Document dnRyOxOwMjg3Zrk945Gwvq1Lq

FUGITIVE EMISSIONS OF A PVC POLYMERISATION UNIT : EXPERIMENTAL EVALUATION Authors : MM. J.-P. BINDELLE (SOLVAY S.A., Brussels) P. DEPRET (SOLVAY S.A., Tavaux Plant, France) VCSA meeting, oct. 6th & 7th, 1993. ABSTRACT: Experimental data on fugitive emissions of a PVC plant show that fugitive VCM emissions are orders of magnitude lower than controlled emissions. THe figures are also much lower than those based on the emissions factors published in the literature for petrochemical industry. The low emission value in PVC polymerisation is probably due to the traditional high standards usedfor leakage reduction, aiming to reduce the exposure ofpersonnel to vinyl chloride. The potential reduction of emission by the possible substitution of conventional valves by bellow valves is negligible. 1. SCOPE OF THE STUDY Atmospheric emissions of petrochemical plants consist of three parts : controlled emissions : normal exhaust gases of the process, scheduled opening of equip ment,... accidental emissions : abnormal emissions of the Pressure Safety Valves (PSV) and other equipment in case of emergency fugitive emissions emissions arising from potentially leaking equipment (valves, flanges, pump seals, compressors, pressure relief valves,..., see appendix 1) Controlled emissions are rather well known. Optimisation of the process and of the operating procedures allow to keep them into acceptable limits. Safety measures and procedures reduce frequency and amount of such accidental emissions to very low values. CTL015366 Fugitive emissions are difficult, indeed even impossible to quantify in all cases. Industry applies "Codes of Good Practice" to reduce them as much as possible. They are : 2 associated with volatil organic compounds potentially present at numerous points throug hout a petrochemical manufacturing unit, randomly dispersed according to various parameters : age and type of equipment used, maintenance programs,... Measurement or estimation of the actual of fugitive emissions in global atmospheric emissions of a chemical unit is rather difficult and extremely expensive. For petrochemical industry in general some authors believe that they could represent between 30 and 70 % of the total air emissions [1,4,10], Emission factors associated with the type of equipment and the level of screening values (portable leak detectors ["sniffers"] near potentially leaking equipment) have been proposed by different authors (leak-no leak method, stratified factors, correlated factors [1,2,3,6]; see appendix 2). The range of factors is quite large (for instance 1 to 100 g/h for valve handling lights liquids [4]). The origin of such disparity is due to many factors : maintenance, conception, preventive procedures, ... [8,10]. Different studies achieved in chemical units show that conventional emission factors used can lead to large overestimations [8,13,14], Overestimated fugitive emissions will likely generate negative public perception. PVC industry, who faced with the ASL problem in 1974, took impresive measures (equipment upgrade, process improvements, special procedures, organization,...) to reduce drastically exposure of personnel to VCM, reducing at the same time controlled and fugitive emissions. For these reasons one may presume that the level of fugitive emissions is probably much lower in PVC industry than in common petrochemical industry. In 1992 SOLVAY went into an experimental study of a PVC polymerisation unit of the Tavaux PVC plant (France), in order to verify the veracity of the assumption, and to give an estimate of the fugitive emissions of the unit compared with controlled emissions. CTL015367 2. MATERIAL AND METHODS 3 2.1. Description of the studied area The studied volume is a closed polymerization room of S80 m3 located at 16 m heigth (appendix 3). It contains the top part of two 120 m3 reactors producing suspension PVC according to the closed process technology. We took advantage of this particularity to perform the measurement program. This plant has been erected at the end of the seventies using the best available technology and has been operated with skilled people applying strict operating and maintenance procedures to reduce fugitive emissions (see the "Code of good practice" proposed by ECVM [12]). The studied volume contains about 2/3 of the critical equipment (manifold, pneumatic valves, flanges, manual valves, end-line caps,...) associated with liquid VCM transfer in the unit. The air is extracted from the polymerization room by 4 air fans running constantly (3000 m3/h each). 2.1 Measurement techniques A rigid cardboard section of about 2 m lenght was adjusted in front of fan propeller to enable sampling in a pseudo isokinetic way (appendix 4). The sampling location and the actual flowrate were checked by measurement of the speed of the air lines across the section (Note : due to their quick diffusion isokinetic sampling is not requested by usual standards for volatil organic compounds gases). From the head of one reactor a known amount of the of tracer gas (SF6) was released at a constant rate and recovered regularly (vacuum bottles) at sampling location to estimate the actual flowrate of the ventilation system. Analysis of SF6 were performed on a FID CHROMPACK P9000 gas chromatograph equipped with column DB1301 of 60 m, Him thickness 1 pm). VCM was collected at each fan by a SIPIN Personal Sampler Pump SP1 (0.5 dm3/h) through adsorption cartridge (PERKIN ELMER ATD50 of 1 cm3). During the first trial, 4 different pumps using different flowrates (0.25-0.50 dm3/min) and sampling duration (20, 30 and 50 min) were used on one of the fan in order to control the sampling conditions. The mesurements were performed three times during running periods of 4 up to 6 hours (total 129 samples taken). Analysis of VCM adsorbed were achieved by chromatographic analysis on FID (Flame Ionization Detector) PERKIN ELMER ATD50 equipment (detection level of 0.5 pl/m3 for a sampling period of 8 hours, i.e. 2.5 ppb). CTL015368 The level of VCM was also monitored by the fixed system (GC FED CARLO ERBA serie 1000) normally used for the continuous control of hygiene at workplace, and by 8 passive cartridges (open end without pump) located at each 1/8 of the unit and adsorbing during the whole sampling period. 23 Bagging measurements In order to quantify the leak rate of some individual equipment, some bagging trials were achieved by isolating the equipment with a triple layer (PEHD/A1/PEHD) sheet (CLARYL, RHONE POULENC LYON). The air tight tent was purged with dry air to control any loss of pressure, then emptied by a vacuum pump, and filled again with dry air; sample was taken after a defined period of equilibrium with the tested equipment. 3. RESULTS 3.1 Flowrate and replacement rate of the extract air The air speed in the sampling section vary from 5.3 to 5.9 m/s and is not symetric. The air flowrate extracted by each fan is 3630 m3/h (average value) with standard deviation of 895 m3/h. The measured replacement rate is about 25 times/hour. The sampling was located at the point where a mean velocity was measured. 3.2. Tracer gas 94 % of the emitted tracer was identified in the measurements, 97 % at the two fans located near the reactor where the tracer was emitted (appendix 5). Concentration of SF6 decreases rapidly after the end the injection. All those facts lead to conclude that: ventilation system is very efficient and measurements on relatively short pe riods reflect the actual variations in the emissions rates. Moreover, cross contamination between the two reactors heads and samples was shown insignificant. 33. VCM measurements VCM concentration varies from 5 to 50 pl/m3 (see an example in appendix 5). The compre hensive emission computed on these values is 0.7 g/h in average or 6 kg/year. Related to production, such emission is < 0.1 g/tpvc. to be compared with some tens g/t of controlled emissions. CTL015369 This very low value was measured in the plant working at full capacity (without incident during the trials). (Remark : in case of leakage, detected by the continuous GC CARLO ERBA, a correction procedure is applied by the operators to restablish the normality). 5 Considering the chronologic evolution of the measured emissions according to the different polymerization conditions, it can be observed that VCM emission increases with some operations like filling, emptying, degassing, ... For instance, the average concentration increases from about 10 /d/m3 (base level) to 50 /xl/m3 during VCM introduction. When considering the number of pneumatic valves (in contact with VCM) actuations during the sampling periods one can estimate that VCM release is approximately : - 50 mg/valve, hour, or - 25 mg/actuation. 3.4. Consistency with the continuous GC monitoring During the 3 sampling periods, practically no usable peak was observed by the continuous chromatograph (detection level = 0.1 ppm) : this is consistent with measured levels. 3.5. Consistency with the fixed passive samplers The average level measured by the fixed passive samplers (35 /il/m3) gives an estimate of total emissions (* 1.3 g VCM/h) higher but consistent with SIPIN measurements. 3.6. Consistency with bagging measurements - Emission factors The above actual emissions values are very low and we would like to confirm them by spot bagging measurements several times on three typical equipments handling liquid VCM : one flange, one manual valve and one automatic valve. Bagging measurements are very expensive and the aim was not to perform a exhaustive statistical study but to to check if the emissions factors of the equipment were consistent with the low emission measured (see appendix 7). The emission factor for the flange (SUPRANITE OIL gasket of SEEM) is 30-40 /*g/h. For the manual valves the range observed is between 300 and 1400 /tg/h. The automatic valve (TUFUN without bellow) has an emission factor varying from 2/tg/h in stand-by to 10 mg/h in operation. The result confirms publications in the literature [11] saying that bellow valves are not essential to reduce fugitive emissions. CTL015370 6 Compared with published results, the values measured by SOLVAY are very low and confirm DOW data on Chlorine emissions [13]. They are consistent with the low emission figures in the area. An explanation of the good results can be the preventive maintenance procedure and the criterion for intervention. Years ago US procedures recommanded repairs (for safety reasons) when the "OVA" detected more than 10000 ppm at one point of the leaking device. More recently EPA lowered the recommandation to 500 ppm. At this SOLVAY Tavaux plant, 230 critical points are checked monthly and the intervention level for maintenance is 20-50 ppm. 4. CONCLUSIONS Although the study is not exhaustive (2/3 of the critical material was checked, no incident during measurement, number of experiments,..), consistency and very low figures lead to following conclusions : The level of fugitive emissions is significantly lower than the level of controlled emis sions. This results justifies fully ECVM recommandations for reduction of fugitive emissions [12]. The actual level of emissions shows that potential reduction by the possible substitution of conventional valves by bellow valves is negligible. The best available technology and the Code of good practice, both proposed by ECVM, give for PVC industry lower level of emission than those usually admitted by the literature for general petrochemical industry. In particular, preventive monitoring of fugitive emissions and good maintenance by skilled people is a suitable practice which keeps fugitive emissions at a very low and acceptable level. CTL015371 Bibliography : [1] HUGHES T.W., TIERNEY D.R., KHAN Z.S. Measuring fugitives emissions from petrochemical plants Chem. Eng. Progr., August 1979, pp 35-39 [2] ROSEBROOK, D.D. Fugitive hydrocarbon emissions. Chemical Engineering, October 1977, pp 143-149. [3] FREEBERG C.R., AARNI C.W. (Chevron Research Co) Hydrocarbon emission control in petroleum refineries. Chem. Eng. Progress, June 1982, pp 35-39. [4] SCHAICH J.R. Estimate fugitive emissions from process equipment. Chemical Engineering Progress, august 1991, pp 31-35. [5] STUCKER T. Tracking ever-shrinking emissions. Chem. Engineering, October 1991, pp 90-99. [6] WALLACE M. Controlling fugitive emissions. Chem. Engineering, August 1979, pp 78-92. [7] FARANT J.P., Me KINNON D., ROWLANDS N. Mise au point de m&hodes utilisant un gaz traceur pour determiner le rendement des systfemes de ventilation dans divers milieux de travail. Travail et sant6. Hiver 1985. 6 p. [8] BERGLUND R.L., ROMANO R.R., RANDALL J.L. Fugitive emissions from the ethylene oxide production industry. Env. Progr., 9, Nl, feb. 1990, pp 10-17 [9] UPTON S. Fugitive emissions control in packed valves. Chem. Eng. Progress, August 1990, pp 70-76. [10] SURPRENANT N., Shutting off fugitive emissions. Chemical Engineering, Sept. 1990, pp 199-202. [11] J.F. GARDNER (Xomox Corp.) Selecting valves for reduced emissions. Hydrocarbon Processing, August 1991, pp 87-92. [12] European Council of Vinyl Manufacturers Guidelines for environmental improvement of PVC-Production (Suspension PVC), April 1993. (non published) CTL015372 [13] Ph.A. ROSS (Dow Chemical Company) Survey equipment for leakage. Chemical engineering, February 1993. [14] ROGERS T.R., KURYLA L.A., HOGAN M.P. (PPG Ind. Inc.) Development of unit specific predictive emission equation. Proceedings of the Air and WMA Annual meeting 1991, 84 th V8, paper 91/91.4. 18 pp [15] J.B. WRIGHT (Neles-Jamesbury, Inc.) Avoid valve leaks. Chemical engineering progress, June 1993, pp 62-64. [16] Th. SPOCK (XOMOX Corp.) Control fugitive emissions from valves. Chemical engineering, February 1993, pp 82-86. CTL015373 Proportion des equipements presentant des taux d'emissions significatifs (cad > a 0.5 g/h) en % du n total d'equipements selon [1] industrie industrie raffineries oxfde butadiene petrole ethylene Equipement vannes garnitures de pompes garniture compresseurs clapets de securite brides 4.5 15 16 39 50 95 0 36 30 13 36 54 23 1 CTL015374 Emissions atmosphEriques fugitives. Comparaison des facteurs d'emission selon diverses sources en g.h-- 1. source-1 Equipement vannes gaz vannes liq lgers vannes liq lourds scion SCHAJCH. 1991 SOCMI L NL facteurs stratifies > 10000 < 10000 1000110000 011000 ______ 01 ______ [2L_ 5.6 45.1 7.1 85.2 . 0.23 0.23 ______ [H_ 0.48 1.65 1.71 9.63 0.23 0.23 0.14 0.28 0.23 selon BERGLUND el al ind C2H40 [5] 0.2 0.25 f61 0.22 0.63 garnitures pompes liquides tigers liquides lourds 49.4 21.4 437 12 33.5 1.98 389 13.5 92.6 3.8 19.4 19.4 garnitures de com presseurs gaz/vap clapets de security gaz/vapeur 228 1608 89.4 264 11.3 104 1691 44.7 279 11.4 0.35 0.35 0.07 0.07 brides 0.83 37.5 0.06 8.75 0.02 0.25 0.32 bouche-trous 1.7 11.9 1.5 8.8 0.13 0.49 0.49 prises echantillons 15 12 1.5 *" "" -- [1] facteurs SOCMI (recommandEs par la Society of Organic Chemicals Manufacturing Industry) [2] facteurs applicables aux sources considqrEes comme fuyardes (L) {controle analyseur > 10 000 ppm) [3] facteurs applicables aux sources considers non fuyardes (NL) (controle analyseur < 10 000 ppm) [4] facteurs stratifies (controle analyseurs 0-1000 ppm, 1000-1000 ou > k 10000 cfr [3]) [5] facteurs moyens mesures dans I'industrie de I'oxyde d'EthylEne [6] facteurs tenant compte d'une concentration "par defaut* lors du controle des equipments de 8 ppm CTL015375 dJiv^Xj3Mj&JTCrr\ cUf\ ALCV~L Ur oUA pcry'iA^o fa r^Awu ** VTL H M X VTL3. n ~1 y* 1 VTL VTL 2. "1 IT i n X ix *r^\ rL\ H /- 1 x--^ i I tL * _ --i X _J______ i 1 X -- i_ -- 1 vT 1 *1 1 <\4aVvuia^^Yv^ ll\ VC CCqtud^- Ur cKar^rw.iAA: Ac 1 ofozexi 6c** CeCuu^ 'U rv\ VTL acH E] 1 VTL VTL l 2_ ieuX|x.iv\RlLto . JU4A/\ tlrTVL p-arflAJb d'CiA^ihCtv^'. flUA cU>U/i .1 A .COWllA VCM jrdl(Za.0ULA ..JKAOWVflijCo .CflO^\VvvA-_ _ _ _ _ _ _ _ _ _ S 0 ` v'tujh^aJbLun' &J a idbrachm H uovvlu^ fa^o uxCi.'.' 7- X prita- dx /TjuVtwumr pott Hxjl j CTL015376 ?h~YV Qf Y k I rwbfoiMJtoJv iv\ PinAfeiA d&JM) ^TL dXQ tctif) oO tto o^vvjcUaaJ^ cvc &V t /vuta.1: dcA M-'cu^aAoJ^V^ ^Cto CTL015377 frpevd/x g MESURE DES EMISSIONS FUGITIVES P79 ESSAI DE TRACAGE AU SF6 DU 27.2.1992 resultats exprimes en pi SF6.m-3 HEURE VTL1 VTL2 IlhOOm debut injec ion SF6 11h20m 11h40m 12h00m 70 20 12h20m 100 12h40m 140 12h45m 20 13h00m 280 13h20m 250 13h30m 170 13h40m 110 14h00m 370 14h15m 25 14h20m 170 15h00m 110 40 15h20m 30 15h40m 65 15h45m arret 16h00m <10 35 16h20m <10 16h30m <10 16h40m <10 17h00m <10 17h10m <10 17h15m <10 17h20m <10 VTL3 <10 20 <10 10 <10 <10 VTL4 <10 <10 <10 moyenne pendant duree effective de tracaqe (12:00 a 16:00' MOY4VTL VTL1 VTL2 VTL3 VTL4 49 144 46 6 0 UNE MESU RE EN POS1`E FIXE (C6) 15 SECTION 0.196 m2 d6bit d'air extrait par VTL (m3/h) TOTAL NOMINAL 3500 3500 3500 3500 14000 MESURE 3534.3 quantity de SF6 extraite par les VTL TOTAL en pl.m-3 505 160 21 0 686 eng 14.1 4.5 1.5 0.0 20.0 soit en % 74 23 3 0 100 masse SF6 emise (pesee de la bouteille) % de recuperation : 21.4 g 94 % CTL015378 ro 3 fo H UU>1 -0 lO cone on VCM on p l.m -- M v/V 3dd MESURES DES EMISSIONS FUGITIVES VCM extrait par les 4 VTL 45 r 40 35 30 25 20 15 10 5 0 09:00 AM 11:00 AW essai du 31--mars-92 VTL1------- VT12 l/rt*---------|/7Z.', . SOLVAY USINE DE TAVAUX SERVICE ENVIRONNEMENT MESURE DES EMISSIONS FUGITIVES SUR DIVERS EQUIPEMENTS DATE TYPE D'APPAREIL TEMPS VOLUME CONCENT DEBIT DE EQUILIBRE EMBALLE EN VC1 FUITE fheurel [dm3] [ml.m-3] [^g.h-i] 21 -Avr-92 bride VC liquide 22-Avr-92 bride VC liquide 23-Avr-92 bride VC liquide 2.00 2.08 16.67 1.3 18 1.2 26 0.9 259 33 42 39 28-Avr-92 vanne autom [1 ] 28-Avr-92 vanne autom [2] 0.75 16.3 2.75 2.5 342 hi 3499 < 5 < 2.1 30-Avr-92 vanne manuelle 05-Mai-92 vanne manuelle 06-Mai-92 vanne manuelle 18.33333 17 41 10 193 294 45 191 1411 40 370 1007 [1] une manoeuvre effectuee pendant la duree de mesure [2] aucune manoeuvre effectuee pendant la duree de mesure [l] >toocj/K CTL015380