Distillation is the crucial separation method of liquid mixtures. However, its energy demand, currently provided by burning fossil fuels, is very high. Therefore, it is very important to optimise the design and the operation of distillation technologies. Optimisation is often performed by evolutionary (e.g. genetic) algorithms coupled to flowsheet simulators. However, such algorithms require many simulations, making the optimisation very time-consuming.

Surrogate models are simpler, fast-to-evaluate models fitted to simulation results. During optimisation, the flowsheet simulator is replaced by the surrogate models, making the procedure much faster.

Tools to be used: CHEMCAD (flowsheet simulator), ALAMO (machine learning-based tool to fit surrogate models), Python and VBA (in Excel) to enable automatic communication between CHEMCAD and ALAMO for adaptive sampling (optional). Knowledge of Python and VBA is an advantage, but this part can be omitted.

Disclaimer: This is an assignment of the Department of Building Services and Process Engineering. Contact Dr. László Hégely (hegely.laszlo@gpk.bme.hu) for more details.