Article pubs.acs.org/EF
Bio-oil Stabilization and Upgrading by Hot Gas Filtration Robert M. Baldwin* and Calvin J. Feik National Renewable Energy Laboratory (NREL), Golden, Colorado 80401, United States ABSTRACT: The objective of this research project was to test the hypothesis that separation of char with its associated mineral matter from pyrolysis vapors before condensation will lead to improved bio-oil quality and stability with respect to storage and transportation. The metric prescribed by the U.S. Department of Energy (DOE) to evaluate stability in this case was a 10-fold reduction in the rate of increase of viscosity as determined by an accelerated aging test. The primary unit operation that was investigated for this purpose was hot gas filtration. A custom-built heated candle filter system was fabricated by the Pall Corporation and furnished to the National Renewable Energy Laboratory (NREL) for this test campaign. This system consisted of a candle filter element in a containment vessel surrounded by heating elements on the external surface of the vessel. The filter element and housing were interfaced to NREL’s existing 0.5 MTD pyrolysis process development unit (PDU). For these tests, the pyrolysis reactor of the PDU was operated in the entrained-flow mode. The hot gas filter (HGF) test stand was installed on a slipstream from the PDU, so that both hot gas filtered oil and bio-oil that was not hot gas filtered could be collected for purposes of comparison. Two filter elements from Pall Corporation were tested: (1) porous sintered stainless-steel (PSS) metal powder and (2) sintered ceramic powder. A sophisticated bio-oil condensation and collection system was designed and fabricated at NREL and interfaced to the slipstream filter unit. The test campaign on vapor-phase filtration of biomass-derived pyrolysis oil demonstrated that a bio-oil with substantially improved properties can be obtained by application of hot gas filtration. The ceramic filter element and test stand supplied by Pall Corporation and the vapor condensation and collection system designed and fabricated by NREL both demonstrated very good operability. Application of periodic blowback was shown to be effective in maintaining the filter element pressure drop within acceptable limits, and filter plugging was never experienced. A bio-oil with greatly reduced alkali and alkaline earth metals and very low solids content was produced. Bio-oil obtained by hot gas filtration with a PSS element had elevated iron content, suggesting that the material of construction is not suitable for this application. The PSS-filtered bio-oil also did not pass the viscosity metric of a 10-fold reduction in the rate of viscosity increase as determined by the accelerated aging test at 80 °C. Bio-oil obtained by hot gas filtration with a ceramic (Dia-Schumalith sintered ceramic powder) filter element was also low in alkali and alkaline earth metals and total solids and did not exhibit high iron content. The ceramic-filtered oil passed the viscosity metric, indicating that this oil should be much improved with respect to storage and transport stability. Total mass loss because of hot gas filtration was estimated to be in the range of 10−30% by weight.
■
INTRODUCTION Alkali and alkaline earth metals, principally potassium, sodium, and calcium, are present in biomass. The concentration of total inorganics can range from 0.5 to 15 wt % depending upon feedstock. Alkali and alkaline earth metals in biomass are typically in the range of 0.2−1% by weight. Early research on the effects of these materials was based on reducing the flammability of cotton fiber and wood by impregnation with inorganics.1,2 Over 20 years ago, removal and back-addition of salts of calcium and potassium were shown to have a significant effect on pyrolysis yields from woody biomass and cellulose.3 Removal of acid-soluble minerals increased the thermal decomposition maximum rate temperature and decreased the yield of char by approximately 15−20% for woody biomass under slow pyrolysis conditions and slightly over 33% for cellulose. Another work has reported the influence of mineral species on the rate of the pyrolysis reaction.4 When biomass is pyrolyzed, mineral matter that is present in the sample is usually found to be incorporated in the char.5−8 Alkali and alkaline earth metals have been shown to participate in or catalyze pyrolysis reactions, leading to higher yields of char.7,5 In addition, it has long been suspected that alkali and alkaline earth metals may catalyze deleterious chemical reactions during storage, leading to increases in viscosity and giving rise to the well-known lack of stability for biomass pyrolysis oil.9,10 Accordingly, a working hypothesis is that © XXXX American Chemical Society
separation of char and the associated alkali and alkaline earth metals from the primary pyrolysis vapors by filtration prior to condensation would lead to a biomass pyrolysis oil (bio-oil) with improved storage and transportation characteristics. Char from fast pyrolysis reactors consists of a mixture of course and fine material, with a small but significant fraction having a mean diameter below 5 μm. Char and mineral matter can be removed from bio-oil using conventional cyclone separators, but these devices are not effective for particle sizes below about 10 μm. Removal of char and minerals from pyrolysis oil for the production of biomass-derived boiler and turbine fuels has been demonstrated at Solar Energy Research Institute (SERI)/ National Renewable Energy Laboratory (NREL) using a ceramic cloth hot gas filter (HGF).6,11,12 Results showed that bio-oils with alkali and alkaline earth metal concentrations below 10 ppm could be produced but stability of the oil was not demonstrated or quantified. The HGF media used for these tests was a ceramic cloth fiber, which was found to “blind” very quickly, leading to excessive back-pressure on the pyrolysis reactor; regeneration by periodic blowback was not successfully demonstrated, and Received: February 7, 2013 Revised: April 20, 2013
A
dx.doi.org/10.1021/ef400177t | Energy Fuels XXXX, XXX, XXX−XXX
Energy & Fuels
Article
Figure 1. Filter element and housing from Pall Corporation.
Figure 2. NREL entrained-flow pyrolysis PDU with HGF.
accordingly, stable operation of the filter assembly was never reached during these early tests. Results of tests carried out on
hot gas filtration at the University of Aston have also been reported.13 In these tests, some improvement in storage stability B
dx.doi.org/10.1021/ef400177t | Energy Fuels XXXX, XXX, XXX−XXX
Energy & Fuels
Article
Figure 3. P&ID of the slipstream filter test stand showing HGF and supporting equipment consisting of the filter and char receiver (green), bio-oil condensation and collection (blue), and bio-oil sample points (yellow).
Figure 4. P&ID of the filter test stand showing all subsystems and controls and instrumentation. C
dx.doi.org/10.1021/ef400177t | Energy Fuels XXXX, XXX, XXX−XXX
Energy & Fuels
Article
Figure 5. OPTO-22 control system in HGF scrubber mode (filter is online).
Figure 6. OPTO-22 control system in blowback mode (filter is offline for regeneration by blowback with superheated nitrogen).
for hot gas filtered oil was found. Work on in situ filtration of biooil using a filter element installed directly in a fluid-bed reactor has also been reported;14 little improvement in the properties of the resulting bio-oil was reported. For the tests carried out under this project, filter elements from the Pall Corporation were obtained and tested at NREL. The Pall Corporation manufactures commercial gas/solid separation systems, which use a very high efficiency barrier filter to separate particles from a hot gas stream. In commercial systems, solids that collect on the filter element are periodically removed from the filter surface through reverse flow (blowback) in situ cleaning. Use and evaluation of advanced materials and operating conditions for vapor-phase filtration of pyrolysis oil were some of the primary objectives of this project.
Table 1. PDU and HGF Operating Parameters, with the PSS Element HGF conditions (set points)
PDU run conditions
filter temperature = 450 °C feed line temperature = 500 °C exit gas flow rate = 28 ALPMa blowback N2 flow rate = 4.8 kg/h filter online time = 60 s blowback time = 3 s
nitrogen carrier gas flow rate = 20 kg/h biomass feed rate = 10 kg/h superheater temperature = 400 °C pyrolysis reactor temperature = 500 °C eductor pressure (set point) = 50 kPag heat trace temperature = 500 °C
a
ALPM = actual liters per minute.
■
EXPERIMENTAL SECTION
The hypothesis that was tested in this task was that separation of char with its associated mineral matter from pyrolysis vapors before D
dx.doi.org/10.1021/ef400177t | Energy Fuels XXXX, XXX, XXX−XXX
Energy & Fuels
Article
condensation will lead to improved oil quality and stability with respect to storage and transportation. The metric used to evaluate stability in this case was a 10-fold reduction in the rate of increase of bio-oil viscosity. This metric is arbitrary but was mandated by the U.S. Department of Energy (DOE) in the original Funding Opportunity Announcement [DE-PS36-08GO98018, Biomass Fast Pyrolysis Oil (Bio-oil) Stabilization]. Viscosity for all samples was measured following ASTM D445. Accelerated aging of the bio-oil to simulate extended storage times was performed by heating the bio-oil in closed containers for 8 h and an additional 16 h at 90 °C, with viscosity measurements made after each heating interval. The primary unit operation that was investigated for this purpose was hot gas filtration. A custom-built heated candle filter system was fabricated by the Pall Corporation and furnished to NREL for this test campaign. This system consisted of a candle filter element in a containment vessel surrounded by heating elements on the external surface of the vessel; a schematic diagram of the Pall Corporation filter subsystem (without heaters) is shown in Figure 1. Basic operating parameters for the filter and housing include the following: (1) maximum working temperature of 600 °C, (2) maximum working pressure of 30 psig, (3) maximum filter ΔP of 2 psid, (4) design face velocity of 2−9 ft/min (actual operation was 4 ft/min), (5) pyrolysis vapor design flow rate of 2.2 actual cubic feet per minute (ACFM), and (6) blowback flow rate of 6.6 ACFM. The filter element and housing was interfaced to NREL’s existing pyrolysis process development unit (PDU); a schematic of the PDU and filter subsystem is shown in Figure 2. For these tests, the pyrolysis reactor of the PDU was operated in the entrained-flow mode. The HGF test stand was installed on a slipstream from the PDU, so that both hot gas filtered oil and bio-oil that was not hot gas filtered could be collected for purposes of comparison. Two filter elements supplied by Pall Corporation were tested in this project: (1) grade H 310SC porous sintered stainless-steel (PSS) and (2) DiaSchumalith 10-20 ceramic. PSS media is a relatively thick monolithic layer of sinter-bound metal particles, which forms both the cylindrical structure of the element and serves as the filtration mechanism. DiaSchumalith ceramic media is a ceramic-based material comprised of a relatively thick structure of coarse silicon carbide, which supports a thin ceramic filtration membrane. Both elements were fabricated as classic candle filter designs with single open-ended cylindrical construction. The open end of the element was affixed to a metallic tubesheet, which supported the element and also served to separate the filtered process vapor and gas from the unfiltered process stream. Both filter elements were designed to perform as surface-capture filters, meaning that solid particulates entrained in the process gas are collected along the outer surface of the filter media and gradually form a particulate cake. As time progresses, this cake serves to decrease the permeability of the filter and thereby needs to be removed by reverse flow cleaning. During reverse flow (or blowback) cleaning, a short duration gas flow is initiated against the downstream side of the filter element. This action results in a reverse flow through the filter medium, which, in turn, dislodges the particulate from the outer surface of the filter. As successive forward and reverse flow cycles occur, a permanent filter cake is established at the filtration surface, which serves to both stabilize the filter media permeability and improve the overall removal efficiency of the element. A sophisticated bio-oil condensation and collection system was designed and fabricated at NREL; a simplified piping and instrumentation diagram (P&ID) for the complete HGF system is shown in Figure 3. The main subsystems for the HGF test stand consisted of the filter including blowback for char collection (shaded in green), superheated nitrogen for blowback, and the bio-oil condensation and collection subsystem (shaded in blue). Points where filtered bio-oil were collected are shaded in yellow. When the test stand was “online”, a slipstream of raw pyrolysis vapors, including char and associated mineral matter from the PDU entrained-flow biomass pyrolysis reactor, was admitted directly to the filter subsystem. The filter element was allowed to operate for a fixed amount of time, after which the inlet valve from the PDU to the filter stand was closed and a short blowback of superheated nitrogen was used to regenerate the filter. Char dislodged from the filter
by the blowback was designed to be collected in the char receiver, which was isolated from the filter during operation but open to the filter housing (containment vessel) during the blowback. After initial operations, the valve separating the receiver from the filter housing failed and the valve was left open for the remainder of the test campaign. Gas and vapor from the filter subsystem were admitted directly to the condensation/collection subsystem. The first element of this subsystem was a spray drum, where the gas and vapor products from the filter were contacted with a recirculating stream of chilled liquid dodecane (C12H26). Gas and the condensed vapor products then passed to a chilled cyclonic gas/liquid separator and then to a gravity-settling tank, where phase separation took place. Non-condensable gaseous products were vented through a 2 μm filter for collection of any residual aerosol and then to a totalizing dry-gas meter for flow rate measurement. Gaseous products were vented to the PDU thermal oxidizer. The top liquid phase in the settling tank consisted of dodecane, which was removed from the settling tank via a centrifugal pump, chilled in a shelland-tube heat exchanger by contact with a recirculating stream of cooling media at 10 °C, and then pumped back into the spray chamber. The bottom phase consisted of bio-oil and water in a single phase; a separate water phase was never experienced under the conditions used in this testing program. The control system was designed to measure and control temperatures of the filter and all transfer lines independently and to facilitate automated switching between filtration and blowback. A complete P&ID of the instrumented test stand as built showing all control systems components is shown in Figure 4. All instrumentation components and control software were OPTO-22; other hardware components (thermocouples, pressure sensors, valves, filters, fittings, pumps, etc.) were standard off-the-shelf items that were readily available. Two views of the control system in the scrubber mode and blowback mode are shown in Figures 5 and 6, respectively. Valves shown in green are open, and black designates a valve that is closed. Typical operating parameters for the PDU and slipstream HGF are given in Table 1. For these tests, the cycle time on the filter element was set to an unrealistically low value (1 min online followed by a 3 s blowback) to obtain as many cycles on the element as possible during a given test run. This mode of operation allowed for a steady-state filter cake to be accumulated more rapidly, thus more accurately simulating the operation of the filter in an industrial HGF application.
Table 2. Analysis of White Oak Feedstock wt % C H N O S ash volatile matter fixed carbon water aluminum calcium chromium iron magnesium manganese molybdenum nickel phosphorus potassium silicon sodium zinc E
48.55 5.11 0.05 39.92 0.03 0.53 79.05 14.61 5.81 wt % in ash 0.91 42.4 ND 0.71 2.02 ND ND ND 2.62 18.9 3.82 0.43 ND
dx.doi.org/10.1021/ef400177t | Energy Fuels XXXX, XXX, XXX−XXX
Energy & Fuels
Article
Figure 7. Filter cycle screenshot showing the pressure drop as a function of time (PSS element). White oak was used as the feedstock to the pyrolysis reactor for all experiments; analytical data on this material is shown in Table 2.
Table 3. Properties of Raw and Hot Gas Filtered (PSS Element) Bio-oils for Pyrolysis at 500 °Ca
■
raw oil (500 °C)
RESULTS AND DISCUSSION PSS Filter Element. Testing of the system began with the PSS element installed in the filter housing. After some preliminary system shakedown to iron out operational problems and control system functions, a successful run of approximately 8 h in duration was made with the PSS element. Operating conditions for the PDU and HGF units were as shown previously in Table 1. Typical operation of the filter with the PSS element at 450 °C is shown in Figure 7. Data displayed in this figure includes the pressure drop (green line) and bio-oil level in the phase separator (blue line); the periodic pressure “spikes” relate to the introduction of superheated nitrogen gas for the blowback. The filter, condensation, and control systems all operated mechanically very well throughout the duration of the run. The pressure drop across the filter element increased slowly. As shown in Figure 7, the pressure drop was below 4 kPa for the 20 min period represented by this screenshot. Blowback was demonstrated to successfully maintain the filter pressure drop well below the manufacturer’s recommended cutoff value of 13.8 kPa (2.0 psid). Chemical and physical analyses of the raw and HGF oils produced from the PSS element test are shown in Table 3 and Figures 8 and 9. Several important observations can be drawn from these data. First and most importantly, it is apparent that the oil does not pass the stability test metric. As shown in Figure 8, while the viscosity of the HGF oil is lower than that of the raw oil, the HGF oil does not exhibit the desired 10-fold decrease in the rate of viscosity increase because of aging, which was set as the primary metric for this application. For this set of samples, the carbonyl and acid contents of the raw and HGF oils were also determined at 0, 8, and 24 h into the accelerated aging test. As shown in Figure 9, the acid content was essentially unchanged, while the total carbonyl content decreased. For the raw (unfiltered) oil, the carbonyl content decreased in a nearly linear fashion with time, while for the HGF oil, the carbonyl content decreased initially (from 0 to 8 h) and then remained constant. The change in the carbonyl content can be attributed to the presence of carbonyl condensation reactions, which, in turn, lead to the change in viscosity and give rise to the lack of stability shown in Figure 8. It is important to note that the “raw oil” in Table 3 and Figures 8 and 9 has been filtered after condensation to remove particulates from the
C H N O S ash volatile matter fixed carbon waterb aluminum calcium iron magnesium manganese phosphorus potassium sodium titanium zinc
HGF oil (450 °C)
wt %
wt %
44.94 7.29 0.08 47.66 0.01
