Advanced applications: batch process optimisation
The dynamics of batch processing make recipe optimisation a challenging task unsuited to traditional steady-state simulation tools.
gPROMS ProcessBuilder combines dynamics, high-fidelity process models and optimisation capabilities to allow you to truly optimise batch processes.
By optimising process design and operating policy you can enhance product quality, minimise batch time, increase asset utilisation and easily develop recipes for different equipment or product qualities.
The example below shows an industrial batch esterification process. The unoptimised process results in around 60 tonnes per batch with 96.2 wt% purity, over a batch time of 8 hours and 10 minutes.
Figure 1 – Batch process flowsheet & recipe
There is significant scope for optimising the process to minimise batch time (in order to maximise throughput), minimise cost of raw materials and utilities and maximise profitability, while simultaneously improving product purity.
Some key constraints and potential decision variables are shown on the diagram in red and green respectively.
Figure 2 – Batch process, showing potential decision variables in green and constraints in red
The batch process was simulated then optimised in the following steps:
Step 1 – Create the process model
The process flowsheet, shown in Figure 3, was created using gPROMS ProcessBuilder's drag & drop flowsheeting capabilities.
Initial reaction kinetic parameters were taken from literature. For additional accuracy the paraeters were refitted to recent experimental data using the gPROMS parameter estimation capability.
Figure 3 – batch process flowsheet
Step 2 – Create the operating policy
The batch operating policy was defined using the gPROMS graphical TASK language. The schedule is shown on the right.
Step 3 – Simulate a batch
The simulation was then executed, with the operating policy defined in Step 2 applied to the process flowsheet to dynamically simulate a batch. The plots below show the purity and batch temperature over the course of the simulation.
It can be seen that the batch takes around 8 hours and 20 minutes (just under 30,000 seconds) to complete.
Intermediate and final product purity
Figure 4 – Selected results from the batch simulation
Step 4 – Define and run the batch optimisation
The simulation results reflect the process designers' best first guess at a schedule. In many cases, if this produces product satisfactorily (as is the case here) it will be applied and used without modification.
However many schedules can be significantly improved by using optimisation techniques to minimise batch time while maintaining or improving quality.
The optimisation was set up as follows:
- Define the objective function. The objective specificed in the gPROMS optimisation setup was to minimise the operating cost.
- Define constraints. These were the items shown in red in figure 2 – e.g. batch vessel maximum temperature and pressure, product purity, etc.
- Choose decision variables. These are taken from the items shown in green in figure 2 – for example feed addition rates, steam rates and so on.
The optimisation was then executed.
The optimisation results show that it is indeed possible to improve the operation significantly.
Optimisation shows that batch time can be reduced to just over 7 hours, a saving of over an hour. At the same time product purity increases from 96.2% to 97.1%, providing a higher-value product.
Batch temperature showing optimal batch length (red) vs original (blue)
Optimised energy (steam and cooling) profiles
Figure 5 – Selected results from the batch process optimisation
The profiles for the optimal case vs the original simulation case are shown above.
The changes doubled the profit margin for the product.