Antoine Kemperman

Native protein recovery from potato fruit juice by ultrafiltration

Potato fruit juice, i.e. the stream resulting after the extraction of the starch from the potato, contains up to 2.5% [w/w] of proteins that are potentially valuable for the food market. However, today the recovery of protein from the potato fruit juice with reverse osmosis membranes results in a protein concentrate that is not suitable for human consumption. The described research shows that the use of ultrafiltration with additional diafiltration is able to produce a higher quality protein. Tests with the produced protein show that the quality depends on the rate of diafiltration used and that the product has functional properties that are equal or better than the compared commercial food product that are currently used.

Fouling control mechanisms of demineralized water backwash: Reduction of charge screening and calcium bridging effects

This paper investigates the impact of the ionic environment on the charge of colloidal natural organic matter (NOM) and ultrafiltration (UF) membranes (charge screening effect) and the calcium adsorption/bridging on new and fouled membranes (calcium bridging effect) by measuring the zeta potentials of membranes and colloidal NOM. Fouling experiments were conducted with natural water to determine whether the reduction of the charge screening effect and/or calcium bridging effect by backwashing with demineralized water can explain the observed reduction in fouling. Results show that the charge of both membranes and NOM, as measured by the zeta potential, became more negative at a lower pH and a lower concentration of electrolytes, in particular, divalent electrolytes. In addition, calcium also adsorbed onto the membranes, and consequently bridged colloidal NOM and membranes via binding with functional groups. The charge screening effect could be eliminated by flushing NOM and membranes with demineralized water, since a cation-free environment was established. However, only a limited amount of the calcium bridging connection was removed with demineralized water backwashes, so the calcium bridging effect mostly could not be eliminated. As demineralized water backwash was found to be effective in fouling control, it can be concluded that the reduction of the charge screening is the dominant mechanism for this.

Tight ceramic UF membrane as RO pre-treatment: The role of electrostatic interactions on phosphate rejection

Phosphate limitation has been reported as an effective approach to inhibit biofouling in reverse osmosis (RO) systems for water purification. The rejection of dissolved phosphate by negatively charged TiO2 tight ultrafiltration (UF) membranes (1 kDa and 3 kDa) was observed. These membranes can potentially be adopted as an effective process for RO pre-treatment in order to constrain biofouling by phosphate limitation. This paper focuses on electrostatic interactions during tight UF filtration. Despite the larger pore size, the 3 kDa ceramic membrane exhibited greater phosphate rejection than the 1 kDa membrane, because the 3 kDa membrane has a greater negative surface charge and thus greater electrostatic repulsion against phosphate. The increase of pH from 6 to 8.5 led to a substantial increase in phosphate rejection by both membranes due to increased electrostatic repulsion. At pH 8.5, the maximum phosphate rejections achieved by the 1 kDa and 3 kDa membrane were 75% and 86%, respectively. A Debye ratio (ratio of the Debye length to the pore radius) is introduced in order to evaluate double layer overlapping in tight UF membranes. Threshold Debye ratios were determined as 2 and 1 for the 1 kDa and 3 kDa membranes, respectively. A Debye ratio below the threshold Debye ratio leads to dramatically decreased phosphate rejection by tight UF membranes. The phosphate rejection by the tight UF, in combination with chemical phosphate removal by coagulation, might accomplish phosphate-limited conditions for biological growth and thus prevent biofouling in the RO systems.

Hydrogel-coated feed spacers in two-phase flow cleaning in spiral wound membrane elements: A novel platform for eco-friendly biofouling mitigation

Biofouling is still a major challenge in the application of nanofiltration and reverse osmosis membranes. Here we present a platform approach for environmentally friendly biofouling control using a combination of a hydrogel-coated feed spacer and two-phase flow cleaning. Neutral (polyHEMA-co-PEG10MA), cationic (polyDMAEMA) and anionic (polySPMA) hydrogels have been successfully grafted onto polypropylene (PP) feed spacers via plasma-mediated UV-polymerization. These coatings maintained their chemical stability after 7 days incubation in neutral (pH 7), acidic (pH 5) and basic (pH 9) environments. Anti-biofouling properties of these coatings were evaluated by Escherichia coli attachment assay and nanofiltration experiments at a TMP of 600xa0kPag using tap water with additional nutrients as feed and by using optical coherence tomography. Especially the anionic polySPMA-coated PP feed spacer shows reduced attachment of E. coli and biofouling in the spacer-filled narrow channels resulting in delayed biofilm growth. Employing this highly hydrophilic coating during removal of biofouling by two-phase flow cleaning also showed enhanced cleaning efficiency, feed channel pressure drop and flux recoveries. The strong hydrophilic nature and the presence of negative charge on polySPMA are most probably responsible for the improved antifouling behavior. A combination of polySPMA-coated PP feed spacers and two-phase flow cleaning therefore is promising and an environmentally friendly approach to control biofouling in NF/RO systems employing spiral-wound membrane modules.

Desalination of brackish groundwater and concentrate disposal by deep well injection

In the province of Friesland (in the Northern part of The Netherlands), problems have arisen with the abstraction of fresh groundwater due to salinization of wells by upcoming of brackish water. A solution to this problem is to intercept (abstract) the upcoming brackish water, desalinate it with a brackish water reverse osmosis installation, and dispose the concentrate in a deeper, confined aquifer. The fresh-brackish interface in the source aquifer is stabilized by simultaneous abstraction of the fresh and brackish parts. After desalination, the abstracted brackish water provides an additional source for drinking water. To demonstrate the feasibility of this concept a pilot study was set up. In one year about 220.000 m3 of concentrate was produced and injected. The reverse osmosis installation was carried out under anaerobic conditions without pretreatment and antiscalant dosing. Despite the high iron concentrations (40 mg Fe/l) in the feed water, the installation performed was very stable. Although the concentrate was supersaturated toward carbonate and phosphate minerals (SI > 1), scaling or fouling of the membranes did not occur at recoveries of 50, 70, and 75%. The mass transfer coefficient or normalized flux (at 10°C) was stable at 0.85·10-8 m s-1 kPa-1. Water quality changes in the target aquifer were monitored by two observation wells, at 12 and 24 m distances from the injection well. Also the injection of the supersaturated concentrate did not lead to mineral precipitation in the target aquifer, indicating that deep well injection is technically feasible without risks of injection well or aquifer clogging. The fresh-brackish water interface in the source aquifer remained stable by the simultaneous abstraction of fresh and brackish water. This showed that the so called “fresh-keeper” concept works in practise, providing a successful remedy against salinization of fresh water abstraction wells. Overall, the pilot study showed that brackish groundwater provides an excellent, additional source for drinking water in The Netherlands or in other coastal areas worldwide, where fresh groundwater is scarce or where fresh water wells are threatened by salinization

SDI: Is it a reliable fouling index?

The ASTM considers the silt density index (SDI) test as a standard test for fouling potential of RO and NF feed waters. Up to date, the SDI is used at many full- and pilot-scale installations. The design and choice of the applied RO pretreatment is to a large extent based on the SDI test on the raw feed water. Comparing and monitoring UF/MF membrane performance is another SDI application. From a practical point of view, the SDI of RO feed water preferably should be lower than 3. The SDI has several disadvantages making it an unreliable test. The SDI has a non linear relationship with the colloidal concentration in the water and it is not corrected for the feed water temperature. Besides that, the SDI is not based on any filtration model. SDI is trying to simulate the RO fouling using dead-end MF membrane. The modified fouling index (MFI) is another fouling index. The MFI is based on the cake filtration model, can be corrected for pressure and temperature and is therefore used as a promising alternative for the SDI. Nevertheless, the procedure of measuring a MFI is more difficult and not directly suitable for carrying out “in the field”. The objective of this study therefore is to determine a theoretical relationship between SDI and MFI, and to validate this with experimental results. This relationship can be used to investigate the influence of membrane and testing parameters on SDI under cake filtration conditions, implying this model is valid for cake filtration mechanism and a particle rejection of 100%. In order to calculate the SDI, the times t 1 and t 2 for collecting the first and second sample are predicted using the measured MFI value and the MFI definition. In this research, the influence of several parameters (such as temperature, membrane resistance, etc.) on the SDI will be shown. The experimental results show a good agreement with the theoretical work, but only if the cake filtration start builds up directly at the beginning of the experiment. In general, this work clearly demonstrates that SDI currently is not reliable test for RO fouling. Either corrections for the SDI are necessary to give a more reliable index, or a new index has to be developed.

Dominant factors controlling the efficiency of two-phase flow cleaning in spiral-wound membrane elements

Two-phase flow cleaning has been successfully applied to control fouling in spiral wound membrane elements. This study focuses on its experimental optimization using a Taguchi Design of Experiment method (L-25 orthogonal arrays) to elucidate the influence of different factors and to reveal the important one(s) affecting the cleaning efficiency of two-phase flow cleaning. All possible combinations of the factors, i.e. feed type, spacer geometry, gas/liquid ratio, and liquid velocity, each at five levels were evaluated. The main effect of each factor on the efficiency of two-phase flow cleaning was measured by determining the performance response (mean of cleaning efficiency) and by calculating the mean signal-to-noise ratio. An analysis of variance was applied to calculate the relative contribution of each factor on the efficiency of two-phase flow cleaning. The results showed that the feed type is by far the most essential factor contributing to the cleaning efficiency. The spacer geometry is ranked second, followed by the gas/liquid ratio and the liquid velocity, which both have an only very minor effect on the cleaning performance. In terms of practical application, the operator should consider first the type of foulant prior to taking a decision on whether or not two-phase flow cleaning will be effective. Once the foulant type is defined, the use of the highest gas/liquid ratio, the highest liquid velocity, and the thickest feed spacer (diamond type) are recommended to achieve maximum two-phase flow cleaning efficiency.

Sensitivity of SDI for experimental errors

Silt density index (SDI) testing is a widely-accepted method for estimating the rate at which colloidal and particle fouling will occur in water purification systems when using reverse osmosis (RO) or nanofiltration (NF) membranes. However, the SDI has several deficiencies. For example, the SDI has no linear relationship with the particle concentration, is not based on any fouling mechanism, and is not corrected for temperature, pressure and membrane resistance. The accuracy and reproducibility of the SDI is often questioned. In this study, mathematical models were developed to investigate the sensitivity of SDI for the following types of errors: errors due to inaccurate lab or field equipment, systematic errors, and errors resulting from artifacts and personal observations and experience. The mathematical results were verified experimentally. Both the mathematical models and experimental results show that the membrane resistance RM has the highest impact on the SDI results. The allowable ASTM variation in RM is responsible for a deviation in SDI between 2.29 and 3.98 at a level of SDI = 3. Besides that, a 1 s error in measuring the time to collect the second sample t2 results in ±0.07 at SDIO = 3. The artifacts and personal experience also influence the SDI results. The total error in measuring SDI was estimated to be equal to ±2.11 in the field and only ±0.4 in the lab in level of SDIO = 3. Furthermore, several recommendations are mentioned based on these theoretical results and our personal experience. This study demonstrates the sensitivity of the SDI for errors in RM and the accuracy of the equipments, and explains the difficulties in reproducing SDI results for the same water.