ABSTRACT
High-quality gasoline may be synthesized by catalytic reaction of methanol over zeolite ZSM-5 between 350-450 C at a liquid
methanol volumetric space time of 0.6-1 hour (Chang, 1983). This methanol-to-gasoline (MTG) process provides an alternative to crude oil
as a source of gasoline if petroleum becomes too expensive.
A bench scale vertical tubular packed bed reactor is used to investigate the MTG process. The inner diameter of the tube is 2.6 cm,
and when filled with 100 g of catalyst pellets, the bed is 33 cm tall. Experiments designed to determine the influence of reactor temperature,
methanol flow rate, and catalyst weight are performed.
In one series of experiments, the methanol flow rate is varied while the other conditions are held constant. The methanol flow rate is (in
different experiments): 60, 30, 15, and 7.5 mL/hr. Meanwhile, the other variables are held constant: the heating jacket set point are 600 C,
the nitrogen flow rate is ~70 cc/min, the condenser chiller set point is 1 C, and the total amount of methanol injected in each experiment is 20
mL.
In another series of experiments, the catalyst weight was varied while the other conditions are held constant. The catalyst weight is (in
different experiments): 100, 75, and 50 g. Meanwhile, the other variables are held constant: the heating jacket set point is 600 C, the
nitrogen flow rate is ~70 cc/min, the condenser chiller set point is 1 C, the methanol flow rate was 15 mL/hr, and the total amount of
methanol injected in each experiment is 20 mL.
A final series of experiments is aimed at varying the reactor temperature while the other conditions were held constant. However, the
temperature variation in the reactor precludes this experiment.
Undesirable temperature settings and behavior are noted and corrected. A mid-project review of the literature reveals optimal reaction
temperature and space time, and those conditions are replicated in the bench-top reactor. Liquid gasoline is produced. Issues that should be
solved in order to adequately address the MTG process are discussed.
In the appendix, scale-up calculations for a 4 bbl/day (gasoline production) fixed bed plant and a 100 bbl/day (gasoline production)
fluidized bed plant are made. Progress memos and safety reports are archived. Printouts and calculations of GC/MS analysis of experimental
aqueous and hydrocarbon products are also provided.
INTRODUCTION During the OPEC oil embargos of the mid 1970's to early 1980's, the high prices of petroleum prompted development of alternatives to fuels derived from crude oil. Up to that time, only two processes of fuel synthesis had had any commercial significance. The first was the Bergius process that used an oil/coal slurry and an iron catalyst to produce synthetic crude oil. The second was the Fisher-Tropsch process, which produced hydrocarbons from coal. Both of these processes produced hydrocarbons with poor selectivity and quality. This problem was overcome by Mobil’s methanol-to-gasoline (MTG) process of methanol conversion over a highly selective zeolite catalyst. The MTG process makes possible the synthesis of a high quality, high octane gasoline without the need for expensive post-production processing. If oil prices stay high, MTG may be a competitive method of producing gasoline. Our project studied the MTG process on a bench scale unit. Methanol is be fed to a fixed-bed flow reactor in the presence of the zeolite catalyst, ZSM-5. The product gasoline is analyzed by gas chromatography method to find the composition.
BACKGROUND The MTG process was discovered by accident by researchers at Mobil corporation. They had been trying to use zeolite ZSM-5 to convert methanol into a fuel additive. The process instead produced dimethyl ether, which with increasing space time next produced olefins (alkenes), and finally paraffins (alkanes) and aromatics. This mixture of paraffins and aromatics is commonly known as gasoline. MTG was extensively studied by Chang and other researchers at Mobil Corporation beginning in the mid 1970s. It reached it commercial zenith in the early 1980s when a 14,000 barrel/day (gasoline product) facility was scheduled to be built in New Zealand (Meyers, 1984). However, when oil prices dropped again in the mid 1980s, MTG was no longer as economic and has not been used on a significant scale since.
PROCESS CHEMISTRY The overall stoichimetric equation in the conversion of methanol to hydrocarbons is shown below: n (CH3OH) à (CH2) n + n (H20) For every 100 kg of methanol, 58 kg of water and 42 kg of hydrocarbons (gasoline) are produced (Chang, 1983). The MTG reaction is very exothermic--the heat of reaction is 1.74 MJ/kg (Meyers, 1984). It is run at 350-450 oC (Chang, 1983) at a pressure of one or several atmospheres (Chang, 1983; Meisel, 1988). The yield of by-products such as CO, CO2, H2, and coke is very small, typically less than 0.5 percent by weight. Only small quantities of oxygenates are present in the aqueous-phase product. Of the hydrocarbons, approximately 75% are in the C5+ gasoline fraction, 22% is n-butane and n-propane, and less than 3% is methane and ethane. Further alkylation of propane, butane, and isobutene increases the yield of high-octane gasoline to 99+% conversion of methanol to gasoline (Meisel, 1976). Almost no hydrocarbons are produced higher than C10 because of the shape-selective nature of the zeolite. Bulky higher-molecular-weight aromatics could be formed, but they probably undergo subsequent isomerization and other reactions to lower-molecular-weight compounds of the proper size to exit the catalyst. The methanol to gasoline conversion over ZSM-5 has been studied by the Temperature Programmed Surface Reaction (TPSR) technique. The technique is able to delineate the two steps in the process: the dehydration of methanol to dimethyl ether and the subsequent conversion of dimethyl ether to hydrocarbons (Jayamurthy et al., 1995). The general reaction pathway is represented by the sequence:
(n/2)[2CH3OH] = CH3OCH3 + H2O à CnH2n à n [CH2]
where [CH2] is the average formula of a paraffin-aromatic mixture.
The mechanism of the acid-catalyzed dimethyl ether (DME) formation from methanol and the olefin aromatization were widely investigated many years before the discovery of the MTG reaction. The intermediate in DME formation is a protonated surface methoxyl, which is subject to nucleophilic attack by methanol. Aromatization of olefins is believed to proceed along classical carbenium pathways, with concurrent hydrogen transfer. The primary mechanism for forming a C-C bond remains unsolved (Chang, 1983).
QUALITY Extensive testing of MTG gasoline has shown that its properties and performance are highly satisfactory. MTG gasoline contains typically 60 percent by volume saturated (paraffins and naphthenes), 10 percent by volume olefins, and 30 percent by volume aromatics. Most of the saturates are branched paraffins. The gasoline contains essentially no sulfur or nitrogen compounds. In a consumer-type test carried out with 34 cars (23 U.S., 6 Japanese, and 5 European) over a temperature range of –12 to 32oC, the driveability performance of MTG gasoline compared well with high-quality petroleum-derived gasoline (Meyers, 1984).
ECONOMICS The cost of methanol contributes about 90 percent of the total cost of producing MTG gasoline (Meyers, 1984). Thus, it is a simple calculation to find the price of MTG gasoline and compare it with the current price of gasoline. Of course, the volatility of natural resource prices must be taken into account. Whether or not MTG is actually economically competitive depends on market factors beyond the scope of this investigation.
EXPERIMENTAL APPROACH Our investigation involves building and testing a bench scale vertical packed bed tubular reactor in order to study the feasibility of the MTG process. The apparatus consists of a fixed bed tubular reactor with electric heating jacket and a shell and tube condenser. The inner diameter of the tube is 2.6 cm, and when filled with 100 g of catalyst pellets, the bed is 33 cm tall. A schematic is shown in Figure 2. The reactor is filled with zeolite ZSM-5 catalyst. The reactor jacket, which has a PID controlled temperature, is brought to steady state, while nitrogen gas flows through the catalyst bed. Methanol is continuously pumped into the reactor using a single syringe infusion pump, with refills as appropriate. Thermocouples are used to monitor the reactor temperature. The product vapors are sent to a condenser. Part of the organic vapor noncondensate can be absorbed by bubbling through liquid dodecane (n-C12) if necessary. The aqueous and hydrocarbon products collected are analyzed using gas chromatography and mass spectroscopy. The first experimental approach involved varying the following variables over the range available to examine their influence on the product yield and composition: Residence time. This is proportional to the amount of catalyst divided by the flowrate of reactants. Reactor temperature. This was controlled by changing the reactor heating jacket set point temperature. In one series of experiments, the methanol flow rate was varied while the other conditions were held constant. The methanol flow rate was (in different experiments): 60, 30, 15, and 7.5 mL/hr. Meanwhile, the other variables were held constant: the heating jacket set point was 600 C, the nitrogen flow rate was ~70 cc/min, the condenser chiller set point was 1 C, and the total amount of methanol injected in each experiment was 20 mL. In another series of experiments, the catalyst weight was varied while the other conditions were held constant. The catalyst weight was (in different experiments): 100, 75, and 50 g. Meanwhile, the other variables were held constant: the heating jacket set point was 600 C, the nitrogen flow rate was ~70 cc/min, the condenser chiller set point was 1 C, the methanol flow rate was 15 mL/hr, and the total amount of methanol injected in each experiment was 20 mL. A final series of experiments was aimed at varying the reactor temperature while the other conditions were held constant. However, the temperature variation in the reactor seemed very difficult to control. This is explained in the Results section below.
RESULTS It was hypothesized that the initial dip in temperature was due to vaporization of methanol, followed by the swift rise in temp as the exothermic reaction proceeded, followed by a more gradual decline as the reaction stopped. However, it was not clear how to control this behavior. If the reactor temperature could not be held steady, then all experiments where temperature is supposed to be constant would be nearly worthless. There were other problems as well. The catalyst was very black with coke deposits, there were virtually no products coming out, and what did emerge seemed to be only water. GC analysis of the water showed very small amounts (a few ppm) of benzene and toluene present. We tried adding a dodecane trap, but we only collected small amounts of aromatics as before. When this project first started, there was only one thermocouple in the reactor vessel. Unfortunately, it was not located inside the catalyst bed. Instead, it was several inches below the bottom of the catalyst bed, so that it “saw” the temperature of the product gases and not the temperature at which the MTG reactions actually took place. In an attempt to make this thermocouple see temperatures which the literature had said was ideal for the MTG process (around 400 C), we raised the set point of the reactor heating jacket to 600 C. However, this resulted in large deposits of coke on the catalyst, which indicated that the reaction was not proceeding as desired. It was not until additional thermocouples were installed that a more complete picture emerged. One thermocouple was placed in the condensate stream, just before the product flows into the collection flask. Another thermocouple was installed at the top of the reactor in such a way that it could be pushed deep into the catalyst bed. Moving this “top thermocouple” in and out, we could gauge the temperatures at different locations in the catalyst bed. Since the catalyst bed was approximately 30 cm tall (for 90 g of catalyst), this ability to measure the temperature profile proved very useful. The first experimental run for which we could measure the temperature inside the catalyst bed showed that we had been running the reactor at temperatures far higher than desired. Trial N, on 5-23-00, had the same operating conditions typically used in previous runs: a methanol flowrate of 15 mL/hr with a simultaneous flowrate of nitrogen of 70 cc/min, and a heating jacket set point of 600 C. The “top thermocouple” was placed in the middle of the catalyst bed (16 cm into a bed 32 cm deep). For start up, the reactor was initally run with just nitrogen until it reached a steady state, at which point the methanol injection was begun. Two things were evident from this experiment. First, it was clear that the reactor could indeed run at a steady state quite well. It only took about forty minutes to reach a steady state temperature inside the catalyst bed. Our earlier temperature variations had probably been due to the fact that we had first begun and then stopped injection of methanol before the steady state had been achieved. Second, the reaction was running much too hot: it was at 700 degees centigrade instead of the 400 C we wanted. This explained the large coke deposits we had seen. In order to fix the temperature problem, the next reaction run, Trial N, on 5-24-00, used a heating jacket set point of 340 C. First, the reactor was allowed to reach a steady state temperature while nitrogen gas flowed through the catalyst bed at 70 cc/min to activate the catalyst. After reaching a steady state, a methanol flow of 15 mL/hr was started. The top thermocouple’s depth was repeatedly changed to measure the temperature at different places in the catalyst bed. The reactor reaches a steady state temperatures after about an hour or two of only nitrogen flow, at which the nitrogen flow is stopped and methanol flow is begun. At that point it only takes about twenty minutes for the temperatures to reach a new steady state, and the reaction proceeds smoothly for the last hour. The temperature as a function of location is difficult to discern from Figure 5. It is easier to look at the temperature profile as a function of axial position in the catalyst bed. It is apparent that the temperature inside the reactor varies, and these variations might be expected to have an important effect in the product composition. However, the reactor temperature is still largely within the desired range for the MTG reaction. (Chang, 1983). We will discuss this point in more detail in the discussion section. At this point, we turned our attention to the other parameter which was still not in the range which the literature indicated was desired. A review of the literature had indicated that a space time of about 1 hour was optimal for complete conversion of methanol to gasoline (Chang, 1983). In order to obtain this space time for the 90-100 g catalyst bed we were typically using, calculations showed we would need a methanol feed flowrate of 130 mL/hr. These calculations are given in the discussion section. Therefore, the next two trials P and Q, on 5-25-00 and 5-30-00 respectively, were run using a 130 mL/hr methanol flow rate, after activation of the catalyst with nitrogen with the heating jacket set at 340 C. The liquid products which emerged were quite clearly gasoline; there were two phases in the product collection flask, with the organic phase on the top and the aqueous phase below. The product smelled strongly of gasoline. The first third of the catalyst bed (the first 5-10 cm) is effectively acting as a vaporizer and superheater for the methanol feed at 60-350C, while the rest of the bed is the “active catalyst” at 350-450 C. One could argue that our reactor is effectively at a lower space time because the first third of our reactor is not in the proper temperature range of 350-450 C. That would bring our space time to about 0.6 or 0.7. However, formation of the desired final products (paraffins and aromatics) are achieved in the reactor temperature range of 350-450oC at a space time of 0.6 hr (Chang, 1983). Thus, even though the temperature inside the reactor varies, these variations actually have little impact on the product composition under the conditions at which our reactor is operating. Of course, we would want to verify these assertions with chemical analysis of the product. Using gas chromatography with mass spectroscopy, it was determined that the product from Trials O, P and Q include propane, butane, pentane, methyl butene, cyclopentane, methyl pentane, hexane, methyl cyclopentane, benzene, methyl hexane, and toluene. Additionally, as in earlier trials, in the aqueous phase contained trace amounts of toluene and benzene. It was attempted to determine the exact concentration of the each component in the final product. The usual method is as follows: determine the mass of each component by using the known density and volume. After injecting approximately 1 micro liter of the standard into the gas chromatography, the integration of the peaks of the chromatogram will give the area under the curve. The ratio of the mass to the area of each component gives a calibration factor for each component. The area under the curve for each component is measured and multiplied by this factor to give the mass of each species present in the product. The standard is compared with the sample in question. However, the standard that was available did not contain all the species found in the product gasoline. Thus, it was not possible to make quantitative calculations of its composition, only qualitative visual estimates from the graphs. (Calculations are given in the appendix.)
DISCUSION Initially, it was not clear what conditions (temperature and residence time) the reactor should be run under. Since the previous group had seen little success in producing liquid products, the operating conditions were at first intended to maximize the residence TIME . The reactor tube was filled with catalyst (~100 g of catalyst) and 20 mL of methanol was injected at a rate of 60 mL/hr. The heating jacket temperature was set at 400 C. However, since the previous group had reported problems in actually producing liquid product, the nitrogen flow rate was increased to 70 cc/min in order to “push” the products out of the catalyst bed and through the condenser tubing. This approach did indeed liquid products, so the next few experiments used variations on the above conditions. Additionally, the one thermocouple inside the reactor was registering temperatures around 220 C, far below the desired temperature of 370 C. Thus, it was decided to change the heating jacket set point to 600 C. When this was done, the thermocouple inside the reactor registered temperatures around 360 C, which seemed better. These were the conditions used as the basis for the series of experiments that followed. The initial series of experiments varying reactor temperature, methanol flow rate, and catalyst weight did not directly produce useful results, since their conditions were so far from optimal: temperatures were so high (around 700 C instead of the optimal 350-450 C) and highly variable; space time was too long; the nitrogen carrier gas evaporated virtually all organic products. For the purposes of this project, the results of those experiments have essentially been thrown out. However, they did provide some indirect utility in that their problems provided clues as to what needed to be changed. The lessons learned are expanded upon below.
TEMPERATURE One of the biggest problems was that reactor temperatures had been far too high. Trial N proved that with the heating jacket set point at 600 C (as had been used for all the experimental series), the reactor temperature was around around 700 C in middle of catalyst bed, much higher than the 350-450 recommended (Chang, 1983). Thus, changing the reactor heating jacket set point to 340 C was an improvement, as seen in Trials O, P and Q. Another problem was the varying reactor temperature. This was solved by letting the reactor come to a steady state temperature using the inert flow for 1-2 hour before beginning methanol flow. Upon reflection, it was noted that if given enough time, the temperature profile resembles a well-designed PID controller that overshoots and then comes to the setpoint (see Figures 3 and 4). By happy coincidence, there is little change in temperature when the N2 flow is stopped and the methanol injection begins. Of course, the catalyst temperature does rise at this point due to the high exothermicity of the MTG reaction, but not much. One interesting problem with the measurement of temperature in the catalyst bed is that at the higher methanol flowrate (130 mL/hr), the top thermocouple would produce wildly varying temperatures at a height of around 20 cm in the bed. It seems likely that this was due to a lack of effective contact between the thermocouple probe and the catalyst at those times. Instead, the thermocouple’s tip was probably being cooled by reactant vapors and/or liquids running along its stem in the channel created by moving it up and down in the bed. This could be corrected by pushing the tip of the thermocouple an extra cm or so into the catalyst bed to ensure better contact between the thermocouple and the local catalyst. However, it would be best to design the reactor in such a way that the thermocouples do not interfere with flow inside the reactor, perhaps with dedicated thermocouples every cm or so inside the bed. An assumption made in this project was that of negligible radial temperature variation within the reactor. Verifying this assumption would require additional thermocouples at different radial locations within the catalyst bed.
SPACE TIME The space time (ST) is commonly used in the place of residence time as a parameter in reactors where the density changes with conversion. It is generally defined as ST = V/vo where V is the volume of the reactor and vo is the volumetric flow of feed. Note that the space time is equivalent to residence time in a reactant with constant density where v = vo (Schmidt 1997). At 350-450 C, a space time of 0.6-1 hour gives good conversion of methanol to gasoline and water (Chang 1983). These conditions seemed optimal, so the methanol feed rate was changed to match this space time while keeping the amount of catalyst the same. First, a simple calculation was done to determine the volume that the catalyst occupies inside the reactor. Next, we assumed that the definition of the volume of catalyst includes the void volume: density of catalyst = 0.693 g/mL. The mass of catalyst occupying the reactor is assumed constant at 90g. The mass of the catalyst (90 g) divided by its density (0.693 g/mL) is its volume (130 mL). The desired flow rate of methanol can then be determined as follows: Flow rate of methanol = volume of the catalyst/hourly space time Flow rate of methanol = (130 ml)/(1 hr)= 130 ml/hr This represents the required flow rate of methanol to be injected inside the reactor in order to achieve the desired space time and the desired product. For the purposes of this project, we assume all catalyst is active for our reaction. This requires assuming that the temperature variations with location (axial) and time (catalyst aging) are negligible. On a more practical level, it also meant that fresh catalyst was used for every reaction run, to eliminate the effects of coking and catalyst deactivation. The flow rate of methanol was assumed to remain constant, which more accurately means negligible disruption during refills. We estimated 30 sec for refillis out of a ten minute run cycle, which reduces the flow rate by about about 5%, and thus increases the space time by the same amount. Additionally, since the refilling process is done by hand, there is also a small amount of variance in flowrate introduced. However, we have neglected the effects of these disruptions here. They could be easily eliminated, by using pumping equipment designed to deliver a constant, continuous flow of methanol into the reactor.
PRODUCT COLLECTION The chiller was run as low as it would go: the set point was set at 1 C, just above the freezing point. In scaled up plants, one would just use cooling water (assumed at near room temperature) and good heat exchangers in order to effectively condense the gasoline product vapors. One large problem found early on was the fact that we had been collecting virtually no liquid organic product. This had first been thought to be due to product collection temperatures that were too high (essentially room temperature, ~25 C). However, flash calculations (done on Pro/II) had indicated that our temperatures should be adequate to condense gasoline. Upon further reflection, it was hypothesized that that the large amounts of nitrogen gas flowing through the product collection flask were evaporating the hydrocarbon liquid components. Even though they were not boiling, they do have a high vapor pressure and will rapidly evaporate if given large volumes of air. Eliminating the nitrogen flow during reaction in Trial P greatly reduced the evaporative losses of gasoline products, since it was seen that about a third of the product in the collection flask consisted of the organic phase, just what we would expect. Further reductions of evaporative loss were observed in Trial Q when the product collection flask was bathed in ice water to reduce the vapor pressure of the gasoline. Accurate mass balances could not be done to quantify the recovery, however, because of the uncertainty of the volumes of methanol injected (refills were done by hand), because of the difficulty of separation of the organic and aqueous phases (also done by hand), and uncertainty about the amount of product still remaining inside the condenser lines after the reaction run. Effective separation of the aqueous and organic phases as a separate stage in the product collection is key to the industrial production of MTG gasoline, and this equipment should also be incorporated into the benchtop reactor.
CATALYST ADSORPTION AND ACTIVATION Mass balances of catalyst before and after reaction had shown about a 3% drop in mass after reaction, even though there was obviously coke deposits on the catalyst which should have increased the catalyst mass instead. This phenomenon had puzzled us; was it due to loss of catalyst during handling (spills), or something else? Dr. Herz had suggested that the catalyst may have adsorbed moisture from the atmosphere which was subsequently desorbed in the high temperatures of reaction, reducing the apparent mass of the catalyst. To investigate this, a small amount of catalyst was carefully massed, dried, and massed again. The catalyst sample weighed 13.547 g before drying, and 13.088 g after drying—a drop of 3.4%. This seems to support Dr. Herz’s theory of water adsorption on the catalyst, which would explain the change in apparent mass of the catalyst. In the later part of the project, it became standard practice to activate the catalyst with 1-2 hours of nitrogen flow with the heating jacket at 340 C. This would certainly desorb any water on the catalyst, and would probably also remove adsorbed oxygen that would lead to undesirable products.
The project started with the idea of varying several reaction parameters in order to determine their influence on the product. However, it soon became clear that the reactor’s operation was not suitable for these experiments, and their results were thrown out. Two mid-quarter developments improved the performance of the bench-top MTG reactor. First, a review of the literature showed the optimal temperature and space time conditions. Second, installation of more thermocouples allowed much better measurement of the temperatures in the reactor, which in turn allowed much better control and therefore better selectivity. At this point, we have essentially replicated the work done by Chang in terms of operating conditions. The MTG reactor could be used effectively next year, since it’s finally at a good starting point for student research. Here are our recommendations: More thermocouples in reactor and in heating jacket: It is crucial to get a good picture of what’s really going on inside the catalyst bed. With appropriate placement of thermocouples, one could measure radial and axial distributions in the catalyst bed and in the heating jacket. It would also be useful to measure the pressure and liquid or vapor flowrates at different parts of process. The more information is available about the operation of a reactor, the better. Better temperature control in the reactor could be accomplished by using multiple stages in the heating jacket. Additionally, the use of a separate unit for the vaporization and preheating of the methanol feed would allow the entire catalyst bed to be “active” at the optimal reaction temperatures of 350-450 C, instead of the ad hoc feed vaporization and heating currently taking place inside the bed itself. Quantitative GC analysis requires the use of standards which match all the species seen in product. Make sure that standards used for calibration meet this requirement. Product collection would be facilitated by the use of equipment designed to minimize evaporative loss of organic vapors. One could put the product collection flask in an ice bath, or use a dodecane bubble trap to try to capture any vapors before lost to the vent. Better yet would be the use of something like a jerry can, designed to prevent the loss of gasoline vapors at all. Additionally, process equipment designed to continuously separate the organic and aqueous phases would aid mass balances and GC analysis.
SCALE-UP DESIGN CALCULATIONS Note: for these calculations we assume that the density of mixtures are linear combinations of the densities of the pure components (i.e. they are ideal liquids). Mass Balance and Scale-Up MTG products are (from a feed of 100% methanol) 18 wt% aromatics, 24 wt% paraffins, and 58 wt% water (Chang 1983). That means that 18 + 24 = 42 wt% of the methanol fed in becomes gasoline. Or, for 1 g of methanol feed, we get 0.42 g of gasoline. The density r of methanol is 0.80 g/mL. The volume of 1 g of methanol is ___(1g)___ = 1.25 mL. (0.80 g/mL) For the gasoline fractions in question, rparaffins is 0.65 g/mL, and raromatics is 0.87 g/mL (Hancock 1985). Thus, rMTG gasoline = (18/42)(0.87) + (24/42)(0.65) = 0.744 g/mL. Therefore, for 1 g methanol fed in, we get 0.42 g of gasoline, whose volume is __(0.42 g) _ = 0.565 mL. (0.744 g/mL) Or, in order to obtain one barrel of gasoline out, we feed _(1.25 mL MeOH)_ = 2.21 barrels pure methanol feed. (0.565 mL gasoline) What if we are using methanol which is not 100% pure? Crude methanol produced from natural gas is typically 83 wt% methanol and 17 wt% water (Meyers, 1984). For one barrel of gasoline out, we feed (2.21 barrels pure MeOH)*(0.80 kg/L)/(0.008386 barrel/L) = 210.8 kg pure methanol which comes with (17/83)(210.8 kg MeOH) = (43.2 kg H2O)*(0.008386 barrel/L)/(1 kg/L) = 0.36 barrels of water. So, for one barrel of gasoline out, we feed (2.21 barrels MeOH) + (0.36 barrels H2O) = 2.57 barrels of crude methanol from natural gas. Remember that space time (ST) is defined for our system as ST = V vo where V is the volume of the reactor catalyst bed and vo is the reactant liquid volumetric feed rate. Rearranging, V = (ST)( vo) and the weight of the catalyst (Wcat) that we need is Wcat = (rcatalyst)(ST)(vo). Assuming the use of the same catalyst pellets we used in our experiments, rcatalyst = 0.693 g/mL, where the void fraction of the catalyst bed is included in its volume. For a space time of 1 hour, Wcat = (0.693 kg/L)(1 hr)(1 barrel gasoline/day) = 3.44 kg catalyst (24 hr/day)(0.008386 barrel/L) in order to obtain 1 barrel/day of gasoline. To obtain more, we simply scale up proportionally. This calculation assumes all catalyst is being used all the time (i.e. a fixed bed reactor with 100% capacity factor and negligible deactivation with time). Since this is not very realistic, let’s look at two more plausible scenarios. For 4 barrel/day of gasoline production in a fixed bed reactor, let’s assume a plant capacity factor of 95%. That means that 5% of the time the plant is shut down for catalyst regeneration and so on. Thus, Wcat = (4 barrels/day gasoline)(3.44 kg catalyst/barrel/day gasoline) = 14.5 kg catalyst. (0.95) For 100 barrel/day gasoline production in a fluidized bed reactor, let’s assume a plant capacity factor of 95%, and assume that an additional 5% of the weight of the active catalyst is being regenerated in an activation loop on a continuous basis. Then, Wcat = (100 barrels/day gasoline)(3.44 kg catalyst/barrel/day gasoline)(1.05) = 380.2 kg catalyst. (0.95)
PRACTICAL ISSUES There are several sources and sinks of thermal energy in the MTG process. With appropriate heat integration, the sources and sinks could be coupled to minimize the use of external heating and cooling loads. By following the recommendations below, making the appropriate assumptions, and modeling the system in Pro/II or the equivalent, a reasonable heat balance can be obtained. In the envisioned system, methanol liquid feed at ~30 C is vaporized and superheated by countercurrent heat exchange with product vapors at ~400 C, which also condenses the product vapors and brings them to room temperature (~30 C) for storage. Additional heat for superheating the methanol feed vapor to optimal reaction temperature (~350 C) can be obtained by cocurrent shell-and-tube heat exchange with the reactor tube (350-450 C) to recover the heat of reaction (1.74 MJ/kg methanol). Working fluids (e.g. steam) may be used if necessary. If additional external heating or cooling sources are necessary (especially at startup and shutdown) these should be designed and sized appropriately. Electric heating rods and evaporative cooling towers are probably best for these functions. For this type of catalytic reaction, it would be best to use a tubular reactor with a small diameter and good external heat transfer, to minimize radial temperature gradients and keep the catalyst in the desired temperature range of 300-450 C, to minimize hot spots and optimize selectivity. Heat transfer experiments and design calculations would yield the numbers necessary to satisfy these requirements. Additionally, to facilitate heat transfer internal to the catalyst bed, a fluidized bed is appropriate. A fluidized bed also makes it easier to run the catalyst through a regeneration loop, by continuously diverting a small amount from one end of the reactor, regenerating it separately, and then adding it back to the other end. Catalyst aging and deactivation is an important factor in catalytic reactors, especially one like the MTG process which does create coke deposits on the catalyst. The separation of aqueous and organic liquid products should be nearly 100% efficient, with appropriate treatment for the trace amounts of hydrocarbons remaining in the water byproduct. Product collection equipment should be designed to prevent vaporization losses of hydrocarbons, much like a “jerry can” prevents losses of gasoline vapors. Of course, the presence of large quantities of highly volatile and very combustible hydrocarbons mandates high levels of safety inherent to the design. Additionally, environmental considerations dictate that emissions be strictly controlled.
