Neural network models basically rely on a large number of experimental data [108, 109]. This model connects the input and output of a process unit with less knowledge of the system phenomena compared to the equilibrium and kinetic approaches. Guo et al. developed a neural network model for biomass gasification in a fluidized bed and emphasized its success and applicability for this process.

Symbols and Nomenclature

A = Cross-sectional area of the gasifier (m2)

C = Molar concentration (mol/m3)

Cp = Specific heat

D = Diffusion coefficient (m[1] [2]/s)

F = Molar flow rate (mol/s)

Qgasification, Qconv, Qrad, and Qmass = Energy transfer due to gasification, convection, radiation (kW/m[3] [4] [5] [6] [7] [8] [9] [10] [11] of bed)

R = Reaction rate (kg/m3 s)

X = Char conversion u = Gas velocity (m/s)

W = Solid flow rate (kg/s)

Subscripts

as = Ascending phase b = Bubble be = Bubble-emulsion c = Char

ds = Descending phase f = Freeboard g = Gas phase

i = Gaseous components in the product gas k = Reaction number w = Wake

The sustainability of biorefineries is very sensitive economically and their main objective is to generate the pyrolysis products that maximize the plant profitability.

11.6.1 Pyrolysis Products for Biorefineries

Bio-oil is considered as the most commercially valuable pyrolysis product if it contains specialty chemicals in high percentages. Therefore, maximizing bio-oil production can be an objective during process development. Table 11.3 summarizes the main specialty chemicals present in bio-oil obtained from the pyrolysis of selected feedstocks along with reactor type and operating conditions.

Fatty acids, phenolic compounds and aldehydes/ketones are among the high-value chemicals used for crops. On the other hand, woody biomass generates high amounts of levoglucosan, which is a promising biopolymer building block. Also, catechol and its derivatives have a high market value. In addition, hard wood has different pyrolysis behaviour than soft wood through their bio-oil composition (Table 11.3: pine wood vs oak wood).

Chemical species containing double bonds and oxygen can be detrimental to the stability of the bio-oil. These chemicals polymerize with time and increase the bio-oil viscosity as well as modify other properties. Acetic acid is one common pyrolysis product which can cause this phenomenon. Acetic acid is derived from cellulosic biomass and process optimization will generally focus on reducing its production. Moisture fraction has been shown to influence acetic acid production [64]. Another method consists of using metal oxides catalysts [47].

Lignin Content

Lignin content is an important recalcitrant factor to enzymatic hydrolysis of ligno — cellulosic biomass, evidenced by the obvious enhancement of enzymatic hydrolysis yield and rate after removal of lignin [25-27]. The negative effect of lignin in the biomass against enzymatic-catalyzing reaction is caused by the enzyme adsorptive loss to lignin, enzyme deactivation by lignin [28, 29], and steric hindrance between the enzyme and the substrate [30, 31]. To reduce or remove the lignin content from

the lignocellulosic biomass, organic solvents were employed for deligniflcation due to their good solubility for lignin. In the previous studies, organosolv pretreatment could achieve higher deligniflcation yields on most kinds of biomasses than any other methods. The enzymatic digestibility had been consequently more or less en­hanced. The relationship between the lignin content in pretreated materials and their enzymatic digestibility was not found to be always proportional [22]. In some cases, significant reduction of lignin (>95 %) can negatively affect cellulose digestibility. In the absence of lignin, decreased enzymatic accessibility was caused by aggregat­ing cellulose microfibrils [32]. Therefore, the process should be optimized in terms of the balance of enzymatic hydrolysis enhancement and deligniflcation.

Hemicellulose Compositions

Hemicellulose is a branched matrix polysaccharide in compound middle lamella to­gether with cellulose and lignin [33]. It forms enzyme-impenetrable cross-links that are the barrier for enzyme-catalyzed deconstruction of cellulose [34]. Because the hemicellulose is amorphous, it is more easily deconstructed by the hydrothermal effect during organosolv pretreatment. Then, the hydrolyzed sugars can be ex­tracted in the aqueous organic solvent. The removal of hemicellulose by organosolv pretreatment could reduce the steric hindrance of enzymes.

Acetyl Group

The enzyme-impenetrable crosslinks were created by covalent bonds between the side chains of branched hemicellulose (e. g., acetyl, uronic acids, and arabinose) and lignin or unbranched hemicellulose that attached to cellulose [34]. Several studies showed that the removal of acetyl groups from side chains of hemicellulose greatly enhanced the cellulose and xylan digestibility [35, 36]. Removal of acetyl groups could cause splitting of the cross-linking structure and thereby accelerate the enzyme recognition to the substrate.

15.4.2.1 Nano-Silica

In recent years, silica-supported acid catalysts are gaining considerable attention because of their high activities due to large surface area, and high selectivities. Silica-supported acid catalysts have many advantages over liquid catalysts such as high mechanical and thermal stabilities, easy handling, low toxicity, easy separation, and reusability of the catalysts, which make them promising for both academic study and industrial application [102].

In a typical preparation process, first, silica was calcined at high temperature (e. g., 400 °C) and used as support. Silica-supported acid catalyst was then prepared by impregnating the silica support with calculated amount of an aqueous solution of acid functional groups to achieve the required loading. Ghanbaripour et al. [103] used nano-silica-supported H3PW12O40 to catalyze the self-condensation of acetophe­nones. The products (1,3,5-triarylbenzenes) were obtained with high yield (80-93 %) in very short reaction time (18-30 min). Moreover, the catalyst can be used at least five consecutive times without significant loss in product yield.

Although nano-silica-supported acid catalysts have enormous potential applica­tion in cellulose hydrolysis, their improvement in the catalytic stability is needed. In most cases, nano-solid acid catalysts were prepared under mild conditions. As a result, surface passivation occurred, leading to the high density of surface active sites and high specific surface area [104]. However, large specific surface area and surface tension result in an absorption/aggregation process and affect the mass transfer. The combination of ILs and nano-solid acids can be used to resolve these issues, be­cause ILs can help to dissolve cellulose and hemicellulose completely and facilitate catalysts to approach the в-1,4 glycosidic bonds in cellulose.

Unfortunately, some Si-O-Si links were easily ruptured by H2O under moderate conditions and caused the formation of Si-OH groups [105]. Wang et al. [106] proposed an effective method to improve the surface chemical reactivity of a magnetic SiO2 catalyst support. More Ti-OH groups were bonded on the silica surface due to the formation of Si-O-Ti. Therefore, the incorporation of metal oxides such as TiO2, Fe2O3, and Al2O3 can greatly enhance the stability of nano-silica supported acid catalysts.

Pretreatment means the applied stages required in the plant in order to process feed­stocks prior to their conversion to biodiesel. Such stages typically involve reducing factors of negative impacts on the biodiesel production process such as water, gums, suspended particles, polymers, and mostly FFAs. Water normally leads to formation of increased concentration of soaps during alkaline transesteriflcation, reacts with the alkaline catalyst sodium methylate to form methanol and sodium hydroxide, and also shifts the equilibrium reaction toward hydrolysis under acid-catalysis condi­tions. The soaps can solidify or freeze up and clog lines which results in equipment lost time (downtime). One pretreatment method involves reacting caustic soda with FFAs; however, there will be significant yield losses in such pretreatment method. On the other hand, the acid pretreatment doesn’t lead to the formation of soaps and therefore will be minimal yield losses. Polymers, gums, and particulates in oil feed­stocks impose also negative impacts on the biodiesel production process as they lead to destroying of the catalyst and also implications on phase separation of oil/glycerol phases. The biodiesel industry worldwide has adopted several pretreatment methods that are described below:

• Liquid acid treatment—Pretreatment by esterification of FFAs with a liquid acid catalyst (Sect. 18.6.1).

• Distillation—Removal of FFAs by distillation (Sect. 18.6.2).

• Blending—Blending low FFAs feedstock with higher FFAs feedstock (Sect. 18.6.3).

• Glycerolysis—Glycerol reaction with FFAs (Sect. 18.6.4).

• Acid esterification with solid catalysts—Lower FFA with ion exchange (Sect. 18.6.5).

• Removal of FFA with solid adsorbents (Sect. 18.6.6).

• Pretreatment with enzymes (Sect. 18.6.7).

• Degumming—Removal of gums from crude oils (Sect. 18.6.8).

Organosolv pretreatment is a promising pretreatment strategy, since it has demon­strated its potential for lignocellulosic materials [121]. Numerous organic or aqueous solvent mixtures can be utilized, including methanol, ethanol, acetone, ethylene gly­col, and tetrahydrofurfuryl alcohol, in order to solubilize lignin and provide treated cellulose suitable for enzymatic hydrolysis [122]. Comparing to other chemical pre­treatments, the main advantage of organosolv process is the recovery of relatively pure lignin as a by-product [122], which can be used as a substitute for polymeric ma­terials, such as phenolic powder resins, polyurethane foams, and epoxy resins [123]. In some studies, these mixtures are combined with acid catalysts (HCl, H2SO4, oxalic, or salicylic) to break hemicellulose bonds. A high yield of Xyl can usually be obtained with the addition of acid. However, this acid addition can be avoided for a satisfactory deligniflcation by increasing process temperature (above 185 °C) [124]. Usually in the organosolv pretreatment, high lignin removal (>70 %) and minimum cellulose loss (less than 2 %) could be achieved [121].

Particulate impurities in the product gas typically originate from ash, dust, carry-over bed materials, and unconverted char. The particulates can cause corrosion and plug­ging in the downstream process equipment. The most commonly used techniques for particulate removal are cyclones for the initial cleaning of larger particles, barrier filters (low and high temperature operations), electrostatic filters (ESP), wet scrub­bers, and alkali salts. The application of the proper removal technique depends on the concentration of impurities, particle size distribution, and the particulate tolerance of the downstream application.

Drying constitutes another relevant pre-treatment for bark and wood residues for most pyrolysis reactor technologies. Biomass moisture fraction is generally controlled by the use of rotary drum dryers in the industry [31]. It uses the same principles as conventional clothes dryers: a conditioned air stream with low humidity enters the drum and is charged with moisture evacuated from biomass. Industrial rotary drum dryers can reach volumes as high as 200 m3. Using such drying equipment allows one not only to reach very low moisture fractions, but also attain a given moisture fraction set point that is desired for pyrolysis.

11.2.2.1 Sorting

Large pieces of inorganic material (e. g. metals and glass) can be present in significant quantities in MSW. Therefore, sorting of MSW is normally required since inert material will only consume useful volume, while some metal can act as catalyst to produce more pollutants [32].

11.2.2.2 Pre-Treatment for Sewage Sludge

In sewage sludge, organic matter is diluted in water and micro-scale inorganic ele­ments can be present. Once the sludge has been chemically and biologically stabilized at the wastewater treatment plant (pH neutralization and microorganisms’ denatura- tion), dewatering is the first step to recover municipal biosolids. Many techniques can be employed for dewatering and centrifugation is commonly used since it can easily yield suspensions of 20-25 wt% biosolids from 0.5 to 3 wt% diluted sewage sludge with an overall solids recovery of over 90 % [10]. Rotary drum filtration can also be used to remove water at lower levels. Depending on the operation, a controlled flow of air can be injected into the rotary drum filter, which can be used as a dryer to obtain moisture fractions below 10 wt% [10]. The wall of a rotary drum filter is meshed so the gas flow will be radial, as clogging of the Alter is avoided by constant scrub­bing mechanisms. As pure water can be obtained through the wastewater treatment process, the sludge sequesters impurities, which are then very difficult to remove.

Let us discuss presented plasma gasification modes of chicken manure with various consumptions and oxidants—1-4 (according to the data of Fig. 12.1). As it is seen the

CO content in syngas and the chemical energy yield increase while the temperature grows at the stoichiometric gasification mode. It is generally caused by the decrease of the graphite yield which becomes less than 0.1 g/kg only at temperatures more than ~ 1,400 K, and its conversion to carbon monoxide. On modes 3 and 4, change of composition is caused by the shift of equilibrium toward H2 and CO2 formation according to stoichiometry of water-gas shift reaction and insignificant methane content increase (up to ~1.2-1.6 %mol) at temperature decrease. The influence of the process temperature on a chemical energy yield is insignificant, as volumetric heating values of H2 and CO are close, and during water-gas shift reaction one is replaced by another.

Increase of energy consumption with the temperature growth in all modes is mainly caused by the increase of the sensible heat of the system, and also with the graphite gasification on stoichiometric modes. The influence of this effect on energy balance upon reaching 1,200 K decreases to ~2 kJ/kg K.

The syngas composition is a major factor defining rate of the Fischer-Tropsch synthesis. During the water-gas shift reaction, the sum of concentrations of hydrogen and carbon monoxide in syngas decreases (steam during the conversion is replaced by carbon dioxide), therefore the more initial H2/CO ratio close to stoichiometric, the higher is the rate of Fischer-Tropsch synthesiswhile using the converted syngas. It causes lower rates of synthesis for syngas produced by air plasma gasification in comparison with the steam plasma gasification. Rate decreases at the reduction of the plasma energy.

Thecontent of contaminants (such as H2S, COS, NH3, HCN, HCl, soot, tars, BTX, volatile metals, and dust) in syngas should be limited to prevent the accelerated aging of catalyst [58]. The residual content and residence time of these contaminants define catalyst lifetime, therefore the higher space velocity is the more valuable products can be made by the catalyst pellets before its deactivation. Usage of the steam plasma instead of the air plasma allows increasing space velocity by 15-59 % that leads to a decrease of financial expenses for the catalyst pellets in the Fischer-Tropsch process.

Alterations in liquid fuel yield are caused by changes in syngas composition and resemble the dependence of the chemical energy yield; however, they are not proportional as methane is inert to the Fischer-Tropsch synthesis.

The energy consumption (associated with energy used for plasma generation) is about 5.0-6.2 kW h for the steam and 1.9-5.2 kW h for the air oxidant to produce 1 kg of synthetic fuels by plasma gasification.

Carbon dioxide is a byproduct of synthetic fuel production from biomass. CO2 content in the residual gas can be very high (more than 90 %), while using steam plasma that makes it very attractive for innovative applications [59]. For exam­ple, CO2 reforming of methane [60] or synthesis of CH3OH by non-equilibrium electrocatalysis plasma reactor [61].

Further work to improve the OES-precipitating pretreatment may be focused on the fractionation of softwood by new kinds of OES because the substrate applicability of H3PO4 + organic solvent precipitating technology is by far limited to hardwood, herbaceous, and bamboo plants [97, 112, 114].

[AMIM]Cl is the only IL that has broad solubility to dissolve both the softwood and hardwood completely among the 96 ILs [85]. It can compete with the lignocellu — lose components for hydrogen bonding, attributed to its anions, especially chloride anions [85, 101]. However, the n-electrons which are exhibited within imidazolium ring and on side-chain of [AMIM]Cl cation, could have n-n interactions with the aromatic compounds of lignin [86, 101]. It reveals that [AMIM]Cl could take a stronger and more efficient effect on the lignin structures in softwood than other dissolving solvents. As a result, [AMIM]Cl could be employed to dissolve the soft­wood first, and then a bicomponent or tricomponent OES is homogeneously formed to deposit cellulose and hemicellulose by adding ethanol, or acetone and DMSO, or acetone and PEG mixtures. After flltrations, lignin would be separated from the OES by water and further modified to other products.

Therefore, OES is a green and efficient solvent to be employed in the pretreat­ment producing a high-reactivity lignocellulosic substrate suitable for enzymatic saccharification as well as high-quality lignin polymers for materials. OES has many advantages such as lower viscosity, rapid and selective solubility, generating less by-products, and recyclable. To meet the requirement of high biomass productivity for biofuels and biorefinery, further research with respects to the OES pretreatment should be focused on:

1. Designing new bicomponent or tricomponent OES which has broad applicability on different biomasses.

2. Developing efficient OES recycling techniques.

3. Conducting continuous flow process by combining unit operations of dissolving, precipitation and solvent recycling.

4. Optimizing the process and evaluating the economics of the entire process.

Acknowledgments This work was funded by Chinese Academy of Sciences (BairenJihua and Knowledge innovation key project (KSCX2-YW-G-075)), Yunnan Provincial Government (Baim — ing Haiwai Gaocengci Rencai Jihua and Provincial Natural Science Foundation), and China National Natural Science Foundation (No: 21076220). One of the authors (C. XU) would like to acknowledge the funding from Natural Science and Engineering Research Council of Canada (NSERC) through the Discovery Grant.