Berty

Ethylene oxide catalyst research is expensive and time-consuming because of the need to break in and stabilize the catalyst before rehable data can be collected. Computer controlled tubular microreactors containing as Httle as 5 g of catalyst can be used for assessment of a catalyst s initial performance and for long-term life studies, but moving basket reactors of the Berty (77) or Carberry (78) type are much better suited to kinetic studies. 

Berty, Laboratoty reactors for catalytic studies , in Leach, ed.. Applied Industrial Catalysis, vol. 1, Academic, 1983, pp. 41-57. 

Gill, W.N., Garside, J. and Berty, J. M., Editors, 1989, Special Issue on Kinetic Model Development, Chem. Eng, Comm. 76. 

CH3CH=CHCH0 + 2 H2 —> CH3CH2CH2CH2OH which, for conditions selected here, is practically irreversible (Berty 1983.)... 

Figure 1.3.1a, b illustrates an idea proposed by the author (Berty 1974) the basis for geometric similarity should be what the catalyst sees on the inside, instead of the gross features of a reactor that can be observed from the outside. Identity as well as similarity can be achieved on various scales if the investigator bases the criteria on the direct surroundings of the catalyst where flow, partial pressures, and temperature all are important. 

Figure 2.2.4 (Berty 1983) shows a tubular reactor that has a thermosiphon temperature control system. The reaction is conducted in the vertical stainless steel tube that can have various diameters, 1/2 in. being the preferred size. If used for fixed bed catalytic studies, it can be charged with a single string of catalytic particles just a bit smaller than the tube, e.g., 5/16 particles in a l/2 O.D. tube. With a smaller catalyst, a tube with an inside diameter of up to three to four particle diameters can be used. With such catalyst charges and a reasonably high Reynolds number— above 500, based on particle diameter—this reactor...

The thermos phon circulation rate can be as high as 10 to 15 times the coolant evaporation rate. This, in turn, eliminates any significant temperature difference, and the jacket is maintained under isothermal conditions. In this case, the constant wall temperature assumption is satisfied. During starting of the thermosiphon, the bottom can be 20-30°C hotter, and the start of circulation can be established by observing that the difference between the top and bottom jacket temperature is diminishing. Figure 2.2.5 (Berty 1983) shows the vapor pressure-temperature relationship for three coolants water, tetralin, and Dowtherm A. 

As the name implies, these reactors are mostly used for the study of exothermic reactions, although they can be applied to endothermic reactions, too. Figure 2.2.6 shows a liquid-jacketed tubular reactor (Berty 1989). 

Figure 2.3.2 (Kraemer and deLasa 1988) shows this reactor. DeLasa suggested for Riser Simulator a Fluidized Recycle reactor that is essentially an upside down Berty reactor. Kraemer and DeLasa (1988) also described a method to simulate the riser of a fluid catalyst cracking unit in this reactor.

The older internal recycle reactors of Berty et al (1969), and Berty (1974) are shown on Figures 2.4.3 a, b. The reactor of Romer and Luft (1974) uses no mechanical moving parts. The recirculation is generated by the feed gas as it expands through a nozzle. A major disadvantage of using a jet is that feed rate and recirculation rate are not independent. Due to the low efficiency of jet pumps, recycle rates are quite low. 

These reactors all work on very similar principles and will be discussed based on the example of the Berty reactor, of which more than 500 are in operation around the world. The Berty reactor shown in Figure 2.4.3 a has much empty volume and is laborious to open and close. Another version of the Berty reactor (made by Basic Technology, Inc.) is shown in Figure 2.4.3 b. This 2-inch model was developed for quick exploratory studies on small samples of catalysts. The maximum catalyst sample volume is 15... 

Carbcny and Berty reactors were made by Autoclave Engineers, Inc., Eric, Pennsylvania. 

The operational characteristics of the older Berty reactors are described in Berty (1974), and their use in catalyst testing in Berty (1979). Typical uses for ethylene oxide catalyst testing are described in Bhasin (1980). Internal recycle reactors are easy to run with minimum control or automation. 

Various experimental methods to evaluate the kinetics of flow processes existed even in the last centuty. They developed gradually with the expansion of the petrochemical industry. In the 1940s, conversion versus residence time measurement in tubular reactors was the basic tool for rate evaluations. In the 1950s, differential reactor experiments became popular. Only in the 1960s did the use of Continuous-flow Stirred Tank Reactors (CSTRs) start to spread for kinetic studies. A large variety of CSTRs was used to study heterogeneous (contact) catalytic reactions. These included spinning basket CSTRs as well as many kinds of fixed bed reactors with external or internal recycle pumps (Jankowski 1978, Berty 1984.)... 

Another view is given in Figure 3.1.2 (Berty 1979), to understand the inner workings of recycle reactors. Here the recycle reactor is represented as an ideal, isothermal, plug-flow, tubular reactor with external recycle. This view justifies the frequently used name loop reactor. As is customary for the calculation of performance for tubular reactors, the rate equations are integrated from initial to final conditions within the inner balance limit. This calculation represents an implicit problem since the initial conditions depend on the result because of the recycle stream. Therefore, repeated trial and error calculations are needed for recycle... 

The original recycle reactor developed at Union Carbide Corporation in 1962 (Berty et al 1968) was modified or adapted by several people to different projects. Many recycle reactors were also designed by others for... 

In Chapter 1, Figure 1.4.1 (Berty et al, 1969) shows the actual measurement results of the older 5 diameter recycle reactor performance, using two different types of equipment. 

Figure 3.6.1 (Berty 1979) is a Sankey (1898) diagram, used in power engineering, where the bandwidth is proportional (here qualitatively only) to the flowing masses. This illustrates the calculation results for a rather extreme case of an NOx reduction problem. The case is extreme because the catalyst particle has a dp=0.2mm, i.e., 200 microns. Flow resistance is very high, therefore an L=1 mm deep bend is used only. Per pass concentration drop is still high, Ci-C=1.2ppm, or Dai=0.11. This was tolerated in this case, since it is between 11.2 and 10.00 ppm concentration, and nothing better could have been achieved.

Temperature gradient normal to flow. In exothermic reactions, the heat generation rate is q=(-AHr)r. This must be removed to maintain steady-state. For endothermic reactions this much heat must be added. Here the equations deal with exothermic reactions as examples. A criterion can be derived for the temperature difference needed for heat transfer from the catalyst particles to the reacting, flowing fluid. For this, inside heat balance can be measured (Berty 1974) directly, with Pt resistance thermometers. Since this is expensive and complicated, here again the heat generation rate is calculated from the rate of reaction that is derived from the outside material balance, and multiplied by the heat of reaction. 

For a first test of the reactor and all associated service installations it is recommended that experiments for methanol synthesis should be carried out even if this reaction is not especially interesting for the first real project. The reason for this recommendation is that detailed experimental results were published on methanol synthesis (Berty et al, 1982) made on a readily available catalyst. This gives a good basis of comparison for testing a new system. Other reactions that have been studied in detail and for which the performance of a catalyst is well known can also be used for test reactions. 

More details of operation in an actual study can be seen in Berty et al, (1982). In tliis work, a condenser and a liquid-gas separator were used in the product line before the pressure let-down. Keeping the products all in the vapor phase was difficult. Other improvements later included a continuous, four-component, feedgas make-up system with a compressor. 

In actual experiments (Berty et al, 1989) the method recommended by Bhasin,et al, (1980) was used, and the following conditions applied ... 

Silva (1971) used the Berty reactor to execute exploratory measurements on vapor-phase hydrogenation of organic substrates that had little vapor pressure at room temperature. The substrate was measured by weight in a small ceramic boat and put on the catalyst screen beside a few particles of catalyst, also measured by weight. Then the stirring started, and the autoclave was heated to the reaction temperature. Finally the desired hydrogen pressure was applied suddenly and the reaction started. 

Tests according to Bhasin s recommended experiments were executed at the laboratory of Berty Reaction Engineers, Ltd. on a test unit built for an export order. (The test unit was shown in Figure 4.4.1.) Results of this study were reported by Berty et al, (1989) and are reproduced here in the table on Figure 5.1.2. 

Complete or very high conversion requires the study of catalyst at very low concentrations. At such conditions, close to equilibrium (Boudart 1968), all reactions behave according to first order kinetics. Study at very low concentrations is also helped by the very small heat generation, so these studies can be executed in small tubular reactors, placed in simple muffle furnaces. Such studies were made by Kline et al (1996) at Lafayette College and were evaluated by Berty (1997). 

Reprinted with permission from Berty, 1997 American Chemical Society... 

These equations hold if an Ignition Curve test consists of measuring conversion (X) as the unique function of temperature (T). This is done by a series of short, steady-state experiments at various temperature levels. Since this is done in a tubular, isothermal reactor at very low concentration of pollutant, the first order kinetic applies. In this case, results should be listed as pairs of corresponding X and T values. (The first order approximation was not needed in the previous ethylene oxide example, because reaction rates were measured directly as the total function of temperature, whereas all other concentrations changed with the temperature.) The example is from Appendix A, in Berty (1997). In the Ignition Curve measurement a graph is made to plot the temperature needed for the conversion achieved. 

---