Electrodialysis is used in food and bio-based industries for selective separation and recovery of target ions and neutral molecules. Anion exchange membranes (AEM) are prone to fouling due to their complex interactions with organics. Currently, research has focussed mainly on fouling tests using model organic foulants. In this study, three different complex industrial feed streams: corn steep liquor (CSL), citric acid by-product (CA), and cheese whey (CW), were used to foul two commercial AEM types: Fujifilm type-10 (homogeneous) and RALEX (heterogeneous). A 1.5 - 2.3 fold higher membrane resistance after fouling was a result of a difference in membrane structure, ion exchange capacity, water uptake, and thickness (depending on the feed used). Liquid Chromatography - Organic Carbon Detection (LC-OCD) analysis of feed revealed highly hydrophilic (CDOC) fractions in CSL (85%), CA (87%), and CW (100%). This analysis was extended to foulant desorption solutions (NaCl (35 g/L) and EtOH-water-H2SO4 mixtures), which revealed that > 90 % of humic substances (HS) and building blocks (BB) desorbed when using NaCl (35 g/L), following a (concentration wise) desorption order: HS > BB > LMWN > BP > LMWA. For the EtOH-water-H2SO4 solution, low molecular weight neutrals and acids (LMW(A + N)) were desorbed majorly (>60%) and followed a desorption order: LMWN > LMWA > BB > BP > HS. Desorption due to the ion-exchange mechanism was dominant for NaCl (35 g/L) solution, while foulantdesorbant polar interactions were prevalent for the EtOH-water-H2SO4 solution. Upon foulant desorption using NaCl (35 g/L) solution, RALEX AEM owing to the presence of nanovoids in them, showed a 1.5 - 4 fold higher concentration difference vs. FF10 AEM, while this difference was not dominantly observed when EtOHwater-H2SO4 mixture was used. This study revealed that LMW fractions could possibly enter into the membrane matrix whereas HS and BB fractions owing to their high molecular weight, majorly adsorbed on surfaces.

When electrodialysis (ED) is applied to process streams that are known to foul and scale the system, it can be operated with an oscillating electric field. This procedure is known as pulsed electric fields (PEF) and has been proven to reduce the fouling susceptibility of the system. In literature, the suppression of fouling is attributed to a combination of three effects: the reduction of concentration polarisation, the promotion of electroconvection and the suppression of water splitting. However, these effects have yet to be substantiated quantitatively. Here we study how much of the fouling suppression during PEF can be attributed to the reduction of concentration polarisation. A Nernst–Planck and Kedem–Katchalsky modelling approach is adopted to simulate the evolution of concentration profiles of a ternary solution of sodium, chloride and dodecyl sulfate (DS) during PEF. The effect of the pulse parameters on the boundary layer concentration of sodium dodecyl sulfate (SDS) is studied along with an evaluation of the current efficiency and energy consumption. Our results illustrate the counterproductivity of low-frequency pulses and the trade-off between the current efficiency and fouling suppression but fail to explain the improvement when compared to the constant current operation. Fouling layer relaxation is put forth to complement to previously mentioned fouling suppression effects and is relevant for future research.

Fouling of the ion-exchange membranes by colloidal substances present in bio-based process streams is an important hurdle for electrodialysis. The development of a model is challenged by the limited availability of experimental data and the complexity of the underlying physics. This research addresses this challenge by combining a mechanistic description of the transport processes with a machine learning model to describe the complex phenomena of colloidal aggregation and attachment to the surface of ion-exchange membranes. After validation with fouling experiments using acrylamide as colloidal foulant, it was found that this hybrid model improves the predictive power of the model while reducing the need for experimental data. An analysis of both mechanistic and machine learning models showed that the attachment probability of anion polyacrylamide (APAM) is influenced by the current density and the size of the fouling layer in a non-linear manner. An increase in current density leads to an increase in the attachment probability while the opposite holds for the size of the fouling layer. This research shows that machine learning can complement mechanistic models where fundamental knowledge is lacking or computational resources are limiting. The combination maintains the interpretability and generalisability of mechanistic models while harnessing the accuracy of machine learning.

Cooling of thermal power stations requires large amounts of surface water and contributes to the increasing pressure on water resources. Water use efficiency of recirculating cooling towers (CT) is often kept low to prevent scaling. Partial desalination of CT feed water with membrane capacitive deionization (MDCI) can improve water quality but also results in additional water loss. A response surface methodology is presented in which optimal process conditions of the MCDI-CT system are determined in view of water use efficiency and cost. Maximal water use efficiency at minimal cost is found for high adsorption current (2.5 A) and short adsorption time (900 s). Estimated cost for MCDI to realize maximal MCDI-CT water use efficiency is relatively high (2.0–3.1 € m−3evap), which limits applicability to plants facing high intake water costs or water uptake limitations. MCDI-CT pilot tests show that water use efficiency strongly depends on CT operational pH. To allow comparison among pilot test runs, simulation software is used to recalculate CaCO3 scaling and acid dosage for equal operational pH. Comparison at equal pH shows that MCDI technology allows a clear reduction of CT water consumption (74%–80%) and acid dosage (63%–80%) at pH 8.5.

The attachment of colloids to the ion-exchange membranes in electrodialysis is an important hurdle when processing bio-based process streams. Previous research showed that fouling strongly depends on the crossflow velocity, the current and the salt concentration of the medium. Predicting the influence of these variables on the fouling rate is challenging due to the complex physics at play and optimising the process conditions to reduce fouling remains a challenge. The objective of this study is the development of a model to predict the dynamic behaviour of electrodialysis fouling under varying process settings to facilitate this optimisation. A neural differential equation is fit to experimental data of an electrodialysis pilot undergoing humic acid fouling. We show that this model can predict the fouling rate even when using a limited set of experimental data. The robustness of the model is demonstrated by a simulation study and a sensitivity analysis indicates that the crossflow velocity is the most important variable influencing the fouling rate (approximate to 40%). Both the effect of the current (approximate to 20%), the salt concentration (approximate to 13%) and their interaction effects are considerable. With the model, the evolution of the stack resistance as a result of membrane fouling can be simulated, facilitating process control or decision-making.

This data paper aims to provide data on the effect of the process settings on the fouling of an electrodialysis pilot installation treating a sodium chloride solution (0.1 M and 0.2 M) in the presence of humic acid (1 g/L). This data was used by "Colloidal fouling in electrodialysis: a neural differential equations model" [1] to construct a predictive model and provides interpretive insights into this dataset. 22 electrodialysis fouling experiments were performed where the electrical resistance over the electrodialysis stack was monitored while varying the crossflow velocity (2.0 cm/s - 3.5 cm/s) in the compartments, the current applied (1.41 A - 1.91 A) to the stack and the salt concentration in the incoming stream. The active cycle was maintained for a maximum of 1.5 h after which the polarity was reversed to remove the fouling layer. Additional data is gathered such as the temperature, pH, flow rate, conductivity, pressure in the different compartments of the electrodialysis stack. The data is processed to remove the effect of temperature fluctuations and some filtering is performed. To maximise the reuse potential of this dataset, both raw and processed data are provided along with a detailed description of the pilot installation and sensor locations. The data generated can be useful for researchers and industry working on electrodialysis fouling and the modelling thereof. The availability of conductivity and pH in all compartments is useful to investigate secondary effects of humic acid fouling such as the eventual decrease in membrane permselectivity or water splitting effects introduced by the fouling layer.

Industrial processes such as e.g. oil, fat and resin production and saponification generate large volumes of aqueous NaCl-glycerol containing streams. With typical glycerol concentrations ranging from 5 to 15 % (m/m) and a high salt content, these streams pose a challenge to further processing, both to end-of-pipe treatment as to possible reuse scenarios for organic components and/or salts. Electrodialysis is evaluated as an alternative technology to remove salt from these streams. Salt, glycerol and water transport is studied experimentally through desalination of synthetic solutions for 7 commercial ion exchange membrane pairs, Fumatech (FAB-FKB, FAM-FKM, FAS-FKS), Neosepta (AMX-CMX), PCA (MVA-MVK, SA-MV) and Ralex (AMH-CMH). A phenomenological model is applied i.e. model coefficients are derived from test data through regression analysis to evaluate transport phenomena. Salt transport is found to be related to electromigration only and the NaCl transference number is determined to be highest in AMH-CMH (0.96 +/- 0.04) and FKS-FAS (1.01 +/- 0.03). Limiting current density is highest for FKS-FAS and lowest for FAM-FKM. Water transport is attributed to electroosmosis and osmosis. Water transference numbers found are similar to those in literature (t(w) similar to 7-10). The value of the osmotic coefficients (P-W) differs strongly among the different membrane pairs and is highest for MV-SA Q380 (0.99 x 10(-5) ms(-1)). Glycerol transport is attributed to diffusion and water flux induced cotransport. The diffusion of glycerol differs significantly among all membrane pairs (range 2.06 x 10(-8) ms(-1) FABFKB to 12.4 x 10(-8) ms(-1) MV-SA). Reflection coefficients differ significantly among membrane pairs but fall within a relative narrow range (0.36-0.75 ms(-1)). Numerical simulation of batch desalination with the derived transport model identifies current density as the largest contribution to the overall glycerol transport (38% - 64%) followed by osmotic co-transport (16%-41%) and glycerol diffusion (9%-28%). A low glycerol/ NaCl flux is desirable, this occurs when both initial glycerol/NaCl and NaCl concentrations are low while current density is high.

Water management is becoming increasingly challenging and several technologies, including membrane distillation (MD) are emerging. This technology is less affected by salinity compared to reverse osmosis and is able to treat brines up to saturation. The focus of MD research recently shifted from seawater desalination to industrial applications out of the scope of reverse osmosis. In many of these applications, surfactants or oil traces are present in the feed stream, lowering the surface tension and increasing the risk for membrane wetting. In this study, the technological boundaries of MD in the presence of surfactants are investigated using surface tension, contact angle and liquid entry pressure measurements together with lab-scale MD experiments to predict the wetting resistance of different membranes. Synthetic NaCl solutions mixed with sodium dodecyl sulfate (SDS) were used as feed solution. The limiting surfactant concentration was found to be dependent on the surface chemistry of the membrane, and increased with increasing hydrophobicity and oleophobicity. Additionally, a hexadecane/SDS emulsion was prepared with a composition simulating produced water, a waste stream in the oil and gas sector. When hexadecane is present in the emulsion, oleophobic membranes are able to resist wetting, whereas polytetrafluoretheen (PTFE) is gradually wetted by the feed liquid.

While many models exist in the literature for description of lab-scale direct contact membrane distillation (DCMD) performance, only a handful of modelling attempts at full-scale have been described. In this article a method is presented for the simulation of the flux and energy efficiency of a full-scale, counter-currently operated, spiral-wound DCMD module. The model is based on a previously calibrated lab-scale model. The geometry of the full-scale, spiral-wound module is discretized into small sections and the lab-scale model is applied in each section. It was found that the membrane used in this module compacts significantly under the operational conditions. This effect cannot be neglected, therefore the model had to be extended to account for it. It is noteworthy that, apart from this extension, no additional model calibration at full-scale was needed. Given its validity at full-scale the model can be safely used for extended scenario analysis with regard to the optimization of the module design. The developed model and method are powerful tools to decide on system design and operation.

Key performance indicators for characterization of nanofiltration performance are well developed, similar key performance indicators for electrodialysis reversal are however underdeveloped. Under the E4Water project Dow Benelux BV and Evides Industriewater BV operate a pilot facility to compare both technologies for their application to mildly desalinate a variety of brackish water streams. Normalized pressure drop, normalized current efficiency and normalized membrane resistance proved to be a useful tool to interpret process performance and to initiate a cleaning procedure if required. The availability of these normalized key performance indicators enables optimization and process monitoring and control of electrodialysis reversal independent of the continuously changing conditions of the feed water.

For chemical industries, fresh water availability is a pre-requisite for sustainable operation. However, in many delta areas around the world, fresh water is scarce. Therefore, the E4 Water project (www.e4water.eu) comprises a case study at the Dow site in Terneuzen, The Netherlands, which is designed to develop commercial applications for mild desalination of brackish raw water streams from various origins to enable reuse in industry or agriculture. This study describes an effective two-stage work process, which was used to narrow down a broad spectrum of desalination technologies to a selection of the most promising techniques for a demonstration pilot at 2-4 m³/hour. Through literature study, laboratory experiments and multi-criteria analysis, nanofiltration and electrodialysis reversal were selected, both having the potential to attain the objectives of E4Water at full scale.

Ion-exchange tap water demineralization for process water preparation results in a saline regeneration wastewater (20-100 mS cm(-1)) that is increasingly problematic in view of discharge. A coupled nanofiltration-membrane distillation (NF-MD) process is evaluated for the recovery of water and sodium chloride from this wastewater. NF-MD treatment of mixed regeneration wastewater is compared to NF-MD treatment of separate anion- and cation-regenerate fractions. NF on mixed regeneration wastewater results in a higher flux (30 L m(-2) h(-1) at 7 bar) compared to NF on the separate fractions (6-9 L m(-2) h(-1) at 30 bar). NF permeate recovery is strongly limited by scaling (50% for separate and 60% for mixed, respectively). Physical signs of scaling were found during MD treatment of the NF permeates but did not result in flux decline for mixed regeneration wastewater. Final salt composition is expected to qualify as a road de-icing salt. NF-MD is an economically viable alternative compared to external disposal of wastewater for larger-scale installations (1.4 versus 2.5 euro m(-3) produced demineralized water for a 10 m3 regenerate per day plant). The cost benefits of water re-use and salt recuperation are small when compared to total treatment costs for mixed regenerate wastewater.

Tank truck cleaning (TTC) activities generate highly complex wastewater. In a previous study, we found that a significant ecotoxic effect was still present in biologically treated TTC wastewater. The aim of the present study was therefore to investigate the removal of acute toxicity from TTC wastewater by a sequence of technologies routinely applied for industrial wastewater. Acute toxicity was assayed with the widely applied and standardized Vibrio fischeri bioluminescence inhibition test. During a 5-month period, raw wastewater was grab-sampled from a full-scale TTC company and treated by the different unit operations on a laboratory scale. Chemical pretreatment of the wastewater by coagulation with FeCl3 removed approx. 38% of the influent chemical oxygen demand (COD) and reduced the bioluminescence inhibition by 8%. Biological treatment with activated sludge subsequently removed another 77% of the remaining COD. This treatment step also reduced the bioluminescence inhibition but the removal efficiency varied strongly from 5 to 92% for the different samples. Powdered activated carbon almost completely removed the remaining COD and inhibition in all samples. The results suggest that conventional technologies did not suffice for complete removal of toxicity from TTC wastewater, and that advanced wastewater treatment technologies such as activated carbon are required for a satisfactory detoxification.

Girdling can be used as a valuable research tool to improve our understanding of the tight coupling between water (xylem) and sugar (phloem) transport. Therefore, double girdling was applied on young oak trees (Quercus robur L.) to manipulate the sugar flow by mechanically removing a complete band of bark at two different heights. The double girdling effects on both the water and sugar transport were investigated by analysing stem diameter variations, photosynthesis, xylem sap flow and content of carbohydrates. The double-girdled oak trees were divided in three stem zones: (1) the upper stem zone (U) still receiving new assimilates from the leaves, (2) the lowest stem zone (L) receiving only stored sugars from the roots, and (3) the middle stem zone (M) completely isolated from crown and roots. As downward carbon transport was interrupted by girdling, the stem expansion and carbohydrate content increased in U, indicating that U became the major sink instead of the roots. In contrast to U, stem expansion and carbohydrate content decreased in the two lower stem zones (M and L). Furthermore, a decrease in photosynthesis and sap flow rate was observed, which could be attributed to an indirect effect of girdling.

The internal cleaning of tank trucks gives rise to a highly variable and complex wastewater often containing significant amounts of persistent organics and toxic micro-pollutants. This paper evaluates the use of partial chemical oxidation to remove COD and estrogenic activity from a tank truck cleaning generated wastewater concentrate. Following ranking of processes regarding removal efficiency was established: direct partial ozonation (pH 7.5) > advanced partial ozonation (pH 11.5) > partial hydrogen peroxide oxidation. Wastewater estrogenic activity (1.78 ng/L 17β-estradiol equivalents) is effectively lowered below the detection limit of the applied MCF-7 cell proliferation assay by all studied oxidation processes.

This article compares multi-stage ozonation-biological treatment processes [O-3 - Biodegradation - O-3 - Biodegradation] and conventional partial ozonation processes [O-3 -Biodegradation] for the removal of S-COD and toxicity from a recalcitrant and toxic wastewater concentrate. Both direct (pH 7.5) and advanced (pH 11.5) ozonation are evaluated. Short-term estimation methods for biodegradability (respirometry) and toxicity (30 min Vibrio fisheri luminescence inhibition) are evaluated for their use in process control.

A kinetic model for multicomponent substrate removal by the partial ozonation process is presented. The model consists of a component describing the co-evolution of COD and BOD as a function of ozone dosage and a mass transfer based component describing ozone dosage as a function of time. A multiple zero order reaction concept is used to describe the multicomponent kinetic behaviour. The model has been verified experimentally by comparing stoichiometric ratio and ozone reaction rate of conventional partial ozonation processes and partial ozonation processes with intermittent biodegradation. The model is found to effectively describe the change in stoichiometry and reaction rate that typically occurs during ozonation processes. Intermittent biodegradation is found to have no effect on the instantaneous stoichiometric ratio or the instantaneous ozone reaction rate. This implicates that intermittent BOD removal results in an additional decrease in required ozonation time and ozone demand compared to the conventional partial ozonation process and in addition to the expected decrease resulting from BOD removal.