65 A publication of CHEMICAL ENGINEERING TRANSACTIONS VOL. 3, 03 Guest Editos: Eddy De Rademaeke, Buno Fabiano, Simbeto Senni Buatti Copyight 03, AIDIC Sevizi S..l., ISBN 978-88-95608--8; ISSN 974-979 The Italian Association of Chemical Engineeing Online at: www.aidic.it/cet A Theoetical Appoach to Risk Evaluation in Connection with Oxygen Releases Emilio Palazzi, Fabio Cuò, Buno Fabiano* DICCA Civil, Chemical and Envionmental Engineeing Depatment - Univesity of Genoa, Via Opea Pia, 5 645 Genoa, Italy. * bown@unige.it Oxygen eniched combustion offes seveal advantages, including highe combustion efficiency (Hu et al., 00) and highe CO concentation in the flue gas, allowing an easie and cheape CO sepaation fom the gas itself (Hu et al., 000). In addition, oxygen combustion is consideed in the IPCC list as a categoy of CCS options fo lage point souce emittes. Oxygen eniched pocess can be adopted to enhance sulphu ecovey in the effluent steam adopting modified Claus pocesses in efineies. Accoding to intenational histoical data, accidents involving gas pipelines still happen and often esult in sevee consequences. This implies that measues need to be adopted in ode to adequately quantify and theeby to mitigate the isks. In elative tems, the poblem of HazMat pipeline isk assessment does not come with hazad analysis, o the estimation of failue fequency, but with the calculation of the consequences. In this pape, we conside the peculia case of oxygen instantaneous o continuous eleases fom a pipeline and consequent evolving scenaios. A novel modelling appoach to isk evaluation is elaboated stating fom the athe common situation of a pipe-way whee seveal gas/liquid lines ae localized, in a athe congested industial aea. As an applicative case-study, it is consideed a downsteam oil industy whee the possibility of eplacing the ai feeding to Claus plant with oxygen eniched line is investigated, with the aim of enhancing sulphu ecovey yield. This technical option fom one side can be seen as an appoach to pocess intensification, fom the othe side poses obvious safety issues.. Oxygen accident statistics Figue : Numbe of accidents involving gas and liquid oxygen, soted by accident location. Figue : Numbe of accidents involving gas and liquid oxygen, soted by activity.
66 Table : Refeence composition and consequent taget dilution fo diffeent eleases. Release Refeence composition Taget dilution (v/v) O 75% O y d.in = 0.68 50% O y d50 = 0.37 5% O y d5 = 0.05 O + CH 4 HFL y dh = 0.5(+k)/(-.8k) Stoichiometic condition y ds = 0.095(+k)/(-0.36k) y dl = 0.05(+k) CH 4 HFL y dhc = 0.5 Stoichiometic condition y dsc = 0.095 y dlc = 0.05 The majoity of companies in the pocess secto ecognize the impotance of leaning fom accidents by epoting all incidents with the potential to cause injuy o loss and by using analytical techniques to establish the oot cause of incidents so as to identify technical, oganizational and human factos which may have contibuted to the incident (Fabiano and Cuò, 0). The accident cases, which wee used to cay out the statistical suvey, wee taken fom the database FACTS (Failue and Accidents Technical infomation System) managed by managed by the Unified Industial & Habou Fie Depatment in Rottedam-Rozenbug (NL). A total of 40 accidents wee selected and validated: 83.3 % ae efeed to gas and 6.7 % to liquid. Statistical esults in fom of immediate eadability ae summaized in Figues and, espectively accounting fo accident location and activity. Ove a total of 68 accidents ecoding injuies (4.9 % of the total), 4 involved gas and 7 liquid. In events (30.4 % of the total) it was ecoded at least one fatality: 07 involved gas and 5 liquid oxygen. It is notewothy noting that on the contay to pocess plant statistics, gas phase is by fa the most fequent condition connected to accidents.. Theoetical The isk levels connected to an oxygen elease, possibly in pesence of othe hazadous eleases (inhalation, deflagation, detonation), depend upon the inteaction with the envionment and the consequent dispesion o build-up in semi-confined egions. Oxygen clealy plays a ole detemining maximum flame tempeatue and themal powe in given scenaios, e.g. pool fie (Palazzi et al., 0). As a shot-cut appoach, with a hazad analysis oientation, thee kinds of scenaios/elease time modes can be consideed, following a loss of containment in a pipeway whee an oxygen line is pesent: pue O neainstantaneous o continuous elease and ove-oxygenation of the atmosphee; concuent elease of O and HC, of nea instantaneous o continuous duation and consequent fomation of explosive mixtues; build-up of O and HC in semi-confined spaces and fomation of explosive pockets. The methodological appoach hee outlined is based on the following steps: identification of the citical efeence concentations fo the diffeent scenaios (health effect fo pue O elease on the basis of vulneability to human beings and explosive limits fo O /HC eleases); analytical calculation of the dilution equied to attain the efeence concentations; shot-cut dispesion modelling fo continuous and instantaneous eleases to develop contou map descibing the gas concentation; inteaction of O and HC eleases (as an example we efe to methane). The esults of ai dilution needed to attain efeence compositions stating fom pue oxygen and HC eleases, calculated on the basis of mass balance and stoichiomety ae summaized fo the sake of bevity in Table. The following analytical appoach consides a neutally buoyant dispesion, consideing the slight gas density diffeence between oxygen and ambient ai, as well as the absence of consideable auto-efigeation duing elease, unde the given conditions.. Continuous elease In ode to evaluate gound concentation (v/v) of a continuous elease of given elease ate, we stated fom the well-known Gaussian model and dispesion coefficients available fom BNL (Bookhaven National Laboatoy) (Pasquill and Smith, 983): y ηy Q β d( 0) X,Y, e X = () β X ε The isopleth equation coesponding to the gound level concentation boundaies in connection with the given dilution facto is attained fom Eq. () as follows:
67 Y = β η X ln β ε yd X Q () The maximum downwind distance, at gound level, whee the given taget dilution is attained, can be calculated as: Q β X = ε (3) Making efeence to the wost conditions (atmospheic stability D, accoding to BNL) and to the coesponding set of paametes: u = m s - ; α = 0.3; β = 0.7; γ = 0.06 ;ε 0.058;η 5., (Pasquill and Smith, 983) pevious equations can be modified espectively as: Q β 7 3 X =. (5) 0 7 7. 0 44 Q Y =. X ln (4). 4 X Consideing a given elease, we define as a hazad indicato the extension of the flammable egion at the gound level. In this espect, the quantitative effect of an oxygen elease can be evaluated making efeence to the elative incease of the hazadous aea, compaed to the absence of the oxygen elease, accoding to Eq. (6). Ai Aic Δ Ai = (6) A ic It must be noticed that in the most common situation when egions chaacteized by gas concentation exceeds HFL, the aeas A ic and A i can be calculated by subtacting the aea with bounday coesponding to isopleth to the aea with bounday HFL. Stating fom Eq. (), the maximum is easily found as: Q εyd Q Y = α. (7) max At last, we can calculate the hazad index coesponding to the elative incease of the gound level flammable egion: 0 [ ( 8 ) ].. k Δ A = 0. 38 (8) i. Instantaneous elease Again, we evaluate gound concentation (v/v) of a nea-instantaneous elease of given volume, by the Gaussian model and coesponding dispesion coefficients by BNL (Pasquill and Smith, 983): y V ( ) η [( X x) + Y X,Y, = e ] d 0 ε X i 3β The isopleth coesponding to the gound level concentation boundaies in connection with the given dilution facto ae cicle with cente C(x,0) and adius, both vaying with time. Fom Eq. (9), one can wite: (9) = β η X ln 3β ε yd X Q (0) The maximum downwind distance, at gound level, whee the given taget dilution is attained, can be calculated fom Eq. (9) fo X = x e Y = 0: V 3β X = ε () i yd
68 Table : Analytical fomulae of the chaacteistic distance fo diffeent elease duation. Release Refeence composition Chaacteistic distance [m] Continuous elease Chaacteistic distance [m] Instantaneous elease O 75% O X 75 = X in = 9.6Q X 75 = X in = 5.V 50% O X 50 = 5Q X 50 = 6.9V 5% O X 5 = 59Q X 5 = 7V O + CH 4 HFL X H = 8[(-.8k)Q 3 ] X H = 0[(-.8k)V 3 ] Stoichiometic condition X S = 38[(-0.36k)Q 3 ] X S = 3 [(-0.36k)V 3 ] X L = 59Q 3 X L = 7V 3 CH 4 HFL X Hc = 8Q 3 X Hc = 0V 3 Stoichiometic condition X Sc = 38Q 3 X Sc = 3V 3 X Lc = 59Q 3 X L = 7V 3 As a consevative estimate, Eqs. (9) and (0) ae utilized efeing to the wost meteoological conditions (atmospheic stability D, accoding to BNL) and to the coesponding set of paametes: u = m s - ; α = 0.3; β = 0.7; γ = 0.06 ;ε 0.058;η 5., so that one can wite: 0. 47 Q 0 7. = 0 44 V X = 4. 3 ln. x ln () y (3) d. 3 x Also dealing with the instantaneous elease we can define the hazad index as the elative incease of the flammable egion at the gound level; fom eq. (6) the index is evaluated by: i ( ) 0.. 8k 93 Δ A = ln (4) A summay of the analytical fomulae obtained with staightfowad calculations fo the chaacteistic distances in connection with the diffeent elease modes, is shown in Table. 3. Case-study 3. Plant layout As an applicative case-study, we conside a pipeline suitable to feed a Claus plant located in a middle size oil efiney. Due to stingent envionmental egulations, an upgade to oxygen-eniched pocess is consideed, in ode to incease the oveall sulphu ecovey. The technical data of the designed oxygen pipeline at an opeative pessue p m ae summaized in Table 3. Accoding to the famewok outlined in pat, following scenaios ae consideed: Continuous point souce elease: oxygen elease ate coesponding to the maximum designed value of the pipe: Q = 0.788 m 3 s - at 98 K and atm. Instantaneous point souce elease: oxygen dischaged volume coesponding to 00 m of the pipe: V = 0 m 3 at 98 K and atm. Table 3: Designed technical data of oxygen pipeline and coesponding opeative conditions. Pipeline length and diamete L =.5 km; d = 0.03 m Mean pessue p m = 6.6 0 5 Pa Tempeatue T = 98 K Maximum flow ate Q = 600 Nm 3 h - 0.788 m 3 s - at 98 K and atm Oxygen mola faction y = 0.995 Oxygen volume in the pipeline V c = AL = 8 m 3 Oxygen volume at 98 K and atm V = V c p atm /p m 500 m 3 Oxygen flow ate at 98 K and atm Q = 0.788 m 3 s - Oxygen density at p M ρ M = p M M /RT 8.4 kg m -3 Table 4: Chaacteistic distances coesponding to oxygen eleases. Chaacteistic distance [m] Continuous elease Instantaneous elease X 75 = X in 8. X 50 8 X 30 50 7
69 Table 5: Chaacteistic distances fo oxygen/methane elease, compaed to methane elease. Release Chaacteistic Continuous elease Instantaneous elease composition distance [m] Q 3 (m 3 s - ) V 3 (m 3 ).3.58 7.88 3.5 40 00 O + CH 4 k =0.64 k=0.5 k=0. k=0.64 k=0.5 k=0. X H 09 7 39 9 X S 3 38 9 53 59 7 X L 67 80 46 86 96 05 CH 4 k=0 k=0 k=0 k=0 k=0 k=0 X Hc 37 45 57 58 67 54 X Sc 44 5 6 66 74 57 X Lc 67 80 46 86 96 05 Table 6: Hazad index incease fo oxygen/methane elease, compaed to methane elease. Release Flammable egion Continuous elease Instantaneous elease composition incease [%] Q 3 (m 3 s - ) V 3 (m 3 ).3.58 7.88 3.5 40 00 O + CH 4 k =0.64 k=0.5 k=0. k=0.64 k=0.5 k=0. ΔA i 3 5 5 3 83 Table 4 summaizes the esults of the chaacteistic distances obtained utilizing the dilution factos in Table, by means of Eqs. (5) and (3), espectively fo the continuous and instantaneous dischage mode. Table 5 shows the esults of the chaacteistic distances calculated fo oxygen and methane eleases and pue methane eleases, by means of Eqs. (5) and (3), utilizing the coesponding values of the dilution factos. 3. Flammable egion extension Table 5 summaizes the esults of the hazad index calculation pefomed stating fom Eqs. (8) and (4), fo diffeent popotions of the concuent hazadous eleases into ambient ai. 4. Discussion and conclusions Coeteis paibus, the pobability of fie popagation inceases as the flammable egion inceases. As clealy depicted making efeence to the applicative example in Figues 3 and 4, an oxygen elease modifies the hazadous aea by exeting its effect on the isopleth coesponding to HFL, so as to educe the zones occupied by mixtue with highe HC concentation. Also taking into account Table 6, following consideation can be dawn, with espect to the hazad index: consideing methane eleases compaable o lowe than oxygen ones (k>0.64), the oxygen effect is negligible as the souce concentation is lowe than HFL; when a massive methane elease (k<0.) is concened, again the oxygen contibution to the hazad incease can be consideed negligible. 0.0 5.0 0.0 Stech HFL 5.0 0.0-5.0 0 50 00 50 00 50 X [m] 300-0.0-5.0-0.0 Figue 3: Maximum isopleths fo thee taget concentations (dotted lines) and cuves enveloping all maximum concentations (solid lines) fo instantaneous O elease (V =0 m 3 ). Figue 4: Isopleth cuves (lowe flammable limit, uppe flammable limit and stoichiometic conditions) efeed to a continuous elease of oxygen and methane (Q =0.788 m 3 s -, k=0.5).
70 The aea of the flammable egion and, theefoe, the hazad index hee defined, can be significantly enhanced by an oxygen elease only when the HC elease (total volume o dischage ate is two thee times the oxygen one, with an incemental value eaching 00 % in case of nea instantaneous elease, fo example due to thid paty inteaction. In any case, it can be noticed that the incemented hazadous aea is limited to appoximately 00 m fom the elease point. Dealing with the seveity of a lage fie and theefoe the intensity of the effects and the vulneability of people and popety, it is known that an oxygen elease can enhance the fie themal powe and the flame tempeatue. This item can be citical in pesence of vulneable components in a congested pipeway: e.g. high intensity fie exposue of steel flanges may easily esult in futhe escalation, as bolt on flanges ae pe stessed (nealy 50 % of ultimate stength) (Opstad and Bege, 00). Moeove, pope isk pevention measues, including the safety management system, can be consideed in ode to educe the fequency of the pipeline failues (Milazzo et al., 00). On the basis of the theoetical appoach, it follows that the incemental effect is negligible both in connection with HC eleases of small volume and in case of catastophic HC eleases. In fact, in the fome event, atmospheic envionment supplies the equied oxidant fo combustion in the neighbouhood of the elease point, so allowing a fully developed fie. In the latte, eleased oxygen epesents only a small faction of the oxidant needed to ensue a complete combustion of a massive quantity of HC. The most significant effects in tem of elative flame tempeatue incease and the consequent additional hazads ae elevant to HC elease of medium seveity and ae limited to 0-0 m fom the souce point, as clealy indicated by the isopleths cuves depicted in Figues 3 and 4. Results obtained by the analytical model wee compaed with available integal models, showing output diffeences not exceeding 0 %. The mathematical appoach hee outlined allows obtaining, by means of explicit fomulae, the safety distances of sensible tagets fom the elease location fo both pue oxygen and concuent oxygen/hc elease, so as to identify incemental hazads and set-up technical/manageial measues to avoid ignitions, explosions and seious toxic effects. Nomenclatue A i = flammable aea fo a elease of oxygen and hydocabons, i.e. Q = (+k)q 3, m A ic = flammable aea fo a elease of hydocabons, i.e. Q = Q 3, m k = atio of oxygen to hydocabons ate in the elease, - - Q = oxygen volumetic ate at 98 K and atm, m 3 s Q 3 = hydocabon volumetic ate at 98 K and atm, - m 3 s - Q = elease volumetic ate at 98 K and atm, m 3 s = adius of the isopleth fo instantaneous elease, m u = wind velocity at 0 m above gound level, m s - V = oxygen volume at 98 K and atm, m 3 V 3 = hydocabon volume at 98 K and atm, m 3 V = elease volume at 98K and atm, m 3 x = downwind distance fom the mass cente of a elease at time t, m X = downwind distance fom the emission point, m Y = tansvesal coodinate, m y d = elease dilution, v/v α = paamete defined accoding to BNL, - β = paamete defined accoding to BNL, - γ = paamete defined accoding to BNL,- ε = paamete defined as ε=παγu, - ε i = paamete defined as ε i =(π 3 ) / α γ, - η = paamete defined as η=/(α ), - Refeences Hu Y., Naito S., Kobayashi N., Hasatani M., 000, CO, NO x and SO emissions fom the combustion of coal with high oxygen concentation gases, Fuel 79, 95-93. Hu Y.Q., Nikzat H., Nawatu M., Kobayashi N., Hasatani M., 00, The chaacteistics of coal-cha oxidation unde high patial pessue of oxygen, Fuel 80, -6. Fabiano B. and Cuò F., 0, Fom a suvey on accidents in the downsteam oil industy to the development of a detailed nea-miss epoting system, Pocess Safety and Envionmental Potection 90, 357-367. Milazzo M.F., Maschio G., Uguccioni G., 00, The influence of isk pevention measues on the fequency of failue of piping, Intenational Jounal of Pefomability Engineeing 6, 9-33. Opstad K., Bege G., 00, Fie integity of flanges, Chemical Engineeing Tansactions 9, 69-74, DOI: 0.3303/CET0908 Palazzi E., Fabiano, B., 0, Analytical modelling of hydocabon pool fies: Consevative evaluation of flame tempeatue and themal powe, Pocess Safety and Envionmental Potection 90, -8. Pasquill F., Smith F.B., 983, Atmospheic diffusion, 3 d edition. John Wiley & Sons, New Yok, USA.