The objective of this thesis was the development of an extraction process for the removal of multiple aromatics from several petrochemical streams by means of an ionic liquid. Due to environmental legislation, the demand of ‘clean’ fuels is increasing and most likely will increase even more towards fuels with almost zero content of certain aromatics, e.g. benzene and toluene. In particular, the concentration of benzene has to be reduced to = 0.1 wt-% in carburant fuels. Furthermore, the sulphur content of gasoline and diesel fuel has to be decreased to <10 ppm. However, the separation of aromatic and aliphatic hydrocarbons is complicated due to overlapping boiling points and azeotrope formation. Therefore, the conventional processes for this type of separation are extraction or extractive distillation with polar, organic solvents such as NFM (Uhde Morphylane® process), Sulfolane® (UOP/Shell Sulfolane® Process), NMP, ethylene glycols etc. Another class of solvents, which are considered promising to replace organic solvents in industrial processes, are ionic liquids and the use of them as solvents for different processes on industrial and pilot plant scale gains more and more interest. Moreover, for feeds with low aromatic contents, ionic liquids can be superior to conventional extraction solvents. Additionally, process simulations showed that an ionic liquid based process can be economical more beneficial than conventional processes. However, an industrial extraction process is not reported yet, since, ionic liquids reported in the literature are mostly not suitable for processes on larger scale. Therefore, the aim of this work is the development of an extraction process based on an improved ionic liquid for the separation of aromatic hydrocarbons from various petrochemical streams. Ionic liquids are liquid salts consisting of large, mostly organic, cations and a great variety of anions. Their positive properties are a wide liquid temperature range (~300 K), low vapour pressure and the ability of tailoring. The extremely low vapour pressure allows for easy recovery of enclosed solutes while the tailorability enables the design of an ionic liquid as extraction solvent for a specific separation problem. The first step to obtain a suitable ionic liquid for the aimed process is the detailed screening of possible candidates. In this context, the comprehension of conventional solvents is indispensable and the application of the obtained knowledge to new systems a prerequisite. For this purpose the screening of ionic liquids for the investigated model feeds FCC gasoline, Reformate and Diesel in comparison with conventional solvents has been carried out by means of the quantum chemical based tool COSMO-RS and subsequently experimentally validated. It was shown that by means of the, in this work, developed COSMO-RS ¿-profile screening method it is suitable to identify promising candidates by separate cation and anion screening. Ionic liquids based on the cations imidazolium, pyridinium and pyrrolidinium and on the anions [SCN]-, [DCA]-, [TCM]- and [TCB]- have been identified as most promising candidates. This could be experimentally validated for a toluene/heptane model feed. Moreover, it was shown that the optimal alkyl chain length on the above mentioned cyclic cations is a butyl chain. Also for more complex, multiple aromatic feeds it was shown that the same ionic liquids as described above are suitable candidates. In order to determine a suitable extraction solvent, ionic liquids have been evaluated by the COSMO-RS ¿-profile screening for the aromatic components present in FCC gasoline, reformate and diesel. The experimental validation has been carried out by means of model feeds that comprise only the most important aromatic, olefinic and paraffinic components from the aforementioned feeds. It was shown that the extraction is in the order benzene > toluene > p-xylene > cumene > 1-hexene > n-hexane > n-heptane and for higher aromatics and cyclic aliphatics 9,10-dihydrophenanthrene > naphthalene > tetralin > decalin. Based on these screening results the ionic liquid [3-Mebupy][DCA] has been chosen for further evaluation due to the high capacity (Dbenzene,[3-Mebupy][DCA] = 0.60 [g/g]) and reasonable selectivity (¿[3-Mebupy][DCA],benzene/n-hexane = 35.3). Subsequent, the two ionic liquids [3-Mebupy][TCM] and [3-Mebupy][TCB] became available, which exhibit an even higher capacity and comparable or slightly lower selectivity (Dbenzene,[3-Mebupy][TCM] = 0.70 [g/g] and Dbenzene,[3-Mebupy][TCB] = 0.74 [g/g]; ¿[3-Mebupy][TCM],benzene/n-hexane = 34.8 and ¿[3-Mebupy][TCB],benzene/n-hexane = 27). Additionally, as petrochemical streams contain numerous components, including heterocycles, the affinity of heteroatoms towards ionic liquids has been studied in comparison to mono-aromatics. Therefore two model feeds containing sulphur aromatic components or nitrogen aromatics and toluene, tetralin and n-heptane have been investigated for the extraction with the same ionic liquids that showed to be promising for feeds containing only aromatic and aliphatic hydrocarbons. It was found that the sulphur and nitrogen containing hetero aromatics thiophene and dibenzothiophene and pyrrol, indole and carbazole are significantly better extracted than aromatic hydrocarbons. In one extraction step up to 80 % of the thiophene and up to 90 % of the dibenzothiophene can be removed while > 99 % of the nitrogen containing aromatics has been removed. Furthermore, the results for the model feeds are compared to real feed experiments in order to investigate the influence of a real petrochemical stream mixture compared to a model feed with a limited number of components. In all cases it was observed that the removal of the aromatic components from the real feed was less, due to competing influences of other components, but still promising. For the real feed experiments the ionic liquid [3-Mebupy][DCA] has been chosen. Whit a suitable candidate defined, the subsequent step is the development of an extraction process based on this ionic liquid. Therefore, a process design for the FCC gasoline model feed based on [3-Mebupy][DCA] comprising the main extraction column with the additional separation and solvent recovery units has been developed with Aspen plus together with an economical feasibility study of the process. The process comprises the main extraction column, a back extraction column for recovery of the ionic liquid that is withdrawn in the raffinate phase by entrainment, an extractive stripper in order to remove the co-extracted aliphatic components from the extract phase and a flash evaporator for separation of the aromatic product from the extraction solvent. Since [3-Mebupy][DCA] is hydrophilic, in contrary to the compounds present in the raffinate phase, the ionic liquid in the raffinate phase can be easily back-extracted by means of water. The results for investment and operational costs for the ionic liquid based process have been compared to a process using sulfolane as extraction solvent, since this is the most conventional solvent for aromatics extraction. It is shown that the investment costs for the ionic liquid based process are up to 42 % lower than for sulfolane and the annual costs for the [3-Mebupy][DCA] process are only 17.8 M€ compared to 32.6 M€ for a sulfolane process. This is due to the higher capacity of the ionic liquid which results in smaller process streams and therewith smaller equipment. The process design is based on the ternary diagrams that can be derived from the components present in the FCC gasoline and reformate model feeds and [3-Mebupy][DCA]. The ternary data have been determined experimentally and correlated with the NRTL model. The data regression for the two model feeds is in good agreement with the experimental data and the RMSD values are in general <0.0324. Additionally, the ternary diagrams for toluene/n-heptane with the three ionic liquids [BMIM][DCA], [BMIM][SCN] and [3-Mebupy][DCA] have been determined. The RMSD-values in this case are <0.0076. Furthermore, an experimental study on the scale-up of the FCC gasoline model feed and a real feed (LCCS) to a rotating disc contactor (RDC) pilot plant with the solvent [3-Mebupy][DCA] provided insight in mass transport and hydrodynamic effects, which is valuable information for an industrial process. Analogous to the LLE-measurements, it was shown that the extraction performance of the RDC column is higher for the model feed than for the real feed. From the FCC gasoline model feed 89 % benzene and 75 % toluene removal was observed while for the real feed 81 % benzene and 71 % toluene could be removed, respectively, with a solvent-to-feed ratio S/F = 4 and 800 rpm. This is due to competing effects of the multiple components in the real feed, which also hampers the mass transfer. Therefore, the mass transfer performance of RDC-column is also higher for the model feed than for the real feed. Besides, comparable hydrodynamic behaviour of the model and real feed has been observed. Since the densities and viscosities of both feeds are comparable, this explains the observed similar data in terms of Sauter means size diameter, hold-up and operational window. The conclusions that can be drawn from this work confirm that ionic liquids are potential solvents for the extraction of aromatic hydrocarbons as well as hetero aromatics containing sulphur and nitrogen. It has been reported that these components can be removed selectively from components as, e.g. olefins, aliphatics and cyclic aliphatics. Moreover, the results obtained with model feeds could be validated by means of real feed experiments on lab scale as well as pilot plant scale. Furthermore, from the conceptual process design based on [3-Mebupy][DCA] it is evident that an ionic liquid based extraction process can be energetically, and thus economically, more favourable than a sulfolane process. However, for the implementation of an ionic liquid extraction process on industrial scale further research has to be carried out in particular with view of the ionic liquid recovery and aromatics removal form the extract phase.
|Qualification||Doctor of Philosophy|
|Award date||23 Jun 2010|
|Place of Publication||Eindhoven|
|Publication status||Published - 2010|