Kinetics of the Reactions of Ethylene Oxide with Water and Ethylene Glycols Georges A. Mothem, Aue lane, Mar. Levin’ Harold Gish Senon Chippet Surendra K. Singh’ ond Peter. Chipman’ “Aur. Lin, ne, Carrie, MA 2160 “Dow Chena Cpa 20. Br 8351, Sou Chk, WV 2503 gun Exes LC, PO, Bo 18, Hest, TX 725 “Ution Cid Caps, BO Box 8651, Sach Caron, WV 25303 Sh Cem Comps BO, Bos 138, Housag, TKTPST itu of be wer contamina roc of eb side wa conducted hy ArburD ile, Ia funding from, and er th pice the Ebene Orie Indu) ‘aoc prtofte american Chom Counc. Scant ‘perimental and cnc conrson were ao made by ‘af fom Sl Conical Westbalow Tcbaiy Cer Houston, Tas and non Carbide Corporation's Ressarch Contr South Gave, West Vrgbna. Unique fort ‘onder Kt for the reaction of een cide wth ter, ‘tndtinne ide uth ten has vr dred ave a. aed, asuore bint forte osc cf a ese xide tad toe decompstion of yon eke The ater data was cceperte to reaction model ful fr th determina. tin elon cde strage aby and presse lef "em ds dr water ontaminaton conan. secon “The reactions of ethylene oxide with water, and ethylene oxide with ethylene glycols, to produce Iagher molecular weight yeas are widely practiced tin the chemical proces indsres,Etylene oxide Iust be stored snd Shipped to meet the demands for this versitile chemical. Contamination of tis highly reactive chemical with almost any ther substance Should be avon, ‘A numberof incidents cesuking from contamina tion with water have been reported ‘432.00 gallen tanec containing 107,000 pounds ofa 60-40 weight percent eiyene oxidesver water Proce Safe Progress (Ve 20, No) (pit ~ 68) mimure, with an estimated intial empers tore of 23", was alowed to remain on a singin the plant for approximate 25 days. A severe runaway reaction occured, the safe rele valve malfuncooned, fd the tank car rupted de to overpresre. The ine from fist low through the salty eel walve C75 pie sec pressure) to rupture estimated 1,200 pi) of the ‘an car way seven minutes I ' 24,000 gallon tank car containing 193,000 pounds fof 1585 percent by weight ethylene oxide-brackish (owston Ship Chana iver water mise wi 4 high st content and assumed nial tenperatse of 27 pated alter 14 hours. The ak ca led a ocaton ‘whece a mixed layer of ethylene oxide-water would hve been in contact with the shel, The crumsances ‘of foading water ato the tank cr for cleaning sugwest ‘hat the wer was ler below the preexisting ety lee oxide in the tank car por othe Inadent. Temper ures at the lnterace wete high enough to cause the resaling ncdent (2 This fepod presen, ln deta the kinetics, heats of reaction, stoichiometry, physieal properties and ‘por lid equiva necessary to identi and mit tite hizarcous watercontamination reactions of et lene oxide and essentially neural pH water. The {aformation contained herein can also be used by knowledgeable persons to size pressure rele devices for inadvertent water contamination of vessels com taining ethene onemc one renee "To succesfully understand and model the behavior of mites of ethene axide and we, aging fom tute to neat etlene oxide, several chemical esc tions should be considered Thylene Oxide + Water ~> tylene Glycol Ethylene Oxide ¢ Ethylene Glycol -» Diethylene Giyeol Gand higher leas) 3. Btylene Oxide -> Actaldehyde 4 Ethene Oxide + Decomposition products 5. Bihvene Oxide > Bthyiene Oxide Polymer é [Bylene One Pohmer =» Decomposition products Gyeals-> Decomposition precise ylene Oxide and Water Reactions (Neural Meda) Ethylene onide reacts with water and etvene gy cols by addition polymeszation to for higher ei ne geal. As the mater concentration increases the Snset (detection) temperature forthe reaction inthe ‘Automatic Pressure Tracking Adiabati Calorimeter (APTAC® drops feom approximately 200° € (neat ethylene oxide) to approximately 60" C at 25-38 tveight pereent water ‘The detected exothermic onset temperature then increases aga a the concentation fof ethylene oxide becomes to low to stu 4 ree ‘om of suficentsate 2 lowe tempersture Monoethylene geo s produced by reaction of eh lene oxide with water Higher ethylene glyats ace formed by the mccessveadalton of eileneoxice to ‘welbrepresesed by 3 Weibull Nyeander dtu 3 Tavthis model a single activation energy describes the ‘temperature dependence forall diene oie sion, ‘ections, but ne rte i employed for characterizing ‘monoethylene glycol formation. anda different rate oefielent applied for forming il sibecquent gs ‘The rato ofthe higher glen rate cocfen to tat for rmonoethsiene gol delined a he Weibull: Njeane der valve. A C-value of two has heen previously Feponed (3 and confined inthis sd {Flory (Poisom dst predits that the ate of ac of thee seaione ate cal. A Nata dist in predicts thatthe ates ofthe successive reactions arediferet 3 "The eat of reaction of etylene oxide with wats is lower than the heats of feacton of ethylene oxide wih mono and higher glycols And, the neutral ey ne oxide eeaction with tater det toward a aciic pH and faster kinetics as the reaction proceeds 4, ‘The diference between unbuffered and buffered cite constants is only a few percent Buttered solutions ‘were nosed dung th suc [thylene Oxide and Water oF Ethene Geol Reactions (Acie Mediom) ‘Theses of reaction ofetlene oxide with water, and exnlene oxide with ethylene glycols a produce higher molecular weigh ethylene glycols by sdtion polymerization increase asthe pH decteases, al are ‘Sguficanty faster in an aide meu (pt 1-2) than "ales ina peal medium The Weill Nyeander 252, December 2001 mode! apps: Neural glycol Klnetes occur pall (othe adi-caalyzed kinetic, dhylene Oxide and Water or Faye Glycol -Reatons fase Mediu) The rates ofreacion of ethylene oxide with water, and etiviene ode with ethylene glycols to produce higher molecule weight ethylene glycols by atonal polymerization, ae significa ase in 2 basi mee {im than ina acutl medium, The rection rates, how tre are slower than i an adic mes, The Wetbul ‘Nycander model again apples Newtal glo! kinscs ‘ecurin pall othe base tly nei ihylene Oxide and Aqueous Sodium Hydroxide Reactions Blene oxide reacts with the water ntodced by contamination with aqueous sodiim hydroxide t0 produce monoetylene glycol This fexetion i accel. trated by he presence of sodium hydroxide. The Sodium then associates wth the glycols to produce ‘Sodium gheoltes, and edylene oxide reacts ith the ‘Sodium gheoates to produce higher glycol: The rate fof reaction of ehylene oxide with water to produce tivene ljcol is ulferent from that of etiplene oxide ‘wath sodium gheote to produce higher yok Ce 'Sconsistnt wih the Weibull Npcander ode, “Two liquid phases ayers) can form wien agcous secu hyehoaide contaminates etilene ene Kiet foe the teraur ate pial assaf listed ‘A tangule pase cages, snowy olen ie regions, willbe present as Rate ep Polymerization Inthe absence of contaminants, neat, commer grade, ebene oxige wn previous found undergo 3 fel-polymenzation reacion starting a approximately 200°C in an Atcelerting Rate Calocimeter(ARC®). A Subsequent decmpenticn reason fos 5 woscossascrenenos “The kinetics for the reactions of ethylene oxice with neviral © < pit « 9) water and ethylene oxide ‘wih ethylene slyeols to proce higher yeas, ave previously appeared inthe open bteratore 67,8 Determining the reaction osers ofthese kinetics has bbe heavy inuenced by the commercial concent UUons of water (6-15 water-to-ethylene oxide mass fati) used to maximize the production of eter iyeol. Pseudo frstorder kines wl ft uch cond tions, Incorporating the typical water concentation Into a fist onder kinetic expression rests in second onder kines, which also tthe experimental dts, ‘Unimolecular Expressions "The Iteature [4,6 9, 10) provides rate expressions foe reaction of etlene cide with water ote fell inate seoten{—E] ” Process Safety Progress (V2, No)where JRO/s the acy concentration of ethylene oxide Inthe misture, fis the frequency or preexponemal fc tor and Fis de acaton energy forthe reaction. Note thats expression docs not cotan a ter for concen trton of any reactant ater than eile oxide. Tiss deuat or solutions sth los concentrations of eth he bide wheres the inking escant, and gener Sy valid in commercial glycol reactors, where there Is freacess of water, This expression wil yield naceurte resus or solutons containing high concentations of fsjlene oie ‘imoteclae Expressions “The following expresson, proposed in the hers tunis more seal for soltions containing High con entrations of eaylene oxide r= Alco HO} ] = RT. a Most of the experimental data in the literature appear to be tom solutions with low ethslene oxide COncentrations Expressions for base: and aes ca lyzed eacsions are wpically given in the form: i @ ta =Ateoprj-£] ar Aris AO DAML SFE tan extensive calorimetie and modeling sugy was conducted onthe ethylene oviewater reaction 5) tem. Experiments Were catied outa the ARC and [APTAC aisbatie calorimeters om misties of eivlene xidewater, ethylene oxide glycols, neat ethylene ‘ide and neat glycols. The experimental dla were ‘nal using Superchents™, Yesion 4 In adton, Independent analysis of the experimental ta, 25 wel 5 confirmation ofthe derived Kinetic mode, were ‘Conducted using other computer programs (e.g SARE, Accelerating Rate Calorimeter (ARO® “The accelerating eae calonmete (ARC) san las ‘ment that can provide sel heat rte vers tempers {re-ime dita required for charactenzing eeactions The adiabatic environment is accomplished by cate tng the temperstore ofthe cell srmoendings match the temperature ofthe cell The AKC cza be Used 2 btu tnfornation on the thermal behavior of reac: tions and detected exothermic onset temperatures, marily or liquid-phase seacuve systems is aso Process Sey Pogess (E20, No“ ‘used for safety and/or performance evaluation of explosives and propellans. Self eat rates 2 or above ‘on Clmin ean be detected. This instrament (se Fg lure 1, deveribed by Towasend and Tou [11 4s known to provide thermokinetic data applicable to the design and safery perfomance evaluation of exc ors and sorsge vessels. Such thermokinetic data include * sate of sett heating, ¢ dlabadi time to maxim rate, 1 ete of pressive nse, 2 imax rate of rection, 1 Hinetic parameter, uch as preexponental Factor, scvaion energy ad eeacuon oraer) nd + het of reaction ‘The reation minture 10 be examined i introduced Jogo a epherieal eel witha volume of approximately 10 ml. The cells equipped with «thermocouple ‘outed externally on the wall Pressure is seceded bby means ofa wansducer Various cells which have presse ratings from 4500 pri to 15,000 ps, can be {sed As the pressure rating hai, wall Uickness for {cell increcs, the relative amount of eat alee Dy the’ cell use, a5 indscated by the phitactos, increases The tempersire and presse of tet wi 2 high phfactor peciic heat ofthe sample pls the test cell divided by the specific heat ofthe test ar ple) mst be corrected by established techniques 10 {he low phc-tactor concitions types! of many com ‘mercial operations The high ph lacor ean potentially ‘ask exothernte that can occur a higher tempers The significant advantage thatthe ARC offers over other calorimeti techniques sts high- sensi {exothermic onset detectabilty of 002° Cran. The ‘maximum scfchent rate i ited t9 approximately Automatic Pressure Tracking Adiabatic Calorimeter Grr, “The APTAC (ce Figure 2) combines the featues of the Design Institute For Emergency Relict Systems (DIERS) bench-scale apparatus Gow phi-factor and high eitheat aes) and the ARC Gow exces onset, tection) ino a single insrament “The primary use ofthe APTAC is obtain infers ton on the thermal behavior and rates of reachions Information obeined inches aciabatc temperature 2d pressure profes for actions. The base data 2 ‘manipulated to give selt-heat ard pressure rates 3s ‘nctons of empersture and te "The reaction nature 1 be examined is inodced Jnto.a spherseal cell wth a volume of approximately 130 The cel equipped wih a intemal thes couple which diceetly measures the tempersire of the feaction miature. Four independent heaters with, Dib cascaded contol maintain acabatie conditions Pressuze is messused by means of 2 wansducer. The calls placed ina fourlter, high-pressure containment ‘vesel fate to 2500 psig Non-sobarc pressure rack {ng by means of low contrl valves prevents the cell ‘fom roping. Tempest and pressre res of Up 0 400° C/min and 10,000 psvmincan be tracked, The Deere 2001. 255Figue1. The accelerating « Relief Valve Magnadrive stirrer Exhaust gure 2. Aucomatic pressure tracking sdabate calorimeter CAPTAC™)APTAC ie low phifictor instrament (ess than 115), ‘pith exatheam detection eapabites of 04" C/min 12 superchems Computer Ruperchems 13, 14, 14) san advanced (ool for ncrmal hazards Assessinent, pestire reli design, lind consequence analysis. The DIERS Eton ist: idred to perform dynamic simulations of runaway ‘hemes! reactions, and design emergency teil sy tens for two-phase, vaporliquié lows ran suet She experiment closer dats ae summarized infer 23 and 4 Calormeny dat ete collec inthe ARC ad APTAC on four Sten {Eden one and water, 2 iene oxide an pols ono and 8) 5 Neatetsiene ode A Net alveoks The eiylone oxide-water tee (oe Table 1) were calles overa wide range of pi factrs and ety fe cnide meth aches (9 209 weight 90. Rep SSnive sithest ted data from Selected tests ae Sow in gute 3. These tes were conducted using Side ange of tert cel sates of constuction Including tts, Haslloy-C, ad gle. Detected rset tonperatures measured ingase were higher {fam tow metsured in anlam or astly.C pe {nll orconcenmted soln of etilene one snd Satovyook tuscan be sebuted to posible rae Dita the cllwal toma inet ccs ‘Ahhiong detects exodbermic onset temperatures acuta sseting the scopy fs tem ‘Sfunimay texcion experiment data must be acd ‘elvan appreciation othe limits ofthe mesg Tistment Adiabatic calorimeters employed for "Suing heal subi can innodce ast, {SSuiiy and henna nea cect abun is dete th rscton of he reaction orgy blag rnc i sample vers the cel a2 itvant An sdabt ervknument son n which he ‘ican het is nether ow fom nor ene ed ‘Name bom na Eertey appara Saab 8 thelr suse of eat ose the emonmen. “cosy ithe sblty of aerator instrument to manure (tect) the thermal parameters Const temperate, et) ofa ronaway heme faction. For ‘Supe, se ARC ha detection limi for messing Stoo emperature comtesponding 1 0.02" Cin ot SC day" obeousiy the measured onset temper tre woul be mach ta high to ety use i make Ingesey determinations. The messed daa and the Pacis sf hemalexpionon thoy aul be used Sime the temperate exeapelatons equed for ‘Stay reed dione “Female she rt of te al heat of ec seat the eat sry he een ny {torino fabio apse 2TScan heat sinks teal neta therefore te ‘hesmnal capac ofthe system ample pas sample Som) ced byte theta capac ofthe sample Se, Theron inet ase fet of darpening he Iagoiudeof an obsews rnaway reaction When he Process Slay Popes (WoL 2, Now thermal inertia ofthe laboratory instrament i higher than the plant-scale equipment, the measured onset temperate can be too high, the rte of reaction too Slow. the adiabatic temperature rise too low, andthe pressure rise fo0 low compared to what would be ‘served ins placate enviorent Aga, the mee ‘Sire dita and the pons of thea explosion theo- fy should be used oma the parameter extrapaations equed for safey-eated stds ‘Accondingy,expersmental results cannot be used lecly a8 2 measure ofthe appearance of reac {he time-to-manimum rate the vate of feacuon, the ‘Miabatic temperature sis, or the maxim sll beat fate in commercat-cale equipment Consult quliied Personnel for interpretation of the daa and gundance [ithe analysis of ermal ablty runaway reaction behaiorn process veel Figure 4 lysates the impact of ethylene oxide ‘weight action onthe detected exothermic onset tem peritutes, The detected onset temperature goes through a minimm around 33, and tends towards the value measured for neat elkylene oxide at 100%, “Oral ofthe compotion teed the 2538 weight 9 compositions have the lowest detected exothermic fnsettempertires.Eliylene axide-water solutions In thts compostion range have sifiiet energy to seach the glycol decomposition temperature range of Lppectanutely 30? C. The glycol decomposition reac tons are disci elsewhere in tis eport and mest Special consideration Because they are Very energetic fe proce lange amounts of gs “rhe eylene one and gots tess (see Table 2) sete conicted mony th tant test cells, The eb ene oxide composition ranged from 1610 61 weight percent. The primary objective of conducting thee lene oxide-gvcos tests was to extract specie rate Sta on the ethylene oxideglycal reaction. Once this ‘wis eatublished the impact ofthe reaction of etiylene ‘rie with water was eal quantified ‘The ethylene oxide-water system isa highly non {deal vaporliguid equilibrium system. In order to propery interpret the experimental calorimetry dat, Ey emulating the acl fests and scaling up de dat, the miature PVT behavior must be well epresenred ‘The vapor liquid equim dara for all ofthe binary ‘Stems of interest were oained 2s [llows Collect published iteraure data on all the systems [Experimentally measuse the vapor pressure data of pure alot nthe APTAC. |5 Perform VLE thermodynamic consistency checks 4 sig the Gibbs Duhem relation Fit the binary interaction parameters using Soperchems Exper wih the Nelhem modiieaton ofthe Peng-Rabinson equation-of state) Experimental measurements ofthe vapor pressure curves forthe pure glycols at low temperatures ace necessary becnse they decompose at elevated tex perature The choke of ciel properties is important‘able 1. APTAC™ rxperimental calorimetry data sommary forthe ethene oxide-water sytem, Ta [Teer] Tam | O Mir] Taeta number | 6G) | “eS | Mae omnas 108 woy)—TiraNNT aontsa an] io | 529 | TIraNIUM Aonils son | x sem | TITANIUM cwaizs | ss00 | to | 109 so | _TITANTUM rnwyzies | 55268 | oe | 133 09 | HasTELLON-c coaomz | coas |e | 132 4842 ‘Lass ADoI2e en | 30 | 130 haus | rraNUM ‘aop123 son | ie | 130 ne | TitANUM Cacaosig | ose | Set | 138 an GLASS ‘00170 me | 5 | 138 ais | aman ‘ores ao | Na | 13 258 | TITANIUM Cuansise | saa | NA | Lae nas LASS pois |11209 | NA | tet anss GLASS ‘oar tise | ona | 139 2128 | tran cmansio 387 | NA | te 236 GLASS ‘os gam | NA | us 97 | TTANIUM penn me] _a__| to wwnt_| _TmaNum Thermal inertia “Tp= Detected onset emperatre 06°C: cect to 0 of “pa = Measured maim ree enperatie; NOT soeted fo thera nea Me™ Toa sample mast of lylone ore und water Altes were sed Tobie 2. APTAC™ Experimental calorimetry data summary forthe ethylene oxide. goo ystems Tet amber [Toes [| Kip | Mao | Meo | Mies | Mas [Meas | Tora cc wi | wo | @ | w | wr | | Masia NWT fosrr [12] teat | 207] sr sy [TiraNst Aooisn [44a | 14 | 2183 | 1247 | 4438 Seas | TITANIUM onsen [6420] 114 | 2346 | 1299 | ar 58.2 | TIRANIUM mnwy2i20 |7357] 114] 2500 | 1500 | 48.0 {000 | TITANIUM comaaris9” | 5803] 114 | 2802 | isa | 3500 (ou | TITANIUM nawy2iis | 7003] 114 | 2526 | isi | 4070 S081 | TITANIUM aNWY24I9 | 169] 113 | 2678 | 1602 | $843 | 6265 | TITANIUM JcMaosa39p | 6237 | 127 | 3347 | teo0 | 3130 | a0 | GLass aooisi |$830| 146 | 3sa7 | Leo | 3347 S137 | THANIUM faoorsia |7629 | 115 | 463s | 2667 087 5754 | TITANIUM ‘ooisea [rss | tis 489 | 29.77 | nas ‘oan | TtaNIN femnotose | 6008 | 120 | .22 | 2105 | anos 110 | TITANIUM eaizzas |6776| 113 | sooo | joss | dass 6170 | THrANIUM emaoiii9 |640s | 130) sooo | iso | 1840 2820 | TTRANTUM. muwv2air |s574| Vis | 5259 | doze | 2748 5796 | TIEANTUM aooies— |39s| 129) ose | isa? | 198 hos | HIKANIUM ‘oman [sis] 114 1688 | in8t ss20 38) TITANIUM ‘ooiss [225 is | 3221 | isos no S604 | TITANIUM soos | s31] 145 | 3216 | 1223 | 20 79 | 3805 TITANIUM © = Theo insta ‘y= Datei ns! temperatures 0.04°Cmin; eet 10 af Mya = Tol trp mess Alen were sro 236, December 2001 Proce Sefey Progress Vol No.)Table 3. Experimental clonmetry data summary for ne st ethylene oxide Tea Tet [Tom | OY Meo Tet Car Nanber_| Type } | Materia ein ance | 37-| $a7 | 30} HASTELLOV.C coss09 ace | 25 | sis | 326 | uastéLLoy.c £03269 ‘arco | 24 | 483 | 60 | nastettoy.c cours ance | 23 | 59 | 475 | wastetLov.c cmoai9a | atace | 210 | 146 | 3000 GLASS, woos | xptace | 210 | 129 | 3200 ‘TITANIUM awios | xptac: | 258 | 144 | 2000 {ITANIUM awior | xptace | 240 | a7 | 2000 “HITANIUN ‘woos | xptace | 283 | tas | 2000 TITANIUM ‘Awoiion | aptace | 247 | a7 | 1937 ‘TITANIUM awa | sptace | 265 | 147 | 19.60 [TITANIUM | > Thermal inertia ‘To Detected onset temperature at 0.02°Cinin (ARCS) or 0.4"Cinin (APTAC™) AILAPTAC™ tests were tine all ARC tests were wistined Tobie 4 Experimental calorimetry dats summary for neat alycols Tat Tat [The | te | @ | Sheol | Mw | Testa wumber_| type _| ee) | 0. Type | (gh | Matera ‘Guaosii98 [AprAc%[3I4 [333 [10 [Mono | 40.08 [TITANIUM ‘EMAOSII9A |APTAC™| 340 [303—[ 1.21 [D1 | 4001 — [TITANIUM Siso9 | APTAG!| 365 [32921 Di [40.00 | TTANIUW. ‘cmaosi@o—[aptac™[ sai [312 [1.20_[ta_—_[ 40.08 _[TTTANTUM. ‘APTAC™| 340 [328 [1.20 [Tera — [4095 —[ TITANIUM ‘ARGB_[330 [304 [1.34 | Mono [402 [TITANIUM ‘ARGe [350 [27 —}136 [pi [Aon [TITANIUM ‘ARo# [340 [305 [134 [te | 400 TITANTUMT ‘AROS [340 [3121.88 [Tera [400 [TITANIUM © = Thema ineria “To = Detocod onset temperature a 0.02°Ciin (ARC) or 0.04°Cinin(APTAC™) Detected one empertue at dP/dt = 0.1 pti A APTAC™ toss were ted and conducted a 10°C increments AARC ents were unstired for expesimentl data interpretation of the ehylene Ethylene oxide-nater—k=-01044, 2." 00895, fide-watergivol reacting mixtre phase behavior. Ethene lyeolwater_ Ke =-00883, 0 Tolusrae the highly am ieal phase Behavior of ethylene icolwater | Ka 0107," = 0 the system, Figures 5 and.6 show 4 comparison Trethylene gical water K-08, 2-0 between experimental VE data at O13 bus fore? Ethylene oxide ene oxide water, and model predictions of the data ethylene glycol kro, no Using deal behavior and the Peng-Robinson equation Ethene oxide Df stae. An del phase behavior assumpricn will lead cletiyene glycol k= 00196, 2-0, 1 the wrong prediction ofthe vapor ga spt the ‘easel ands ava resule to an eaoncous energy balance EMNENEODE WDAEONS RACE OE 2nd tempetatute pressure Behavior Thin tr, can ‘Rudo incomeet predic 0 ele! ster action Lireature yelerences for various component pair ‘vapor igus equa dota ste provided in Tale 5. "The binary interaction paranors used in the Hal simone ave a follows Proce Sey Pogess (Vol No? "tne data presente Recin se believed to cove the widest range of concentations and eeacion temper ures avallable ithe open ikerture. The publised Kinetic models presentee ner Experiment Appar: ‘sand Dats Anal Software, however, do aoe ade ‘quately Be our ene data St (ee gure 4) In papers December 2001. 237Shesornen ace ‘Sow come mone inencata | || Eh 2 St ‘ua Eom ou AD. ‘aie toon ett Cini on AS 100007) Figure 3 Selcheat rate daa for sleced ethylene oxide-water APTAC™ tess. wei e0 Figute 4. comparison of measured and predicted onset emperatures (Foe a specified self heat ate) asa funtion of reaction oder fr the tyne oxide-water sytem, by vinanen 7 3) reaction kinetics wid higher order Water dependencies were developed on mechanistic srgunentsand ito expormeyal data. We have cho Sen to expove higher oder zenction modes that bear some simdactes to Vitanen's (auch as an overall fourhvorder model dependence), bat which so con 288, December 2001 "ain some key distinguishing features, The ace and base catalyzed reactions of ethylene oxide and water to produce eslene glycols wil be covered in seps ‘ate proposed papers “These kinetics were fi by the following smltion proces Process Suey Progress (oL20, NOable ,Litsature sources for thermadyamlc component pair dats Wee ee [ra] [epee | TEs aD = = Es qo aa ae SS ST teas mca ceri de i: i tH uy Figure 5. Comparison of experimental TXY data and Heal behavior predictions forthe etiylne oxide- ‘Nate pom 91013 bare ea able 5). 1. Select reaction orders and activation energies 25 Savung estimates. 2. Simulate the ARC/APTAG tests, including the heat- ‘walesearch ses 11 3, Compare the model predictions of temperanuse fie ate temperature, pressure ise ate-emper Ture, pressure cemperature,tempersure time, ad prestre time wo experimentaly-neasued vales 4 the node predictions donot agree ith dhe mea ‘ued ales, go hack to step 1 cerwise op, Panicuar emphasis was pliced on the ability of the smedel to preget the detected onset temperature both on” fn high concentrations of eilene oxide. Reaction orders of one, two and thee were examined fs candidates ‘These models performed reasonably ‘well inthe lowe concentration range, but underpredict the desecced oncet temperatures at high ethylene bide concentrations The over ourtvotder reac tion sate expression was Found to accurately predict the reaction Kinetics of eylene oxide with neural ‘water (beeen 10 and 90 W196 EO), etiylene oxide trith ethylene lycol (between 18 and 61 Wt 96 FO), 3nd some explorsory rons wih daelene glycol Kinetics idols Reaction Ethylene oxide eeacts with water inthe fiquid phase to produce monoethylenepiveat according 10 Process Sey Progress Vol, No) Figure 6. Comparson of experinersal TAY dat and Penge Robinson equation of Sate predictions forthe lone cece pets 1003 as (See Tae 5) the following soichiomery: GHO+ 10 GO, ° © align 1 Ren ‘The rate of reaction is given by the following expresso = Rls irl tROEEY o where nis in kool Sefined und isan Artes factor a aenf-E | tr so on ee eh eeeao ROM = 110) elRGt + DEG + TRA saa + 2580 + ee ap ‘The value of the preexponential factor 4 has an uncertainty of #108. Note that the [ROT teem atcount fo the total molar concentration of indepen Sent hydroxyl groupe in solution. The units the above expressions ate in SI, ie, m, kml, 5, te Te ‘mporant to emphasize that in Viunen'staéchanisic evelopment (7 only water wa considered to asc ste with eiiylene oxide. Our ova modeling more Sccurate beesuse it considers al species containing Ihydeoxyl groups. Moreover, the curent study also takes into account the impact ofthe subsequent reac” Hons ofthe ethylene glycols on the overall Kinetics resus. ‘Gea! Reaction Tylene oxide reacts with glyols to make higher- ‘order alycls and polygiycls. The reactions a ey ‘ene obides wth glycols secur based on the following ‘Scichiomexy” GHO* Carats, Ce.2, “me reactions Derwcen etylene oxide and they col proceed according tothe following expression as 6 ap p= PRUE Cay gO) IROHE, 62,6 “where hand [ROM ate defined aries. Mechanistic sification “These leinetcr are consistent withthe following ‘wo reactions scsrng in sequence TEaiylene oxide i assumed to rapidly associate with ‘so moles of hylrony to form an intermediate, 0s 2R0Hte 1 ao “The concentration of ROHis defined in Equation 11, The glycol concenuation terms la Equation 1 are ‘multiplied by two to accoune for the presence of to Ihydroxyl groups ea each alveo! molecule “This reaction s presumed tobe s fast that chemi ‘al equilbium is atained Hence as ‘where Kisthe equillrum constunt for this reaction 2. This intermediate is presumed to reat slowly with ‘water to form EG, wth #6 to form DEG, with DEG {0 form TEG, etc, regenerating the two moles of OH. 240 December 2001 Trg” MyBO ‘where the subscript EG implies he formation rte From 10 and does not evaluate dhe consumption of EG in subsequent reactions. IF BG being converted to DEG, the following reaction and retction rate appr 1+ 86 DEG+ 28OH ry an as) ogo BNE a9) ‘The facor of wo applies because ether ofthe rw haydrexyls onthe ethylene glycol ean undergo reac Lon. Equations 18 and 19 apply sn the exact same manner forall the heavier gheols. The Inermedate oncentrton [f has not heen measured in ths dy, btean be expressed per Bquation 155 b= Ktean won en ‘This expression for [can be substituted into Eau tions 16 and 19, which Yields for the reactions ‘esrb herein as By REG HO ROH? en ‘ims * 2h ECE ROME cco Iris convenient to combine & Kas cod es ‘The final ate expression is pve by the following equations ag ~ MEO H01 ROH Yoag" MEO) LROHE ‘Model Comparisons "The aggregate dependence on species concen astions is fourth otder The samme energy of activa tion and the same preexponential factor were found to predict adequately the onset temperanure and heat ate for both the hydrolysis reaction of tihylene oxide with Water, 2nd the ethoxylation feaction of ethylene oxide with glycols. A single Fourthorder reaction model accurately predicts the experimental results ‘Consideration was aso given to thind order reac. tion model. thiel-order reaction snechaniin implies that only one hycroxyl group associates wi ethylene ‘oxide (o form the intermediate vs. two hydroxyl |roups in the fourth-order reaction mechani ‘A comparison of predicted onset temperatures 006" Cin heat rate) fora doer reaction anda Tourth-cederreacson Versus Weight percent ethylene foude Is shown in igure 4 The expenimental detected ‘onset temperatures (unadjusted for thermal neti) eo es Proce Safety Progress (V2, NoTeble 6. Measured heats of reaction fom calorimetry dt, Siena “ie TTT Name (Bike FO) | (@TUM EO) iiyleacaxle (FO) | w4058 Wane 0) | 1016 tylenegyeol EG) | 62.068 2068 7 Diciyien gest (DEG) | 106.122 208 oe Tieiyene heat (TEG) | 150.175 as] at ‘etetylene gyeo! 4EO) | 196208 Zato| —“kons| Pataethylene een (PSEG) | 3820 aia] “Lon Hecatylne yet (HEC) | 28230 ea) The Polytnene gol 25 aio | “toro tithe st of oman 225°C ‘Suburp ane ii pase ‘Sebecht gids gt pase sth tea eat vaporization a 23°C Table 7 Comparison of calculated formation energies from APTAG data wit reported erature vals Chemical Phase | AH; | Accuracy (v/-) [Source ‘Name (etikg) |__C4/Ke) Ethylene glycol Gas} 6251 32.22 | NIST Review Gas | 6354 45.11 | NIST Review Gas | 6272] Unknown | DIPPR Gas | 6.172 ‘This work Liquid | 73301 1950, NIST Review Liquid | 7.411 45.11 NIST Review Liquid | 7322 ‘This work Diethylene glycol Gas | 5382 161.46 | DIPPR Gas | $256 ‘This work Liquid | 5,923 2.42 | NIST Review Liquid | 5,931 ‘This work “Triethylene glyco! Gas | 4828 14484 | DIPPR Gas | 4.710 ‘This work Liquid | 5,406 ‘This work Tetracthylene glycol | Gas | 4,543| Predicted | DIPPR Gas | 4,486 This work Liquid | 5,085 23.68 | NIST Review Liguid_|_5.127 "This work ‘were added to the plot ia order to show thatthe Fourth-osder seaction mechanism predicts the data tore sccutstely pariculey a high ethylene oxide Concentration. AL 10 to 40 wt ethylene oxide cos entation the rites of tesctton fr a thic-oeder seac tion and foumtrorderreacuon are very sil. How vee a divergeace occurs once the concentration af fthlene oxide is greater that 40 01%, 2 pity ploc appears ia Figure 9, consiing of ind vidual points from the observed Selirheat rate data Prces Safety Progress (Wala, No) ‘between onset temperature and up to 200° © (or the pesle heat ene) se compared to de slebeat rates cl Slates from the mode. Excellent agreement is seen Derween the ates measured and those predicted rom the propased kine mode Hest of Reaction Data ‘Tae present model terminates to an oligomer with a molecae weight of 326 (see Table 6). The complete Sciles of ethylene onde addon veacions ae defined December 2001. 241Predicted SeitheatRat, GTi (*Cinin) Measured Set heat Rate, Ti (°Cmin) Figure 7. Pasty poe of setter rates measured in the APTAC™ vs, elPheat ates pred mel Figue 8. Nessured vs + pressure data for ethylene glycol and diethylene glycol 22 December 2001 Process Sey Popes (Va 20, No.8Tobe 8. Gas chromatograph characteristics. [ Instrument Tab 1 Tab? Tastrament HP 5890 HP 5890, Injector Split (200:1 ratio) Injector Temperature (°C) 250 250 Column Type J&W bonded phase | RTX S-Amine DB-4 fused-siliea Column Length ™ 30 30 Column Diameter (mm) 032 032 Column film Thickness (mm) 10 10 Carrier Gas Helium Helium ‘Carrier Gas Flow Rate (ml/min) 30 35 Pressure Ramp No Initial Temperature (°C) 100 38 Ranyp Rate ‘Cimin) 10 15 Final Temperature co 260 220 Retention Times (ain) EG 20 67 DEG 43 n3 TEG 89 149 ‘T4EG 126 188 Detector HD HD Detector Temperature (°C) 300 300 Table, Measured va prediced produc dsebuton analy for selected ts, ‘Weigh Fanaa 7) | Te Hae oma —[—Awoieo [Acute NWT | WET Siting coaposion tiylee oxide ssa 218s 1sa6 isa. 2856 ‘Water a ‘000 aus boo 00 Xie heal 00 701 ‘a0 sis__| naa ding cinposton | Mae Pra Nas’ Prad| Weas/iod | Mes red | Mens Ped as1i282 | veavooo | m20/79 | Navo0 | NAINA ovate | arsos | isms | ears | s6or9s9 pariauy | 313/363 | 330/300 | 302/22 | 335/376 foarte | 9arins | “2ar0s" | goo/720 | iserns Terschecesieo! | 290/390 | TEG, and 2Propanol = TG," Ind. Bag. Chem. Res, 37,3, Pp 1ce5-169, 1998, ‘Hu, H. ad ¥. tw. "Equation of Sate for Polymer Systems Futd Phase Buia, 150-151, 9p 687 or. 188, Fllass, A. H. Modarress, and G.A. Mansoor, ‘Measurement of Activity of Water in Aqueous Poly(ethylene giyel) Solutions (eect of Eacest Volume on the Fory-Hugsine cParamtetes) (Chem rg Dat, 341, pp 5255, 199, Merskowiez, Mand Mt. Gottlieb, "vapor Liquid Equllbum in Aqueous Solutions of Various Cly- foals and Polyethylene glycols), 3. Polytesyene fByeols)" hem Eng Data, 30, 2, 9p 233236, 1985, Virtanen, POLL, “The Variation ofthe Rate of "yelp of Fyienc Orie mith the Water Con tent ofthe Solvent” Suomen Kemisilcht 385, 7 Spo 133-142 1965, Leting J.C and 1.6. Fisher, Runaway Reaction Charscierzation’ A Round-Robin Sey of Three ‘Adalonal Sjsems" ueational Symposium Runaway Reactions, Presure Relief Design and [Afluent Handing, GA. Melnem apd 6. Pst, iors pp 109-134, AICRE, DIERS, New York, Ny, 1908 Fisher, H.G.,et ab, “Emergency Relief System Design Using DIERS Technology—The DIERS Project Manual,” AICHE, DIERS, New York, NY, 190, Gutdelies for Pressure Rel Design and Epluent Handling Systems, AICHE, Center for Chemical Process Sey, New York, NY, 1958 Process Sar Progress (e420, NO)