Minimum Liquid Discharge (MLD) and Zero Liquid Discharge (ZLD) schemes have been widely proposed in the recent scientific literature not only as a possible solution to brine disposal but also as a non-conventional sustainable source of raw materials. Nevertheless, very few works have pushed the idea towards a real demonstration activity, and this somehow limits the reliability that such schemes have with respect to the real implementation potential at the industrial scale. In this work, for the first time in the literature, an integrated treatment chain for the sustainable production of freshwater and minerals has been demonstrated at a pre-industrial scale, in the island of Lampedusa (Italy). The treatment chain included a Nanofiltration (NF) step to separate monovalent and bivalent ions, followed by a Multi-Effect Distillation (MED) unit, powered by waste heat from a Thermal Power plant, generating high-quality water (<30 μS/cm) and an ultra-concentrated brine. The latter was treated in Evaporation Ponds (EPs) to generate high purity NaCl (>99 %). On the other side, the NF retentate was treated to selectively recover magnesium and calcium hydroxides (Mg(OH)₂ purity up to 98 %) in a novel Multiple Feed-Plug Flow Reactor (MF-PFR). The resulting brine fed an ElectroDialysis with Bipolar Membranes unit (EDBM), generating in-situ alkaline and acidic solutions: chemicals needed for internal usage in the plant. All units were successfully tested, reaching satisfactory performance indicators. Furthermore, the stability of each unit during the daily operational run was assessed and successfully achieved, demonstrating not only the technical feasibility of the proposed demo plant, but also the feasibility of MLD as a sustainable alternative for minerals recovery.

This work presents a joint experimental and simulation campaign aimed at characterizing two nanofiltration membranes (TS80 and NF270) in the presence of a multi-ionic water solution simulating the spent regenerant of cationic ion exchange resins employed for water softening. We identified the membrane parameters, which allowed for predicting the performances through the Donnan Steric Pore Model with Dielectric Exclusion. A good agreement between model and experimental trends of rejection as a function of the applied pressure was observed (error < 15%). The analysis of trans-membrane fluxes and exclusion coefficients showed that dielectric exclusion was the crucial mechanism for the ionic partition. In fact, the lower pore dielectric constant found for TS80 justified the higher rejections to divalent cations with respect to NF270. Moreover, negative charge densities were found for both membranes, because of the high concentration of chloride ions in the feed, which likely adsorbed onto the membrane. However, it was observed that the experimental rejections did not change significantly with the feed pH. This result, in line with the minor role of the Donnan exclusion resulting from the model, suggested that the membrane performances were not much affected by the charge density at high feed ionic strengths (~1 M).