Air Gasification of Biomass in a Downdraft Fixed Bed: A
Comparative Study of the Inorganic and Organic
Products Distribution I . De Bari ,* D. Bari sano, M. Cardi nal e, D. Matera, F. Nanna, and D. Vi ggi ano Biomass Laboratory, RenewableEnergy Division, ENEA, I talian Agency for New Technology, Energy and Environment, Policoro (MT) 75025, I taly Received November 30, 1999 Thi s paper deal s wi th the gasi fi cati on of agri cul tural resi dues such as al mond shel l s and wood at hi gh temperature (850 C) i n a smal l -scal e gasi fi cati on pl ant coupl ed wi th a reci procated i nternal combusti on engi ne. Detai l ed i nvesti gati on of both organi c (tar) and i norgani c (NH 3 , HCN, metal s, etc.) products di stri buti on i n the process streams are provi ded. Thi s research ai ms to assess the exi stence of certai n rel ati ons between feedstock composi ti on and the observed products di stri buti on i n the steady state. Bi omass feedstock i s characteri zed for i ts el emental composi ti on, i ts content of metal l i c speci es, and l i gni n, cel l ul ose, and hemi cel l ul ose fracti ons. A di fferent feedstock composi ti on was found to be associ ated wi th some i mportant vari ati ons i n the process moni tored parameters such as the gasi fi cati on rate, the tar content, and the char yi el d i n the fl ue gas. Al so a rel ati on was found between the rel ati ve amount of ammoni a and cyani de speci es i n the fl ue gas at the operati ng condi ti ons of the gasi fi er (oxygen-to-bi omass rati o and acti vated carbon support). Spectroscopi c features of the sl udgy condensate formed i n the upper part of the gasi fi er duri ng the earl y stage of the process together wi th the characteri zati on of char, fl y ash, and aci di c speci es i n the fl ue gas shed l i ght on some mechani sti c aspects of the gasi fi cati on process. Introduction Gasi fyi ng agri cul tural resi dues i s a promi si ng opti on especi al l y i n vi ew of el ectri ci ty producti on. Bi omass gasi fi cati on i s a compl ex thermochemi cal process that converts l i gnocel l ul osi c materi al s i nto a val uabl e gas by means of a parti al oxi dati on process at hi gh tempera- ture. The gas mi xture from the process can be uti l i zed to produce el ectri cal energy and thermal power, gi vi ng a posi ti ve contri buti on i n terms of energeti c and envi - ronmental pol i ci es. Under typi cal gasi fi cati on condi ti ons, the gas compo- si ti on mai nl y consi sts of CO (18-21%), H 2 (10-16%), O 2 (1.5-2.5%), CH 4 (1-3%), and N 2 (40-54%) together wi th traces of hydrocarbons (ethane, etc.), vol ati l e ash, organi c condensabl e compounds (tar), and i norgani c consti tuents such as HCN, NH 3 , whose burni ng causes hi gh ni trogen oxi de emi ssi ons. I n the past, consi derabl e research has been carri ed out to study how the exact composi ti on of the produced tar at the gasi fi er exi t vari es as a functi on of gasi fi er geometri es 1 and process- i ng vari abl es. 2 Most recent research studi es on the chemi stry of tar formati on from bi omass have poi nted out the exi stence of a dependence of tar yi el ds from both the temperature and the type of feedstock l oaded. 3 Gas temperature and fuel chemi cal composi ti on al so pl ay a deci si ve rol e i n determi ni ng the amount of ni trogen compounds (NH 3 and HCN) i n the fl ue gas 4 as wel l as the ash composi ti on and vol ati l i zati on duri ng the gas- i fi cati on process. 5 Better knowl edge of the tar evol uti on and the fate of i norgani c consti tuents of the fl ue gas has hi ghl i ghted the i mportance of studyi ng the anal yti cal framework of the gasi fi cati on process i n order to i m- prove the process technol ogy by choosi ng tai l ored strat- egi es, i ncl udi ng bi omass pretreatment and gas cl ean- i ng. 6 Despi te the sci enti fi c and techni cal knowl edge gai ned about the contami nants embedded i n the fl ue gas, to date, very l i ttl e attenti on has been devoted to the rel ati on between the i ni ti al composi ti on of the bi omass feedstock (both i n i ts organi c and i norgani c consti tuents) and the mi nor products at the exi t of the gasi fi er or among the process products themsel ves. To the best of our knowl edge, onl y a few papers have deal t wi th the acti on of i norgani c consti tuents i n the bi omass as * To whom correspondence shoul d be addressed. Fax: +39 0835- 974210. E-mai l : de_bari @tri sai a.enea.i t. (1) Rajvanshi , A. K. Bi omass Gasi fi cati on. Alternative energy in agriculture; CRC Press: Boca Raton, FL, 1986; Vol . 2, pp 83-102. (2) Ki noshi ta, C. M.; Wang Y.; Zhou J. Tar Formati on Under Di fferent Bi omass Gasi fi cati on Condi ti ons. J . Anal. Appl. Pyrolysis 1994, 29, 169. (3) Brage, C.; Qi zhuang, Y.; Guanxi ng, C.; Sjostrom, K. Tar Evol u- ti on Profi l es Obtai ned from Gasi fi cati on of Bi omass and Coal . I n Biomass for Energy and I ndustry, Proceedings of the I nternational Conference Wurzburg; Kopetz, H., Weber, T., Pal z, W., Charti er, P., Ferrero, G. L., Eds.; C.A.R.M.E.N.: Germany, 1998; pp 1634-1637. (4) Leppal ahti , J. Formati on and Behavi our of Ni trogen Compounds i n an I GCC Process. Bioresour. Technol. 1993, 46, 65-70. (5) Baxter, L. L.; Mi l es, T. R.; Mi l es, T. R., Jr.; Jenki ns, B. M.; Mi l ne, T.; Dayton, D.; Bryers, R. W.; Ol den, L. L. The Behavi or of I norgani c Materi al i n Bi omass-Fi red Power Boi l ers: Fi el d and Laboratory Experi ences. Fuel Process. Technol. 1998, 54, 47-78. (6) Hasl er, P.; Buhl er, R.; Nussbaumer, T. Eval uati on of Gas Cl eani ng Technol ogi es for Bi omass Gasi fi ers. I n Biomass for Energy and I ndustry, Proceeding of the I nternational Conference Wurzburg; Kopetz, H., Weber, T., Pal z, W., Charti er, P., Ferrero, G. L., Eds.; C.A.R.M.E.N.: Germany, 1998; pp 272-275. 889 Energy & Fuels 2000, 14, 889-898 10.1021/ef990243g CCC: $19.00 2000 Ameri can Chemi cal Soci ety Publ i shed on Web 06/08/2000 i ndi genous catal ysts toward bi omass pyrol ysi s and gasi fi cati on reacti ons. 7-9 I n parti cul ar, DeGroot and Ri chards 8 observed that the gasi fi cati on of char woul d occur more qui ckl y i n the presence of a hi gher metal i on content (al kal i and al kal i ne earth metal s), whi l e Agbl evor and Besl er 10 concl uded that the i norgani c consti tuents i n the bi omass act as nucl eati on centers toward the char formati on. Moreover, l i terature data about the evol uti on of the i norgani c speci es NH 3 and HCN mai nl y concern coal and peat, 11-14 and l ess data are avai l abl e on agri cul tural resi dues gasi fi cati on on smal l -scal e fi xed bed pl ants. Thi s paper descri bes the fate of both organi c and i norgani c consti tuents duri ng the gasi fi cati on of bi omass i n a fi xed bed gasi fi er, sheddi ng new l i ght on the chemi stry i nvol ved i n thi s bi omass processi ng, and expl ores how the organi c and i norgani c fuel consti tuents coul d i nfl uence each other duri ng the process. Fi nal l y, an attempt at descri bi ng HCN and NH 3 behavi or as a functi on of the feedstock l oadi ng and the gasi fi cati on rati o (ai r-fuel rati o) i s performed. Comprehensi ve test runs were carri ed out i n a fi xed bed downdraft gasi fi er usi ng some common European bi ofuel s (oak, turkey oak, al mond shel l s). The gasi fi ca- ti on of l i gnocel l ul osi c resi dues by means of such si mpl e and versati l e pl ants i s di ctated by the necessi ty of devel opi ng technol ogi es capabl e of processi ng di fferent bi omass feedstocks for smal l -scal e power producti on (e.g. agroi ndustri es and remote zones). The empl oyed bi om- ass was characteri zed for i ts i norgani c consti tuents, sugars, extracti ves, and l i gni n content, whi l e a detai l ed anal ysi s of products from pl ant runni ng was carri ed out fol l owi ng a sui tabl e gas sampl i ng. The avai l abl e standard methods for fuel anal ysi s mai nl y concern coal and coke. I n recent years, wi th i ncreasi ng i nterest i n the energeti c val ori zati on of bi omass, some l aboratori es 15 cooperate i n tryi ng to establ i sh a best practi ces l i st for bi omass fuel and ash anal ysi s by sel ecti ng and adapti ng the avai l abl e stan- dard methods. Hence, as there are not yet standardi zed methods avai l abl e for sampl e preparati on and anal ysi s of bi omass fuel s and ashes, opti mi zed anal yti cal proce- dures were devel oped i n our l aboratori es i n order to determi ne the feedstock composi ti on as wel l as the consti tuents embedded i n the fl ue gas. I n thi s work several anal yti cal techni ques were used to get a more compl ete pi cture of the formati on of products from gasi fi cati on of certai n bi omass. Experimental Section Pilot Plant Used. Al l the experi ments were carri ed out on a 120 kWth gasi fi er whi ch operates i n downdraft mode uti l i zi ng ai r as the gasi fyi ng agent. Scheme 1 reproduces the major components of the gasi fi cati on pl ant used. The pl ant core (Scheme 2) i s a doubl e wal l cyl i ndri cal chamber whose combusti on zone i s made of stai nl ess steel wi th a doubl e coni c shape. The i nner part of thi s zone contai ns a pl ane wi th nozzl es. The doubl e chamber compensates for heat l oss and al l ows a preheati ng of bi omass. The cl eani ng secti on consi sts of four fi l trati on stages: cycl one fi l ter, scrubber fi l ter, demi ster system, and chi ps fi l ter. The uti l i zati on secti on contai ns a power uni t equi pped wi th a gas fuel ed modi fi ed FI AT di esel engi ne coupl ed wi th a 25KVA al ternator. The generated power i s consumed by means of el ectri c resi stances pl aced i n a water bath. The maxi mum consumabl e power i s 24 kW, and the effecti ve uti l i zed power can be vari ed between 1 and 24 kW by 1 kW steps. I n parti cul ar, duri ng our experi ments the di ssi pated power was settl ed at 21 kW. The gasi fi er was operated at a temperature of about 850 C. Ai r was fed i nto the gasi fi er at an average fl ow rate of 36 N m 3 /h. Sampling. Gas sampl i ng was performed i soki neti cal l y accordi ng to a geometry reproduced i n Scheme 3. Thi s geom- etry was eval uated to be the most sui tabl e to ensure the col l ecti on of the fl ue gas mai n consti tuents. The gas mi xture was automati cal l y sampl ed wi th a fl ow rate of 5 L/mi n. To avoi d hi gh mol ecul ar wei ght tar deposi ti on, a heati ng el ement pri or to the parti cl es col l ector ensured a mi ddl e temperature of al most 200 C. The organi c and i norgani c sampl i ngs were executed separatel y to avoi d any foul i ng of the sampl i ng system caused by the sol i di fi cati on of water from the i norgani c sol uti ons on the di chl oromethane col l ecti ng sol uti on. Wi th regard to the tar sampl i ng, gas was bubbl ed i n a fi ve washi ng bottl es system fi l l ed wi th di chl oromethane. The fi rst two bottl es were cool ed i n an i ce bath whi l e the l ast ones were cool ed i n a -10 C thermostati c bath. Onl y fresh sol uti ons were consi dered for tar anal ysi s i n order to avoi d any modi fi cati on i n composi ti on caused by repol ymeri zati on reacti ons or evapo- rati on of organi c sol vent. A three bottl es system was empl oyed to col l ect the i norgani c speci es embedded i n the fl ue gas. The aci di c consti tuents were determi ned by col l ecti ng the fl ue gas on Na2CO3/NaHCO3 (2.7 mM/0.3 mM), whi l e ammoni a was col l ected onto a 5 mM H2SO4 sol uti on. Both sol uti ons were cool ed at 0 C. The col l ecti ng equi pment was previ ousl y ri nsed wi th abundant bi -di sti l l ed water and fi nal l y ri nsed wi th the col l ecti ng sol uti on. Al l the concentrati ons were normal i zed for the col l ected gas vol ume referred to normal condi ti ons. Sl uggi ng materi al was al so col l ected from the upper part of the gasi fi er i n order to eval uate i ts chemi cal composi ti on. Three di fferent pl ant zones were chosen for the ammoni a content eval uati on: (1) after the gl ass fi l ter; (2) i n the scrubber sol uti on; (3) i n the condensed phase onto the chi ps fi l ter. Analyses of Anions. To determi ne el ements of i nterest i n the tested bi omass and i n the fl ue gas, i on chromatography anal yses wi th a HPI C DX 300 system were performed. Accuratel y wei ghed bi omass sampl es of approxi matel y 0.3 g were fi rst combusted i n 30 bar oxygen usi ng a Mal her bomb (I SO-587, ASTM 2361-66). The sol uti ons contai ni ng the ani ons from bi omass (Cl , S) were then el uted through an AS 12 i onpac col umn. Tests on several standards i ncl udi ng 3-chl ori ne, 4-fl uori ne ani l i ne, and bromothymol bl ue showed reproduci bl e recoveri es of al most 90% for chl ori ne and 60% for sul fur speci es. (7) DeGroot, W. F.; Ri chards, G. N. I nfl uence of Pyrol ysi s Condi ti ons and I on -Exchange Catal ysts on the Gasi fi cati on of Cottonwood Chars by Carbon Di oxi de. Fuel 1988, 67, 352-360. (8) DeGroot, W. F.; Ri chards, G. N. The Effect I nfl uence of I on- Exchanged Cobal t Catal ysts on the Gasi fi cati on of Wood Chars by Carbon Di oxi de. Fuel 1988, 67, 345-351. (9) DeGroot, W. F.; Kannan, M. P.; Ri chards, G. N.; Theander, O. Gasi fi cati on of agri cul tural resi dues (bi omass): I nfl uence of i norgani c consti tuents. J . Anal. Appl. Pyrol. 1984, 38, 320-323. (10) Agbl evor, F. A.; Besl er, S. I norgani c Compounds i n Bi omass Feedstocks. 1. Effect on the Qual i ty of Fast Pyrol ysi s Oi l s. EnergyFuels 1996, 10, 293-298. (11) Leppal ahti , J.; Kurkel a, E. Behavi our of Ni trogen Compounds and Tars i n Fl ui di zed Bed Ai r Gasi fi cati on of Peat. Fuel 1991, 70, 491- 497. (12) Leppal ahti , J.; Kurkel a, E.; Si mel l , P. Catal yti c Conversi on of Ni trogen Compounds i n Gasi fi cati on Gas. Fuel Process. Technol. 1991, 29, 43-56. (13) Leppal ahti , J. Formati on of NH 3 and HCN i n sl ow-heati ng-rate i nert pyrol ysi s of peat, coal and bark. Fuel 1995, 74, 1363. (14) Leppal ahti , J.; Kol jonan, T. Ni trogen Evol uti on from Coal , Peat and Wood Duri ng Gasi fi cati on: Li terature Revi ew. Fuel Process. Technol. 1995, 43, 1-45. (15) Best practi ces l i st for bi omass fuel and ash anal ysi s; ECN, Report No. ECN-C-98-0875; Netherl ands Energy Research Founda- ti on: Petten, The Netherl ands, 1998. (16) Sander, M. L.; Eri csson, T. Verti cal Di stri buti ons of Pl ant Nutri ent and Heavy Metal s i n Sal i x Vi mi nal i s Stems and Thei r I mpl i cati ons for Sampl i ng. Biomass Bioenergy 1998, 14 (1), 57-66. 890 Energy & Fuels, Vol. 14, No. 4, 2000 DeBari et al. The Na2CO3/NaHCO3 col l ecti ng sol uti ons from pl ant runs were anal yzed for thei r content i n the fol l owi ng ani ons: Cl - , NO2 - , NO3 - , SO4 2- , PO4 3- , CH3COO - (acetates), and HCOO - (formi ates). An el uti ng mi xture of Na2CO3/NaHCO3 (2.7 mM/ 0.3 mM) was sel ected to separate Cl - , NO2 - , NO3 - , SO4 2- , and PO4 3- , whi l e an el uti ng mi xture of B2O7 2- /NaOH (75 mM/100 mM) was shown to be sui tabl e to di scern the contri buti on from formi ates and acetates. I denti fi cati on and quanti fi cati on were performed by compari son of retenti on ti mes and responses wi th those of standard sol uti ons. FTIR Mesurements. The sl udgy condensate col l ected i n the upper part of the gasi fi er was subjected to several treatments to obtai n sampl es easy to handl e. I n thi s regard, several sol ubi l i zati on procedures were eval uated. The more effecti ve one provi ded a prel i mi nary treatment wi th an al kal i ne sol uti on 1 N to separate the pol ar from the hydro- phobi c fracti on. Mass bal ance esti mati ons reveal ed that the sol i d hydrophobi c fracti on separated i n thi s way i s at l east 1%of the col l ected materi al , and the FTI R spectrum i ndi cated the char features of thi s fracti on. The basi c sol uti on was then aci di fi ed wi th 1 N H2SO4 unti l pH 2 was reached. I n thi s Scheme 1. Fixed Bed Gasification Plant Used a a I n the scheme are al so i ndi cated the fl ue gas sampl i ng poi nts (SP1 and SP2). Scheme 2. Fixed Bed Gasifier Scheme 3. Equipment Adopted for Tar Sampling a a Anal ogous equi pment was used for i norgani c sampl i ng wi th the excepti on of the - 10 C secti on. Biomass Air Gasification in a Downdraft Fixed Bed Energy & Fuels, Vol. 14, No. 4, 2000 891 condi ti on a sol i d materi al preci pi tated. Fol l owi ng fi l trati on on a gl ass fi ber fi l ter, thi s materi al was col l ected, dri ed i n an oven at a temperature of 90 C for 4 h, and then used to prepare KBr pel l ets i n l ess than 1wt % concentrati on. I R spectra were recorded wi th a Perki n-El mer 2000 spectrometer i n the transmi ssi on mode, adopti ng a resol uti on of 2 cm -1 . Spectra were al so recorded i n ATR mode. No si gni fi cant di fferences were observed. Elemental Analysis(C, H, N, O) andFixedCarbon. For the determi nati on of the el emental composi ti on a Perki n- El mer 2400 mul ti el ement anal yzer was empl oyed. Approxi - matel y 3-4 mg of dri ed and pul veri zed sampl e was wei ghed wi th an accuracy of 0.01 mg and pl aced i n an al umi num contai ner. Hel i um was used as carri er gas. Fi xed carbon was determi ned accordi ng to ASTM D3175 and D3172 standard procedures. Ash Determination. Ash determi nati on was performed wi th a sampl e of ca. 1 g i n a furnace accordi ng to ASTM D1102. Extracts, Lignin, and Carbohydrates. The sampl es (ground at 50 mesh and dri ed to 105 C) were extracted i n a Soxhl et wi th an ethanol /tol uene (1/2) sol uti on for 7 h. Thi s procedure (ASTM D1106, D1107, D1108) ensured the extrac- ti on of resi ns, fats, wax, oi l s, and catechol s. The percentages of the extracts were referred to the dri ed bi omass. Li gni n quanti fi cati on was performed on the sampl es previ - ousl y extracted wi th the ethanol /tol uene mi xture. The amount of Kl ason l i gni n was quanti fi ed accordi ng to the Mod. Tappi T-13 m-54 and ASTM D1106 standard method. The sampl es were treated wi th concentrated sul furi c aci d (72% (w/w)). The rel evant sol uti on was then di l uted (3%), heated unti l i t reached ebul l i ti on, and kept i n ebul l i ti on for 4 h at constant vol ume. An i nsol ubl e resi due of l i gni n (Kl ason l i gni n) was obtai ned. I ts percentage composi ti on was referred to the dry matter wi th the extracti ves. The aci di c sol uti on obtai ned from the previ ous step was anal yzed by means of a HPI C DX 300 equi pped wi th a Carbopac 100 col umn i n order to eval uate the total amount of carbohydrates. The anal yti cal resul ts were corrected wi th experi mental factors to account for the sugars degradati on occurri ng duri ng the hydrol ysi s step. Analysisof Heavy andAlkali-AlkalineEarth Metals. The metal content of the bi omass used was establ i shed on ground and homogeni zed sampl es di gested i n ACV cl osed vessel s (vol ume, 100 mL) wi th a mi xture of HNO 3/HF (10:1). Quartz fi l ters from the pl ant were combusted i n a muffl e operati ng at 600 C to compl etel y remove the organi c fracti on (char). The i norgani c fracti on was then subjected to mi crowave di gesti ons i n ACV cl osed vessel s. The aci d vol ume used i n the di gesti on procedure was a mi xture of 10 mL of HNO3 and 15 mL of HF. The di gesti on temperature was al most 180 C, and the durati on of the mi neral i zati on procedure was 40 mi n. Di gesti on of two bl ank fi l ters was al so executed under the same condi ti ons of the sampl es i n order to eval uate the contri buti on from the col l ecti ng medi um. The di gesti ons were performed wi th a CEM-2000 oven equi pped wi th both a pressure and a temperature control uni t. Prepared i norgani c sampl es were anal yzed i n a Vari an Ul tramass I CP-MS spec- trometer. Before the determi nati on, sampl es were di l uted 500 ti mes. Sampl es were stored i n bottl es made of PC prel i mi nari l y condi ti oned i n a 20% HNO3 sol uti on and then abundantl y ri nsed wi th bi -di sti l l ed water. Ammonia Determination. The ammoni a content was determi ned accordi ng to the Nessl er method. Aqueous sol u- ti ons from the wet scrubber were fi rst fi l tered on an acti vated carbon bed to remove the organi c speci es, maki ng the sol uti on l ook col ored, and subsequentl y anal yzed accordi ng to the standard procedure (ASTM D1426-79). CyanideDetermination. The cyani de content of the fl ue gas was determi ned on the col l ecti ng basi c water sol uti on by means of i oni c chromatography. A sol uti on of C2O4 2- /NaOH (7.5 mM/0.2 mM) was used to i socrati cal l y el ute sampl es through a Carbopac 100 col umn. The empl oyed el ectrochemi cal detector was an ED40 si l ver worki ng el ectrode anodi cal l y pol ari zed at 0.1 V. SamplePreparation for GC-MS Analysis. Tar sol uti on from the pl ant was pl aced i n a separator funnel to al l ow water- phase separati on. Wi th the excepti on of a few runs no si gni fi cant amount of the water phase condensed onto the di chl oromethane sol uti on. Prepared tar sampl es were anal yzed i n a HP 6890 GC-MS. A HP-5MS (cross-l i nked 5% PH ME si l oxane, 30 m 0.25 mm 0.25 m fi l m thi ckness) col umn was used for tar anal ysi s. GC anal ysi s provi ded an expl orati on of the mass regi on from 35 up to 300 amu. I denti fi cati on of each component was based on the anal yte retenti on ti mes from a certi fi ed mi xture of pol ynucl ear aromati c hydrocarbons (EPA 610 cal i brati on standard, Supel co I nc.), BETX, and phenol standards (certi fi ed by ULTRA Sci enti fi c). Al so, the quartz fi l ters used to col l ect parti cul ate were extracted for 2 h wi th di chl oromethane i n a Soxl et apparatus to eval uate the pres- ence of aromati c compound on the parti cul ate by means of GC- MS anal ysi s. The obtai ned resul ts showed no presence of tar traces on the quartz fi l ter. Materials. The aci ds used i n the di gesti on of bi omass sampl es were HNO 3 (70%, E. Merck, suprapur) and HF (40%, E. Merck, suprapur). Al l other chemi cal s were of anal yti cal grade. Bi omass sampl es for chemi cal characteri zati on were prepared by fi rst gri ndi ng a si gni fi cant amount of bi omass and then subsampl i ng for anal ysi s. Surface Area Measurements. Mul ti poi nt BET anal yses of surface area were determi ned by N 2 adsorpti on at -6 C usi ng a Quantacrom Nova 2200 apparatus. Table 1. Elemental Composition (wt %) and Fixed Carbon on Dry Matter of the Three Biomass Feedstocks feedstock C H O N S Cl fi xed carbon al mond shel l s 45.0 5.0 48.0 0.40 0.0170 ( 0.0012 0.0100 ( 0.0005 16.32 oak 50.0 6.0 43.0 0.10 0.027 ( 0.005 0.04 ( 0.01 14.39 turkey oak 50.0 6.0 43.0 0.10 0.101 ( 0.005 0 a 15.22 a Onl y trace el ements were detected, and the rel evant si gnal s were bel ow the determi nabi l i ty l i mi ts. Table 2. Mean a Trace Metal Contents (mg/Kg in ash) of Three Different Biomass Feedstocks feedstock metal al mond shel l s oak turkey oak Na b 5.8 4.0 4.3 Mg b 10.7 12.6 12.28 Al b 3.7 0.56 2.66 K b 330 38 65 Ca b 55 250 400 Fe b 3.6 9.0 10.79 Cd 4.0 3.9 3.2 Cr 740 194 345 Cu 490 172 105 Mn 540 980 Nd Ni 1220 340 450 Pb 2000 269 260 Zn 3700 1180 2020 a Data reported for al kal i and heavy metal s are affected by two sources of errors: the vari ance wi thi n-sampl e and the vari ance between sampl es. The fi rst val ues are usual l y negl i gi bl e (0.5%), whi l e the second val ues are of the order of 10%. Thi s resul t coul d be expl ai ned by taki ng i nto account that pl ant nutri ent and heavy metal concentrati on coul d vary al ong the wood trunk. 16 b Rel evant amounts are di vi ded for 10 3 . 892 Energy & Fuels, Vol. 14, No. 4, 2000 DeBari et al. Results and Discussion CharacterizationAnalysisof BiomassResidues. Tabl e 1 summari zes the el emental composi ti on (C, H, O, N, S, Cl ) and the fi xed carbon content of the three bi omass feedstocks l oaded, whi l e Tabl e 2 gi ves the rel evant metal composi ti ons. The resul ts reported i n Tabl es 1 and 2 suggest i mmedi atel y that al kal i metal s i n the bi omass matri x cannot be onl y l i ke i norgani c sal ts such as KCl . I n fact, the potassi um concentrati on i n the anal yzed bi omass i s at l east 1000-fol d the chl ori ne concentrati on. The presence of appreci abl e amounts of heavy metal s (Cd, Ni , etc.) extraneous to the i ndi genous composi ti on of the tested bi omass coul d be due to a bi osorpti on process whi ch mi ght i nvol ve the carboxyl , the ami ne, and the ami de groups of the bi omass. The compari son of the bi omass composi ti ons detai l ed i n Tabl e 2 suggest that whi l e oak and turkey oak show a qui te si mi l ar i norgani c composi ti on, al mond shel l s are di fferent. I n parti cul ar, i t i s i mmedi atel y evi dent from the data reported above that cal ci um, whi ch i s l i kel y to be predomi nantl y present i n cel l wal l materi al , i s more abundant i n woody bi omass compared to al mond shel l s. An opposi te resul t was found for potassi um i on. Tabl e 3 i s a compi l ati on of the detai l ed composi ti on of the measured sugar content i n the tested bi omass, whi l e Tabl e 4 i ncl udes the percentages of the other consti tuents. Consi deri ng that summary was affected by an error of at l east 10%, the mass bal ance for the anal yzed bi omass can be consi dered to be al most 100%. On the basi s of these resul ts, i t can be concl uded that al though woody bi omass and al mond shel l s contai n a comparabl e amount of sugars, the content of the most abundant components, gl ucan and xyl an, i s al most i nverted by passi ng from woody bi omass to al mond shel l s. Other di fferences resi de i n the l i gni n and ash content. I n parti cul ar al mond shel l s contai n hi gher l i gni n content than the anal yzed woody bi omass, whi l e they gi ve a l ower ash yi el d. The observed di fference i n the l i gni n content i s al so consi stent wi th the recorded FTI R spectra of the dri ed raw bi omass shown i n Fi gure 1. I n parti cul ar, useful i ndi cati ons about the hol ocel l u- l ose (cel l ul ose and hemi cel l ul ose) to l i gni n rati o can be deduced by measuri ng the i ntensi ti es rati o of the absorpti on bands at 1738 and 1510 cm -1 , respecti vel y (Owen and Thomas 17 ). As al ready reported by Owen, the i ntensi ty measurements were carri ed out on a base l i ne reference measured by eye for each sampl e. The rati os esti mated wi th thi s procedure are comparabl e to that cal cul ated on the basi s of the chemi cal method. Inorganic ConstituentsCompositionof theGas. Fi gure 2 di spl ays the composi ti on of al kal i -al kal i ne earth and heavy metal s i n the gas mi xture rel evant to the three feedstocks. The resul ts seem to be not si gni fi - cantl y di fferent. The gas concentrati ons for heavy and al kal i metal s i ndi cate that the major amount of i nor- gani c compounds remai ns at the ash bed i n the bottom (17) Owen, N. L.; Thomas, D. W. I nfrared Sudi ed of Hard and Soft Woods. Appl. Spectrosc. 1989, 43 (3), 451-457. Table 3. Sugar Percentages into the Three Biomass Feedstocks araban gal actan gl ucan xyl an cel l obi an a al mond shel l s 1.31 ( 0.13 1.12 ( 0.11 23 ( 2 33.09 ( 1.1 2.08 ( 0.08 oak 1.18 ( 0.12 1.13 ( 0.14 32.8 ( 1.3 23.7 ( 1.8 2.21 ( 0.19 turkey oak 1.24 ( 0.12 1.45 ( 0.17 33.0 ( 1.8 23.1 ( 1.3 2.6 ( 0.2 a The cel l obi an i n the anal yzed sol uti on i s due to an i ncompl ete hydrol ysi s of the cel l ul ose backbone. Table 4. Percentage Composition of Three Biomass Feedstocks a sugars l i gni n ash b extracti ves c al mond shel l s 60.6 29.0 ( 1.2 1.38 2.9 oak 61.02 24.4 ( 0.4 2.13 5.6 turkey oak 61.39 23.0 ( 0.5 2.17 4.3 a Data reported are referred to the bi omass dri ed at 105 C. b The reported resul ts are affected by standard errors of 5%. c Thi s extracti ves percentage compri ses compounds such as resi ns, fats, wax, oi l s, and cathecol s. The reported resul ts are affected by errors of 7%. Figure 1. FTI R spectra acqui red from turkey oak (upper curve) and al mond shel l (l ower curve) dri ed raw materi al . Figure 2. Metal s di stri buti on i n col l ected fl y ash sampl es from gasi fi cati on of woody bi omass. Biomass Air Gasification in a Downdraft Fixed Bed Energy & Fuels, Vol. 14, No. 4, 2000 893 of the gasi fi er. I t i s worth noti ci ng that whi l e heavy metal s retai n al most the same rel ati ve di stri buti on to one another i n the fl y ash wi th respect to the anal yzed composi ti on of the raw bi omass, a predomi nant presence of potassi um can be observed i n the fl y ash. The hi ghest vol ati l i ty of KCl among the i norgani c compound present i n the bi omass was often restated to expl ai n the potas- si um presence i n the col l ected fl y ash. 18 However, KCl vol ati l i zati on mi ght not be the onl y mechani sm, proven by the fact that turkey oak feedstocks gi ve fl y ash wi th more potassi um than that from oak despi te the l ower chl ori ne content. Thi s fi ndi ng i ndi cates that addi ti onal reacti ons mi ght be i nvoked. Among others, organi c sal ts such as R-COOK on the surface of char coul d be precursors of the formati on of potassi um i norgani c speci es i n the fl y ash. The FTI R spectrum of a typi cal char carri ed i n suspensi on by the gas shows a wi de absorpti on band i n the regi on of 1600 cm -1 , where carboxyl -carbonate functi onal i ty fal l s. 19 Tabl e 5 l i sts characteri sti c absorpti on bands recogni zabl e i n the FTI R spectrum of a col l ected char sampl e wi th the assi gnments rel evant to the chemi cal functi onal groups. Unfortunatel y, al though the reported assi gnments con- fi rm the presence of oxygenated functi onal i ty, they are weak and cannot be consi dered concl usi ve, especi al l y i n the part of the spectrum contai ni ng O-H stretchi ng absorpti on bands. I n fact, as a resul t of the hi gh hygroscopi ci ty observed for these materi al s, the pres- ence of water traces i s expected. Whi l e the features from the acqui red FTI R spectra do not vary wi th the bi omass l oaded, an approxi matel y 2-fol d i ncrease was observed for the char col l ected from al mond shel l s wi th respect to woody bi omass. Thi s resul t, whi ch pri mari l y refl ects a di fferent percentage of fi xed carbon (Tabl e 1) i n the tested bi omass, can be expl ai ned not onl y i n l i ght of a di fferent l i gni n percentage 20 but al so because of the di fferent ash content and composi ti on. More i n detai l , a hi gher ash content, especi al l y cal ci um i ons, i n woody bi omass was al ready found to be responsi bl e for a hi gher catal yti c effect toward char gasi fi cati on i n a CO 2 atmo- sphere. 7 Fi gure 3 di spl ays the concentrati ons of acetates and potassi um i n the fl ue gas for woody bi omass. I t seems i nteresti ng the fact that the amount of potassi um quanti fi ed on the quartz fi l ter i s comparabl e to the amount of acetates col l ected i n the ani oni c sampl i ng sol uti on. Thi s resul t suggests a common source i n the formati on pathway of these speci es. Some hypotheses about thermal degradati on of al kal i ne carboxyl ates have al ready been formul ated by Zi ntl et al . 21 Unfortunatel y, due to the hi gh concentrati on of aceti c aci d detected duri ng the test runs wi th al mond shel l s, i t was not possi bl e to veri fy the above correl ati on al so i n these experi ments (vi de i nfra). Fi gure 4 di spl ays the measured gas composi ti on for some ani ons of i nterest. The reported errors were computed on the number of the test runs. I t i s apparent that whi l e no si gni fi cant di fferences exi st among the exami ned bi omass for the concentrati on of chl ori ne, (18) Kaufmann, H.; Nussbaumer, T. Characteri sti c and Formati on of Fl y Ash Parti cl es i n Bi omass Furnaces. I n Biomass for Energy and I ndustry, Proceedings of theI nternational ConferenceWurzburg; Ko- petz, H., Weber, T., Pal z, W., Charti er, P., Ferrero, G. L., Eds.; C.A.R.M.E.N.: Germany, 1998; pp 1327-1329. (19) Fanni ng, P. E.; Vanni ce, M. A. A Dri ft Study of the Formati on of Surface Groups on Carbon by Oxi dati on. Carbon 1993, 31, 721- 730. (20) Zanzi , R.; Sjostrom, K.; Bjornbom, E. Rapi d hi gh-temperature pyrol ysi s of bi omass i n a free-fal l reactor. Fuel 1996, 75, 545. (21) Zi ntl , F.; Stomberg, B.; Bjorkman, E. Rel ease of chl ori ne from bi omass at gasi fi cati on condi ti ons. I n Biomassfor Energyand I ndustry, Proceedings of the I nternational Conference Wurzburg; Kopetz, H., Weber, T., Pal z, W., Charti er, P., Ferrero, G. L., Eds.; C.A.R.M.E.N.: Germany, 1998; pp 1608-1611. Table 5. Functional Groups on the Surface of Char Collected during Biomass Gasification absorpti on freq (cm -1 ) assi gnt a 1690 CdO stretch of aromati c aci d 1631 CdC stretch 1383 C-O stretchi ng vi brati ons 1270 C-O-C asymmetri c stretchi ng 1120 C-O-C symmetri c stretchi ng a Assi gnment regi ons are reported i n ref 18 and i n the refer- enceds ci ted therei n. Figure 3. Compari son between acetate and potassi um con- centrati ons i n the fl ue gas for woody bi omass. Figure 4. HPLC anal ysi s of the fl ue gas i norgani c composi - ti on (chl ori ne, ni tri te, ni trate, sul fate, acetate). 894 Energy & Fuels, Vol. 14, No. 4, 2000 DeBari et al. ni tri tes, ni trates, and sul fates i n the fl ue gas, the concentrati on of acetates i n the fl ue gas from al mond shel l s i s 60-fol d that from woody bi omass. Measures of pH i n the water phases condensed on the di chl o- romethane duri ng the tar sampl i ng step confi rm the predomi nance of the aci di c consti tuents i n the fl ue gas from al mond shel l s. More exactl y, water phases from al mond shel l s have pH 5, whi l e the ones from woody bi omass have about pH 9. A reasonabl e expl anati on for thi s di fference coul d be the di fferent di stri buti on of the hol ocel l ul ose consti tuents al ready di scussed on the basi s of the resul ts l i sted i n Tabl e 3. I n parti cul ar, the hi ghest content of xyl an i n al mond shel l s can justi fy the di ffer- ence i n the measured aceti c aci d concentrati on i n the fl ue gas from the two di fferent feedstocks consi deri ng that xyl anes i n hemi cel l ul ose are acetyl ated i n the O-2 posi ti on. Thi s experi mental resul t corroborates the theory proposed by Evans and Mi l ne 22 to expl ai n the presence of aceti c aci d i n the mass spectral patterns rel evant to the bi omass pyrol ysi s products. I n the anal ysi s of products di stri buti on from gasi fi ca- ti on processes, attenti on must be devoted to the study of i norgani c ni trogen compounds i n the fl ue gas because of the probl ems connected wi th thei r burni ng. I n fact, i t i s evi dent from Fi gure 5 that onl y a smal l part of the ammoni a contai ned i n the fl ue gas was removed by the wet scrubber before enteri ng the chi ps fi l ter secti on. The anal ysi s of water phase condensed on the chi ps fi l ter confi rmed the presence of an appreci abl e amount of ammoni a. Fi gure 6 provi des the ammoni a and cyani de concen- trati on i n the fl ue gas as a functi on of the gasi fi cati on rati o for the two woody bi omass and al mond shel l s. Data reported make evi dent that ammoni a evol uti on from the two bi omass cl asses i s di fferent. Less pronounced di f- ferences can be observed i n the cyani de trends. I n parti cul ar, the ammoni a concentrati on for al mond shel l s i s al most 2-fol d that for the two woody bi omasses at the same gasi fi cati on rati o. To i nvesti gate the rel ati on between the ni trogen content of the tested bi omass and the amount of ammoni a produced duri ng the gasi fi ca- ti on process, the el emental anal ysi s was performed (Tabl e 1). As expected, data obtai ned show that the ni trogen percentage i n al mond shel l feedstocks i s 4-fol d that i n woody feedstocks. On the basi s of these resul ts, i t can be concl uded that ammoni a evol uti on i s pri mari l y affected by the amount of ni trogen i n the l oaded feedstock, al though the eval uated percentage conversi on of the fuel -ni trogen i s bel ow 1%. Besi des thi s i nforma- ti on, Fi gure 6 al so i ndi cates that ammoni a and cyani de concentrati ons i n the fl ue gas are al so i nfl uenced by the gasi fi cati on rati o, to say they are i nfl uenced by the rel ati ve amount of ai r (oxygen) i njected i nto the gasi fi er wi th respect to the l oaded bi omass. I t i s worth noti ng that whi l e ammoni a concentrati on i ncreased wi th the i ncreasi ng of the gasi fi cati on rati o, the cyani de concen- trati on dropped to zero i n a comparabl e range of val ues. Thi s behavi or coul d be expl ai ned by the oxi dati on of cyani de by mol ecul ar oxygen i n ai r, 23 taki ng i nto account that the presence of woody charcoal and copper speci es i n the same reacti on medi um can make thi s thermody- nami c oxi dati on ki neti cal l y possi bl e. The potenti al reac- ti on pathways are reported bel ow: Measures of BETi sotherm were carri ed out to expl ore i f the gasi fi cati on condi ti ons are effecti ve toward a certai n acti vati on of the woody charcoal so maki ng i t possi bl e the descri bed reacti ons. Tabl e 6 di spl ays the measured surface area for the charcoal i ntroduced i n the gasi fi er necessary to burn the bi omass i n the starti ng up stage and the charcoal at the end of a typi cal pl ant runni ng. The compari son between the reported resul ts suggests that, at the operati ng condi ti ons se- l ected for the test runs, an appreci abl e i ncrease of about 1 order of magni tude of the charcoal surface area has been observed Compari ng the resul ts of thi s study wi th that reported by Leppal ahti et al . 12 for gasi fi cati on of wood i n updraft fi xed bed gasi fi ers, some val uabl e di fferences can be outl i ned. I n parti cul ar, data col l ected from the updraft gasi fi er showed a hi gher percentage conversi on of fuel ni trogen both i n the case of ammoni a and cyani de formati on. Thi s resul t i s coherent wi th the downdraft gasi fi er geometry that provi des the passage of ammoni a and cyani de speci es over the ash bed i n the boi l er where certai n decomposi ti on reacti ons can be favored from the presence of i norgani c speci es such as CaO, MgO, and Ni O. Tar Content andComposition in theGas. Fi gure 7 provi des a typi cal chromatogram for a tar sampl e, whi l e i n Tabl e 7 the detected and quanti fi ed consti tu- ents of the three bi omass feedstocks are l i sted. I t i s evi dent from the reported amounts that tol uene and i ts deri vati ve compounds such as xyl ene are the most abundant consti tuents fol l owed by phenol , naphthal ene, and traces of pol yaromati c hydrocarbons. Moreover, tar composi ti on does not exhi bi t appreci abl e amounts of ni trogen deri ved compounds i n the cases of both woody bi omass and al mond shel l s. For a better i l l ustrati on of the changes i n the di stri buti on of tar consti tuents i n (22) Evans, R. J.; Mi l ne, T. A. Mol ecul ar Characteri zati on of the Pyrol ysi s of bi omass. 1. Fundamental s. Energy Fuels 1987, 1, 123- 137. (23) Chen, Y. You, C. Cyani de Destructi on by Catal yti c Oxi dati on. Metal Finishing 1991, 89 (11), 68-71. Figure 5. Compari son between the ammoni a and cyani de concentrati ons i n the fl ue gas at sampl i ng poi nts 1 and 2 (SP1 and SP2, respecti vel y, of Scheme 1). 2CN - + O 2 ) 2CNO - 2CNO - + 3 / 2 O 2 + H 2 O ) N 2 + 2CO 2 + 2OH - CNO - + 2H 3 O + ) CO 2 + NH 4 + + H 2 O Biomass Air Gasification in a Downdraft Fixed Bed Energy & Fuels, Vol. 14, No. 4, 2000 895 the three bi omass feedstocks, the quanti fi ed compounds were di vi ded i nto four cl asses: one ri ng compounds, phenol and phenol type compounds, two and three ri ngs compounds (Fi gure 8). Resul ts show that woody bi omass and al mond shel l tar composi ti ons di ffer onl y i n the rel ati ve percentage of phenol and phenol type com- pounds. The magni fi cati on di spl ayed i n Fi gure 8 enabl es thi s di fference to emerge. Thi s resul t i s coherent wi th Figure7. GC-MS chromatogram of tar speci es col l ected on a CH2Cl 2 sol uti on duri ng a gasi fi cati on of oak bi omass. Table 6. Measured Surface Areas (m 2 /g) surface area charcoal i ntroduced i n the gasi fi er 16 charcoal at the end of the process 110 Table 7. Mean Percentage Compositions of the Identified Tar Constituents for Almond Shells, Oak, and Turkey Oak a compd al mond shel l s oak turkey oak (1) tol uene 41.29 46.07 62.46 (2) ethyl benzene 22.71 10.52 10.71 (3) p-xyl ene 4.24 7.73 5.74 (4) styrene 27.86 32.56 18.16 (5) phenol 1.32 0.35 0.19 (6) i ndene 0.10 0.41 traces (7) o-cresol 0.56 0.07 traces (8) p-cresol 1.04 0.19 traces (9) naphthal ene 0.38 1.06 1.72 (10) 1-methyl naphthal ene traces 0.16 traces (11) 2-methyl naphthal ene traces 0.11 traces (12) acenaphthyl ene 0.12 0.38 0.32 (13) acenaphthene traces traces traces (14) fl uorene traces traces traces (15) phenantrene traces 0.15 0.25 (16) anthracene traces traces traces (17) fl uoranthene traces traces 0.13 (18) pyrene traces traces 0.19 a From the chromatogram, typi cal vari ances wi thi n-sampl e were about 5% for the more abundant compounds (benzene and naphthal ene types) and about 15-20 for the trace consti tuents. The esti mated vari ance between runs was typi cal l y not more than 20%. Figure6. Ammoni a and cyani de concentrati ons i n the fl ue gas from woody bi omass and al mond shel l s versus gasi fi cati on rati o. Note that the reported concentrati ons for ammoni a and cyani de speci es do not correspond to the same sampl i ng. Figure8. Rel ati ve di stri buti on of tar compounds grouped i nto four pri nci pal cl asses: one ri ng type (tol uene, ethyl benzene, xyl enes, and styrene), phenol type (phenol and cresol s), two ri ngs type (naphthal ene, 1-methyl naphthal ene, 2-methyl naph- thal ene, and i ndene), three ri ngs type (phenanthrene, an- thracene, acenaphthyl ene, acenaphthene, and fl uorene), and four ri ngs type (fl uoranthene and pyrene).The i nsert magni fi es the phenol compounds regi on of the i stogram. 896 Energy & Fuels, Vol. 14, No. 4, 2000 DeBari et al. the pattern observed i n the l i gni n content (Tabl e 4). More pronounced di fferences i n tar produced from the three bi omass feedstocks are evi dent i n Fi gure 9, where the total amounts of produced tar for the tested bi omass are compared. Here, the amount of col l ected tar from al mond shel l s i s al most 10-fol d that from woody bi om- ass. These resul ts coul d be tentati vel y expl ai ned by the catal yti c acti vi ty of the i ndi genous al kal i ne earth metal s such as cal ci um toward the cracki ng of hi gh mol ecul ar tar i n vi ew of thei r acti vi ty i n catal ysts such as dol omi te and cal ci te. I n parti cul ar, at the gasi fi er runni ng tem- perature of 850 C, i norgani c speci es such as cal ci um and magnesi um i n the bi omass feedstock are l i kel y to be present as CaO and MgO, to stay i n the acti ve form to catal yze tar cracki ng. Consi deri ng that woody com- posi ti on i s ri cher i n cal ci um content than al mond shel l s, the woody fuel coul d be subjected to a cracki ng operated by i ndi genous catal ysts. Al so i t cannot be excl uded that a di fferent content i n the l i gni n component mi ght have affected the amount of produced tar. I n fact, besi des the fact that the l i gni n backbone coul d be a source of phenol compounds for the produced tar, i t i s wel l -known that i n the bi omass matri x i t pl ays the rol e of l i nki ng agent for the hol ocel l ul ose component. A major amount of l i gni n can account for a more di ffi cul t defi beri zati on process of the bi omass i n the process starti ng up. I n parti cul ar, a major amount of l i gni n coul d make the hol ocel l ul ose backbone l ess avai l abl e i n the pyrol ysi s zone causi ng a hi gher uncombusted parti al l y cracked product fl ow from the combusti on to the reducti on zone of the gasi fi er i tsel f. Characterization of Sludgy Material. Fi gure 10 di spl ays the FT-I R spectrum of a typi cal sl udge sampl e that tends to condense i n the upper part of the gasi fi er duri ng the earl i er phases of the process. The col l ected spectrum l ooks si mi l ar to a typi cal spectrum from a l i gni n sampl e so confi rmi ng the exi stence of a defi ber- i zati on step i n the gasi fi cati on process duri ng whi ch some fragments of l i gni n remai n unal tered. Gasification Rate. Fi gure 11 reproduces the di mi - nuti on of the bi omass feedstock l oaded as a functi on of the ti me. The reported sl opes for the three bi omasses i nvesti gated are a measure of the gasi fi cati on rates. The compari son of the reported val ues suggests that the gasi fi cati on of al mond shel l s i s sl ower than that of woody bi omass. Thi s resul t i s di fferent from that expected on the basi s of the fact that al mond shel l s are smal l er than woody materi al and, thanks to a major uni formi ty of the temperature i n the reacti on zone of the gasi fi er, they shoul d gasi fy more qui ckl y. Al so i n thi s case the di fferent amount of l i gni n measured i n the two anal yzed feedstocks coul d expl ai n the observed di fference. I n parti cul ar, a di fferent l i gni n percentage i s l i kewi se refl ected i n the bi omass morphol ogy. I n fact, al mond shel l s have a morphol ogy harder than wood other than l ess porous. Detai l ed exami nati on of the i norgani c composi ti on l i sted i n Tabl e 2 seems to i ndi cate that i f al kal i and al kal i ne earth metal s pl ay a rol e i n determi ni ng the gasi fi cati on rate, cal ci um i s the more effecti ve speci es. Thi s resul t agrees wi th the bench experi ments run by DeGroot and Ri chards, 7 whi l e i t does not match perfectl y data reported from some authors 24,25 i n the study of other carbonaceous materi al s about the potassi um reacti vi ty. The gai ned i nformati on confi rms that the i norgani c content of the bi omass cannot expl ai n at al l the observed phenomena and that i t must be i nserted i n the context of a compl ete structural anal ysi s of the substrate. Other i nvesti ga- Figure 9. Compari son among the total amount (mg/NL) of tar from oak, turkey oak, and al mond shel l s. Figure 10. FT-I R spectrum rel evant to gasi fi cati on products condensed i n the upper part of the gasi fi er. Biomass Air Gasification in a Downdraft Fixed Bed Energy & Fuels, Vol. 14, No. 4, 2000 897 ti ons are currentl y progressi ng to extend these studi es to other bi omass feedstocks. Conclusions A pi l ot pl ant for fi xed bed gasi fi cati on of bi omass i s bei ng operated at the ENEA Researcher Centre of Tri sai a (I tal y). A detai l ed survey of both i norgani c and organi c composi ti ons of the fl ue gas was carri ed out. The di fference observed under si mi l ar condi ti ons i n the product di stri buti on from the gasi fi cati on of the three di fferent bi omass feedstocks are i nterpreted i n l i ght of i nteracti ons between the organi c backbone and the i norgani c consti tuents of the tested materi al s. The fol l owi ng concl usi ons are drawn: 1. The composi ti on of the tested bi omass feedstocks shows si gni fi cant di fferences wi th regard to both the organi c and i norgani c consti tuents. The anal ysi s of the metal composi ti ons reveal s that, among the mai n con- sti tuents, the woody bi omass i s ri cher i n cal ci um than al mond shel l s whi l e poorer wi th regard to the potassi um concentrati on. I nteresti ng di fferences have al so been recorded i n the di stri buti on of the mi nor consti tuents. The anal ysi s of the organi c matri x composi ti ons i ndi - cates that woody bi omass has l ower percentages of both xyl an uni ts and l i gni n content than al mond shel l s. 2. There are at l east three i mportant rol es that mi ght be assi gned to cal ci um toward the gasi fi cati on process: (i ) to operate a certai n decomposi ti on of ammoni a, (i i ) to promote an i nternal cracki ng of the produced tar, and (i i i ) to catal yze char gasi fi cati on. 3. Di fferent l i gni n percentages i n the bi omass feed- stocks are refl ected i n di fferent phenol concentrati ons i n the fl ue gas and i n di fferent gasi fi cati on rates. 4. The hi gh percentage of xyl an uni ts from al mond shel l feedstocks wi th respect to woody ones are found to be associ ated wi th a hi gher aceti c aci d concentrati on i n the fl ue gas. Some other eval uati ons have been carri ed out by studyi ng the dependence of the products di stri buti on as a functi on of the gasi fi cati on rati o. I n thi s regard, oxygen was found to i ncrease the amount of ammoni a i n the fl ue gas whi l e decreasi ng that of cyani de. A reacti on pattern was proposed to expl ai n the observed trends. Acknowledgment. Thi s work was performed by the Tri sai a ENEA (I tal i an Agency for New Technol ogi es, Energy and Envi ronment) Research Centre wi thi n the conventi on between ENEA and LI ER (Li aoni ng I nsti - tute of Energy Resources, Chi na) for the gasi fi cati on of bi omass resi dues and energy producti on i n the Popul ar Republ i c of Chi na. We thank Eng. Ma Long Long for hi s parti ci pati on i n the performed pl ant runs. Fi nal l y we thank Dr. Coppol a Ti zi ana for hel pful di scussi ons about surface area measurements. EF990243G (24) Chen, S. G.; Yang, R. T. The acti ve surface speci es i n al kal i - catal yzed carbon gasi fi cati o: Phenol ate (C-O-M) groups vs cl usters (parti cl es). J . Catal. 1993, 141, 102-113. (25) Chen, S. G.; Yang, R. T. Mechani sm of al kal i and al kal i ne earth catal yzed gasi fi cati on of graphi te by CO2 and H2O studi ed by el ectron mi croscopy. J . Catal. 1992, 138, 12-23. Figure11. Rate of bi omass consumpti on duri ng a gasi fi cati on process. 898 Energy & Fuels, Vol. 14, No. 4, 2000 DeBari et al.