On the fast pyrolysis of torrefied woody biomass
Alexander Louwes is a PhD student in the department Thermal Engineering. (Co)Promotors are prof.dr.ir. G. Brem and dr.ir. A.K. Pozarlik from the faculty of Engineering Technology.
Producing biofuels from woody biomass via torrefaction and fast pyrolysis can play a key role in the transition to a sustainable economy. Products like bio-coal and bio-oil have the potential to replace fossil fuels for the production of energy as well as chemicals and materials. However, bio-coal and bio-oil do not have the same properties as their fossil counterparts. For example, bio-oil from fast pyrolysis has a lower quality in terms of heating value, acidity, viscosity, and storage stability. A combination of a torrefaction pre-treatment with subsequent fast pyrolysis could improve the bio-oil’s quality, in the direction of fossil fuels.
This thesis aims to provide a better insight in the possible combination of the torrefaction and fast pyrolysis technologies. Torrefaction and fast pyrolysis are both thermochemical conversion methods that take place in the absence of oxygen. Torrefaction typically transpires at temperatures of 200-300 °C with residence times of 15-60 min and low to moderate heating rates; its main product is bio-coal, and about 70-80 % of the mass is typically converted to this main product. Fast pyrolysis on the other hand takes place at a significantly higher temperature of about 500 °C, with residence times of a few seconds and a high heating rate of up to 1000 °C/s. Here, the main product is bio-oil, with up to 70 % of the feed being converted to this liquid product, and the other 30 % to char and gaseous byproducts.
Five research objectives were formulated that are addressed in this thesis. The first objective was to characterize the physical and chemical properties of the coal-like product from the torrefaction of raw biomass (bio-coal) in terms of elemental composition, fuel characteristics, and grindability. To this end, a torrefaction screw reactor was designed, built, and successfully used to torrefy beech wood and rape straw samples under various torrefaction conditions. This resulted in a series of bio-coal samples which had a significantly higher heating value compared to the raw samples (up to 40 % higher) – at the cost of a mass loss of up to 65 %, although this mass loss consisted mainly of water and other low heating value compounds. The bio-coal samples were analyzed and compared with fossil coal samples regarding elemental composition and grindability. A Van Krevelen diagram showed that the most severely torrefied samples (at 300-325 °C for 15 minutes) were very close to the coal-region regarding both O/C and H/C. In addition, to be able to compare the biomass samples regarding grindability, a new Modified Hardgrove Grindability Index (MHGI) was devised. The MHGI graph showed a strong increase in grindability for the torrefied samples, and the particle size distribution plot underscored this by showing a significant increase of smaller particle size fractions (<200 µm) compared to raw or lightly torrefied samples. In conclusion, the torrefaction of raw biomass was able to produce a bio-coal with coal-like qualities in terms of heating value, elemental composition and grindability, making it an interesting candidate to (partially) replace fossil coal in coal-fired power plants – or to act as a precursor for the production of higher quality bio-oil via fast pyrolysis.
The second research objective was to upgrade the cyclonic thermogravimetric analyzer (TGA) by extending it with a condenser unit to determine product yield in addition to primary fast pyrolysis kinetics. For this objective, design optimizations were added to the existing cyclonic TGA-setup, including an air-operated valve for sample introduction, a fast-responding electronic balance, and a removable condenser add-on. Two experimental campaigns were conducted to measure conversion times and kinetics (with raw beech wood samples of three different particle size regimes, all <500 μm) as well as oil yields (with both raw and torrefied beech wood samples of 6 different particle sizes regimes, all <500 μm). To determine the kinetics, the fast pyrolysis temperature was varied between 450-550 °C in the first experimental campaign. The fast pyrolysis temperature was kept at 500 °C for the second experimental campaign. The results from the first campaign showed that the optimized cyclonic TGA operates stably and gives reproducible results. Varying the particle size did not lead to different results, indicating that the process is kinetically controlled. Furthermore, varying the gas flow rate also did not influence the results, indicating that in addition, there was no limitation of external heat or mass transfer to or from the particles. Particle full conversion times varied from 4 seconds at 550 °C to around 12 seconds at 450 °C. Regarding kinetic constants, the activation energy () was calculated at close to 80 kJ/mol, and the pre-exponential factor () at about 100,000. The condenser add-on performed well and resulted in oil yields of up to 73 % for raw wood and 62.5 % for wood torrefied at 260 °C. A trend of decreasing yield with decreasing particle size was also noticed and was attributed to a difference in chemical composition between different particle sizes. To achieve the highest oil yields for both raw and torrefied wood, designs for fast pyrolysis plants should aim for particle sizes no less than 400 µm – which in addition is beneficial regarding grinding costs.
The third research objective was to develop an analytical model which describes the fast pyrolysis conversion of a single biomass particle based on measurements of fast pyrolysis kinetics with the cyclonic TGA. To this end, a large experimental campaign was conducted with the cyclonic TGA, involving three different woody biomass species (beech, spruce, and ash wood) ground down to eight particle sizes between 0-425 μm. Next to the raw samples, the beech wood samples were torrefied in-house using an auger screw reactor at six different temperatures with a residence time of 15 min. The spruce and ash wood samples also came in raw and torrefied variants, and here torrefaction took place off-site in a directly heated moving bed pilot plant, at 260 °C and 280 °C for spruce and 250 °C and 265 °C for ash wood. The cyclonic TGA was operated at five different fast pyrolysis temperatures between 450-550 °C. The results showed similar conversion times (around 1 s) at high fast pyrolysis temperatures (> 500 °C) for the various biomass types, but a larger spread (3-5 s) at lower fast pyrolysis temperatures (< 500 °C), suggesting that at higher fast pyrolysis temperatures, the conversion is more dominated by heat transfer as the biomass samples had similar thermodynamic properties but diverged more regarding kinetic properties. In general, both high temperatures and smaller particle sizes led to faster conversion, and both variables played a significant role in the regimes studied. Regarding a torrefaction pre-treatment, the effect on conversion times was more significant for smaller particles, indicating that a different chemical composition or internal thermal conductance/resistance was the cause. Torrefaction leads to a slightly increased conversion time, more noticeable at lower fast pyrolysis temperatures (< 500 °C) – and more severe torrefaction leads to more significant increases in conversion time. The results were then converted to a model which splits the overall conversion time into the time necessary for a particle to heat up plus the time necessary for a particle to convert based on kinetics . The model showed a good fit with both raw beech and raw ash wood at fast pyrolysis temperatures of 450-550 °C. The model did underpredict conversion times of torrefied variants and would need to be adapted to fit the results of torrefied samples more accurately. In designing fast pyrolysis reactors, the model could be employed as an aid to calculate required residence times and estimating the main dimensions.
The fourth research objective was to determine the quality of fast pyrolysis bio-oils from raw and torrefied biomass sources as well as the energetic viability of a torrefaction pre-treatment, based on laboratory experiments. Raw and torrefied hardwood (ash wood), softwood (spruce) and mixed waste wood samples were used to perform fast pyrolysis experiments in a 1 kg/h entrained down-flow reactor with a length of 4.2 m. The hardwood samples were torrefied in a directly-heated moving bed reactor at two temperatures (250 °C and 265 °C), and were delivered in two variants next to the raw wood chips: torrefied chips and torrefied pellets. The softwood samples were torrefied in the same reactor at 260 °C and 280 °C, and the mixed waste wood samples were torrefied in a Torbed reactor at a temperature of 280-300 °C. The fast pyrolysis experiments resulted in oil yields of 42-45 % for the raw wood samples, and 25-36 % for the samples that had undergone a torrefaction pre-treatment – with solids yield increasing from (on average) 25 % for bio-oil from raw samples up to 40-50 % for the pre-treated samples. An elemental analysis of the bio-oil samples mainly revealed differences in the carbon and oxygen content, with the torrefaction pre-treated bio-oil samples gaining about 5-10 % in carbon content and losing about the same amount in oxygen content. Moreover, the heating values increased on average with 20 %. Van Krevelen diagrams of the various bio-oils showed that torrefaction results in a shift towards the origin of the diagram (and towards fossil counterparts): significantly lower O/C ratios (down from 0.7-0.8 to 0.5-0.6) and slightly lower H/C ratios. Although the O/C ratio of a bio-oil which has undergone a post-deoxygenation treatment is not quite reached (about 0.15), a significant step in the direction was made. The bio-oil samples’ storage stability was then evaluated by employing an accelerated aging test while measuring the change in water content and viscosity. Both raw and pre-treated samples showed a similar increase in water content, but the pre-treated samples became much more viscous (about 2-2.5 times) than the raw bio-oil samples (about 1.25-1.5 times). Regarding the energetic viability, two routes to produce bio-oils of similar quality (regarding oxygen content) were compared: one that was experimentally studied here, in which woody biomass is torrefied, ground, and then subjected to fast pyrolysis, and one comparison route in which woody biomass is ground, dried, subjected to fast pyrolysis, and subsequently subjected to a hydro-deoxygenation (HDO) post-treatment to arrive at a similar decrease in oxygen content. The results showed a significant increase in energy efficiency from feedstock to the sum of all products for the torrefaction and fast pyrolysis route over the conventional chain plus HDO, with a 15 % increase for hardwood, 25 % for softwood, and over 50 % for mixed waste wood. Looking only at energy represented in bio-oil, the same trend held, and a maximum chain energy efficiency of 25 % was reached in the pre-treated hardwood bio-oil. Overall, it was concluded that the most severely torrefied softwood sample yielded the best bio-oil quality (in terms of oxygen content, heating value and storage stability), while the hardwood samples outperformed the other samples in terms of energy efficiency, although all samples pre-treated with torrefaction performed better than their raw counterparts. The significant increase in char yield with pre-treated samples does require finding (commercial) purposes for the char, such as activating it to sell it on the market.
The fifth research objective was to determine the economic and environmental feasibility of a combined torrefaction and fast pyrolysis chain to produce electricity from biomass compared to a conventional chain that employs biomass combustion. A case study was conducted to convert wood from Brazil into electricity in the Netherlands, based on the two chains mentioned. In the conventional chain, wood is harvested from plantations near the Amazon river, shipped to the river port of Santarém where it is dried, chipped and pelletized, before transporting the pellets to the seaport of Macapá for the ocean journey to the port of Rotterdam in the Netherlands. After arriving in Rotterdam, the pellets are combusted for the production of electricity. On the other hand, for the torrefaction and fast pyrolysis chain, wood is harvested at ten decentral locations and torrefied and ground on-the-spot, before being transported to the central fast pyrolysis plant at the river port of Santarém. The produced bio-oil is then, similar to the conventional chain, shipped via Macapá to Rotterdam to be combusted for the production of electricity. To set up a dynamic techno-economic analysis, both chains were translated into mass and energy balances, and CAPEX, OPEX and revenue values were associated with all variables in the balances. With an input of 1,000 kton/a of wood and a weighted average cost of capital (WACC) of 5.4 %, projected discounted cash flows for the economic lifetime of the projects (20 years) were calculated. The results showed a slightly higher annualized CAPEX for the torrefaction and fast pyrolysis case (33 M€ vs 27 M€) mainly due to the additional number of (small torrefaction) plants, but a significantly lower OPEX (140 M€ vs 164 M€) mainly due to savings in electricity costs for grinding as well as transport costs. As the annual revenues are relatively equal (194 M€ for the conventional case and 192 M€ for the torrefaction and fast pyrolysis case), the Net Present Value (NPV) of the torrefaction and fast pyrolysis case is much higher (167 M€) compared to the conventional case (16 M€). Sensitivity analyses did reveal that both projects would be significantly impacted by changes in both the wood purchasing prices as well as the electricity selling prices: a 20 % increase in the wood purchasing price (set to 50 €/ton) or a 10 % decrease in the electricity selling price (set to 100 €/MWh) would make the torrefaction and fast pyrolysis case unviable. For the conventional case, even a 5 % change (increase for wood or decrease for electricity) would lead to the project becoming unviable, highlighting the importance of securing long-term fixed contracts. An environmental analysis showed that most of the CO2 emissions from both chains are biogenic (90-95 %) and could be mitigated by replanting the trees (or a carbon sink could be created by capturing the CO2 at the combustion plants). The fossil CO2 emissions were twice as high for the conventional case (0.08 kg CO2/kWh electricity generated) and consisted mainly of transport fuel combustion as well as electricity consumption for grinding. For the torrefaction and fast pyrolysis case, transport fuel combustion was lowered by two thirds, although it still made up about half of its fossil CO2 emissions. Using the fast pyrolysis bio-oil as transport fuel could mitigate fossil CO2 emissions even further.
In conclusion, this thesis shows the technical feasibility and economic viability of the combination of a torrefaction pre-treatment with fast pyrolysis for the production of bio-oils. Although torrefaction on its own also has an interesting use case by producing bio-coal, the combination with fast pyrolysis leads to a higher quality bio-oil that can be used much more flexibly than bio-coal: as transport fuel, as fuel to produce electricity in large-scale power plants or in small-scale and possibly remote locations, or as drop-in fuel in oil and bio-refineries. More research into these various applications of bio-oil, as well as further research into increasing bio-oil yield and quality are recommended to both improve the understanding of the processes involved as well as to open the way for commercial use of the technology.