Production of lignocellulosic gasoline using fast pyrolysis of biomass and a conventional refining scheme

Introduction

Brazil is recognized globally for the use of renewable fuels since the 1970s, with the start of the Proalcool program with its strong commitment to energy independence. Proalcool enabled Brazil to use ethanol in its energy matrix and become one the first countries to eliminate lead from the gasoline. Currently more than 90 % of the new light vehicles sold in Brazil are flex fuel, which are designed to run either on E20–E25 gasoline or 100 % of hydrous ethanol fuel. More recently, in the diesel vehicles segment, Brazil established a mandate for using 5 % of biodiesel (B5), obtained from the transesterification of triglyceride oils, in its diesel fuel. The oil is produced from a variety of oilseed crops produced in Brazil, such as palm, soybeans, cotton, canola and sunflower.

Many types of non-food biomass are abundant including co-processing streams from the sugar cane industry. The sugar cane plant can be divided into three, almost equal parts: bagasse, sucrose and leaves (or straw). At present, only sucrose is fermented to produce ethanol. Nevertheless, bagasse, tops and leaves can also be transformed into energy or biofuels. In Brazil, the sugar cane bagasse cogeneration installed capacity reached 1830 MW in 2005. But the process of turning certain lignocellulosic materials into biofuels still presents many challenges and is much more expensive and difficult.

The use of lignocellulosic biomass in a refinery necessitates prior densification, since its density and energy content are much lower than those of petroleum. Fast pyrolysis of biomass is one of the most promising thermochemical routes to liquefy agricultural wastes such as sugar cane bagasse, wood chips and other field residues, by converting the solid raw biomass into a liquid named bio-oil, used today to produce specialty chemicals or to generate power. Liquefaction via pyrolysis of solid raw biomass is the key factor that makes it possible to use this renewable lignocellulosic resource in a refining scheme.

Petrobras has pioneered the processing of renewable raw materials in its refineries. Soybean oils were commercially used as a feedstock in one of Petrobras catalytic cracking units in the 1980s (REMAN refinery in the State of Amazonas, 1983) to increase gasoline production [1, 2]. The resulting gasoline had a higher octane number (a standard measure of the performance of an Otto cycle motor fuel) and much lower sulfur and nitrogen levels than that obtained from a fossil gasoil stream. The oxygen content of the product was negligible and the product was as stable as any other naphtha obtained from fluid catalytic cracking (FCC). This kind of operation was technologically undemanding, owing to the relative chemical simplicity of the soybean oil, but it was not economically attractive, since soybean oil prices, as well as those of other vegetable oils based on triglycerides, are extremely high relative to comparable feedstocks derived from fossil streams.

Bio-oils are a much lower cost alternative to triglyceride oils, since their feedstocks derive from agricultural residues, wood residues, and lignocellulosic materials, which are not used for human consumption. However, processing of bio-oils is considerably more complex than of triglyceride oils processing, since the latter are straightforward esters, whereas bio-oils constitute a complex blend of many classes of different oxygenated compounds and share some of the characteristics found in petroleum residues.

This paper discusses how some existing refining schemes can be modified to co-process bio-oil, transforming it into a lignocellulosic transportation fuel.

Fast pyrolysis of biomass

Fast pyrolysis is a thermal process that uses sand in a fluidized bed in the absence of air to transform biomass into a product named bio-oil, along with smaller amounts of gas and char. The char is generally burned to provide energy for the reaction by heating sand, which is recirculated to the (fluidized bed) reactor. The sole role of the solids is to transport heat to the biomass aiming at vaporizing it in a few seconds. Depending on the source of the raw biomass, operating conditions and type of reactor, it is possible to produce up to 75 % of bio-oil in weight basis (dry biomass feed). Some desirable conditions to carry out biomass pyrolysis aiming at bio-oil maximization were listed by Bridgwater [3]:

• High heat transfer rate from the sand to the biomass, aiming at its complete primary degradation.

• Moderate reaction temperature between 450 and 550 °C

• Low pyrolysis vapors residence time (< 2 s) and rapid condensation of pyrolysis vapors after the reactor to minimize secondary reactions.

Fast pyrolysis is already a commercial technology dedicated mostly to specialty applications such as food flavorings production. These commercial units work in a much smaller scale than regular refining units. The largest fast pyrolysis biomass units range between 100 and 200 tons per day, while small refineries deal with thousands of tons per day of fossil fuel streams. Bio-oils are generally unstable, viscous and acidic. However, today’s pyrolysis oil producers use filtration or hot filtration technologies to improve the stability of the bio-oil. The filtration removes char and some of the alkaline metals present in the bio-oil, which would promote secondary reactions during storage. Thus, bio-oil samples may be transported and stored for longer periods of time, even years. Still, bio-oil chemical characteristics require its further chemical conversion into a transportation fuel.

Properties of typical bio-oil

Table 1 shows the properties of liquid streams produced from different sources. Bio-oil properties are compared with those found in the soybean oil (triglyceride) and a typical vacuum gasoil (VGO). Triglyceride oils, such as soybean oils, are very different from bio-oils. Triglycerides present heating values comparable to those found in fossil fuels streams and do not contain water. Additionally, bio-oils are also much more acidic, viscous and contain higher amounts of oxygen.

Metals such as nickel and vanadium, which are sometimes present at high levels in fossil fuel streams, are not present in bio-oils. On the other hand, alkaline metals are found at high levels and may be critical for some of the processes used inside a conventional refinery. Further, bio-oil heating value is less than half of the one found in fossil fuel oil, due to the high levels of water and oxygen.

Although petroleum industry has a wide experience in characterizing petroleum streams and derivatives, its experience with oxygenated compounds is very limited. Petroleum acidity, for instance, is much lower than the one present in bio-oil, while the oxygenated compounds present in bio-oils are much diversified, including ketones, acids, aldehydes and phenolics [5, 6]. Oxygenated compounds are responsible for the bio-oil polar characteristics, making it immiscible with petroleum fractions.

Bio-oil has already received ASTM standard specification (ASTM D7544-09) so it that commercialization as a renewable fuel oil for use in industrial and commercial boilers can start. The industrial burners must be especially equipped to handle these types of fuels. The specification does not cover other uses as, for instance, residential heaters, small commercial boilers, engines, or marine applications that require additional development.

A refining scheme proposal

Although it may be feasible to stabilize bio-oil, bio-oil upgrading is still a challenge. One of the most studied routes uses catalysts (instead of sand) to improve bio-oil properties [7, 8]. The use of ZSM-5 zeolite [9], for instance, produces a much better quality bio-oil: less acidic, more stable and highly aromatic. Theoretically the treated bio-oil can then be more easily hydrogenated and transformed into a transportation fuel, especially high octane gasoline. It is worth noting that the catalyst must be cheap and robust enough to handle the high level of metal contaminants present in the raw biomass, namely sodium, calcium, potassium, magnesium and iron. Otherwise, the process will not be economically viable. Thus, this route poses the disadvantage of using an expensive catalyst, instead of sand and zeolitic catalyst deactivates with most of these metals.

Another route to upgrade bio-oils involves the catalytic cracking (FCC) process, which has been widely employed for decades in petroleum refineries, for the production of high octane gasoline. Catalytic cracking units consist of two separated reaction zones interconnected by catalyst stand-pipes. Catalyst circulates between the catalytic reactor named riser, where the catalytic cracking reactions takes place, and the combustion zone, where the coke formed during the reaction and deposited on the catalyst is burnt.

The refining scheme proposed is a two-step route: fast pyrolysis followed by catalytic upgrading. In the first step, a stable bio-oil is produced by fast pyrolysis and then it is catalytically converted into gasoline, liquefied petroleum gas (LPG), LCO (a diesel range cut) and fuel oil. The gasoline fraction is further hydrotreated to comply with Brazilian specification as shown in Fig. 1.

Fig. 1

Bio-oil co-processing in an existing refinery.

Bio-oil catalytic cracking

Catalytic cracking (FCC) hardware presents many similarities to the hardware used by the pyrolysis of biomass. As in the pyrolysis, FCC units have a reactor and a regenerator section interconnected by standpipes. Catalyst, instead of sand, circulates between these two sections. FCC can be used to process a large variety of refinery internal streams, including heavy distillation cuts such as atmospheric residues or vacuum residues.

The catalytic cracking reactions proceed in the riser reactor where the catalyst in the form of fine particles makes contact with feedstock. As the catalyst promotes cracking reactions throughout the riser reactor length, it is also deactivated by the coke formed as a by-product of the reactions along the riser. The catalytic cracking reactions, mostly endothermic, are carried out in only 2 s. After the reactor a series of cyclones separate the reaction products and the deactivated catalyst is rectified by injecting steam, which separates the volatile hydrocarbon products carried by the catalyst. The deactivated catalyst then goes to a regenerator, where the coke deposited on the surface of the catalyst is burnt, resulting in a regenerated catalyst which returns at a high temperature to the reactor, where it provides heat for the cracking reactions and initiates a new cycle of catalytic cracking reactions.

The typical FCC catalyst is constituted by faujasite zeolite (Y-zeolite), alumina, silica, and kaolin. Some ZSM-5 may also be used. The zeolite present in the catalyst is strongly and permanently deactivated by alkaline metals. In the FCC process, catalyst is continuously added and removed from the FCC unit in order to renew the catalyst inventory and maintain catalyst activity. Thus the smallest amount of alkaline metals present in the bio-oil will lead to the lowest FCC catalyst deactivation and the lowest catalyst addition rate to keep catalyst activity constant.

A biomass fast pyrolysis liquid (bio-oil) from pine woodchips produced by BTG (Netherlands) was used in the experiments [10]. Its main characteristics have been presented in Table 1. A commercial FCC vacuum gasoil (VGO) from Petrobras was used as reference. Its characteristics are found in Table 2. All tests were carried out in a catalytic cracking demonstration-scale unit in Six (São Mateus Sul, State of Paraná, Brazil). The bio-oil and the Brazilian commercial vacuum gasoil (Table 2) were co-processed using a regular Y-zeolite FCC equilibrium catalyst (Table 3) from a Petrobras refinery.

The demonstration-scale unit was used to carry out the co-processing experiments. It consists of a vertical reactor riser type reactor (50 mm ID, 18 m riser height), a fluid bed regenerator reactor, a stripper and a lift line. The gasoil feed and bio-oil are immiscible and were injected at two different heights into the riser reactor, at a combined feedrate of 150 kg/h. The tests were performed under conditions close to that used in commercial FCC units. Reactor temperature was kept at 540 °C in all experiments. Two different bio-oil/VGO weight ratios were used: 10/90 and 20/80.

The yields were calculated as the weight percent of feed. Gaseous products were analyzed by gas chromatography (hydrogen, C1-C5+). Liquid samples were collected and analyzed by simulated distillation (ASTM D2887). The cut-off points between naphtha (gasoline range), LCO (light cycle oil) and bottoms (decanted oil) were used as follow: for gasoline, C5 up to 220 °C; for LCO, 220 °C up to 343 °C; and decanted oil, 343 °C and higher. The water content in the bio-oil was determined by the method of Karl Fischer (ASTM D4928).

The coke yield was calculated through the analysis of the flue gas from the regenerator obtained by gas chromatography.

The oxygenated products (water, carbon monoxide and dioxide) increases in the gas composition when the amount of bio-oil processed is increased, due to decarbonylation and decarboxylation of oxygenated compounds present in the bio-oil. Likewise, the yield of water also increases as a consequence of dehydration reactions. In all tests, carbon monoxide yield is higher than carbon dioxide, suggesting that decarbonylation reactions prevail over decarboxylation.

Some of these tests results are shown in Table 4. Fuel gas, LPG, gasoline, LCO and bottoms are regular FCC products. Carbon monoxide and dioxide are found only in small quantities in the fuel gas produced by a conventional FCC as a result of the drag of the combustion gases from the regenerator to the riser reactor. Gasoline yields obtained from Brazilian vacuum gasoil are c.a. 40 % wt/wt. The gasoline yield obtained in co-processing is approximately the same when 10 % of bio-oil is added to the fossil feedstock, but decreases when 20 % bio-oil is introduced. Analyzing the coke yields obtained with pure gasoil and the different bio-oil levels, it is possible to observe that the coke yield is kept constant at 10 % bio-oil, but increases with 20 % bio-oil. Nevertheless, this increase is not as high as those found in literature [9] obtained at bench scale with the same or even lower bio-oil addition levels. The differences in coke yield obtained at different scales are possibly related with difficulties in the introduction and dispersion of the bio-oil at smaller scales. If the bio-oil is not adequately atomized into small droplets, it will be rapidly transformed into coke on the catalyst, blocking its active sites and deteriorating the FCC yields. It is worth noting that the coke formation is of extreme importance in the FCC process, since the ratio between the catalyst that circulates in the riser reactor and the hydrocarbons is controlled by the energy balance in the unit.

There was no detectable difference in the water content between liquid effluents produced from bio-oil and those produced from pure gasoil. All liquid effluents contain around 400 ppm of water. Likewise, elemental analysis did not indicate oxygen (obtained by difference) in the liquid effluents. Although oxygen could not be detected, a high level of phenolic compounds could be measured in the liquid product using an adapted UV technique from UOP (UOP 262–59). Phenolics content ranged from 12 000 up to 16 000 ppm in the total liquid effluent, depending on the operating conditions used.

Co-processing experiments require assessing the renewable carbon content in each product, since it is desirable to concentrate the renewable material in the most valuable streams, such as gasoline and LCO (diesel range). Nevertheless, the only reliable method to distinguish a fossil carbon from a renewable one is the ¹⁴C isotopic analysis (ASTM D6866-12), which can be applied to any carbon-containing product [11]. The ¹⁴C is a slightly radioactive carbon isotope naturally present in any living substance in a certain predictable trace level. On the other hand, fossil substances do not contain ¹⁴C, since it decays by half in approximately 5730 years, i.e., its half-time life is 5730 years. The measurement of ¹⁴C level is the basis of radiocarbon dating used to determine the age of organic materials up to about 60 000 years. ASTM D6866-12 method uses an Accelerator Mass Spectrometer combined with Isotope Ratio Mass Spectrometry to quantify the amount of biobased carbon. This method was used to determine the amount of renewable carbon content in the catalytic cracking products: gasoline range yield, LCO (diesel range) yield, bottoms and coke. Samples obtained from several liquid effluents and the catalyst, which contains the coke formed during the reaction, were analyzed to calculate directly the bio-oil contribution to the yields profile. Only the renewable carbon content in gaseous products, fuel gas and LPG, were not measured. Results indicated that the liquid products contain around 5 % of renewable carbon when 20 % of bio-oil is added to the feed.

Renewable naphtha hydrotreatment

Due to the high phenol levels in the cracker naphtha range when bio-oil is used as feed, it was extremely important to investigate if the oxygenate compounds present in the cracker naphtha would interfere on final sulfur removal in naphtha hydrogenating step, since Petrobras usually hydrodesulfurizes under moderate pressure the naphtha produced by using a sulfided cobalt-molybdenum (CoMoS) catalyst, aiming at reducing sulfur level down to 50 ppm.

Therefore, the cracker naphtha produced in the demonstration scale unit was hydrogenated under same commercial conditions and catalyst used in the Brazilian refineries. Naphtha properties obtained with 20 % of bio-oil were analyzed before and after the hydrogenation step aiming at verifying the impact of the hydrogenation on naphtha quality. The impact of the hydrogenation on octane numbers, Motor (ASTM D2700) and Research (ASTM D2699), is the same observed in naphthas obtained from fossil streams, i.e., there is some loss on octane numbers, but the product still complies with Brazilian gasoline specification. It was also possible to reach 50 ppm in sulfur, the future specification to be introduced in Brazil in 2014.

Although no problems were detected for sulfur removal, hydroxyl groups were not completely hydrogenated and a high level of phenolic compounds still remained in the naphtha. Future work will evaluate the effect of the presence of these phenolic compounds on naphtha stability, especially the type of phenolic compound present in the naphtha, since their effect on naphtha stability may depend on some specific characteristics, such as the number of hydroxyl groups present in the compound.

Conclusions and future works

The pyrolysis route to convert lignocellulosic wastes into biofuels via catalytic cracking proved to be feasible at demonstration scale and it was possible to successfully hydrotreat the cracker naphtha. But there are still some concerns to be addressed, especially those related to the presence of alkaline metals in the bio-oil, which may deactivate the catalysts used in some processes, including the catalytic cracking. It is also necessary to perform life cycle assessment and techno-economic analysis of the complete route, from the pyrolysis unit up to the production of the final biofuel.