reactor modeling tools - an overview

Refinery Process Stream Purification Refinery Process Catalysts Troubleshooting Refinery Process Catalyst Start-Up / Shutdown Activation Reduction In-situ Ex-situ Sulfiding Specializing in Refinery Process Catalyst Performance Evaluation Heat & Mass Balance Analysis Catalyst Remaining Life Determination Catalyst Deactivation Assessment Catalyst Performance Characterization Refining & Gas Processing & Petrochemical Industries Catalysts / Process Technology - Hydrogen Catalysts / Process Technology – Ammonia Catalyst Process Technology - Methanol Catalysts / process Technology – Petrochemicals Specializing in the Development & Commercialization of New Technology in the Refining & Petrochemical Industries

Web Site: www.GBHEnterprises.com

GBH Enterprises, Ltd.

Process Engineering Guide: GBHE-PEG-RXT-817

Reactor Modeling Tools - An Overview

Process Disclaimer

Information contained in this publication or as otherwise supplied to Users is believed to be accurate and correct at time of going to press, and is given in good faith, but it is for the User to satisfy itself of the suitability of the information for its own particular purpose. GBHE gives no warranty as to the fitness of this information for any particular purpose and any implied warranty or condition (statutory or otherwise) is excluded except to the extent that exclusion is prevented by law. GBHE accepts no liability resulting from reliance on this information. Freedom under Patent, Copyright and Designs cannot be assumed.



Process Engineering Guide: Reactor Modeling Tools CONTENTS 1 SCOPE 2 OPTIONS IN REACTOR MODELLING

2.1 General 2.2 Level of Complexity of Model 2.3 Mode of Operation of Model 2.4 Deterministic versus Empirical Modeling 2.5 Platforms for Model 2.6 Steady State versus Dynamic Model 2.7 Dimensions Modeled in Reactor 2.8 Scale of Modeling for Multiphase Reactors 2.9 Writing and Using the Model

APPENDICES A CHARACTERISTICS OF DIFFERENT REACTOR MODELS B NEEDS FOR MODELLING AT DIFFERENT SCALES IN

HETEROGENEOUS CATALYTIC REACTORS C REACTOR MODELS EMPLOYED WITHIN GBHE DOCUMENTS REFERRED TO IN THIS PROCESS ENGINEERING GUIDE



1 SCOPE This Guide provides a general overview of reactor modeling needed for reactor design and manufacturing improvement, and the issues that need to be addressed before starting to write a reactor model. Appendix C gives a list of the reactor models that are available within GBHE, with some basic information and contacts. More detailed information on some of these programs is given in their respective guides. 2 OPTIONS IN REACTOR MODELLING 2.1 General

There are a large number of options and types of reactor model and the choice of model for a particular reactor will depend on the objectives of the model. It is normally the case that a given reactor will have a number of different models with each model set up to achieve various objectives.

The options for reactor modeling are listed below and the issues are dealt with in more detail in 2.2 to 2.9. (a) Level of complexity of model:

(1) Equilibrium model. (2) Simple kinetic model. Apparent kinetics plus residence time

distribution (and heat input distribution). (3) Mass transfer/kinetic model. Intrinsic kinetics plus interphase

mass and heat transfer plus flow patterns (use of computational fluid dynamics).

(b) Mode of operation of model:

(1) Design of reactor. (2) Rating of existing reactor. (3) Performance evaluation.

(c) Deterministic versus empirical modeling.



(d) Platforms for model:

(1) Standalone program. (2) Spreadsheet package. (3) Flowsheeting package;

(i) Standard reactor module. (ii) User supplied reactor module.

(e) Steady state versus dynamic model.

(f) Number of dimensions modeled in reactor.

(g) Scale of modeling for multiphase reactors.

(h) Writing and using the model.

2.2 Level of Complexity of Model

It is normal to start with simple models for initial evaluation of new processes and progress to more complicated models as the knowledge to build the model and the detail required in the reactor design increases.

The level of complexity will depend on:

(a) What variables need to be explored?

For example, is the model to be used to determine the size of reactor for a given conversion, or is it also going to be used to calculate by-product formation?

(b) What mechanistic features need to be included?

For example, does it need to determine the reactant residence time distribution, or will this be assumed? Does it need to calculate a heat transfer coefficient within the model or will it be calculated off-line?

(c) How much time (and effort) is available for model development?

Three basic levels of complexity are given in 2.2.1 to 2.2.3.



2.2.1 Equilibrium Model

The equilibrium model is used for the evaluation of new process routes in order to estimate the conversion attainable in the reactor.

Equilibrium calculations can be performed using standalone models e.g. STANJAN as part of a short cut design study, or are supposedly able to be carried out within modern flowsheeting packages as part of the flowsheet model.

2.2.2 Simple Kinetic Model

A simple kinetic model will typically be one dimensional, modeling apparent kinetics (which includes mass transfer effects) within an imposed residence time distribution. If the reactor is not adiabatic or isothermal a heat input distribution is also imposed.

Simple modeling of this type can be performed using spreadsheets e.g. Excel. However a common problem with these and more complicated models is 'stiffness'. This arises when the various differential equations in a simultaneous set have time constants that differ by orders of magnitude. The effect is that in ordinary integration algorithms the step length for accuracy and stability is very small and computation would take too long.

Hence more advanced tools or models become necessary. Options are:

(a) Build kinetic equations into a general model framework e.g.

KINPACK or BATCHCAD.

(b) Write bespoke Fortran model with advanced solver. This may take longer, but gives more flexibility for moving to more complicated models and integrating to flowsheeting programs.



2.2.3 Mass Transfer/Kinetic Model

The mass transfer/kinetic model has a higher level of complexity than the simple kinetic model and will include any or all of:

(a) Intrinsic kinetics plus calculated mass transfer.

(b) Calculated flow patterns and residence time distributions.

(c) Calculated heat transfer.

(d) Two or three dimensional modeling.

There is probably an infinite level of complication that can go into these models, so it is important to understand the objectives of the model and the important factors affecting the reactor performance, so that a reasonable level of complexity can be used.

2.3 Mode of Operation of Model

A reactor model can be used in a variety of modes. These affect which are the input variables and which are the output variables of the model.

2.3.1 Design of Reactor

The model is set up, for example, to calculate the size of reactor needed in order to achieve a given conversion, i.e. the model input is the conversion and the model output is the reactor size. The model needs to be run in this mode for design optimization.

2.3.2 Rating of Existing Reactor

When a reactor is designed it is usual to calculate how it would perform under a range of operating conditions. The model is set up, for example, to calculate the conversion from the input streams and reactor geometry. The model needs to be run in the rating mode for operation optimization.

This is the easiest way to set up a model.



2.3.3 Performance Evaluation

The model is set up to check or monitor the performance of the reactor compared with what it is expected to achieve. The model inputs are therefore the input and output streams and reactor geometry and the model output is some reactor performance measure e.g. catalyst activity.

It will be normal for the model to be operated in all the different modes over the life of a project. Since the natural way to write a model is for rating mode, there is an issue of how to run the model in the design mode and performance mode. There are three ways in which to do this:

(a) the model has to be run a number of times changing model input

variables;

(b) the model is set up with internal iteration loops to automatically adjust reactor input variables to obtain the reactor output variables that are set;

(c) the model is integrated into a flowsheet modeling platform such as

Aspen HYSYS V8, PRO II or SPEED-UP and the convergence routines within these platforms are used to adjust reactor inputs to obtain the required output variables.

A lot of time can be saved by considering all the input and output variables that may be needed when the model is first written, so that changes do not have to be made later on. 2.4 Deterministic versus Empirical Modeling

A table with the characteristics of different types of model is given in Appendix A.

2.4.1 Deterministic or Mechanistic Modeling

Deterministic or mechanistic modeling involves building a model completely from known data obtained from kinetics experiments or the literature. The model can then be used to predict the effect of design and operational changes. These models can only be written for reactors where the technology is fairly well established.



2.4.2 Semi-empirical Modeling

Semi-empirical modeling uses standard equations or relationships, which are known to fit the reactor performance, but a small number of constants in the model are fitted by comparing the model predictions to plant data. It can be used predicatively to some extent, but care has to be taken and its accuracy questioned for extrapolation outside the conditions for which the constants were fitted.

2.4.3 Empirical Modeling

Empirical modeling involves characterizing a reactor from data obtained from the reactor itself. It can only be used for predicting performance within the design and operating conditions for which data has been fitted. Empirical modeling is used for plant control and can also be used to establish which input variables have an effect on the output variables. The latter is useful as a prelude to writing a more complex model to identify all the relationships that a more advanced model needs to include. Empirical modeling can use a variety of techniques ranging from multivariate regression to neural networks. Empirical modeling has to be used for plants where the technology is an art rather than a science and is too difficult to be modeled deterministically.

2.5 Platforms for Model

It is important to choose the appropriate platform or platforms upon which the model is going to be run. For example it is normal to start off with a model as a standalone model that will be run on its own, but it will later be required to integrate the model into a flowsheeting system.

2.5.1 Standalone Model

A standalone model is used at any stage of reactor modeling. If the model does not take too long to solve, it can be incorporated into a flowsheeting system - see 2.5.3.



2.5.2 Spreadsheet Model

A spreadsheet model can be used at the early stages of design for simple modeling or scouting and for running empirical models. However, it has limitations in that it can lead to problems with iteration and it cannot be interfaced into flowsheeting packages. Since the structure of spreadsheet models is difficult to check, spreadsheet models cannot be used with confidence, except by the person who wrote them and should not be made generally available.

2.5.3 Flowsheeting Package

When a reactor model is built into a flowsheeting package, see Appendix C.6, it gives a number of advantages:

(a) The reactor model can use the flowsheet package to access

physical properties, rather than having to supply them separately. However, care is needed when using these packages as there is a tendency for some commercial software to offer default data without appropriate warnings when results are given outside the valid range of the data.

(b) The algorithms within the flowsheeting package can be used

to do iteration around the reactor, for instance in design mode, to adjust the reactor size to obtain a given performance.

(c) A whole system around the reactor, including separation

equipment, can be modeled so that the effects of changes in reactor performance can be followed through the flowsheet.

(d) When the financial drivers associated with the reactor and

the reactor system are added, the algorithms within the flowsheeting package reactor model can be used to do a design or performance optimization of the reactor system.



There are two ways that the reactor model can be built into the flowsheeting package: (1) Standard Reactor Modules

Flowsheeting packages such as ASPEN and PRO II have standard reactor modules which can be used to model a reactor. However, these are generally inflexible and do not accommodate the peculiarities that need to be modeled to make the reactor model realistic.

(2) User Supplied Reactor Module

All sequential modular flowsheeting packages such as GENIE, FLOWPAK, ASPEN and PRO II allow Fortran models of reactors to be incorporated into the flowsheet package, see Appendix C.6. A harness needs to be written around the standalone reactor model in order to incorporate it into a flowsheet package.

Equation based flowsheeting packages such as SPEED-UP, need the reactor equations to be written into the package and cannot accept Fortran subroutines.

2.5.4 Computational Fluid Dynamics (CFD) Packages

CFD packages can be extended to be used as reactor modeling programs. The CFD package itself gives an accurate residence time distribution (RTD) of reactants in a reactor and kinetics can be added to describe the reaction. This approach is applicable when the RTD is crucial to the reactor performance and idealized RTDs such as plug flow and well mixed are not appropriate. This is the case in liquid reactors when the reactions are fast and are therefore controlled by mixing rate, or in fixed bed reactors where there is poor flow distribution.



2.6 Steady State versus Dynamic Model

While it may seem to be ideal to write reactor models so that they can be run either in the steady state or dynamically, this is not usually the case. There is quite a lot of extra investment required in reactor data, model writing time and run times as well as problems with robustness for a dynamic model compared with a steady state model. Dynamic models are normally written to answer specific questions influencing the design of the reactor under transient conditions, or to identify control problems. The effects of reactor start-up and shut down on the rest of the flowsheet can usually be assessed by running a steady state reactor model at different points in the startup sequence. The dynamic models that are written usually use simplified kinetics or contain fewer features than steady state models, see Appendix C.4.

2.7 Dimensions Modeled in Reactor

Most reactor models are zero or one dimensional models. For example a stirred tank is a zero dimensional model, while a plug flow reactor is a one dimensional model.

It is sometimes necessary to go to two or even three dimensions. These multi-dimensional models are normally only needed as standalone models and can be used to calculate corrections or other approximations to be added to a one dimensional reactor model to be used for incorporation into flowsheeting packages.

Examples are: (a) A two dimensional model is needed to determine the heat transfer from

the fluid to the wall of a packed tubular reactor, but once this has been calculated at two or three different conditions, the information can be captured in a one dimensional model as a heat transfer coefficient between a bulk fluid temperature and the wall temperature. Checks may subsequently be needed using the two dimensional model to calculate maximum or minimum fluid temperatures.



(b) Two or three dimensional models are needed to model fluid mixing using computational fluid dynamics, but the results of the mixing modeling can be used to calculate a residence time distribution, or used to characterize the reactor as a series of zones, which can then be used in a one dimensional kinetic model. If the mixing is such that a one dimensional kinetic model is not adequate, then a two dimensional kinetic model is needed. However if this is the case for a new reactor, the reactor design is probably wrong and the mixing intensity needs to be increased.

2.8 Scale of Modeling for Multiphase Reactors

While most reactor models are at the reactor scale, there is often a need to model at smaller scales. These sorts of models can cover a large range. A table of how the performance of heterogeneous reactors can be affected by the properties at different scales and hence the potential need for modeling at these scales is shown in Appendix B. The models that are written at the smaller scale are nearly always standalone models for parts of the reactor that are used to calculate parameters that are subsequently used in the full reactor model.

2.9 Writing and Using the Model

Writing and using the model will depend on the type of model:

(a) Spreadsheet Models

Spreadsheet models will normally be written and used by the process engineer or scientist who is collecting the data or designing the reactor.

(b) Standalone Models

Standalone models should ideally be written jointly by the process engineer or scientist and a mathematician. With simple models, the mathematical input will be small and can be done by consultation. As models become more complicated a greater mathematical input is required. It is best if the process engineer or scientist supplies the equations and relationships and the mathematician builds the model.



Mathematicians are able to use the most appropriate algorithms to solve problems with stiffness and will use standard codes as much as possible.

(c) Integrating into Flowsheeting Packages

Integrating user supplied modules into flowsheeting packages can be done more efficiently by an experienced person. If there is not an experienced person in the group, help should be sought from an experienced Process Systems Engineering Specialist.

(d) Empirical Models

There is a potential danger with empirical modeling that although a model fits existing data very well, the predictive power of the model is very poor. This is especially the case with neural network models. Because they are very flexible, they are able to fit the errors in data. Advice should be sought from a statistician or mathematician to get a view on the empirical technique to be used and the uses to which the model can be put.



APPENDIX C REACTOR MODELS EMPLOYED WITHIN GBHE C.1 EQUILIBRIUM MODELLING EXCEL 2007: Multiple Regression Data Analysis Add-In Platform PC aspenONE Engineering V8.4 Platform PC STANJAN Platform PC C.2 KINETIC MODELLING C.2.1 Kinetics Fitting Packages GNU Octave Platform PC BatchCAD 7.0 (Rate) Platform PC C.2.2 General Kinetic Models EXCEL 2007: Multiple Regression Data Analysis Add-In Platform PC Can be used to model simple kinetics in idealized residence time distributions (RTDs) of one or more phases. Chemical WorkBench Platform PC KINPACK Platform PC – Fortran Incorporates complicated kinetics in idealized plug flow, stirred pot or batch reactors.



BatchCAD (REACTION) Platform PC Dynamic kinetic model for batch reactors. CHEMKIN Platform PC Gas phase and gas/solid chemical kinetic and equilibrium modeling of aeronomic (Atmospheric, etc.) Chemistry, combustion and heterogeneous catalysis, for stirred tanks, CSTR, shock tube and flames. C.2.3 Specific Kinetic Models REFORMER - SimSci-Esscor’s Reformer Reactor Model REFORMER – Bozedown Model Platform PC Dedicated to aromatics reformers. Uses fundamental kinetics, and integrates reactors into a dedicated flowsheet package including heaters and compressor. CHEMCAD Platform PC Dedicated to benzene hydrogenation reactor. C.3 KINETIC MODEL WITH MASS AND/OR HEAT TRANSFER Models are reactor type dependent. C.3.1 Mass Transfer Computational Fluid Dynamic (CFD) Codes Open FOAM Platform PC Kinetics can be added to CFD codes in order to model the kinetics in an accurate real residence time distribution.



COMSOL Multiphysics Platform PC C.3.2 Liquid/Gas Reactors CHEMCAD Platform PC COMSOL Multiphysics Platform PC XREACT Platform PC Models kinetics and mass transfer with real RTDs of both phases using RTD information from an associated CFD package. One, two or three dimensions. C.3.3 Fluid/Solid Catalytic Reactors These models are currently all process specific, but it should be possible to amend an existing model for a new application. ADv5 Platform PC Design and performance of steam raising shift and methanol converters with catalyst in tubes or on the shellside. Kinetics includes pore diffusion and heat transfer coefficients are calculated on both shell and tube sides. OTHERS KHIMERA Platform PC Chemical kinetics simulation software tool developed by Kintech Lab Wolfram SystemModeler Platform PC Modeling and simulation software based on the Modelica language



C.4 DYNAMIC MODELS SPEED-UP Platform PC General dynamic modeling platform. Computational Fluid Dynamics Packages AUTODESK CFD Fluid Flow Simulations Platform PC Can be used for dynamic modeling of kinetics and fluid flow. Russian Reactor Model Dynamic model of fixed bed reactor for modeling reactor dynamics or catalyst deactivation. C.5 SMALLER SCALE MODELS These are reactor type dependent. C.5.1 Liquid/Gas Reactors Micro-CFD Models the mass transfer around individual bubbles or droplets. C.5.2 Fluid/Solid Catalytic Reactors EFFPAK Platform PC Calculates kinetics, mass transfer and heat transfer within porous catalyst particles given pellet transport properties.



DIFFCALC Platform PC Will calculate particle diffusion properties from structural parameters of particle. C.6 FLOWSHEETING PACKAGES These will provide the capability for solving a reactor integrated into a flowsheet. C.6.1 Sequential Modular Aspen Plus, Aspen HYSYS, Aspen Custom Modeler Standard modules supplied in the packages can be used for simple models. User supplied subroutine can be built into packages, but need own internal solver. C.6.2 Equation Based SPEEDUP Platform PC Package solves reactor at the same time as it solves the flowsheet using the same solving routine. The reactor can be split into a number of sections within the package. Can be used as a platform for XREACT - see C.3.2. C.7 BATCH REACTORS - HEAT TRANSFER PROGRAMS ProSim BatchReactor Platform PC Cooling an agitated batch reactor with an external jacket with a non-isothermal service fluid in turbulent flow. Heating an agitated batch reactor with an external jacket with steam.



STM LIMP Heating an agitated batch reactor with limpet coils with steam PRO/II is a steady-state process simulator (process simulation) for process design and operational analysis for process engineers in the chemical, petroleum, natural gas, solids processing, and polymer industries. It includes a chemical component library, thermodynamic property prediction methods, and unit operations such as distillation columns, heat exchangers, compressors, and reactors as found in the chemical processing industries. It can perform steady state mass and energy balance calculations for modeling continuous processes.



DOCUMENTS REFERRED TO IN THIS PROCESS ENGINEERING GUIDE This Process Engineering Guide makes reference to the following documents. PROCESS ENGINEERING GUIDES GBHE-PEG-RXT-800 How to use the Reactor Technology Guides GBHE-PEG-RXT-801 Chemical Process Conception GBHE-PEG-RXT-802 Residence Time Distribution Data GBHE-PEG-RXT-803 Data and Computer Programs for Calculating

Chemical Reaction Equilibria GBHE-PEG-RXT-804 Physical Properties and Thermochemistry for

Reactor Technology GBHE-PEG-RXT-805 Solid Catalyzed Gas Phase Reactor Selection GBHE-PEG-RXT-806 Fixed Bed Reactor Scale-up Checklist GBHE-PEG-RXT-807 Reactor and Catalyst Design GBHE-PEG-RXT-808 Solid Catalyzed Reactions GBHE-PEG-RXT-809 Homogeneous Reactors GBHE-PEG-RXT-810 Gas - Liquid Reactors GBHE-PEG-RXT-811 Novel Reactor Technology