Document Z4ZrBp3zmJ8702ObnBV59vbnY

Hydrocarbons Reactive Chemicals Hazards NOx Containing Process Streams Fluid Catalytic Cracking (FCCJ units often export an olefins containing Mlight ends" stream to ethylene plants for separation. This refinery gas typically contains nitric oxide (NO) which is formed during regeneration of the catalytic cracker catalyst. Regeneration consists of burning coke which contains varying amounts of nitrogen depending upon FCC unit feed. N20 and N02 can be completely removed in amine and caustic scrubbers. N203, however, dissociates to NO and N02. Low levels of oxygen react with NO to form nitrogen dioxide: 2N0 + 02---------------2N02 Nitrogen dioxide reacts with nitric oxide to from nitrogen trioxide: N02 + NO---------------N203 Formation of NOx Gums NOx compounds will react with 1-3 butadiene or cyclopentadiene to form an amorphous gummy polymeric material. These compounds are unstable and will ignite or explode at low temperatures. Incident On February 22, 1990, Shell's Berre, France ethylene plant experienced an explosion inside their cold box. As a result, the plant was out of service for 5 months. The cause of the explosion has been attributed to the deflagration of gums present in the low-pressure methane pass of their cold box. The gums contained 14-21% nitrogen with numerous nitro and nitrosocomponents on short hydrocarbon chains. Precautions Ethylene plants should not process FCC streams without special treatment to remove nitrogen and nitric oxides. Reference "Cold Box Explosion at Shell Steam Cracker in Berre, France", by Jacques Kohler. AIChE Spring National Meeting, April 7-11, 1991 DO A 030/49 CONFIDENTIAL 27d COLD BOX EXPLOSION AT SHELL STEAM CRACKER IN BERRE, FRANCE Jacques Kohler FOR PRESENTATION AT AlChE Spring National Meeting Houston, Texas, April 7 - 11, 1991 Copyright Author(s)/ Employer(s) AlChE shall not be responsible for statements or opinions contained in papers or printed in its publications. DO 030750 `-OWFrDETNTTAL COLD BOX EXPLOSION AT SHELL STEAM CRACKER IN BERRE, FRANCE JACQUES KOHLER , SHELL BERKS, FRANCE 7 )L ANUSCRIPT CT.7. MAR 7 1991 ABSTRACT An emergency shutdown occured on February 22nd 1990 at the Shell Berre 420 000 tonnes/year steamcracker in Southeast France after an explosion inside the cold box. As a result, the plant vas out of operation for nearly S months, the cold box having to be completely replaced. After an investigation by a team of experts and exchanges with other relevant industrial specialists, the cause of the cold box explosion has been attributed to the deflagration of gums present in the lov-pressure methane pass, inside one of the fin-plate heat exchangers of a twin pair. The gums contained 14 to 21X of nitrogen with numerous nitro and nitroso components on short hydrocarbon chains. The source of NOx vas identified as NOx components present in the catalytic cracker dry gas vhich had been part of the Berre steam cracker feed since 1982. Corrective measures have been taken to avoid a similar incident. INTRODUCTION The Berre cracker had been taking refinery gas from the neighbouring Shell refinery since 1982 as part of its feed, at the level of the cracked gas compressor, upstream of the caustic vash tover. This stream included gas from one Fluid Catalytic Cracking unit (FCC) from 1982 to 1987 and of tvo FCC units since the end of 1987 to the day of the incident. The Berre cracker is of a standard hummus design with no front-end deethaniser before the cold box. No special treatment vas applied to these refinery gases, like a dedicated cold box, to remove methane, nitrogen or hydrogen and nitric oxides The practice generally applied by the steamcraeker industry (ref.l) vas used, especially during the shutdowns (every four years) and the resulting cold-box thaws. I. BACKGROUND 1. CHARACTERISTICS OF THE FCC GAS CONTAMINANTS The concentration levels of the contaminants in the tvo gas streams did not differ significantly. Low levels of nitrogen oxides, 'knovn to be present in these refinery gases, vere measured using the GRIESS-SALTZHAN method. Nitrogen dioxide (N02) vas never measured (detection limit 0,01 ml/m3). Lov levels of nitric oxide (NO), ranging between 0,02 and 0,3 ml/m3, were currently observed. 2. In general, the ammonia (NH3) concentration vas belov 0,1 ppm but values of 2 to 4 ppm were encountered. Also, concentrations of around 10 ml/m3 of oxygen were found to be present in these gases. 2. COLD BOX STATUS BEFORE FEBRUARY 22nd 1990 A scheme of the Berre cold box is given in Appendix 1. No real abnormal temperature or pressure profile through the cold box had been noticed before the last shutdown in Harch 88 However in 1987, a slight decrease of the hydrogen purity had been observed due to higher temperatures in V 308 (- 155C versus a design of - 165C), as well as some difficulties to evacuate the liquid stream of V307 to the demethaniser. A procedure, consisting of heating the cold box, downstream of V306, to -95/-100*C was applied, in order to melt or sublimate any NOx present. Then flushing with a counter flow of methane from the demethaniser through E 308 to the liquid draining header, permitted the recovery of the right temperature profile. During the Harch 88 shutdown and the resulting cold box thaw at ambient temperature, up to 35 ppm of NOx were measured in V 307, working normally at - 129*C and a very small amount of NOx gums was recovered on a E 309 inlet filter (deposit 0 in appendix 1). In Hay 88, the presence of "blue liquid" was observed for the first time in the LRCV of V 308 and a procedure to purge this valve with nitrogen was applied. From Hay 88 to the day of the incident, fouling of the LRCV of V 307 and V 308 occured several times resulting in difficulties to evacuate the liquid from these vessels and an increase of 2,3 bar of the differential pressure between V 307 and the demethaniser was observed. Also the hydrogen purity decreased, and temperature of V 308 vas more often at - 157C. Two trials to recover the design conditions by heating the last parts of the cold box to - 95C and purging vith methane from the demethaniser did not result in the intended reduction of differential pressure between V 307 and the demethaniser. n. DESCRIPTION OF THE INCIDENT The incident occured after an upset of the steam-cracker operation which resulted from the opening of a cracked gas compressor relief valve. After one hour with the cracked gases essentially to the flare and vith the cold columns on total reflux, the cold box temperature increased by 25C. At that moment, both ethylene and propylene compressors tripped, due to high liquid levels in the suction vessels, and remained out of operation. C>0 A CoNprnJ' tAi 3 About one hour before the explosion, and 3 hours after the start of the incident,the pressure of the demethaniser was lovered by 2 bars, in order to empty the vessels V 304 to V 308 vhich had high levels of liquid. The intention vas also to heat the dovnstteam part of the cold box to about -70C in order to get rid of the high differential pressure between V 307 and the demethaniser. During the investigation, it vas established that during that period, a flov of 30 to 60 metric tons per hour of cracked gases entered the cold box, creating a heat front vhich varmed up the cold box (see appendix 2) and that the temperatures of V 307 and V 308 increased at first to - 58*C (a temperature higher than originally intented). Vhen, finally, the exchangers E 310 - E 311 and 313 lost their liquid ethylene hold-up, the temperature rose at a higher rate (about 4*C/min) to - 25 / - 30*C for the coldest parts of the cold box, and the explosion occured. Operators immediately located the origin of the explosion at the cold box. A gas cloud estimated at 20 tons of hydrocarbons plus the perlite, escaped and covered a large part of the steamcracker. The cloud vas confined by vater spraying, especially from the furnaces vhich had been tripped immediately. Larger quantities of NOx vere measured (up to 200 ppm) around E 308 and the first visual inspection of the cold box could only take place 2 veeks after the incident due to continous NOx release over that period. The visual inspection shoved that the explosion had occured inside the E 308N (see photo 1 and appendix 1) vhich presented a large spherical cavity. The E 3085 vas damaged by the explosion of E 308N, the LP methane header vas torn avay and broken in pieces. The casing and frame of the cold box vere distorted by the explosion (See photo 2). At a first look, the other exchangers seemed less damaged, but a more thorough inspection shoved the necessity to replace the entire cold box. III. DETERMINATION OP THE CAUSES OP THE EXPLOSION The investigation vas done by a team of Shell experts both from Shell CHIMIE and Shell Internationale Chemie Maatschappij. Vhere required, advice vas sought from outside specialists and laboratories familiar with explosive reactions and the analysis of components causing explosive reactions. 1. EQUIPMENT AND EXPLOSION STUDIES The first results of the investigations confirmed that the burst vas not the result of a rupture by overpressure or thermal constraint. The explosion had occured in the lov-pressure methane pass, inside one of the fin-plate heat exchangers of a tvin pair (the E 308N). The melting temperature of aluminium (660"C) vas reached at specific points of the damaged heat exchanger. A study vas later done to evaluate : A 0307S3 C nfidetntta, 4. - the quantity of product (in TNT equivalent) vhieh created the explosion. - the type of the explosive reaction To quantify the explosion pover, various calculations have been done, taking into account i - the level of the sound of the explosion perceived in the control room. - the study of the perlite inpact - the study of the deformation of the frane of the cold box and of an experimental model exposed to an explosion. The conclusions of these studies vere that the explosion phenomenon vas of a deflagration type vith a most probable equivalent mass of TNT betveen 17 and 28 kg. Hovever some local detonations could also have taken place. The perlite had a very important damping effect during the explosion. 2. ANALYSIS OP THE DEPOSITS Tvo different deposits vere recovered (see appendix 1) Deposit 1 : on a filter in the bottom line from V 307 to E 308, at the entrance of E 308. Deposit 2 : at the entrance of E 308 in the lov pressure methane line coming from E 309 (See photo 3). The deposit 1 has very clearly been identified as ammonium salts. It consists essentially of nitrite, bicarbonate and small quantities of nitrate of ammonium. It is completely soluble in vater. The deposit 2 looks like a red-brovn gum, vith a honey like consistency. This unstable product decomposes slovly at ambient temperature, releasing NOx. This deposit is very heterogeneous, vhieh explains the fact that the different laboratories involved found a nitrogen content varying from 14 to 21X. The product can be divided into a vater soluble inorganic phase, and an organic phase. The inorganic phase has a pH of 2 to 3 and contains essentially ammonium nitrate.(Ionic chromatography - IR spectrometry) The volatile fraction of deposit 2 has been characterized by gas chromatography coupled vith mass spectrometry. Of about 10 products, appearing on a total ionic chromatogram, about 8 have been identified (see Appendix 3) DO A 03o7S4 CONFIDENTIAL 5. IV. EXPLANATION OF THE CONDITIONS LEADING TO THB gPLOSlON 1. FORMATION OP AMMONIUM SALTS The deposit 1, vhich plugged the filter at the entrance of E 308, vas mainly responsible for the differential pressure increase betveen V 307 and the demethaniser. The heating and methane purging procedure, previously described in 1-2, had no effect on the ammonium salts. With the presence of ammonia in the refinery gas, the formation of ammonium nitrite can occur following the reaction : 2 NH3 + N203 + H20 > 2 NH4 N02 (1) and that of ammonium nitrate, following the reaction : 2 NH3 + 2 N02 + H20 <.--.> NB4 N03 NH4N02 (2) As for the formation of ammonium bicarbonate, it vas explained by a small leakage, in January 1990, of C02 in the cracked gases out of the caustic tower : C02 + NB3 H20 NH4 H C03 (3) 2. NOx ACCUMULATION IN THE COLD BOX The NOx in refinery gas is essentially nitric oxide (NO). It is formed during the regeneration of the catalytic cracker catalyst when combustion of coke containing variable amounts of nitrogen (depending on the residue fed to the FCC unit) takes place. N20 and N02 should be completely removed by the dry gas treatment (DEA + caustic) and N203 is, at ambient temperature, entirely dissociated into N0+N02. Low levels of oxygen are present in refinery gas resulting in oxygen concentration, in the stream entering the cold box, one or twice the level of the nitric oxide concentration. Oxygen reacts vith nitric oxide to form nitrogen dioxide : 2 NO + 02 <*> 2 N02 Both the rate and equilibrium of this reaction are favoured by the cryogenic temperatures and relatively high pressure in the cold box. According to V.M. BENSTOCK (ref.2) not all the nitric oxide is oxidised to nitrogen dioxide, the reaction being rate limited rather than equilibrium limited. Therefore nitrogen dioxide reacts with the excess nitric oxide to form nitrogen trioxide : 00 A 030755 N02 * NO <....*> N203 CONFIDENTIAL 6. The equilibrium for this reaction favors N203 at lov temperatures and high pressure. Nitrogen trioxide is solid at temperatures < - 102*C and insoluble in the hydrocarbons present. At the prevailing cold box temperature, it vill be present as a solid, inducing plugging in the coldest parts of the cold box. In conclusion, since the untreated FCC gases alvays contained small quantities of NO and oxygen, and in viev of the physical properties of the NOx (see Appendix 3), it vas unavoidable to find NOx in the cold box. This vas recognised after 1987 vhen partial plugging and "Blue liquid" appeared. N203 vas therefore present in V 307 and V 308, plugging from time to time the LRCV of these vessels. 3. FORMATION OF NOx GUMS It is veil knovn that NOx compounds in general and nitrogen dioxide in particular, vill react vith certain unsaturated hydrocarbons to form amorphous gummy polymeric material, generally called "NOx gums" (see Ref.3). Studies (See Ref.4) have also been done on the reaction of unsaturated hydrocarbons vith nitric oxide, nitrogen dioxide and mixtures of nitric oxide and oxygen. They noticed that the reaction rate betveen mono-olefins, propadiene or acetylene and NOx vas very slov and that ignition vas not possible belov 0*C. They shoved also that unstable nitro and nitroso compounds could be formed vhen N02 or N203 reacted vith conjugated dienes like 1-3 butadiene or cyclopentadiene and that these compounds vere apt to ignite or explode at loir temperatures. V. COLD BOX EXPLOSION HYPOTHESIS NOx had accumulated in the coldest parts of the cold box since the last shutdovn in March 1988. The heating procedure of V 306 - V 307 - V 308 at a temperature of - 95C may have displaced monoolefins like ethylene, propylene, 1-butene or acetylene from V 306 dovntream and may have formed'some "stable gums"as partially identified in components of Appendix 3. During the steamcracker upset of 22 February and the resulting cold box heating , the composition of the different vessels V 308 to V 304 vas enriched in heavier components and most likely, as simulations shoved aftervards, in components like 1-3 butadiene and 1-3 cyclopentadiene. This meant that unstable gums could have been formed. 7. When, finally, the temperature of the coldest parts of the cold box increased at a fast rate to - 25*C, due to the loss of the ethylene refrigerant, most probably the unstable NOx gums exploded folloved by the stable gums, since no product vas subsequently found in the exploded E 308 N. For some unknovn reasons, the tvin exchanger E 308 S, vhich presumably vent through the same treatment and vhich contained NOx gums, did not explode. This explanation is veil supported by the literature vhere explosions have oecured in other cryogenic gas processing units (Ref 4.5.6). These explosions vere attributed to the presence of unstable nitrogen compounds vhich had been formed also from the reaction betveen trace amounts of NOx and hydrocarbons accumulated over time, vith concentration sometimes lover than 1 ppm (Ref 7). CONCLUSIONS NOx compounds, vhen present at the entrance of a cold box, vill accumulate in the coldest part of the cryogenic equipment. If during a partial heating or cooling procedure of the cold box or during a plant upset, the concentration profile of the various mono and diolefins in the different parts of the cold box is changed, these NOx may form stable or unstable gums vith the hydrocarbons. At that moment, the potential for a very violent reaction exists, and an explosion can then occur, initiated for example by a fast heating rate of the cold box, as in the Berre case. To prevent the occurence of a similar incident in the Berre cracker, it has been decided to eliminate the presence of NOx in the cold box. The direct consequence of this decision means that FCC gases vill not be taken back as feed to the Berre cracker until a process can be developed vhich vill remove NOx from these gases. 8. REFERENCES 1. Plant Operations Progress Vol 6, Nc4, October 1987. 2. Henstock, V.H., "NOx in the cryogenic section of an ethylene plant" 1989 AICHE Spring National Meeting ; April 4, 1989, Houston, Texas. 3. Henstock, V.B., "NOx in the cryogenic Hydrogen Recovery Section of an Olefins Production Unit". Plant Operations Progress Vol 5, N4, October 1986. 4. Haseba, S, et al., "Nitric Oxide Explosion", Chemical Engineering Progress, 62 (4), P.92 (1966). 5. Bohlken, S.F., "Heat Exchanger Explosion at a Nitrogen Wash Unit", Chemical Engineering Progress, 57 (4), P.49 (1961). 6. Alexander, D.S. and C.M. Finigan, "Inert Gas Poor Piping * Explosion", Petroleum Refiner, 38 (5), P.285 (1959). 7. Haseba S, Kubo, Kitagava, "The Danger of Nitrogen Oxides in Cryogenic Gas Separating Plant", Anzen Kagaku (Safety Engineering), Vol 8, N1, P.22 (1969). DO A 030758 CONFIDENT! AL. ucttuhc up UtTHUtf HP ucnuHC lp HfONOCtH APPENDIX 1 : AUBETTE COLD BOX r> 6o 2: o *n -f y> D '"f? o 2! CO -t o ^N > cn " vO ftCCVCUD CTHWK CRACKCO cuts DO A 0 3 0 7 6 1 CO NFID ENTIAL TEMPERATURE' T"17 11 .* APPENDIX 2 : COLO BOX TEMPERATURE PROFILE r-- 14. 3 15.BJ 02311001/023M602/923h883/023tie84/e23t1Q0S A , T MOl CRACKED GASES TEMPERATURE INLET 306 T M02 .CRACKED GASES TEMPERATURE TO V304 T M03 CRACKED GASES TEMPERATURE INLET 310 v 7 M04 CRACKED GASES TEMPERATURE INLET 311 T M05 .CRACKED GASES TEMPERATURE TO V305 APPENDIX 3 NOx GUMS COMPONENTS 1) FORMIC ACID 2) ACETIC ACID CS3 3) C3H70N02 or > CH0N02 CB3 4) C3H7 O-COOH 5) CH3-C-C-CH3 H It 00 (proposed structure) 6) N02 0 CH2 (CH2) 0 N02 n=*l,2... 7) N02 0 CH2-CH-0N02 CH3 8) CF3 CH2 0 N02 PHYSICAL PROPERTIES OF NOx 1 MELTING PT NO - 161* C N02 - 9,3* C N2 03 - 102* C BOILING PT (At Ato P) - 151 C (At Working P) - 108 C 21* C - 3,5" C - DO A 000762 C 0 N' h 7 D F N T T A i PHOTO 1 E 306 N AFTER EXPLOSION PHOTO 2 COLD BOX FRAME AFTER EXPLOSION PHOTO 3 DEPOSIT 2 ON E 30S S A 030763 C.ONF I DFiVT T Ai