Ind. Eng. Chem. Res. 2010, 49, 10883–10888
10883
Scale-Up of Microchannel Reactors For Fischer-Tropsch Synthesis Soumitra R. Deshmukh,* Anna Lee Y. Tonkovich, Kai T. Jarosch, Luke Schrader, Sean P. Fitzgerald, David R. Kilanowski, Jan J. Lerou, and Terry J. Mazanec Velocys, Inc., 7950 Corporate BouleVard, Plain City, Ohio 43064
The scale-up of a microchannel reactor for Fischer-Tropsch (FT) synthesis has been demonstrated at multiple scales using four reactors with different lengths and number of channels. An FT catalyst provided by Oxford Catalysts, Ltd. was tested in single channel microreactors with catalyst bed length ranging from ∼4 to ∼62 cm. The same catalyst was also tested in a pilot reactor with 276 parallel process channels (∼17 cm in length). Equivalent process performance was observed across each scale as determined by the metrics of CO conversion, selectivity to byproduct, and the chain growth probability (R). The overall C5+ productivity for these reactors of disparate scales spanned several orders of magnitude ranging from ∼0.004 gallons per day (GPD) to ∼1.5 GPD. The operational flexibility of these microreactors was illustrated by varying syngas feed composition and flow sheet conditions (pressure, temperature, dilution, etc.). The elusive premise of numbering up microchannels has been demonstrated, enabling the scale-up of reactor capacity. Introduction Fischer-Tropsch (FT) chemistry is based on the pioneering work of Franz Fischer and Hanz Tropsch in the 1920s and 1930s that aims at creating long chain paraffinic hydrocarbons by polymerizing a mixture of carbon monoxide (CO) and hydrogen (H2), commonly referred to as “synthesis gas” or “syngas”, at an elevated pressure and temperature in the presence of a catalyst.1 The reaction can be represented by the following chemical equation: nCO + (2n + 1)H2 f CnH2n+2 + nH2O. The reaction is highly exothermic and selectivity to the primary byproduct, methane, and the rate of catalyst deactivation are temperature sensitive; thus control of catalyst bed temperature is important for high-yield operation. The primary products of FT synthesis are paraffins with a carbon number distribution,2-4 usually following the Anderson-Schulz-Flory distribution with the exception of carbon numbers less than C4. Typical byproducts are liquefied petroleum gas (LPG) and naphtha. After the FT process, heavier hydrocarbons can be hydrocracked to produce distillate products, notably diesel and jet fuel.5,6 FTderived transportation fuels are typically referred to as synthetic fuels. Recently, there has been an increased interest in synthetic fuels for various reasons including increasing world demand for fuels, the clean burning nature of synthetic fuels due to the absence of impurities, the increased desire to monetize stranded gas as energy prices rise, and the growing interest in producing renewable, environmentally sustainable fuel from the gasification of nonedible biomass. Depending on the source for syngas, today’s FT synthetic fuels facilities rely on utilization of coal (CTL) and stranded natural gas (GTL) resources for larger scale plants. On the other hand, applications such as biomass-toliquids (BTL) and waste-to-liquids call for smaller-scale plants that leverage the benefits of the FT technology in conjunction with localized and smaller feed sources (e.g., municipal waste). Owing to their compact and modular nature resulting from intensified heat and mass transfer, microchannel reactors are ideally suited to address the challenges of small-scale BTL plants and also the large-scale offshore GTL platforms. The demonstration of nearly isothermal performance of FT in a single microchannel has been described early in the literature by Wang * To whom correspondence should be addressed. E-mail:
[email protected]. Tel.: +1-614-733-3376. Fax: +1-614-7333301.
and co-workers.7-9 Several groups continue to demonstrate microchannel reactors for FT on a laboratory scale.10,11 A multichannel FT reactor was described by Jarosch and coworkers,12 and the performance was compared to predicted performance. This paper compares the performance of multiple microchannel reactors evaluated with an identical cobalt-based FT catalyst to clearly demonstrate the scale-up of microchannel reactors. It is shown that neither channel length nor number of channels substantially changes the catalyst performance, when operated in a nearly isothermal microchannel reactor. Experimental Section To demonstrate the scalability of microchannel reactors, experiments are carried out in four microchannel reactors using an identical catalyst. Each reactor is operated with a fresh catalyst charge. The reactors and the experimental setup used in this study are described next. The first microchannel reactor (Short) has a characteristic dimension (gap) of 1 mm and a width of 0.8 cm. A 3.8 cm long catalyst bed is situated in the center of the 7 cm long process channel. Inert beds of SiC placed upstream and downstream of the catalyst bed make up the remainder of the process channel. The heat generated in the reaction is removed by a hot oil (Marlotherm SH) flowed cocurrently in two coolant channels, on either side of the process channel, using a Julabo pump to maintain isothermality. The second reactor (Long-A) contains one process microchannel with a dual gap of 1 mm for two-thirds of the channel width and 0.5 mm for one-third of the channel width. The channel has a width of 0.6 cm and a catalyst bed length of 61.6 cm. A 1.9 cm long SiC bed is placed upstream of the catalyst bed to promote preheating of the reactants to the desired reaction temperature. Similar to the short microchannel reactor, two coolant channels flowing hot oil play an important role in maintaining the reaction isothermality. The third reactor (Long-B) is similar to the second reactor (Long-A) with one process channel and two cooling channels. However, the process channel in this case has a uniform gap of 1 mm. The other process channel dimensions are identical to the second reactor (Long-A). A 1.9 cm SiC bed and a 61.6 cm catalyst bed make up the entire process channel length in this reactor.
10.1021/ie100518u 2010 American Chemical Society Published on Web 08/05/2010
10884
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
Figure 1. Pilot-scale FT microchannel reactor with 276 parallel process channels and 132 coolant channels as seen inside a partially open pressure containment shell. The reactor is operated in a cross-flow mode with down flowing process feed and coolant flow coming out of the plane of the paper.
The fourth reactor (Pilot) has 276 process channels with each process channel having a width of 0.3 cm and gap of 1 mm. Each of the process channels has a 1.9 cm SiC bed at the inlet followed by a 17.1 cm long catalyst bed. At this larger scale, a more commercially applicable cooling approach is tested. Instead of a hot oil coolant system (as in the three single channel reactors), water is employed as a cooling medium in the pilot reactor under partial boiling conditions, that is, the heat of reaction from the Fischer-Tropsch reaction is utilized to generate steam from the cooling water. In the pilot reactor, layers with catalyst containing process microchannels are flanked on either side by layers containing coolant microchannels. Process and coolant microchannels are oriented with their major axes orthogonal to produce a cross-flow architecture. Headers and footers are attached for the coolant and process channels for external connections to macroscale, that is, larger, piping. The final reactor assembly is enclosed within a pressure containment shell (PCS), as shown in Figure 1, which also serves as a phase separation unit for the water and steam. The testing setup for the reactors Long-A and Long-B is shown in Figure 2. The reactors are well insulated to minimize heat losses to the ambient. A similar setup, adjusted to their scale of operational flows, is used for the Short and Pilot reactors. The flow and composition of synthesis gas (syngas) fed to the FT synthesis microchannel fixed-bed reactor is controlled using mass flow controllers to precisely set the flows of the individual component gases. The system pressure is controlled by a backpressure regulator. System pressure and temperature data is recorded using pressure transducers and thermocouples, respectively. A LabView system is used to monitor the real time process performance. The product stream is quenched and collected in two vessels, the first maintained at ∼100 °C collects the waxes, and the second maintained at ∼0 °C collects the lighter liquids. The reactor tail gas is analyzed for noncondensable hydrocarbons using an Agilent micro gas chromatograph (Agilent 3000A RGA) using molsieve 5A, PlotQ, alumina, and OV-1 columns. The collected hydrocarbon product (wax and liquid) is analyzed to determine the chain growth probability, R, for each reaction condition. The nitrogen diluent in the reaction mixture is used as an internal standard for quick mass balance estimation. A detailed material balance is also performed by collecting the product (hydrocarbon and
Figure 2. Microchannel FT reactor testing experimental setup for the Long microchannel reactors.
aqueous) streams for a long period of time (>50 h). The overall material balance for all reactors was found to be within 98-102%. Each reactor was loaded with a catalyst obtained from Oxford Catalysts, Ltd. (OCL). The high cobalt loading catalyst is based on the patented organic matrix combustion (OMX) method for producing highly active and selective FT catalysts discussed in several patents13-16 and is used for carrying out the FischerTropsch synthesis in microchannel reactors.17,18 The mean particle diameter of the catalyst particles loaded in these reactors ranges from 200 to 300 µm. The amount of catalyst loaded in these reactors ranges from 4000 h on stream operation and several regeneration cycles. Several different syngas feed compositions were run successfully with
>60% CO conversion,
