Assessment and potential of membrane cascades for organic solvent nanofiltration of hydroformylation media through a graphical representation composed of performance maps

Abstract

et al.. Assessment and potential of membrane cascades for organic solvent nanofiltration of hydroformylation media through a graphical representation composed of performance maps. Chemical Engineering Science, Elsevier, 2018, 183, pp. AbstractThis paper aims at proposing a graphical representation composed of several performance maps to help to answer to some current questions that can puzzle membrane end-users facing the arrangement of membranes in cascade in order to better master separation of complex media. Indeed, different compromises can be highlighted according to realistic goals for the separation such as quality and recovery yield of each fraction/component, energy consumption and required membrane area. This representation needed first the systematic simulations of cascades of pre-selected configurations. These last ones were chosen thanks to the target application field, namely the organic solvent nanofiltration of a final synthesis media of hydroformylation that is a homogeneous catalysed reaction. We voluntary assumed the a priori limitation of the number of stages to 5, anticipating that more complex cascades will probably be too expensive (both operating and capital costs). The graphical representation by itself is based on sets of six 2D-maps. Each map highlights relationships selected in an appropriate way between two of the six selected criteria: extraction/recovery, retentate/permeate quality/purity, membrane filtering area and overall energy consumption. For sake of illustration, the separation of 2 components C and A was considered. C/A has been chosen in a 1/1000 molar ratio, where C corresponds to the less retained component of the catalytic system that must be recovered in the retentate and A corresponds to the less transmitted product to extract in the permeate. In realistic nanofiltration conditions achieved in toluene, the rejections were experimentally determined on the initial media to filter. C has a high rejection (88%) whereas A has a low one (30%). The simulations of cascades were established using these constant values for rejection and the experimental permeate flux. For sake of an illustration of the use of the graphical representation, a case study was finally discussed regarding a given target of recovery for the two desired components, namely at least 99% of C recovery and better than 70% of A extraction. A complementary multi-criteria analysis was added aiming at facilitating the decision-making.2

Coupling Rhodium‐Catalyzed Hydroformylation of 10‐Undecenitrile with Organic Solvent Nanofiltration: Toluene Solution versus Solvent‐Free Processes

“…For all filtrations, the permeate flux decreased to some extent depending on the ligand, but remained stable up to VRR=2. According to our knowledge, besides TMP and the cross‐flow velocity, the flux was mainly governed by the affinity between ( i ) the membrane and the solvent and ( ii ) the overall solute concentration (≈1 M), each compound being able to have a different impact . The same flux decrease was observed with mixtures containing Biphephos and A4N3 ligands (J OSN /J 0 =0.70).…”

“…Changing 10‐undecenitrile for mixtures of the reaction products and by‐products would have a little impact on flux, as evidenced from experience . To overcome the lack of sufficient amount of all components, preventing from an experimental determination, we have intensively studied the Biphephos/Aldehyde separation on a theoretical point of view . Assuming constant rejection of 88 % for Biphephos and 30 % for the aldehyde, or taking into account rejection variations with the concentration, resulted in similar results in the 1–10 VRR range.…”

“… Overall recoveries of the Biphephos ligand in the retentate and the target Aldehyde in the permeate at different VRR based on OSN performances with the Biphephos system in toluene and Pervap 4060 membrane at 10 bar adapted from simulation …”

“…First, if OSN is achieved as a single step process at TMP=10 bar with a continuous extraction of the permeate up to VRR=5 (Figure ), then 82 % of the catalytic system could be recycled after one reaction‐OSN cycle and only 80 % of the feed volume (permeate) would enter the first distillation column, knowing that toluene would be 82 % of this permeate volume. We estimated that the OSN OPEX would be 0.5 kWh ⋅ m −3 . In that case, the distillation would be achieved on 66 % of the initial feed volume and the energy saving during distillation would be high.…”

“…One has to keep in mind that, generally, this is more an efficient enrichment step than a very fine purification process such as chromatography. An increase in performances can be envisioned by complexifying the OSN scheme with a membrane cascade (see Discussion) . Moreover, the membrane ability to separate the target product(s) from the unreacted substrate(s) and other by‐products have to be evaluated.…”