Process Analysis of Waste Animal Fat Pyrolysis and Fractional Distillation in Semi-Batch Reactors: Influence of Temperature and Reaction Time

Abstract

Waste animal fat (WAF) can be converted to distillate fractions similar to petroleum solvents and used as solvents via pyrolysis and fractional distillation. Pyrolysis oil from triglyceride materials presents adequate viscosity and volatility, compared to petroleum fuels, but shows acid values between 60–140 mg KOH/g, impeding its direct use as biofuels without considerable purification of its distillates. Fractional distillation can be applied for the purification of bio-oil, but only a few studies accurately describe the process. The purpose of this study was to evaluate the effect of temperature in the conversion of waste animal fat into fuel-like fractions by pyrolysis and fractional distillation in a semi-batch stirred bed reactor (2 L) according to reaction time. Waste animal fat was extracted (rendering) from disposed meat cuts obtained from butcher shops and pyrolyzed in a stainless-steel stirred bed reactor operating in semi-batch mode at 400–500 °C. The obtained liquid fraction was separated according to reaction time. The pyrolysis bio-oil at 400 °C was separated into four distinct fractions (gasoline, kerosene, diesel, and heavy phase) by fractional distillation with reflux. The bio-oil and distillate fractions were analyzed by density, kinematic viscosity, acid value, and chemical composition by gas chromatography coupled to mass spectra (GC-MS). The results show that, for semi-batch reactors with no inert gas flow, higher temperature is associated with low residence time, reducing the conversion of fatty acids to hydrocarbons. The distillate fractions were tested in a common application not sensible to the fatty acid concentration as a diluent in the preparation of diluted asphalt cutback for the priming of base pavements in road construction. Kerosene and diesel fractions can be successfully applied in the preparation of asphalt cutbacks, even with a high acid value.

1. Introduction

The use of petroleum-based fuels and solvents is detrimental to the environment due to the disruption of the natural balance of CO₂ in the atmosphere, since combustion and emissions of volatile organic compounds (VOC) generate greenhouse gases (GHG) that were stored in the molecular structure of fuels and solvents far underground in oilfields. In the case of biofuels and solvents, there is no net generation of GHG because they are produced from available biomass in the atmosphere [1,2]. There are varied ways to produce biofuels and solvents, including transesterification, pyrolysis, hydrocracking, hydrothermal carbonization/liquefaction, and others [3], that demand different equipment and technical expertise and have different production costs. Pyrolysis is a simple method that can be applied to upgrade combustible matter (organic matter) into better, more efficient fuels [4,5]. It consists of heating the feedstock in an oxygen-deficient atmosphere, achieving more thermal decomposition than combustion of the sample and generating a solid (char), a liquid (bio-oil), and gases (CO, CO₂, H₂, and light hydrocarbons), and this can be achieved in simple stainless steel semi-batch reactors. Due to its robustness, it can be applied to different types of feedstock, and the construction of the reactor system is relatively simple, lowering investment costs and making it ideal for local small-scale waste treatment that is capable of upgrading wastes, such as lignocellulosic biomass [6], fatty wastes [7,8], municipal solid waste [9], and plastics [10].

Fatty wastes are promising for the production of biofuels and bio-solvents by pyrolysis due to the similar molecular structure of triglycerides and fatty acids to linear hydrocarbons and their wide availability. It generates a considerable quantity per year of fatty wastes as used frying oil in deep frying, grease traps, abattoirs, and meat shops, animal food production, and the vegetable oil industry as its effluents and wastes, such as palm-oil mill effluent (POME) and palm fatty acid distillate (PFAD) [11,12]. One of the established biofuel industries is the biodiesel industry, where lipid-based material is transformed into fuel by the transesterification of animal fats and vegetable oils, with 23% of the demand met by using fatty wastes as feedstock [13].

The pyrolysis of fatty wastes can be conducted in simple semi-batch reactors operating at atmospheric pressure with or even without inert gas flow, but the process does not lead to the complete conversion of triglycerides to hydrocarbons, increasing acidity of the formed bio-oil due to the formation of fatty acids, which are reaction intermediates in the mechanism of lipid pyrolysis [14]. A high acid value in fuels is detrimental to compression–ignition (CI) engines and is related to coke deposits in the engine, corrosion, and leaks [15]. The optimization of process parameters can be conducted to improve the hydrocarbon composition of the reaction products, such as temperature [16,17,18,19,20,21,22,23], heating rate [17], and residence time [19,20]. The main papers concerning the transformation of waste animal fats and vegetable oils by pyrolysis are presented in

Table 1. Different studies concerning the effect of process parameters in pyrolysis of waste cooking oil.

Feedstock	Reactor	Operation Mode	Investigation	Results	Reference
Waste frying oil (palm)	Borossilicate glass reactor (500 mL)	Semi-batch	Temperature influence (350–450 °C) on product yields	Temperature elevation gives higher bio-oil production, increasing diesel production and reducing kerosene and gasoline. Bio-oil presented 0.91 g/cm3, 8.03 mm2/s, and 128 mg KOH/g, containing over 50 wt.% of fatty acids.	[16]
Lamb fatty waste	Stainless-steel fixed bed reactor (200 g)	Semi-batch	Temperature (400–550 °C) and heating rate (5–15 °C/min) influence on products	Bio-oil yield tends to increase until a maximum value is reached and tends to reduce with increasing heating rate; Bio-oil presented acid values of 120 mg KOH/g with 0.89 g/cm3 and 5.2 mm2/s.	[17]
Waste frying oil (animal and vegetable)	Stainless-steel fixed bed reactor (500 mL)	Semi-batch	Temperature influence (300–700 °C) on product yields	Bio-oil yields tend to increase with temperature, and thermal pyrolysis oil contained over 33 area.% of fatty acids.	[18]
Waste vegetable oil	Stainless-steel evaporator followed by fixed bed reactor (39 g/h)	continuous	Temperature (450–500 °C) and residence time influence when using CaO as catalyst pellet	Bio-oil yields tend to decrease with increasing temperature and tend to increase with reduced residence time. Bio-oil presented 0 acid value, density of 0.82 g/cm3 and 1.97 mm2/s.	[19]
Waste cooking oil	Stirred pressure reactor (30 g)	batch	Temperature (360–420 °C) and time (0–120 min) effect reaction products	Temperature increased the quantity of fatty acids formed when considering 0 min of reaction time. Increasing reaction time increased hydrocarbon yield and organic gases, lowering FFA.	[20]
Waste bleaching earth oil	Fixed bed stainless-steel reactor	Semi-batch	Temperature (450–600 °C) influence on reaction products	Temperature increased bio-oil yield until maximum at 550 °C, and higher temperature favored higher concentration of hydrocarbons and lower FFA. Temperature increased lower BP compounds and decreased higher BP compounds; bio had 0.89 g/cm3, 6.36 mm2/s, and 13 mg KOH/g.	[21]
Waste cooking oil	Stainless-steel fixed bed reactor (10 L)	Semi-batch	Temperature influence (350–550 °C)	Bio-oil yield increases until 450 °C; Gas yield reduced with increase in temperature; Bio-oil had 0.85 g/cm3 and 3.19 mm2/s;	[22]
Soybean oil	Stainless-steel fixed bed tubular reactor (10 kg)	continuous	Temperature influence (450–600 °C)	Bio-oil yield decreased with increasing temperature;	[23]

, along with their main investigations and results.

Temperature possesses a remarkable influence on changing product yield and affecting the chemical composition of triglyceride pyrolysis oil, but one needs to consider carefully what it means to use different temperatures of conversion depending on the mode of operation of the reaction. For semi-batch processes, the vapors formed are constantly flowing out of the heated reaction zone and entering the condenser, virtually stopping any further cracking of the molecules. This usually means that the reaction is conducted in distillation mode and atmospheric pressure or a vacuum, meaning that, depending on the heat-transfer mode and heating rate, the reactor temperature is connected to the boiling point of the mixture of vapors formed during pyrolysis. If a reactor is not capable of supplying excess heat, the temperature tends to remain around the boiling point of the mixture of compounds formed [24], and one should observe temperature as controlling the extent of the reaction, i.e., the time needed to convert all the available material in the reactor chamber [5]. This is valid only for semi-batch processes where the feed is charged in the reactor before any heating, and the process changes dynamically. A different case is observed when the feed is slowly pumped into a heated zone, like most continuous processes [19,23], or when the vapors are further cracked in a secondary heated zone defined as vapor-phase cracking. Further complexity is observed when a semi-batch process is coupled to vapor-phase cracking due to a change in the vapor composition, depending upon reaction time [4,5]. The complexity of the process analysis of semi-batch pyrolysis reactions is the cause of the apparent contrary effects of temperature in reaction yields or product composition, which are displayed in Table 1 for semi-batch and continuous processes.

In semi-batch reactors, the flow rate of the vapors formed is connected to the available heat or the heat-transfer capacity and power applied in the reactor. A high heating rate means a higher flow of vapors out of the reactor and less residence time, meaning that the heating rate should negatively affect the cracking of molecules. Hassen et al. reported that increasing the heating rate reduced bio-oil yield and increased syngas yields, showing that there is an intrinsic effect of heating rate over the products of the reaction, even though residence time is reduced, and this is corroborated by the slight increase of bio-char for the experiments of Hassen et al. [17]. The effective control and measurement of residence time is hard to achieve due to innumerous parameters affecting it, such as oil and vapor flow, heated zone, phase changes, presence of inert gas, geometry of the reactor, and heating mode. Only a few studies tried to evaluate the effect of residence time in reaction products [19,20]. Chang et al. conducted pyrolysis of waste cooking oil in continuous mode by pumping the oil to an evaporator followed by a heated tube containing CaO catalyst and reported that a lower residence time tends to increase bio-oil yields due to incomplete cracking [19]. Ito et al. effectively controlled the residence time of pyrolysis products by conducting pyrolysis in batch mode in a pressurized autoclave and showed that higher reaction time (residence time) increased hydrocarbon yields and reduced fatty acid content, but it is not specified if the formed products are part of the liquid bio-oil or non-condensable gases [20].

As temperature rises in a semi-batch reactor, the feed starts to react and vaporize and the composition of the feed, as well as the vapors formed, changes at any given moment of the reaction, meaning that to analyze the effect of a given parameter such as temperature in reaction products, one needs to consider the effect of reaction time as well. This can be achieved by fractionating the collected bio-oil during pyrolysis experiments according to reaction time. In previous works, we evaluated the effect of reaction time on product yields and the chemical composition of the catalytic pyrolysis of waste animal fat using activated carbon [4] and red-mud pellets [5] for catalytic upgrading second-stage vapor-phase cracking. But due to the complexity of analyzing a semi-batch two-stage vapor-phase catalytic cracking, we now present the process analysis considering one-stage thermal cracking in order to better explain the methodology and how it can be applied for chemical process improvement and design.

Fractional distillation can also be used to improve the quality of the obtained pyrolysis bio-oil of many feedstocks, such as lipids [25,26,27,28], algae [29,30], lignocellulosic biomass [31,32,33,34,35], polyolefins (plastic) [36,37], tires [38,39], municipal solid waste [40]. This is conducted by separating it into cuts better suited to conventional fuels, such as gasoline, kerosene, and diesel, but only a few studies accurately describe the process of fractional distillation of animal fats and/or vegetable oils pyrolysis oils in batch mode [25,27,28]. As in the case of pyrolysis, there is a variation in the feed composition as it is distillated, making it harder to establish standard parameters for the production of adequate fractions. In addition, pyrolysis bio-oil is a complex mixture of compounds, and its composition is related to feed composition and, also, processing parameters, increasing even more the complexity of analysis. In this work, a concise description is provided of fractional distillation of WAF pyrolysis bio-oil in order to improve knowledge of distillation in lab-scale and batch mode.

High acid value fuels are deleterious to compression–ignition engines [41], and the thermal pyrolysis of animal fats and vegetable oil produces free fatty acids that are encountered in liquid bio-oil [14]. Due to the complexity of the mixture, including the formation of azeotropes and close boiling-point compounds, most of the time, it is not possible to separate all the fatty acids present from liquid distillates, and they still possess acidity over the specified limit of 0.5 mg KOH/g [27]. Then, we applied the obtained distillate fractions in a conventional application in Brazil where acidity is thought to have only a minor or even good effect: as asphalt cutback, for priming base layers of pavement during road construction [42]. In this application, volatility and viscosity are the keys to producing a penetrating and rapid-curing impermeable layer where the asphalt cement is applied. A perfectly applied priming layer provides a hydrophobic surface and increases cohesion between the base layer and the asphalt cement, reducing slippage and wear on the road [43]. The results are thought to be enlightening about the use of pyrolysis bio-oil in non-fuel applications.

In this work, we aim to improve the process analysis of waste animal fat pyrolysis by analyzing the effect of temperature and reaction time in semi-batch pyrolysis processing on product yields, the physical–chemical properties, and the chemical composition. Furthermore, it provides a description of the fractional distillation of the WAF bio-oil into four fractions, and its distillate fractions are tested in a conventional non-fuel application, namely as asphalt cutback, which is used to prime base pavement in road construction.

2. Materials and Methods

2.1. Methodology

The methodology applied to study the pyrolysis and fractional distillation of WAF in semi-batch mode is described in Figure 1. Meat waste was obtained from local butcher shops and shredded into small pieces to be subjected to a rendering process in order to extract its lipid content. The meat was heated in a fat medium composed of pork fat in order to avoid charring of the waste, and it was rendered until full shrinkage and filtered to separate the extracted meat (animal meal) and the waste animal fat (WAF) used to produce bio-oil through pyrolysis in a fixed bed reactor (2 L) at different temperatures (400–500 °C). The collected bio-oil was fractionated in samples according to reaction time. Afterward, each time fraction of the bio-oil was distilled in a batch fractional distillation column equipped with a glass reflux condenser to obtain distilled fractions at the boiling points (BP) of gasoline (BP < 175 °C), kerosene (175 < BP < 235 °C) and diesel (BP > 235 °C). The bio-oil time fractions and their respective distilled cuts were analyzed to obtain their physical–chemical properties (density, kinematic viscosity, and acid value) and chemical composition by GC-MS. The distilled fractions were used to prepare diluted petroleum asphalt (DPA), and the resulting products were characterized according to standard procedures of DPA evaluation as Saybolt–Furol viscosity, flash point, and priming essay. Furthermore, it was evaluated on the dynamic viscosity of the mixtures.

2.2. Acquisition, Pre-Treatment, and Rendering Process

Meat waste was collected in the free market located in the Guamá district of Belém-PA, Brazil. The collected meat waste was trimmed in order to contain mostly fat tissue before being cut into small pieces to be fed to an LPG-heated stirred bed reactor (60 L). One kg of pork fat was added initially to create a heating medium and avoid charring. The meat had its fat rendered for several hours before being filtered and separated by a strain filter. The resulting product was WAF. Figure 2 shows the LPG-stirred bed reactor and fat separation.

The obtained WAF was characterized according to its acid value and kinematic viscosity, and the results are shown in

Table 2. Physical–chemical properties of WAF.

Property	This Work	Olubunmi et al. [44]	Kleinberg et al. [45]	Okwundu et al. [46]
Acid Value (mg KOH/g)	2.64	4.82	2.77	1.07
Kinematic viscosity (mm2/s)	27.25 (60 °C)	24.6 (40 °C)	9.58 (100 °C)	32.37 (40 °C)

. This was compared with other works [44,45,46], where a similar feedstock was characterized in order to have a better knowledge of the feed to the pyrolysis reactor. The results show similar values to beef tallow obtained through similar methods, being semi-solid at room temperature, with an acid value lower than 5 mg KOH/g, a kinematic viscosity in the range of 30 ish mm²/s at 40 °C, and reducing with increasing temperature.

2.3. Pyrolysis

The pyrolysis of WAF was conducted in a stainless-steel fixed bed reactor (2 L) (Implementação LTDA, Rio de Janeiro, Brazil) described in detail elsewhere [4,5] and visualized in the schematics of Figure 3. Briefly, 800 g of WAF are fed manually into the reactor before closing it. The reactor was heated to 200 °C and maintained at this temperature for 30 min in order to allow for complete distribution of heat and temperature of the system before being heated to a specified temperature, with a heating rate of 10 °C/min. No flushing of the system was performed using inert gases, and pyrolysis was conducted in an air atmosphere. The experiments were conducted using 3 different temperatures (400, 450 and 500 °C). The pyrolysis of triglycerides usually forms 4 products, char in the reactor, organic bio-oil, an aqueous phase, and non-condensable gases (NCG). The vapors formed (bio-oil, aqueous phase, and NCG) in the reactor are condensed in a tube condenser and separated from NCG in a stainless-steel drum. As vapors start to form in the reactor, the reaction time is counted, and this is marked as 0 min of reaction. Every 10 min, a sample is collected, weighed, and transferred to a glass separating funnel for separation from the aqueous phase. A total of 3 or 4 samples is obtained for each pyrolysis experiment.

A mass balance was conducted in order to obtain the yields of pyrolysis products, i.e., char, bio-oil, and non-condensable gases. The initial feed weight is recorded, and after the experiment, the amount of bio-oil and removed char from the reactor is weighted. The amount of non-condensable gases is obtained by difference, as in the mass balance of Equation (1). It is important to keep in mind that any errors or losses involved in the process will be observed in the non-condensable gases yield. The yields were calculated in relation to the initial weight of the feed.

$$m_{gases}=m_{feed}-m_{bio-oil}-m_{char}$$

(1)

2.4. Fractional Distillation

The fractional distillation of the bio-oil obtained at 400 °C and 10 min of reaction time was conducted in batch mode in a borosilicate glass distillation apparatus composed of a heating mantle, a 500 mL two-way round bottom flask, an 11-stage Vigreux column (Laborglass, Belém, Brazil), a reflux condenser, and a separating funnel. K-type thermocouples are used to measure the top and bottom temperature of distillation in the reflux condenser and round bottom flask, respectively. The flask and distillation column were wrapped with aluminum foil to thermally insulate the system. Figure 4 displays the fractional distillation apparatus used. Around 250 mL of bio-oil are loaded into the flask and heated with the reflux valve completely closed until distillation starts. The reflux valve is maintained in a closed position for 20 min in order to allow a homogenous distribution of temperature along the distillation column before being partially opened and collecting the distillates. The distilled fractions were separated based on the top distillation temperature into three fractions: gasoline (BP < 175 °C), kerosene (175 < BP < 235 °C), and diesel (BP > 235 °C). Distillation was stopped when there was a visual sign of thermal decomposition and a low flow of distillates to the separating funnel.

A material balance of the distillation process was conducted in order to observe the separation and yields of the distillation process. Since bio-oil was cut into three distilled fractions, a heavy phase remains in the boiling flask at the end of distillation, and there is a formation of non-condensable gases and/or gas losses through the vacuum adapter of distillation. The material balance is described by Equation (2). Since the initial feed, the gasoline, kerosene, diesel, and heavy fraction were weighted; the gas loss is obtained by the difference in the material balance.

$$m_{feed}=m_{gasoline}+m_{kerosene}+m_{diesel}+m_{heavyphase}+m_{gasloss}$$

(2)

2.5. Preparation of Diluted Petroleum Asphalt (DPA) Mixtures

The distilled fractions were used to prepare DPA mixtures composed of 60 wt.% petroleum asphalt cement (PAC) 50/70 and 40 wt.% of distillates. The PAC (Figure 5a) was kindly donated by Companhia Brasileira de Asfaltos da Amazônia (CBAA) (Ananindeua, Brazil), and its physical–chemical characterization is displayed in

Table 3. Physical–chemical properties of PAC.

Property	Method	Specification	Result
Penetration	D 5	50–70 mm	55 mm
Softening Point	D 36	46 °C	49 °C
Brookfield Viscosity (135 °C-SP21 20 rpm)	D 4402	274 cP	322 cP
Brookfield Viscosity (150 °C-SP21 20 rpm)	D 4402	112 cP	160 cP
Brookfield Viscosity (177 °C-SP21 20 rpm)	D 4403	57–285 cP	61 cP
RTFOT Retained Penetration	D 5	55%	65%
RTFOT Softening Point Increase	D 36	8 °C (max)	4 °C
RTFOT Ductility 25 °C	D 113	20 cm	>150 cm
Trichloroethylene solubility	D 2042	99.5 wt.%	100 wt.%
Flash Point	D 92	235 °C	312 °C
Thermal Susceptibility Index	X 018	−1.5 to 0.7	−1.2
Relative density (20 °C)	D 70	1	1.009
Heating to 177 °C	X 215	No foaming	No foaming

. The standard distillate used in the preparation of DPA is commercial kerosene (Figure 5b), with its physical–chemical properties shown in

Table 4. Physical–chemical properties of Apache commercial kerosene.

Property	Value
Physical state	Liquid
Color	Colorless
Aspect	Translucent
Odor	Solvent
Distillation range	150–300 °C
Flash Point (Closed vessel)	40 °C
Flash Point (NBR-DNER)	60 °C
Vapor Pressure (38 °C)	1.4 kPa
Vapor density	4.5 (a = 1)
Solubility	Water and organic solvents
Auto-ignition temperature	238 °C
Inferior explosive limit	0.7%
Superior explosive limit	5.0%

. A total of 12 mixtures were prepared, varying the content of distillates in the DPA according to

Table 5. DPA prepared sample proportions.

Sample	PAC 50/70 (wt.%)	Gasoline (G) (wt.%)	Kerosene (K) (wt.%)	Diesel (D) (wt.%)	Commercial Kerosene (CK) (wt.%)
0	60	0	0	0	40
1	60	10	0	0	30
2	60	20	0	0	20
3	60	30	0	0	10
4	60	40	0	0	0
5	60	0	10	0	30
6	60	0	20	0	20
7	60	0	30	0	10
8	60	0	40	0	0
9	60	0	0	10	30
10	60	0	0	20	20
11	60	0	0	30	10
12	60	0	0	40	0

. The DPA mixtures were prepared by heating PAC on a hot plate to 120–140 °C, in order to reduce its viscosity, and mixed with a calculated distillate weight until complete dissolution of PAC into the solvent.

The solvent used to prepare MC-30 asphalt cutback is not fixed as commercial kerosene, and other distillates can be used. The requirement is that the prepared mixture complies with viscosity, flash point, and penetration depth minimums. Since distillates from WAF pyrolysis can present similar characteristics of density, kinematic viscosity, acid value, and even chemical composition among all three fractions, depending on characteristics of the fractional distillation process such as column height, vapor flow, and reflux ratio, it was tested for all pure distillates (gasoline, kerosene, and diesel) as full substitutes of commercial kerosene, as well as mixed proportions of commercial kerosene and the same distillates, observing which mixtures could attend to the DNER requirements.

2.6. Characterization of Products

WAF, pyrolysis bio-oil, and distilled fractions were characterized through an adaptation of the official method for acid value (AOCS Cd3d-63). Briefly, 0.2 g (1.0 g in the case of WAF) were dissolved in 50 mL of a combined solvent of toluene and isopropanol 1:1 and titrated, with standard 0.1 N KOH solution prepared in isopropanol and phenolphthalein as an indicator. A blank titration was conducted with just 50 mL of solvent. The acid value was then calculated through Equation (3).

$$\mathrm{A}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{V}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}\left(\mathrm{m}\mathrm{g}{\displaystyle \frac{\mathrm{K}\mathrm{O}\mathrm{H}}{\mathrm{g}}}\right)={\displaystyle \frac{56.1\times {\mathrm{C}}_{\mathrm{K}\mathrm{O}\mathrm{H}}\times {\mathrm{f}}_{\mathrm{c}}\times ({\mathrm{V}}_{\mathrm{g}}-{\mathrm{V}}_{\mathrm{b}})}{{\mathrm{m}}_{\mathrm{s}}}}$$

(3)

where C_KOH is the concentration of potassium hydroxide solution used in mol/L, f_c is the correction factor of the solution, V_g is the volume of titrating solution dispensed, V_b is the blank volume, and m_s is the weight of the sample. Kinematic viscosity of WAF, pyrolysis bio-oil, and distilled fractions were measured using a Cannon–Fenske glass viscosimeter (no. 300, Schott-Gerate, Blaufelden, Germany) in a thermostatic bath at 40 °C. The readings were conducted manually using a manual pump and chronometer. The readings were conducted twice in order to obtain a better accuracy of results. The densities of samples were measured for bio-oil and distilled fractions using a 5 mL pycnometer calibrated with distilled water. Bio-oil and distilled fractions were further characterized with respect to their chemical composition by GC-MS, as described elsewhere [34]. Briefly, 1 µL of sample is diluted in 1 mL of acetone and injected in split-mode (1:50) in a gas chromatography equipment (Agilent Technologies, Model CG-7890B, Santa Clara, CA, USA) coupled to a mass spectrometer (MS-5977A, Agilent Technologies, Santa Clara, CA, USA) in a capillary column of fused silica SLBTM-5ms, (30 m × 0.25 mm × 0.25 mm, Sigma-Aldrich, St. Louis, MO, USA). The gathered spectra were compared with the NIST database, and no internal standard was used. The chemical composition is reported in area.% of chromatograms.

DPA mixtures were characterized to check for key parameters in their application, as defined in the DNER Standards [47,48,49,50,51,52], defining the minimum requirements for medium cure DPAs, including the criteria for approving or rejecting the mixture. The methods are listed in

Table 6. Standard test methods and specifications of regular DPA.

Property	Unit	Method	Value
Kinematic viscosity (60 °C)	cSt	ME 151/94	30–60
Saybolt–Furol viscosity (25 °C)	s	ME 004	75–150
Flash Point	°C	NBR-5765	38

, as well as their defined limits. Dynamic viscosity was determined in a rotational viscosimeter (HAAKE Viscotester, model VT550, Rheology Solutions Pty Ltd., Melbourne, Australia) and converted to kinematic viscosity, considering the weighted average density of DPAs calculated from the densities of pure components and their respective mass fractions. It used a coaxial cylinder configuration for the cup and spindle. Samples were loaded into the cup and inserted into a thermostatic bath (accuracy 0.1 °C) coupled to the equipment, and the temperature was set to 60 °C. The shear rate varied between 0–600 s⁻¹, and the time of analysis was 180 s. The results were obtained via the software of the equipment (Thermo Scientific, Waltham, MA, USA, HAAKE RheoWin Measuring and Evaluation Software v.2.94).

Saybolt–Furol viscosity was determined with appropriate equipment (Petrodidatica, Guarulhos, Brazil), and basically, the sample was loaded into a vessel immersed in a thermostatic bath. The sample is allowed 10 min in order to obtain a homogeneous temperature distribution, and a small cork is removed from the bottom of the vessel, starting a chronometer and allowing the sample to flow into a 60 mL cup. The time was stopped when the cup reached its mark, and determinations were conducted twice to obtain better accuracy. The flash point was determined by heating the sample in a small beaker over a hot plate. A k-type thermocouple was used to check the temperature of the DPA + beaker system. A small LPG flame was passed above the surface of the sample periodically, and when the first flame was visualized, the temperature reading was taken as the flash point.

The priming assays were conducted only with DPA prepared with a single distilled fraction (commercial kerosene, WAF gasoline, kerosene, and diesel samples) comprising the samples 0, 4, 8, and 12 from Table 5. In order to check the penetration of DPA into soil, samples of adequate composition similar to what is used in the pavement of roads must be used to prepare some test specimens of soil to be coated with the prepared DPAs. It used a laterite sand soil donated by a road pavement company (RODOVIÁRIO VILAÇA LTDA, Pedra Branca do Amapari, Brazil) presenting optimal humidity of 9.9 wt.%, defined as H₀, and 1.732 g/cm³ of dry specific weight. The soil is first completely dried at 60 °C for 24 h, then sieved (opening 2.0 mm) in order to remove large soil chunks and stones, and then mixed in order to get a homogenous distribution of moisture. The priming assay is investigated by adjusting the soil moisture around the optimal value of H₀ in steps of 2%. The soil samples are divided into 500 g quantities, and the necessary amount of distilled water is added to obtain moisture values near the optimum moisture value of the sample, in this case 5.9% (H₀ − 4%), 7.9% (H₀ − 2%), 9.9% (H₀), 11.9% (H₀ + 2%), and 13.9% (H₀ + 4%). The humidified samples are mixed and passed in a 4.76 mm opening sieve to obtain the correct distribution of moisture in the samples. Afterward, the soil samples are stored in sealed plastic bags to avoid moisture loss and conditioned for 12 h. The soil samples are mixed again before being used to mix any condensed water back into the soil.

The soil test specimens were prepared in accordance with standard ME-59/03 [53] in a miniature piece of equipment (MiniProctor, Didática Artigos Para Laboratório Ltda, Guarulhos, Brazil), which is, basically, a miniature version of a pile-driver used in construction (Figure 6a). The procedure to prepare one test specimen is as follows. A plastic circular sheet (Figure 6b—4) and, afterward, a metallic ring (Figure 6b—3) are inserted into the cup (Figure 6b—1), and 120 g of the adjusted moist soil is loaded into the chamber of MiniProctor lubricated with mineral oil. The loaded soil is then lightly compacted using a small piston, as in Figure 6c. After inserting another metallic ring, a plastic mold (Figure 6b) is positioned above the soil in order to create a depression for DPA application, and another plastic sheet is inserted in order to avoid any sticking of the soil in the MiniProctor’s parts. Twelve strikes are conducted with its hammer, and the test specimens are extracted with the aid of a mounted lever and then coated with paraffin at the sides and bottom. The prepared test specimens (Figure 6d) are stored for 60 h before use. After, the top portion of the test specimens are coated with distilled water in the proportion of 0.5 L/m² with the aid of a small pipette and left for 15 min to homogenize the applied water. The DPAs were applied at the top surface in a proportion of 1.2 L/m² and left for 72 h of curing time before checking the penetration of the mixtures. The test specimens were cracked vertically at their non-primed surface, and the degree of penetration of DPAs into the soil was measured in mm using a caliper at six different points for each sample.

3. Results

3.1. Pyrolysis Process

As it was initially commented, it is difficult to analyze and compare the pyrolysis processes conducted in semi-batch mode due to the need for parameter observation together with reaction time. The intrinsic dynamics of semi-batch pyrolysis means that an observed result has to take into consideration the reaction time where that event was observed. As an example, it can be considered the FFA content of the bio-oil (acid value) can be considered. While it is possible to obtain bio-oil with near zero acidity, this could be achieved only in later reaction stages, where most of the feed was already cracked and vaporized, meaning that overall acidity removal is low for the whole bio-oil. This was observed by Ferreira et al., who pyrolyzed waste animal fat in two-stage semi-batch catalytic cracking experiments and could obtain near-zero acidity bio-oil in the later stages of the reaction, where most of the feed already vaporized, presenting higher acidity and FFA content. Calculations revealed that the overall acid value of the pyrolysis oil was around 60 mg KOH/g of sample [5]. This particularity needs to be considered in all evaluations of the influence of the process parameters in reaction products (yields and chemical composition), and this even changes how to interpret some defined aspects of pyrolysis in semi-batch mode. For instance, we can cite how temperature is connected to the extent of the reaction (as it will be detailed further) and the impossibility of effectively controlling residence time in semi-batch pyrolysis. Since vapors flow out of the reaction zone constantly, by removing the driving force behind the reaction (heat), the residence time is low for the formed vapors, and high residence times only for obtained bio-oil in later stages of the reaction, not all of the feed, are observed. For semi-batch mode, it is possible to increase the overall residence time for all of the feed when some form of reflux exists that is capable of recycling the undesired unconverted material back to the reaction zone or when a secondary cracking zone is included (vapor-phase cracking), as suggested by Chang and Wan [14].

Even the analysis of product yields is further improved when considering reaction time. Lo and Tsai fractionated pyrolysis products of cottonseed oil pyrolysis according to reaction time and observed that most of the non-condensable gases are formed in the initial stages of the reaction. The gas fraction is rich in carbon dioxide, suggesting that decarboxylation and decarbonylation reactions occur at the beginning of the process [14].

Table 7. Product yields of WAF pyrolysis at 400–500 °C.

Product Yield	400 °C	450 °C	500 °C
Bio-oil (wt.%)	76.4	79.4	87.7
Char (wt.%)	4.2	6.6	5.2
Non-condensable gases (wt.%)	12.9	2.5	5.2
Water (wt.%)	6.5	11.5	1.9

and Figure 7 present the obtained yield of reaction products (bio-oil, bio-char, and non-condensable gases) for the three different temperatures (400, 450, and 500 °C). Since it is not possible to measure the weight of feed at each sample collection, and the non-condensable gases yield is obtained by mass balance, the obtained weight of the liquid phase collected and the observed temperature of the reactor are presented for each experiment (400, 450, and 500 °C) versus reaction time in the graphs of Figure 8, Figure 9 and Figure 10, respectively.

It can be observed that there are consistent data with other works concerning the pyrolysis of triglycerides and derivates, where bio-oil yields increase with higher temperatures [16,17,18,20,21,22]. As was mentioned, in semi-batch atmospheric pyrolysis with no inert gas flow, the heat applied (power) is the driving force behind the evolution of reaction and vaporization and the flowing of formed vapors. Since cracking reactions and vaporization are endothermic processes, and the reactions occur with no inert gas flow and atmospheric pressure, all the heat supplied is used to complete the reaction and vaporize the products of pyrolysis, and the temperature stalls until most of the reaction products are distillated out of the control volume of the reactor, as is shown in the graphs of Figure 8, Figure 9 and Figure 10. For all process temperature curves, it is observed that there is a tendency to maintain a temperature constant until most of the bio-oil is collected in the flash drum.

In order to make observations about the differences in conducting the pyrolysis of WAF at different temperatures, one might think that only the later stages of the reaction are different among different temperatures, but this is not the case. In order to avoid problems with temperature control due to the variability of research, the temperature is set manually at the control panel and no automatic control is conducted for the heating rate control. The heating rate is set indirectly by choosing the setpoint and comparing it with the current reactor temperature. The calculated difference generates the output of current to the electrical heater, as per any PID feedback control scheme [54]. In order to heat the reactor effectively, we found out through experimentation that an adequate heating rate is obtained by choosing a setpoint 20 °C higher than the current temperature, and the established protocol for differentiating the experiments by temperature is when cracking starts to occur with vapor formation. The setpoint was chosen as the designed temperature for the experiment. This can be observed in the graphs of Figure 8, Figure 9 and Figure 10, where the setpoint temperature is shown together with the reactor temperature and vapor flow. The greater the difference in temperature means a higher vapor flow out of the reactor, and Figure 8, Figure 9 and Figure 10 show increased vapor flow with an increasing temperature.

That is why it is observed to increase the bio-oil yield with an increase in temperature, but a similar trend is not observed for bio-char, NCG, and water yields. It seems that maintaining the reaction temperature at 400 °C favors the formation of non-condensable gases, while maintenance of the temperature at 450 °C increases the water formed in the reaction, probably by decarbonylation reactions. A further increase in temperature reduces bio-char yield once it starts to crack into more liquid and vapors, as shown by yields of 500 °C. It is not clear why the experiment conducted at 500 °C produced less water than all the others. It seems that aqueous phase yields are largely influenced by the moisture content of the feed, and the variability of this parameter may be the cause of inconsistent results.

A similar analysis can be conducted for chemical composition and for determining the physical–chemical properties of the obtained bio-oil fractions. As the samples were fractionated according to reaction time, the data are presented in the form of graphs (Figure 11, Figure 12 and Figure 13) of said physical–chemical parameter versus reaction time. In order to evaluate the overall influence of different pyrolysis temperatures, weighted averages of the measured properties are calculated based on the determined property of a time-fraction and its respective weight of bio-oil collected. The results of calculated density, kinematic viscosity, and acid value for each experiment are presented in

Table 8. Physical–chemical properties of WAF pyrolysis at 400–500 °C.

Property	400 °C	450 °C	500 °C
Density (g/cm3)	0.84	0.85	0.85
Kinematic viscosity (mm2/s)	5.67	7.81	9.02
Acid Value (mg KOH/g)	62	73	96

. The specific measurements of density, kinematic viscosity, and acid value for each time-fraction sample are presented in

Table 9. Physical–chemical properties of each time fraction (10, 20,30, and 40 min) of WAF pyrolysis bio-oil at 400–500 °C.

Experiment	Reaction Time (min)	Bio-Oil Weight (g)	Density (g/cm3)	Kinematic Viscosity (mm2/s)	Acid Value (mg KOH/g)
400 °C	10	65	0.85	8.1	135
	20	271	0.84	5.7	54
	30	111	0.83	3.6	7
	40	58	0.82	3.0	2
450 °C	10	294	0.86	9.8	131
	20	312	0.85	6.9	43
	30	176	0.85	6.3	34
	40	12	0.84	5.5	32
500 °C	10	207	0.86	11.1	154
	20	486	0.85	9.5	104
	30	120	0.84	5.4	19
	40	64	0.85	5.5	2

, along with the collected liquid phase for that reaction time.

It can be observed in Table 8 and Figure 11, Figure 12 and Figure 13 that all physical–chemical parameters reduced with reaction time, since higher reaction times allow for the transformation of triglycerides and fatty acids into hydrocarbons, displaying reduced density, viscosity, and acid value. Furthermore, as was visualized in Figure 8, Figure 9 and Figure 10, the experiments presented increased vapor flow, depending on the setpoint temperature, and consequently, there was less time for conversion of fatty acids to hydrocarbons [4,5]. The increased process temperature in semi-batch reactors affects negatively the residence time of vapors, which is of ultimate importance for the full conversion of fatty acids to hydrocarbons. In addition to the increased heat and vapor flow, this may allow for distillation of the original oil with no or only partial cracking, exhibiting a higher content of fatty acids, density, viscosity and acid value. It is observed from Table 8 that density, viscosity, and acid value presented higher values with increased process temperature, and the reaction tended to complete itself faster, with increased vapor flow, indicating a lower residence time of vapors for increased temperature and affecting the total transformation of the WAF into hydrocarbons. For semi-batch pyrolysis, in order to increase conversion, what is needed is not only temperature and heating rate but a way of maintaining the necessary residence time for conversion. This may be attained by reflux or the correct geometry of the reactor. Wiggers et al. conducted the pyrolysis of soybean oil in a continuous plant where the heated zone consisted of a long electrically heated tube and observed the opposite effect displayed here, with a reduction in the yield of bio-oil with an increasing temperature. This was achieved by feeding the reactor continually into an evaporator, and pyrolysis only occurred in the vapor phase along the heated tube. Since the variation in temperature exists only in the tube, it does not affect residence time. It is governed by the heat supplied by the evaporator and maintained at a constant temperature. When the feed rate of the oil is increased, there is a reduction in residence time, and the same effects observed in this work are repeated [23].

In the same manner, the chemical composition of obtained bio-oil is presented in two ways. Weighted averages of the area.% of the chemical functions observed in the chromatogram of each fraction (alkanes, alkenes, cyclic hydrocarbons, ketones, aldehydes, and fatty acids) are presented in

Table 10. Chemical composition of WAF pyrolysis bio-oil.

Chemical Function (area.%)	400 °C	450 °C	500 °C
Alkanes	41	31	26
Alkenes	40	35	26
Cyclic hydrocarbons	3	9	7
Fatty acids	7	19	35
Other oxygenates (ketones, aldehydes)	9	6	6

. The detailed chemical composition of bio-oil at different temperatures (400, 450, and 500 °C) and reaction times (10, 20, 30, and 40 min) are presented as Supplementary Material in the form of Tables S1–S11. The change in the chemical composition of bio-oil at different temperatures (400, 450, and 500 °C) is observed in Figure 14, Figure 15 and Figure 16, respectively. For simplicity of analysis, data are presented as separating the compounds in bio-oil by hydrocarbons and oxygenates, and the variation of the acid values of the samples are observed in the other axis. The acidity of triglyceride pyrolysis oil is connected to the fatty acid content of the sample, and since fatty acids are intermediate in the conversion of triglycerides into hydrocarbons by pyrolysis [14], it is a simple measure of the conversion, which serves as a good parameter to compare and validate the chemical composition change observed with the increase in reaction time.

As suggested by the tendency observed for density, viscosity, and acid value, it seems that by conducting pyrolysis under high temperature, in this case meaning under high energy transfer by heating and consequently high flow of vapors out of the reaction zone, there is a decrease in residence time in the heated zones and a reduction of the total conversion to hydrocarbons. The increase in liquid bio-oil yield is from distillated original material or only partially converted one, as shown in Table 10 by the increase in fatty acid concentration from 7 to 35 area.% of the chromatogram by increasing the temperature from 400 to 500 °C. Indeed, the experiment conducted at 400 °C generated a higher yield of gases (12.9 wt.%), and deoxygenation of pyrolysis oil (conversion of fatty acids to hydrocarbons) is related to the production of carbon dioxide and carbon monoxide by decarboxylation and decarbonylation reactions, respectively. The observed chemical composition may also explain why the experiment conducted at 450 °C generated more aqueous phase, as it presented a higher concentration of alkenes. Its production is related to decarbonylation reactions, where water is also formed [14].

Thermal decomposition of triglycerides and fatty acids occurs at mid-to-high temperatures (300–500 °C) and follows a complex reaction mechanism involving cracking reactions, deoxygenation, aromatization, and condensation reactions [14,55]. From a simplified perspective, the mechanism can be divided into three condensed steps: 1—initial cracking of triglycerides into ketenes, acrolein, and fatty acids; 2—fatty acids deoxygenation into hydrocarbons through decarboxylation and decarbonylation reactions; and 3—further reactions of aromatization and condensation of products [4]. Another pathway involves cracking in the double bonds present in the hydrocarbon’s chains of both triglycerides and fatty acids, producing smaller, more volatile, compounds of the parent initial chemical function and gaseous hydrocarbons [55]. It is not wrong to assume that both reaction pathways occur simultaneously, even if it is not to the same degree, so a wide distribution of chemical composition of the products is obtained for bio-oil, gas, and char fractions [4].

The chemical composition of bio-oil produced at 400 °C and 10 min of reaction time is presented in

Table 11. Chemical composition of WAF bio-oil at 400 °C and 10 min of reaction time.

Chemical Compound	Retention Time (min)	area.%
Alkanes	-	23.35
Heptane	4.504	0.650
Octane	6.803	0.874
Nonane	9.723	0.899
Decane	12.913	0.506
Undecane	16.107	0.730
Dodecane	19.180	1.036
Tridecane	22.100	2.506
Tetradecane	24.860	2.255
Pentadecane	27.454	6.816
Hexadecane	29.961	1.947
Heptadecane	32.416	5.106
Alkenes	-	32.68
1-Heptene	4.350	0.806
1-Octene	6.572	0.573
1-Nonene	9.457	0.550
1-Decene	12.633	0.614
1-Undecene	15.841	1.181
5-Undecene	16.243	1.405
2-Undecene, (E)-	16.553	0.654
1-Dodecene	18.937	1.036
4-Undecene, 3-methyl-, (Z)-	19.314	0.149
1-Tridecene	21.871	1.803
1-Tetradecene	24.654	2.696
1-Pentadecene	27.274	2.629
7-Hexadecene, (Z)-	29.596	1.228
Cetene	29.775	2.335
8-Heptadecene	32.036	5.181
1-Heptadecene	32.290	1.815
1-Octadecene	33.832	2.572
9-Hexacosene	53.020	5.454
Alkynes	-	4.524
13-Hexacosyne	52.535	4.524
Cyclic compounds		11.420
Cyclopentane, nonyl-	26.403	0.357
n-Nonylcyclohexane	29.112	0.744
Cyclotetracosane	48.767	10.319
Fatty acids	-	23.657
Propanoic acid	3.975	0.162
Butanoic acid	5.903	0.212
Pentanoic acid	8.654	0.416
Hexanoic acid	11.760	0.615
Heptanoic acid	14.895	0.776
Octanoic acid	17.979	1.265
Nonanoic acid	20.903	0.650
n-Decanoic acid	23.737	5.432
Tetradecanoic acid	33.451	2.250
n-Hexadecanoic acid	38.011	9.238
Oleic Acid	44.012	1.051
Ketones, Alcohols, Aldehydes	-	3.812
2H-Pyran-2-carboxaldehyde-3,4-dihydro-	13.967	0.381
2-Heptadecanone	36.536	1.596
Behenic alcohol	45.115	1.835
Non-identified compounds	-	0.581

in the form of the major compounds present, and it shows that bio-oil is majorly composed of 72 area.% of hydrocarbons (linear alkanes, alkenes, alkynes, and cyclic hydrocarbons), 23 area.% of fatty acids (deca-, tetra-, and hexadecanoic acids) and 4 area.% of ketones, alcohols, and aldehydes (heptadecanone, behenic alcohol, and a dimer of acrolein). Only a small fraction of compounds remained unidentified. It can be observed that a significant portion of the bio-oil is formed by linear compounds presenting 14 to 17 carbons with a high content of C15 and C17 hydrocarbons. Although vegetable oils and animal fats are different and present different physical properties, it can be argued that the majority of lipid-based materials are composed of triglycerides derived from palmitic (C16) and oleic (C18:1) acids, and their differences arise from the differences in the minority composition of other triglycerides and fatty acids. Even though animal fats display a higher content of stearic (C18) acid-based triglycerides, they still contain significant amounts of palmitic (27 wt.%) and oleic (48 wt.%) acids, corresponding to over 75 wt.% of their chemical composition [41]. As detailed before, triglycerides submitted to thermal decomposition are first converted to fatty acids and then deoxygenated to hydrocarbons, releasing carbon dioxide or carbon monoxide depending on the mechanism of deoxygenation (decarboxylation or decarbonylation), losing one carbon atom and turning into C15 and C17 hydrocarbons. The presence of smaller-chain hydrocarbons can be related to an analogous mechanism, with the other triglycerides composing the starting material, and also due to the cracking of double bonds in the hydrocarbon chain of fatty acids, as suggested by Benson et al. [55]. The presence of ketones can be explained by the mechanism of the decarboxylation of fatty acids. Although not fully explained, the academic literature suggests that ketones are intermediaries in the conversion of fatty acids to hydrocarbons through a mechanism of ketonic decarboxylation, as in the case of the conversion of adipic acid to cyclopentanone and cyclopentene [56].

It is important to observe that the area of the total ion count (TIC) of the chromatogram does not reflect the actual wt.% of the detected compounds in the sample, since different organic functions can present different responses in determined methods. The use of certified internal standards can be conducted in order to convert the area.% into wt.%, but the high number of detected compounds in the sample allied with the high value of some certified standards restricts its actual use. Some distortions can be observed, for example in the actual acidity of the WAF bio-oil. Even though the chromatogram shows 23 area.% for fatty acids, the actual acidity of the sample is far higher, since it presents an acid value of 135 mg KOH/g. It is estimated that, for oleic acid-derived triglycerides and fatty acids (the case of the WAF and WAF bio-oil), the acid value represents double the wt.% of fatty acids in the sample [41], so WAF bio-oil with an acid value of 135 mg KOH/g should contain approximately 60 wt.% of fatty acids. The obtained composition is in accord with other waste vegetable oil and animal fat pyrolysis results [4,16,17,57,58]. Bernar et al. obtained a very similar composition for grease-trap pyrolysis in the semi-pilot reactor at 400 °C [4], demonstrating that the initial wt.% of free fatty acids (FFA) of the feedstock only slightly influences the obtained FFA content of the pyrolysis oil due to the mechanism of conversion of triglycerides into hydrocarbons [14,55]. The waste fat used in this work has a much lower acid value of 2.6 mg KOH/g compared to the one used by Bernar et al. of 70 mg KOH/g. Indeed, the WAF used in this work is similar to beef tallow, largely composed of triglycerides and having low fatty acid concentration. Bernar et al. used waste fat from a kitchen grease trap, and even with characteristics defined as good for grease traps (the grease trap only collected effluent from meat cooking), it still suffers from biological (bacteria) and chemical degradation (hydrolysis), increasing its acid value [4].

3.2. Fractional Distillation

Batch fractional distillation of the WAF bio-oil produced was conducted in order to further purify the bio-oil into distilled fractions (Figure 17), with physical–chemical properties such as viscosity in the range of the commercial kerosene normally used for the preparation of DPA. The distillation temperature profile and the applied power in the heating mantle are presented in the graph of Figure 18, and it suggests the volatility of distilled fractions when compared to the distillation curve (Figure 19) collected for commercial kerosene.

Distillation was carried out under partial reflux in order to obtain better purification of products and separated according to the top distillation column temperature. In a multi-component mixture, it is not possible to separate all the individual compounds present, especially when the boiling point of the compounds being separated is near each other (less than 20 °C) [27]. As shown in Table 11, WAF bio-oil is composed of more than 20 different compounds, and some of them present close boiling points. It is more feasible to separate the mixture into a boiling-point range in order to obtain lighter and purified fractions. Distillation of bio-oils from the pyrolysis of triglycerides can be achieved using different boiling-point ranges [27,28,59,60]. Lima et al. pyrolyzed vegetable oils in a stainless steel 5 L reactor and distilled its liquid fractions into four temperature ranges, namely BP < 80 °C; 80 °C < BP < 140 °C; and 140 °C < BP < 200 °C, and a heavy fraction (BP > 200 °C) [59]. Anis et al. pyrolyzed waste cooking oil in a microwave-assisted reactor and distilled the bio-oil in a single-stage distillation, separating it into three fractions: gasoline (BP < 180 °C), kerosene (180 °C < BP < 230 °C), and diesel (230 °C < BP < 340 °C) [60]. Suota et al. separated the bio-oil obtained from the pyrolysis of waste cooking oil into two fractions: light (BP < 220 °C) and heavy (BP > 220 °C) fractions [28]. In a previous study, we pyrolyzed crude palm oil in a pilot plant and conducted fractional distillation of the bio-oil, separating it into three fractions, namely gasoline (BP < 175 °C), kerosene (175 °C < BP < 235 °C), and diesel (BP > 235 °C) [27], so the same temperatures were used for separation of the fractions of this work’s bio-oil.

The applied power in the heating mantle was adjusted in order to obtain a smooth and continuous flow of distillates to the separating funnel. With the aid of the reflux valve, it is possible to regulate the flow of distillation, and an increase in the applied power only increases the reflux rate if the valve remains unchanged, obtaining better separation but increasing process time [27]. High power in the heating mantle coupled with high reflux rates is not advisable though, since it can lead to temperature and vapor-flow increases in the condenser, possibly reducing liquid-fraction yields through losses in the vacuum adapter. With low applied power and a high reflux rate, better separation is achieved, and the temperature profile presented is a good indicator of the true boiling point of the compounds in bio-oil.

Table 12. Physical–chemical properties of bio-oil, distilled fractions, and commercial kerosene from WAF bio-oil at 400 °C and 10 min of reaction time.

Physical Property	Bio-Oil	Gasoline	Kerosene	Diesel	Commercial Kerosene
Yield of distillation (wt.%)	-	11.85	11.73	35.00	-
Density (g/cm3)	0.85	0.77	0.82	0.82	0.77
Kinematic viscosity (mm2/s)	8.10	0.66	1.78	3.77	1.43
Acid Value (mg KOH/g)	135.0	11.7	83.8	95.8	5.3

compiles the physical–chemical properties of WAF, bio-oil, and distilled fractions at 400 °C and 10 min of reaction time. It can be observed that fractional distillation further reduces the density and viscosity of bio-oil in the distilled fractions by separating it from the heavy fraction.

Table 13. Yields of distillation products from WAF bio-oil at 400 °C and 10 min of reaction time.

Yield (wt.%)	Value
NCG	7.33
Gasoline	11.85
Kerosene	11.73
Diesel	35.00
Bottoms	34.10

presents the yields of distilled fractions along non-condensable gases (NCG) and bottoms of distillation, representing the heavier fraction of bio-oil, corresponding to over 34 wt.%, and is probably responsible for a large part of the viscosity of 7.58 mm²/s of the bio-oil. When compared with the values presented for commercial kerosene, it can be seen that the kerosene fraction presents similar viscosity to commercial kerosene, even though it presents higher values of density. From the acid values, it can be seen that distillation could reduce the acidity of gasoline fraction but not of kerosene and diesel, showing that acidic compounds have boiling points inside the range of kerosene and diesel, impeding the use of these distilled fractions for fuel without better or further purification.

Since WAF bio-oil is rich in C15 and C17 compounds, a high amount of diesel and bottom fractions of almost 70 wt.% were obtained, with NCG, gasoline, and kerosene corresponding to 30 wt.% of the bio-oil. Ferreira et al. conducted fractional distillation under partial reflux for crude palm-oil bio-oil and found yields of 5.8, 13.7, and 12.0 wt.% for the NCG, gasoline, and kerosene fractions, respectively, when using a distillation column of 30 cm [27]. Even though crude palm oil is different than WAF and the pyrolysis was conducted in the presence of a catalyst (Na₂CO₃), both processes show similar fraction yields, since most oils and fats have similar composition. And, distillation yields are highly correlated to equipment and processing variables. Since similar conditions were used, like distillation column height and the use of partial reflux, similar yields were obtained. Similar physical properties were obtained for the different fractions. Ferreira et al. produced a gasoline fraction with 0.75 g/cm³ and 1.25 mm²/s, while the one produced in this work has a higher density of 0.77 g/cm³ but a lower viscosity of 0.66 mm²/s. Since WAF contains a higher proportion of saturated compounds, the differences in density and viscosity could be associated with the presence of a higher amount of saturated small volatile compounds in WAF gasoline than in palm bio-oil. The presence of small quantities of water in gasoline and kerosene fractions could also be related to higher densities when compared to the work of Ferreira et al., producing a systematic error in density measurement. Nevertheless, the kerosene and diesel fractions presented almost identical kinematic viscosity, and this parameter is important in the application of distillate fractions of waste oils and fats in the production of asphalt cutbacks. Kinematic viscosity is highly related to the capacity of the cutback to penetrate deeply into the base pavement layer and effectively create the prime coat and bind the asphalt cement layer [61,62]. When one is using waste as a feedstock of a determined chemical process, one should expect higher variability of composition and properties of the feed, and adequate characterization is paramount. Since crude palm bio-oil and WAF bio-oil can produce kerosene and diesel fractions of similar kinematic viscosity, reducing the variability of feedstock properties, it shows one advantage of upgrading waste oils to their bio-oils counterparts via pyrolysis process for this application.

The chemical composition of gasoline, kerosene, and diesel fractions of WAF bio-oil are presented in

Table 14. Chemical composition of WAF gasoline fraction of bio-oil at 400 °C and 10 min of reaction time.

Chemical Compound	Retention Time (min)	area.%
Alkanes	-	35.62
Nonane	9.700	1.197
Decane	12.893	1.222
Undecane	16.081	2.426
Dodecane	19.156	3.857
Tridecane	22.076	7.108
Tetradecane	24.830	5.267
Pentadecane	27.431	9.420
Hexadecane	29.932	2.284
Heptadecane	32.396	2.840
Alkenes	-	39.22
1-Decene	12.613	1.575
1-Undecene	15.816	3.406
2-Undecene, (E)-	16.219	4.165
2-Undecene, (Z)-	16.530	2.065
1-Dodecene	18.914	3.787
1-Tridecene	21.852	6.018
1-Tetradecene	24.627	7.682
1-Pentadecene	27.249	4.802
Cetene	29.748	2.969
8-Heptadecene, (E)-	32.013	1.324
8-Heptadecene, (Z)-	32.270	1.424
Cyclic compounds		8.20
9-Oxabicyclo[6.1.0]nonane	24.630	6.572
1-Nonylcycloheptane	32.102	1.629
Fatty acids	-	21.03
Butanoic acid	5.989	2.733
Hexanoic acid	11.824	4.803
Heptanoic acid	14.919	3.871
Octanoic acid	17.958	3.479
Nonanoic acid	20.855	1.272
n-Decanoic acid	23.673	4.868
Ketones, Alcohols, Aldehydes	-	1.537
2H-Pyran-2-carboxaldehyde-3,4-dihydro-	13.967	1.537

,

Table 15. Chemical composition of WAF kerosene fraction of bio-oil at 400 °C and 10 min of reaction time.

Chemical Compound	Retention Time (min)	area.%
Alkanes	-	33.97
Undecane	16.081	1.681
Dodecane	19.156	2.906
Tridecane	22.076	6.693
Tetradecane	24.830	5.419
Pentadecane	27.431	11.646
Hexadecane	29.932	2.168
Heptadecane	32.396	3.454
Alkenes	-	35.23
1-Undecene	15.818	2.420
2-Undecene, (E)-	16.219	2.898
2-Undecene, (Z)-	16.535	1.679
1-Dodecene	18.915	2.927
1-Tridecene	21.853	4.958
1-Tetradecene	24.630	6.572
1-Pentadecene, (E)-	27.106	1.731
1-Pentadecene, (Z)-	27.248	4.422
Z-8-Hexadecene	29.561	1.521
Cetene	29.744	2.492
8-Heptadecene	32.016	1.977
2-Methyl-Z-7-hexadecene	32.102	1.629
Fatty acids	-	30.81
Hexanoic acid	11.817	2.513
Heptanoic acid	14.947	3.794
Octanoic acid	18.022	6.817
Nonanoic acid	20.904	3.683
n-Decanoic acid	23.743	14.001

and

Table 16. Chemical composition of WAF diesel fraction of bio-oil at 400 °C and 10 min of reaction time.

Chemical Compound	Retention Time (min)	area.%
Alkanes	-	43.72
Undecane	16.081	1.540
Dodecane	19.156	2.437
Tridecane	22.076	4.258
Tetradecane	24.830	4.934
Pentadecane	27.431	12.689
Hexadecane	29.932	5.295
Heptadecane	32.396	10.988
Octadecane	34.244	1.574
Alkenes	-	39.25
1-Undecene	15.842	1.513
1-Dodecene	18.930	2.063
1-Tridecene	21.871	3.495
1-Tetradecene	24.649	4.666
1-Pentadecene	27.266	5.698
Z-8-Hexadecene	29.579	2.161
Cetene	29.771	4.757
8-Heptadecene	32.033	3.042
2-Methyl-Z-7-hexadecene	32.117	3.193
2-Methyl-E-7-hexadecene	32.286	3.835
8-Heptadecene	32.482	1.538
E-7-Octadecene	33.925	1.355
Z-7-Octadecene	34.124	1.931
Fatty acids	-	13.05
Nonanoic acid	20.878	1.356
n-Decanoic acid	23.731	9.348
Tetradecanoic acid	33.439	2.332
Ketones, Alcohols, Aldehydes	-	4.01
2-Heptadecanone	36.528	4.005

, respectively. As was commented on in Section 2.4, these results are from bio-oil obtained at 400 °C and 10 min of reaction time. It is possible to observe that distilled fractions are enriched in hydrocarbons, and acid content is minimized when WAF bio-oil is distillated. Maybe more important is the reduction in high boiling-point compounds that are left in the boiling flask as a heavy oil. The area of the chromatogram of fatty acids went from 23% to 21, 31, and 13% for the gasoline, kerosene, and diesel fractions, respectively. Strangely enough, the acid value of the distilled fractions shows the opposite direction, with a lower acid value for gasoline and a higher for diesel. This difference can be explained by the nature of the fatty acids present for each fraction. Gasoline and kerosene are rich in lower molecular weight fatty acids (C5–C9 fatty acids), while diesel contains higher molecular weight fatty acids, such as decanoic and tetradecanoic acid. Nevertheless, all distilled fractions contain decanoic acid as the major fatty acid present.

The gasoline fraction presented an 83 area.% of hydrocarbons, and it is the only fraction containing cyclic hydrocarbons. Indeed, cyclic hydrocarbons usually present boiling points in the gasoline range, and fractional distillation was capable of separating all cyclic compounds formed in the gasoline fraction. It can be observed that the enrichment of lower boiling-point compounds, such as C8–C12 hydrocarbons, improves the volatility and reduces the viscosity of the WAF gasoline. Nevertheless, the compound with a higher area.% is pentadecane in all three distilled fractions.

The kerosene fraction presented a higher area.% of fatty acids, showing that a majority of the fatty acids present in WAF bio-oil are in the kerosene boiling-point range (175 < BP < 235 °C). Distilled fractions containing high amounts of fatty acids and high acid values should present an interesting solvent in the preparation of asphalt cutbacks. Moreover, conventional MC-30 asphalt cutbacks are usually prepared with petroleum kerosene [61], so a solvent with physical–chemical properties similar to petroleum kerosene should produce the best results with regard to penetrative ability and curing time. The obtention of a bio-solvent similar to kerosene but with a high acid value (rich in fatty acids) is an interesting way to check if the acidity of the solvent could improve or deteriorate the quality of the prime coat applied using the bio-solvent.

The diesel fraction is shown to contain the higher boiling-point compounds, and as was shown in Table 13, it presents the majority of compounds produced in WAF pyrolysis, together with the bottom distillation fraction. It is the fraction with larger amounts of heptadecane and the only fraction containing octadecane. The physical–chemical properties of this diesel fraction and its chemical composition reveal that diesel presents similar properties to the kerosene fraction and is a light diesel fraction, rich in compounds near the beginning of the distillation cut (BP > 235 °C). Indeed, the distillation curve of fractional distillation shows that the majority of the diesel compounds were collected below 270 °C, and the diesel fraction is a good candidate for a diluent in the preparation of asphalt cutbacks. It was possible to separate 2-heptadecanone from the other fractions, and diesel was the only fraction where this compound was detected.

3.3. Application in Asphalt Cutback

The prepared asphalt cutbacks were characterized according to Saybolt–Furol viscosity, dynamic viscosity, and flash points before conducting the priming assays. The Saybolt–Furol viscosities of commercial kerosene and their mixtures (10 to 40%) of distilled fractions are presented in the form of a graph in Figure 20. Commercial kerosene presented a standard value of 150 s for its viscosity and 59.5, 141.5, and 136 for pure gasoline, kerosene, and diesel, respectively. Except for the gasoline fraction, kerosene and diesel presented viscosities inside the range of 75–150 s, as defined in the DNER standard for DPAs [47,48,49,50,51,52]. As for the mixtures, a synergic effect is observed, where the mixtures presented lower values than the pure DPAs. It was expected that mixtures would present values between the ones presented by commercial kerosene and the pure distilled fractions.

Analogous to the way Saybolt–Furol viscosities are presented, the kinematic viscosity determined from the dynamic viscosity obtained with the rheometer and calculated considering the average density of samples is presented in Figure 21, and the flash points are presented in Figure 22. The same synergic effect was observed for the kinematic viscosity of the mixtures. The kinematic viscosity of standard DPA (prepared with commercial kerosene) was 45.6 mm²/s, and except for diesel, only the mixtures prepared with pure or almost pure (30%) kerosene presented values inside the kinematic viscosity range of standard DPAs. The diesel mixtures all presented values inside the range, showing that they could be used as a full substitute for commercial kerosene in the preparation of DPAs. Indeed, the graph in Figure 18 shows that the diesel fraction was obtained near boiling points of lighter fractions, from 235 to 260 °C, and it corresponds to a lighter fraction of diesel, similar to commercial kerosene. Pure gasoline presented a kinematic viscosity of 70 mm²/s, and this could indicate that gasoline was not capable of adequately dissolving and diluting the petroleum asphalt.

Flash point is a measurement of the volatility and flammability of liquids, and a minimum value of 38 °C is necessary for the adequate evaporation of the DPA solvent when conducting the soil priming. Lower flash points could be dangerous when applying the mixtures due to fire hazards. Even though the standard does not establish a maximum value, high flash points mean that the diluent is of insufficient volatility, and priming could take longer or never happen. In order to avoid that distilled fractions DPAs with a similar volatility to the standard DPA (61 °C) should be used. It can be seen from Figure 22 that an increase in the gasoline fraction tends to lower flash-point values, since a high degree of volatility is found for gasoline. Meanwhile, for kerosene and diesel, an increase in distillate substitution increases the flash points. Nevertheless, even with the use of pure kerosene and diesel fractions, the flash point is still near the flash point presented by the standard, of 64 and 77.6 °C, respectively.

3.4. Priming Assays

Since some mixtures of commercial kerosene and WAF bio-oil distillate fractions presented synergic effects for the measurements made and could not meet the standard of viscosity, it was decided that only the pure distillate fractions would be used in the priming assays. The results of the DPA’s penetration (mm) in the soil are presented in the graph of Figure 23 as functions of the moisture content of the soil. The priming assays with different distillates were also used to evaluate the influence of the acid value of the samples at penetration depth. As shown in Table 12, distillates have different acid values, with 12 for gasoline, 84 for kerosene, and 96 mg KOH/g for diesel distillates, against near-zero acidity in the case of commercial kerosene (5 mg KOH/g).

It can be observed that there are similar behaviors for commercial and pyrolysis kerosene and increasing penetration with the increase of moisture levels in the soil. For commercial kerosene, values between 4.61 and 5.50 mm were observed. It is important to note that penetration was higher for pyrolysis kerosene, indicating the superior performance of the application. As observed for gasoline and diesel, an inferior penetration is observed, probably due to the higher volatility of gasoline, curing the DPA in less time and, consequently, reducing soil penetration. The diesel samples showed lower penetration values due to the presence of long-chain compounds, exemplified by the higher boiling point of diesel and increased viscosity when compared with the other distillates. This increased viscosity should interfere with the penetration of petroleum asphalt into the soil, especially during curing time. It is important to note though that all distillates presented adequate penetration values when compared with the academic literature [63,64]. Zhang et al. evaluated the penetration of traditional emulsified asphalt and found a value of 4.2 mm after 72 h [63]. Ouyang et al. evaluated the penetration in the soil of asphalt emulsions and found values in the range of 3.85–7.76 mm [64]. These results show that all distillate fractions could be used as substitutes for commercial kerosene in the preparation of DPA when considering only the penetration depth in laterite soils.

When considering the acid value of the samples, no difference in curing time was observed, with all prime coats cured in the specified time of the method. At penetration depth, as was already discussed, the penetration depth of pyrolysis kerosene was higher than commercial kerosene, and the synergistic effect observed when mixing commercial kerosene with pyrolysis distillates could be related to the fatty acid content of pyrolysis oil. Elkadri et al. studied the influence of different emulsifiers in prime coat systems and showed that cationic emulsifiers produced better results for penetration depths due to superior mixing and the consequent reduction of the droplet sizes in the emulsion [43]. While emulsion prime systems are different than cutbacks, a possible interaction of fatty acids and hydrocarbons present in kerosene could explain why mixtures of commercial kerosene and pyrolysis oil show reduced viscosity beyond even the separate mixtures. The penetration depths also show a different trend when compared to the work of Barroso et al., which showed a decreasing penetration depth with the increase in soil moisture [65]. Meanwhile, for most samples containing pyrolysis oil, the penetration depth increased with the moisture levels. The penetration depths in the test specimens are related to water surface tension, which is related to the physical–chemical properties and composition of the tested soil. Different soils generate different dynamics of diffusion of the cutback into the base layer, generating different results of penetration depending on the moisture level. The work of Mantilla and Button, explaining the effect of moisture levels on the penetration depths of prime coat systems, shows practically no effect of moisture levels in the penetration of standard MC-30 cutback, pointing once more to the variability of this type of data [61].

The asphalt cutback prepared with pyrolysis oil kerosene presented a higher penetration depth of between 8 and 14% of soil moisture when compared to commercial kerosene, its closest in kinematic viscosity and volatility. In Table 12, the kinematic viscosity of pyrolysis kerosene is 1.78 mm²/s, and for commercial kerosene, it is 1.43 mm²/s. Meanwhile, the flash-point measurement (Figure 19) shows flash points of around 60–65 °C for both solvents. The higher penetration in intermediate values of moisture could be related to the fatty acid content of pyrolysis kerosene. Fatty acids are known emulsifiers due to their mixed polar and non-polar nature [41]. The carboxylic group of fatty acids may interact with water in the soil and change its surface tension, improving penetration. Even though they display acidity, gasoline and diesel produced prime coat systems with less penetration than commercial or pyrolysis kerosene. This could be related to the low acid value of gasoline and added volatility, reducing penetration once the cutback cures faster. In the case of diesel, it shows the highest acidity of all the solvents but also increased viscosity (3.77 mm²/s), reducing penetration. Nevertheless, all solvents produced prime coats capable of penetrating 4–5 mm deep into the base layer, which is considered acceptable for the application of asphalt cutbacks.

4. Discussion

The problem with conducting process analysis of pyrolysis of fatty wastes in semi-batch reactors is that a steady-state operation is never achieved, and there is a constant change of chemical composition of the feed and produced vapors with reaction time. In order to effectively analyze the pyrolysis in semi-batch reactors, one needs to draw samples according to reaction time and measure the physical–chemical properties and chemical composition of each fraction [4,5]. In this work, we conducted pyrolysis of waste animal fats in a one-stage non-catalytic semi-batch stirred reactor at different temperatures (400–500 °C), fractionating the obtained liquid fraction according to reaction time (every 10 min). This was conducted in order to understand how the process can be controlled and improved.

The obtained results revealed that the temperatures chosen as setpoints actually represent different rates of reaction. As soon as the reactor reaches the moment where the feed starts to pyrolyze, it is rather difficult to increase the reactor temperatures, since lipid pyrolysis and vaporization are endothermic processes. Most of the energy is directed to the cracking or vaporization of formed vapors, and higher setpoint temperatures mean a higher flow of vapors out of the reaction zone, reducing residence time and lowering the conversion of fatty acids to hydrocarbons. Liquid bio-oil yield increased with temperature, but physical–chemical properties and chemical composition were negatively affected by temperature increase. The experiment conducted at 400 °C produced the best results when considering the quality of the bio-oil (0.84 g/cm³, 5.67 mm²/s, and 62 mg KOH/g), generating a modest amount of non-condensable gases (13 wt.%), which are associated with decarboxylation and decarbonylation reactions [14]. Chemical composition presented 84 area.% of the hydrocarbons, mostly containing C15 to C17 hydrocarbons.

This reveals the difficulty and care needed to analyze pyrolysis in semi-batch reactors, leading to false conclusions and directions. Temperature control is key to obtaining bio-oils rich in hydrocarbons, but if it is accompanied by a reduction in residence time, it could mean lower conversion. The same can be said of the utilization of inert carrier gases that improve conversion by oxygen elimination but could also lower the residence time to obtain an adequate conversion. This knowledge can be used to improve the design of semi-batch or even continuous processes, much like the reactor designed by Wiggers et al. [23], where vapor-phase cracking is used to improve the conversion of fatty acids to hydrocarbons. Other means could also be used, such as partial reflux (dephlegmators) or pressurized reactors to achieve better conversion.

Fractional distillation was capable of partially refining the 400 °C bio-oil, improving the volatility, density, and viscosity of the obtained distilled fractions (gasoline, kerosene, and diesel) but could not reduce fatty acid content and fractions still contained considerable acidity, especially in kerosene and diesel cuts. Chemical composition showed the presence of fatty acids of a wide dispersion of carbon number, mainly decanoic, tetradecanoic, hexadecanoic, and oleic acids, presenting boiling points in the kerosene and diesel range. Fractional distillation under reflux can still be improved by choosing different column heights and packings and also different reflux rates. A “tailored” cut regime can also be adopted in batch distillation in order to produce fractions with a minimized acid value, and these studies are the subject of upcoming research.

Fractional distillation is a powerful separation technique, but it demands careful control of some parameters, such as the heating rate applied, column height, reflux ratio, condenser temperature, and even adequate insulation of the column. Unfortunately, most works where pyrolysis oil is distillated provide very little information about the distillation process. In this work, precise information was supplied about some of the parameters, such as the power applied and the temperatures of the bottoms and tops of the column in the presence of partial reflux. Even then, more information is still needed, such as the actual reflux ratio, demanding that the vapor flow rate be measured.

Kerosene and diesel fractions proved to be effective for the preparation of asphalt cutback, showing that acid value does not heavily influence the process, and volatility and viscosity are keys to the correct penetration depth and curing time. Kerosene and diesel fractions could produce asphalt cutbacks complying with the defined road standard [47,48,49,50,51,52].

In pavement preparation for asphalt application, it is more common to use stabilized water emulsions to prime the base layers of the pavement, due to the inherent pollution of asphalt-cutback application, since the solvents used release VOC into the atmosphere and many countries have banned its application. Nevertheless, the application of bio-solvents obtained from WAF could be successfully applied as a conventional asphalt cutback, exhibiting the same characteristics as a standard prepared with commercial kerosene, despite its high fatty acid content, which could reduce the VOC release.