Surrogates for Liquid-Liquid Extraction

Abstract

Fuel surrogates are mixtures that mimic the properties of real fuels with only a small number of components, simplifying the calculation and simulation of fuel-related processes. This work extends a previously published surrogate optimization algorithm toward the generation of fuel surrogates with a focus on liquid-liquid extraction characteristics. For this purpose, experimental liquid-liquid equilibrium data from batch extraction experiments are incorporated into the calculation procedure as an additional constraint. The use of the method is demonstrated by optimizing a surrogate for the catalytic reformate. Application of the surrogate to an extraction process and comparison with experimental data demonstrate that the resulting surrogate accurately depicts the properties of the real mixture with regard to liquid-liquid extraction performance. This demonstrates that the use of such surrogates is of particular interest for mixtures used as extracting agents for biofuels.

Figure 1

ASTM D86 and TBP curves of reformate, TBP curve calculated according to Daubert from ASTM D86 data provided by OMV Downstream GmbH.

Figure 2

Sixth order polynomial fit for the TBP curve.

Figure 3

LLE flash calculation flowsheet, adapted from Gmehling et al. Page 274. 2019. Copyright 2019 Wiley-VCH GmbH. Reproduced with permission.

Figure 4

TBP boiling curve of the resulting surrogate without the LLE criterion.

Figure 5

Ternary diagram of system 1-propanol–water–reformate: comparison between experimental LLE data and simulated results with conventional surrogate without LLE criterion, simulation conducted in AspenPlus V11 with NRTL-UNIFAC21.

Figure 6

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data and simulated results with LLE-optimized surrogate for the calibration set, simulation conducted in AspenPlus V11 with UNIFAC-LL.

Figure 7

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data and simulated results with LLE-optimized surrogate for the calibration set, simulation conducted in AspenPlus V11 with NRTL-UNIFAC21.

Figure 8

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data and simulated results with LLE-optimized surrogate for the calibration set, simulation conducted in AspenPlus V11 with NRTL-UNIFACAspen.

Figure 9

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data and simulated results with LLE-optimized surrogate for the calibration set, simulation conducted in AspenPlus V11 with NRTL-UNIFACDMD21.

Figure 10

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data and simulated results with LLE-optimized surrogate for the calibration set, simulation conducted in AspenPlus V11 with NRTL-UNIFACLL.

Figure 11

Ternary diagram of system 1-propanol–water–reformate, comparison between experimental LLE data of the validation set with simulated results with and without LLE-optimized surrogate, simulation conducted in AspenPlus V11.

Figure 12

RMSRE for 1-propanol and water content in the extract for the validation set, comparison between LLE-optimized and conventional surrogates.

Figure 13

RMSRE for 1-propanol and water content in the raffinate for the validation set, comparison between LLE-optimized and conventional surrogates.

Figure 14

Stage construction for a two-stage cross-flow extraction process based on experimental LLE data; conjugation line for the construction of additional tie lines not shown in the diagram.

Figure 15

Flowsheet of the two-stage cross-flow process simulated in AspenPlus V11, S1: solvent stage 1, M1: mixing point 1, SETT1: settler stage 1, RAFF1: raffinate stage 1, EXT1: extract stage 1, S2: solvent stage 2, M2: mixing point 2, SETT2: settler stage 2, RAFF2: raffinate stage 2, EXT2: extract stage 2, and EXTtotal: combination of extracts from stages 1 and 2.