You're using an outdated browser. Please upgrade to a modern browser for the best experience.
Low-Pressure Membrane Systems: Comparison
View Latest Version
Please note this is a comparison between Version 2 by Dean Liu and Version 1 by José M. Gozálvez-Zafrilla.

Peristaltic pumping is used in membrane applications where high and sterile sealing is required. Control is difficult due to the pulsating pump characteristics and the time-varying properties of the system. Three artificial intelligenge techniques (artificial neural networks, fuzzy-logic expert systems and fuzzy-loguc local models) were used to regulate pressure and crossflow velocity in a microfiltration process with humic acid fouling. All techniques were able to control de plant but with different control performance.  three artificial three artificial intelligence control strategies (artificialneural networks (ANN), fuzzy logic expert systems, and fuzzy-integrated local models) were usedto regulate transmembrane pressure and crossflow velocity in a microfiltration system under highfouling conditions.intelligence control strategies (artificialneural networks (ANN), fuzzy logic expert systems, and fuzzy-integrated local models) were usedto regulate transmembrane pressure and crossflow velocity in a microfiltration system under highfouling conditions.  three artificial intelligence control strategies (artificialneural networks (ANN), fuzzy logic expert systems, and fuzzy-integrated local models) were usedto regulate transmembrane pressure and crossflow velocity in a microfiltration system under highfouling conditions.

  • low-pressure driven process
  • peristaltic pump
  • microfiltration
  • intelligent control

1. Introduction

Peristaltic pumping is used in membrane processes, such as microfiltration (MF) or ultrafiltration (UF), when high sealing without contact of the solutions with the lubricating fluids or friction elements is required. Typical applications of this configuration are in food processing industry separations, membrane bioreactors, membrane medical applications, and laboratory experimentation.
The use of peristaltic pumps in membrane bioreactor systems also has the advantage of diminishing the effect of tearing stress. For example, anaerobic sludge membrane reactors using peristaltic pumping for pressure application and recirculation have been used to treat municipal or industrial waters [1,2][1][2]. This kind of pumping allows special membrane applications such as the use of fluidized glass beads to improve membrane performance [3].
In medical systems using membranes, mainly in artificial kidneys or microdialysis sensors, pumps must not introduce harmful agents and must maintain sterility. Peristaltic pumps can meet this requirement with suitable flow rate characteristics [4]. However, problems related to the formation of particles may occur. These particles may be caused by wear of the tube due to spallation [5] or degradation of the solution. An example of the latter situation is the formation of protein particles caused by the effect of the tube tearing on proteins adsorbed on the tube walls [6].
The combination of peristaltic pumps and membranes is also widely used on a laboratory scale. For example, multi-channel peristaltic pumps can perform many membrane experiments at the same time [7] or experiments in which two peristaltic pumps work in opposite directions [8].

2. Modelling of Peristaltically Pumped Low-Pressure Driven Membrane Systems

The development of control systems to meet permeate flow specifications is essential for the proper functioning of the above-mentioned applications.
Most control systems are, in some way, model based. Therefore, the existence of a computational model deduced from the physical behavior of the system can facilitate control development. However, modeling and control of peristaltically pumped MF or UF membrane systems can be more challenging than other membrane configurations.
The behavior and performance of peristaltic pumps can be approximately modeled by lumped models based on physical considerations [9] or by using computational fluid dynamics [10]. Disturbance models can also be used for control purposes [11]. However, leit uis supposed that more accurate time-varying models are needed for control. In that case, the models could be improved by taking into account changes in feed properties and mechanical characteristics of the tubing material.
Membrane modelling must describe the permeate flux and component rejection and its evolution over time. In MF, retention depends on particle size and pore size distribution. The flow through the pores can be described by Poiseuille’s law. In UF, modelling of solute and solvent transport is based on hydrodynamic equations describing hindered diffusion and convection [12]. For most processes, fouling adds additional resistance to flow. Moreover, the effect of fouling involves longer time dynamics than that those caused by feed or operational variations. To describe the different fouling mechanisms (pore blocking or gel layer formation), empirical models such as those developed by Hermia can be used [13]. Other modelling difficulties are the different fouling potential of the solutions [14], the important effect of spacer design on fouling [15], and the fact that the membrane performance does not fully recover after the cleaning procedures. For both processes (MF and UF), the situation can be more complex in the case of a non-constant flow [16], as in the case of the pulsating flow produced by a peristaltic pump.

3. Intelligent Control Approaches for Low-Pressure Membrane Systems

In conclusion, accurate modeling of peristaltically pumped low-pressure driven membrane systems is challenging due to the changes in system performance over time. In general, models derived from first principles with sufficient accuracy are very complex. Therefore, classical control techniques cannot be applied directly. On the other hand, those models simple enough to apply well-known control techniques cannot describe the long-term behavior of the system. This fact led the authors to consider the use of artificial intelligence (AI) and system identification methods as the most appropriate approach.
Most attempts to control peristaltically pumped membrane systems have come from the field of hemodialysis, where accurate control of fluid delivery is especially critical [17,18][17][18]. In this application, the use of hierarchical adaptive and supervisory control has allowed adjusting the pump inputs to patient monitoring data [19].
In other fields where MF or UF have been applied, the modeling and control focus on the interaction of the membrane and the pump, but on the membrane performance. Niu et al. have recently carried out a critical review on the use of different AI methods in fouling prediction [20]. They found that the most modeled features were transmembrane pressure, flux dynamics, and flux decline by fouling. They divided the AI techniques into single and hybrid algorithms. They indicated that the most employed single algorithm was artificial neural networks (ANN), but others such as fuzzy logic, genetic programming, or support vector machines were also used. Hybrid algorithms were usually built by combining these techniques with a search algorithm. Jawad et al. conducted a review that found that permeate flux is the most modeled feature in the different MF or UF applications. The same study showed that the most used model input is transmembrane pressure (TMP), and that the composition-related parameters were also used as input in the different works [21]. Recently, machine learning has been used to model the dynamics of filtration and backwashing of UF by a back propagation ANN [22]. Ultrafiltration of protein solutions has been effectively modeled using ANN for a tubular crossflow membrane [23]. ThIt is work  used as inputs: operational time, pH, and ionic strength; and as outputs: filtrate flow and protein transmission. The authors compared the performance of ANN with that of the Hermia models [24] and obtained similar performance results for the fitted experiments, but better ANN extrapolation capability for experiments not included in the fitting process. Other AI-based modelling tools, such as fuzzy logic, have proven to be an alternative to ANN modelling [25].

4. Selection of Study Case for Comparison of Control Approaches

Given the promising results found in the literature on AI methods applied to low-pressure driven processes, tresearchis work ers aimed to compare the performance of a set of AI modeling techniques for controlling peristaltically pumped membrane processes subjected to strong fouling. The specific process case studied used a ceramic MF membrane with a pore size close to the maximum UF pore size range. The operating conditions of the MF system to be controlled were average crossflow velocity and average transmembrane pressure. Forced membrane fouling was expected to produce permeability variability due to both reversible and irreversible fouling.
The chosen techniques cannot only provide an accurate system description but can also deliver models that are good for subsequent control design under all the expected operating conditions. This balance between modeling and control could lead to a better performance of the controlled MF system. Therefore, a comparison between the available methods was made to determine which can be more effective in controlling the operating conditions in a changing membrane system.
In order to obtain the data necessary for the creation of the models, each experiment consisted of an operational step with fouling followed by a cleaning step. For the first step, the selected foulant was humic acid at a relatively high concentration. Humic acid fouling is one of the main factors limiting MF in water treatment [26,27][26][27] with humic acid aggregates being responsible for most of the fouling [28]. For the second step, it was considered that fouling by organic matter can be removed from ceramic membranes by cleaning-in-place (CIP) procedures using alkalis. When NaOH is used, typical concentrations are in the range of 0.5–2% wt. [29,30][29][30]. The combination of rapid fouling with humic acid and subsequent cleaning with NaOH allowed short dynamic experiments with a rapid decrease in flow rate suitable for model identification and control design.

References

  1. Ozgun, H.; Tao, Y.; Ersahin, M.E.; Zhou, Z.; Gimenez, J.B.; Spanjers, H.; van Lier, J.B. Impact of Temperature on Feed-Flow Characteristics and Filtration Performance of an Upflow Anaerobic Sludge Blanket Coupled Ultrafiltration Membrane Treating Municipal Wastewater. Water Res. 2015, 83, 71–83.
  2. Muñoz Sierra, J.D.; Oosterkamp, M.J.; Wang, W.; Spanjers, H.; van Lier, J.B. Comparative Performance of Upflow Anaerobic Sludge Blanket Reactor and Anaerobic Membrane Bioreactor Treating Phenolic Wastewater: Overcoming High Salinity. Chem. Eng. J. 2019, 366, 480–490.
  3. Düppenbecker, B.; Engelhart, M.; Cornel, P. Fouling Mitigation in Anaerobic Membrane Bioreactor Using Fluidized Glass Beads: Evaluation Fitness for Purpose of Ceramic Membranes. J. Memb. Sci. 2017, 537, 69–82.
  4. Armignacco, P.; Garzotto, F.; Bellini, C.; Neri, M.; Lorenzin, A.; Sartori, M.; Ronco, C. Pumps in Wearable Ultrafiltration Devices: Pumps in Wuf Devices. Blood Purif. 2015, 39, 115–124.
  5. Liu, Y.; Faria, M.; Leonard, E. Spallation of Small Particles from Peristaltic Pump Tube Segments. Artif. Organs. 2017, 41, 672–677.
  6. Deiringer, N.; Friess, W. Proteins on the Rack: Mechanistic Studies on Protein Particle Formation During Peristaltic Pumping. J. Pharm. Sci. 2022, 111, 1370–1378.
  7. Kazemi, A.S.; Patterson, B.; LaRue, R.J.; Papangelakis, P.; Yoo, S.M.; Ghosh, R.; Latulippe, D.R. Microscale Parallel-Structured, Cross-Flow Filtration System for Evaluation and Optimization of the Filtration Performance of Hollow-Fiber Membranes. Sep. Purif. Technol. 2019, 215, 299–307.
  8. Yuan, P.; Le, Z.; Zhong, L.; Huang, C. Dynamic Dialysis: An Efficient Technique for Large-Volume Sample Desalting. Prep. Biochem. Biotechnol. 2015, 45, 588–595.
  9. McIntyre, M.P.; van Schoor, G.; Uren, K.R.; Kloppers, C.P. Modelling the Pulsatile Flow Rate and Pressure Response of a Roller-Type Peristaltic Pump. Sens. Actuators A Phys. 2021, 325, 112708.
  10. Gendy, M.E.L.; Anwar Bég, O.; Kadir, A.; Islam, M.N.; Tripathi, D. CFD Simulation and Visualization of Netonian and Non-Newtonian Transport in a Peristaltic Micro-Pump. J. Mech. Med. Biol. 2021, 21, 2150058.
  11. Medvedev, A.; Hillerström, G. External Model Control of a Peristaltic Pump. IFAC Proc. Vol. 1994, 27, 517–522.
  12. van der Bruggen, B. Microfiltration, Ultrafiltration, Nanofiltration, Reverse Osmosis, and Forward Osmosis. Fundam. Modeling Membr. Syst. Membr. Process Perform. 2018, 2016, 25–70.
  13. Iritani, E.; Katagiri, N. Developments of Blocking Filtration Model in Membrane Filtration. KONA Powder Part. J. 2016, 2016, 179–202.
  14. Kecili, K.; Habarou, H.; Suty, H.; Croué, J.P.; Pontié, M. Approche Analytique Intégrée Du Colmatage de Membranes MF–UF Par Les Matières Organiques Naturelles (MON). Partie 1: Caractérisations in Situ et Modèles de Colmatage. Comptes Rendus Chim. 2006, 9, 1178–1191.
  15. Sreedhar, N.; Mavukkandy, M.O.; Aminabhavi, T.M.; Hong, S.; Arafat, H.A. Fouling Mechanisms in Ultrafiltration under Constant Flux: Effect of Feed Spacer Design. Chem. Eng. J. 2022, 446, 136563.
  16. Winzeler, H.B.; Belfort, G. Enhanced Performance for Pressure-Driven Membrane Processes: The Argument for Fluid Instabilities. J. Memb. Sci. 1993, 80, 35–47.
  17. Flendrig, J.G.; Carpay, W.M.; Dekkers, W.T. The Accurate Control of Ultrafiltration. Artif. Organs. 1978, 2, 144–146.
  18. Roberts, M.; Winney, R.J. Errors in Fluid Balance with Pump Control of Continuous Hemodialysis. Int. J. Artif. Organs. 1992, 15, 99–102.
  19. Morales, E.O.; Polycarpou, M.M.; Member, S.; Hemasilpin, N.; Bissler, J.J. Hierarchical Adaptive and Supervisory Control of Continuous Venovenous Hemofiltration. IEEE Trans. Control. Syst. Technol. 2001, 9, 445–457.
  20. Niu, C.; Li, X.; Dai, R.; Wang, Z. Artificial Intelligence-Incorporated Membrane Fouling Prediction for Membrane-Based Processes in the Past 20 Years: A Critical Review. Water Res. 2022, 216, 118299.
  21. Jawad, J.; Hawari, A.H.; Javaid Zaidi, S. Artificial Neural Network Modeling of Wastewater Treatment and Desalination Using Membrane Processes: A Review. Chem. Eng. J. 2021, 419, 129540.
  22. Zhou, Y.; Khan, B.; Gu, H.; Christofides, P.D.; Cohen, Y. Modeling UF Fouling and Backwash in Seawater RO Feedwater Treatment Using Neural Networks with Evolutionary Algorithm and Bayesian Binary Classification. Desalination 2021, 513, 115129.
  23. Ibáñez, R.; Almécija, C.; Guadix, A.; Guadix, E.M. Modelling the Ultrafiltration of Protein Solutions by Artificial Neural Networks. In Artificial Neural Networks (Mathematics Research Developments); Nova Science Publishers, Inc.: New York, NY, USA, 2011; pp. 361–384. ISBN 9781617615535.
  24. Corbatón-Báguena, M.-J.; Vincent-Vela, M.-C.; Gozálvez-Zafrilla, J.-M.; Álvarez-Blanco, S.; Lora-García, J.; Catalán-Martínez, D. Comparison between Artificial Neural Networks and Hermia’s Models to Assess Ultrafiltration Performance. Sep. Purif. Technol. 2016, 170, 434–444.
  25. Yusuf, Z.; Wahab, N.A.; Sudin, S. Soft Computing Techniques in Modelling of Membrane Filtration System: A Review. Desalination Water Treat 2019, 161, 144–155.
  26. Yuan, W.; Kocic, A.; Zydney, A.L. Analysis of Humic Acid Fouling during Microfiltration Using a Pore Blockage–Cake Filtration Model. J. Memb. Sci. 2002, 198, 51–62.
  27. Yuan, W.; Zydney, A.L. Effects of Solution Environment on Humic Acid Fouling during Microfiltration. Desalination 1999, 122, 63–76.
  28. Yuan, W.; Zydney, A.L. Humic Acid Fouling during Microfiltration. J. Memb. Sci. 1999, 157, 1–12.
  29. Gruskevica, K.; Mezule, L. Cleaning Methods for Ceramic Ultrafiltration Membranes Affected by Organic Fouling. Membranes 2021, 11, 131.
  30. Alresheedi, M.T.; Barbeau, B.; Basu, O.D. Comparisons of NOM Fouling and Cleaning of Ceramic and Polymeric Membranes during Water Treatment. Sep. Purif. Technol. 2019, 209, 452–460.
More
© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/)
Academic Video Service
Feedback