Model-Based Engineering

Integrating models and data to accelerate engineering

Chemical Engineering November 2008 modeling article

How does MBE work? See our article in Chemical Engineering

A Model-based Engineering (MBE) approach applies high-fidelity predictive models in combination with observed (laboratory, pilot or plant) data to the engineering process.

The objective is to be able to explore the design space as fully and effectively as possible and support design and operating decisions with accurate information.

Typical application is in new process development, including scale-up, or optimisation of existing plants.

Closely-related is Model-Based Innovation, where a similar approach is used at R&D level to bridge the gap between experimentation and subsequent engineering design, and capture corporate knowledge in usable form, typically for early-stage process or product development.

What does model-based engineering involve?

Repsol's new HPPO process design is an excellent example of MBE in practice. Simultaneous whole plant optimisation using high-fidelity models resulted in millions of Euros' improvement in process economics and the elimination of two distillation columns.

MBE places high-fidelity predictive models at the heart of process design or operational analysis.

Initial project effort is put into constructing a high-fidelity model of the plant or process that is predictive over the entire range of interest.

This model is then used to optimise design or operation, exploring a wide design space rapidly and at low cost, and applying optimisation techiques to determine answers directly rather than by trial and error simulation.

MBE is based on three core approaches:

First-principles modelling, where all relevant phenomena are described to an appropriate level of chemical engineering first principles representation. This typicaly involves detailed mass transfer, heat transfer and reaction equations.
Multiscale modelling, where phenomena at all relevant scales are taken into account. The diagram on the right shows, for example, the scales that need to be taken into account for a multitubular reactor. The phenonema occurring at a microscale in a catalyst pore can have a significant influence on the overall (macroscale) reactor design.

Integration with experimental data, by applying a model-targeted experimentation approach to refine the model and at the same time maximise the effectiveness of the experimental programme.

The guiding principle of model-targeted experimentation is "use experimentation to improve the accuracy of the model (rather than the process itself), then use the model to optimise the process design or operation".

Why apply Model-based Engineering?

Key objectives of MBE are to:

Accelerate process or product innovation, by providing fast-track methods to explore the design space while reducing the need for physical testing.
Minimise (or effectively manage) technology risk by allowing full analysis of design and operational alternatives, and identifying and addressing areas of poor data accuracy.

We made rapid progress in bridging experiments and process design for lactide production

PURAC Biochem,
PSE AM 2008

Integrate R&D experimentation and engineering design in order to maximise effectiveness of both activities and save cost and time.
Reduce operating costs or improve throughput and product quality through high-accuracy analysis and model-based optimisation techniques.
Reduce the amount of experimentation, through better-targeted and better designed experiments
Reduce (but not eliminate) the requirement for pilot plant testing. Many design options can be explored and eliminated using predictive modelling before the best candidates are chosen for pilot plant testing.

How does MBE work in practice?

MBE applies a family of methodologies centred on high-fidelity predictive models in a structured way.

First-principles models and experimental data are combined to create a high-fidelity predictive representation of the key phenomena occuring in any process, often (particularly in the case of reactor design) through model-targeted experimentation.
These sub-models of the phenomena are then used to build high-fidelity models of the full-scale process.
The model is used in steady-state and dynamic simulation and optimisation studies to achieve the project objectives.

If necessary it is possible to deploy a variety of well-established techniques – for example, combining the physics and chemistry models derived above with Computational Fluid Dynamics (CFD) hydrodynamic information, or using population balance models – to take into account mixing and other effects introduced at larger scales.

Examples of application

For comprehensive examples of application, see the following articles:

LG Chem article Optimize terephthldehyde operations [Hydrocarbon Processing, April 2007], describing the model-based design of a multitubular reactor
Süd-Chemie article Enhanced methods optimize ownership costs for catalysts [Hydrocarbon Processing, June 2007], describing the application of MBE techniques to catalyst deactivation analysis for methanol synthesis and optimisation of catalyst loading
Repsol article Improve engineering via whole-plant design optimization [Hydrocarbon Processing, December 2010], describing the simultaneous optimisation of reactor and separation section that improved plant economics by tens of millions of Euros.