Reactor modelling, simulation and optimisation

Optimising reactor design and operation with high-fidelity models

gPROMS provides all the capabilities necessary for reaction modelling within a single package.

Reaction is a fundamental process in virtually all sectors of the Chemical Process Industries – from pharmaceuticals through chemicals and petrochemicals to refining.

Despite this, the modelling of reaction processes remains relatively unsophisticated:

Much current process flowsheeting simulation application uses equilibrium or stoichiometric reaction modelling.
More sophisticated reaction models tend to be executed in isolation, without the benefit of flowsheeting or optimisation capabilities.

PSE's gPROMS^® is rapidly becoming the reaction modelling environment of choice for many of the world's leading process companies.

As a company we are focused on becoming the world's leading supplier of reaction modelling technology and services.

The gPROMS advantage

The gPROMS advantage is that it provides everything necessary for advanced reactor modelling in a single package. Click on the graphic for more information.

Microkinetics of a steam reforming reaction – click to enlarge

Microkinetics of a steam reforming reaction (sample) – click to enlarge

Kinetic parameter confidence fit from model-based data analysis

Kinetic parameter confidence fit from parameter estimation and subsequent model-based data analysis

Reactions take place around, on, and in a catalyst particle – click to enlarge

Reactions take place around, on, and in a catalyst particle – click for detail

Hybrid modelling schematic overview – click for details

Links to CFD models – click for more information

Tubular reactor model from the AML:FBCR – click to enlarge

Tubular reactor model from the AML:FBCR – click for more information

Reaction sets and reaction kinetics

The key to successful modelling of reacting systems is to determine the set of reactions occuring and the rate mechanisms for those reactions.

PSE provides a number of tools and services for this:

the Advanced Model Library for Fixed-Bed Catalytic Reaction (AML:FBCR) provides a detailed framework for the modelling of catalytic reaction.
many other detailed reaction models that can be customised as required. These can be supplied on request.
gPROMS' built-in model validation facilities, which use parameter estimation and model-based data analysis techniques to fit kinetic parameters and provide related confidence information.
parameter estimation can also provide model discrimination information to choose the best of a number of proposed reaction sets
a well-proven project methodology to ensure rapid project delivery.
Model-Based Innovation guidance on integration of R&D and modelling/engineering. This includes advice on experimentation programmes, including model-based experiment design service to minimise the number and cost of experiments.

Microkinetics

Where sufficient data exists in the literature, or sufficient experimental data is available, it is possible to model the reaction set at a microkinetic level. Providing the kinetic parameters can be estimated correctly, the resulting reaction set definition is 'universally true' for virtually any set of conditions.

This involves breaking down the reactions into fundamental reactions, as shown on the right for a steam reforming process.

gPROMS' numerical solution capabilities are capable of handling the hundreds of implicit reaction equations (or tens of thousands, for distributed systems) resulting from this approach.

Determining kinetic parameters using models and data

Another key to successful modelling of reaction systems is accurate kinetic parameter values.

Providing suitable experimental data exists, these can be determined using gPROMS' extensive model validation facilities.

gPROMS' advanced parameter estimation uses optimisation techniques to extract high-accuracy parameter information from experimental data. Selected model parameters are adjusted to give the best fit to observed data.

Multiple parameters can be extracted simultaneously using measurements from any number of dynamic and/or steady-state experiments.

Maximum likelihood techniques allow the errors inherent in practical experimentation to be taken directly into account by estimating them simultaneously with the model parameter values.

If gaps in the data are identified, gPROMS' model-based experiment design facilities can be used to design the most effective experimentation programme, in order to maximise parameter information at minimised time and cost.

The determining of kinetic parameters follows a well-established methodology, which deploys customers' own experimental data where possible for maximum accuracy.

In addition to parameter values, model-based data analysis yields estimates of the accuracy of these values. Modern risk analysis techniques can translate this information into an assessment of the risk involved in using these parameter values for model-based decisions.

Modelling the complexity of catalytic reactions

The complex interactions between the various components of a reaction system at a microscopic level ultimately govern the equipment requirements and operating envelope at the macroscopic level.

For example, the outcome of a plant-wide optimisation of a chemical plant can depend strongly on the rate of diffusion of reactants through microscopic pores in the catalyst particles.

In order to capture the effects of these interactions, it is sometimes necessary to model systems in great detail.

At their most comprehensive, PSE's catalytic reaction models include reactions:

in the bulk fluid
in the film
on the catalyst surface
within the catalyst pores
on the catalyst surface within the pores

The models takes into account the rate-limiting diffusion of reactants and products to and from catalyst sites.

Generally, the approach taken in PSE's reactor modelling methodology is to determine a suitable level of detail given the model requirements; a detailed design model has very different performance and accuracy requirements from an operator training model.

Modelling the complexities of reactors

Once the reaction set has been defined and validated, it is included in a reactor model.

gPROMS' advanced modelling capabilities make it possible to cover a variety of equipment geometries and configurations. If necessary, gPROMS models can be linked to CFD models to provide additional accuracy for irregular geometries.

The gPROMS AML:FBCR provides detailed fixed-bed models such as the tube shown on the right. PSE can supply many reactor configurations, such as fluidised bed models, catalyst monolith, falling film reactors and more.

gPROMS modelling environment advantages

gPROMS has many general advantages as a modelling environment, in addition to capabilities such as parameter estimation mentioned above.

Flowsheeting is a core functionality of gPROMS, as is steady-state and dynamic simulation

Polymer bimodal molecular weight distribution

Distributed systems: bimodal molecular weight distribution for polymer

Powerful modelling language. gPROMS language allows the description of virtually any reaction phenomenon. You are not restricted to a set of predefined relationships.

Hierarchical modelling capabilities. In gPROMS, models can be embedded within other models ad infinitum. For example, it is possible to include a catalyst model within a tube model, which may then be included within a full reactor geometry model, which may in turn be included in a full plant model that incorporates cooling systems, separation sections and recycles, and so on.

Distributed systems modelling. gPROMS allows easy definition and solution of distributions to any practical number of dimenstions. For example, a fixed-bed catalytic reactor tube is described using three dimensions:

axial, representing the variation of properties and conditions along the tube
radial, representing the variation of properties and conditions from the centre of the tube to the radius
intra-pore, representing the variation of properties and conditions along the length of catalyst pores

Distributions can also be in terms of properties – for example, polymer molecular weight or age of particles in a fluidised bed reactor – as well as spatial.

Flowsheeting capability, allowing you to simulate and optimise the entire reaction flowsheet rather than just a single unit, or include separation units and recycles.

Steady-state and dynamic modelling. gPROMS' inherent dynamic capabilities means that you can perform dynamic simulation and dynamic optimisation easily.

Powerful numerical solution capabilities. gPROMS was specifically designed to solve large numbers of equations, for steady-state or dynamic solution. Some reactor models number hundreds of thousands of equations.

Optimisation capabilities. gPROMS has industry-leading capabilities for steady-state, dynamic and mixed-integer optimisation. These can be used to provide direct answers – rather than by trial-and-error simulation – to design and operational questions such as "what is the optimal grade change policy?" or "where should I locate the feed tray to this column given the likely feed disturbances?".