.| Mcjsee Date January 18|, l|j77 1 I JAN 20 1977 CONFIDENTIAL subject Trip Reporj: oj* Meeting with Akzo Representative Meetings with Mr[ Anthony Beekwilder, Chief Process Engineer for Akzo, were held 1 on November 30 and December 1 at the Lake Charles VCM Plant. | The objectives of the meetings were to evaluate the Akz<i) HCl recovery technology, discuss the operat ing performance if their installation, and in general to utilize Mr. Beektlrilcier's technical assistance on chlorinated hydrocarbon incineration and HCl recovery. Mr. Beekwilder proved to be extremely knowledgeable in this field. Certainly his visit was well Worthwhile considering the immediate appli cation for several if his recommendations. Conoco personnel Hattending the meeting were: J. H. McClilley - VCM Plant (part-time) J. R. Holetomb - VCM Plant (part-time) C. R. iar>:er - CED M. D. kcGie - PED Akzo*s incineration land HCl recovery unit is located in their VCM plant in Rotterdam, Netherlands. The capital cost of their incinerator and HCl recovery unit, excluding the tank farm and site preparation,] was $5.0-5.5 MM (1976 dollars). The unit has operated only 6 months since completion in 1972 because economics have favored dispjosall of chlorinated wastes at sea on ships equipped to incinerate the wastes. Akzo currently plans to begin operating the |unit continuously in February, 1977. Feed to the unit is a mixture of tars produced in the VCM plant and waste chlorinate^ hydrocarbon liquids from outside sources. Akzo employs the plitltetu submerged combustion technology for incineration. Thle riel recovery process was developed by Akzo and represents a .significant departure from other available recovery methods.) A process flow diagram of the incineration and HCl recovery unit is attached. The most unique aspects of their process arel 1) | a partial condenser is used to remove most of HCl from the qUenched incinerator flue gases and 2) the HCl stripper is operated at elevated pressure (37 psig) which reduces the azeotpopp composition to 18 weight percent HCl (from 20.2% HCl ait 1 atm.). Mr. Beekwilder did not think their process could be economically applied to our application because of the relatively] low HCL/H2O ratio in our combustion gases. A preliminary engineering study would be required to determine the VVC 000012203 File EV-857.6 Page 2 January 18, 1977 utilities requirements and approximate capital costs of their system. Mr. BeebwiLder estimated the approximate cost of the study at $20,000 ank about 6 weeks required for completion. Further consideration of this alternative does not appear warranted in light kf Mr. Beekwilders' recommendation that the calcium chloride process is more suited for our installation. Additional information obtained during the meeting is listed in the following sections. INCINERATION Sodium salts in| the feed liquids can cause rapid failure of the firebrick in the incinerator. Sodium carbonate forms low freezing point eutectic mixtures which vigorously attack the brick. Akzo established a maximum sodium concektrAtion of 100 ppm on contracted liquids to prevent r^pik deteriration of the brick lining. The tars from Ak?o's VCM plant contain very little sodium < 100 ppm) Water washing ol the liquids to remove the salts was attempted in the laboratory. This approach was abandoned because a stable emulsi on was formed by the wash. Installation of I sodium resistant brick in the incinerator was also rejected because of the extremely high cost. Also, sodium resistant) brick are susceptible to deterioration from thermal shdck. Akzo expects |15, 000-20,000 hours service life from the Molite fire brie k installed in their incinerator. The following spelcif ications were placed on the firebrick. CaO + MgO < 0.6 wt. % A|lumina - 72-75 wt. % Porosity < 16 % Two types of (sodium resistant brick were suggested; a fused Molite (veby low porosity), Corhart-manufactured by Corning Glhss. Gradual heatikg and cooling of the incinerator during start-up and shut-down (< 180F/hour) is recommended to avoid thermal shbek damage. The waste liquids pass through an 80 mesh (177 microns) basket-type filter before injection into the incinerator Filter elements ^re changed once per day. Smaller mesh size elements wejre tested and plugged rapidly. vvc 00001220* File EV-857.6 Page 3 January 18, 197 INCINERATION (Cdnti.nued) Burner guns are designed to produce liquid droplets below 200 microns in (diameter. No problems with burner plugging have been encountered by Akzo. One of the Nittetu installations i n Japan has experienced nozzle plugging which was atltri buted to inadequate liquids filtration. Low-alloy, centrifugal waste liquid feed pumps were originally installed. These pumps have been replaced, due to excessivk wear, with other centrifugal pumps with a single mechanical seal. Filtered liquid is used as a flush fluid. Mr. Beekwilder was not sure of the current material of honitruction, but indicated he would send this information. Reciprocating pumps were not selected because of pressure fluctuations and valve sticking dur ing shut-down. 10 Air atomization of the waste liquids was strongly recommened over stean|i atomization. One of the Nittetu installations had experienced sever corrosion of the burner tip with steam atomization. The steam hydrolyzes the chlorinated liquids to fdrm HC1 which is highly corrosive at the ele vated temperature of the burner. Polymerization of any isoprene present in the tars could also be caused by steam atomization. Water in the feed tars in concentrations above 200 ppm will also lead to burner corrosion due to hydrolysis. 11 An air atomizktibn rate of 4.3 - 5.2 SCF per pound of waste was recommended. 12. Atomization air is supplied by an oil sealed, rotating vane compressor atomization air to liquids ratio is not controlled 13 No automatic Controls are utilized on the incinerator. Combustion ai* aikd steam rates are adjusted manually to maintain the desired excess oxygen (i.e. 15-20% excess air) and chlorine levels in the combustion chamber. During start-up the ieshonse of the incinerator was too complex to allow automatic control. Once the system has reached steady-state, Beekwilder thought automatic control would be feasible. Thd line between the ample point and the oxygen analyzer ^hould be as short as possible 14. The chlorine aonc|entration leaving the incinerator has ranged from 1000-j2000 ppm (normally between 1000-1500 ppm) which is approximately 4-5 times the equilibrium value predicted by the Deacon equation. Nittetu has developed some data which shows the ratio of actual chlorine VVC 000012205 File EV-857.6 Page 4 January 18, 1977 INCINERATION (CoAtiAued) production to the theoretically production decreases as the residence time in the incinerator increases. Since the data is proprietary Nittetu information we could not obtain a copy of the correlation. The basic shape of the curve is shown below: Residence time, seconds 15. Their incinerator design incorporated an elaborate inter lock system for start-up and emergency shut-downs. The primary objectives of the system are to prevent explosive conditions iA the firebox and to limit the rate of heat ing and cooling of the incinerator. Two flame scanners were utilized? one monitors the main flame and the other is on the pilot. The start-up is aborted by the inter lock if the oildt does not ignite within four seconds. Initial startl-up attempts were thwarted by this feature. A material, dpaque to ultraviolet light, was orginally installed by erior in the flame monitor window. The inter-lock would not allow start-up because it appeared the pilot had net ignited. If the main flame is lost, the incinerator is purged with 5-7 volumes of air before re-ignition i,s attempted. The purge gas rate is 170 SCFM. Nitrogen is available as a standby purge gas because the purge air blowers are not equipped with emergency power. If the firebox temperature does not drop below 2300F the main burner can be reignited without using the pilot burner vvc 000012206 File EV-857.6 Page 5 January 18, 1977 INCINERATION (Coi|itiJiUed) 16. During shutJdoim, purge air is circulated by the wind- box, inspec^ioA windows, manholes and any other metal surfaces toiprovide cooling from back radiation from the firebrick. I 17. Cooling of the i downcomer tube between the incineration chamber and lth4 quench chamber is from splashing of acid onto the outfsicje of the tube. Normally the Nittetu sub merged combustion units can operate no lower than 60% of capacity beqaude at low-gas rates the decreased turbulence does not prcjvicte complete wetting of the tube. Akzo has circumvented this limitation by diverting high rates of air into the downcdmer. 18. The primary 'operating problem encountered with the Akzo incineration! section is incomplete combustion of small carbon particles (80-120 micron diameter) present in the feed liquid.1 ^hey are currently testing a 'gorator' pump which w|illl grind the particles to less than 30-40 microns in diameter. There is adequate residence time in the firebbx to completely combust of this diameter particle. Preliminary investigations show rapid wear of the internal gears may present problems with this approach. Tney^ are planning on testing burner nozzles with 90 and| greater spray angles to produce a short flame front and, possibly increase the combustion of soot. The original nozzles produced a 60 spray angle. 19. 20. 21. Feed liquids* to, the incinerator are blended to provide a homogeneout composition prior to pumping to incinera tion. A feed storage tank volume of 560,000 gallons provides approximately 2^7 days of residence time at the maximum feed rate (about 4 tons/hr.). The feed tank is equipped with jet mixers to keep the suspended solids in suspension. A large manhole tO provide access for clean-out was suggested. A radiation pyrometer, preferably with aluminum sheathing, is the only typ4 of temperature sensor Beekwilder thought could be succfesiful applied in this application. Mr. Beekwildir was quite interested in our incinerator design and wis Anxious to find out the eventual performance of the unit (expecially the waste heat boiler). Based on their evaluation that was made in 1970-71, he expressed some reservations on certain aspects of the system. The service life'of ;the waste heat boiler with liquids firing was one of his primary reservations. The higher chlorine VVC 000012207 File EV-857.6 Page 6 January 18, 1977 INCINERATION (Cohtiiiued) . concentrations I resulting from liquids combustion will substantially increase the corrosion rates of carbon steel in the boilir.I He cited the high corrosion rates of carbon steel by chlorine above 300F. At 450F, ignition occurs whenlpujie chlorine contacts carbon steel. Presumably the ignitioA is caused by ferric chloride formed by attack of chlorine|on carbon steel. The presence of salts in the liquids and difficulty of maintaining a constant fire box temperatures were two additional aspects of liquids firing that he believeld would have a deleterious effect on the life cf a waste heat bailer. Chlorine formation in the Nittetu installations ils suppressed by steam addition, rather than water addition specified in our design. He indicated there is a potenti al for spalling of the fire brick from contact with water n ear the injection port during normal operation, Also, if tig|ht shut off is not maintained, leaks of water would cause spa lling during shutdown. Shell had a waste heb.t boiler in chlorinated hydrocarbon liquids incineration service that was started up in 1966. The performance of the boiler was so poor that it limited the steam factor of the plant. The unit has now been shut down. Shell specified during long term incineration negotations with Akzo that they would not enter into an agreement if a waste heat boiler was utilized in the unit. General JANHYDROUS HC1 RECOVERY Akzo evaluated both the calcium chloride and the typical above the azeotrope distillation processes (as licensed by Hoescht-Uhde and Union Carbide). The above-the azeotrope distillation Processes were rejected because both the capital and operating costs were higher than for the process developed by ^kz4 The calcium chloride process normally utilized by Nittetu was rejected because at Akzo's high cost of boiler feed water quality water required for scrubb ing and the high iron content of the feed liquids (1000 ppm) (Based on information presented by Nittetu, this iron con centration would require a complete purge of the calcium chloride solutiiod every three weeks to avoid exceeding the solubility limit of iron chloride). Akzo visited three of Nittetu* s installations in Japan at VCM plants own|ed iby Kashima VCM, Mitsubishi-Monsanto and Mitsubishi Chemicals. They were not allowed to walk through the Kashima unit, however, technical discussions were held with operation s and engineering personnel at the plant site, At the time of their visits none of the units were producing vvc 00001220a File EV-8576 Page 7 January 18, 1977 ANHYDROUS HC1 RECOVERY (Continued) anhydrous HCj. Rather, the vapors from the water cooled condenser were absorber in a water scrubber to produce a 35 weight percent acid. The reason for producing a strong acid was the hidh price of 35% HCl and not any operating pro blems associated with anhydrous production. Equipment plugging was encountered only at the Mitsubishi-Monsanto installation, which was the only plant which did not use a boiler feed water quality (low TDS) scrubbing water. The river water used for scrubbing the| Hcjl contained a high sulfate ion concentration. Calcium sulfate precipitated on the packing in the stripper anc formed a scale on the reboiler tubes. Inorganic impurities in the feed acid do not interfer with the ability of calcium chloride to b|reat the azeotrope. Only when the concentration exceeds the solubility do inorganics crease any operating problems Nittetu originally installed polypropylene telleretes packing in the caustic scrubber which failed due to embrittlement from hypochlorite kttack. The packing was replaced with PVC teller etes which al^o failed from embrittlement. They are now using trayed columnt with titanium internals. The only other operating problem the Japanese companies have experienced with the recovery section is lelkage of HCl past the single mechanicals seals on centrifugal, Acid pumps. 3. The only instillation of the Union Carbide process that is purported to havd produced anhydrous HCl is B. F. Goodrich's unit at Calvert city, Kentucky. According to Beekwilder the unit was never able to manufacture anhydrous HCl. Materials of Construction 4. All of the vessells in the Akzo recovery unit are constructed of rubber-lined steel. The HCl stripping tower is also lined with twd layers of brick-acid resistant brick on the inside with calrbdn brick next to the rubber-lined walls. No corrosion plroblems have been encountered in any of the vessels. 5. Low quality rubber-lined pipe was installed through out the system except fori carbon steel lines on the anhydrous HCl service. All of phe original lined pipe had to be replaced because of holes present in the lining. A high quality rubber lined pipei was used for replacement. 6. For hot acid service, he recommended teflon lined pipe for our system. For cold acid service (< 160F) Kynar-lined pipe was recommended for diameters less than 6 inches. VVC 000012209 File EV-857.6 Page 8 January 18, 1977 ANHYDROUS HC1 RECOVERY (Continued) 7. Expansion joints between all major pieces of equipment in hot acid service was strongly recommended. HC1 Absorber .8 The adiabatic absorber contains three packed sections, each with 10 feet f 2 inch polypropylene pall rings. The minimum liqui< leading is approximately 4 gpm per square feet. Warm s< :rubbing water (149F) is feed to the top section of th< absorber to avoid fog formation which results from cold wat< ir Addition. Liquid leakage through the trapout trays is the only operating difficulty experienced with the absorber. 9. A polypropyler.fe or Haveq, sieve tray absorber would be soecifi ed in their ne xt design rather than the packed absorber, The primary ad vantage with the seive tray is they can operate at lower liqui d to gas ratios which eliminates the necessity for recirculat ioh of acid. Approximately 10 actual trays would be requi red in the absorber. Stacked, removable trays were recommende d. Cemented Haveg trays have caused cracks at the tower wjall s due to shrinkage of the trays as they cool. .10 Less than 1 pe rcent at the total HC1 formed by incineration remains in the exhaust gases from the absorber. .11 The acid leaviig bhe bottom of the absorber is at its boiling point. Forty minutes of liquid residence time is provided in th<j bottom of the absorber Pumps, Piping and Instrumentation 12. Two automatic .ici^H strength monitors were installed in the recovery system; neither of which are not being used. A Quick-fit analyzer which monitored condenser outlet acid concentration by boiling point determination was destroyed during start-up and has not been replaced. The acid strength in the top sectioili of the absorber is checked by specific gravity analyses in hand drawn samples. The continuous analyzer usingjconductivity measurement which was installed on this stream]isjnot used because it was not accurate. Plant personnel did not work on the instrument to determine if the accuraci could be improved. Although Beekwilder thouglt the anaj-yzir could be adjusted to function properly, he did not believe the high amount of maintenance needed was justified ginqe the concentration varied little after start-up. vvc 000012210 File EV-857.6 Page 9 January 19, 1977 ANHYDROUS HCl RECCjVE^Y (Continued) 13. All of the acid transfer pumps are solid epoxy construction, Each of the acid pumps are spared. For hot acid service (>165F). Eith er impervious graphite or stoneware construction was re commended. The Nittetu installations have had good perform ance from the stoneware pumps except when water hammer lis encountered the impeller is rapidly destroyed Double mechan Leal seals with a water purge is recommended. 14. Saunders membrane (pinch) control valves were recommended for acid service with a teflon or glass lined body ^nd te flon membrane. ,Uczo has experienced good performance with teflon lined hall values in HCl service. 15 Glass-lined t leri^owells with a tantalum tip were specified by Akzo and have performed satisfactorily. Hastelloy C relieves valves and rupture discs were specified in their unit. Ruptu e discs are installed on both sides of the relief valve since fchiy relieve to the HCl absorber. Caustic Scrubber 16 A spray-type austic scrubber was specified because of the problems Nitte tu Experienced with packed towers. The caustic consumption ra te is 0.1 pounds of NaOH per pound of HCl produced. App roximately 2.5 times the stoichiometric caustic reguir ed for HCl and Cl? neutralization is fed to the scrubber, The excess caustic is consumed by carbon dioxide absorbed from the flue gases. Beekwilder cautioned agai nst| attempting to partially spend a strong caustic soluti on in a scrubber. Their experience indicates that any excess caustic added will be spent by the resulting increased carbon dioxide absorption. 17 The vent gases contain less than 10 ppm chlorine. 18 Spray nozzles are constructed of titanium. 19 Rubber lining of the scrubbing tower was specified to provide corrosion protection during periods of upset in the HCl absorber. I Du ing normal operation carbon steel would be adequate fo th'e caustic service. However, rapid corrosion rates w :.ll result if large amounts of acid are carried over from the absorber. 20 Sulfite, thiosulfate and urea have been used to reduce the hypochlorites formed by scrubbing chlorine with caustic. At current prices in Europe, thiosulfate is the least expensive reducting agent. yve 000012211 File EV-857*6 Page 10 January 13, 1977 ANHYDROUS HC1 RECOVER]? (Continued) .21 Beekwilder thought porcelin Intallox Saddes was the best packing choice for the scrubber. At temperatures above 150F and pH aboive 10, silica will dissolve out of the packing. .22 Both the stripping column preheater and reboiler are impervious graphitef cylindrical exchangers. The preheater shell is teflon^^ined steel; the reboiler has a carbon steel shell. The preheater has not performed satisfactorily because of byj-pass at the baffles. A plate-type, impervious graphite prehiat^r would be specified in their next design. Great care must be taken in fitting the cylindrical blocks together to avoid leaks between the shell and tube side fluids. An impervious graphite, shell and tube reboiler is now preferred because of the lower cost of this exchanger type. The maximum shell side pressure is 175 psig from Holland manufactures which contrasts to the 75 psig maximum available in the U.S. 23 Acid cleaning of the graphite exchangers is recommended, rather than m<. chanical cleaning which may damage the exchangers 24 The overhead condensers and the product interchanger are impervious gr<aphite shell and tube exchangers. 25 The product in terchanger is designed to heat the HCl to 40F above its dew point to avoid corrosion in the compressor 26 A carbon steel air-cooled exchanger equipped for air recirculation is used for intercooling the HCl during compression. Air cooling was selected to avoid cooling of the HC1 below its dew point in the winter. 27 The overall he at transfer coefficient in the reboiler is 200-250 BTU/hr ftrF. At the outlet end of the acid cooler the overall coafficient is 20-25 BTU/hr ft2F. 28. The HCl chiller and refrigeration unit should be designed with enough spiire, capacity to accomodate upsets in the HCl condenser. Exc:esk capacity also allows anhydrous HCl production during the initial stages of start-up. 29 Ethylene glycol is used as an intermediate heat transfer fluid in the HCl chiller, Ammonia is the working fluid in the refrigeration unit. 30. None of the exchangers are spared. Replacement tubes and cylindrical exchanger sections are kept on hand. VVC 000012212 File EV-857* 6 Page 11 January 18, 1977 ANHYDROUS HCl REC OVERY (Continued) HCl Stripper 31. A total of ttaenrty feet of 2 inch, impervious graphite raschig ringls ajre included in the tower. Raschig rings, which are made from the excess length cut from exchanger tubes, were chosen because of their low cost. 32. Above 275F :ermic packing will be dissolved by dilute hydrochloric acid. 33. Two hours of feed storage capacity is provide in the bottom of the HCl condenser. 34 No problems have been encountered with using anhydrous HCl in the o nychlorination unit. 35 The maximum Allowable temperature of anhydrous HCl in contact with cafbon steel is 212F. Above this temperature ferric hloride has an appreciable vapor pressure. Since the ferric chloride film is continuously evaporating, a high corrosion rate results. 36 During start-up the column is operated at total reflux. Nitrogen is circulated through the compressor with a small purge to tihe absorber. The line to the oxychlorination unit is pudged with nitrogen and blocked in until the overhead reaches temperature and pressure. Once the out let gas temperature from the absorber reaches 5F the pro duct HCl is dry enough for compression. Approximately 1220 hours is required for the system to reach equilibrim during start-up. 37 A two stage crsntfifugal compressor manufactured by Sundyne is used by Ak5 o. The internals are 316 stainless steel, A spare comprs ssor is kept in the shop. Approximately 6 hours is required to install the spare. Reciprocating compressors are considerab ly more expensive than the centrifugal type for this serv:L ce; 38. The product HCl line is steam traced to avoid condensation during the winter. 39. The product HCl contains less than 100 ppm water. 40 Approximately `.8-24 months would be required for mechanical design and cor struction of their unit, The process design would require at least 3 months. vvc 000012213 Approved by: File 857.6 VVC 00001221^