467819 A Systems Approach for the Development of Osn Membrane Cascades

Monday, November 14, 2016: 2:43 PM
Monterey I (Hotel Nikko San Francisco)
Vincentius Surya Kurnia Adi1, Marcus Cook2, Ludmila G. Peeva2, Andrew G. Livingston2 and Benoit Chachuat1, (1)Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London, United Kingdom, (2)Department of Chemical Engineering, Imperial College London, London, United Kingdom

The processes involved in separating the components of chemical mixtures into pure or purer forms, such as distillation, account for 10-15% of the world’s energy consumption. Methods to purify chemicals that are more energy efficient could, if applied to the U.S. petroleum, chemical and paper manufacturing sectors alone, save 100 million tonnes of CO2 emissions and US$4 billion in energy costs annually [1]. Membrane-based separation methods, or other non-thermal ones, can be an order of magnitude more energy efficient than heat-driven separations that use distillation, but these methods are underdeveloped [2].

Given the recent advances in organic solvent nanofiltration (OSN) membranes [3], the focus of this presentation is on OSN membrane cascades for separating organic/hydrocarbon mixtures. As a case study, we investigate the design of a lab-scale membrane cascade for purifying a binary mixture of heptane and hexadecane, with a 75-25 weight fraction. The main objective is to maximize heptane purity in the permeate stream by using a maximum of 10 identical membrane units, each having an effective area of 14 cm.

In a first step, we use a superstructure optimization approach based on mixed-integer nonlinear programming (MINLP) to determine the most promising cascade configurations, in terms of the number of stages in the cascade, the number of membrane units in each stage, and the interconnections thereof, along with certain operating parameters [4,5]. Constraints are imposed for limiting the transmembrane pressure drop and the stage-cut in each unit. The permeate and retentate flux and composition for each membrane unit are predicted by a classical solution-diffusion model, where the permeability coefficients for heptane and hexadecane are estimated based on dedicated experiments for the membranes at hand, and the activity coefficients of heptane and hexadecane are approximated using the UNIFAC method for different mixture compositions. An optimal 3-stage cascade obtained with this approach is shown on the figure below, with GAMS 24.5.6 and the global optimizer BARON 15.9 used to solve the MINLP problem.

In a second step, we implement the optimised OSN membrane cascades experimentally in order to verify the predictions. For the 3-stage cascade shown on the figure, the mismatch between the model predictions and the measurements remains small, within a few percents, thereby validating the model-based optimization approach.


1. U.S. Department of Energy, Advanced Manufacturing Office, "Bandwidth Study on Energy Use and Potential Energy Saving Opportunities in U.S. Petroleum Refining", June 2015.

2. D.S. Sholl, R.P. Lively, "Seven chemical separations to change the world," Nature 532:435-43, 28 April 2016. U.S. Petroleum Refining (US Dept. Energy, 2015).

3. P. Marchetti, M.F.J. Solomon, G. Szekely, A.G. Livingston, "Molecular separation with organic solvent nanofiltration: A critical review," Chemical Reviews 114:10735-10806, 2014.

4. R.H. Qi, M.A. Henson, "Membrane system design for multicomponent gas mixtures via mixed-integer nonlinear programming," Computers & Chemical Engineering 24:2719-2737, 2000.

5. V.S.K. Adi, M. Cook, L.G. Peeva, A.G. Livingston, B. Chachuat, "Optimization of OSN membrane cascades for separating organic mixtures," Proc. 26th European Symposium on Computer Aided Process Engineering (ESCAPE), 12-15 June 2016, Portorož, Slovenia.


Extended Abstract: File Not Uploaded
See more of this Session: Process Intensification
See more of this Group/Topical: Computing and Systems Technology Division