MASS TRANSF TRANSFER ER AND CA CAT TAL ALYSTS YSTS
General reactor model improves improv es HPI applications Applying new tools, refiners can optimize revamp, and modernize existing process units S. Dutta and R. Gualy Gua ly,, GTC Technology Corp., Houston, Texas
M
any commercially available predictive models focus on improving the entire processing unit. Very few applications concentrate on the reactor within the processing unit. In fact, the economical, profitable management of petrochemical and refining facilities is largely based on the design and operation of processing reactors. Traditionally, reactor conditions are set, and it runs. Optimization opportunities are missed, because the HPI facility does not focus on refining reactor performance in conjunction with other upgrades and modernization changes made within the processing unit. Several examples describe the key features of a general model and how to apply it when re-evaluating re-evaluating,, optimizing, revamping and modernizing HPI reactor. Fig. 1. Schematic of reactor improvement steps.
Unseen opportunities. Very fe w ref r ef ine inerie riess have h ave examined and developed in-depth modeling tools for process reactors. Conversely, several petrochemical producers have developed very reliable reactor models, while others still use simple calculation methods when optimizing processing operations. Developing models for refinery reactors with classical reaction engineering is cumbersome. It is a complex problem due to the many non-stoichiometric reactions involved and very large number of components and pseudo-components used in traditional simulations. In-house reactor models are typically based on accumulated experience and operating databases. Such models are totally empirical in nature, applicable to very specific feedstocks and designs and predict performance within very limited operating ranges. These models fail when there are variations in the feedstock quality or composition, increased demand for productivity, or changes are necessary to produce new products. Thus, the refiner, relying on past experience and old databases, is forced to use intuitive and nonquantifiable tools. Current predictive models for cost-effective operation of a ref inery or a petrochemical plant do not con-
sider optimizing the reactor reactor.. Instead, pseudo-optimized plant conditions are evaluated based on the changes from other units, products, prices and feedstocks. If the reactor operation could be evaluated and optimized in addition to these variables, facility profits could be maximized. The situation is even more critical when revamp or modernization is required for an existing production unit. Using existing and often old designs severely limits the scope of revamp and modernization projects and does not offer the best chance for success. In today’s increasingly competitive operations, this situation is no longer feasible. With better characterization of feedstocks and products, and advent of high-speed computation, a general reactor modeling package, applying the classical approach, can now be built. It is applicable to both refineries and petrochemicals. Such a program can be readily applied or customized to fit varying demands of revamp and modernization projects. More than ever, refiners must adopt a petrochemical mindset and move rapidly in that direction.
HYDROCARBON HYDROCAR BON PROCESSIN PROCESSING G / JULY 1999
General model—key features. For any given reactor, the general model should predict the performance for any known reaction system and catalyst activity. Conversely, for a given reaction system and catalyst activity, the model should predict the performance for a variety of known reactor configurations. Thus, the model is applicable to any combination between reaction and reactor systems. The user can use the model to optimize the operating conditions that best meet a given demand from an existing reactor. Also, the user can determine the best possible reactor configuration by examining various possible new design options or design modifications of the existing unit. For example, assuming the same catalyst activity for a catalytic gas-solid reaction system, the user can immediately determine the possible advantages of switching from an existing fixed-bed to a fluidized-bed reactor for the next revamp or modernization plan. The model should be based on fundamental principles of reaction engineering and reactor hydrodynamics. Therefore, its application should not be restricted to narrow operating ranges or conditions of empirical models. It should predict the reactor performances, particularly the product composition, for a wide variety of designs and operating conditions, including those used or are expected. It should also cover design options such as product recycle, multiple feed injections, multiple reactor types in series or parallel, and various internal flow circulation or mixing modes within the reactor. The model should cover conditions beyond normal operations and can predict upset, offspec and unsafe situations. The model should have a minimum number of adjustable/experimental parameters and can be solved by standard mathematical routines requiring minimum execution time. It should be easy to integrate into other commercial or in-house simulation packages, and into both off- and online control algorithms. Finally, it should evolve through extensive validation by experiments in appropriate test methods. Is there such a model? Models are available; they apply to gas-solid and homogeneous reaction and reactor systems of virtually any kind and
Ten frequently asked questions concerning reactor modeling þ
Q.1 Does the reactor currently have a deficiency to address? It is a difficult question. However, many old reactors are obvious candidates for close scrutiny of their performance. A performance analysis with a modeling tool can determine the improvement potentials for a given reactor. This is same as the performance diagnosis of most other unit operations or segments of the process plants that are subject to revamps and modernization. This diagnosis can be done with the help of commercial simulation packages and other modeling tools developed in-house or available in the market.
þ
Q.2 How much information is needed? Any model is as good as the quality and quantity of data available. The model applicability is also dependent on the ranges of key variables covered by the data on which the model is built. The larger the ranges covered by the data, the better the applicability of the model. A working model for an acceptable range of operations and design options can be quickly built with the data and information that is available from the plant archive and open literature. This model obviously gets refined as more data are gathered from operating the reactor and fed back to the model.
þ
Q.3 How should nonconventional designs, such as FCC, be handled? An FCC can be viewed as a series combination of a completely mixed reactor (CSTR) near the hydrocarbon feed point, followed by a circulating fluidized-bed reactor in the rest of the riser section and a bubbling fluidized-bed reactor at the disengagement zone at the top. A schematic representation of this configuration is shown in Fig. 4. Such a system can be easily modeled by a general model package that includes all three reactor types and can be executed as a combination in series. Most other reactors of a nonconventional design can similarly be modeled by using the existing reactor modules within the package and by making minor modifications or adjustments to fit the exac t needs.
þ
Q.4 What is the most critical step? The most critical and often the most time-consuming step is determining the reaction mech anism and kinetics, especially the latter. A large number of data must be carefully screened and analyzed for this purpose. Gaps in the data must be filled with new tests. Also, tests must be done to replace bad data.
þ
Q.5 What about data confidentiality? Confidentiality of all information and data can be maintained by standard secrecy agreements between the client, any external laboratory and third party involved in the process.
þ
Q.6 How long/expensive are experimental test campaigns? This depends on three factors: 1) quality and quantity of background information available on the reaction system, 2) quality and quantity of lab or pilot plant data available, and 3) complexity of the reaction system. Reliable data covering adequate ranges of the four major variables—temperature, pressure, feed composition and flowrate—are needed for application of the general reactor model. These test campaigns may run from a few weeks to a few months depending on the situation. The cost can eas ily be justified from both shortand long-term benefits derived from this effort.
þ
Q.7 How do you assess the benefit? The reactor is the heart of most process plants and refineries. Modeling and optimizing the performance of almost all unit operations or segments of plants has become a routine matter. However, the reactors remained relatively untouched in most cases. We have lived with these old reactors or old designs, particularly in refineries, for too long. Significant improvement potentials of these reactors exist, both on a short- and a long-term basis. Using a good modeling package could only assess the benefits from these improvements.
þ
Q.8 What is the risk? There is no risk in implementation modeling, since most of the effort is done on the sideline. Total confidence is gained from a thorough model validation in the test facility or controlled step tests in the operating plant. Only then can the modifications and improvements be carried out in stages in the operating plant, and in revamp and modernization.
þ
Q.9 Will the effort succeed? A well-tested, practical and comprehensive general reactor model has a higher probability of success and a return on investment than most trial-and-error procedures normally practiced today. Furthermore, similar to many simulation tools used today, the reactor modeling will also become refined and more powerful with time and its value and usefulness will increase.
þ
Q.10 What are the time and costs involved? Time and cost involved in building such a tool are negligible compared to both short- and long-term benefits expected from its applications. The expected improvements in yield and productivity alone, ignoring many other benefits cited earlier, could justify these costs. For revamp or modernization the justification is even stronger, since this tool could minimize, if not eliminate, much more expensive pilot plant campaigns normally considered necessary for such a job.
HYDROCARBON PROCESSING / JULY 1999
Fig. 2. General reactor model applications that respond to market demands.
complexity. They handle any G× R×S reaction system, where G is the number of gaseous reactants and products, S is the number of solid reactants and products and R is the number of reactions for any complexity and kinetics. They apply to both catalytic and noncatalytic gas-solid reactions, and also to nonstoichiometric reactions of refineries and coal conversion processes. Conventional gas-solid reactor type—fixed-bed (various types), f luidized-bed (bubbling, turbulent, circulating or entrained) or moving-bed, or any combination—can be simulated. They also can evaluate many design modifications such as, product recycle, multiple feed injections, multi-staging and temperature programming. The model should include all necessary correlation and methodologies for heat and mass transfer, and hydrodynamic calculations necessary for each reactor configuration. The model must be customized for a unique design feature(s) not covered by conventional reactor configurations. Modeling—a powerful tool. Incorporating reactor modeling in process operations has several benefits: Boost your yield and productivity. Using a model, the operator can predict the optimum processing conditions for operating unit reactors. Switching to those conditions can increase yield and productivity. Different operating conditions can be evaluated via the model without interfering with normal operation. Conditions outside the normal operating envelope can be studied for stability, safety and productivity. In some cases such as in acetylene converters in ethylene plants, the converters are operated at extreme conditions to avoid breakthrough of unwanted contaminants. At breakthrough, valuable process products are lost. In such cases, a good reactor model integrated with advanced control system could allow plants to back up conditions
to avoid production losses and boost productivity. Custom-tailor products. With a general model, users can change operating conditions or modify an existing reactor. For example, adding a new recycle or feed stream, or programming of reactor temperature profile can be simulated when developing new or varying product composition or quality. This is a burning need in today’s refineries. A general model can be used to develop modified design and/or operating conditions promptly and reliably. Modernize the reactor. Superior reactor designs can be simulated for future plant expansions for a new facility. For example, converting from a conventional fixed-bed to a fluidized-bed reactor can substantially decrease capital costs and gain dramatic improvement in both yield and productivity. Currently, this opportunity exists in many production schemes that involve partial oxidation reactions. For a very fast reaction system, for example cracking, switching from a fluidized bed to an ultra-rapid reactor system, can boost the yield of desired products. A general model can critically examine the pros and cons of each conversion. Equally important, it can be used to develop the detailed design for the new system. Revamp for a new product line. A new catalyst or a different reaction system can be a good candidate for an operating reactor that is no longer profitable. The general model allows the user to promptly evaluate this situation. The model can then be used to revamp the reactor for a new product line based on the new catalyst and reaction system. Improve product quality. Processing facilities can add a secondary booster or polishing reactor to improve conversion or selectivity, and minimize impurities or harmful traces. Often, environmental regulations are the drivers for polishing and waste-reduc-
HYDROCARBON PROCESSING / JULY 1999
tion efforts. Different designs for the booster reactor can be evaluated quickly and with no risk or interruption to current production. A schematic representation of reactor revamp and modernization is shown in Fig. 1. It represents an improvement for the current reactor to a fully modernized one after progressing through successive steps of optimization and revamps. A general reactor model is a powerful tool; it can guide the user through these improvements at every step. Prerequisites. There are three prerequisites to using this tool type: Reactor performance data. Data covering a reasonable range of the four major operating variables— temperature, pressure, feed composition and feedrate— are key prerequisites for a general model. The plant data archive often is a valuable source. However, a plant normally runs at almost fixed conditions; hence, a large variation in the operating conditions may not be found in the archive data. Original lab data or pilot-scale information on the reaction system that usually cover a wide variety of operating conditions are more useful for this purpose. Good data sometimes can also be obtained from operating the plant by controlled step changes, and then monitoring the effects from such changes. Background information on reaction mechanism and kinetics. For reactions of commercial interest, adequate information is usually available in open literature. Background literature in the plant archive can also provide valuable data on this issue. After a careful analysis of these literatures, the user can usually formulate a satisfactory reaction mechanism for the system. For complex nonstoichiometric refinery reactions, the reaction mechanism is represented through lumping of compounds. The compounds in turn are represented either by pseudo- (CxHyOz for example) or model compounds. In contrast, information on the kinetics of refinery reactions is fragmentary. However, a careful analysis of this fragmentary information should identify many important rate parameters, particularly reaction activation energies. A general model can be used to determine the pre-exponential factors of reactions by using the available data and known activation energies, and improve these factors with additional data that are available from the test facility. Appropriate parameter estimation procedure is used to determine the rate parameters. In the absence of satisfactory information, the rate equation for each reaction can be expressed as first order with respect to each reacting component for most reaction systems, at least for a preliminary model. Small test facility for model validation. To achieve total confidence in the model projections, a small test facility to carry out a limited number of experiments to validate the model is needed. Data from this facility would also provide missing information on reaction mechanism and kinetics. This facility can be located in an in-house laboratory or at an external laboratory contracted to do the job under confidential agreement. Such data sometimes can also be obtained from the operating plant by controlled step changes in the operation and monitoring the effects of such changes. This arrangement can be a substitute for the test facility.
Table 1. Preliminary reaction mechanism and rate expressions for acrylonitrile production Assumed reaction mechanism for ammoxidation of propylene to acrylonitrile 1 2 O2 à C3H6 + NH3 + 1 ⁄ C3H3N + 3 H2O Rxn1 1 1 1 2 O2 2 CH3CN à C3H6 + 1 ⁄ 2 NH3 + 1 ⁄ 1 ⁄ + 3 H2O Rxn2 à C3H6 + 3 NH3 + 3 O2 3 HCN + 6 H2O Rxn3 à C3H6 + 3 O2 3 CO + 3 H2O Rxn4 1 2 O2 à C3H6 + 4 ⁄ 3 CO2 + 3 H2O Rxn5 1 2 O2 à C3H6 + 1 ⁄ C3H4O2 + H2O Rxn6 C3H6 = Propylene C3H3N = Acrylonitrile CH3CN = Acetonitri le C3H4O2 = Acrylic acid
The rate expressions used r1* = k 1* × CPR0.48 CO
CNH
0.19
r2* = k 2* ×
CNH
0.47
CNH
0.47
r3* = k 3* × r4* = k 4* × r5* = k 5* × r6* = k 6* ×
0.48 2 CPR0.43 CO 0.52 2 0.43 CPR CO 0.52 2 CPR CO 0.50 2 CPR CO 0.50 2 CPR CO 0.50 2
3 3 3
where: r1* – r6*
Rates of reaction 1 to 6
k 1* – k 6*
Rate constant of reaction 1 to 6 ( k 1* = k 01* –E2*/RT e –E1*/RT , k and so on) 2* = k 02* e
k 01* – k 06*
Pre-exponential factor of reaction 1 to 6
E1* – E6*
Activation energy of reaction 1 to 6
R
Gas constant
T
Temperature
CPR, C0 , CNH 2
3
Molar concentration of propylene, oxygen and ammonia
Five steps to use this tool. A typical general model has five components, which vary between reaction and reactor systems. They are: • Reaction system—stoichiometry and rate parameters, etc. • Catalyst and reactor specifications—catalyst size, shape, porosity, bulk and particle densities, and reactor diameter, height and internals, etc. • Feed conditions—pressure, temperature, flowrate and composition, etc. • Thermophysical properties of reactants and products, etc. • Heat transfer system specifications—heat transfer coil/jacket design, and coolant flow, properties and inlet temperature, etc. The five steps to run a typical general reactor model for a single reactor type are: ♦ Choose the reactor type that best represents the existing reactor or the alternate design(s) of interest from the menu of the model package ♦ Input all five components of data representing the reaction and reactor system ♦ Input the rate expression for each reaction ♦ Execute the model ♦ Change one or more design and/or operating variables of the second step and execute the model. Repeat the process until the best design and operating condition(s) are found. Alternatively, interface this package with an optimization routine to find the optimum automatically by repeated execution of the model.
HYDROCARBON PROCESSING / JULY 1999
Table 2. Performance projections of three reactor configurations for acrylonitrile production, capacity: 100 tpd** Reactor type
Inside dia., cm
Height, cm
Temp., °C
Propylene conversion, %
Acrylonitrile selectivity, %
Acrylonitrile yield, %
MTY 1, (Mass time yield)
Steamcooled multitubular fixed bed
3.812
800
Feed 307 Peak 360 Exit 327
87.1
89.3
77.8
0.608
# of Tubes–18,000 Catalyst particle 1 8 in. size– ⁄ Inlet gas velocity– 45 cm/s4
Bubbling fluidized bed
500
5003
442
98.9
81.3
80.4
1.945
Catalyst particle size–100 micron Inlet gas velocity– 56 cm/s4
Circulating fluidized bed
280
2,000
4425
80.7
75.5
60.9
5.946
Catalyst particle size–75 micron Inlet gas velocity– 865 cm/s4 Catalyst loading– 51.54 kg/m3 gas
Other key design/ operating features
** Assuming same feed composition as follows for all reactor types: C 3H6 8.3%, O2 16.8%, NH3 9.6%, H2O 1.5% 1 2
MTY: tons of acrylonitrile produced per ton of catalyst per day Each tube Dense bed height Superficial velocity at inlet conditions Feed temperature 6 Based on bed inventory only and does not include catalyst in the rest of circulation loop 3 4 5
EXAMPLE 1—ACRYLONITRILE PRODUCTION In this case study, a general model was used to evaluate the commercial performance projections of three different reactor configurations in a typical petrochemical process. The reaction involved producing acrylonitrile from propylene by vapor-phase catalytic ammoxidation. Preliminary reaction mechanism and rate expressions as derived from a brief study of open literature and are listed in Table 1. These values were used for this example. Approximate activation energies for the reactions are also derived from the literature search. A general reactor model is then applied to estimate the pre-exponential factors of this 6-reaction × 10-component system. This is done by simulating the performance of a typical commercial bubbling fluid-bed reactor. The preexponential factors ( k01*– k06*, particle volume basis) and the activation energies ( E1*– E6*) used are summarized below:
Fig. 3. Heavy hydrocarbons conversion reaction network.
For an alternate reactor type, choose it in the first step and repeat the steps. Repeat the whole process for all desired reactor types and operating modes (say with recycle, multi-staging, reactor network and temperature programming, etc.) until the optimum of all possible designs is reached. Using a general model, reactor performance is obtained in the form of detailed axial prof iles for all major variables including temperature, pressure, composition, production rates, conversions, selectivities, heat-generation rate, and other physical and hydrodynamic conditions. The model outputs terminate with all conditions at the reactor outlet. Fig. 2 shows a schematic representation when applying a general reactor model in revamp and modernization to meet market demands.
k01*(gmol/cm3 /s) = 0.1710E+00 k02*(gmol/cm3 /s) = 0.1120E+00 k03*(gmol/cm3 /s) = 0.1630E+00 k04*(gmol/cm3 /s) = 0.5290E–01 k05*(gmol/cm3 /s) = 0.1100E+00 k06*(gmol/cm3 /s) = 0.2150E+00
E1*(cal/gmol) = E2*(cal/gmol) = E3*(cal/gmol) = E4*(cal/gmol) = E5*(cal/gmol) = E6*(cal/gmol) =
20,000 20,000 25,000 35,000 35,000 25,000
Using the general model, commercial projections for a steam-cooled tubular fixed-bed reactor and a circulating f luidized-bed reactor are made. Activity and selectivity of the catalyst are assumed to be same for all projections. Performance projections of the three reactor configurations are summarized in Table 2. This table immediately indicates that a bubbling fluidbed (BFB) reactor is perhaps the most attractive configuration between the three choices. It offers the highest conversion and yield in a reasonable size reactor. The BFB reactor also offers a significantly higher MTY (mass time yield: tons of acrylonitrile production per ton of catalyst) compared to the fixed-bed reactor due to operation at a much higher and uniform temperature level. This is due to higher reaction rates at elevated temper-
HYDROCARBON PROCESSING / JULY 1999
atures. The fixed-bed reactor cannot Table 3. Heavy hydrocarbons conversion reaction mechanism be operated at as high temperatures With symbols With model as intended due to heat transfer limicompounds tation, which leads to a temperature Hv n-Par + H 2 ßà Lt n-Par + Gases nP-C 9H20 + H2 ßànP-C6H14 + C3H8 runaway within the reactor tube. A circulating fluidized-bed (CFB) reacHv i-Par + H 2 ßà Lt i-Par + Gases iP-C 9H20 + H2 ßàiP-C6H14 + C3H8 tor may offer equal or better MTY than ßà Hv i-Par ßàiP-C9H20 Hv n-Par nP-C 9H20 a BFB since the CFB can also be operLt n-Par nP-C6H14 ßà Lt i-Par ßàiP-C6H14 ated at a higher temperature. HowHv Nap + H 2 ßà Lt Nap + Gases N-C9H18 + H2 ßàN-C6H12 + C3H8 ever, the CFB would require a very tall ßà Lt Aro + Gases Hv Aro + H2 A-C9H12 + H2 ßà A-C6H6 + C3H8 reactor to achieve a conversion comßà Lt Nap Lt Aro + H 2 A-C6H6 + 3H2 ßàN-C6H12 parable to BFB. Lt n-Par + H2 ßà Gases nP-C 6H14 + H2 ßà2C3H8 In the projections shown in Table Lt i-Par + H2 ßà Gases iP-C 6H14 + H2 ßà2C3H8 2, the same feed gas composition has ßà Gases Lt Nap + H 2 N-C6H12 + 2H2 ßà2C3H8 also been assumed for all three reactor ßà Gases Lt Aro + H 2 A-C6H6 + 5H2 ßà2C3H8 types. In reality, due to safety reason, the allowable propylene content for the fixed-bed reactor has to be much lower than the value used. With this more diluted feed, the fixed-bed reactor would be even more unattractive than that indicated by the table. A performance projection by a general reactor model, as shown in Table 2, thus allows the user to promptly evaluate design alternatives and select the best design for primary consideration. Developing a suitable catalyst that has resistance to attrition when used in a BFB or CFB reactor, with minimum loss of activity and selectivity, is a parallel consideration. Other factors, such as mechanical aspects, cost of the reactor, reactor safety and operational flexibility, would play a role in the final decision. The results depend on the actual mechanism and kinetics that may vary substantially with actual catalyst formulation. However, once a given catalyst performance data are available or generated through a small test facility, commercial projections (as shown in Table 2) can be made with a fair degree of accuracy. The best reactor configuration or the optimum operating conditions of an existing unit can then be easily established. The general model can be used to screen alternate reactor design concepts for many other catalytic and noncatalytic reaction systems. These include: butane to maleic anhydride, naphthalene to phthalic anhydride, oxidative coupling of methane, hot gas desulfurization by zinc titanate and partial oxidation of paraffins. The model can be used to simulate lab, pilot-plant and commercial data from other reaction and reactor systems that include FCC catalyst regeneration in a CFB reactor, methanol synthesis in a double-wall tubular super-converter and ammoFig. 4. FCC simulation with a general reactor model. nia synthesis in multistage adiabatic quench reactor. EXAMPLE 2-CONVERSION OF HEAVY HYDROCARBONS In this example, the reactor model is used as a kinetic modeling tool for a typical refinery process that catalytically converts heavy hydrocarbon feedstocks. It is carried out in the vapor phase where feedstocks undergo many reactions including: hydrocracking, isomerization, alkylation, cyclization, hydrogenation and de-hydrogenation to form lighter products. A schematic reaction network for the process is shown in Fig. 3. To formulate a preliminary reaction mechanism for the product compositions, numerous process compounds are first divided into seven compound groups, i.e., heavy paraf-
Reaction number
1 2 3 4 5 6 7 8 9 10 11
fins (Hv Par), light paraffins (Lt Par), heavy naphthenes (Hv Nap), heavy aromatics (Hv Aro), gaseous hydrocarbons (Gas), light naphthenes (Lt Nap) and light aromatics (Lt Aro). Heavy and light paraffins are again subdivided into normal and iso- paraffins (Hv n-Par, Hv i-Par, Lt n-Par and Lt i-Par). The compound groupings are: Hv Par: C9+, i-C9+ Lt Par: C5–C8, i-C5 to i-C8 Hv Nap: Nap C9+ Hv Aro: Aro C9+ Lt Nap: Nap C5–C8 Lt Aro: Aro C6–C8 Gas: C3, C4, i-C3, i-C4, C3=, C4=, product gases
Each compound group is then represented by a model compound so that reaction stoichiometry can be easily
HYDROCARBON PROCESSING / JULY 1999
defined for both modeling and reactor material balance. This representation also allows estimation of the reaction equilibrium constants from standard free energies of formation of the model compounds. Although model compounds have been used to represent compound groups, actual molecular weights of these compound groups as estimated from the detailed component analysis of the feedstocks are used in the reactor model to simulate actual gas densities and mass throughputs into the reactor. A reaction mechanism based on model compounds is listed in Table 3. Rate expression for each reaction is obtained assuming an elementary reaction. For instance, the rate expression for reaction 1 becomes: r1
=k
e 01
− E1 / RT
C C − C C K
C1C7 1
2
1
where: r1 k01, E1
4
7
1
rate of reaction 1 pre-exponential factor and activation energy of reaction 1 R, T gas constant, reaction temperature C1,C7,C2,C4 molar conc. of nP–C9H20, H2, nP–C6H14 and C3H8 respectively K 1 equilibrium constant of reaction 1. Preliminary values of reaction activation energies are found in open literature on the kinetics of similar reaction systems. Using these values of activation energies, the reaction pre-exponential factors are then obtained by matching a preliminary set of experimental data with model predictions. All rate parameters ( k0 and E values) are finally refined by using all prescreened experimental data and a parameter estimation routine. Pilot plant performance of the above process using several feedstocks and a wide range of operat ing conditions were modeled satisfactorily. In these examples, the general model was used extensively to develop kinetic models for other reaction systems. The data used in the development of these kinetic models were again obtained from a variety of reactor configurations at lab, pilot plant and commercial-scale levels, and not just from ideal test units like a differential and or an integral reactor. These developments were possible because the model is able to extract true or intrinsic kinetics from almost any data source irrespective of influences of various mass and
heat transfer effects and hydrodynamic conditions of different reactors. The parameter estimation routine is helpful when determining the kinetic parameters for any reaction or reactor system. Array of applications. A general reactor model can apply to any combination between reaction and reactor systems. A general gas-solid reactor model such as the one used in the above examples should apply to: • Catalytic processes, such as oxidation, oxychlorination, ammoxidation, reduction, reforming, catalytic cracking (including FCC), isomerization, hydrogenation, dehydrogenation, shift/methanation, catalytic burners and ultra-rapid pyrolysis • Homogeneous gas-phase systems, such as gas burners, combustors, incinerators, thermal crackers and reactors • Noncatalytic gas-solid systems, such as catalyst regeneration, hot gas desulfurization, mineral processing, and gasification and combustion of solids.
Dr. Dutta is a senior staff consultant for
GTC Technology Corp., specializing in designing and implementing reactor models to industry. He has published more than 30 papers and a book chapter describing work in reactor modeling. He has served numerous times on advisory boards and research teams across the globe, addressing aspects of chemical engineering and process modeling. Dr. Dutta has a long history of industry experience, spanning over 30 years, having held senior positions for Simulations Sciences, SABIC Research and Development, Engelhard, Fluor Daniel and CE Lummus Crest. Ron Gualy is the business unit manager for the dimethyl terephthalate (DMT) busi nes s for GTC Technol ogy Cor p. He is respons ible for all marketing , research and application activities for DMT licensed technology worldwide. Mr. Gualy holds a BS degree in chemical engineering from Texas A&M University and is a registered professional engineer. He is a member of AIChE, American Chemical Society and The American Management Association. Mr. Gualy holds four patents concerning carboxylic acid recovery and has other patents pending. He has been with GTC Technology since its beginning six years ago.
Reprinted from Hydrocarbon Processing ® magazine, July 1999 issue, pgs. 45-53. Used with permission. F/1M/899 Copyright © 1999 by Gulf Publishing Company. All rights reserved. Printed in USA.
