Version of Record: https://www.sciencedirect.

com/science/article/pii/S0378382017314236
Manuscript_99feb702d19e4d5a811bb1dc9442414d

1 Evaluation of Fast Pyrolysis Feedstock Conversion with a Mixing Paddle Reactor

2
3 S. Zinchik1, J.L. Klinger2, T.L. Westover2, Y. Donepudi1, S. Hernandez2, J.D. Naber1, E. Bar-
4 Ziv1*
5
6 1 Michigan Technological University, 1400 Townsend Dr., Houghton, MI 49931
7
8 2 Idaho National Laboratory, 1955 N. Fremont Avenue, Idaho Falls, ID 83415
9
10 *Corresponding author
Contact information: Ezra Bar-Ziv
Michigan Technological University
815 R. L. Smith Bldg
Houghton, MI 49931

Phone 906-487-3151
E-mail [email protected]
11
12
13 Abstract
14 We have developed a pyrolysis reactor based on a unique auger-paddle configuration
15 with heat transfer material (HTM) and proved to achieve high heating rates and fast pyrolysis.
16 We tested ten different biomass types and obtained bio-oil yields ranging from approximately
17 40% for thermally treated wood, to approximately 57% for crop residues (corn stover) and 67%
18 yield for woody feedstocks (tulip poplar). These results, as well as the solid char yields, are
19 similar to those obtained for the same feedstock using a circulating fluidized bed. Tests
20 conducted without HTM resulted in lower bio-oil yields (ranging from 8 to 18% decrease in
21 yield) and higher char yields with similar changes in magnitude, which is indicative of slow
22 pyrolysis. In addition, a comprehensive study and analysis of the material residence time and
23 mixing characteristics of the novel auger-paddle system is presented. These results demonstrate
24 that an auger-paddle configuration is capable of achieving the high heating rates required for fast
25 pyrolysis.
26
27 Keywords: Fast pyrolysis; biomass; bio-oil; augers; paddle reactor
28
29 Highlights
30 • A novel single shaft auger was developed to achieve uniform solid particle mixing.
31 • The enhanced mixing reactor was validated for fast pyrolysis of biomass.
32 • Material residence times and process parameters established mixing performance.
33

© 2018 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
34 1 Introduction
35 Pyrolysis is a thermal degradation process that occurs in the absence of oxygen. It is a
36 promising route for creating a liquid fuel that can be upgraded to transportation fuels or used
37 directly as a fuel (Mohan, Pittman, and Steele 2006). Of specific interest, biomass pyrolysis has
38 been heavily investigated for the generation of bio-oil for upgrading to fuels with a significantly
39 reduced or neutral carbon footprint. Balat et al. divided pyrolysis into three categories:
40 conventional pyrolysis, fast pyrolysis, and flash pyrolysis (Balat et al. 2009) based upon their
41 reaction temperature, heating rate, solid residence time and particle size. Conventional pyrolysis
42 is generally applied to relatively large particles (5-50mm) because conduction heat transfer limits
43 the heating rates to less than 1 K/s. Fast pyrolysis is typically applied to small particles (<1 mm)
44 for which heating rates of 10-200 K/s can be achieved. The residence time of fast pyrolysis
45 reactions are usually in the range of 0.5-10 s. Flash pyrolysis is typically applied to particles that
46 are smaller than 0.2 mm at heating rates >1000 K/s and residence times <0.5 s (Balat et al.
47 2009).
48 Several recent reviews have investigated various production technologies for pyrolysis
49 including ablative (coil, mill, plate, vortex, etc.), circulating fluidized bed, entrained flow,
50 fluidized bed, moving bed (vacuum, transported, stirred, horizontal, etc.), rotary hearth,
51 microwave, and rotating cone (Bridgwater, Meier, and Radlein 1999; Bridgwater 2003; Butler et
52 al. 2011; Venderbosch and Prins 2010). Regardless of the technology, however, perhaps the
53 most important factor for the conversion reactor is the heat transfer rate stemming from the
54 reactor itself (reactor wall in ablative pyrolysis, gas or wall contact in transport bed or entrained
55 flow), or from use of a heat transfer medium (HTM) such as the bed material in a fluidized bed.
56 According to Briens et al., 2008, and Butler et al., 2011, the only current technologies
57 that can be commercially applied for bio-oil production are the bubbling fluidized bed (BFBs)
58 and circulating fluidized beds (CFBs), although auger reactors also have high market attraction
59 because of their simplicity, robustness, and their long established history as effective conversion
60 reactors (Briens, Piskorz, and Berruti 2008; Butler et al. 2011). Particularly the Lurgi-Ruhrgas
61 twin-screw mixer has been extensively investigated in the past for thermal treatment (focusing
62 on coal degassing), and has been demonstrated at a capacity of at least 50 ton/day for pyrolysis
63 (Butler et al. 2011; Venderbosch and Prins 2010). In addition, McGee and Miao recently detailed
64 the application of auger feeders in fast pyrolysis systems and difficulties in continuous feed of
65 biomass systems (McGee 2012; Miao et al. 2014). It was found that the established and simple
66 characterization tests, such as bulk density, particle size, and angle of repose did not correlate
67 well with established auger correlations and must be proven empirically. Mixing paddle reactors
68 offer additional advantages over standard mixing auger reactors because they provide greater
69 control of the radial and axial mixing patterns. This increased control offers opportunities to
70 decrease the amount of HTM required to maximize liquid production, while still maintaining the
71 simplicity, robustness and market attractiveness of auger-based reactors. For these reasons, this
72 work publishes the fast pyrolysis performance of several feedstocks in a custom auger-paddle
73 reactor.
74 Specifically for thermal treatment of materials, augers have been heavily studied in
75 disposal and recycling of waste materials, degassing of coals, and gasification. Chun et al.
76 studied the pyrolysis gasification of sewer sludge in a pilot screw reactor (Chun, Kim, and
77 Yoshikawa 2011). Many studies have been done on the recycling/disposal of automotive waste
78 and shredded tires in pilot conversion units ranging from 100 g/h to 8 kg/h (Aylón et al. 2010;
79 Aylón et al. 2008; Day, Shen, and Cooney 1999; Haydary et al. 2016; Martinez et al. 2013), and

2
 80 larger commercial units, such as that studied by Day et al. at 200 kg/h (Day, Cooney, and Shen
81 1996). In study of the larger unit, Day et al. concluded that although the auger reactor is a good
82 method of resource recovery and waste disposal, and is energetically self-sustaining, further
83 development is required and there is vast room for improvement. Other studies have found
84 similar results from medium density fiberboard scraps and aseptic packaging (Ferreira et al.
85 2015; Haydary, Susa, and Dudas 2013). The range of studies that have been conducted using
86 auger reactors demonstrates that such systems are robust for a wide variety of materials and
87 processing conditions.
88 Many researchers have shown that auger reactors operating without using HTM can
89 liquid achieve yields that are comparable those obtained by fluidizied beds (Bosong et al. 2014;
90 Liaw et al. 2013; Liaw et al. 2012; Dahmen et al. 2012; R. Zhou, Lei, and Julson 2013; S. Zhou
91 et al. 2013; Puy et al. 2011; Morgano et al. 2015; Garcia-Perez et al. 2007; K Promdee and
92 Vitidsant 2014; Kittiphop Promdee and Vitidsant 2013). However, the liquid composition can be
93 significantly affected from primary and secondary reactions arising from low particle heating
94 rates and vapor-phase residence times.
95 In order to increase the heating rate of the biomass particles within auger reactors to
96 avoid such secondary charring/cracking reactions or long particle/vapor residence times, some
97 development has been done to include a heat transfer medium (HTM). Through this addition,
98 biomass particles are not only heated through contact with the hot reactor wall, but mostly
99 through contact with the preheated HTM. Several HTM have been investigated including silica
100 sand and quartz (Henrich et al. 2016; Pfitzer et al. 2016), steel shot (Brown and Brown 2012;
101 Henrich et al. 2016), and various catalysts (clay minerals such as bentonite and sepiolite, and
102 oxides of alkali metals such as calcium oxide) referenced above. Henrich et al. used a twin screw
103 auger (40 mm diameter screws, 1.5 m total length) with various heat transfer media to process
104 hardwood, softwood, wheat bran, and straw (Henrich et al. 2016; Pfitzer et al. 2016). Average
105 yields of 66.5%, 69.1%, 60.0%, and 51.4% were achieved for the respective feedstock with feeds
106 of approximately 10 kg/h raw biomass and 1,150 kg/hr heat transfer media. Brown and Brown
107 also found that auger-type reactors are well suited for bio-oil production, and achieved yields of
108 73% in a surface-response analysis of their 1 kg/hr twin auger system (25.4 mm diameter screws,
109 0.56 m total length) with an HTM feed rate of 18 kg/hr and 63 RPM (Brown and Brown 2012).
110 They report that the addition of a HTM reduces the solid residence time by more than 95%, and
111 achieves 25% more liquid yield compared to other auger studies that use external heating (Brown
112 and Brown 2012). Both studies used short solid residence times (ranging in 5-15 s) to
113 demonstrate that HTM used within twin screw auger reactors exhibits sufficient heat transfer to
114 perform fast pyrolysis, and produce yields comparable to those of fluid bed reactors.
115 The present study undertakes the development of a single-shaft auger-type reactor with
116 cuts in the flighting and additional paddles to optimize the mixing between the feedstock and the
117 HTM, in order to reduce the motor power requirement and the wear on moving parts. Here we
118 show that this reactor design produces bio-oil with similar properties and yields as other leading
119 technologies, such as bubbling fluidized bed (BFB) and circulating fluidized bed (CFBs)
120 systems, while potentially reducing capital and operation costs. Other objectives of this study
121 are to determine: (1) residence times as functions of rotation frequency of the reactor shaft; and
122 (2) the quality of mixing, or rather heat transfer, in this configuration.
123 We tested ten distinctly different feedstock materials with this system at a reaction
124 temperature of 500 °C to evaluate the pyrolysis yields in comparison to other literature. It is
125 shown that the unit produces quantitatively similar liquid yields when compared to other studies,

3
126 as well as good qualitative representation of yields obtained from varying feedstock in other
127 reactor systems.
128
129 2 Materials and Methods
130 In this work we show the reactor configuration and parameters that determine residence
131 time and mixing (heat transfer) rates. Results for ten biomass feedstock that were tested within
132 the mixing auger reactor system are presented to evaluate conversion performance and pyrolysis
133 characteristics. The feedstocks include switchgrass, corn stover, hybrid poplar, clean loblolly
134 pine, sorted construction and demolition (C&D) wood waste, thermally treated loblolly pine,
135 miscanthus, tulip poplar, piñion juniper, and a blended feedstock consisting of clean loblolly
136 pine, tulip poplar, and switchgrass. These feedstock were tested in a pilot scale reactor described
137 later in this section and compared to results obtained from a circulating fluidized bed reactor.
138
139 2.1 Biomass Collection and Preparation
140 Ten biomass feedstock were screened for conversion in this study. Ten biomass samples
141 (approximately 5-10 kg) were prepared at the Idaho National Laboratory as representative
142 samples from larger biomass piles. The samples were ground using a knife mill equipped with a
143 2 mm screen (Thomas Wiley Laboratory Mill Model 4, 1 hp; Thomas Scientific, NJ).
144
145 2.2 Mixing Paddle Reactor System
146 The fast pyrolysis system consists of four main parts: (1) the HTM dosing system (2) the
147 biomass dosing system (3) the heating zone for the HTM (4) fast pyrolysis reactor zone similar
148 to the HTM heater. The current system is semi-continuous and is shown schematically in Figure
149 1, but is easily adaptable to continuous operation; the cartoon in the figure is not drawn to-scale.
150 The solid bio-char stream exits the system into a sealed container (not drawn). The gas stream
151 (condensable bio-oil and non-condensable gases) flows through a heated transfer line and
152 through a condenser (described further below) that collects the liquid product into a sealed tank
153 (not drawn). After the condenser, the cold non-condensable gases pass through a cold water bath
154 to capture any remaining bio-oil (not drawn). The system is kept inert with a sweep stream of
155 nitrogen. The nitrogen flow rate is adjusted so that the residence time of the gases in the transfer
156 line does not exceed 2 seconds.
157 An important aspect of any similar system is the consistent and continuous flow of both
158 HTM and feedstock. The HTM feed system consisted of a bin that flood-feeds a standard 2.54
159 cm diameter regular screw auger flight with a pitch of 2.54 cm. A pneumatic agitator was placed
160 inside the feed bin to avoid bridging. This agitation was essential for the smooth and continuous
161 operation of the system as it ensures rather constant mass flow rates. A similar agitation
162 configuration was used for the feedstock. The shaft of the dosing augers reduces in diameter
163 from the flood fed bin area to the main delivery shaft to allow for a flood fed mouth and avoid
164 material plugging (depicted in Figure 1). The shaft has a diameter of approximately 1.6 cm under
165 the feed bin that decreases to 1.27 cm after the material feed bin. The feed bin can accommodate
166 approximately 3.5 L of material and is sealed after material charging. The feed rate is controlled
167 with a motor equipped with a variable frequency drive. The shaft has packing gland seals with
168 graphite packing on both ends of the auger housing to prevent air from entering the system. The
169 biomass is metered into the system in an identical manner. The pyrolysis reactor acts to: (i) heat
170 the HTM, (ii) mix the HTM with biomass, and (iii) react the biomass using fast pyrolysis
171 conditions.

4
172 Heating of the mixing paddle reactor was accomplished by 12, 1-inch-wide heating bands
173 of 250 W each down the length of the system (approximately 45 cm length), with attached
174 thermocouples that are inserted inside the reactor and touching the moving material without
175 touching the auger paddles. Figure 1 also shows a cartoon of the heating-control configuration.
176 Each thermocouple measures the material temperature at the respective location and controls the
177 operation of the respective heating band. A programmable logical controller (PLC) monitors all
178 temperatures and uses a proportional-integral-derivative (PID) controller to operate the heating
179 bands independently. Both temperature and heating duty-cycles are recorded continuously to
180 interpret heating rates of the material.
181 The mixing auger extends through the HTM pre-heat and pyrolysis zones and is
182 comprised of two mixing elements: (1) cuts in the auger flighting and (2) four mixing paddles
183 within each flight pitch. These features are known to enhance mixing rather significantly. These
184 features were selected due to the necessity of superior mixing required for fast pyrolysis that is
185 not attainable in conventional single auger configurations. The diameter of the mixing auger is
186 2.54 cm with a pitch of 5.08 cm and a shaft diameter of 1.27 cm. The flighting cuts consist of
187 five equal segments of approximate length and spacing of 36 degrees each. Approximately 40%
188 of the flighting area within the 36 degree of the flight is cut away, and leads to material
189 effectively not being conveyed forward and being mixed with the incoming materials from the
190 previous pitch. Four paddles were also placed within each pitch to push the material both
191 sideways and forward, and also lift and retain the solids. The paddles act as an internal mixing
192 device within each pitch and are placed at 45 degrees to the shaft. The paddles are of similar
193 dimensions to the flighting cuts, but extend from auger shaft to the same diameter as the normal
194 flighting. Four paddles are spaced evenly within each pitch. Features of the reactor configuration
195 are shown in Figure 1. Cut flightings and paddles slow down significantly the conveyance of the
196 particles and enhance solids mixing.
197 The pyrolysis gases and vapors are carried with high purity nitrogen sweep gas to a single
198 pass shell-and-tube condenser with three condensing tubes. The jacket of the condenser is
199 chilled with cold water with ice maintained at approximately 0 °C. Pyrolysis gases and
200 uncondensed vapors are then bubbled through an ice water bath maintained at approximately 0
201 °C to capture additional volatile organics before being vented. Oils from the exchanger were
202 collected in standard plastic bottles and stored at approximately 0 °C away from light sources.
203 The solids fall under gravity into a sealed collection bin after the pyrolysis zone. The char
204 collector is not insulated and remains at a relatively low temperature (<200 °C) compared to the
205 reactor, ensuring that the material does not continue to react.
206
207 2.3 Mass Flow rates and Residence Time
208 Mass flow calibration tests were performed with the HTM and biomass materials to
209 determine the rotation frequency required in the dosing feeders to deliver the required mass
210 flows. In these experiments material was flown from the respective feed bins by the dosing
211 auger, at a certain shaft rotation and weighed continuously by a balance with a resolution of
212 0.01g. The material weight was monitored continuously by a PC. Mass flow rates were readily
213 available by automatic derivative of the weight transients. A similar process was used to
214 estimate material residence time in both dosing auger and the reactor.
215
216 2.4 Pyrolysis Experimental Procedure

5
217 During the pyrolysis trials, the system was brought to an initial thermal steady state prior
218 to flowing the HTM, in which the HTM preheat zone with all eight heating bands (heaters 1-8 in
219 Figure 1) were controlled to a set-point of 550 °C. The pyrolysis reactor zone, with all four
220 heating bands (heaters 9-12 in Figure 1) was controlled to a set point of 500 °C. The concept is
221 that the larger percentage of HTM and biomass causes the biomass to heat rapidly to achieve fast
222 pyrolysis conditions. The vapor transfer line to the condenser was controlled to 400-425 °C
223 during all trials to prevent condensation of liquid product. After the system reached the
224 respective stable temperatures (approximately 30 minutes), the flow of HTM (sand) was fed at a
225 continuous 1,500 g/hr. This caused the initial heaters to deviate from their set points until the
226 PID controllers could accommodate the temperature drop and increase the duty cycle on the
227 affected band heaters and re-approach a steady state (approximately 15 minutes). After this
228 steady state was reached, the biomass was fed at a continuous rate of approximately 100 g/hr. A
229 high flow of HTM relative to the biomass was used to ensure sufficient heat was available within
230 the mixing solids and provide a mechanism for fast heat transfer. It is likely that the amount of
231 HTM could be significantly reduced through optimization of temperatures and flow rates, but the
232 possibility was not investigated in this proof-of-concept study. Approximately 400 grams of
233 biomass was tested for each trial. Duplicate and triplicate trials were performed on select
234 feedstock to understand repeatability, it was found that liquid product yield varied by less than
235 2% demonstrating very good experimental repeatability. To accommodate the material and
236 provide sufficient mixing, the reactor auger was maintained at a rotation frequency of 50 RPM.
237 Throughout the experiment, high purity nitrogen was passed through the system at a rate of at
238 least 0.24 standard liters per minute to maintain an inert system and ensure short vapor phase
239 times in the transfer line. The system temperature profiles and other characteristics are described
240 in the Results and Discussion section below.
241 After concluding the experiment and allowing the system to cool to <50°C, the unit was
242 opened for cleaning and product recovery. Between each trial, the material feed hoppers were
243 completely emptied of material and weighed to factor into the mass balance. The system was
244 also opened and cleaned at the various drop tubes, transfer lines, seals etc. for collection of
245 fugitive materials to refine the trial’s mass balance. Typical values for collected unreacted solids
246 were less than one percent. The solids collector was weighed before and after each trial in
247 addition to the solids recovered from the collector. Small amounts of char and HTM (<1% of the
248 combined masses) were also collected and considered from the tube connecting the solid
249 collector to the reactor. Total liquid yield was comprised of the sum of the oil collected in the
250 condenser bottle (>85% liquid mass), the difference in weight of the water bubbler/scrubber
251 before and after a trial (~8% liquid mass), and oil residues remaining on the condenser tubes and
252 transfer line. As the high-purity nitrogen (dry) passed through the liquid impinger system, some
253 moisture was carried away with the permanent gas stream. Off-line tests were performed with
254 only the nitrogen/impinger system and the water mass was found to change by approximately
255 0.69 g/h. The nitrogen stream during an experiment should be somewhat saturated with water
256 and organics from the reaction chemistry, and the amount of mass loss in the water scrubber
257 should be decreased during actual trials as the single condenser system (~0 °C) cannot
258 effectively remove all moisture/organics from the gas stream (as mentioned above, ~8% of
259 recovered liquid is captured in the scrubber). This water mass loss was not accounted for in the
260 mass balance due to the relatively low magnitude (1-2%), and difficulty in accurately measuring
261 the loss during experiments. Overall the liquid recovery values presented in this work are bias

6
262 low due to this phenomena, however the loss in mass from the water bath was considered
263 negligible compared to the overall recoverable condensate masses (150-250 g).
264
265 3 Results and Discussion
266 3.1 Particle Flow Analysis
267 Weight transients of biomass as a function of time was measured by the weighing system
268 with data acquisition rate of 1-10 Hz. Calibration of the two dosing augers were done by
269 selecting a feedstock and measuring the feed rates as described above as functions of the auger
270 shaft rotation frequency. The mass flow rates for the HTM (sand) with results fit to a straight
271 line with a slope of 412 g x h-1 x RPM-1 (R2 = 0.998), and 41.7 g x h-1 x RPM-1 (R2 = 0.999) for
272 mixed sawdust. Because the mass flow rate depends on material density and other physical
273 characteristics, each biomass type at the specific particle size distribution was calibrated
274 separately, although those results are not shown here. After the dosing augers were calibrated,
275 the approximate feed rates of the materials were controlled by setting the respective shaft
276 rotation velocities to the target values.
277 Figure 2 shows schematics of the system with the respective rotation frequencies (ν) and
278 the residence times (t) nomenclature for each component. The system has four characteristic
279 residence times: tHTM and tbio, which are the residence times for the HTM and biomass augers,
280 respectively, and theat, and tpyr, which are the residence times in the heating and pyrolysis zones,
281 respectively. The sum of latter two is the residence time in the reactor. The system also has
282 three rotation frequencies as variables: νHTM and νbio are the frequency of the HTM and biomass
283 dosing augers respectively; and νreactor, which is the rotation frequency of the reactor for both the
284 heating and pyrolysis zones. The figure also shows the balance that measures the mass of
285 biomass coming out of the reactor outlet.
286 The residence time in a regular screw is approximately given by
287 = (1)
288 where and c are the rotation frequency and the number of pitches in the auger. Numerous
289 experiments (see supplemental material) were done with the two dosing augers (HTM and
290 biomass) to determine that cHTM and cbio were determined to equal 5.72 pitches for both augers.
291 Because of the complex structure of the auger-paddle reactor, Eq. (1) must be modified to
292 correctly predict material residence time in the reactor. In fact, the residence time in the reactor
293 should depend on two main factors: (i) the percent filling in the reactor, which in turns depends
294 linearly on the rotation frequency of the dosing auger; and (ii) the rotation frequency of the
295 reactor. The reactor has two characteristic residence times−one that relates to the entire reactor,
296 treactor (see Figure 2) and another that relates to the pyrolysis-zone time, tpyr (see Figure 2). An
297 empirical correlation was assumed for the dependence of each of the residence times on the
298 respective dosing auger rotation frequencies (in other words the percent filling of the reactor) and
299 the rotation frequency of the reactor, analogous to that of Eq. (1) by:
( , )
300 = (2)
301 where ( , !"# ) is analogous number of pitches of the reactor that depends on the
302 HTM and respective frequencies. The parameter ( , !"# ) can also be called the
303 effective number of pitches of the reactor. Because the effective number of pitches is dependent
304 on rotation frequency, a similar relation (inverse or power-law model) to those in Eq. (1) and (2)
305 was assumed as:
( %
306 , !"# ) = $ !"# (3)

7
307 where areactor is a constant, and & and ' are exponents related to the dosing and reactor
308 frequencies respectively. Introducing Eq. (3) in (2) yields
% ()
309 = $ !"# (4)
310 Equation (4) has three parameters, ', &, and $, which can be determined experimentally
311 by measuring treactor at various reactor and HTM frequencies and carrying out a non-linear
312 regression fitting procedure. To determine these parameters, the following experiment was
313 conducted: (i) the HTM was filled when it was stopped; (ii) the reactor auger was turned on
314 continuously; (iii) the HTM auger and balance were turned on simultaneously. The rotation
315 frequency of the HTM auger was changed in the range 2-18 rpm and reactor frequency was
316 varied in the range 20-200 rpm. Figure 3(A) shows the measured residence time results in the
317 reactor, treactor, vs. the calculated time by Eq. 4 for best fit values which were a=9.68, m=-0.25,
318 and n=0.50. Thus the effective number of pitches in the entire reactor equals (where ceff is
319 dimensionless)
( (../0 ..0
320 , !"# ) = 9.68 !"# (5)
321 The same analysis was conducted for the pyrolysis zone with similar power terms,
322 however with a different scalar. The effective number of pitches in the pyrolysis zone was found
323 to be
324 12 ( , !"# ) = 3.67 5(../0 ..0 (6)
325 The residence time data for the entire reactor and the pyrolysis zone were plotted in
326 Figure 3(B) and show a perfect fit to a straight line with a slope of almost unity. It is important to
327 note that the constant, $, likely can be further defined to contain an additional geometry
328 parameter, such as auger length or the ratio of length to diameter. In comparison of equations (5)
329 and (6), the length of the reactor is approximately three times the length of the pyrolysis zone
330 and could account for a majority of the difference in the constant value. This is a strong
331 indication that the flow of material within the reactor reached orderly and steady flow at a much
332 shorter distance than the reactor length. In other words, the residence time within the reactor is
333 linear with the reactor axial coordinate. For the presented system geometry, the approximate
334 length to diameter ratio for the respective zones are 18 and 6. If this additional term is included,
335 the value of the constant reduces to 0.54 and 0.61 for eqn. (5) and (6) respectively. Due to the
336 closeness of the fit constant values and the fixed geometry, it may be reasonable to use an
337 average constant of 0.58 for both and include the ratio described above.
338 As expected, the effective number of pitches depends on both frequencies, however they
339 are varied in the range of 33-88 pitches. In fact, the effective number of pitches sheds light on
340 the mixing quality, and hence the heat transfer rate. Clearly, higher the effective number of
341 pitches implies better mixing and higher heating rates. One can define a new parameter that
342 perhaps can be used to determine the mixing quality as the ratio of theoretical effective pitches in
343 ratio with the actual number of pitches for the geometry. This can be represented as:
344 6= ( , !"# )/ 8 9 (7)
345 or by introducing Eq. (3) into (7) yields
346 6=$ ' &−1
:;< =>$ ?= / 8 9 (8)
347 where cactual is the number of pitches of the designed configuration.
348 The parameters ceff (Eq. 3) and p (Eq. 8) referred to as the pitch number, can probably
349 serve as a measure of quality of mixing because higher p values indicate better material
350 retention, and mixing. The values of cactual in this particular system is 9 pitches; ceff and p were
351 calculated by Eqs. (3) and (7), respectively and results for p are shown in Figure 4. The value ceff
352 varies in the range 33-104 for HTM shaft frequency in the range 3-18 rpm and reactor shaft

8
353 frequency in the range 50-200 rpm. Respectively, p varies in the range 3.7-12 for the same
354 conditions, which indicates a significant increase of the mixing propensity of the paddle
355 configuration. The value of p also indicates a ratio of the equivalent axial length (and thus
356 equivalent time) solid particles would spend in a traditional screw reactor at the same rotation
357 frequency. A value greater than unity suggests that the design footprint of the reactor can be
358 significantly reduced, while simultaneously enhancing the time solid particles mix and comingle.
359 The results presented here suggest that given a constrained residence time (such as that required
360 for pyrolysis), the size of a reactor can be reduced by up to an order of magnitude at high rotation
361 frequency (material throughput) compared to traditional screws. Additional data on the analysis
362 are shown in the supplemental material.
363
364 3.2 System Thermal Characteristics
365 In order to demonstrate that this reactor configuration can indeed perform fast pyrolysis,
366 it is essential to understand the thermal characteristics of the system. As indicated above, the
367 heating zone was kept at a set point of 550 oC (T1-T8 in Figure 2), while the pyrolysis zone was
368 maintained at 500 oC (T9-T12 in Figure 2). Table 1 shows: (i) the set points (second column) at
369 the two zones; (ii) the temperature after heating was turned on and system reached steady-state
370 (third column), showing a stable temperature of approximately +0.1 oC and negligible difference
371 from the set points except in a few points; (iii) the temperatures after flowing the HTM and
372 reaching a second steady state, showing negligible changes in the temperatures (this is of course
373 attributed to the fact that the heating bands were capable of delivering sufficient heat); (iv) after
374 feeding the biomass and reaching a third thermal steady state again showing negligible changes
375 in the temperatures in all zones.
376 Figure 5(A) is a graphical portrayal of the temperature of thermocouples 1-8 (top) and
377 duty-cycle of heaters 1-8 (bottom) in the preheating zone. We note the following: (i) A start-up
378 period, starting at t=0, where only the metal is being heated with no material flowing, thermal
379 steady-state was achieved after about (both temperatures and duty cycles) was stabilized about
380 1.50 hours. (ii) We kept the system stable for about 1.2 hours which is indicated by very stable
381 temperatures and duty cycles in all zones. (iii) At t ~ 2.6 hr HTM was introduced, causing a
382 sudden drop in temperatures of T1 and immediate increase of duty-cycle of heater 1. Other
383 zones were affected, but not as much. The first heater accounted for greater than 80% of the
384 power supply to heat the HTM. (iv) After about 20 minutes the system reached again thermal
385 steady-state, noticed by constant temperatures and duty cycles of all heaters. (v) At t ~ 3.4 hr
386 biomass was flown with very small changes both in temperatures and duty cycles. (vi) After
387 about 40 minutes, the system reached again thermal steady-state, noted by no change in
388 temperatures and duty cycles.
389 Figure 5(B) is a graphical portrayal of the temperature of thermocouples 9-12 (top) and
390 duty-cycle of heaters 9-12 (bottom) in the pyrolysis zone. The behavior here follows that in the
391 preheat zone, except that the changes are very small: At t ~ 2.6 hr HTM was introduced, causing
392 a small drop in the temperature of T9 and small changes in duty-cycles of the heaters. At t ~ 3.4
393 hr biomass was flown with very small changes both in temperatures and duty cycles.
394 One can gain insight into the heating rate between the HTM and the biomass by studying
395 the behavior of the duty cycle for each heater. In fact, what is important in examining the duty
396 cycles of each heater, is not the absolute value of the duty cycle, but the difference between the
397 final steady-state when both HTM as well as biomass are introduced, and the initial steady-state
398 for each heater without material. Heat losses from the reactor to the surroundings vary

9
399 significantly in different regions of the reactor; the heaters compensate for these heat losses by
400 operating in different duty cycles to maintain constant temperatures in the various regions in the
401 reactor. Once the temperature profiles reach steady-state, the heat losses depend only on the
402 temperature differences between reactor and surroundings. Therefore, the difference between
403 the final steady-state value of the duty cycle in each heater and its initial steady-state value
404 depends only on the requirement to heat the material in the specific location. Figure 6 shows
405 typical results for the duty cycle difference vs. the location of the heater. The difference in duty
406 cycle in heater 1 is significant because this is where the HTM is heated from ambient
407 temperature to the required 550 oC (we note that material at this location did indeed reach this
408 temperature as shown in Figure 5 and Table 1). Heaters 2-7 display little contribution to the
409 overall power consumption for heating the HTM. This means that the first heater was mostly
410 responsible for heating the HTM. To help follow the results an exponential decay was fit. The
411 difference in duty cycle of Heater 8 increases to about 8% and that of 9 increased by about 3%.
412 The difference in duty cycles for heater 9 is ~ 3.1+0.6%. Heaters 10-12 display zero duty cycle
413 difference. The important implication of these interesting results is the heating rate of the
414 biomass. The fact that the increase in the duty cycle differences in Heaters 8 and 9 are very
415 small is not surprising because the HTM provides sufficient heat to increase the temperature of
416 the biomass from ambient to 500 oC. What is important to indicate, is that heater 8 provided
417 >80% of the required heat (power consumption of heater 8 divided by the total power
418 consumption of pyrolysis zone heaters) for the incoming biomass, and was fully heated after
419 heater 9. Recalling from the design discussed above, biomass is fed between heater 8 and 9, and
420 is therefore fully heated when it is measured after heater 9 (approximately 3.2 cm). This can be
421 roughly translated into time (see detailed residence time analysis above) into approximately 4.3
422 seconds (~0.73 cm/sec axial reactor velocity at 50 RPM in the reactor) required to mix and heat
423 the surface of the biomass particles from ambient temperature to 500 oC. This yields a heating
424 rate of approximately 110° C/s (approximately 6600 °C/min), which is significantly faster than
425 the required heating rate for fast pyrolysis (~15-20° C/s).
426
427 3.3 Pyrolysis and Product Mass Balance
428 As mentioned above, the main objective of the present paper is to demonstrate the
429 applicability of our proposed reactor system to achieve fast pyrolysis. We selected to compare
430 our pyrolysis results with those of NREL’s 2 inch fluidized bed reactor (2FBR) using the same
431 biomass feedstocks (Howe et al., 2013; Carpenter et al., 2016). Comparison was made only for
432 the measured char and bio-oil yields. Pyrolysis of the ten tested biomass feedstock was carried
433 out with the following conditions: 2.5-3 RPM for HTM (Sand) yielding approximately 1,500
434 g/h, and 3-4 RPM of biomass yielding approximately 100g/h flow of biomass. The reactor was
435 maintained at 50 RPM for all trials. The target conversion temperature for all feedstocks was
436 500°C. This is a ratio of 15 mass units of sand to 1 mass unit of biomass. These values are
437 consistent with those used by (Brown and Brown 2012) (18:1) and by (Henrich et al. 2016), and
438 (Pfitzer et al. 2016) (11:1). We selected a ratio of 15:1 because this corresponded to a particle
439 ratio of 5:1 under the studied sand and biomass sizes. Figure 7 displays char and bio-oil yields
440 plotted vs. NREL’s results for the same materials (Westover et al. 2013; Howe et al. 2015;
441 Carpenter et al., 2016). The yield of char and liquid (char yield %, liquid yield %) for
442 switchgrass, cornstover, clean pine, thermally treated pine, blend (clean pine, tulip poplar,
443 switchgrass), pinion-juniper, tulip poplar, hybrid poplar, C&D waste, and miscanthus were
444 (19.7,64.8), (8.6, 56.8), (10.8, 64.6), (33.2, 39.6), (13.2, 58.5), (12.1, 66.0), (8.5, 66.7), (7.6,

10
445 61.6), (11.3, 59.6), and (11.4, 58.1) respectively. The species with the highest liquid yield was
446 tulip poplar (66.7%), while the lowest yield was obtained from the thermally treated pine
447 (39.6%), followed by corn stover (56.8%). The highest char production was observed from the
448 thermally treated pine (33.2%) followed by switchgrass (19.7%), while the lowest char was
449 observed from hybrid poplar (7.6%) The solid line in the plot represents a perfect fit (slope of 1).
450 The dashed line is the best linear regression fit of the results with a slope of 1.00 and an intercept
451 of 0.02, indicating an excellent fit between our results and those of NREL’s for the same
452 materials. This fit represents the entire measured mass balance, meaning the fit includes data for
453 both the dry, ash free char and liquid product values. If the fit is restricted to the char alone, the
454 fit parameters are 1.02 and -0.02 respectively. If the fit is restricted to the liquid product alone,
455 the fit parameters are 0.87 and 0.06 respectively. Importantly, comparing the results from
456 replicates tests from NREL’s 2FBR (Howe et al.; Carpenter et al.; as well as unpublished results)
457 indicates that oil yields of individual tests can vary by 4% or more, so that a substantial portion,
458 and perhaps even most of the disagreement between results observed from the two systems is
459 likely due to replication error of the 2FBR.
460 We also carried out trial pyrolysis runs without HTM, which are described in the
461 supplemental materials. If the geometry of the mixing reactor is able to use the solid
462 biomass/char particles, with forward and backward mixing to effectively act as an HTM, then
463 several industry relevant concerns can be avoided (discussed below).
464
465 3.4 Scaling-up considerations
466 The dimensionless Froude number (Fr) relates the forces exerted on particles to
467 gravitational forces. In reactor engineering, it is an important number to understand reactor
468 design and scaling. For example, the Froude number can be used in fluidized bed reactors to
469 determine the minimum sweep gas velocity required to ensure particles act as a 'fluid'. At this
470 stage, the forces exerted by the sweep gas is approximately the same as the gravitational forces.
471 When the Froude number is the region of 1-3, the particles can be considered fluidized. At
472 higher values, substantial turbulent blow-off is expected, and lower values would result in a more
473 'aggregate' or 'granular' system flows.
474 Radial mixing in a mechanically rotating system can be described by a form of the
475 dimensionless Froude Number of the form (Henrich et al. (2016)):
476 B= = (2 )2=/D (9)
477 where ν is the shaft rotational frequency, r is the outer screw radius, and g is gravitational
478 acceleration. The Froude number is a critical parameter to consider for scaling of a well-mixed
479 system such as the one presented here. Henrich et al. (2016) used the Froude number to examine
480 mixing in a twin screw reactor indicating that good mixing is achieved when the centrifugal
481 forces at the outer screw radius and gravitational forces of the heat carrier are effectively equal
482 and the radial velocity of biomass particles are that of the rotating screw. In that study Fr = 1 was
483 used as an indication for good mixing. For example, for a 0.35 m diameter rotator used in a
484 mechanical mixer, the rotation frequency should be 3.71 Hz to obtain Fr = 1, or that the
485 minimum rotational speed that should be considered for the solid particles to be considered 'free
486 flowing' or fluidized is approximately 220 RPM.
487 The chemical industry uses paddle mixers because of their ability to effectively mix two
488 or more particle streams to a very uniform blend. In fact, we tested for mixing quality of a
489 HSB1484 (Scott Equipment Company) paddle mixer of 0.35 m diameter and 2.1 m long mixer
490 that can rotate its shaft up to 600 rpm, and a residence time of 10 seconds, which can convey

11
491 1200 kg/h biomass. The Froude number at these conditions is about 7, which is indicative of
492 excellent mixing. This mixer was tested by flowing two streams of biomass materials into the
493 mixer, one stream was oven bone dry the other had 50% moisture content. Thirty samples (~1.0
494 g) were collected from various location and measured for moisture content, resulting in an
495 average moisture content half of the dry biomass with a relative standard deviation of +2%,
496 which is indicative to excellent mixing. The purpose of testing the Scott paddle mixer was to
497 ensure that a device similar to the one proposed in this study is indeed scalable, as was proven by
498 the results.
499
500 3.5 Attrition
501 One of the major problems of the current mode of operation is the use of silica sand as an
502 HTM. It is well known that this material causes significant attrition of the reactors walls, shaft,
503 flightings and in time imposes drastic degradation that require maintenance (Dahmen et al. 2012;
504 Henrich et al. 2016). In fact, after operation of about 1000 hours of this system, the shaft was
505 replaced with a new one. We are currently working with other materials to reduce attrition, and
506 to optimize reactor conditions to significantly reduce or eliminate the need for HTM. Sand was
507 used in these tests to facilitate comparison but is not intended for future tests at larger scales.
508 Preliminary trials without HTM are displayed in the supplemental materials. With further
509 optimization of the mixing/scaling discussion above, future work will focus on pyrolysis liquid
510 yields in optimized scenarios.
511
512 4. Summary and Conclusions
513 We have developed a pyrolysis reactor system that is based on mechanical conveying
514 using a unique auger-paddle configuration with HTM as an essential part for achieving high
515 mixing rates and as a consequence high heating rates. This, in turn, enables fast biomass
516 pyrolysis. In this study the focus was on the proof of concept and the characterization of
517 residence times and heating rates. Residence times were measured in the reactor and a
518 correlation was developed for the residence time and reactor shaft frequency. Further, we have
519 conducted a comprehensive thermal characterization of the system and showed that heating rates
520 are approximately 110 oC/s which is significantly faster than those required for fast pyrolysis.
521 We tested ten different biomass types and obtained char and bio-oil yields similar to those
522 obtained by NREL using a circulating fluidized bed with the same feedstocks. Thus, our system
523 with the current operation mode has high heating rates that enable fast pyrolysis.
524
525 Acknowledgments
526 The authors would like to acknowledge the U.S. Endowment for Forestry and Communities and
527 its project known as the Consortium for Advanced Wood to Energy Solutions (CAWES, grant
528 #CW-4), and the NSF-STTR (grant # 1412015). We are grateful to the APS Labs, Mechanical
529 Engineering-Engineering Mechanics Department, Michigan Technological University for
530 providing us with their facility, and lab equipment’s to carry out this research. This manuscript
531 has been coauthored by Battelle Energy Alliance, LLC under Contract No. DE-AC07-
532 05ID14517 with the U.S. Department of Energy. The United States Government retains and the
533 publisher, by accepting the article for publication, acknowledges that the United States
534 Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or

12
535 reproduce the published form of this manuscript, or allow others to do so, for United States
536 Government purposes.
537
538 References
539 Agirre, I., T. Griessacher, G. Rosler, and J. Antrekowitsch. 2013. “Production of Charcoal as an
540 Alternative Reducing Agent from Agricultural Residues Using a Semi-Continuous Semi-
541 Pilot Scale Pyrolysis Screw Reactor.” Fuel Processing Technology 106 (July 2016): 114–
542 21. doi:10.1016/j.fuproc.2012.07.010.
543 Aramideh, Soroush, Qingang Xiong, Song Charng Kong, and Robert C. Brown. 2015.
544 “Numerical Simulation of Biomass Fast Pyrolysis in an Auger Reactor.” Fuel 156. Elsevier
545 Ltd: 234–42. doi:10.1016/j.fuel.2015.04.038.
546 Aylón, E., A. Fernández-Colino, R. Murillo, M. V. Navarro, T. García, and A. M. Mastral. 2010.
547 “Valorisation of Waste Tyre by Pyrolysis in a Moving Bed Reactor.” Waste Management
548 30 (7). Elsevier Ltd: 1220–24. doi:10.1016/j.wasman.2009.10.001.
549 Aylón, E., A. Fernández-Colino, M. V. Navarro, R. Murillor, T. García, and A. M. Mastral.
550 2008. “Waste Tire Pyrolysis: Comparison between Fixed Bed Reactor and Moving Bed
551 Reactor.” Industrial and Engineering Chemistry Research 47 (12): 4029–33.
552 doi:10.1021/ie071573o.
553 Balat, Mustafa, Mehmet Balat, Elif Kirtay, and Havva Balat. 2009. “Main Routes for the
554 Thermo-Conversion of Biomass into Fuels and Chemicals. Part 1: Pyrolysis Systems.”
555 Energy Conversion and Management 50 (12). Elsevier Ltd: 3147–57.
556 doi:10.1016/j.enconman.2009.08.014.
557 Bosong, Li, Lv Wei, Zhang Qi, Wang Tiejun, and Ma Longlong. 2014. “Pyrolysis and Catalytic
558 Upgrading of Pine Wood in a Combination of Auger Reactor and Fixed Bed.” Fuel 129.
559 Elsevier Ltd: 61–67. doi:10.1016/j.fuel.2014.03.043.
560 Botten, A. J., A. S. Burbidge, and S. Blackburn. 2003. “A Model to Predict the Pressure
561 Development in Single Screw Extrusion.” Journal of Materials Processing Technology 135
562 (2–3 SPEC.): 284–90. doi:10.1016/S0924-0136(02)00859-2.
563 Bridgwater, A. V. 2003. “Renewable Fuels and Chemicals by Thermal Processing of Biomass.”
564 Chemical Engineering Journal 91 (2–3): 87–102. doi:10.1016/S1385-8947(02)00142-0.
565 Bridgwater, A. V., D Meier, and D Radlein. 1999. “An Overview of Fast Pyrolysis of Biomass.”
566 Org. Geochem. 30 (12): 1479–93. doi:10.1016/S0146-6380(99)00120-5.
567 Briens, Cedric, Jan Piskorz, and Franco Berruti. 2008. “Biomass Valorization for Fuel and
568 Chemicals Production -- A Review.” International Journal of Chemical Reactor
569 Engineering 6 (1). doi:10.2202/1542-6580.1674.
570 Brown, J. N., and R. C. Brown. 2012. “Process Optimization of an Auger Pyrolyzer with Heat
571 Carrier Using Response Surface Methodology.” Bioresource Technology 103 (1). Elsevier
572 Ltd: 405–14. doi:10.1016/j.biortech.2011.09.117.

13
573 Butler, Eoin, Ger Devlin, Dietrich Meier, and Kevin McDonnell. 2011. “A Review of Recent
574 Laboratory Research and Commercial Developments in Fast Pyrolysis and Upgrading.”
575 Renewable and Sustainable Energy Reviews 15 (8). Elsevier Ltd: 4171–86.
576 doi:10.1016/j.rser.2011.07.035.
577 Chen, Yen Chang, and Yung Ning Pan. 2011. “Effects of Grain Size and Rotational Speed on the
578 Bio-Oil Yield in a Fast Pyrolysis Reactor Employing a Single Tapered Screw Extruder.”
579 Advanced Materials Research 347–353: 153–56.
580 doi:10.4028/www.scientific.net/AMR.347-353.153.
581 Chun, Young Nam, Seong Cheon Kim, and Kunio Yoshikawa. 2011. “Pyrolysis Gasification of
582 Dried Sewage Sludge in a Combined Screw and Rotary Kiln Gasifier.” Applied Energy 88
583 (4). Elsevier Ltd: 1105–12. doi:10.1016/j.apenergy.2010.10.038.
584 Dahmen, Nicolaus, Edmund Henrich, Eckhard Dinjus, and Friedhelm Weirich. 2012. “The
585 Bioliq® Bioslurry Gasification Process for the Production of Biosynfuels, Organic
586 Chemicals, and Energy.” Energy, Sustainability and Society 2: 1–44. doi:10.1186/2192-
587 0567-2-3.
588 Day, M., J. D. Cooney, and Z. Shen. 1996. “Pyrolysis of Automobile Shredder Residue: An
589 Analysis of the Products of a Commercial Screw Kiln Process.” Journal of Analytical and
590 Applied Pyrolysis 37 (1): 49–67. doi:10.1016/0165-2370(96)00938-2.
591 Day, M., Z. Shen, and J. D. Cooney. 1999. “Pyrolysis of Auto Shredder residue: Experiments
592 with a Laboratory Screw Kiln Reactor.” Journal of Analytical and Applied Pyrolysis 51:
593 181–200.
594 Dupont, C., Chiriacb, R., Gauthiera, G., Toche, F., 2014. Heat capacity measurements of various
595 biomass types and pyrolysis residues. Fuel, 115: 644-651.
596 Efika, Chidi E., Chunfei Wu, and Paul T. Williams. 2012. “Syngas Production from Pyrolysis-
597 Catalytic Steam Reforming of Waste Biomass in a Continuous Screw Kiln Reactor.”
598 Journal of Analytical and Applied Pyrolysis 95. Elsevier B.V.: 87–94.
599 doi:10.1016/j.jaap.2012.01.010.
600 Ferreira, Suelem Daiane, Carlos Roberto Altafini, Daniele Perondi, and Marcelo Godinho. 2015.
601 “Pyrolysis of Medium Density Fiberboard (MDF) Wastes in a Screw Reactor.” Energy
602 Conversion and Management 92. Elsevier Ltd: 223–33.
603 doi:10.1016/j.enconman.2014.12.032.
604 Gao, Hao Jie, Yue Zhao Zhu, Hai Jun Chen, Chuan Hua Liao, Yang Du, and Hao Wu. 2014.
605 “Pyrolysis of Hailar Lignite: Experiments with a Tubular Reactor and a Continuous Screw
606 Kiln Reactor.” Applied Mechanics and Materials 672–674: 665–71.
607 doi:10.4028/www.scientific.net/AMM.672-674.665.
608 Garcia-Perez, Manuel, Thomas T. Adams, John W. Goodrum, Daniel Geller, and K. C. Das.
609 2007. “Production and Fuel Properties of Pine Chip Bio-Oil/biodiesel Blends.” Energy and
610 Fuels 21 (4): 2363–72. doi:10.1021/ef060533e.
611 Haydary, J., D. Susa, and J. Dudas. 2013. “Pyrolysis of Aseptic Packages (Tetrapak) in a
612 Laboratory Screw Type Reactor and Secondary Thermal/catalytic Tar Decomposition.”

14
613 Waste Management 33 (5). Elsevier Ltd: 1136–41. doi:10.1016/j.wasman.2013.01.031.
614 Haydary, J., D. Susa, V. Gelinger, and F. Čacho. 2016. “Pyrolysis of Automobile Shredder
615 Residue in a Laboratory Scale Screw Type Reactor.” Journal of Environmental Chemical
616 Engineering 4 (1): 965–72. doi:10.1016/j.jece.2015.12.038.
617 Henrich, E., N. Dahmen, F. Weirich, R. Reimert, and C. Kornmayer. 2016. “Fast Pyrolysis of
618 Lignocellulosics in a Twin Screw Mixer Reactor.” Fuel Processing Technology 143.
619 Elsevier B.V.: 151–61. doi:10.1016/j.fuproc.2015.11.003.
620 Howe, Daniel, Tyler Westover, Daniel Carpenter, Daniel Santosa, Rachel Emerson, Steve
621 Deutch, Anne Starace, Igor Kutnyakov, and Craig Lukins. 2015. “Field-to-Fuel
622 Performance Testing of Lignocellulosic Feedstocks: An Integrated Study of the Fast
623 Pyrolysis-Hydrotreating Pathway.” Energy and Fuels 29: 3188-97.
624 doi:10.1021/acs.energyfuels.5b00304.
625 Ingram, Leonard, Dinesh Mohan, Mark Bricka, Philip Steele, David Strobel, Brian Mitchell,
626 Javeed Mohammad, et al. 2008. “Pyrolysis of Wood and Bark in an Auger Reactor :
627 Physical Properties and Chemical Analysis of the Produced Bio-Oils Pyrolysis of Wood and
628 Bark in an Auger Reactor : Physical Properties and Chemical Analysis of the Produced Bio-
629 Oils,” Energy and Fuels 15: 614–25. doi:10.1021/ef700335k.
630 Kelkar, Shantanu, Christopher M. Saffron, Li Chai, Jonathan Bovee, Thomas R. Stuecken,
631 Mahlet Garedew, Zhenglong Li, and Robert M. Kriegel. 2015. “Pyrolysis of Spent Coffee
632 Grounds Using a Screw-Conveyor Reactor.” Fuel Processing Technology 137 (July 2016):
633 170–78. doi:10.1016/j.fuproc.2015.04.006.
634 Liaw, Shi Shen, Zhouhong Wang, Pius Ndegwa, Craig Frear, Su Ha, Chun-Zhu Li, and Manuel
635 Garcia-Perez. 2012. “Effect of Pyrolysis Temperature on the Yield and Properties of Bio-
636 Oil Obtained from the Auger Pyrolysis of Douglas Fir Wood” Journal of Analytical and
637 Applied Pyrolysis 93: 52–62.
638 Liaw, Shi Shen, Shuai Zhou, Hongwei Wu, and Manuel Garcia-Perez. 2013. “Effect of
639 Pretreatment Temperature on the Yield and Properties of Bio-Oils Obtained from the Auger
640 Pyrolysis of Douglas Fir Wood.” Fuel 103. Elsevier Ltd: 672–82.
641 doi:10.1016/j.fuel.2012.08.016.
642 Lin, L., S.J. Khang, and T.C. Keener. 1997. “Coal Desulfurization by Mild Pyrolysis in a Dual-
643 Auger Coal Feeder.” Fuel Processing Technology 53 (1): 15–29.
644 doi:10.1080/15435070701583110.
645 Martinez, Juan Daniel, Ramon Murillo, Tomas Garcia, and Alberto Veses. 2013. “Demonstration
646 of the Waste Tire Pyrolysis Process on Pilot Scale in a Continuous Auger Reactor.” Journal
647 of Hazardous Materials 261. Elsevier B.V.: 637–45. doi:10.1016/j.jhazmat.2013.07.077.
648 McGee, Eddie. 2012. “Continuous Combustion: Plug Screw Feeder Technology for Biomass
649 Pyrolysis System.” Bulk Solids Handling 32 (2): 44–46.
650 Miao, Zewei, Tony E. Grift, Alan C. Hansen, and K.C. Ting. 2014. “Flow Performance of
651 Ground Biomass in a Commercial Auger.” Powder Technology 267: 354–61.
652 doi:10.1016/j.powtec.2014.07.038.

15
653 Mohan, Dinesh, Charles U. Pittman, Mark Bricka, Fran Smith, Ben Yancey, Javeed Mohammad,
654 Philip H. Steele, Maria F. Alexandre-Franco, Vicente Gomez-Serrano, and Henry Gong.
655 2007. “Sorption of Arsenic, Cadmium, and Lead by Chars Produced from Fast Pyrolysis of
656 Wood and Bark during Bio-Oil Production.” Journal of Colloid and Interface Science 310
657 (1): 57–73. doi:10.1016/j.jcis.2007.01.020.
658 Mohan, Dinesh, Charles U Pittman, and Philip H Steele. 2006. “Pyrolysis of Wood / Biomass for
659 Bio-Oil : A Critical Review.” Energy & Fuels 20 (4): 848–89. doi:10.1021/ef0502397.
660 Morgano, M.T., H. Leibold, F. Richter, and H. Seifert. 2015. “Screw Pyrolysis with Integrated
661 Sequential Hot Gas Filtration.” Journal of Analytical and Applied Pyrolysis 113: 216–24.
662 Pfitzer, Cornelius, Nicolaus Dahmen, Nicole Troeger, Friedhelm Weirich, Jörg Sauer, Armin
663 Günther, and Matthias Müller-Hagedorn. 2016. “FAST PYROLYSIS OF WHEAT STRAW
664 IN THE BIOLIQ® PILOT PLANT.” Energy & Fuels 30 (10): 8047–54.
665 doi:10.1021/acs.energyfuels.6b01412.
666 Pittman, Charles U., Dinesh Mohan, Anthonia Eseyin, Qi Li, Leonard Ingram, El Barbary M
667 Hassan, Brian Mitchell, Hua Guo, and Philip H. Steele. 2012. “Characterization of Bio-Oils
668 Produced from Fast Pyrolysis of Corn Stalks in an Auger Reactor.” Energy and Fuels 26
669 (6): 3816–25. doi:10.1021/ef3003922.
670 Promdee, K, and T Vitidsant. 2014. “Applied Thermal Pyrolysis of Cogongrass in Twin Screw
671 Reactor 1” 61 (8): 612–17. doi:10.1134/S0040601514080102.
672 Promdee, Kittiphop, and Tharapong Vitidsant. 2013. “Synthesis of Char, Bio-Oil and Gases
673 Using a Screw Feeder Pyrolysis Reactor.” Coke and Chemistry 56 (12): 466–69.
674 doi:10.3103/S1068364X13120107.
675 Puy, Neus, Ramón Murillo, María V. Navarro, José M. López, Joan Rieradevall, G. Fowler,
676 Ignacio Aranguren, Tomás García, Jordi Bartrolí, and Ana M. Mastral. 2011. “Valorisation
677 of Forestry Waste by Pyrolysis in an Auger Reactor.” Waste Management 31 (6). Elsevier
678 Ltd: 1339–49. doi:10.1016/j.wasman.2011.01.020.
679 Roedig, M.-K., and W. Klose. 2009. “Modelling of Coal Pyrolysis Using a Twin Screw
680 Reactor.” Revue de Metallurgie 106 (10): 404–9.
681 Sirijanusorn, Somsak, Keartisak Sriprateep, and Adisak Pattiya. 2013. “Pyrolysis of Cassava
682 Rhizome in a Counter-Rotating Twin Screw Reactor Unit.” Bioresource Technology 139.
683 Elsevier Ltd: 343–48. doi:10.1016/j.biortech.2013.04.024.
684 Venderbosch, RH, and W Prins. 2010. “Fast Pyrolysis Technology Development.” Biofuels,
685 Bioproducts and Biorefining 4 (2). John Wiley & Sons, Ltd.: 178–208.
686 doi:10.1002/bbb.205.
687 Veses, A., M. Aznar, J. M. López, M. S. Callén, R. Murillo, and T. García. 2015. “Production of
688 Upgraded Bio-Oils by Biomass Catalytic Pyrolysis in an Auger Reactor Using Low Cost
689 Materials.” Fuel 141. Elsevier Ltd: 17–22. doi:10.1016/j.fuel.2014.10.044.
690 Veses, A., M. Aznar, I. Martinez, J. D. Martinez, J. M. Lopez, M. V. Navarro, M. S. Callen, R.
691 Murillo, and T. Garcia. 2014. “Catalytic Pyrolysis of Wood Biomass in an Auger Reactor
692 Using Calcium-Based Catalysts.” Bioresource Technology 162. Elsevier Ltd: 250–58.
16
693 doi:10.1016/j.biortech.2014.03.146.
694 Westover, Tyler L, Manunya Phanphanich, Michael L Clark, Sharna R Rowe, Steven E Egan,
695 Alan H Zacher, and Daniel Santosa. 2013. “Impact of Thermal Pretreatment on the Fast
696 Pyrolysis Conversion of Southern Pine.” Biofuels 4 (1): 45–61. doi:10.4155/bfs.12.75.
697 Zhou, Rui, Hanwu Lei, and James Julson. 2013. “The Effects of Pyrolytic Conditions on
698 Microwave Pyrolysis of Prairie Cordgrass and Kinetics.” Journal of Analytical and Applied
699 Pyrolysis 101. Elsevier B.V.: 172–76. doi:10.1016/j.jaap.2013.01.013.
700 Zhou, Shuai, Daniel Mourant, Caroline Lievens, Yi Wang, Chun Zhu Li, and Manuel Garcia-
701 Perez. 2013. “Effect of Sulfuric Acid Concentration on the Yield and Properties of the Bio-
702 Oils Obtained from the Auger and Fast Pyrolysis of Douglas Fir.” Fuel 104. Elsevier Ltd:
703 536–46. doi:10.1016/j.fuel.2012.06.010.
704

17
705 Table 1: Temperature set-points and stabilized temperatures at various stages of operation
706 Figure 1: Process schematic showing major inlet/outlet streams, heating, and the mixing features of the
707 mixing paddle reactor.
708 Figure 2: Residence time in each part of the system.
709 Figure 3: (A) Measured treactor vs. calculated time by Eq. 5. (B) Measured residence time in the entire
710 reactor and pyrolysis zone vs. calculated time.
711 Figure 4: ceff/cactual vs. reactor shaft rotation frequency.
712 Figure 5: (A) Top – temperature transients of the pre-heating zone (1-8 in Figure 1) in a typical
713 run in which all heaters were controlled according to temperature set points listed in Table 1.
714 Bottom – corresponding duty-cycles of heaters 1-8. (B) Top – temperature transients of the
715 pyrolysis zone (9-12 in Figure 2) in a typical run were all heaters were controlled according to
716 temperature set points listed in Table 1. Bottom – duty-cycles of heaters 9-12.
717 Figure 6: Difference between final and initial duty-cycles for each heater (after reaching steady
718 state operation.
719 Figure 7: Plot of char and bio-oil yields vs. those of NREL for the same materials. The ratio
720 between HTM and biomass was 15:1.

18
 Pneumatic Material Pneumatic Material
agitator inlet agitator inlet

Agitator Agitator

Regular auger: Heat

transfer material (HTM) Regular auger: feedstock
dosing system dosing system

Heated transfer line

Biomass
HTM

HTM heating zone Pyrolysis Zone

HTM heating zone Pyrolysis zone

Bio-oil
Mixing paddle auger

Biochar
Mixing paddle auger

3-D model cutaway showing the mixing features

photograph of paddle reactor

details of cut flights and paddles

 tHTM tbio

νHTM νbio Heated transfer line

Biomass
HTM
HTM heating zone Pyrolysis zone

Bio-oil
νreactor Mixing paddle auger

Biochar
theat tpyr
treactor
Balance
 1 1
(A) (B)

Measured time, min

Measured time, min

0.8 Slope = 0.974 0.8 Slope = 0.973

R² = 0.98 R² = 0.99
0.6 0.6

0.4 a = 9.68 0.4

m = 0.25 m = 0.25
0.2 n = 0.5 0.2 n = 0.5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Calculated time, min Calculated time, min
 13
HTM @ 3 rpm
HTM @ 6 rpm
11 HTM @ 12 rpm
HTM @ 18 rpm
ceff/cactual

3
50 100 150 200
νreactor, rpm
 (A) HTM Biomass
Steady-state in in
(St.-S.) St.-S. St.-S.
800

600
Temperature (°C)

400 T1 T2 T3
T4 T5 T6
200 T7 T8

0
0 1 2 3 4 5
100 time (hr)

80
Duty, %

H1 H2
60

40

20

0
0 1 2 Time, hr 3 4 5

(B)
St.-S.
Steady-State HTM Biomass
(St.-S.) in St.-S. in
600
Sd.-St.
500
Temperature (°C)

400
T9 T10
300 T11 T12
200

100
0
100 0 1 2 3 4 5
time (hr)
80
H9
Duty (%)

60
H10
40

20

0
0 1 2 time (hr) 3 4 5
 100

80
Duty cycle difference, %

60

40

20

0
1 3 5 7 9 11
Heater positions
 100%
Mixing Auger (yield, d.a.f)

Silica Sand HTM

80% 15 sand:1 biomass
60%

40%
Liquid
20% y = 1.00x - 0.02 Char
R² = 0.95 Linear Regression
0%
0% 20% 40% 60% 80% 100%
Fluidized Bed (yield %, d.a.f)
 Set-point Empty System With HTM HTM and Feed
(°C) (°C) (°C) (°C)
T1 550 550.0±0.1 550.4±1.0 550.2±0.9
T2 550 558.4±0.1 557.3±0.5 556.5±0.4
T3 550 550.0±0.1 550.4±0.3 550.0±0.1
T4 550 550.0±0.2 550.3±0.4 549.9±0.2
T5 550 550.0±0.3 550.3±0.6 550.0±0.2
T6 550 558.0±0.1 555.0±0.5 558.3±0.1
T7 550 573.5±0.1 562.0±0.4 575.6±0.1
T8 550 550.0±0.1 550.3±0.4 549.9±1.4
T9 500 500.0±0.1 514.6±0.5 500.0±1.6
T10 500 500.0±0.1 500.4±0.3 500.2±0.5
T11 500 498.6±0.1 494.2±0.2 496.5±0.3
T12 500 500.0±0.1 500.2±0.1 500.1±0.1