Fast oxidative pyrolysis of eucalyptus wood residues to replace fossil oil in pulp industry

This study evaluates the potential of the oxidative fast pyrolysis (OFP) of eucalyptus wood residues (EWR) for producing bio-oil to replace fossil fuels in the lime kilns of the pulp industry. OFP is an alternative to inert-atmosphere fast pyrolysis where separate heat addition stage is not required. OFP was studied by characterizing the fuel using proximate and elemental chemical analyses, thermogravimetric analysis and heating value. Then, OFP experiments in a pilot-scale autothermal fluidized reactor were done with EWR. Pyrolysis products were gases, bio-char and bio-oil (heavy and light). The gases were burnt, and the energy used for heating the fluidization air. Heavy bio-oil energy yield of 30% and 21.4 MJ kg (cid:0) 1 lower heating value indicate good potential for fuel applications. The results were used to model and evaluate industrial-scale cases. Integration with the pulp mill recovery boiler and steam cycle allows easy recovery of the considerable waste heat from the process itself, as well as the combustion of solid and gaseous residues. Economic analysis indicates profitability for OFP of fine EWRs from the mill. A higher net present value, but longer payback period, was obtained for a larger OFP plant using purchased feedstock. Stand-alone production was found unprofitable.


Introduction
Climate change due to anthropogenic greenhouse gas (GHG) emissions is considered one of the biggest challenges for many decades to come. Energy supply is the single greatest origin of the CO 2 , the main GHG, accounting for more than one third of the total emissions [1] according to the International Panel on Climate Change (IPCC). The IPCC considers sustainably produced bioenergy as having significant potential for climate change mitigation: the majority of pathways restricting warming to 1.5 • C include considerable amount of bioenergy, with potential of several gigatons of CO 2 emission mitigation annually through fossil fuel replacement [2].
The pulp and paper (P&P) industry is currently the single largest industrial bioenergy user and producer, surpassing both bioethanol and biodiesel production [3]. Of all pulp production, the kraft process contributes more than two thirds [4]. At over 30 million air-dried tons (ADt) annually, South America is a significant producer; more than half of this is from Brazil alone. In Brazil eucalyptus is the main feedstock, representing 75% of the 9 million hectare of total planted forestry in 2019 [5].
The forestry and forest industry generate considerable amounts of solid residues; as much 30-40% of an eucalyptus tree is lost as such [6]. This represents a significant and still largely unused potential, for use as raw material for bioenergy production [7].
In a modern kraft pulp mill, the majority of the energy used is already based on renewable biomass, and the process is self-sufficient in energy, with many mills selling considerable amounts of excess electricity to the national grid. The lime kiln, however, is a significant exception; according to the last reported global survey on lime kilns in 2008, almost 90% of the surveyed mills used fossil oil and/or natural gas in their lime kilns; tall oil from softwood was the main fuel in the majority of the rest [8]. The lime kilns thus remain a GHG emission source in an otherwise almost carbon-neutral industry.
Despite research on renewable alternatives, fossil fuel replacement has proven challenging: the thermal decomposition of lime mud CaCO 3 into CaO requires very high temperatures, and to avoid contaminants, the fuel should have very little ash. Many renewable alternatives have been considered, all having clear drawbacks: untreated wood and bark have low heating values and must be dried and hammered to fine powder; torrefied to biochar these are easier to powder and have slightly better heating values, but ash contents are even higher; and hydrogen from electrolysis needs low electricity prices to be economical [9]. Producer gas from gasification and pyrolysis oils are promising alternatives, but these, too, have their problems: acidic, corrosive fuel (fast pyrolysis), low heating values of the fuel (gasification), and expensive investment (both) [9].
Oxidative fast pyrolysis (OFP) is a little studied method of making bio-oil. Whereas in typical inert-atmosphere fast pyrolysis (IFP) the aim is in maximizing the bio-oil yield requiring an additional heat input step to maintain the reaction temperature, in OFP one oxidizes some biomass at pyrolysis stage to simplify the process. The main difference is that OFP is less complicated, but requires more biomass for the same bio-oil thermal output.
The goal of this study is to evaluate the techno-economic viability of oxidative fast pyrolysis of eucalyptus wood residues (EWR), with the main focus on pulp mill lime kiln fossil fuel replacement. As well as reducing GHG emissions, using bio-oil to replace heavy fossil oil in the lime kiln can be expected to reduce sulfur dioxide emissions, while not adversely affecting other emissions [10]. Since many factors (e.g. high density, homogeneity, fast growth and good availability as side streams of the forestry and forest industry sectors) combine to make EWR a promising feedstock for liquid biofuel production, also the wider potential of EWR as feedstock for renewable bio-fuel production through OFP is evaluated.
To address the lack of comprehensive reviews, and to provide context and comparison basis for the results obtained in this work, thorough reviews of existing data on OFP experimental studies, as well as fast pyrolysis plant investment costs, were first conducted. An experimental study was then performed in an auto-thermal pilot-scale plant, similar to that used by Refs. [11,12]. Thermogravimetric analysis was also performed to confirm the suitable thermal degradation performance of EWR for bio-oil production through OFP.
The experimentally obtained yields and product characteristics were then used in the techno-economic analysis part of the study. This was performed by first designing four different flowsheet models, each representing a different case: one where only the fine residues from pulp mill wood handling are used as a feedstock, one where additional wood is purchased for increased bio-oil production, and also two cases of biooil production for sales: a stand-alone plant, and one integrated to a biomass-fired power plant. The modelling results were then used as input for the final economic analysis of the considered cases. Experimental results of fast oxidative pyrolysis, characterization of the produced oil, and the industrial plant process designs considering the integration of fast pyrolysis process to a South American pulp mill and/or power plant process have not been published before. Although cautious assumptions are used throughout the work to ensure conservative estimates, the results indicate that OFP process provides cost savings in comparison to inerta-atmosphere FP, and can be an economically viable way of replacing fossil oils in kraft pulp mill lime kilns. Unless process parameters of an industrial-scale process can be significantly improved from those of the pilot-scale plant with which the experiments were carried out, the profitability of biocrude production for sales profit appears unlikely, however.

Fast pyrolysis process
Untreated wood is in many ways a problematic fuel. Low energy density increases logistics costs, and the modest heating value and high moisture content combine for low heating values and adiabatic flame temperatures [13]. Fast pyrolysis is a thermochemical conversion process that takes place in the temperature range 400-650 • C and transforms the solid low energy density biomass into a high energy density liquid (bio-oil).
During pyrolysis, biomass is heated in the presence of little or no oxygen, generating solid, gaseous and liquid fractions. Unlike slow pyrolysis, where residence times are typically long, up to several days to maximixe the char formation [14,15], in fast pyrolysis for bio-oil production residence times are usually 0.5-5 s [16,17]. Liquid yields vary in 30-65% range depending on feedstock and process parameters, with 450-525 • C temperatures, short residence times and clean wood feedstock typically resulting in the highest oil yields [18]. Rapid removal and cooling of the oil vapors are important for good liquid yields [19]. Ash content reduces the yield: potassium in particular catalyzes oil decomposition to gases, char and water [18].
Biomass fast pyrolysis has been extensively studied using mostly inert-atmosphere fluidized bed reactors at scales of 0.15-20 kg/h, different residence times and temperatures, and in bubbling, spouting and circulating fluidization modes [19][20][21]. Screw feeder driven reactors have also been used [21], but fluidized bed reactors are predominant due to their superior mass and heat transfer properties [19]. Inert-atmosphere fast pyrolysis (IFP) is a clearly endothermic process with a reaction heat estimated at almost 2 MJ kg − 1 [22]; in fluidized bed reactors heated sand is typically used as heat carrier to maintain desired bed temperature. Although most research has focused on IFP to prevent oxidative degradation of the oil [23], oxidative fast pyrolysis (OFP) with some oxygen in the fluidization gas offers the advantage of a simpler process by allowing autothermal operation without need for heat carrier or other means of reactor heating [23,24], reducing the energy consumption and thereby the operating costs in the pyrolysis plant.
A summary of experimental parameters and product characterization under oxidative pyrolysis conditions is given in Table 1. While the previous works of techno-economic assessment have mainly focused on either fully inert pyrolysis, or less often autothermal operation which can be achieved with an oxygen input of 10% or slightly less of the stoichiometric oxygen demand [18,19], the current work aims at simplifying the process even further by using a slightly higher oxygen input of approximately 20%. This has the advantage of supplying enough heat to operate also the drier. When integrated with an existing boiler plant, the fluidization gas temperature can be reduced to such levels that the gas can be taken as a suitable mixture from the boiler flue gas stream before the stack, after the heat transfer surfaces and electrostatic precipitators, and combustion air to the furnace. This allows a simpler process without the need of a heat carrier to maintain reactor temperature. Particularly in case of the recovery boiler, the use of hotter flue gases before the gas cleaning would likely be problematic in practice. While bio-oils from fast pyrolysis of eucalyptus have been evaluated in several studies [20][21][22][23][24][25][26], almost all of these focus on IFP; information on properties of oil produced by OFP of eucalyptus is still sparse. Addressing this research gap is one of the objectives of this work.

Fast pyrolysis plant economics
Similarly to experimental studies, plant design and techno-economic analyses largely focus on inert-atmosphere processes. Hamaguchi et al. [36] considered fast pyrolysis of eucalyptus residues as one of the possibilities for additional revenue streams in a South American pulp mill. Two options were evaluated: one where the mill would have a fluidized bed biomass boiler that could both supply hot sand as pyrolysis heat carrier and incinerate the pyrolysis gases, and one without, requiring a separate combustor for the same purposes. The integration with biomass-fired combined heat and power (CHP) plants has been investigated in several studies [37][38][39], demonstrating efficiency benefits from heat integration, as well as economic benefits from increased full load operation hours of the cogeneration plant due the presence of additional heat user. All of the above have considered only IFP, but Amutio et al. [24] reported results of simulating a stand-alone autothermal OFP plant.
The cost of producing bio-oil via fast pyrolysis applying varying assumptions for biomass cost, plant capacity, reactor technology, among other variables, have been reported in several studies; many of these dates back to the 1990's, well prior to the first commercial plants. In order to establish a basis against which to compare the results obtained in this study, a literature survey on fast pyrolysis plant costs was conducted. Some of the considered plants include further processing of the raw biocrude through e.g. cracking and hydrotreating; these are included when sufficient data is provided for estimating the pyrolysis plant investment alone. The data is summarized in Table 2. Investment costs range from $22 million to $793 million for 200-2000 bone-dry metric tones per day (tBD/d) capacities [40,41]. Recent studies have indicated EWR as a promising feedstock for second-generation biofuels [6], but also that economic viability of fast pyrolysis will likely require co-location and integration with P&P or sugarcane industries [40].

Feedstock preparation and characterization
Eucalyptus wood (Eucalyptus urophylla x Eucalyptus camaldulensis) residues were collected from a pulp and paper mill in São Paulo state, Brazil. Granulometric analysis was performed according to ASTM E828-81 (2004) standard, using Tyler series sieves of different meshes, with shaking time of 15 min and 80 Hz frequency. The particle size distribution was not uniform: 67% of the particles had diameters of <0.71 mm, the remainder 0.71-1.41 mm.
To evaluate the potential for fast oxidative pyrolysis process, the EWR was characterized by chemical and physical analyses. Moisture (MC), volatile (VM) and ash (AC) contents were determined as mass percentages by proximate chemical analysis according to procedures of EN 14774-217, EN 15148.18 and EN 14775.19 standards, respectively. Fixed carbon (FC) content was determined from The higher heating value (HHV) of biomass samples was determined in a Parr 6300 bomb calorimeter, according to DIN51900-1 standard. The analyses were carried out in duplicate and a mean value was reported. •Dry bio-oil with less than 1 wt% water and over 93 wt% total energy •Compared to inert pyrolysis, the yield and energy content of dry bio-oil was reduced by 22%-31% and 25%-34%, respectively.
• Bio-oil produced in oxidative pyrolysis reported lower heating value, tar content and heavy compounds content, while increasing viscosity and phenolic concentration, than those produced without oxygen. •Addition of oxygen in the pyrolysis increase the production of gases i.e., CO, CO 2 .
•Pyrolysis temperature 500 • C was preferred for bio-oil quality [ Ultimate analysis was performed using a TruSpec Micro -LECO Instruments 628 Series CHN elemental analyzer with oxygen and sulfur module. Prior to the analysis, the standard samples (ethylenediaminetetraacetic acid for CHN and coal for S measurements) were first analyzed to verify the experimental error within ±1% for the elements. The analyses were carried out in triplicate and a mean value corrected for moisture content was reported. Oxygen content was determined by difference of 100% and the sum of carbon, hydrogen, nitrogen, sulfur and AC on dry basis.
The suitability of EWR for fast pyrolysis was confirmed through thermogravimetric analysis (TGA) and differential thermal analysis (DTGA), carried out under inert and oxidative atmospheres at different heating rates, using SHIMADZU DRG-60H system. As expected, the results showed the OFP resulting in more rapid reactions and less char residues for than IFP; full results and analysis can be found in supplementary material A.

Oxidative fast pyrolysis
Fast oxidative pyrolysis experiments were carried out in a fully controlled, continuously operated auto-thermal SDB-20 pilot-scale plant (Fig. 1). It consists of the fluidized bed reactor (200 mm inner diameter, 935 mm height), with systems for biomass feeding, char collection, vapor condensation and bio-oil recovery.
The gas flow rates and the reactor temperature were monitored in order to control and maintain the variables at desired values. EWR feed rate was 15.06 kg/h, determined by measuring the mass difference in the storage hopper before and after the experiment. 480 ± 8 • C temperature was maintained, and reactor and condenser pressures were monitored during the tests to check for instabilities or obstructions in the condenser. The fluidization agents were air, supplied by a blower at 13 Nm 3 /h, and recirculation gases, supplied by a fan at 7 Nm 3 /h. The gas entered the reactor through a dispersion plate. The condensable and non-condensable gaseous products were sent to sequential cyclones to remove bio-char particles. The condensable vapors were cooled, condensed, and collected as bio-oil. A centrifugal device on top of the condenser collected the low-moisture, high-HHV heavy bio-oil phase, enabling the light organic and aqueous liquid phases to be collected separately. The non-condensables were burned in the combustion chamber to heat the fluidizing air. Quartz sand (Quartzo Brasil Minas 403/050) of 1300 kg/m 3 particle density was used as the bed material. Operating conditions were defined based on previous studies using a similar pilot-scale plant [12,24,35]. The pressure and temperature were registered using pressure transducers and thermocouples located along the reactor height to determine temperature and pressure profiles during the experiments. At the end of an experiment and cooling the system, the liquid fractions were removed and weighed.

Composition analysis and energy efficiency
The mass yields of products were obtained using the ratio between the quantity of each by-product produced by the oxidative fast pyrolysis and the amount of fed biomass. The non-condensable gases were directly burned without any auxiliar fuel in the combustion chamber to heat the fluidizing air in a continuous process, thus the flow rate of noncondensable gases was not quantified. The energy balance was determined considering the mass balance and higher heating value (HHV) data of the EWR and products from fast pyrolysis.
The HHV of the heavy bio-oil was measured using the Parr 6300 calorimeter. The char was characterized by proximate analysis (MC, VM, AC and FC) and HHV using the same procedures and standards as with the feedstock. Liquid and solid product densities were measured using a scale and a graduated cylinder.

Process simulation
For the techno-economic analysis, industrial-scale plants were simulated using commercial IPSEpro process simulation software. Four plant configurations were considered: two different scales of plant integrated to the CHP plants of a large South American kraft pulp mill for lime kiln fossil fuel replacement, and two plants for producing oil for sales: a stand-alone plant, and one integrated with a small biomass-fired power plant. All cases consider a fluidized-bed pyrolysis reactor using the same parameters as in the experiments: 480 • C temperature, equivalence ratio ER = 22% and gas-to-fuel ratio G/F = 2.0. Ambient temperatures of 20 • C (water) and 25 • C (air, feedstock) are assumed. Reaction heat was determined by assuming 10% total losses in the pilot •The maximum bio-oil yield was achieved at 470 • C. The bio-oil obtained has a low oxygen content and high heating value.
•The presence of organic acids compounds confers an acidic character to bio-oil.
•Brazil produce large amount of sugar cane straw residues gerenating a considerable environmental impact.
Fast pyrolysis have been proved as a potential residual management solution.
plant [24]; using the parameters listed in supplementary material B, the heat balance considering sensible and evaporation enthalpies was solved for reaction heat, yielding an exothermic value of h r = − 2540 kJ kg − 1 .
The energy balance including the heating values was then solved to obtain the HHV of the pyrolysis gas, 5.6 MJ kg − 1 .
The stand-alone plant (Case SA; Fig. 2) was then designed first, to serve as the basis for integrated plants. 200 t BD /d feed of 45% moisture EWR is dried in a belt drier (1.2 kWh/kg evap heat consumption) heated a The higher cost and higher combustion fraction is the figure listed in the reference which includes hydrocracking and related steam production equipment. Smaller costs in parentheses excludes hydrocracking and boiler and feedwater-related equipment needed for producing the steam; these make up approximately 90% of total combustion system cost. b Includes electric works, piping, automation and HVAC. c Operating or under-construction plant, approximate investment costs based on investor/operator statements and published estimates.
by flue gases from a combustor using char as auxiliary fuel to dispose of the pyrolysis gases. The losses are estimated in three parts: radiation heat loss, mass loss, and other unaccounted losses. Radiation loss was estimated with boiler standards for a solid-fuel steam boiler with maximum continuous rating equal to the feedstock LHV rate. 1% of incoming feedstock LHV was assigned to other unaccounted losses, and 1% of feedstock mass was assumed lost. These can be considered conservative estimates, unlikely to underestimate actual losses.
The gases and bio-oil vapors from the reactor cyclone are first cooled to stop further degradation. An exit temperature of over 300 • C and cocurrent heat exchanger arrangement are used to prevent condensation. In the stand-alone plant, the coolant is air, some of which is used in combustor and reactor fluidization. The 350 • C gas-oil vapor mixture is then condensed in a spray tower condenser; cooled oil is sprayed from the top, and oil vapors condense on the droplets. The oil is cooled to 60 • C with cooling water before pumping to the spray tower.  The pyrolysis plant, modified as necessary for heat recovery, was then integrated with the recovery boiler and steam cycle of a large pulp mill, and a small-scale biomass-fired condensing power plant (CPP), as seen in Fig. 3. The boiler and steam cycle modelling is described in Refs. [53,54].
The recovery boiler produces power and steam for the pulp mill. Steam is available as 4 bar low-pressure (LP) steam, and 16.5 and 10 bar medium-pressure (MP). Remaining LP steam expands in a condensing turbine. Steam pressures are maintained in controlled extractions, and condensing turbine flow throttled as necessary to maintain pressure and flow to steam consumers. The simpler CPP produces only power; without steam consumers, the LP turbine valve is kept fully open. Extraction pressures change due changing steam flow are estimated according to ellipse law as described in Ref. [43].
Three integrated plants are considered: pulp mill integration cases PM1 (only fines from mill are processed) and PM2 (large 400 tBD/ d plant using additional purchased feed), and condensing power plant case CPP (200 tBD/d). All share the same pyrolysis operations with the stand-alone plant, but use LP steam for drier heat supply, and steam cycle water streams serve as heat recovery heat sinks. The PM1 has the simplest heat recovery configuration as its the smallest one, both absolutely and more importantly relative to the steam cycle. Char and gases are disposed of in the boilers. The char would make a poor energy product to be sold as its ash content is high due to feedstock ash becoming enriched in the char fraction, and impossible to pelletize without additional adhesives. Both are also unsuitable for lime kiln fuel, due to ash content (char) or low HHV (gas). The integrated pyrolysis plants are described in more detail in supplementary material C.
Based on the simulation results, integration benefits are estimated from the main performance indicators of the plants: total conversion ratio CR of energy products to energy inputs, and electrical efficiency η el (for PM and CPP cases), defined in Eqs. (2) and (3): where Φ LHV,HBO , Φ LHV,EWR , and Φ LHV,f are the LHV heat rates of heavy bio-oil, EWR feedstock, and boiler fuel, respectively; Φ steam is the sum of process steams to pulp mill excluding pyrolysis plant (case PM); P el,net is the net power generation (cases PM and CPP) after deducting auxiliary consumptions of the boiler, steam cycle and pyrolysis equipment but not the pulp mill consumption, and P el,p is the purchased power (case SA).

Investment cost
The majority of plant equipment cost data is based on [41], which presents detailed breakdown of the purchased equipment costs of a 2000 tBD/d inert fast pyrolysis plant. The main exceptions are the oil condenser system cost, based on [45] which considers similar spray tower condenser as in this study instead of the shell-and-tube condenser in Ref. [41], and the dryer plant, which is based on vendor quotes. The costs of the heat exchangers, pumps and fans are based on [55]. Equipment cost is scaled by the well-known scaling rule where C and S refer to cost and scale, subscripts 0 and 1 refer to the initial equipment and equipment being considered, respectively, and n is the cost exponent. Values of n vary depending on equipment. Typically 0.6 < n < 0.7; here 0.65 is used for most equipment. The calculation process for determining the total capital investment is shown in Fig. 4. The installed equipment cost C IE is calculated from the purchased equipment FOB cost, estimated at 1.1 times purchase cost C PE , using an installation factor f i of 2.0. The C IE includes the costs associated with transportation, tolls and taxes, installation, instrumentation, and yard improvements on the site. To obtain total direct costs C DC , the following additional costs are estimated as percentages of C IE : buildings 10%, piping 20%, electrical work and automation 10%, service facilities 40% (greenfield) or 20% (existing pulp mill or power plant site), and 2% land (greenfield only). The total indirect costs C IDC , including engineering and supervision, construction-time expenses, legal and contractor fees, are estimated at 0.5C IE . The direct and indirect costs, and a contingency estimated at 20% of the C IE , form the fixed capital investment C FCI , to which is added 5% of working capital to obtain the total capital investment C TCI . The breakdown of the investment costs of each of the plants can be found inSupplementary Material D.

Profitability evaluation
The profitabilities of the considered cases at different energy price assumptions are evaluated with net present value (NPV), internal rate of return (IRR) and payback period (PBP). NPV is often considered the best evaluation metric if only one is used [56]. All methods generally give the same accept/reject decision, but in case of mutually exclusive projects of different scope, they may suggest different ranking of the options; using more than one metric is thus often useful [56]. The NPV of a project is the sum of the total present worth of future cash flows during the Fig. 4. Calculation process for determining total capital investment. economic project lifetime of n years discounted at an interest rate of i, minus the value of total capital investments C TCI .
where the annual energy outputs and inputs E and Q are calculated using the process simulation model, assuming annual operating time t of 8400 h. The IRR is found using Eq. (3) and solving iteratively such interest i that NPV = 0. The PBP is found similarly by setting the NPV = 0 in Eq. (5), and solving for n. Table 3 shows the values of economic parameters in Eq. (5) when not treated as variables. Three different scenarios were considered for market price of sold electricity. Constant electricity price throughout the plant operating time was assumed in each case. Electricity price was based on the average prices in Brazil during 2019: approximately 70 USD/MWh for spot price, and 120 USD/MWh for an industrial consumer's purchased electricity. The annual operating and maintenance cost C O&M is determined as a fraction of the C TCI . 2-3% estimates are common in the energy industry; 2.5% was used here. Table 4 includes the results of proximate and ultimate analyses, and HHV of EWR. Moisture was within the recommended range for the pyrolysis process, typically 7 … 15% [57]. High MC decreases the useful energy and results in bio-oil with elevated moisture levels, reducing its value as a fuel [57]. The HHV of the sample was similar to other lignocellulosic biomasses already evaluated as raw materials in pyrolysis [12,35,58]. High volatile matter and low ash contents, typical for woody biomass, were observed. The VM is of particular interest as it is related to the lignocellulosic fractions of the biomass, which during heating are thermally degraded, generating liquid and gaseous products seen in fast pyrolysis [59]. The FCC indicates the extent of nonvolatile organic matter in the biomass. EWR had a 12.89% FCC, close to values of vegetable biomasses, which are generally in the 10-25% range [60,61]. Feedstocks with high VM and low FCC are more susceptible to degradation [62], requiring less time during the conversion process. For several eucalyptus clones, the FCC, VM and AC are in the range of 11-12.4 wt%, 75.2-79.7 wt% and 0.1-0.25 wt%, respectively, similar to the values found for EWR. Slightly higher AC, but still below typical non-woody biomass values, was found, likely due to the presence of bark. At high AC, many of the ash minerals may cause problems for thermal conversion at moderate or high temperatures, e.g. corrosion, fouling and deactivation of selective catalytic reduction [62]. High AC (>2 wt%) can also influence product yields and compositions.

Characterization of feedstock and products
The ultimate analysis indicates high values of carbon and hydrogen in EWR, elements that contribute the most to the HHV of the fuel [60]; low O/C ratios are desirable in biofuels, because the energy found in C-C bonds has higher chemical energy than C-O bonds [63]. Cellulose is the most oxygenated and saturated constituent of wood, while lignin is the most unsaturated. Relatively high H/C and O/C were observed. Similarly to Ref. [61], small fractions of nitrogen (N) and sulfur (S) were found.
Most of the water produced in pyrolysis as well as feedstock moisture ends up in the light bio-oil. High water content results in higher density, and reduced heating value and combustion temperature. On the other hand, the presence of water reduces viscosity, improves atomization properties and reduces NOx formation in combustion [64]. The heavy bio-oil with low water content could be visually observed as a viscous, opaque black liquid, in contrast to the lighter, more transparent light oil.
In oxidative atmosphere, part of the char in pyrolysis are combusted, contributing to the formation of gaseous products. Addionally, the volatile material content related to the lignocellulosic fractions of the biomass are thermally degraded, generating vapors and gases, and inducing the formation of liquid and gaseous products in pyrolysis [11]. Characterization of the biochar, energy efficiency and energy density of EWR products can be found in supplementary material E.

Plant simulation results
The main operating characteristics of the considered plants are listed in Table 5; the resulting annual production and consumption of feedstock and energy are listed in Table 6. Cases PM0 and CPP0 refer to the basic recovery boiler and power plant configurations, without pyrolysis. The pyrolysis plant consumes both power and steam from the power plants, while producing additional boiler fuel (pyrolysis gas and char) and waste heat.
In cases PM1/2, the increased fuel input (char and pyrolysis gas) increases the steam generation. Increases in PM1 are very small; <1% of boiler fuel heat rate Φ f and turbine inlet flow ṁ ti , and a 2.4% increase of flue gas flow ṁ fg . These could very likely be accommodated in existing boiler and turbogenerator. PM2 increases are larger at 3.2% Φ f and 4.7% ṁ ti , and, with constant excess air ratio, a 14% ṁ fg increase. Particularly the flue gas flow increase may require modifications to the boiler. The increased main steam flow to the turbine can likely be accommodated, but will manifest itself as increased extraction pressures requiring some throttling and consequent losses. The condensing turbine steam flow change is negligible for PM1, and clearer but still likely within swallowing capacity for PM2, but will depend on the turbogenerator details of the considered mill.
The additional boiler power more than compensates for the power and steam consumptions of the pyrolysis plant and the slightly reduced turbine efficiencies: increased net power generations are obtained with both PM1/2.
In case of CPP, it is assumed that the basic CPP0 plant would operate at the maximum rated power, and the char and pyrolysis gas will replace some boiler fuel rather than increasing steam generation. The net result is thus a reduced power generation.

Economic analysis
The investment costs of the four OFP plants are shown in the context of the literature review results of existing scientific data and available information on planned and operating commercial plants in Fig. 5. The figure on the right demonstrates the specific costs increase dramatically at small-scale plants, although this is not immediately obvious from the As no detailed cost analyses of OFP plants exist, all comparison data points represent IFP. The cost savings due to the combination of not needing the heat input step, and co-location and process integration, are clear. Details of the investment cost analysis can be found in the supplementary material D.
The annual net cash flows of the studied cases are summarized in Table 7. Two different scenarios are considered in conditions of the investment amortization of 10% and 5% in equal annual payments within the project economic life. Both PM1 and PM2 are profitable at 10%, while CPP and SA fail to reach break-even at even 5%. The PM1/2 scenarios benefit from a combination of better performance, and being able to convert pyrolysis gas and char to electricity, rather than fuel savings as in CPP. In case of PM1 this benefit can likely be realized as neither boiler nor turbine would likely become bottlenecks for such small changes. With PM2 considerably greater uncertainty exists on whether this would be the case: depending on the given mill and boiler, either increased investment for boiler modifications, or reduced revenue due to not-fully-realized power generation increases could reduce the profitability. Case CPP is unprofitable, but at 5% interest rate only slightly, whereas the stand-alone plant is far from break-even, and thus excluded from further analyses. Fig. 6 depicts the variation of IRR, NPV and PBP as function of bio-oil worth, at three different C TCI scenarios: baseline, and±30% pessimistic/ optimistic scenarios. As expected, the pulp mill cases perform best within all three profitability metrics. Despite the small scale of the PM1 plant increasing the specific investment cost, the ability to use mill residues without feedstock purchase results in higher IRR and shorter PBP than the PM2. The larger capacity of the PM2 scenario yields a clearly higher NPV at most of the oil price and C TCI ranges considered, making it probably the preferred investment as long as the power generation increase is achievable without costly boiler modifications.   Within the variation range of the parameters, CPP reaches profitability only with a combination of optimistic C TCI scenario and very high oil price and/or low IRR. Unless the oil yield could be improved considerably, the bio-oil used locally as in PM1/2, or considerable subsidies are established for bio-oil production, the profitability of CPP appears highly unlikely.
In order to estimate how much uncertainties of the estimated parameter values affect the profitability analysis results, a sensitivity analysis was conducted by varying the investment, oil price, electricity price, feedstock price and O&M by ±20% and ±40% of baseline, and evaluating the impact on IRR. The results are shown in Fig. 7. All plants show greatest sensitivity to oil price and investment cost; between the plants, the small PM1 with high specific investment cost due to small scale is particularly sensitive for the investment cost. With the CPP the electricity price effects are reversed, and the impact of oil price is the greatest, as additional power and thus revenues cannot be generated using char and pyrolysis gas. The feedstock price is of moderate impact for both PM2 and CPP. The O&M cost variation had low impact with all cases.

Bio-oil combustion in a lime kiln
While the combustion properties of the bio-oil and modifications required on the kiln and/or the burners were not the focus of this study, based on literature, some conclusions can still be drawn. It appears likely that an existing lime kiln could be adapted for bio-oil with relatively small changes. The most probable changes would be due to the acidic nature of the pyrolysis oil; because of this, any material coming to contact with the oil, e.g., pipes, burners, pumps and tanks must be constructed of stainless steel [65]. Where temperatures and pressures permit, most plastics would also be chemically suitable. Silicon is a suitable material for gaskets and seals. Like heavy fossil oils, low-moisture pyrolysis oil is highly viscous and must be pre-heated for atomization. Unlike fossil oils, once-heated bio-oil cannot be returned to the storage tank, however, as it would not remain homogeneous. The bio-oil also degrades in storage, a factor favouring on-site use with comparatively short storage times.
The front-end temperatures of pulp mill lime kilns mostly fall within 900-1200 • C range [8]. As the kilns are typically almost 100 m long, the combination of high temperatures and long residence times makes near-complete combustion with low CO emissions relatively easy to achieve with different fuels. While fossil oils and natural gas are globally by far the most common fuels [8], many alternative fuels have also been successfully used, including wood-based solids and syngas, petcoke, treated animal fats, sludges, and tall oil pitch. Tests conducted at the University of British Columbia (UBC), Canada, confirmed in a pilot-scale kiln that fast pyrolysis bio-oil atomizes and burns well, with flame similar to natural gas [10]. Bio-oil combustion thus appears likely to be unproblematic.
In terms of emissions, the lime kilns can be significant sources of SO 2 and particulates at pulp mills [66]. While the particles are mostly lime dust and thus unlikely to be significantly affected by the fuel choice, the SO 2 typically originates mainly from the fuel. Replacing some of the heavy fuel oil with an almost zero-sulfur bio-oil can be expected to reduce emissions. The tests at UBC, Canada, showed CO, NOx and SO 2 emissions similar to natural gas combustion [10]. A summary of bio-oil characteristics compared to fossil oils and a some other fuels used in lime kilns is presented in supplementary material F.

Summary and conclusions
The viability of industrial-scale oxidative fast pyrolysis (OFP) of eucalyptus residues was investigated. Primary focus was replacing fossil oil as lime kiln fuel in pulp mills, but possible wider applicability in liquid bio-fuel production was also studied.
For context and comparison purposes, thorough reviews of existing data on fast pyrolysis process, and fast pyrolysis plant investment costs, were conducted first, as there were no pre-existing recent reviews of either. While a comprehensive body of literature does exists on inert fast pyrolysis, data on OFP is sparse; in case of industrial-scale plant technoeconomic analysis, non-existent prior to this work.   Fig. 6. Results of NPV, IRR, PBP analyses at different bio-oil worth (20-60 USD/MWh) and investment cost (±30%) scenarios. Fig. 7. Absolute IRR change as percentage points as a result of ±20% and ±40% changes of parameters.
An experimental study was then conducted in an auto-thermal pilotscale plant. Based on experimental data, flowsheet models of four industrial-scale plants were created: two different scales of plant integrated to a pulp mill, one integrated with a biomass-fired power plant, and a stand-alone plant. Experimental results of eucalyptus residue OFP, and results of plant designs, process integration and techno-economic analyses have not been published before.
The following main conclusions were reached from the results of the study: • There is economic potential for using bio-oil from oxidative fast pyrolysis to replace fossil oil in kraft pulp mill lime kilns. Although assumptions were chosen cautiously in order to obtain conservative estimates, both cases of integrating pyrolysis plants with the pulp mill were clearly profitable. Processing only the wood handling residues had the highest value of IRR; purchasing additional feedstock increased the NPV, but reduced IRR and increased payback period. • The clearly exothermic fast oxidative process increases also electricity sales of the mill; the profitability of the larger plant depends on the boiler and turbogenerator having the capacity to exploit this. Replacing all fossil fuels in the lime kiln would likely require modification, or greenfield design. • Producing oil for sales revenue would be profitable only if the performance of the pilot plant could be significantly improved in an industrial-scale implementation, especially in case of stand-alone production. • With a lower heating value of 21.4 MJ kg − 1 , oxidative fast pyrolysis bio-oil has good energy carrier properties and appears suitable for replacing fossil oils in a pulp mill lime kiln.