Gas separations are often required in chemical processes, e.g. air separation, ethylene production, etc. These are often challenging and costly processes because of the low temperature and high pressure needed if vapour-liquid phase separations are involved. This thesis focuses on hybrid membrane-distillation separations as an opportunity to develop more energy-efficient separation processes. In a typical ethylene plant, recovery, the separation and purification of the cracked product are economically important. The focus of this thesis is on the ‘C2splitter’ which separates the desired product, ethylene, from ethane. Cryogenic distillation, which is currently used to separate the binary ethylene-ethane mixture, is extremely expensive in terms of both capital and operating costs, especially because of refrigerated cooling requirements. Hybrid membrane-distillation processes are able to effectively separate low-boiling compounds and close-boiling mixtures and to reduce energy consumption, relative to cryogenic distillation. However, hybrid membrane-distillation processes present challenges for process modelling, design and operation. There are two major challenges associated with the modelling of hybrid processes for low temperature separations: i) the complex interaction between the process and the refrigeration system and ii) the large number of structural options, e.g. conventional column, membrane unit or hybrid membrane-distillation separation, where the distillation column can be integrated with the membrane unit to form a sequential, parallel, ‘top’or ‘bottom’ hybrid scheme. This thesis develops a systematic methodology to design, screen, evaluate and optimise various design alternatives. Schemes are evaluated with respect to energy consumption, i.e. power consumption of process and refrigeration compressors, or energy costs. In the methodology, process options are screened first for feasibility, based on numerous simulations and sensitivity analyses. Then, the feasible options are evaluated in terms of energy consumption and compared to the performance of a conventional distillation column. Finally, economically viable schemes are optimised to identify the most cost-effective heat-integrated structure and operating conditions. The methodology applies models for multi-feed and multi-product distillation columns, the membrane, compressor and refrigeration system; heat recovery opportunities are systematically captured and exploited. For the separation of relatively ideal mixtures, modified shortcut design methods, based on the Fenske-Underwood-Gilliland method are appropriate because they allow fast evaluation without needing detailed specification of column design parameters (i.e. number of stages, feed and side draw stage locations and reflux ratio). The modifications proposed by Suphanit (1999) for simple column design are extended to consider multi-feed and/or multi-product columns. The complex column designs based on the approximate calculations method are validated by comparison with more rigorous simulations using Aspen HYSYS. To design the hybrid system, a reliable and robust membrane model is also needed. To predict the performance of the module model, this work applies and modifies detailed membrane model (Shindo et al., 1985) and approximate method (Naylor and Backer, 1955) to avoid the need for initial estimates of permeate purities and to facilitate convergence. Heat integration opportunities are considered to reduce the energy consumption of the system, considering interactions within the separation process and with the refrigeration system. A matrix-based approach (Farrokhpanah, 2009) is modified to assess opportunities for heat integration. The modified heat recovery model eliminates the need to design the refrigeration cycle and uses a new simple, linear model that correlates the ideal (Carnot) and a more accurately predicted coefficient of performance. This work develops a framework for optimising important degrees of freedom in the hybrid separation system, e.g. permeate pressure, stage cut, side draw molar flow rate and purity, column feed and side draw locations. Heat recovery options between: i) column feeds and products; ii) the membrane feed and products and iii) the associated refrigeration system are considered. A deterministic and a stochastic optimisation algorithm are applied and compared for their efficiency of solving the resulting nonlinear optimisation problem. The new approach is demonstrated for the design and optimisation of heat-integrated sequential and parallel hybrid membrane-distillation flowsheets. Case study results show that hybrid scheme can reduce energy cost by 11%, compared to distillation, and that parallel schemes have around 8% lower energy costs than sequential hybrid schemes. These results suggest hybrid membrane-distillation processes may be competitive with distillation when applied for ethylene-ethane separations, but that further development of suitable membranes may still be needed.

The current work focused on defining the material science options to fabricate novel, high performing ethylene/ethane (C₂H₄/C₂H₆) separation carbon molecular sieve (CMS) dense film membranes. Three polymer precursors: Matrimid®, 6FDA-DAM and 6FDA:BPDA-DAM were used as precursors to the CMS membranes. CMS performances were tailored by way of tuning pyrolysis conditions such as the pyrolysis temperature, heating rate, pyrolysis atmosphere etc. The CMS dense film membranes showed attractive C₂H₄/C₂H₆ separation performance far exceeding the polymeric membrane performances. Semi-quantitative diffusion size pore distributions were constructed by studying the transport performance of a range of different penetrant gases as molecular sized probes of the CMS pore structure. This, in conjunction with separation performance data, provided critical insights into the structure-performance relationships of the CMS materials. The effects of testing conditions, i.e. the testing temperature, pressure and feed composition on C₂H₄/C₂H₆ separation performance of CMS dense films were also analyzed. These studies were useful not just in predicting the membrane behavior from a practical stand-point, but also in a fundamental understanding of the nature of CMS membrane separation. The study helped clarify why CMS membranes outperform polymeric membrane performance, as well as allowed comparison between CMS derived from different precursors and processing conditions. The effects on C₂H₄/C₂H₆ separation in the presence of binary gas mixture were also assessed to get a more realistic measure of the CMS performance resulting from competition and bulk flow effects. The current work thus establishes a framework for guiding research ultimately aimed at providing a convenient, potentially scalable hollow fiber membrane formation technology for C₂H₄/C₂H₆ separation

Olefin/paraffin separation is a large potential market for membrane applications. Carbon molecular sieve membranes (CMS) are promising for this application due to the intrinsically high separation performance and the viability for practical scale-up. Intrinsically high separation performance of CMS membranes for olefin/paraffin separations was demonstrated. The translation of intrinsic CMS transport properties into the hollow fiber configuration is considered in detail. Substructure collapse of asymmetric hollow fibers was found during Matrimidﾮ CMS hollow fiber formation. To overcome the permeance loss due to the increased separation layer thickness, 6FDA-DAM and 6FDA/BPDA-DAM polyimides with higher rigidity were employed as alternative precursors, and significant improvement has been achieved. Besides the macroscopic morphology control of asymmetric hollow fibers, the micro-structure was tuned by optimizing pyrolysis temperature protocol and pyrolysis atmosphere. In addition, unexpected physical aging was observed in CMS membranes, which is analogous to the aging phenomenon in glassy polymers. For performance evaluation, multiple "proof-of-concept" tests validated the viability of CMS membranes under realistic conditions. The scope of this work was expanded from binary ethylene/ethane and propylene/propane separations for the debottlenecking purpose to mixed carbon number hydrocarbon processing. CMS membranes were found to be olefins-selective over corresponding paraffins; moreover, CMS membranes are able to effectively fractionate the complex cracked gas stream in a preferable way. Reconfiguration of the hydrocarbon processing in ethylene plants is possible based on the unique CMS membranes.