Renewable hydrocarbon fuels from hydrothermal liquefaction: A techno-economic analysis

This study demonstrates the economic feasibility in producing renewable transportation drop-in fuels from lignocellulosic biomass 9 through hydrothermal liquefaction and upgrading. An Aspen Plus ® 10 process model is developed based on extensive experimental data to 11 document a techno-economic assessment of a hydrothermal liquefac-12 tion process scheme. Based on a 1000 tonnes organic matter per 13 day plant size capacity, three different scenarios are analysed in order 14 to identify key economic parameters and minimum fuel selling prices (MFSP). Scenario I, the baseline scenario, is based on wood-glycerol 16 co-liquefaction, followed by thermal cracking and hydroprocessing. 17 Results show that a minimum fuel selling price (MFSP) of 1.14 $ per 18 litre of gasoline equivalent (LGE) can be obtained. In Scenario II only 19 wood is used as feedstock, which reduces the MFSP to 0.82 $/LGE. 20 Scenario III is also based on a pure wood feedstock, but investigates a 21 full saturation situation (a maximum hydrogen consumption scenario),

resulting in a slightly higher MFSP of 0.94 $/LGE. A sensitivity anal-23 ysis is performed identifying biocrude yield, hydrogen and feedstock 24 prices as key cost factors a ecting the MFSP. In conclusion, the study 25 shows that renewable fuels, via HTL and upgrading, can be highly 26 cost competitive to other alternative fuel processes. 27 1 Introduction 28 Biomass is the most important renewable carbon source with the ability to 29 replacing current fossil transportation fuels. One way of converting highly 30 diverse, low cost biomass into drop-in liquid fuels is by hydrothermal liq-31 uefaction (HTL). In HTL, biomass is mixed with water and processed at 32 temperatures between 250-450 °C and pressures between 5-35 MPa [1,2,3]. 33 The severe conditions improve water solvent properties, enhancing the pro-  46 Despite the fact that HTL has been proven technically viable, the eco-47 nomic perspectives of the process have only been superficially addressed. 48 The Pacific Northwest National Laboratory (PNNL) established an economic 49 frame of reference for the viability of fast pyrolysis and hydrothermal lique-50 faction followed by upgrading. Fuel costs of 0.82 and 0.53 $/litre of gasoline 51 equivalent (LGE) were estimated for fast pyrolysis and HTL, respectively. 52 The study considered a large scale production facility, processing 2000 met- 53 ric tons per day of dry wood (474 MW) [6]. The assessment was evaluated 54 based on a 8 % dry matter content feedstock, which is rather low compared 55 to other experimental studies that have demonstrated the processing of feed-56 stock having dry matter contents much higher [12, 10,13]. In another study 57 by Zhu et al., based on a similar process but in a current "state-of-technology" 58 scenario, the cost of producing renewable fuels from woody biomass (15 wt.% 59 dry biomass content) using HTL was estimated at 0.74 $/LGE [5]. In the 60 same study a "goal case" scenario was evaluated, projecting a minimum fuel 61 selling price (MFSP) of 0.67 $/LGE. 62 In a more recent study, de Figure 1 shows the main process scheme considered in this work. Starting 123 from the pretreatment, wood is grinded and mixed with glycerol and water, 124 constituting the initial slurry. This mixture is pressurized to 300 bar, then 125 heated to 400 C, prior to entering the HTL reactor. Within the reactor, 126 the organic macromolecules are decomposed into smaller compounds. After-127 wards, the product is depressurized, cooled down and separated into three 128 phases: water phase, gas and biocrude. 129 The gas phase, mainly constituted by CO 2 [16], can be extracted for 130 further utilization either as a combustible source or e.g. for CO 2 or H 2 131 recovery. In the current study, the gas phase is used as a combustible source 132 for process energy. The gas composition can be seen in Table 2. 133 The water phase is rich in organics, and in the case of lignocellulose 3 134 processing, the organic content of the water phase represents a signi cant 135 share of the organic output (> 50 %). In this study, it is assumed that 90 % 136 of the water phase is recovered by recirculating it back to the system input. 137 Other means of water phase utilization has been proposed, such as anaerobic 138 digestion for biogas production [6]. The remaining 10 % of the water phase 139 is lost to a water phase cleaning process. After the HTL core process, the 140 biocrude is separated into a volatile and a non-volatile fraction (residue) by 141 distillation according to [16]. The residue is then thermally cracked and 142 mixed with the volatile stream prior to co-hydrogenation in a hydrotreater. 143 The obtained liquids are referred to as gasoline equivalents.

145
The total system is divided into four core blocks, which are explained in the 146 following.  Table 1.   (1): To summarize the different process conditions, Table 2 collects the main 207 properties and assumptions of the system. Energy is a major concern for an industrial plant, since an optimal configu-211 ration implies economical savings. 212 From the thermal requirements a pinch analysis was performed obtaining 213 the minimum hot and cold utility demands. A minimum temperature ap-214 proach of 20 °C was used. The energy requirements presented are calculated 215 per kilogram of organic matter fed to the system (kJ/kg). 216 The overall hot and cold utility demands, before incorporating any heat 217 recovery, are 7404 and 5325 kJ/kg, respectively. Figure      will behave as well as the pure glycerol for the process. 277 The selected wood feedstock is sawn timber residue. The sensitivity analysis is an economic tool designed to identify the param-291 eters responsible for major cost variations. 292 Three different scenarios have been analyzed. In the rst scenario the or-293 ganic matter fed to the system is constituted by a mixture of 50/50 biomass-294 glycerol. For the second case, the glycerol has been substituted by water, 295 therefore the inlet is reduced to 500 tonne/day of wood, but the overall mass 296 ow is maintained. In the third scenario, the biocrude is totally deoxygenated 297 and saturated. The prices influencing the MFSP for the three scenarios is 298 shown in Table 5. Most of these prices will be used as starting values for the 299 sensitivity analysis.  Table 4. Notice that the yield 307 it is not subject to price variations, thereby is only represented by a bar. 308 The most likely fuel production cost of Scenario I is 1.14 USD/LGE, 309 approx 3 times higher than fossil gasoline. 310 The HTL yield is the most sensitive parameter from the system, produc-311 ing a cost change of 0.13 USD within a 10 % variation. 312 Moreover, its unequal deviance at each side of the base cost shows that a 313 reduction in the yield has a higher impact on the cost per litre than an equal 314 yield increase. 315 Regarding the glycerol, the wide price variation represents a problem to 316 the fuel cost stability. Furthermore, glycerol has also a wide bar that con rms 317 the large effect that it has on the overall cost. 318 Hydrogen has the third major effect on the cost variation, with a price 319 variation which depends on the production technology, where electrolysis 320 represent the upper limit. 321 Despite the high consumption, wood has a small impact on the production 322 cost due to its low cost. Therefore it represents a small percentage in the 323 total cost share.
The remaining parameters have a minor impact on the cost, where the 325 electricity and thermal energy have a low impact on the system with a low 326 cost reduction potential. 327 As stated, glycerol represents a key parameter for the process viability 328 due to its high cost and price range. Thus, it is necessary to study the 329 economic behaviour of the system using only wood as organic feedstock.

334
In this scenario, the glycerol is excluded from the feedstock but the wood feed 335 rate remains unchanged. The system performance is una ected except for 336 the yield, which increases the conversion performance of the HTL by 10 % 337 due to the lack of glycerol [10]. The changes made between scenario I, II 338 and III is shown in Table 6. It has to be noticed that char formation would 339 increase by 25 % when glycerol is excluded [10], which could generate other 340 variations in the process. 341 Figure 7 represents the economical sensitivity of di erent parameters in 342 the system. In this case, the production cost has been reduced to 0.82 USD/LGE, 343 which represents a 28 % decrease compared to scenario I. It has to be noticed 344 that the total mass ow of drop-in fuels is lower than the previous case, as 345 the system is fed only with 500 tonne/day of wood. 346 The HTL yield is still the most dominant factor, maintaining the same 347 behaviour described in the first scenario. The hydrogen consumption is pro-348 portional to the amount of biocrude processed, thus this parameter is not 349 affected by the exclusion of glycerol from the system, but due to the share, 350 it becomes the most sensitive parameter after the yield. 351 On the other hand, the price of wood accounts for a larger effect on the 352 cost variation, as constitutes the only organic inlet of the system. 353 The sensitivity bar for the wood is equal in size to the thermal energy 354 and plant cost, meaning that a variation on these parameters equally affects 355 the fuel cost. 356 The rest of the parameters increase proportionally their share of the pro-357 duction cost, but still having a minor e ect on the price contribution. 358 The beneficial effects of the glycerol in the HTL system has to be further 359 studied as it represents a major cost of the bio-fuels production. Therefore, 10 it has to be evaluated if the reduction of the char formation compensates 361 the reduction in the biocrude yield in views of the relative high price of the 362 glycerol, that could be six times higher than wood. 363 The MFSP ranges between 0.56-1.16 USD/LGE.

365
As for the previous scenario, glycerol is not included in scenario III either. It 366 is assumed that complete deoxygenation and saturation of the biocrude, con-367 sequently the hydrogen consumption increases to 0.0482 kg/kg of biocrude. 368 The complete deoxygenation is the only change between case II and III as 369 shown in Table 6 It is expected an increase of the HHV due to de lower oxy- This comparison does not only shows the sensitivity to this parameter from 415 geographical effects, but also shows the need for low cost feedstock for a 416 feasible business case. 417 In combination with external price parameters such as the three discussed 418 above, internal process performance obviously also plays a role.         . MFSP of a best case scenario using lowest external prices and high dry matter content and yield (in red).