gPROMS ProcessBuilder example

Whole-plant optimisation

Styrene, also known as Styrene monomer (SM), ethenylbenzene, vinylbenzene and phenylethene, is the precursor to polystyrene and several copolymers. Annual demand is nearly 25 million tonnes, and it is one of the chemicals industry's key building blocks.

SM is produced by dehydrogenation of ethylbenzene (see right) in a fixed-bed catalytic reactor. The reactor products are separated by distillation, with large columns and high reflux required because of the closeness in ethylbenzene and styrene boiling points.

Optimising process economics involves maximising the conversion to reduce the amount of ethylbenzene that must be separated. However it is not sufficient simply to optimise the reactor at the expense of the separation system, as there is a large recycle.

The process

The traditional styrene monomer process is shown below. It comprises an axial flow catalytic reactor in which ethylenzene is dehydrogenated in the presence of steam over a catalyst. Unconverted ethylbenzene is separated from the product and recycled.

The challenge

A large component of the cost of the process (and hence its profitability) is influenced by the capital cost of the reactor and separation equipment, and the energy requirements for the large distillation columns, which operate at high reflux.

The main challenge in designing and operating the process is to improve conversion and so reduce the amount of ethylbenzene that must be separated. However, minimising the size, and hence cost, of the reactor can result in increased separaton costs. When determining the optimal design it is essential to consider all key decisions simultaneously, including both the reaction and separation sections.

The approach

The solution is to perform a whole-plant optimisation using high-fidelity reactor and separation models. This determines the optimal values for all the decision variables simultaneously, to arrive at a truly optimal design.

The figure shows operating (green) and equipment configuration (red) decision variables.

Technology background

Whole-plant optimisation is typically performed using an economic objective function that includes annualised capital and operating costs.

Because the whole plant is optimised simultaneously, there are usually many (20–50) decision variables. Typically these include:

  • Equipment dimensions (eg. diameter, length, etc.)
  • Operating conditions (eg. temperatures, pressures, etc.
  • Equipment configuration decisions (feed tray location, number of stages, etc.)
  • Process synthesis decisions (include/exclude equipment, alternative process routes, etc.)

It is essential to use high-fidelity models; using approximate or simplified models will provide meaningless results. In many chemical processes only a small percentage improvement – which may nevertheless represent millions of dollars' improvement in process economics – is possible. Any error introduced by using simplified models can dwarf this.

Models should be validated against data where necessary – for example, to determine key constants such as reaction kinetic parameters.

The optimisation should include all significant equipment/process within the optimisation envelope, as well as cost calculations

The solution

gPROMS ProcessBuilder makes the use of high-fidelity reactor and separation models within an optimisation framework possible for the first time. This opens the potential to add significant value – for example, tens of millions of dollars improvement in process economics in one case.

The objective function used is:
Total annualized profit (MMUSD) = Annual revenue − annualised capital cost − operation cost

Optimisation variables are:

  • reactor radius
  • high-pressure steam flowrate
  • feed stage and total number of stages in the first column
  • boil-up ratio of first column
  • fresh ethylbenzene flowrate
  • superheated high pressure steam flowrate

The constraints are:

  • Conversion of ethylbenzene
  • Selectivity of styrene
  • maximum reactor temperature
  • minimum temperature difference in the two-stream heat exchangers
  • ethylene benzene at top liquid stream of first column of 0.1 mol%
  • ethylene benzene at styrene product of 1000 mol ppm
  • styrene at top liquid stream of 2nd column of 1 mol%

The objective function, constraints and decision variables are defined in the ProcessBuilder optimisation dialogs, and the optimisation executed.

The result

The table below shows the improvement in overall economics that ca be achieved:

Base caseOptimal case
Total annualised profit (million USD) 19.6 24.5
Annual revenue(+) 49.6 50.3
Annualised capital cost(-) 5.6 4.2
Operation cost(-) 24.3 21.6

Control variables

HP steam generation duty (kW) 2500 6998
Feed stage 1st column 9 15
Total number of stage 1st column (V/B) 20 22
Boil up ratio of 1st column (V/B) 1.00 0.62
Superheated steam flowrate (kg/s) 10.0 11.72
Reactor height (m) 12.19 13.00
Reactor radius (m) 1.37 1.25


The optimiser has made significant changes to the configuration of column 1, reducing the number of stages and thus reducing the capital cost. The feed stage has been moved to the optimal location for the new configuration. The reactor diameter has been reduced slightly, and operating conditions changesed significantly.

The revenue has increased by $0.7m per year, and the combined capital and operating costs have dropped by $4m. As a result, the annualised profit has increased from $19.6m to $24.5.

Our website uses cookies so that we can provide a better browsing experience. Continue to use the site as normal if you're happy with this or find out more about cookies