13th International Colloquium Fuels - September 2021 65 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale Tim Schulzke Fraunhofer UMSICHT, Institut für Umwelt-, Sicherheits- und Energietechnik, Oberhausen, Germany Summary Within the framework of the PyroMar collaborative project about the production of low-sulphur bio-based blendstocks for marine fuels, the esterification of condensates resulting from ablative fast pyrolysis of 2 nd generation biomass with higher alcohols derived from lignocellulosic ethanol is investigated. To examine the scale-up of the chemical process three different experimental setups are used: a 3-neck-flask of 250 ml nominal volume, a 3-neck-flask of 2 l and a stirred tank reactor of 20 l. In all setups identical reaction mixtures (relative amounts of pyrolysis oil, butanol and solid acid catalyst) under the same conditions (external heating, operation under reflux, reaction end point indication) were converted and the product mixtures were analyzed. While the smallest setup needed only about 6 hours of operation to reach the end point, the two larger setups required 20 hours. The remaining water mass fraction in all experiments regardless of size and original water content was around 1 wt.-% and the acetic acid mass fraction could be reduced to half the value of the reaction mixture. Further scale-up to cubic metre scale for industrial production seems easily achievable. 1. Introduction The marine sector contributes significantly to global greenhouse gas emissions by burning fossil based residual oil from refineries with low potential for electrification. In addition, these marine fuels traditionally contained high amounts of sulphur resulting in SO x emissions. In recent years, the sulphur content in fuels is increasingly strictly regulated - starting in emission control areas and being extended to the high sea outside. The existing Emission Control Areas (ECA) are the Baltic Sea, the English Channel, the North Sea and coastal areas of United States of America and Canada except for Alaska, Yukon, Northwest Territories, Nunavut and the north-eastern coast of Quebec. Enlargement of ECAs is in discussion for Mexican, Japanese, Norwegian and (southern parts of) Malaysian coast as well as the Mediterranean Sea. The fuel sulphur limit within ECAs is 0.1 wt.-% since 2015 and outside ECAs 0.5 wt.-% since 2020 for vessels without exhaust gas cleaning systems (“scrubbers”) [1]. This leads to the idea to apply biomass-based fuels to reduce the emission of greenhouse gases by a renewable liquid energy carrier that is inherently low in sulphur. A promising candidate for that purpose is fast pyrolysis oil, a liquid that is obtained from condensation of vapours resulting from thermal degradation of solid biomass applying high heating rates at around 550 °C and short vapour residence times in the hot area [2]. The advantages of fast pyrolysis are the relatively simple process and the comparatively high liquid yield of up to even 70 wt.-%. Another important point is the possibility to process 2 nd generation biomass with no food or fodder competition. On the other hand there are some drawbacks: as the input material is composed of oxygen-containing natural polymers such as cellulose, hemicellulose (both hydrocarbons) and lignin, also the products from thermal degradation still include many oxygen-containing functional groups and water is formed, resulting in an overall low heating value [3]. Some of the oxygen-containing compounds are organic acids, aldehydes and ketones. While the acids lead to high Carbon Acid Numbers (CAN) and hence cause severe corrosion on standard material applied in pumps, boilers and engines, the aldehydes and ketones are very reactive, especially under acidic conditions, and result in products with increasing molecular weight even at room temperature over extended storage time, thus leading to increased viscosity. High water concentrations in the condensates, typical for herbaceous biomass like cereal straw, lead to spontaneous phase separation into an aqueous phase with high organic contamination and a very viscous, tarry phase, which still dissolves noticeable amount of water. Due to the oxygen-containing functional groups in the organic molecules forming the liquid product it is quite polar and therefore not miscible with hydrocarbons. In particular in the beginning of an application of such pyrolysis oils as fuels in shipping, its availability would be scarce and restricted to a small number of harbours worldwide, why a separate tank system would be non-economical. 66 13th International Colloquium Fuels - September 2021 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale One possible way to overcome these undesirable properties would be the chemical removal of the oxygen by hydrogenation. This would result in pure hydrocarbons, which are indistinguishable from petroleum-based ones and fully miscible on one hand and on the other hand water forms a second phase that can easily be separated. Such approach is investigated by several groups, e.g. by the BioMates consortium [4]. Unfortunately, this process requires a large amount of renewable hydrogen to remove the oxygen in form of water and quite sophisticated catalysts. An alternative is to chemically mask the compounds that mainly provoke the undesired properties [5]. Luckily all three group of compounds - carboxylic acids, aldehydes and ketones - react with alcohols following the reaction schemes shown in Fig. 1 for acids and in Fig. 2 for aldehydes and ketones. Figure 1: Reaction scheme of esterification Figure 2: Reaction scheme of acetalization The reactions are executed with an acidic catalyst under boiling conditions. The evolving vapour consists of an azeotrope made of water and the alcohol. To avoid the reaction mixture to deplete in alcohol, the vapour is condensed and returned to the reactor. Both reactions, esterification as well as acetalization, are equilibrium reactions. The formed reaction water together with the already present water in the pyrolysis oil prevent the reactions to fully proceed to the product side. To push the reaction towards the products, a very large overspill of alcohol would be necessary, which is expensive and could not be recovered after finishing the reaction, as distillation in presence of water would lead to significant rate of backwards reaction, thus cleaving the esters, acetals and ketals. A better way to shift the equilibrium towards the products would be the continuous removal of water from the reaction mixture. This can be achieved by the application of alcohols with a longer carbon chain. Methanol, ethanol and propanols are all fully miscible with water, but beginning with butanols, higher alcohols exhibit a miscibility gap in the liquid phase and the azeotropic composition always fall into it. By that means the evaporating water can easily be separated from the higher alcohol in a Dean-Stark-Apparatus (Fig. 4 in the middle): the denser water is collected in the lower part of the burette and removed periodically while the lighter alcohol flows back to the reactor. Butanol typically is a product of petroleum refining, but it can be efficiently produced by catalytic condensation from ethanol as developed by Fraunhofer UMSICHT [6]. And if cellulosic ethanol is used for this catalytic condensation instead of starchor sugar-based ethanol, the resulting butanol is as sustainable as the pyrolysis oil. Putting all this together leads to the overall process approach, which is investigated within the framework of the PyroMar collaborative project, which started January 1 st , 2020. The concept is given in Fig. 3. Figure 3: PyroMar concept for low sulphur marine fuels The first line of action within the overall concept starts with cereal straw. After enzymatic decomposition it is fermented to ethanol, which serves as feedstock for the catalytic condensation at Fraunhofer UMSICHT to butanol. In the second line of action a wide variety of 2 nd generation biomass like straw, autumn leaves, landscaping material, rice husk, decomposed and dried sewage sludge is converted to pyrolysis oils applying an ablative fast pyrolysis plant in laboratory scale with a capacity of about 5 kg/ h biomass feed. The pyrolysis oil and the butanol are mixed and used as feedstock for the esterification and acetalization with a solid acid as catalyst. This catalyst is recovered by filtration after the reaction is completed. The project partner LKV (Chair of piston machines and combustion engines) at Rostock University than analyse the reaction products prior to do blend tests with conventional marine fuels. In the end engine tests are envisaged for a suitable blend. The project is round off by ecological and economic studies performed by ifeu (Institute for Energy and Environment Research) in Heidelberg. In the following sections the process step of esterification and acetalization is further described with a strong focus on scale-up based on initial results. 13th International Colloquium Fuels - September 2021 67 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale 2. Experimental Setups Fraunhofer UMSICHT started its work on pyrolysis oil upgrading by catalytic treatment with higher alcohols in the framework of a collaborative project named “Development of a combined process for the upgrading of pyrolysis oil with biogenic alcohols and hydrogen” together with Thuenen-Institute for Wood Research in Hamburg, funded by the German Federal Ministry for Food, Agriculture and Consumer Protection (grant nr. 22018811). The focus in this project lay on intensive catalyst screening and the identification of favourable process conditions. First results were presented during ECI’s conference “Biorefinery I: Chemicals and Materials from Thermo- Chemical Biomass Conversion and Related Processes” in Crete 2015 [7]. Heterogeneous solid acid catalysts are easy to be removed after the reaction in contrast to homogeneous liquid acids like sulphuric acid or p-toluenesulfonic acid, which stay in the product mixture and need to be neutralized, e.g. with caustic soda, thus resulting in a salt residue not tolerable in engine application. From all the tested solid acids, Amberlyst70™, a polystyrene backbone with sulfonic acid functional groups, performed best with lowest water content and lowest TAN after reaction. As that catalyst was no longer available, it was decided to use a similar one from the same group Amberlyst36™ ever since. All these experiments and the majority of the subsequent ones performed in the following years were conducted in a 250 ml 3-neck-flask immersed in a heated oil bath for temperature control shown in Fig. 4 (left photograph). For the engine tests of potential blend fuels at the partner LKV more than 100 l of PyroMar intermediates (esterified pyrolysis oils) are necessary to prepare the blend fuels. Such an amount cannot be supplied from laboratory setup with 250 ml flask and a typical batch amount of 160 ml liquid feedstock. Figure 4: Small (250 ml, left) and medium scale (2 l, right) setup with Dean-Stark-Apparatus in the middle Therefore one focus of the first two years in the Pyro- Mar project is the scale-up of the esterification process to a batch reactor with reasonable size. As about 10 is a typical scale-up factor in chemical engineering, the next size of experimental setup is a 2 l 3-neck-flask (as 2.5 l flasks are not available), also immersed in a heated oil bath for temperature control, shown in Fig. 4 (right photograph). The Dean-Stark-Apparatus for water separation is the same as for the 250 ml setup (Fig. 4, middle). In this larger setup a half-moon impeller is used for stirring, hence the magnetic stir bar can be used to agitate the oil bath. The oil bath temperature is measured via PT100 temperature sensor and controlled by the heater below the oil bath. A small flow of nitrogen is passed through the reactors to inertize the atmosphere in the respective flask. A thermocouple is dunked in the reaction mixture to monitor the temperature progress over time. From that size a real scale-up factor of 10 is achieved by using a 20 l stirred tank reactor with a heating jacket and two impellers stacked on a central shaft, shown in Fig. 5. The thermal oil flowing through the heating jacket is controlled by a thermostat and the reactor is inertized by a small flow of nitrogen as well. A PT100 temperature sensor is also installed to monitor the liquid’s temperature over time. The condensate coming from the top cooler is caught in a Dean-Stark-Apparatus similar to that used in the smaller setups. Here, a capacitive level switch is installed to automatically detect the phase boundary between water and alcohol. Whenever this boundary rises to the switch position, it triggers a solenoid valve, which stays open for a fixed time. The flow is regulated by a needle valve, so that a fixed amount of about 21 ml of water is removed per opening period of the solenoid valve. Figure 5: Large scale (20 l) setup All experiments start with filling the reactors at room temperature with the respective amounts of pyrolysis oil, butanol and solid catalyst. Then the nitrogen flow is started as well as the flow of cooling water to the condenser and the oil bath or heating jacket is set to a temperature between 125 °C and 140 °C. The temperature of the reaction mixture rises fast until it reaches the boiling point at about 97 °C. Then the reflux starts and water evolves in the Dean-Stark-Apparatus from which it is removed 68 13th International Colloquium Fuels - September 2021 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale periodically either manually for the smaller setups or automatically at large scale. With diminishing water fraction in the reaction mixture over time the temperature slowly increases. The reaction mixture is kept boiling until no more water evolves in the Dean-Stark-Apparatus. While for the smallest setup this is achieved within about 6 hours, the other setups need up to 20 hours. As the setups arte not designed for unattended operation, they have to be shut down over night. For that, the heating plate in the medium sized setup is switched Table 1: Communalities and differences of experimental setups 250 ml 2 l 20 l Reactor 3-neck flask 3-neck flask stirred tank Stirrer magnetic bar halfmoon impeller 2 impellers stacked on shaft Heating system oil bath (unstirred) oil bath (stirred) heating jacket Catalyst Amberlyst™36 5 g 40 g 400 g Batch volume 160 ml 1,280 ml 12,800 ml Pyrolysis oil 80 ml / 100 ml / 120 ml 640 ml / 800 ml / 960 ml 6,400 ml / 8,000 ml / 9,600 ml 1-Butanol 80 ml / 60 ml / 40 ml 640 ml / 480 ml / 320 ml 6,400 ml / 4,800 ml / 3,200 ml Oil/ Alcohol-ratio 1: 1 / 5: 3 / 3: 1 1: 1 / 5: 3 / 3: 1 1: 1 / 5: 3 / 3: 1 Water separation DSA ≈ 20 ml, manually DSA ≈ 20 ml, manually DSA ≈ 50 ml, automatically off and in the large scale setup the setpoint for the oil jacket is reduced to 50 °C. In the medium sized setup the stirrer is switched off as well, while in the large scale it stays on. The next day the heating system is set back to the original setpoint and the reflux starts after heating up exactly at the same boiling temperature as the experiment was interrupted the day before. Table 1 lists the characteristics of the three experimental setups for direct comparison. 3. Results Fig. 6 shows the temperature profile for exemplary experiments in small and medium scale while Fig. 7 depicts the temperature profile for the large scale setup. While in the smallest setup the boiling point of the reaction mixture is directly rising, the boiling temperature of in the medium sized setup stays longer around the boiling point of the water-butanol-azeotrope and rises slower. Although the batch volume increased by a factor of 8, the necessary reaction time increased by a factor of less than 4 only (both experiments were performed with the same pyrolysis oil and the same oil/ alcohol-ratio). The further magnification of batch volume by a factor of 10 to the largest setup does not make the necessary reaction time for full water removal increase any further (this experiment had to be performed with another pyrolysis oil of similar composition with the same oil/ alcohol-ratio). Figure 6: Reaction mixture temperature profile over time for small and medium scale setup Fig. 7 additionally shows the profile of the removed water over reaction time. It can be seen clearly that at the beginning the slope is nearly linear and quite steep and after approximately 300 min it flattens more and more. At the end of the experimental run the time between the last two opening periods of the solenoid valve accounted for exactly 4 hours. Figure 7: Reaction mixture temperature and separated water over time for large scale setup 13th International Colloquium Fuels - September 2021 69 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale At the submission deadline for this contribution to the conference proceedings not all results from laboratory analyses were at hand. Therefore, in the following only preliminary results obtained for the small and medium scale experimental setups are presented. Figure 8: Water mass fraction for different experiments Fig. 8 depicts the water mass fraction for several groups of experiments conducted with three different pyrolysis oils. All pyrolysis were based on a mixture of wheat and barley straw at equal share. For the experiment Pytec296-301 the pyrolysis temperature was set to 550 °C and the temperature of the first condenser and electrostatic precipitator to 75 °C. The experiment Pytec302-305 was done with a pyrolysis temperature of 500 °C and a condensation temperature of 70 °C. The last pyrolysis experiment Pytec307-308 was conducted at 450 °C and 70 °C, respectively. As a result of these differing operating conditions in pyrolysis and condensation, the water mass faction of the oils was 13 wt.-% for Pytec296-301, 22 wt.-% for Pytec302-305 and 31 wt.-% for Pytec307-308. After mixing with butanol, the water mass fraction of the reaction mixture is significantly lower due to physical dilution as can be seen clearly from Fig. 8. But regardless of original water mass fraction of the pyrolysis oil and mixing ratio with butanol, after the end of the esterification reaction the remaining water mass fraction is around 1 wt.-% for each experiment. The experiments in the medium sized setup are slightly below and the ones in the smallest equipment are consistently above that value. This could be a significant effect caused by different surface-to-volume ratio and subsequent different heat flux into the reaction mixture, but it could also be owing to the fact that with the very small total amounts of water in the small scale setup it is extremely difficult to decide, when there is no more water evolving in the Dean-Stark-Apparatus. Figure 9: Acids mass fraction for different experiments Another important quality indicator is the acetic acid mass fraction, which is presented in Fig. 9 for the same set of experiments. The acetic acid mass fraction is more or less not influenced by the pyrolysis and condensation conditions as the three pyrolysis oils exhibit very similar values of about 4.4 wt.-%. Again, the reaction mixture shows lower values due to physical dilution. After the reaction the remaining acetic acid mass fraction is approximately half of that of the reaction mixture. As for the water mass fraction the medium sized setup slightly performed better than the smallest equipment. 4. Conclusion The preliminary results of the presented scale-up test campaign are quite promising. The remaining water mass fraction in all scales is around 1 wt.-% and the acetic acid concentration could be reduced to half the value of the reaction mixture. Further scale-up to cubic metre scale seems possible and the conversion might be performed within 20 hours, leaving 4 hours for filling, discharging and cleaning; so every day a batch can be treated to give an upgraded pyrolysis oil with only small amount of water and reduced concentration of acids. 5. Acknowledgement and disclaimer The project behind this contribution to conference proceeding has received funding from the German federal Ministry of Economy and Energy BMWi under grant nr. 03EI5412 based on a decision by the German Bundestag. We thank BMWi for the funding and its funding agency PTJ for the supervision. Additionally, we thank Clariant, subsidiary Straubing for supplying us with the cellulose-based ethanol from agricultural residues. This publication reflects the author’s view only. BMWi and PtJ are not responsible for any use that may be made of the information it contains. 70 13th International Colloquium Fuels - September 2021 Scale-up of batch esterification of pyrolysis oils with higher alcohols from 250-ml-scale to 20-l-scale References [1] International Maritime Organization: IMO 2020 - cutting sulphur oxide emissions. https: / / www. imo.org/ en/ MediaCentre/ HotTopics/ Pages/ Sulphur-2020.aspx (accessed 29.05.2021) [2] Conrad S., Blajin C., Schulzke T., Deerberg G. Comparison of fast pyrolysis bio-oils from straw and miscanthus. Environ Prog Sustainable Energy 39 (2019); e13287; doi: 10.1002/ ep.13287 [3] Bridgwater AV. Review of fast pyrolysis of biomass and product upgrading. Biomass Bioenergy 38 (2012), 68-94; doi: 10.1016/ j.biombioe.2011.01.048 [4] Schulzke T., Conrad S., Shumeiko B., Auersvald M., Kubicka D., Raymakers LFJM. Fuels from Reliable Bio-based Refinery Intermediates: BioMates. Waste Biomass Valor 11 (2020), 579-598; doi: 10.1007/ s12649-019-00625-w [5] Sundqvist T., Oasmaa A., Koskinen A. Upgrading Fast Pyrolysis Bio-Oil Quality by Esterification and Azeotropic Water Removal. Energy Fuels 29 (2015), 2527-2534; doi: 10.1021/ acs.energyfuels.5b00238 [6] Breitkreuz K, Menne A, Kraft A. New process for sustainable fuels and chemicals from bio-based alcohols and acetone. Biofuels Bioprod Bioref 8 (2014), 504-15; doi: 10.1002/ bbb.1484 [7] Schulzke T., Conrad S., Kaluza S., Van Loo T. Upgrading of fast pyrolysis condensates via esterification with higher alcohols. Biomass Bioenergy 103 (2017), 11-20; doi: 10.1016/ j.biombioe.2017.05.010