AIP Con e ence P oceedings 2033, 130013 (2018); h ps://doi.o g/10.1063/1.5067147 2033, 130013
© 2018 Au ho (s).
Expe imen al es ing o mul i- ubula eac o
o hyd ogen p oduc ion and compa ison
wi h a he mal CFD model
Ci e as: AIP Con e ence P oceedings 2033, 130013 (2018); h ps://doi.o g/10.1063/1.5067147
Published Online: 08 No embe 2018
El i a Tapia, Au elio González-Pa do, Al edo I anzo, Al onso Vidal, and Felipe Rosa
ARTICLES YOU MAY BE INTERESTED IN
The mal es s o a mul i- ubula eac o o hyd ogen p oduc ion by using mixed e i es
he mochemical cycle
AIP Con e ence P oceedings 1850, 100009 (2017); h ps://doi.o g/10.1063/1.4984466
Helios a aiming s a egy o 3 cylind ical ca i y- ecei e s in eg a ed in a 750 kW sola owe
hyd ogen plan
AIP Con e ence P oceedings 2033, 130007 (2018); h ps://doi.o g/10.1063/1.5067141
HYDROSOL-PLANT: S uc u ed edox eac o s o H2 p oduc ion om sola he mochemical
H2O spli ing
AIP Con e ence P oceedings 2033, 130010 (2018); h ps://doi.o g/10.1063/1.5067144
Expe imen al Tes ing o Mul i-Tubula Reac o o
Hyd ogen P oduc ion and Compa ison wi h a The mal CFD
Model
El i a Tapia1,a), Au elio González-Pa do2,b), Al edo I anzo1,c), Al onso Vidal2,d),
Felipe Rosa1,e)
1The mal Enginee ing G oup, Ene gy Enginee ing Depa men . School o Enginee ing, Uni e si y o Se ille.
Camino de los Descub imien os s/n, 41092 Se illa, Spain.
2CIEMAT-PSA, Ca e e a de Sene s, S/N Tabe nas, Alme ía, Spain.
a)Co esponding au ho : [email p o ec ed]
b)[email p o ec ed]
c)ai [email protected]
d)al onso. idal@ciema .es
e)[email p o ec ed]
Abs ac . This s udy p esen s a compa ison o he expe imen al es s and CFD esul s o a mul i- ubula sola eac o o
hyd ogen p oduc ion in a pilo plan in he Pla a o ma Sola de Alme ía. This pape desc ibes he me hodology used o
he sola eac o design and he expe imen al es s ca y ou du ing he es ing and cha ac e iza ion campaign o he plan .
The CFD model which has been used o design he sola eac o has been alida ed wi h an e o a ound 10%. CFD
simula ions also allow o sol e he he mal balance in he eac o (ca i y and ubes) and o calcula e he pe cen age o
eac ing media inside he ubes which achie e he equi ed empe a u e o he p ocess. The empe a u e in he
he mocouples is a ound 1200 ºC o he expe imen al da a and a 90% o he e i e inside he ube is abo e 900 ºC. The
mul i- ubula sola eac o , which has been design wi h CFD echniques, has been buil and ope a ed success ully.
INTRODUCTION
Technologies in ol ing high empe a u e endo he mic eac ions o con e ing sola ene gy o chemical uels
ha e been in es iga ed a ound he wo ld. Many o he ac i i ies a e ocused on iden i ying, de eloping, and
assessing imp o ed ecei e / eac o s o e icien ly ca ying ou he he mochemical p ocesses o he p oduc ion o
hyd ogen [1]. In his con ex , sola eac o s ep esen a p omising echnology o a sus ainable ene gy sys em, o
example as an ene gy s o age by p oducing an ene gy ca ie (hyd ogen) o i s use du ing he pe iod when sola
ene gy is no a ailable. A p esen , di e en sola eac o s ha e been demons a ed o di e en chemical p ocesses
and scales.
Nowadays, CFD ools a e inc easingly being used o sola eac o s design and o op imize ope a ing condi ions.
The s udies using CFD ools ha e inc eased du ing he las yea s in o de o be e p edic esul s o empe a u e,
hea lux, p essu e, and eloci y dis ibu ions. Di e en eac o ypes and con igu a ions a e possible, and clea ly he
modelling equi emen s and de ails will a y depending on he eac o ype.
A 100 kW h mul i- ubula ca i y eac o o hyd ogen p oduc ion in eg a ed in a sola owe was designed,
manu ac u ed and es ed in he amewo k o a p ojec called SolH2 (an INNPACTO ini ia i e o he Spanish
Minis y o Economy and Compe i i eness) wi h he main goal o demons a e he echnological easibili y o sola
he mochemical wa e spli ing cycles as one o he mos p omising op ions o p oduce hyd ogen om enewable
Sola PACES 2017
AIP Con . P oc. 2033, 130013-1–130013-9; h ps://doi.o g/10.1063/1.5067147
Published by AIP Publishing. 978-0-7354-1757-1/$30.00
130013-1
sou ces in an emission- ee way. A he end o his p ojec , he owne ship o his plan has been ans e ed o
CIEMAT o u he in-dep h s udies.
The eac o was designed acco ding o he ecei e equi emen s o he capaci y o a gi en hyd ogen
p oduc ion, inle and ou le gas empe a u es, and e iciency. A Compu a ional Fluid Dynamic (CFD) analysis o he
sola ca i y- ecei e was ca ied ou wi hin he design p ocess o de e mine and op imize he empe a u e
dis ibu ion and he abso bed adia ion lux o e he in e nal ca i y walls, which is explained in he nex sec ion. In
his wo k, he CFD modelling o he 100 kW (comme cial scale) mul i- ubula eac o design is alida ed wi h he
expe imen al esul s o his pilo plan in he Pla a o ma Sola de Alme ía (PSA).
PILOT PLANT
The eac o has been ins alled in he SSPS-CRS acili y o he Pla a o ma Sola de Alme ía, loca ed in Tabe nas
dese , Spain. The Pla a o ma Sola de Alme ia is he la ges Eu opean sola ene gy esea ch cen e wi h mo e
es ing acili ies in he wo ld o sola ene gy. The SSPS-CRS plan consis s o an au onomous ield o helios a s and
a me allic owe 43 me e s high. The acili y collec s di ec sola adia ion by means o a ield o 91 39.3-m
2
-su ace
helios a s dis ibu ed in a 150-x- 70-m no h ield in o 16 ows. The owe is 43 me e s high and has wo me al
es ing pla o ms, a 26 and 32 me e s espec i ely. The maximum he mal powe deli e ed by he ield on o he
ecei e ape u e is 2,0 MW. This plan has been used in he pas o accommoda e es ing o small sola ecei e s in
he ange o 200-350 kW he mal powe applica ions. Nowadays, he es acili y has been u ned in o a sui able es -
bed o hos he mos impo an esea ch ini ia i es in sola hyd ogen p oduc ion: HYDROSOL –Plan , SYNPET, e c.
Reac o
Acco ding o he hyd ogen p oduc ion speci ica ions and he ini ial geome y condi ions, a i s design o he
sola eac o was p oposed. This ini ial design consis ed o a semi-cylind ical ca i y wi h 2 m adius and a squa e
opening 30x30 cm
2
. Inside he ca i y he e we e 80 ubes in a s agge ed a angemen in 2 ows (FIGURE 1 (a)).
This design was dismissed o wo main easons: he emp y space in he middle o he ows in on o he opening,
and he oo small opening acco ding o he inciden adia ion, esul ing in an inc ease o he empe a u e di e ence
wi hin ubes.
The nex eac o was designed o a oid he p oblems which we e ound in he ini ial design. In his case, he
opening was inc eased un il 46x46 cm
2
and he ubes dis ibu ion was modi ied. In his case, i was p oposed 3 ows
in s agge ed a angemen inc easing he numbe o ubes un il 104 and using he ex eme ubes as a p ehea ing
in e change (see FIGURE 1 (b)). This design was dismissed oo because he big ca i y adius in ol ed a slow
ansien he mal hea ing, and he e iciency in ex eme p ehea ing in e change was low.
The inal design consis s o a semi-cylind ical shape wi h a 1.5 m in e nal adius and 80 ubes in s agge ed
a angemen in h ee ows, including ubes in on o he opening (see FIGURE 1 (c)).
(a) (b) (c)
FIGURE 1. (a) Ini ial sola eac o design. (b)Middle sola eac o design. (c) Final sola eac o design.
130013-2
All designs we e calcula ed wi h CFD ools using a Mon e Ca lo su ace- o-su ace adia ion model o compa e
he esul s o inciden adia ion lux on he walls and on he ubes, and he esul ing empe a u e dis ibu ions. The
las design was buil as esul o SolH2 p ojec (see FIGURE 2). The ubes ha e a leng h o 1.2 m and a o al olume
capaci y o 566 cm3. These ubes a e illed wi h small pelle s o mixed e i es, cylind ically shaped in o de o
inc ease he eac ion su ace and bene i he gas low h ough hem.
(a) (b)
FIGURE 2. (a) F on al iew o he eac o ins alled in CRS owe plan . (b)View o he eac o being i adia ed by CRS ield.
Flux Measu emen Sys em
This sys em is used o measu e he powe a i ing o he ape u e o he eac o . I consis s o cap u ing he
i adiance dis ibu ion on a mo ing lambe ian a ge wi h a high esolu ion CCD came a [2]. The ba in e cep s he
concen a ed beam in he measu ing plane, which is as close o he ecei e as possible. The dis ibu ion o ela i e
in ensi y, in e ms o g ay-scale map, ep esen s he shape o he lux dis ibu ion a he ecei e . A adiome e is
used o calib a e he sys em, by co ela ing he g ay-scale alues o he image pixels wi h he co esponding
i adiance alue measu ed by he adiome e .
Fu he mo e, ou he mocouples a e included inside he ubes o compa e he expe imen al es s wi h he CFD
esul s. These he mocouples a e ins alled in he middle o he ube, hei posi ions a e ube 14 and 21 in he i s ow
and ube 3 and 12 in he second one.
(a) (b)
FIGURE 3. (a) De ailed iew o he ubes illed wi h e i es and wi h a he mocouple ins alled. (b) Dis ibu ion o he ou
he mocouples a anged inside he ubes.
130013-3
THERMAL TESTS
P e ious expe ience wi h alumina ubes showed ha hey we e e y sensi i e o empe a u e g adien s. In
p e ious es s, some ubes we e b oken due o he empe a u e g adien h oughou hei leng h, which p oduced a
s ess in he alumina ubes. This e idence and he goal o he design o ob ain as homogeneous as possible lux
dis ibu ion did necessa y a good s a egy o he helios a s ield. Ini ially, wo g oups o helios a s we e ocused on
he ape u e and hen he numbe o g oups was inc eased un il he equi ed empe a u e was achie ed. Fo he i s
g oups o helios a s, he hea ing was e y p og essi e, and we did no wai un il he empe a u es s abilized since i
would ha e aken a long ime. When 20 helios a s we e ocused, a o al powe o 44.1 kW was ocused inside he
ca i y and a empe a u e o 750-800 ºC was eached. Finally, wi h a o al o 36 helios a s and a powe abou 80 kW,
1200 ºC we e eached [3]. In addi ion, his sola eac o design concep implied a discon inuous hyd ogen
p oduc ion o e he ope a ion ime. The e o e, a sequen ial mode inside he sola eac o was implemen ed o couple
he ope a ion o he plan wi h he mixed e i es he mochemical cycle, which consis s o wo eac ion s ages.
This s udy was done he e nal equinox, which is a ep esen a i e day o he yea . The es was ca ied ou wi h
ni ogen and wi hou wa e , he o al mass low inside ubes was 40 kg/h and he inle empe a u e a ound 70 ºC. The
p ocess o ocus he g oups o helios a s is desc ibed below, and i was designed o enable he sola adia ion o
impinge on all he ce amic ubes: i s one, he g oups 1 and 2; Then he g oup 3 and is con inued wi h g oups 4 and
5. This pa is done e y p og essi e wi hou wai ing he s abili y o empe a u e. Then, g oups 10 and 11 a e
included and inally g oups 8 and 9 (see FIGURE 4). Each g oup con ains dis an and nea by helios a s o a o he
lux homogenei y [3]. The inal con igu a ion o helios a s allows o measu e he inciden adia ion lux and
di ec ion o ays a he opening which ha e been used in CFD simula ions.
FIGURE 4. G oups o helios a s in which he ield has been di ided.
CFD MODEL
The comme cial so wa e ANSYS-CFX 14.0 [4] has been used o sol e he he mal in e changes ha exis in he
sola mul i- ubula eac o . The geome y o he eac o and he ope a ing condi ions ( adia ion lux a opening and
mass low and empe a u e o he ine gas) a e equi ed o sol ing his me hodology [1]. The me hodology is
ocused on a ubula sola eac o whe e luid-dynamic model (ca i y u bulence) and he mal models (conduc ion,
con ec ion and adia ion) a e conside ed in he eac o modeling. In his con ex , his me hodology p oposes he
Shea S ess T anspo (SST) model o u bulence esolu ion due o he ca i y is opened and a be e accu acy is
achie ed in he he mal bounda y laye . Fo adia ion model, Mon e Ca lo model and S2S (su ace- o-su ace)
model a e p oposed due o he ma ix o ays ( eal di ec ions o a adia ion dis ibu ion a he opening) is included a
he opening, and nonpa icipa ing media is conside ed. Fo a p ope CFD analysis, i is necessa y o p e iously ca y
ou a mesh analysis in o de o ensu e ha esul s a e independen o he mesh used in he simula ions. The main
s eps a e desc ibed below:
130013-4

 The ca i y design is ocused on op ical e iciency and he achie emen o a uni o m empe a u e
dis ibu ion, wi h a minimum empe a u e g ea e han he equi ed empe a u e o he hyd ogen p oduc ion
p ocess. This pa o he me hodology is di ided in o di e en pa s whe e mesh independence analysis,
adia ion ac o independence analysis and bounda y condi ions a e included in he model. A adia ion ac o
independence analysis is equi ed in o de o ensu e ha he numbe o his o ies used in he Mon e Ca lo
model is enough o co ec ly calcula e he adia ion and empe a u e ield (i.e. a highe numbe o his o ies
do no in luence he esul s).
 The eac o design s a s once he ca i y design is closed. The main goals a e a comple e balance o
phenomena which ake pa in he p ocess, iden i ying ho /cold spo s, and de e mining he eac ion olume
which achie es he equi ed condi ions.
In o de o ob ain he op imized compu a ional mesh, g id independence es s a e conduc ed on he mesh models
a he beginning o he simula ion p ocess. The mesh is e ined nea wall and high ( empe a u e and eloci y
modulus) g adien egions. A e mesh and adia ion ac o analysis a e done, i is concluded ha he inal mesh has
1.9 million nodes and 9.0 million elemen s and he minimum adia ion ac o (numbe o his o ies in he Mon e
Ca lo model) is 125·106, which is clea ly highe han he de aul alue o 1·104.
The concen a ed sola adia ion en e s he ca i y ecei e h ough he opening. The inciden adia ion lux and
di ec ion ec o s a e p o ided by CIEMAT acco ding o he expe imen al es s (see FIGURE 5). The o al sola
powe is a ound 80 kW.
FIGURE 5. Di ec ion o adia ion e lec ed om CRS ield (36 helios a s).
A ca ie gas, ni ogen, en e s he inle collec o loca ed on he op o he eac o and goes h ough he ubes un il
he ou le collec o . The luid low inside he ca i y and ubes, ai and ni ogen espec i ely, is assumed as u bulen
and lamina espec i ely, ideal gas and anspa en o adia ion. Reac o walls a e conside ed as di usely emi ing
and e lec ing, and opaque. The Mon e Ca lo model wi h su ace o su ace adia ion model is used. The conduc ion
hea ans e wi hin he ubes and po ous media ( e i e pelle s) is included in he same simula ion whe e he
adia ion beha io is analyzed, he e o e a oiding he need o decoupling he models. The speci ic hea capaci y o
he ni ogen gas a cons an p essu e is 1041 J/kgK and dynamic iscosi y, densi y and he mal conduc i i y a e
included depending on he empe a u e as an ideal gas. The he mal conduc i i y o alumina and silicon ca bide a e
included depending on empe a u e. The su ace emissi i y o he di usely- e lec ing ca i y inne wall (silicon
ca bide) and he ubes (alumina) a e 0.3 and 0.8 espec i ely. The ni ogen gas condi ions ( low a e and
empe a u e) a e he same as he expe imen al es . The e ec o he ca i y insula ion a he ex e nal walls is
included in he model, by de ining an ex e nal hea ans e coe icien o 0.128 W/m2K (equi alen o he insula ion
ma e ial hea conduc ion esis ance and ex e nal hea con ec ion esis ance).
The summa y o physical models and bounda y condi ions used in he simula ion is p esen ed in TABLE 1.
130013-5
TABLE 1. Physical models and bounda y condi ions used in he CFD model.
Model / Bounda y Condi ion alue
Radia ion model (ca i y ai ) Mon e Ca lo / Su ace- o-Su ace
Tu bulence model (ca i y ai ) Shea S ess T anspo (SST)
Tu bulence model (ni ogen po ous media
ubes)
Lamina
Fe i e po osi y in ubes 0.4
Ca i y window (ai ) Opening Bounda y Condi ion [4]
Ca i y window ( adia ion) Ray ma ix p o ile (di ec ion and in ensi y)
de ined om expe imen al da a. 80kW
Tubes emissi i y 0.8 (alumina)
Recei e walls emissi i y 0.3 (silicon ca bide)
Recei e ex e nal walls 0.128 W/m
2
K, 25ºC
Ni ogen ubes inle 40 kg/h; 70ºC
RESULTS
The compa ison be ween he expe imen al es s and he CFD model esul s is shown in his sec ion, also aimed a
p o iding an expe imen al alida ion o he CFD model. The e olu ion o empe a u es du ing he expe imen al es
un il achie ing he s eady s a e o he inal con igu a ion o he helios a ield (80 kW) and he s ead s a e CFD
esul s a e p esen ed in FIGURE 6. The maximum empe a u e di e ence is ob ained a ube 21 o he i s ow and
ube 3 in he second ow. These ubes a e loca ed close o he ca i y ex emes so he inciden adia ion lux is lowe
han a he middle ca i y ubes. A u he analysis o he di ec ion adia ion lux and he opening mesh needs o be
done.
FIGURE 6. E olu ion o empe a u es o expe imen al es and CFD empe a u e esul s. T. ube co esponds o he
he mocouple inside he ubes, in con ac wi h e i es. Numbe o helios a s ocused on he ecei e and he powe measu ed.
130013-6
The e o o he CFD simula ion is shown in TABLE 2. I is obse ed ha he CFD empe a u e esul s a e
smalle han he expe imen al empe a u e da a o all he mocouples, his ac could be caused by la ge con ec ion
losses in he CFD simula ions han in he eal expe imen al es s, as well as di e ence be ween he conside ed and
eal ma e ial emissi i y. The maximum empe a u e e o in CFD simula ions is a ound 11.5% which is in he same
ange as o he s s udies [5, 6, 7, 8]. I can be also obse ed ha e o s a e e y small a cen al ubes, and la ge a he
side ubes. The eason o his di e ence is unde in es iga ion. As he inciden adia ion ecei ed by cen al and
side ubes is no he same (mainly di ec inciden adia ion om he helios a ield o cen al ubes, and e lec ions
and ecei e emissions o side ubes) i is possible ha ei he he ma e ial emissi i y o he accu acy o he adia ion
model is no ully app op ia e o ep esen ing he eal adia ion ield wi hin he ca i y. This is howe e s ill unde
in es iga ion in o de o be e assess he eason o his beha io .
TABLE 2. Compa ison o expe imen al empe a u e and CFD esul s
Expe imen al es
(ºC)
CFD esul s
(ºC)
E o
(%)
Di e ence
(K)
T. ube 12 1
s
ow 1178 1167 0.1 1
T. ube 14 2
n
d
ow 1151 1150 0.9 11
T. ube 21 1
s
ow 1147 1015 11.5 132
T. ube 3 2
n
d
ow 1058 938 11.3 120
The CFD simula ions also allow o e alua e he empe a u e dis ibu ion in ubes. FIGURE 7 (a) shows ha
empe a u e in ubes loca ed in ca i y ex emes a e lowe han he es o he ubes. In addi ion, he e ec o he
ca ie gas mass low inle empe a u e is obse ed, as he uppe pa o ubes is clea ly colde . FIGURE 7 (b)
p esen s he a e age empe a u e a he ou e su ace o he ubes. All ubes a e abo e 900 ºC. I is obse ed ha he
second ow has he same beha io as he hi d ow nea he ca i y ex emes and he i s ow in he middle ca i y
ubes due o he e ec o e lexing adia ion lux and di ec adia ion lux espec i ely. Finally, he empe a u e
di e ence among di e en ubes is a ound 140 ºC. So he empe a u e dis ibu ion can be conside ed homogenous
inside he ca i y.
(a) (b)
FIGURE 7. (a) Tempe a u e dis ibu ion o CFD simula ions. (b) A e age empe a u e in he ube su aces dis ibu ion.
The olume o e i e which achie es he equi ed p ocess empe a u e is calcula ed also om he CFD esul s,
and depic ed in FIGURE 8. This esul is used o analyze he e iciency o he eac o eac ion olume. In his case,
90.4% o he e i e domain achie es he equi ed p ocess empe a u e (900 ºC). This pe cen age could be u he
imp o ed i he ene gy o he ou le ca ie gas low is used o p ehea he inle ni ogen low.
130013-7
FIGURE 8. Volume o e i e (inside ubes) whe e empe a u e is highe han 900 ºC.
The he mal balance is shown in TABLE 3. The e a e adia ion and con ec ion losses in he opening o he
ca i y, ep esen ing a ound 19 kW and 13 kW espec i ely. Fu he mo e, only 6.3 kW a e eaching he eac ion
media inside he ubes. The e o e, he he mal e iciency o he sola eac o is 7.9%.
TABLE 3. The mal balance in sola eac o
CFD esul s
Inciden adia ion (W) 80000
Opening adia ion losses (W) 19032
Opening con ec ion losses (W) 12816
Hea ans e o ube 6340
Op ical e iciency (%)     




76.21
The mal e iciency (%)    
 7.93
CONCLUSIONS
A mul i- ubula sola eac o designed wi h he aid o CFD ools was expe imen ally e alua ed du ing his wo k.
The CFD esul s on he he mal analysis o he mul i- ubula sola eac o a e p esen ed in his wo k, as well as he
esul s o he expe imen al es ing. The pu pose o his analysis is o in es iga e he he mal ecei e pe o mance
and i s ope a ional beha io in he speci ied condi ions. Fo his speci ic wo k, he ope a ion empe a u es ange
om 800 o 1200 ºC acco ding o he e i e cycle. An ope a ion s a egy p e iously pe o med by ay acing
simula ions was applied o supply he equi ed powe wi h he op imal lux dis ibu ion on o he alumina ubes. The
CFD model was used o he calcula ion o he empe a u e and adia ion lux p o iles a he ca i y and ecei e
walls, and was alida ed based on he esul s o he he mal es ing campaign. In addi ion, he empe a u e
dis ibu ion inside he ubes was calcula ed in o de o analyze i s homogenei y and whe he he equi ed
empe a u e is achie ed. The scope o he analysis is o assess he he mal e iciency as well as empe a u es o e
he ecei e . The alida ed CFD model p o ides use ul in o ma ion o he assessmen o design pa ame e s and o
op imize he he mal pe o mance o he sola ca i y.
ACKNOWLEDGMENTS
The au ho s would like o hank he Spanish Minis y o Science and Inno a ion (Na ional Plan o Scien i ic
Resea ch, De elopmen and Technological Inno a ion, 2008-2011), he Eu opean Regional De elopmen Fund
(ERDF) o he inancial suppo gi en o he SolH2 p ojec (IPT- 2011-1323-920000) and he FEDER Ope a ional
P og am o Andalusia 2007-2013 o he inancial suppo gi en o he RNM 6127 p ojec .
130013-8