- Email: [email protected]
Water-washing to reduce metals in oils extracted from Nannochloropsis algae for potential FCC feedstock
Fuel 155 (2015) 63–67
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Water-washing to reduce metals in oils extracted from Nannochloropsis algae for potential FCC feedstock Chengrong Wang a,⇑, J. Scott Buchanan b, Wayne R. Kliewer b, Kuangnan Qian b a b
ExxonMobil Research and Engineering, 600 Billingsport Road, Paulsboro, NJ 08066, United States ExxonMobil Research and Engineering, 1545 Route 22, Annandale, NJ 08801, United States
h i g h l i g h t s Organic oils can be extracted from algae using solvents such as ethanol. The oils metals levels are too high for inclusion in refinery cracker feedstock. Washing the extracted oil with water was ineffective at removing the metals. Washing the algae with water prior to solvent extraction greatly reduced metals. The resulting oil could be co-fed to a cracking unit at a 2% level in the feed.
a r t i c l e
i n f o
Article history: Received 2 May 2014 Received in revised form 26 March 2015 Accepted 30 March 2015 Available online 8 April 2015 Keywords: Algae Metals Refining Catalytic cracking
a b s t r a c t Algae cultivation offers the potential to produce biofuels with much higher productivity than conventional soil-based crops. Organic oils can be extracted from algae using solvents such as alcohols and paraffins. A challenge is to define the most economical means to process these extracted oils into products such as chemical intermediates and liquid fuels. One approach is to add algal oils directly into the petroleum-based feeds to existing refinery processes such as fluid catalytic cracking (FCC) or hydroprocessing. Maximum acceptable levels were defined for contaminants such as sodium, potassium, calcium, phosphorus, and nitrogen in the feed to an FCC unit, based on refinery experience. Cracking runs in a laboratory unit confirmed the large decrease in conversion for high levels of sodium and potassium. For a laboratory sample of Nannochloropsis algae, the level of metals was far too high in the whole algae or in oils extracted from the algae using ethanol or n-heptane to allow their inclusion in FCC feedstock, even at a level of 2% algal product in the FCC feed. Washing the extracted oil with water removed only modest amounts of metals. It was found that washing the algae with water first, and then doing solvent extractions, resulted in oils with metals levels low enough for co-feeding to an FCC unit at a level of 2%. The oil yields by mass from the starting algae from this route were 21% for ethanol solvent, and 10% for heptane solvent. If an FCC unit were completely dedicated to bio-feeds without the necessity of co-processing petroleum feedstocks, the unit might operate with a different catalyst and different set of operating constraints, but that is beyond the scope of this study. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Algae are fast-growing organisms with the potential to produce high volumes of biofuels [1–4]. Some algae can grow on non-productive land and utilize brackish water to minimize concerns of competing with agriculture for limited natural resources [1,2]. To obtain fuels that are useful in application such as internal combustion engines, some sort of processing must typically be ⇑ Corresponding author. Tel.: +1 856 224 2060; fax: +1 856 224 3633. E-mail address: [email protected] (C. Wang). http://dx.doi.org/10.1016/j.fuel.2015.03.074 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
applied to the bulk algae. Various approaches have been proposed to transform the whole algae into useful liquid or gaseous fuels and chemicals. These approaches include hydrothermal processing with high pressure and high temperature, and organic solvent extraction to recover the algae oil. For solvent extraction process, algae cell membrane disruption, either via physical or chemical means, is typically necessary to achieve high oil recovery. A typical initial processing step is to remove water (either via extraction or direct drying). The dried algae may then pass through a mechanical milling step to lyse the cell membrane. A solvent extraction may be carried out to recover the algae oil, which can
64
C. Wang et al. / Fuel 155 (2015) 63–67
be further processed via polar and non-polar solvent partition to separate the non-polar lipids (glycerides) from polar components. In some cases, the algae cell membranes can be ruptured during the solvent extraction and thus the mechanical lysing step may be omitted. Adding algal-derived materials to the feeds of existing refinery units offers a cost-effective means to insert bio-feeds into the existing system for producing fuels and chemicals. ‘‘Algal-derived materials’’ here could be whole algae (with or without rigorous dewatering), or some fraction or extract of the algae. Key conversion units in refineries include the fluidized catalytic cracking (FCC) and hydroprocessing (HDP) units, which utilize catalysts to promote desired reactions such as cracking and the removal of heteroatoms by hydrogen insertion. A number of reviews have appeared on the catalytic processing of bio-derived feedstocks [5–8]. Idem et al. [9] cracked canola oil over various acidic and basic catalysts in a fixed-bed reactor, and compared reaction yields to those from an empty reactor (thermal cracking). The choice of catalyst had little impact on the initial cracking of canola oil to long chain hydrocarbons and oxygenates, but did affect the secondary reactions. Secondary cracking was enhanced by the catalysts which were amorphous and non-basic (silica, alumina, and silica–alumina), resulting in more gas and light liquid products. ZSM-5 catalysts, with no aluminum and with Si/Al = 56, moderated the secondary cracking, producing high yields (26–27 wt%) of aromatics. Basic catalysts (calcium and magnesium oxides) somewhat dampened secondary cracking, giving lower gas yields. Corma et al. [10] cracked biomass-derived oxygenates (glycerol, sorbitol) in a fixed-bed MAT unit. At a typical FCC MAT test temperature of 500 °C, glycerol gave relatively high yields of coke (e.g. about 40% coke at 80% conversion) with alumina and with commercial (faujasite-based) FCC catalysts. The highest yields in gases were: CO > propene > CO2 > ethene. ZSM-5 catalyst gave lower coke and higher ethylene. Sorbitol gave similar yields as glycerol, except for higher CO. When glycerol was co-fed with a vacuum gasoil, both feed components were converted, with gas yields lower than expected from a simple additive effect. Dupain et al. [11] cracked rapeseed oil in a riser-type reactor under realistic FCC conditions. At their short contact time conditions, they observed only water as an oxygenate product, not CO or CO2. The triglycerides were largely converted to fatty acids within 50 ms through radical cracking reactions. The resulting unsaturated fatty acids from rapeseed rapidly aromatized, forming relatively stable heavy liquid products. When a fully-saturated stearic acid feed was used, the aromatics yield was lower, and there was more cracking down to gases and gasoline-range liquids. These prior publications deal mainly with idealized or clean oxygenate feeds. This seems appropriate when treating vegetable oils or animal fats, which consist mainly of triglycerides with low metals contents. Tran et al. [12] reviewed catalytic upgrading of oils from algae. The work discussed there typically involved short-term laboratory experiments. For longer-term runs, however, the oils extracted from algae can have levels of metals that are high enough to interfere with conventional refinery operations. A constraint on these units is the amount of metals and phosphorus that can be tolerated by the catalyst while retaining acceptable catalytic activity and selectivity and longevity. If catalyst change-out intervals or rates are kept within the industry norms, this translates to effective maximum desirable levels of these contaminants in the feed. In typical petroleum operations, the metals in the feeds that are of most concern are nickel and vanadium. These metals are commonly present in algae at levels much lower than the alkalis, so our focus here is on the basic metals rather than nickel or vanadium. Bi and He [13] characterized five green and
three brown microalgae for their potential application in biofuels. The brown algae, with high ash contents (up to 43.4 wt% on dry basis), were deemed less suitable for conversion to biofuels. Multiplying the ash contents by the mineral contents in the ashes, for the five green algae the average levels of sodium, potassium, and calcium in the dry algae were 1.4 wt%, 1.5 wt%, and 0.6 wt%, respectively. For instance for FCC, some typical maximum contaminant values are shown in Table 1. At these levels of metals on catalyst, significant performance deterioration is expected, so normally metals levels are kept below these maximum values. The assumption here is that the catalyst consists of the faujasite-based acidic composite which has proven over many decades to be the most effective material for cracking the primary vacuum gasoil FCC feed. If an FCC unit were completely dedicated to bio-feeds without the necessity of co-processing the petroleum stock, the unit might operate with a much different set of constraints, including a nonfaujasitic base catalyst, but that is beyond the scope of this study. Faujasite is a type of zeolite, with acidic sites formed on the walls of the micropores. Calcium and alkali metals can deactivate the catalyst by titrating these acid sites, and can also accelerate longterm structural breakdown of the zeolite in the FCC unit. The translation in Table 1 from levels on catalyst to levels in feed assumes a relatively high catalyst make-up rate of 0.32 lb catalyst/bbl feed, such that 1 ppm metals in feed gives 1000 ppm (0.1%) on catalyst. For a given maximum level of a contaminant in the overall feed to the FCC, one can calculate the maximum acceptable level of that contaminant in some component of that feed. For instance, if an algal component was present in at a level of 5% by weight in the feed, then in order to maintain the level of sodium below 8 ppm in the overall feed, the sodium level would have to be less than 0.016% in the algal component. If the algal concentration in the feed were reduced from 5% to 2%, then a higher level (0.04%) of sodium could be tolerated in the algal component. These values are also listed in Table 1. They provide a target for metals reduction in the algal component, if that component is intended to be included in the feed to the FCC. High levels of these contaminants are detrimental in hydroprocessing as well as in catalytic cracking. However, it can be more difficult to define specific levels of metals in hydroprocessing feeds that are acceptable or unacceptable than for FCC, since practices with guard bed change-out vary. Rapid change-out, especially with swing operation of two guard beds, can help protect the main catalyst bed from high levels of metals in the feed. To quantify the effects in FCC of the high alkali levels associated with algal components, experiments were done in a laboratory FCC using catalyst which was spiked with sodium and potassium. These experiments verified the degradation of cracking performance with high levels of alkali. Accordingly, research was carried out to determine effective means to remove metals from algal components to prepare them as potential FCC feed extenders.
Table 1 Typical maximum acceptable levels of contaminants by weight on FCC catalyst and in FCC feed; corresponding maximum levels in an algal component in FCC feed, at two concentrations of algal component within the FCC feed. Typ. max. on FCC cat. (wt%)
Typ. max. in FCC feed (ppm)
Max. in algal for 5% algal in FCC feed (wt%)
Max. in algal for 2% algal in FCC feed (wt%)
Na K Ca
0.8 1 0.8
8 10 8
0.016 0.02 0.016
0.04 0.05 0.04
P N
5 –
50 5000
0.1 10
0.25 25
C. Wang et al. / Fuel 155 (2015) 63–67
65
The focus was on liquid–liquid extraction. A challenge was to achieve high levels of metals removal, while retaining high yields of the desired extract. 2. Materials and methods 2.1. Catalyst cracking runs To assess the effects of high levels of basic metals on an FCC catalyst, cracking runs were made in a fixed-bed downflow reactor using 3 versions (treatments) of a commercial FCC catalyst containing rare earth treated ultrastable Y (RE-USY) faujasite. The reactor was a Kayser Technologies Model P ACE unit. The nominal reactor temperature was 524 °C and operating pressure was just above 1 atmosphere. Six grams of catalyst were loaded in the reactor. For each run, a set weight of vacuum gasoil feed was pumped through the reactor in a period of 60 s. Tests were made with catalyst/oil weight ratios of 3, 5, and 7. The catalyst was regenerated with air after each test. Gas was accumulated over a brine solution and analyzed by gas chromatograph. Liquid product were collected in a 0 °C chilled receiver, weighed, and analyzed using a gas chromatograph for simulated distillation. Gas chromatography was done using Agilent 6890 instruments, with non-polar dimethyl silicone capillary columns and flame ionization detectors. This type of micro-activity test (MAT) is widely used for assessing the relative performance of FCC catalysts [14]. The starting total Brunauer–Emmett–Teller (BET) surface area of the catalyst by nitrogen physisorption was 245 m2/g, and starting unit cell size by X-ray diffraction was 24.55 Å. The unit cell size in faujasite may be used to estimate the framework aluminum content [15]. For laboratory testing, FCC catalysts are typically subjected to a steam treatment, to mimic the deactivation that occurs in a commercial cracking unit. Version 1 of the catalyst was obtained by steaming this starting catalyst at 760 °C for 16 h. Version 2 was obtained by taking a portion of the Version 1 catalyst and impregnating it via incipient wetness with 1.5% Na and 1.5% K by weight. Version 3 was obtained by first impregnating with 1.5% Na and 1.5% K, and then steaming at 760 °C for 16 h. By adding the metals both before (Version 3) and after (Version 2) the steaming, these two treatments span the metals/steam interaction that might be expected in the refinery, where the metals accumulate on the catalyst while it is undergoing steam-deactivation as it circulates in the FCC unit. 2.2. Algae extraction and washing An organic solvent Soxhlet extraction process was developed in our study to recover algae oil. A typical extraction procedure is described below: About 60 g of the starting algae (powder or wet paste) was placed into a thimble (43 mm 123 mm). The thimble was then placed in a Soxhlet extractor. About 300 mL HPLC grade denatured ethanol (95%, 190 proof) was added into a round bottom flask and then heated to reflux. The refluxed ethanol was allowed to condense into the Soxhlet extractor to continuously extract the algae until no visible color shown in the extractor solution. The algae oil was obtained from the ethanol solution after removal of ethanol solvent via evaporation. For extracted algae oil water-wash operation, the algae oil was contacted with deionized water (1:10 oil:water ratio by weight) at room temperature. Preliminary water washing tests were made using a high-speed vortex mixer, where the oil/water mixture was placed in a vertically-oriented centrifuge tube and a rapid circular oscillating motion was applied. This vigorous mixing created an emulsion which was very difficult to break. Therefore, a gentler
Fig. 1. Extraction and wash treatments for samples.
mode of contacting was employed for water-washing of the algae oil. The oil/water mixture was placed in an Erlenmeyer flask and mixed at low speed on an orbital shaker table for about 30 min. The algae oil/water mixture was then centrifuged to separate the oil from the water. The washed algae oil was obtained after evaporation of residual water on a steam bath. For the mechanical lysing/water wash operation, the dry algae were first ball-milled in a cylindrical metal container with stainless steel ball through vigorous shaking. The lysed algae were then contacted with deionized water (1:10 ratio by weight) at room temperature for 30 min with stirring. The solution was filtered to obtain the washed algae solids, and the aqueous filtrate was evaporated on a steam bath to recover the water soluble residue. Fig. 1 shows the set of treatments for the samples generated in this study. Further details on these treatments are given in Tables 2 and 3. The starting Sample A was a dried whole Nannochloropsis microalgae. 3. Results and discussion 3.1. Catalytic cracking conversions Fig. 2 shows conversion versus catalyst/oil ratio for catalytic cracking runs. A normal range of conversions for commercially viable catalysts in this test would be around 65–80%. Conversions less than 50% in a FCC test like this indicate a commercially unacceptable catalyst. With the two alkali metals added at 1.5% each, the catalyst activity here is well below 50% for the two protocols of metal addition (before or after steaming). Adding the alkali before steaming gave lower average conversions that adding the alkali after steaming. These results provide confirmation that high levels of basic metals can greatly harm the performance of FCC catalysts. 3.2. Algae extractions and washing The Na, K, Ca, P, and N contents for Samples A–D are shown in Table 2. Sample A was the starting whole algae. The elemental values for Sample A may be compared with the maximum values in an FCC feed given in Table 1. The levels of Na, K, Ca, and P are much too high in the whole algae to allow it as a component in the feed to the FCC, even if this algae were only 2% of the FCC feed. Samples B–H represent various treatments for reducing the metals levels in the algal products. Sample ‘‘B’’ was derived from Sample A by Soxhlet extraction of a lipid fraction using ethanol, followed by evaporation of the ethanol. This lipid fraction amounted to 33% of the original dry weight of the algae. Its contaminant levels were significantly reduced compared to the parent algae, yet Na and K levels remained too high to include Sample B in
66
C. Wang et al. / Fuel 155 (2015) 63–67
Table 2 Elemental analyses of algae Samples A–D. Algal sample Description
A Original dry algae
Yield from precursor, wt% Yield from orig. algae, wt% Elemental, wt% Na K Ca P N
1.16 1.27 0.53 0.86 7.65
B EtOH Soxhlet extract of original algae A
C 1 water wash of Sample B (EtOH extracted lipids)
D 3 water wash of Sample B (EtOH extracted lipids)
33.0 33.0
61.0 20.1
38.3 12.6
0.854 0.485 0.006 0.188 3.10
0.899 0.308 0.014 0.261 2.33
0.396 0.155 0.006 0.173 2.54
Table 3 Further samples derived from Nannochloropsis algae. Algal sample Description
E Original algae A, milled, then water washed
F EtOH Soxhlet extract of Sample E (water washed/milled algae)
G C7 Soxhlet extract of Sample E (water washed/milled algae)
H C7 extract of milled, unwashed original algae A
Yield from precursor, wt% Yield from orig. algae, wt%
71.2 71.2
29.2 20.8
14.1 10.0
16.1 16.1
Elemental, wt% Na K Ca P N
0.172 0.228 0.634 0.450 8.52
0.026 0.014 0.001 0.008 2.10
0.040 0.001 0.019 0.019 0.64
Activity of Catalysts With and Without 1.5% Na/ 1.5% K 90 80
% Conversion
70 Steamed
60 50
Steamed, then Na/K Added
40
Na/K Added, then Steamed
30 20 10 0
0
2
4
6
8
Catalyst/Oil Ratio Fig. 2. Activity of catalysts with and without added sodium and potassium.
FCC feed at even a 2% level (see Table 1). For the Nannochloropsis strain used here, earlier testing showed that for ethanol extractions, pre-milling the algae made no significant difference in the ethanol extraction results, since the ethanol tended to swell and rupture the cells. Thus, no mechanical milling was done in the production of Sample B. A portion of Sample B was subjected to a water wash treatment. It was contacted with water for 30 min, followed by centrifugation, decanting, and drying at 373 K to yield sample C. The product recovery of C from B was 61 wt%. Thus, the yield of sample C from the original sample A was 20.1%, as shown in Table 2. This water wash did reduce the K levels, from 0.484% to 0.308%, but that is still
0.560 0.454 0.080 0.381 1.29
too high for adding to FCC feed in appreciable levels. Surprisingly, the Na, Ca and P levels were not decreased by this treatment. Within the accuracy of the measurements, it appeared that most of the Na, and nearly all the Ca and P, in Sample B were carried through into sample C, and concentrated there. Another portion of Sample B was given a more severe water washing, consisting of three rounds of contacting for 30 min, followed by centrifugation and decanting, with a final drying step, to yield Sample D. The yield from Sample B was only 38.2%, giving a 12.6% yield from the original algae. This more severe water washing of the Sample B lipids did reduce the levels of Na and K, but still not enough to add appreciable amounts to an FCC feed. Only modest reductions were seen in P and N levels. Even with the relatively gentle mixing action during the waterwashing step, some loss of algae oil to the water phase via the formation of a partial emulsion could not be excluded. Also, the algae oil was highly viscous, adhering to the glassware walls. Thus, the low recovery of algae oil seen in Sample D represents loss of bulk algae oil to emulsion and handling losses, in addition to the components removed by being dissolved in the water wash. In light of these results, experiments were undertaken to identify alternative methods for extracting products from algae with lower levels of contaminants. In particular, the concept was explored of pre-washing the algae with fresh water prior to extraction of a lipid fraction with solvents. Table 3 shows results. Sample E was obtained by mechanically milling the original algae A, then contacting with water for 30 min, followed by filtering and drying. The yield of Sample E (dried solids) from Sample A was 71.2%. This treatment gave a large reduction in the wt% content of Na and K in Sample E, with lesser or no reduction in Ca, P, and N. Nonetheless, by material balance some Ca, P, and N were removed by the water wash in going from sample A to E. Apparently the Na, K, Ca, P, and N that were removed in that water wash were the most mobile portions of these contaminants, since when Sample E was extracted with ethanol (with the usual subsequent evaporation
C. Wang et al. / Fuel 155 (2015) 63–67
of the ethanol solvent to recover the liquid organic product), the resulting Sample F had very low levels of all these contaminants (see Table 3). Sample F has acceptable levels of contaminants for inclusion in FCC feed at a 2% level per Table 1. Sample F could even be included in FCC feed at a 5% level, apart from Na which is too high by about a factor of two. Comparing Sample F (water wash first, then ethanol extraction) with Sample C (ethanol extraction, followed by water wash) we find that both treatments gave about the same yields (around 20%) of lipid from the original algae A, but that Sample F has much lower levels of contaminants. Benefits for water-washing the algae prior to solvent extraction were also found for the nonpolar solvent n-heptane. Extracting Sample E with n-heptane instead of ethanol gave Sample G, with a lipids yield of 14.1%, or 10.0% yield from the parent Sample A. Like Sample F, Sample G has contaminant levels low enough to add to an FCC feed at the 2% level, or at the 5% level apart from Na. Sample H was obtained by n-heptane extraction of a milled algae which had not been water-washed. The lipids yield for Sample H is somewhat higher than for Sample G, but the levels of contaminants, especially Na and K, are much higher. The divalent Ca appeared to be firmly bound to the solid portion of the algae. Little Ca was found in the ethanol-extracted lipid fraction (Sample B), and by comparison of Samples A and E, little Ca was removed via water-washing of the solid algae. On the other hand, about 85% of the monovalent Na and K were removed from the algae by washing with water (Sample E). Also, substantial amounts of Na and K were extracted by ethanol into the lipid phase: the levels of Na and K in the ethanol-extracted oil (Sample B) were 73% and 38%, respectively, of the Na and K levels in the starting algae. The low degree of alkali removal from the lipid phase by water washing is likely due to poor oil/water contacting, using extended but moderate stirring of the two phases. It appears that this contacting was not sufficient to allow the expected transfer of alkali between phases, due in part to the high viscosity of the algal oil. 4. Conclusions FCC runs in a laboratory unit demonstrated the enormous decrease in conversion with high levels of Na and K, confirming our estimates from refinery experience. For the sample of Nannochloropsis algae studied here, the levels of metals are far too high in the whole algae, or in oils extracted from the algae using ethanol or n-heptane, to allow their inclusion in FCC feedstock, even at a level of 2% in the feed. Washing the extracted oil
67
with water was relatively ineffective in removing the metals, probably because the oil was too viscous for good contacting with water under the moderate stirring conditions used. Very vigorous mixing gave a troublesome emulsion. It is possible that improved alkali removal might be obtained using other oil/water mechanical contacting regimes than investigated here. It was found that washing the algae with water first, and then doing solvent extraction with ethanol and with n-heptane, resulted in an oil with metals levels low enough for co-feeding to an FCC unit at a level of 2% in the feed, or even (apart from the sodium content) at a level of 5%. The oil yields from the starting algae from this route were about 21 wt% for the ethanol solvent, and 10% for the heptane solvent. References [1] Thomas WH, Seibert DLR, Alden M, Eldridge P, Neori A, Gaines S. Selection of high-yielding microalgae from desert saline environments. In: Aquatic species program review: proceedings of the march 1983 principal investigators’ meeting, Solar Energy Research Institute, Golden, Colorado. pp. 97–122 [SERI/ CP-231-1946]. [2] Thomas WH. Microalgae from desert saline waters as potential biomass producers. Prog Solar Energy 1983;6:143–5. [3] Sheehan J, Dunahay T, Benemann J, Roessler P. A look back at the U.S. department of energy’s aquatic species program—biodiesel from algae. U.S. Department of Energy’s Office of Fuels Development; 1998 [NREL/TP-58024190]. [4] Wijffels RH, Barbosa MJ. An outlook on microalgal biofuels. Science 2010;329:796–9. [5] Huber GW, Corma A. Synergies between bio- and oil-refineries for the production of fuels from biomass. Angew Chem Int Ed 2007;46:7184–201. [6] Smith B, Greenwell HC, Whiting A. Catalytic upgrading of tri-glycerides and fatty acids to transport biofuels. Energy Environ Sci 2009;2:262–71. [7] Lestari S, Mäki-Arvela P, Beltramini J, Lu GQM, Murzin DY. Transforming triglycerides and fatty acids to biofuels. ChemSusChem 2009;2:1109–19. [8] Ong YK, Bhatia S. The current status and perspectives of biofuel production via catalytic cracking of edible and non-edible oils. Energy 2010;35:111–9. [9] Idem RO, Katikaneni SPR, Bakhshi NN. Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape selectivity on product distribution. Fuel Process Technol 1997;51:101–25. [10] Corma A, Huber GW, Sauvanaud L, O’Connor P. Processing biomass-derived oxygenates in the oil refinery: catalytic cracking (FCC) reaction pathways and role of catalyst. J Catal 2007;247:307–27. [11] Dupain X, Costa DJ, Schaverien CJ, Makkee M, Moulijn JA. Cracking of a rapeseed vegetable oil under realistic FCC conditions. Appl Catal B: Environ 2007;72:44–61. [12] Tran NH, Bartlett JR, Kannangara GSK, Milev AS, Volk H, Wilson MA. Catalytic upgrading of biorefinery oil from micro-algae. Fuel 2010;89:265–74. [13] Bi Z, He BB. Catalytic upgrading of microalgae for the purpose of biofuel production. Trans ASABE 2013;56:1529–39. [14] Sedran UA. Laboratory testing of FCC catalysts and hydrogen transfer properties. Catal Rev-Sci 1994;36:405–31. [15] Sohn JR, Decanio SJ, Lunsford JH, O’Donnell DJ. Determination of framework aluminium content in dealuminated Y-type zeolites: a comparison based on unit cell size and wavenumber of i.r. bands. Zeolites 1986;6:225–7.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Water-washing to reduce metals in oils extracted from Nannochloropsis algae for potential FCC feedstock Chengrong Wang a,⇑, J. Scott Buchanan b, Wayne R. Kliewer b, Kuangnan Qian b a b
ExxonMobil Research and Engineering, 600 Billingsport Road, Paulsboro, NJ 08066, United States ExxonMobil Research and Engineering, 1545 Route 22, Annandale, NJ 08801, United States
h i g h l i g h t s Organic oils can be extracted from algae using solvents such as ethanol. The oils metals levels are too high for inclusion in refinery cracker feedstock. Washing the extracted oil with water was ineffective at removing the metals. Washing the algae with water prior to solvent extraction greatly reduced metals. The resulting oil could be co-fed to a cracking unit at a 2% level in the feed.
a r t i c l e
i n f o
Article history: Received 2 May 2014 Received in revised form 26 March 2015 Accepted 30 March 2015 Available online 8 April 2015 Keywords: Algae Metals Refining Catalytic cracking
a b s t r a c t Algae cultivation offers the potential to produce biofuels with much higher productivity than conventional soil-based crops. Organic oils can be extracted from algae using solvents such as alcohols and paraffins. A challenge is to define the most economical means to process these extracted oils into products such as chemical intermediates and liquid fuels. One approach is to add algal oils directly into the petroleum-based feeds to existing refinery processes such as fluid catalytic cracking (FCC) or hydroprocessing. Maximum acceptable levels were defined for contaminants such as sodium, potassium, calcium, phosphorus, and nitrogen in the feed to an FCC unit, based on refinery experience. Cracking runs in a laboratory unit confirmed the large decrease in conversion for high levels of sodium and potassium. For a laboratory sample of Nannochloropsis algae, the level of metals was far too high in the whole algae or in oils extracted from the algae using ethanol or n-heptane to allow their inclusion in FCC feedstock, even at a level of 2% algal product in the FCC feed. Washing the extracted oil with water removed only modest amounts of metals. It was found that washing the algae with water first, and then doing solvent extractions, resulted in oils with metals levels low enough for co-feeding to an FCC unit at a level of 2%. The oil yields by mass from the starting algae from this route were 21% for ethanol solvent, and 10% for heptane solvent. If an FCC unit were completely dedicated to bio-feeds without the necessity of co-processing petroleum feedstocks, the unit might operate with a different catalyst and different set of operating constraints, but that is beyond the scope of this study. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Algae are fast-growing organisms with the potential to produce high volumes of biofuels [1–4]. Some algae can grow on non-productive land and utilize brackish water to minimize concerns of competing with agriculture for limited natural resources [1,2]. To obtain fuels that are useful in application such as internal combustion engines, some sort of processing must typically be ⇑ Corresponding author. Tel.: +1 856 224 2060; fax: +1 856 224 3633. E-mail address: [email protected] (C. Wang). http://dx.doi.org/10.1016/j.fuel.2015.03.074 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
applied to the bulk algae. Various approaches have been proposed to transform the whole algae into useful liquid or gaseous fuels and chemicals. These approaches include hydrothermal processing with high pressure and high temperature, and organic solvent extraction to recover the algae oil. For solvent extraction process, algae cell membrane disruption, either via physical or chemical means, is typically necessary to achieve high oil recovery. A typical initial processing step is to remove water (either via extraction or direct drying). The dried algae may then pass through a mechanical milling step to lyse the cell membrane. A solvent extraction may be carried out to recover the algae oil, which can
64
C. Wang et al. / Fuel 155 (2015) 63–67
be further processed via polar and non-polar solvent partition to separate the non-polar lipids (glycerides) from polar components. In some cases, the algae cell membranes can be ruptured during the solvent extraction and thus the mechanical lysing step may be omitted. Adding algal-derived materials to the feeds of existing refinery units offers a cost-effective means to insert bio-feeds into the existing system for producing fuels and chemicals. ‘‘Algal-derived materials’’ here could be whole algae (with or without rigorous dewatering), or some fraction or extract of the algae. Key conversion units in refineries include the fluidized catalytic cracking (FCC) and hydroprocessing (HDP) units, which utilize catalysts to promote desired reactions such as cracking and the removal of heteroatoms by hydrogen insertion. A number of reviews have appeared on the catalytic processing of bio-derived feedstocks [5–8]. Idem et al. [9] cracked canola oil over various acidic and basic catalysts in a fixed-bed reactor, and compared reaction yields to those from an empty reactor (thermal cracking). The choice of catalyst had little impact on the initial cracking of canola oil to long chain hydrocarbons and oxygenates, but did affect the secondary reactions. Secondary cracking was enhanced by the catalysts which were amorphous and non-basic (silica, alumina, and silica–alumina), resulting in more gas and light liquid products. ZSM-5 catalysts, with no aluminum and with Si/Al = 56, moderated the secondary cracking, producing high yields (26–27 wt%) of aromatics. Basic catalysts (calcium and magnesium oxides) somewhat dampened secondary cracking, giving lower gas yields. Corma et al. [10] cracked biomass-derived oxygenates (glycerol, sorbitol) in a fixed-bed MAT unit. At a typical FCC MAT test temperature of 500 °C, glycerol gave relatively high yields of coke (e.g. about 40% coke at 80% conversion) with alumina and with commercial (faujasite-based) FCC catalysts. The highest yields in gases were: CO > propene > CO2 > ethene. ZSM-5 catalyst gave lower coke and higher ethylene. Sorbitol gave similar yields as glycerol, except for higher CO. When glycerol was co-fed with a vacuum gasoil, both feed components were converted, with gas yields lower than expected from a simple additive effect. Dupain et al. [11] cracked rapeseed oil in a riser-type reactor under realistic FCC conditions. At their short contact time conditions, they observed only water as an oxygenate product, not CO or CO2. The triglycerides were largely converted to fatty acids within 50 ms through radical cracking reactions. The resulting unsaturated fatty acids from rapeseed rapidly aromatized, forming relatively stable heavy liquid products. When a fully-saturated stearic acid feed was used, the aromatics yield was lower, and there was more cracking down to gases and gasoline-range liquids. These prior publications deal mainly with idealized or clean oxygenate feeds. This seems appropriate when treating vegetable oils or animal fats, which consist mainly of triglycerides with low metals contents. Tran et al. [12] reviewed catalytic upgrading of oils from algae. The work discussed there typically involved short-term laboratory experiments. For longer-term runs, however, the oils extracted from algae can have levels of metals that are high enough to interfere with conventional refinery operations. A constraint on these units is the amount of metals and phosphorus that can be tolerated by the catalyst while retaining acceptable catalytic activity and selectivity and longevity. If catalyst change-out intervals or rates are kept within the industry norms, this translates to effective maximum desirable levels of these contaminants in the feed. In typical petroleum operations, the metals in the feeds that are of most concern are nickel and vanadium. These metals are commonly present in algae at levels much lower than the alkalis, so our focus here is on the basic metals rather than nickel or vanadium. Bi and He [13] characterized five green and
three brown microalgae for their potential application in biofuels. The brown algae, with high ash contents (up to 43.4 wt% on dry basis), were deemed less suitable for conversion to biofuels. Multiplying the ash contents by the mineral contents in the ashes, for the five green algae the average levels of sodium, potassium, and calcium in the dry algae were 1.4 wt%, 1.5 wt%, and 0.6 wt%, respectively. For instance for FCC, some typical maximum contaminant values are shown in Table 1. At these levels of metals on catalyst, significant performance deterioration is expected, so normally metals levels are kept below these maximum values. The assumption here is that the catalyst consists of the faujasite-based acidic composite which has proven over many decades to be the most effective material for cracking the primary vacuum gasoil FCC feed. If an FCC unit were completely dedicated to bio-feeds without the necessity of co-processing the petroleum stock, the unit might operate with a much different set of constraints, including a nonfaujasitic base catalyst, but that is beyond the scope of this study. Faujasite is a type of zeolite, with acidic sites formed on the walls of the micropores. Calcium and alkali metals can deactivate the catalyst by titrating these acid sites, and can also accelerate longterm structural breakdown of the zeolite in the FCC unit. The translation in Table 1 from levels on catalyst to levels in feed assumes a relatively high catalyst make-up rate of 0.32 lb catalyst/bbl feed, such that 1 ppm metals in feed gives 1000 ppm (0.1%) on catalyst. For a given maximum level of a contaminant in the overall feed to the FCC, one can calculate the maximum acceptable level of that contaminant in some component of that feed. For instance, if an algal component was present in at a level of 5% by weight in the feed, then in order to maintain the level of sodium below 8 ppm in the overall feed, the sodium level would have to be less than 0.016% in the algal component. If the algal concentration in the feed were reduced from 5% to 2%, then a higher level (0.04%) of sodium could be tolerated in the algal component. These values are also listed in Table 1. They provide a target for metals reduction in the algal component, if that component is intended to be included in the feed to the FCC. High levels of these contaminants are detrimental in hydroprocessing as well as in catalytic cracking. However, it can be more difficult to define specific levels of metals in hydroprocessing feeds that are acceptable or unacceptable than for FCC, since practices with guard bed change-out vary. Rapid change-out, especially with swing operation of two guard beds, can help protect the main catalyst bed from high levels of metals in the feed. To quantify the effects in FCC of the high alkali levels associated with algal components, experiments were done in a laboratory FCC using catalyst which was spiked with sodium and potassium. These experiments verified the degradation of cracking performance with high levels of alkali. Accordingly, research was carried out to determine effective means to remove metals from algal components to prepare them as potential FCC feed extenders.
Table 1 Typical maximum acceptable levels of contaminants by weight on FCC catalyst and in FCC feed; corresponding maximum levels in an algal component in FCC feed, at two concentrations of algal component within the FCC feed. Typ. max. on FCC cat. (wt%)
Typ. max. in FCC feed (ppm)
Max. in algal for 5% algal in FCC feed (wt%)
Max. in algal for 2% algal in FCC feed (wt%)
Na K Ca
0.8 1 0.8
8 10 8
0.016 0.02 0.016
0.04 0.05 0.04
P N
5 –
50 5000
0.1 10
0.25 25
C. Wang et al. / Fuel 155 (2015) 63–67
65
The focus was on liquid–liquid extraction. A challenge was to achieve high levels of metals removal, while retaining high yields of the desired extract. 2. Materials and methods 2.1. Catalyst cracking runs To assess the effects of high levels of basic metals on an FCC catalyst, cracking runs were made in a fixed-bed downflow reactor using 3 versions (treatments) of a commercial FCC catalyst containing rare earth treated ultrastable Y (RE-USY) faujasite. The reactor was a Kayser Technologies Model P ACE unit. The nominal reactor temperature was 524 °C and operating pressure was just above 1 atmosphere. Six grams of catalyst were loaded in the reactor. For each run, a set weight of vacuum gasoil feed was pumped through the reactor in a period of 60 s. Tests were made with catalyst/oil weight ratios of 3, 5, and 7. The catalyst was regenerated with air after each test. Gas was accumulated over a brine solution and analyzed by gas chromatograph. Liquid product were collected in a 0 °C chilled receiver, weighed, and analyzed using a gas chromatograph for simulated distillation. Gas chromatography was done using Agilent 6890 instruments, with non-polar dimethyl silicone capillary columns and flame ionization detectors. This type of micro-activity test (MAT) is widely used for assessing the relative performance of FCC catalysts [14]. The starting total Brunauer–Emmett–Teller (BET) surface area of the catalyst by nitrogen physisorption was 245 m2/g, and starting unit cell size by X-ray diffraction was 24.55 Å. The unit cell size in faujasite may be used to estimate the framework aluminum content [15]. For laboratory testing, FCC catalysts are typically subjected to a steam treatment, to mimic the deactivation that occurs in a commercial cracking unit. Version 1 of the catalyst was obtained by steaming this starting catalyst at 760 °C for 16 h. Version 2 was obtained by taking a portion of the Version 1 catalyst and impregnating it via incipient wetness with 1.5% Na and 1.5% K by weight. Version 3 was obtained by first impregnating with 1.5% Na and 1.5% K, and then steaming at 760 °C for 16 h. By adding the metals both before (Version 3) and after (Version 2) the steaming, these two treatments span the metals/steam interaction that might be expected in the refinery, where the metals accumulate on the catalyst while it is undergoing steam-deactivation as it circulates in the FCC unit. 2.2. Algae extraction and washing An organic solvent Soxhlet extraction process was developed in our study to recover algae oil. A typical extraction procedure is described below: About 60 g of the starting algae (powder or wet paste) was placed into a thimble (43 mm 123 mm). The thimble was then placed in a Soxhlet extractor. About 300 mL HPLC grade denatured ethanol (95%, 190 proof) was added into a round bottom flask and then heated to reflux. The refluxed ethanol was allowed to condense into the Soxhlet extractor to continuously extract the algae until no visible color shown in the extractor solution. The algae oil was obtained from the ethanol solution after removal of ethanol solvent via evaporation. For extracted algae oil water-wash operation, the algae oil was contacted with deionized water (1:10 oil:water ratio by weight) at room temperature. Preliminary water washing tests were made using a high-speed vortex mixer, where the oil/water mixture was placed in a vertically-oriented centrifuge tube and a rapid circular oscillating motion was applied. This vigorous mixing created an emulsion which was very difficult to break. Therefore, a gentler
Fig. 1. Extraction and wash treatments for samples.
mode of contacting was employed for water-washing of the algae oil. The oil/water mixture was placed in an Erlenmeyer flask and mixed at low speed on an orbital shaker table for about 30 min. The algae oil/water mixture was then centrifuged to separate the oil from the water. The washed algae oil was obtained after evaporation of residual water on a steam bath. For the mechanical lysing/water wash operation, the dry algae were first ball-milled in a cylindrical metal container with stainless steel ball through vigorous shaking. The lysed algae were then contacted with deionized water (1:10 ratio by weight) at room temperature for 30 min with stirring. The solution was filtered to obtain the washed algae solids, and the aqueous filtrate was evaporated on a steam bath to recover the water soluble residue. Fig. 1 shows the set of treatments for the samples generated in this study. Further details on these treatments are given in Tables 2 and 3. The starting Sample A was a dried whole Nannochloropsis microalgae. 3. Results and discussion 3.1. Catalytic cracking conversions Fig. 2 shows conversion versus catalyst/oil ratio for catalytic cracking runs. A normal range of conversions for commercially viable catalysts in this test would be around 65–80%. Conversions less than 50% in a FCC test like this indicate a commercially unacceptable catalyst. With the two alkali metals added at 1.5% each, the catalyst activity here is well below 50% for the two protocols of metal addition (before or after steaming). Adding the alkali before steaming gave lower average conversions that adding the alkali after steaming. These results provide confirmation that high levels of basic metals can greatly harm the performance of FCC catalysts. 3.2. Algae extractions and washing The Na, K, Ca, P, and N contents for Samples A–D are shown in Table 2. Sample A was the starting whole algae. The elemental values for Sample A may be compared with the maximum values in an FCC feed given in Table 1. The levels of Na, K, Ca, and P are much too high in the whole algae to allow it as a component in the feed to the FCC, even if this algae were only 2% of the FCC feed. Samples B–H represent various treatments for reducing the metals levels in the algal products. Sample ‘‘B’’ was derived from Sample A by Soxhlet extraction of a lipid fraction using ethanol, followed by evaporation of the ethanol. This lipid fraction amounted to 33% of the original dry weight of the algae. Its contaminant levels were significantly reduced compared to the parent algae, yet Na and K levels remained too high to include Sample B in
66
C. Wang et al. / Fuel 155 (2015) 63–67
Table 2 Elemental analyses of algae Samples A–D. Algal sample Description
A Original dry algae
Yield from precursor, wt% Yield from orig. algae, wt% Elemental, wt% Na K Ca P N
1.16 1.27 0.53 0.86 7.65
B EtOH Soxhlet extract of original algae A
C 1 water wash of Sample B (EtOH extracted lipids)
D 3 water wash of Sample B (EtOH extracted lipids)
33.0 33.0
61.0 20.1
38.3 12.6
0.854 0.485 0.006 0.188 3.10
0.899 0.308 0.014 0.261 2.33
0.396 0.155 0.006 0.173 2.54
Table 3 Further samples derived from Nannochloropsis algae. Algal sample Description
E Original algae A, milled, then water washed
F EtOH Soxhlet extract of Sample E (water washed/milled algae)
G C7 Soxhlet extract of Sample E (water washed/milled algae)
H C7 extract of milled, unwashed original algae A
Yield from precursor, wt% Yield from orig. algae, wt%
71.2 71.2
29.2 20.8
14.1 10.0
16.1 16.1
Elemental, wt% Na K Ca P N
0.172 0.228 0.634 0.450 8.52
0.026 0.014 0.001 0.008 2.10
0.040 0.001 0.019 0.019 0.64
Activity of Catalysts With and Without 1.5% Na/ 1.5% K 90 80
% Conversion
70 Steamed
60 50
Steamed, then Na/K Added
40
Na/K Added, then Steamed
30 20 10 0
0
2
4
6
8
Catalyst/Oil Ratio Fig. 2. Activity of catalysts with and without added sodium and potassium.
FCC feed at even a 2% level (see Table 1). For the Nannochloropsis strain used here, earlier testing showed that for ethanol extractions, pre-milling the algae made no significant difference in the ethanol extraction results, since the ethanol tended to swell and rupture the cells. Thus, no mechanical milling was done in the production of Sample B. A portion of Sample B was subjected to a water wash treatment. It was contacted with water for 30 min, followed by centrifugation, decanting, and drying at 373 K to yield sample C. The product recovery of C from B was 61 wt%. Thus, the yield of sample C from the original sample A was 20.1%, as shown in Table 2. This water wash did reduce the K levels, from 0.484% to 0.308%, but that is still
0.560 0.454 0.080 0.381 1.29
too high for adding to FCC feed in appreciable levels. Surprisingly, the Na, Ca and P levels were not decreased by this treatment. Within the accuracy of the measurements, it appeared that most of the Na, and nearly all the Ca and P, in Sample B were carried through into sample C, and concentrated there. Another portion of Sample B was given a more severe water washing, consisting of three rounds of contacting for 30 min, followed by centrifugation and decanting, with a final drying step, to yield Sample D. The yield from Sample B was only 38.2%, giving a 12.6% yield from the original algae. This more severe water washing of the Sample B lipids did reduce the levels of Na and K, but still not enough to add appreciable amounts to an FCC feed. Only modest reductions were seen in P and N levels. Even with the relatively gentle mixing action during the waterwashing step, some loss of algae oil to the water phase via the formation of a partial emulsion could not be excluded. Also, the algae oil was highly viscous, adhering to the glassware walls. Thus, the low recovery of algae oil seen in Sample D represents loss of bulk algae oil to emulsion and handling losses, in addition to the components removed by being dissolved in the water wash. In light of these results, experiments were undertaken to identify alternative methods for extracting products from algae with lower levels of contaminants. In particular, the concept was explored of pre-washing the algae with fresh water prior to extraction of a lipid fraction with solvents. Table 3 shows results. Sample E was obtained by mechanically milling the original algae A, then contacting with water for 30 min, followed by filtering and drying. The yield of Sample E (dried solids) from Sample A was 71.2%. This treatment gave a large reduction in the wt% content of Na and K in Sample E, with lesser or no reduction in Ca, P, and N. Nonetheless, by material balance some Ca, P, and N were removed by the water wash in going from sample A to E. Apparently the Na, K, Ca, P, and N that were removed in that water wash were the most mobile portions of these contaminants, since when Sample E was extracted with ethanol (with the usual subsequent evaporation
C. Wang et al. / Fuel 155 (2015) 63–67
of the ethanol solvent to recover the liquid organic product), the resulting Sample F had very low levels of all these contaminants (see Table 3). Sample F has acceptable levels of contaminants for inclusion in FCC feed at a 2% level per Table 1. Sample F could even be included in FCC feed at a 5% level, apart from Na which is too high by about a factor of two. Comparing Sample F (water wash first, then ethanol extraction) with Sample C (ethanol extraction, followed by water wash) we find that both treatments gave about the same yields (around 20%) of lipid from the original algae A, but that Sample F has much lower levels of contaminants. Benefits for water-washing the algae prior to solvent extraction were also found for the nonpolar solvent n-heptane. Extracting Sample E with n-heptane instead of ethanol gave Sample G, with a lipids yield of 14.1%, or 10.0% yield from the parent Sample A. Like Sample F, Sample G has contaminant levels low enough to add to an FCC feed at the 2% level, or at the 5% level apart from Na. Sample H was obtained by n-heptane extraction of a milled algae which had not been water-washed. The lipids yield for Sample H is somewhat higher than for Sample G, but the levels of contaminants, especially Na and K, are much higher. The divalent Ca appeared to be firmly bound to the solid portion of the algae. Little Ca was found in the ethanol-extracted lipid fraction (Sample B), and by comparison of Samples A and E, little Ca was removed via water-washing of the solid algae. On the other hand, about 85% of the monovalent Na and K were removed from the algae by washing with water (Sample E). Also, substantial amounts of Na and K were extracted by ethanol into the lipid phase: the levels of Na and K in the ethanol-extracted oil (Sample B) were 73% and 38%, respectively, of the Na and K levels in the starting algae. The low degree of alkali removal from the lipid phase by water washing is likely due to poor oil/water contacting, using extended but moderate stirring of the two phases. It appears that this contacting was not sufficient to allow the expected transfer of alkali between phases, due in part to the high viscosity of the algal oil. 4. Conclusions FCC runs in a laboratory unit demonstrated the enormous decrease in conversion with high levels of Na and K, confirming our estimates from refinery experience. For the sample of Nannochloropsis algae studied here, the levels of metals are far too high in the whole algae, or in oils extracted from the algae using ethanol or n-heptane, to allow their inclusion in FCC feedstock, even at a level of 2% in the feed. Washing the extracted oil
67
with water was relatively ineffective in removing the metals, probably because the oil was too viscous for good contacting with water under the moderate stirring conditions used. Very vigorous mixing gave a troublesome emulsion. It is possible that improved alkali removal might be obtained using other oil/water mechanical contacting regimes than investigated here. It was found that washing the algae with water first, and then doing solvent extraction with ethanol and with n-heptane, resulted in an oil with metals levels low enough for co-feeding to an FCC unit at a level of 2% in the feed, or even (apart from the sodium content) at a level of 5%. The oil yields from the starting algae from this route were about 21 wt% for the ethanol solvent, and 10% for the heptane solvent. References [1] Thomas WH, Seibert DLR, Alden M, Eldridge P, Neori A, Gaines S. Selection of high-yielding microalgae from desert saline environments. In: Aquatic species program review: proceedings of the march 1983 principal investigators’ meeting, Solar Energy Research Institute, Golden, Colorado. pp. 97–122 [SERI/ CP-231-1946]. [2] Thomas WH. Microalgae from desert saline waters as potential biomass producers. Prog Solar Energy 1983;6:143–5. [3] Sheehan J, Dunahay T, Benemann J, Roessler P. A look back at the U.S. department of energy’s aquatic species program—biodiesel from algae. U.S. Department of Energy’s Office of Fuels Development; 1998 [NREL/TP-58024190]. [4] Wijffels RH, Barbosa MJ. An outlook on microalgal biofuels. Science 2010;329:796–9. [5] Huber GW, Corma A. Synergies between bio- and oil-refineries for the production of fuels from biomass. Angew Chem Int Ed 2007;46:7184–201. [6] Smith B, Greenwell HC, Whiting A. Catalytic upgrading of tri-glycerides and fatty acids to transport biofuels. Energy Environ Sci 2009;2:262–71. [7] Lestari S, Mäki-Arvela P, Beltramini J, Lu GQM, Murzin DY. Transforming triglycerides and fatty acids to biofuels. ChemSusChem 2009;2:1109–19. [8] Ong YK, Bhatia S. The current status and perspectives of biofuel production via catalytic cracking of edible and non-edible oils. Energy 2010;35:111–9. [9] Idem RO, Katikaneni SPR, Bakhshi NN. Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape selectivity on product distribution. Fuel Process Technol 1997;51:101–25. [10] Corma A, Huber GW, Sauvanaud L, O’Connor P. Processing biomass-derived oxygenates in the oil refinery: catalytic cracking (FCC) reaction pathways and role of catalyst. J Catal 2007;247:307–27. [11] Dupain X, Costa DJ, Schaverien CJ, Makkee M, Moulijn JA. Cracking of a rapeseed vegetable oil under realistic FCC conditions. Appl Catal B: Environ 2007;72:44–61. [12] Tran NH, Bartlett JR, Kannangara GSK, Milev AS, Volk H, Wilson MA. Catalytic upgrading of biorefinery oil from micro-algae. Fuel 2010;89:265–74. [13] Bi Z, He BB. Catalytic upgrading of microalgae for the purpose of biofuel production. Trans ASABE 2013;56:1529–39. [14] Sedran UA. Laboratory testing of FCC catalysts and hydrogen transfer properties. Catal Rev-Sci 1994;36:405–31. [15] Sohn JR, Decanio SJ, Lunsford JH, O’Donnell DJ. Determination of framework aluminium content in dealuminated Y-type zeolites: a comparison based on unit cell size and wavenumber of i.r. bands. Zeolites 1986;6:225–7.