Simulation and optimization of a batch esterification process
This case shows how simulation and optimization can be used to gain a deeper understanding of a batch esterification process and how its recipe can be optimized to maximize profitability. It highlights the advantages of dynamic simulation over more conventional trial-and-error/manual optimization techniques. The batch time and utility costs are reduced while still maintaining product purity standards.
Business implications and objectives
Specialty chemical processes deliver high purity products to a range of different customers. To be able to change over from one product to another in a cost effective manner, processes need to be agile and efficient from the start. However, reducing production costs while increasing throughput and maintaining product purity is not a trivial task. The aim of this case is to simulate the process to gain futher understanding of key parameters so that it can be successfully optimized with the following aims:
- minimise batch time
- maximize throughput
- minimise cost of raw materials
- maximize profitability
The batch esterification process
The batch esterification process is shown in Figure 1. The reaction is a two-step homogeneous reaction that takes place in water. The first step, producing reactant C from A and B, is very fast while the second step, to produce the product D, is slower and follows a catalytic route.#
The recipe for the process is complex, involving a rangne of process steps for reaction, reflux and storage, recovery, stripping, cooling and steam sparging as show in Figure 2.
Figure 2. The batch esterification recipe. The bottom table shows that the batch time is relatively long. There is also a significant consumption of high pressure steam which accounts for a large part of the overall cost for the process.
Simulation of the batch esterification process – key findings
To gain better understanding of how key process parameters affect process performance a process a simulation in gPROMS Process was set up. During the steam sparging step a couple of throughs in the temperature plot were identified. These could not be explained by a reduction in energy supplied to the reactor by high and low pressure steam. In fact they were the result of pressure drops in the reactor – the system is heavily dependant on pressure as shown in Figure 3.
It was also found the temperature alone is not a sufficient driving force for purity. Purity plots of the intermediate C and product D highlight that although all of intermediate C is converted to product D by the end of the esterification stage, the process benefits greatly in pressure reduction during the stripping of reactant B and water from the reactor and driving the purity of product D as shown in Figure 4.
Figure 4. During the esterification stage, there is a near 100% recovery of saturated reactant B refluxed into the reactor. Once the stripping stage commences and the pressure is ramped down (at 20,100 s) we see the contributing amount of reactant B removal that drives the purity of product D.
Based on the findings of the pocess simulation two appoaches for optimization were used:
- Manual optimization using a trial-and-error approach
- Dynamic optimization using the optimizer in gPROMS Process
For both approaches the constraints of the processe considered.
- Operating parameters:
- the reactor pressure should remain between 0.013 – 1.07 bar at all times
- the reactor temperature: < 220 °C
- reactor holdup total liquid: < 85 m3
- reactor holdup total liquid for stable stirrer operation: 45 m3
- End of batch parameters:
- Desired product purity in the product tank of + 97 wt.%
- Minimising the amount of intermediate C
- Ensure a minimum amount of product is generated, i.e. 57 tonnes
Based on the findings of the simulation a manual optimization of the process was carried out.
The upper limit of the temperature envelope for esterification is 220 °C before there are any adverse effects on product quality. The base case operates at a temperature far lower. Thus, performing simulations with incremental increases in the flowrate of high-pressure steam circulating the reactor jacket until the desired temperature is reached would naturally be a good option.
The manual optimization resulted in an improved product puity. However, the consumption of expensive high pressure steam and the duration of the process also increased resulting in a batch time that was not acceptable to the operator.