J. Chem. Chem. Eng. 9 (15) 176-19 do: 1.1765/1934-7375/15.3.3 D DAVID PUBISHING Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and Extra-heavy Crude Ol Upgradng Technology Rcardo Sardella Palma 1*, Zacarías us, Pava Mguel and Medna Henry 1 1. Infrastructure and Upgradng Department, PDVSA-Intevep, os Teques-Mranda 11, Venezuela. Refnng Department, PDVSA-Intevep, os Teques-Mranda 11, Venezuela Abstract: Bubble column reactors are multphase contactng devces used n a wde varety of ndustral applcaton. Inrtevep S. A. s workng on developng technologes to convert heavy and extra-heavy crude ol usng ths type of reactors. Volumetrc gas hold up, flow pattern, average gas bubble sze, average nterfacal area, RTD (resdence tme dstrbuton), dsperson coeffcent, Péclet number are mportant desgn parameters for a proper scale up of them. Several cold model experments have been proposed to determne the prevously mentoned parameters at atmospherc condtons, usng a plexglas bubble column reactor at plot plant scale unt (1 cm dameter). It was also evaluated our own desgn of nternal trays (plates) n the reactor. Ar-tap water and ar-lght ol systems have been used. A wde operatng condton range was appled, superfcal gas velocty between.5-1 cm/s, lqud flowrate between 15-65 l/h. Generally speakng, workng wthout nternal trays was found that gas hold up ncrease along the reactor and t was possble to dentfy heterogeneous bubble, transton and turbulent flow pattern areas for the ar-lght ol system. Average gas bubble sze ncrease along the reactor at bubble regme from -5 mm but at turbulent regme, stay oscllatng between 1-3 mm. Average nterfacal area ncreases exponentally wth superfcal gas velocty at any reactor heght, tll 1,41 m /m 3 for the ar-lght ol system but, at bubble flow regme, the average nterfacal area s lower than 1 m /m 3, whch negatvely mpact the reactor performance. Internal trays n the reactor always ncrease gas hold up at any condton or system used. Resdence tme dstrbutons curves, Péclet numbers and dsperson coeffcents founded, show that ths reactor wth ths knd of desgn nternal trays stll tends to be a complete mxng reactor under the operatng condtons used. Key words: Flud dynamc, bubble column reactor, resdue upgradng technology, trays. 1. Introducton It s well known that bubble column reactor have a wde range of applcatons such as absorpton, catalyst slurry reactor, boreactons, coal lquefacton, etc. These reactors are preferred because the smplcty of ts operaton, easy constructon and low operatng costs. These man characterstcs make bubble column reactors an excellent choce for extra-heavy ol upgradng technology as Aquaconverson. Venezuela has extensve reserves of heavy and extra heavy crude ol n the eastern part of the country, ncludng conventonal producton areas and the Ornoco Ol Belt. The Ornoco Ol Belt s dvded n * Correspondng author: Rcardo Sardella Palma, magster, research fled: flud-dynamc. E-mal: sardellapr@pdvsa.com. four bg areas named as Carabobo, Ayacucho, Junín and Boyacá, and has 97,7 MMMBS of proved reserves of extra heavy crude ol 77 MKm approx. The Ornoco Ol Belt crude has very hgh denstes (5-9 API (Amercan petroleum nsttute) gravty) whch reduce hs market value and a very hgh vscosty (1-4 Kcst at 5 C) whch mpact the transportaton cost. Ths crude also has hgh sulfur (-5w%) and metal content (4-6 ppmw). Producton of these extra heavy crudes wth gravtes of 8-9 s currently done by dluton wth lght crude (4-3 ) n the well to be lfted to the surface, where t goes through a process scheme ncludng degasfcaton, dehydraton and desaltng to be commercalzed as a dlute crude of 13.5-16 API
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 177 gravty and a vscosty less than 35 cst at 5 C. The quantty of dluents could be around 38%-5%v of the total producton, dependng on the lght crude ol used. One of the man constrants n the current producton of extra heavy crude n eastern Venezuela s the avalablty of lght or medum crudes to be used as dluents. Ths stuaton wll be even worse n the future due PDVSA am to ncorporate more heavy and extra heavy crude ol n ts crude portfolo to reach producton capacty of 6 mllon BPD (Barrel per day) by 19. In order to overcome the constrants mposed by the potental shortage of lght crudes requred as dluent, PDVSA-Intevep has developed a new low-captal-cost technology for moderate converson of resdue and heavy or extra-heavy crude ol called Aquaconverson. Aquaconverson s a thermal catalytc steam crackng technology developed by PDVSA-Intevep to convert resdue and heavy or extra heavy crude ol n the producng feld nto transportable upgraded crude, reducng the use of lght/medum crude as dluents. The process s also utlzed n petroleum refneres to ncrease converson of vacuum resdue wthout losng unconverted bottom-product stablty [1. In spte of the advantages, bubble column reactors used n a technology as Aquaconverson are dffcult to desgn and scale up, because of the flow regme complexty, and ther unknown behavor under dfferent sets of desgn parameters such a dameter, heght, etc. Nonadjustable parameters lke phase holdups, gas-lqud nterfacal area, nterfacal mass transfer coeffcent, dsperson coeffcent and heat transfer coeffcent have a drect effect on the desgn of the bubble column reactor. Therefore, ths study presents some expermental results to understand the flud dynamc behavor of a bubble column at plot plant scale, workng under smlar Aquaconverson operatng condtons of lqud and gas flow rates, but, usng atmospherc condtons of pressure and temperature.. Expermental Sectons.1 Expermental Set Up A schematc dagram of the expermental apparatus s shown n Fg. 1. Ths unt was conformed by several sectons. The lqud secton had three tanks, one for recyclng tests (gas hold up tests) another, for one through tests (dsperson coeffcent test) and the last for dsposal of lqud materal. Ths secton also contaned the pumpng system. The gas secton capable of feedng ar at requred condtons, from the ar plot plant network nto the unt has a pressure control valve, Brooks rotameter and an Endress Hauser mass flowrate meter used to accurately data measurement. In the reactor secton was used a 1 cm dameter plexglas column whch s geometrcally smlar to the real Aquaconverson reactor (see Fg. ). The bubble column s a plexglas tube wth nner dameter of 1.3 cm, the cylndrcal heght of 16 cm, and a cone-shaped bottom also made of plexglas wthout gas sparger. The specal desgn cone-shape column top was made of metal and t was bult as the real apparatus. The column was equpped wth a Rousemount 115 T DPT (dfferental pressure transmtter) whch s connected to ABB Screen Master vdeografc regster to gather the column pressure drop. Ths unt had a pneumatc lqud tracer system whch s used to nject both tracers used n ths study, Thophene and Potassum Dchromate. Those njectons were used to determne the dsperson coeffcent n the column. All tests were conducted at room condtons of temperature and pressure. Two dfferent systems were used n ths study, ar-tap water and ar-lght ol. The lght ol was chosen because of some physcal and chemcal propertes lke, densty, surface tenson and vscosty, whch has to be smlar to the feedstock (vacuum resdue) at operatng, condtons (18 psg of pressure and around 43 C of temperature). These man lqud propertes are shown n Table 1.
R1 178 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and Ar compressor D 3 1 1 1- Plexglas Plexglas column column - Ball value Ball valve 3 3- Check Check value valve 4 4- Rotameter Rotameter 5 5- Mass Mass flow transmtter flow transmtter 6 6- Dfferental Dfferental pressure pressure transmtter transmtter 7 7- Mxng Mxng tank of tank slurry of slurry 8 8- Dscharge Dscharge tanks tanks 9 9- Recculaton Recrculaton pump pump 1- Feedstock pump 1 Feedstock pump 6 5 4 D D 7 8 1 9 Fg. 1 Schematc dagram of the expermental apparatus. Fg. Plexglas reactor. The upgradng process use steam as gas phase, but ar was used because of avalablty and economc aspects. Table shows a comparson of the propertes of the two gases. Akta [ used water and four gases of dfferent denstes, ar, oxygen, carbon doxde and helum n a column wth 15 15 cm square cross secton, showed that the effect of the gas densty on the gas hold up could be neglected. Nevertheless, n very hgh pressure systems, n whch the gas densty can be affected n a hgher proporton, the gas hold up, could ncrease sgnfcantly wth the ncrease of the gas densty [3. A wde operatng condton range was used, n order to reproduce the actual condtons of the plot plan unt. Superfcal gas phase veloctes between.5-1 cm/s and lqud flowrate between 15-65 l/h (lqud phase velocty between.35-.15 cm/s). Three dfferent flowrates were used, 15, 35 and 65 l/h and for each one, a set of superfcal gas velocty were tested from 1 cm/s to 1 cm/s wth 1 cm/s steps. In some cases, t was necessary to operate at.5 cm/s of superfcal gas velocty. It was tested nternal trays nsde the column n order to defne ther mpact on the gas hold up and lqud axal dsperson coeffcent. The nternal trays
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 179 Table 1 qud phase propertes. Flud Method Vscosty (cst) Densty (Kg/m 3 ) Surface tenson (dyn/cm) Vacuum resdue PROII 1.7 843 16.63 ght ol Whhelmy plate -3.5 at 4 C ASTM 815 at 5 C ASTM (D-445) (D-198) 7.36 at 5 C (ab. E&P) Water Perry, 199 1. 996 7.8 Table Gas phase propertes. Vscosty Mol. Flud Densty (Kg/m 3 ) Surface tenson (dyn/cm) (cst) weght Steam.5 18 3.69 4.93 Ar.19 9 1.18 1.18 consst on 1 perforated trays wth 3 mm dameter holes. (Fg. 3). The number of holes change, at the bottom s around 3 holes and ncreases along the column tll reach 64 holes at the top, n order to keep gas velocty constant n the column trough the holes because the gas formaton nsde the column when ths nternal tray s used nsde the real reactor, followng typcal shell nternal trays desgn used n vsbreakng commercal reactor.. Methods Ths part conssts of the descrpton of the procedure used to estmate gas hold-up, flow pattern, dsperson coeffcent and average gas bubble sze and average nterfacal area...1 Gas Hold Up The unt was operatng n recyclng mode. Two dfferent methodologes were used, the frst one consst on determnng the pressure drop n the column and the second methodology was based on phase trappng and measure both phases nsde the column. The pressure drop s determned by usng a DPT n the column. The dfferental pressure profle s deduced from a one-dmensonal bphasc model, where frcton losses and flud acceleraton are neglected. Then, the axal pressure profle n the c olumn can be descrbed by the Eq. (1). Fnally, wth Eqs. (1) and (), t s possble to defne a total gas and lqud hold ups n the secton where the pressure drop s determned. Ths procedure was done n order to Fg. 3 Internal trays tested. determne the total and partal (top and bottom sectons) hold up n the column. Both sectons represent almost half column. dp ( G G ) * g (1) dz where, dp s the pressure dfferental, dz s the heght dfferental, ε and ε G are the lqud and gas holds up, respectvely and g s the gravtatonal constant. 1 () G The phases trappng methodology conssts on stoppng both lqud and gas flow rate sent nto the column and closng the nlet and outlet reactor valves to measure the heght of both phases n the cylndrcal secton of the column. Then, t s possble to determne total gas hold-up as a relaton between the gas volume and the total column volume, as t s shown n Eq. (3):
18 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and V V ctop h G (3) VT where, V c-top, V h,, V T are the volume of the cone at the top of the column, the obtaned volume at specfc heght and the total column volume, respectvely... Flow Regme A pneumatc system was used to nject nto the reactor a pulse of a lqud tracer. Thophene and Potassum Dchromate were used as tracer for ar-lght ol and ar-tap water systems, respectvely. When the pulse s njected, the systems stop workng n recycle mode and the dscharge s sent to the dsposal lqud tank. Reactor outlet s sampled along the tme n order to quantfy tracer concentraton; sulfur concentraton (Tophene) by WD-FRX usng ASTMD-6-1 norm and Potassum concentraton (Potassum Dchromate) by ICP (nduce coupled plasma). Injectons were used to determne the RTD (resdence tme dstrbuton curve) for all the cases. evenspel [4 defned a correlaton between Péclet number and the RTD curve, as follows: 8 t Pe Pe (4) where, Pe s the Péclet number, σ s the varance and t s the average tme. Average tme can be determned usng the next Eq. (5) [4: t n 1 n 1 t C t C t (5) where, C s concentraton of the component, t s the tme dfferent between samples, and t s tme. In the same way, varance can be determned by the followng Eq. (6) [5: n t C t t 1 n (6) 1 C t Fnally, dsperson coeffcent can be obtaned by the correlaton commonly used n reactor desgn between the axal Péclet number based on the lqud superfcal velocty and the dsperson coeffcent [6: Q u Pe (7) A D D R where, u s the lqud lneal velocty, s the column length tangent-tangent, D s the dsperson coeffcent. The prevous procedure was appled at dfferent operatng condtons: lqud flow rate of 35 l/h and three dfferent superfcal gas phase veloctes.5, 3 and 6 cm/s...3 Average Gas Bubble Sze and Average Interfacal Area It was used the DSD (dynamc gas dsengagement) methodology to determne the average gas bubble sze dstrbuton [7. Ths method s based on determnng the average gas bubble sze dstrbuton when the gas bubble s dsengagng of the gas-lqud bed nsde the column. Ths measurement can be lked wth the rsng bubble velocty of dfferent groups of bubble szes. These veloctes could be used to determne the average bubble sze of each group of bubbles and then t s possble to estmate the Sauter dameter and the average nterfacal area. Frst, t s necessary to plot the bed heght level as functon of tme when the gas nlet s closed by the followng Eq. (8): H H ( t) G ( Vc A. H nf s H (1 ) A G ) H H s (8) where, H o s the bed level before closng the gas nlet, H (t) s the bed level any tme, H s s the bed level at the end, A s the columns transversal area, and V c-nf s the nlet cone volume. Accordng to the shape of the curve obtaned, the gas bubble szes have to be defned as b, tro, poly modal dfferent szes of groups of bubbles: small, medum and large group of bubble. For nstance, each group of bubbles should be defned as a secton of the curve obtaned and represented by a straght lne. The slope (S ) and the y-axs cut pont (b ) should be
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 181 determned (Fg. 4). Then, gas bubble hold up for each ndvdual group of bubbles could be determned for as follows: H s 1 1 G (9) H b1 b where, b s the y-axs cut pont of each group of bubble. And the followng equatons could be used to estmate the gas bubble rsng veloctes: H(t)/Ho rato (adm) (1) where, S s the curve slope of each group of bubble. For small bubble, whch s normally used the termnal rsng velocty (u b ) based on a sngle bubble n an nfnte medum determned by Marruc equaton [7: 5 3 u bs 1 GS u (11) b 1 where, u Bs s the small bubble velocty. Then, average gas bubble sze dameter for each group of bubbles could be determned followng Stokes s law, or Peebles and Garber, Clft or Abou-el-Hassan equatons as t s shown n Table 3. Fnally, to estmate the Sauter dameter (d S ) of the bubble and the average nterfacal area (a S ) can be obtaned by the next two Eqs. (1) and (13) [7: a d S S n 1 n n d n d 1 3 b b GS N 1 1 G G N 6 G 6 d d S d G b b (1) (13) where, ρ and ρ G are lqud and gas densty, μ s the lqud vscosty and the σ s the lqud surface tenson. 3. Results and Dscusson To ensure ssystem stablty and qualty of the results, Fg. 4 Tme (s) Gas dsengagement profle on tme. Table 3 Bubble sze dameter equatons. Reference Correlaton Stokes s law 18 ub dbs g G Peebles and Garber.41.78 ub dbs 4.76. 59 g Clft Abou-el-Hassan u bs.5.14.55gd B d B.75log. 5 3 1 3 8 3 V F u B d V B gd F B G 4 1 3 3 3 column pressure drop was montored and regstered for all sets of experments done here. For all the experments conducted, the transent perod of the system was less than 5 mn. It was reached operatonal stablty after that tme. 3.1 Gas Hold Up Profle and Flow Patterns Fg. 5 s a plot of the expermental gas hold up and gas velocty n the bubble column reactor for the ar-lght ol system at 35 l/h of lqud flow rate at dfferent superfcal gas veloctes (lne wth sold trangles). Gas hold up ncreases when superfcal gas velocty ncreases, as expected, followng a potental shape curve ( = aug N type) descrbed by Shah [8..5
18 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 35 Gas hold up (v%) 3.6 9.7 9.1 3 7.7 5.8 5 6.6.4 5.1 3.9.5 17.9.7 18.9 15 13.8 16.5 1 9.8 1.4 Eg top Eg bottom 9.1 Eg total cal 5 3.6 Eg total exp 3.9 4 6 8 1 Superfcal gas velocty (cm/s) Fg. 5 Gas hold up for ar-lght ol system at 35 l/h of lqud flow rate. The equaton found for the ar-lght ol system was ε =.81Ug.5493. Accordng to Shah, f the exponental number s lower than.7, n ths case,.5493, then the flow pattern tend to be church turbulent. Empty trangles lne represents the calculated total gas hold up obtaned usng the pressure drop methodology based on both, top and bottom gas hold up by partal pressure drops, whch was exactly the same value obtaned when the total gas hold up was determned by the pressure drop methodology for the total column. All the results showed n ths work are based on pressure drop methodology used. But, the dfference between the phases trappng and the pressure drop methodologes was only of.9% on average [9. Top and bottom sectons gas hold up are shown at the Fg. 5 on square and cycle lnes, respectvely. It can be seen that gas hold up at the top of the column s almost always hgher than the gas hold up at the bottom. The dfferental between both curves ncrease when the superfcal gas velocty s hgher than cm/s and t s even hgher after 4 cm/s. Only at superfcal gas velocty of.5 cm/s, the gas hold up at the top and bottom are qute de same. Ths result confrm what was observed by Sardella [9 where at.5 cm/s of superfcal gas velocty usng the same condton and system t was defned the flow pattern as heterogeneous bubble flow, where dfferent sze bubbles rse along the reactor almost vertcally wthout changng sgnfcantly bubbles volume or wth a bubble sze ncreasng to the pressure drop on the column. After cm/s of superfcal gas velocty, the gas hold up at the top of the column s clearly hgher than the one at the bottom. As s mentoned by Sardella [9, after cm/s of superfcal gas velocty, the vsual flow pattern can be clearly defned as churn turbulent because t can be seen dfferent sze bubbles whch change the volume along the column, wth random movements, back flow and even axal swrl, wth the bg jet type bubble at bottom of the column whch dsappear along the column breakng down n small one. The results obtaned n ths paper support the prevous explanaton because, bg bubbles rse faster than small one, and faster the bubble lower the resdent tme n the column decreasng the gas hold up. Bllet [1 found smlar results. After 4 cm/s of superfcal gas velocty the dfferent between gas hold up at the top and bottom ncrease. Ths can be explaned based on Sardella [9 where t was dentfy foam formaton at the top of the column and due to the ncreasng of gas velocty, the bg type
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 183 jet bubble at the bottom of the column ncrease. These two phenomena could explan the results obtaned n ths paper. Pno and Gutan found smlar results wth foam formaton [11, 1. 3. Average Gas Bubble Sze and Average Interfacal Area Fg. 6 shows the bed heght level as functon of tme when the gas nlet valve to the column s closed. Ths result was obtaned workng at 6 cm/s of superfcal gas velocty and wth an ntal lqud level of 155 cm n the column (total column). Smlar behavor was obtaned for the rest of the cases evaluated. As t was mentoned at the methodology secton, t was necessary to splt the curve n dfferent lneal segments and categorze t as bg, medum (f apply) and small bubbles. Durng the test, t was observed that bgger the bubbles faster t dsengages from the lqud. Fnally the bed heght level remans constant and equal to the begnnng. Fg. 7 shows the average gas bubble sze usng ar-lght ol system at three dfferent heghts of the column wth 4 dfferent superfcal gas veloctes:.5, 1.5, 3 and 6 cm/s. Frst, t s mportant to pont out that the average gas bubble sze decreases when the superfcal gas velocty ncreases, at any column level. Smlar results were obtaned by Daly [7, usng a.3 cm dameter column workng wth smlar gas-lqud systems. These results confrm also that the bubble breakng phenomena s governng along the column when the gas flow rate s ncreasng, whch s also more obvous because the jet type bg bubble formaton at bottom of the column. Ths s a typcal phenomenon n a church turbulent flow regme. Mors [13 obtaned an average bubble dameter of 1.5-3.5 mm, Shah [8 obtaned 1- mm and Daly [7 obtaned bubble dameter of -5 mm. Those values are smlar to the ones obtaned n ths work. As t was mentoned before, at.5 cm/s of superfcal gas velocty, the average bubble sze tend to ncrease along the column due to the pressure drop confrmng the heterogeneous bubble flow pattern due to the coalescence phenomena. In spte of the average bubble dameter also ncrease along the reactor at 1.5 cm/s, typcal of heterogeneous bubble flow pattern, accordng wth Sardella [9 ths condton was defned as transton zone whch make very dffcult to understand. After 3 cm/s, at the upper sze of the column, t can be observed a breakng-coalescence bubble phenomena and the bubble dameter try to keep almost constants, H(t)/Ho rato (adm) 1. 1..98.96.94.9.9.88.86.84.8.8 y = -.171x + 1. R =.9166 y = -.183x +.89313 R =.9578 y = -.59x +.85941 R =.98419 y = -6E-5x +.8169 R =.993 4 6 8 1 1 14 Tme (s) Tme (s) Fg. 6 Typcal gas dsengagement profle on tme ar-lght ol system at 6 cm/s and 155 cm of heght.
184 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and Sauter dameter (mm) 6 5 4 3 1.5 cm/s 1.5 cm/s 3 cm/s 6 cm/s 9 15 155 Dstance from bottom (cm) Fg. 7 Sauter dameter for ar-lght ol system at dfferent superfcal gas velocty. ncreasng and decreasng just a lttle, whch s a typcal behavor of a churn turbulent flow pattern. At 155 cm of lqud heght level, t can be seen also that the average bubbledameter tend to reach a constant value, after 3 cm/s, whch also can explan the tendency to reach mnmum constant value n the gas hold up at the top of the column shown n the Fg. 5, because t reached a dynamc equlbrum between breakng and coalescence phenomena. Those results also confrm the flow patterns defned by Sardella [9 and the gas hold up profle already shown n ths paper. As t s mentoned on Soong [14 Akta y Yoshda proposed an equaton to estmate average bubble dameter d 3, to be used on column tll 3 cm of dameter, and superfcal gas velocty lower than 7 cm/s, based also on physcal-chemcal propertes and column dameter as t s followng shown: 3 6.5.1. 1 Bo Ga Fr d D (13) C The Fg. 8 shows that, n spte of the value are not exactly the same, the tendency of the average bubble dameter (d 3 ) as a functon of the superfcal gas velocty are very smlar, and the Akta and Yoshda equaton could be used to estmate the average bubble dameter n ths column workng wth ths systems. The comparson was made usng the average bubble dameter expermentally obtaned at 155 cm of lqud heght n the column, because t almost represents the whole column scenaro. Fg. 8 also shows that the average bubble dameter tends to reach a mnmum constant value at hgher superfcal gas veloctes as t was commented prevously. The gas-lqud nterfacal area s a very mportant varable whch s mpacted by the column geometry, the operatng condtons and lqud propertes. The gas-lqud nterfacal area s very mportant n ths knd of reactor because t could mpact the mass transfer. Fg. 9 shows the gas-lqud nterfacal area at dfferent superfcal gas veloctes workng a lqud flow rate of 35 l/h at several lqud levels n the column. Due to the fact the gas-lqud nterfacal area depend on the gas hold up and the average bubble dameter as t s shown n the Eqs. (1) and (13), hgher the superfcal gas velocty, hgher the gas hold up and lower the average bubble dameter, mpactng almost exponentally the gas-lqud nterfacal area as t s shown n the Fg. 9 at any lqud level. Prevous works have found smlar gas-lqud nterfacal area (9-1 m /m 3 ) workng wth norganc lqud and wth superfcal gas veloctes between 3-14 cm/s [15.
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 185 1 1 Sauter dameter (mm) 8 6 4 Cal Exp 1 3 4 5 6 7 8 Superfcal gas velocty (cm/s) Fg. 8 Bubble dameter comparson wth Akta and Yoshda correlaton. 16 Interfacal area (m /m 3 ) Interfacal area (m /m 3 ) 14 1 1 8 6 4.5 cm/s 1.5 cm/s 3 cm/s 6 cm/s 9 15 155 Dstance from bottom (cm) Fg. 9 Interfacal area for ar-lght ol system. Workng at heterogeneous bubble flow pattern at 1.5 cm/s, the gas-lqud nterfacal area obtaned s very low (lower than 15 m /m 3 ), whch could negatvely mpact the mass-transfer coeffcent (Kla), wth a potental decreasng of the vacuum resdue converson at the real reactor, droppng the reactor performance. Workng at church turbulent flow pattern, at 3cm/s or 6 cm/s, the stuaton change dramatcally and the gas-lqud nterfacal area ncrease tll the expected value for ths knd of moderate hdroconverson reactor. As reference, accordng to Trambouze [16, the gas-lqud nterfacal area for moderate hdrotreatment process should be around -1, m /m 3. Fg. 1 shows the comparson between the ar-tap water and the ar-lght ol systems, workng at 155 of lqud level and at three dfferent superfcal gas veloctes. As t can be seen, the average bubble dameter of the ar-tap water system s always hgher (at least 4 tmes) than the ar-lght ol system. The tendency was very smlar to the one found wth lght ol. The average bubble dameter s hgher
186 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and Sauter dameter (mm) 5 15 1 5 ght ol Tap water Fg. 1.5 cm/s 3 3 cm/s 6 cm/s Superfcal gas velocty (cm/s) Ar-tap water and ar-lght ol systems Sauter dameter comparson. at bubble flow pattern than the one found at turbulent flow pattern. Ths result can be explaned because as t s shown n Table 1, lght ol has lower surface tenson than the water. Accordng to emone [15 and Daly [7, when lqud surface tenson decreases, the system become less coalescng and the gas hold up ncrease. Ths was observed by Sardella [9. The lower the lqud surface tenson, the lower the bubble sze, ncreasng the nterfacal area and makng the system easer to transport. Smaller bubble produce larger gas holds up due to ther low rse veloctes. Kelkar and Daly found smlar results [7, 17. Daly [7 found average bubble dameter between -5 mm for tap water systems, but emone [15 found values around 5-9 mm for the same system n a column of 7.8 cm dameter on church turbulent flow regme, whch s more smlar to the values obtaned n ths work. 3.3 Reactor Trays Evaluaton 3.3.1 Reactor Trays Evaluaton Fg. 11 shows the gas hold up wth tap water-ar and lght ol-ar systems at a lqud flow rate of 35 l/h at dfferent superfcal gas veloctes wth and wthout trays nsde the column. As t can seen, trays nsde the column ncrease clearly the gas hold up n both systems tested. Trays nsde the column make the system less coalescng, breakng more the bubbles, creatng small bubbles. As t was mentoned before, smaller bubble produce larger gas holds up due to ther low rse veloctes. Kelkar and Daly found smlar results [7, 17. Alvaré [18 found that trays nsde the column also reduce the lqud recrculaton whch then, reduces ts hold up. 3.3. Flow Regme Fg. 1 shows the obtaned Resdence Tme Dstrbuton curves for the ar-lght ol system workng at 35 l/h of lqud flow rate and three dfferent superfcal gas veloctes (.5, 3 and 6 cm/s) wth nternal trays nsde the column. In order to compare all the systems t was necessary to normalze the RTD curves, constructng whch s normally called E curves. It can be seem that all curves follow the same shape. They have a peak at the begnnng and then fall progressvely. These curves are typcal for well mxed flow regme systems. The hgher the superfcal gas veloctes, the hgher the tracer concentratons, because the lqud hold up s lower. The dfference on tme between all the peaks s very small, but, t s possble to dentfy that at hgher gas velocty, the peak s found faster, as expected. Table 4 shows dsperson coeffcents and Péclet numbers found for ar-lght ol system at 35 l/h of
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 187 Gas hold up (v%) 5 45 4 35 3 5 15 ght ol wth trays 1 Water wth trays ght ol wthout trays 5 Water wthout trays 4 6 8 1 1 Superfcal gas velocty (cm/s) Fg. 11 Gas hold up wth and wthout trays..14.1 Ug =.5 cm/s Ug = 3 cm/s Ug = 6 cm/s.1 E curve (adm) E Curve (adm).8.6.4.. 1 3 4 5 6 7 8 Fg. 1 Tme (s) E curves for ar-lght ol systems at 35 l/h of lqud flow rate wth trays. Table 4 Flow regme parameter for ar-lght ol system at 35 l/h of lqud flow rate. Mean tme Superfcal gas velocty (cm/s) (s) Dsperson coeffcent (cm /s) Péclet number (adm).5 1,97 1.6 6.4 3 1,476 111.6 4.3 6 1,13 34.8 3.15 lqud flow rate and three dfferent superfcal gas veloctes. It can be seen that for all the cases, smlar and small Péclet number values and smlar and hgh dsperson coeffcent values. Based on the Péclet numbers and the RTD curves, t s possble to defne that ths reactor at 35 l/h of lqud flow rate and any gas veloctes tends to be a well mxed flow reactor. Carberry [6. Smlar results were found for the ar-tap water system. Fg. 13 shows a comparson wth and wthout
188 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and nternal trays nsde the column workng at 35 l/h of lqud flow rate and 3 cm/s of superfcal gas velocty, usng Sardella [9 results. Both E curves have the same behavor, but due to the hgher gas hold up on obtaned wth the nternal trays, tracer concentraton ncreases nsde the column and the peak value n ths system s hgher than the value found for the column wth nternal. When t s used nternal trays nsde the column the peak of the tracer concentraton appears later on tme, almost the double of the tme due to the ncreasng on the number of the mxng stages creatng by the trays. It was found a Péclet number of 4.3 s the same to the one obtaned wthout nternal trays Sardella [9, reason why ths system also can be defned as a complete mxng reactor. Another way to see the results s trough the F curves (Fg. 13), whch represent the quantty of tracer removed as functon of tme. For the system wth nternal trays, the quantty of tracer removed out the reactor at the average resdence tme (,119 s) s between 84%, around 1% more than the quantty of materal obtaned wthout trays, but n both cases the tracer reman nsde the column untl 8 s ( h and 13 mn), almost four tmes the average resdence tme defned for the lqud flow rate. Ths result ndcate that the tray nsde the column could mprove products selectvty, because, less materal reman n the reactor after at the desred resdence tme. At operatng condton wth the vacuum resdue then, ths stuaton could decrease coke formaton. The results obtaned wth the nternal trays shows that dead volumes are obtaned by mean of the nternal trays, dsplacng the peak of the curve on tme, wthout changng the flow regme, as t s mentoned on Fogler [19. Alvaré [18 explan that the total free area of the holes should be lower than 1% of the total area of the tray n order to obtan an mportant change n the flow regme. Trays tested n ths work have 14-16% of free area, whch s hgher the value mentoned at Alvaré [18, and t could be the reasons of the results obtaned n ths work. Dead zones are very common on fxed bed reactor, nevertheless, trays can also create dead zone just over t. Hgher the superfcal gas velocty, hgher the backflow mxng between each par of trays, reducng potental dead zone, makng the RTD more smlar to the one obtaned wthout nternal trays. Dead zone are volume wth neglgble veloctes whch could promote coke formaton..1.9.8 1.9.8 E curve (adm) E Curve (adm).7.6.5.4.3.7.6.5.4.3 F Curve F curve (adm) (adm)..1 Wthout Trays Wth Trays..1. 1 3 4 5 6 7 8 Tme (s) Fg. 13 E and F curves for ar-lght ol systems at 35 l/h of lqud flow rate wth and wthout trays.
Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and 189 4. Conclusons Wthout nternal trays nsde the column was found that gas hold up ncrease along the reactor. It was possble to dentfy heterogeneous bubble, transton and turbulent flow pattern areas at dfferent superfcal gas veloctes for the ar-lght ol system. It was also determne that average gas bubble sze ncrease along the reactor at bubble regme from -5 mm but at turbulent regme, stay oscllatng between 1-3 mm. At the same reactor heght level, average gas bubble sze decreases wth the superfcal gas velocty. Average nterfacal area ncreases exponentally wth gas superfcal velocty at any reactor heght, tll 1,41 m /m 3 for the ar-lght ol system but, at bubble flow regme, the average nterfacal area s lower than 1 m /m 3, that perhaps negatvely mpact the reactor performance. Internal plates n the reactor always ncrease global gas hold up at any condton or system used. Resdence tme dstrbutons curves, Péclet numbers and dsperson coeffcents founded, show that ths reactor wth nternal plates used here stll tends to be a complete mxng reactor under the operatng condtons used. The new RTD curve was dsplaced on tme because the death space created by the nternal nsde the column Acknowledgments The authors thank Wllam Jmenez and Davd Mendoza from Intevep for provdng techncal support on the vdeo and photographc secton used n ths work. We thank Wlfredo Mendoza from Intevep for the ad provded on the tracer defnton and chemcal measurements. References [1] Sardella, R., Rojas, J. D., Zacaras,., Delgado, O., Rvas A., ópez E., and Marténez S. 9. Transportable Upgraded Crude: Aquaconverson Technology Overvew. World Heavy Ol Congress, Margarta Island, Venezuela. [] Akta, K., and Yoshda, F. 1973. Gas Hold-up and Volumetrc Mass Transfer Coeffcent n Bubble Columns. Ind. Eng. Chem. Process Des. Develop 1: 76-8. [3] Hoefsloot, H., and Krshna, R. 1993. Influence of Gas Densty on the Stablty of Homogeneous Flow n Bubble Columns. Ind. Eng. Chem. Res. 3 (4): 747-5. [4] envenspel, O. 1981. Chemcal Reacton Engneerng, Second Edton, Reverté s.a. Edtoral. [5] Duns, H. J. R., and Ros, N. C. J. 1983. Vertcal Flow of Gas qud Mxtures from Boreholes. Proceedngs of 6th World Petroleum Congress, 451-4. [6] Carberry, J. Chemcal and Catalytc Reacton Engneerng. Frst Edton, McGraw-Hll Chemcal Engneerng Seres, New York, 149-51. [7] Daly, J., Patel, S., and Bukur, D. 199. Measurement of a Gas Hold Ups and Sauter Mean Bubble Dameter n Bubble Column Reactor by Dynamc Dsengagement Method. Chemcal Engneerng Scence 47 (13/14): 3647-54. [8] Shah, Y. T., Kelkar, B. G., and Godbole, S. P., 198. Desgn Parameters Estmatons for Bubble Column Reactors. AlChe Journal 8: 353-6. [9] Sardella, R., Zacarías,., Pava, M., and Medna, H. 11. Bubble Column Reactor Flud Dynamc Study at Plot Plant Scale for Resdue and Extra-heavy Crude Ol Upgradng Technology. World Heavy Ol Congress, Edmonton, Alberta Canada. [1] Bllet, A. M., Chaumat, H, and Delmas, H. Hydrodynamcs and Mass Transfer n Bubble Column: Influence of qud Phase Surface Tenson. Chemcal Engneerng Scence 6: 7378-9. [11] Pno,. Z., Yepez, M. M., and Saes, A. E. 199. An Expermental Study of Gas Holds Up n Two-phase Bubble Columns wth Foamng quds. Chem. Eng. Comm 89: 155-75. [1] Gutan, J., and Joseph, D. 1998. How Bubbly Mxtures Foam and Foam Control Usng Fludzed Bed. Ind. J. Multphase Flow (1): 1-16. [13] Mors, B., Behksh, A., Men, Z. and Inga, J.. Mass Transfer Characterstcs n a arge-scale Slurry Bubble Columns Reactor wth Organc qud Mxtures. Chemcal Engneerng Scence 57 (16): 337-4. [14] Soong, Y., Harke, F. W., Gamwo I. K., Schehl, R. R. and. Zarochak, M. F. 1997. Hydrodynamc Study n a Slurry Bubble Column Reactor. Catalyst Today 35: 47-34. [15] emone, R., Behksh, A., and Mors, B. 4. Hydrodynamc and Mass-transfer Characterstcs n
19 Bubble Column Reactor Flud-dynamc Evaluaton at Plot-plant for Resdue and Organc qud Mxtures n a arge-scale Bubble Column Reactor for the Toluene Oxdaton Process. Ind. Eng. Chem. Res. 43: 6195-1. [16] Trambouze, P., and Euzen, J. P.. Chemcal Reactor from Desgn to Operaton, Edcón Technp, 4-15. [17] Kelkar, B., Shah, Y., and Carr, N. 1984. Hydrodynamcs and Axal Mxng n a Three Phase Bubble Column. Effects of Slurry Propertes. Ind. Eng. Chem. Process Des. Dev. 3 (): 38-13. [18] Alvaré, J. and Al-Dahhan, M. 6. Gas Holdup n Trayed Bubble Column Reactor. Ind. Eng. Chem. Res. 45: 33-6. [19] Fogler, S. 1. Elements of Chemcal Reacton Engneerng Prentce Hall. Mexco: D.F.