Elsevier

Desalination

Volume 464, 15 August 2019, Pages 48-62
Desalination

Open-source industrial-scale module simulation: Paving the way towards the right configuration choice for membrane distillation

https://doi.org/10.1016/j.desal.2019.04.018Get rights and content

Highlights

  • Open-source simulators were developed for performance prediction of industrial-scale DCMD, S-VMD and X-VMD

  • S-VMD is more susceptible to radial heat loss, whilst axial heat loss determines DCMD and X-VMD performance

  • Undesired heat pinch phenomenon affects large-scale MD performance in the order of: DCMD > X-VMD > S-VMD

  • S-VMD not only shows the highest flux but also is the configuration least sensitive to scale-up

  • Choice of materials is critical for industrial-scale S-VMD, but less essential for DCMD and X-VMD

Abstract

Understanding the impact of configuration selection of industrial-scale membrane distillation (MD) is critical for a productive and energy-efficient operation of this emerging process for desalination. However, it is rarely considered even as new high flux membranes and wider applications are developed. In this context, three open-source simulators were developed on the Matlab GUI platform for the performance prediction of industrial-size direct contact membrane distillation (DCMD), submerged vacuum membrane distillation (S-VMD), and cross-flow vacuum membrane distillation (X-VMD). Using laboratory-scale experimental results as simulation inputs, the developed simulators were able to predict large-scale MD performance. Furthermore, design considerations on appropriate module scale-up for all three configurations were demonstrated. More importantly, the configuration that shows the optimal mass and heat transfer behaviour was revealed through the performance comparisons across different full-size MD configurations. In addition, these simulators are open-source allowing researchers to use these tools for the development of specific scale-up strategies of their own MD membranes — a critical step towards commercialisation.

Introduction

Recent years have witnessed an increasing interest in membrane distillation (MD) for desalination – an innovative concept that utilises membrane to recover pure water from seawater or brackish water. Many advantages have been identified for MD desalination, for instance it can utilise low-grade heat such as solar-thermal and geo-thermal energy. Furthermore, unlike reverse osmosis (RO) process for desalination, the capacity of MD is not limited by osmotic pressure, indicating its potential use in the area where the salinity is too high to be processed by RO [[1], [2], [3], [4], [5]]. In response to the increased industrial attention on MD process, innovative MD development has become a topical research area, with most efforts being devoted to material chemistry. Meanwhile, investigation on MD configuration and module design, particularly in industrial-scale, has received much less attention, despite the efficiency and efficacy of full-scale MD desalination processes being strongly dependent upon an appropriate choice of MD configuration and an optimal module design [6].

Typical MD configurations include direct contact membrane distillation (DCMD), air gap membrane distillation (AGMD), sweeping gas membrane distillation (SGMD), and vacuum membrane distillation (VMD), which can be further divided into cross-flow VMD (X-VMD) and submerged VMD (S-VMD). Many studies can be found in open literature on the performance evaluation and optimisation of individual configurations. However, comparison across different configurations has been barely broached, with only few publications available to date.

Among the limited number of publications on comparing mass and heat transfer behaviours across different configurations, most stated that the VMD offers the highest water flux due to its negligible conductive heat loss [[7], [8], [9], [10], [11], [12], [13], [14]]. However, this conclusion may only valid for certain VMD configurations with forced convection and cannot be extended to S-VMD with natural convection. S-VMD is a less well-known configuration but has some potential benefits in terms of overcoming heat pinch and solids handling by suspending the membrane in the feed tank. This aspect will be elaborated further in the following sections. Following VMD in performance is the DCMD, and its lower water flux is mainly attributed to the conductive heat loss [[15], [16], [17], [18]]. At the other end of the performance spectrum are the SGMD and AGMD. The low water flux provided by these two configurations is mainly due to the extra mass transfer resistance imposed by the air film in the permeate side [19,20]. Furthermore, the stagnant air film of AGMD further amplifies the mass transfer resistance, resulting in AGMD having the lowest flux [20]. Apart from water flux as one obvious performance measure, the choice of configuration should also consider thermal efficiency. For this criterion, VMD, once again, appears to be the best due to its negligible conductive heat loss, and thus the highest thermal efficiency. Similar to VMD, AGMD and SGMD also provide good thermal efficiency as a result of the negligible conductive heat loss [21]. Whereas in the case of DCMD, the direct contact of membrane with the cold liquid leads to additional heat loss, and therefore, a lower thermal efficiency. In addition to the aforementioned factors, other issues in terms of equipment and operating complexity need to be taken into account. For instance, the design and operation of DCMD is considered the simplest, due to the fact that condensation takes place inside the membrane module [3,22]. As for VMD, which has the highest water flux and thermal efficiency, it also has improved mass transfer inside the membrane due to the removal of air in the membrane pores [23,24]. However, despite its many advantages over other configurations, the potential of undesired pore wetting for VMD is higher than other configurations due to the vacuum applied on the permeate side [25,26]. In addition, costs of vacuum systems are also a potential concern, and the need for an external condenser outside the membrane module complicates the design and operation of the VMD system but may offer additional opportunities for heat recovery and compact module configurations [3].

Although the aforementioned studies seem to provide a ranking for the selection of MD configurations, these works were only conducted in lab-scale MD settings with small membrane areas. Many conclusions drawn from these works cannot be extended to industrial-scale MD modules with membrane areas that are orders of magnitude larger. In these large modules, the driving force could diminish over module length because of the heat transfer across the membrane (heat pinch effect). Such phenomenon are difficult to observe in lab-scale settings due to the small membrane size and negligible heat loss along the module length [3]. Our previous simulation study demonstrated that, for desalination with DCMD, when scaling up two vastly different membranes from lab- to industrial-scale, the performance gap between the two membranes quickly closed as the result of this heat pinch phenomenon, with the supposedly “good” membrane only showing marginally better performance than the “bad” membrane at full industrial-size [27]. The significantly different performance between the lab- and industrial-scale DCMD modules suggests that many aforementioned conclusions based on the lab-scale configurations do not address significant heat and mass transfer issues with industrial-size modules. However, to compare the MD performances in industrial-scale for the selection of MD configurations is a task difficult to achieve in most laboratory settings. Fortunately, such a challenge can be tackled by a computer-aided simulation, which can predict mass and heat transfer behaviour inside large MD modules, and thus, providing critical information to supplement MD configuration selection. In this vein, several modelling efforts have been made in the past aiming to predict mass and heat transfer in the MD processes. However, few constraints were found in these studies: (i) most were developed for guiding membrane development in lab-scale (e.g., predicting how the thickness or porosity of a small size membrane affect lab-scale performance), and thus, unsuitable for large-scale module design; (ii) most focused on specific MD configurations and cannot be used for comparison across different configurations [3]; (iii) many were developed based on expensive commercial software such as Aspen HYSYS or ANSYS that are inaccessible to many membrane researchers, and (iv) all these simulators are unavailable to public, which has severely limited the impact of these works within membrane research.

In this context, also considering that the DCMD and VMD are widely recognised as the most promising configurations for desalination [28], the current work aims to study the mass and heat transfer behaviour inside industrial-size DCMD, S-VMD, and X-VMD modules with the aid of computer simulation. These MATLAB-based simulators were developed based on an algorithm that couples finite difference method and black box method and were capable of predicting industrial-scale MD module performance using lab-scale experimental results. Using the results from these simulators, we were able to reveal the relationship between the lab-scale MD performance and industrial-scale MD module design, and therefore, bridging the gap between academic membrane research and industrial MD design. Furthermore, the comparisons across configurations allowed us to examine the effect of undesired heat pinch phenomenon on each configuration, and thus revealing the least affected configuration which should also give the highest pure water productivity. In addition, these simulators are open-source, which allows researchers to use these tools to develop specific scale-up strategies for their own MD membranes — a critical step towards commercialisation.

Section snippets

Simulator development

Similar to our previous work on DCMD simulation [27], all the simulators developed in the current study take lab-scale experimental results as inputs to compute the membrane water permeation coefficient (kg·m−2·Pa−1·s−1) — an intrinsic membrane property that is only affected by the structural and thermo-physical properties of the membrane itself. Subsequently, the simulators take this coefficient as input to predict the large-scale MD performance. The following assumptions were made during the

Experimental

The experiments carried out in this study were to validate the two in-house developed simulators (S-VMD and X-VMD). Note that the validation of DCMD simulator was reported in our previous work [27]. To study the performance response of different membrane types on module scale-up, two types of hollow fibre membranes were used to provide lab-scale simulation inputs, and they are: polypropylene (PP) and polyvinylidene fluoride (PVDF) hydrophobic hollow fibre membranes. The structural and

Simulator validation

The lab-scale validation was done by comparing the experimental and simulation results of two hollow fibre membranes (PP and PVDF) in three configurations (DCMD, S-VMD, and X-VMD). Note that for lab-scale S-VMD validation, no agitation was applied. This decision was made considering (i) the magnetic stirrer used in lab-scale set-up gives different hydrodynamic behaviour from blade paddle or Rushton turbine found in large-scale tank agitation; and (ii) stirring in lab-scale single-fibre setting

Conclusions

Three open-source simulators were developed on the Matlab GUI platform for the performance prediction of industrial-scale DCMD, S-VMD, and X-VMD. The developed simulators were subsequently used to demonstrate selection considerations for the appropriate configurations for industrial-scale MD desalination. By means of a thorough assessment including a wide range of criteria, the following conclusions were drawn.

  • Significant heat loss in radial direction is a key concern for industrial-scale

List of symbols

AMembrane aream2
CmMembrane water permeation coefficientkg·m−2·Pa−1·s−1
cpSpecific heat capacityJ·kg−1·K−1
dDiameter of the agitatorm
hHeat transfer coefficientW·m−2·K−1
JWater flux across the membranekg·s−1
kThermal conductivityW·m−1·K−1
LEffective fibre lengthm
NAgitator speeds−1
nNumber of fibres in one module
pVapour pressurePa
poVacuum pressurePa
qcConductive heat transfer rateW
qvVaporisation latent heatW
rHollow fibre radiusm
T/tTemperatureK/°C
Greek letters
βThermal expansion coefficient
δ

References (54)

  • S. Bandini et al.

    Separation efficiency in vacuum membrane distillation

    J. Membr. Sci.

    (1992)
  • J.I. Mengual et al.

    Heat and mass transfer in vacuum membrane distillation

    Int. J. Heat Mass Transf.

    (2004)
  • L. Eykens et al.

    How to select a membrane distillation configuration? Process conditions and membrane influence unraveled

    Desalination

    (2016)
  • M. Essalhi et al.

    Surface segregation of fluorinated modifying macromolecule for hydrophobic/hydrophilic membrane preparation and application in air gap and direct contact membrane distillation

    J. Membr. Sci.

    (2012)
  • F.A. Banat et al.

    Theoretical and experimental study in membrane distillation

    Desalination

    (1994)
  • A. Cipollina et al.

    Development of a membrane distillation module for solar energy seawater desalination

    Chem. Eng. Res. Des.

    (2012)
  • S. Yarlagadda et al.

    Potable water recovery from As, U, and F contaminated ground waters by direct contact membrane distillation process

    J. Hazard. Mater.

    (2011)
  • M. Khayet et al.

    Theory and experiments on sweeping gas membrane distillation

    J. Membr. Sci.

    (2000)
  • M.A.E.-R. Abu-Zeid et al.

    A comprehensive review of vacuum membrane distillation technique

    Desalination

    (2015)
  • G.C. Sarti et al.

    Extraction of organic components from aqueous streams by vacuum membrane distillation

    J. Membr. Sci.

    (1993)
  • G. Dong et al.

    Open-source predictive simulators for scale-up of direct contact membrane distillation modules for seawater desalination

    Desalination

    (2017)
  • J. Phattaranawik et al.

    Heat transport and membrane distillation coefficients in direct contact membrane distillation

    J. Membr. Sci.

    (2003)
  • L. Martínez-Díez et al.

    Temperature and concentration polarization in membrane distillation of aqueous salt solutions

    J. Membr. Sci.

    (1999)
  • P. Mohan et al.

    Review heat transfer to Newtonian fluids in mechanically agitated vessels

    Exp. Thermal Fluid Sci.

    (1992)
  • W. Zhong et al.

    Superhydrophobic membranes via facile bio-inspired mineralization for vacuum membrane distillation

    J. Membr. Sci.

    (2017)
  • F. Laganà et al.

    Direct contact membrane distillation: modelling and concentration experiments

    J. Membr. Sci.

    (2000)
  • M. Khayet et al.

    Porous hydrophobic/hydrophilic composite membranes: application in desalination using direct contact membrane distillation

    J. Membr. Sci.

    (2005)
  • Cited by (16)

    • Carbon-based material derived from biomass waste for wastewater treatment

      2022, Environmental Advances
      Citation Excerpt :

      This is because carbon-based compounds have only been expressed in lab-scale quantities for screening, functional, regulatory, or structural studies up to now. It can be difficult to scale up a lab-size membrane for outdoor performance testing (Dong et al., 2019). Consistency in membrane quality for large-volume processes, membrane module design and fabrication procedures, harsh operating environments with fluctuating field pollutants, and process facility requirements for offshore applications are among the challenges (Kadirkhan et al., 2019).

    • Scaling mitigation and salt reduction of vacuum membrane distillation using sacrificial zeolites

      2022, Journal of Environmental Chemical Engineering
      Citation Excerpt :

      In addition, the interconnected cavities within the negatively charged framework contain metal cations and water molecules, resulting in great cation exchange properties, with potential for metal uptakes from effluents and seawater [14–16]. Compared to conventional cross-flow MD, submerged MD (SMD) is more beneficial in terms of thermal energy consumption, heat conservation and process integration due to the absence of feed circulation in the configuration [17]. Furthermore, in SMD configuration, zeolites can be charged directly into the membrane-submerged feed tank, generating a compact integration system and increasing residence time of the zeolites in contact with the solutions supporting greater adsorption.

    • Modification of GO-based pervaporation membranes to improve stability in oscillating temperature operation

      2021, Desalination
      Citation Excerpt :

      Furthermore, the fabrication of freestanding dense chitosan/PVA with GO membranesis usually done by the casting method, which requires more volume of chemicals and limited to flat sheet membranes. However, hollow fiber membranes are easier to scale-up by increasing packing density and length of the fibers [32]. In this study, the aromatic diamine with sulfonic groups was used as a guest molecule to cross-link GO nanosheets to increase membrane stability and water transport.

    • Pilot-scale vacuum membrane distillation for decontamination of simulated radioactive wastewater: System design and performance evaluation

      2021, Separation and Purification Technology
      Citation Excerpt :

      In contrast, pilot-scale plants may suffer from a significant temperature drop in the feed side of the membrane module, due to insufficient energy supply resulting from vaporization latent heat demand when the flow rate is too slow. Dong et al. [33] developed a simulator for the performance prediction of industrial-size membrane modules for membrane distillation, indicating that permeate flux decays with the increase of the effective membrane area. The significant heat loss in radial direction as temperature polarization and in axial direction as the heat pinch effect are the main issues for industrial-scale MD plant.

    View all citing articles on Scopus
    View full text