Elsevier

Journal of Membrane Science

Volume 502, 15 March 2016, Pages 171-178
Journal of Membrane Science

Energy efficiency of permeate gap and novel conductive gap membrane distillation

https://doi.org/10.1016/j.memsci.2015.12.017Get rights and content

Highlights

  • Novel conductive gap MD has 2× higher GOR than PGMD, 2.4× than AGMD.

  • Permeate flow in gap countercurrent to cold stream achieves maximum GOR.

  • Experimental validation of higher flux predicted by PGMD and CGMD models.

  • Improvement in CGMD GOR is a result of better energy recovery within module.

Abstract

This work presents numerical modeling results and flux experiments for a novel membrane distillation configuration called conductive gap membrane distillation (CGMD), as well as permeate gap membrane distillation (PGMD). CGMD has a conductive spacer in the gap between the membrane and condensing surface rather than more commonly used insulating materials. Flux measurements with two experimental systems are used to validate the numerical models for PGMD and CGMD. PGMD has 20% higher GOR (energy efficiency) than an air gap membrane distillation (AGMD) system of the same size, whereas CGMD can have two times higher GOR than even PGMD. Increasing gap effective thermal conductivity in CGMD has negligible benefits beyond kgap10W/mK under the conditions of this study. The direction of pure water flow in the gap has a significant influence on overall system energy efficiency, especially in the case of CGMD. Using a countercurrent configuration for the pure water flow in the gap relative to the cold stream leads to 40% higher GOR than flow cocurrent with the cold water stream.

Introduction

Membrane distillation (MD) is a thermal desalination technology that has received increased attention for small scale, renewable-energy-driven desalination applications as well as for treating high salinity brines. Scalability, the ability to handle high salinity feed streams, and relatively high fouling resistance are some of the advantages of the MD process. Several MD configurations with relative advantages and disadvantages have been proposed in the literature. The four common MD configurations include air gap (AGMD), direct contact (DCMD), vacuum (VMD) and sweeping gas membrane distillation (SGMD) [1], [2]. In addition, other configurations have been suggested recently, including the permeate (liquid or water) gap MD (PGMD), and material gap membrane distillation (MGMD) [3]. Several multi-staged configurations based on the above designs have also been proposed in the literature and commercially implemented [4].

The fundamental principle of operation of all these configurations is the same. Separation is achieved through evaporation of the more volatile component of a mixture. When used for desalination, pure vapor preferentially passes through the pores of a hydrophobic membrane that prevents liquid feed from passing through. The different configurations of MD vary based on how this vapor is captured and condensed to obtain pure water. In the case of DCMD, cooler pure water flows on the other side of the membrane, countercurrent to the feed, and the vapor condenses into this stream and warms it up. The warm pure water stream would then be passed through a heat exchanger to preheat incoming salt water [5], [6]. DCMD is the oldest MD configuration proposed [7]. In AGMD, the vapor condenses on a condensing plate that is cooled by the incoming feed. The feed thereby gets preheated within the module by the condensing vapor, hence recovering some of the energy input. The air gap between the membrane and the condensing surface is meant to reduce the sensible heat loss from the hot to the cold side [8].

In the case of SGMD and VMD, the condensation happens outside the MD module. The sweeping air gets heated and humidified before leaving the module in SGMD, and in VMD, the vacuum pump would draw out relatively pure vapor in the case of VMD. This is passed through a condenser or vapor trap to recover pure liquid water [9]. Recently, a system using an aspirator was proposed to more efficiently recover the product water in VMD systems [10], eliminating the vacuum pump and concerns associated with incomplete condensation.

Energy is supplied to MD systems both in the form of heat and work. Work transfer is usually used to achieve fluid flow through the channels. Heat is required to increase the temperature of the feed liquid and thereby enabling evaporation of water. Heat energy consumption by far exceeds pumping power requirements under normal operating conditions. Energy consumption in MD has been relatively higher than that of other desalination technologies as illustrated by Mistry et al. [11]. As a result, MD companies have focused on applications with available waste heat energy, from power plants or other sources [4], [12]. MD is also readily coupled with renewable and waste heat sources due to its ability to operate at low temperatures and still achieve desalination.

Energy efficiency is analyzed as gained output ratio (GOR):GOR=ṁphfgQ̇h

Most MD systems in practice have been restricted to GORs of less than about 4–6 [4], [13]. In contrast, large scale thermal desalination systems such as multi-stage flash and multi-effect distillation, which may have significantly more complicated designs, often have GOR higher than about 7–10 [14]. Increasing the GOR of MD close to these values would make MD more competitive with these conventional technologies.

The GOR of single stage MD configurations has been previously analyzed and compared by Summers et al. [5]. While single stage VMD systems are restricted to GOR below 1, single stage SGMD also does not achieve GOR beyond about 4 [9]. In contrast, DCMD and AGMD are relatively simple systems, and were shown to be capable of achieving higher GORs [5], [15], [16], [17].

Permeate gap MD is also referred to as water gap and liquid gap MD. Simply, it can be understood as a modification of AGMD where the gap between the membrane and the condensing surface is filled with permeate water (Fig. 1a). PGMD has shown improved fluxes compared to AGMD [8], [18], [19] as have other modifications such as material gap MD systems with sand added to the gap [3]. A clear comparison of GOR between AGMD and PGMD has not been established, though commercial spiral wound PGMD modules have achieved higher GOR . Winter et al. [20] suggested PGMD as a modification of DCMD with internal heat recovery by separation of the distillate from the coolant. Therefore the coolant can be any other liquid, such as incoming feed water. They note that “The presence of the distillate channel reduces sensible heat losses due to an additional heat transfer resistance. An additional effect is the reduction of the effective temperature difference across the membrane, which slightly lowers the permeation rate.” PGMD can therefore be intuitively placed between AGMD and DCMD with intermediate thermal resistance of the gap, thereby having lower sensible heat transfer to the permeate as compared to DCMD, but perhaps more heat loss than AGMD. A hollow fiber MD system in PGMD configuration was experimentally analyzed by Singh et al. [21], where the condensate from one AGMD module was passed into another module. Singh et al. note that the second module would perform between AGMD and DCMD when the gaps are filled with condensate. Hollow fiber systems with high packing density may also lead to much lower effective gap sizes [22] and sections of the module may have pure water extending across the narrow gaps between membrane and non-porous fibers. While the focus of the present work is on modeling flat sheet and similar spiral-wound MD systems, the overall trends are applicable to hollow fiber MD systems as well.

Conductive gap MD (CGMD) is suggested as a novel MD configuration (Fig. 1b). In CGMD, the overall thermal conductance of the gap is increased. One way of achieving this is to insert a high conductivity material (such as a metal mesh) into the gap of a PGMD system. Increasing the conductivity of the gap in hollow fiber MD systems by inserting high conductivity materials was suggested by Ma et al. [23] in 2010. This is in contrast to other systems proposed in the literature such as material gap MD where low conductivity materials such as sand were added to the gap region [3]. This also contradicts the conventional wisdom and the historical development in the field towards developing MD configurations with lower sensible heat loss than DCMD [24], such as the evolution from DCMD to AGMD or PGMD. Possible implementations involve using a metal spacer instead of a plastic spacer, or implementing fins on the conductive surface extending up to the membrane to increase the net conductivity of the gap. The gap is filled with pure water, similar to the PGMD system. Vapor would condense immediately as it leaves the membrane. The energy is conducted through the gap into the cold stream, preheating it. While this configuration may have a higher sensible heat loss than even PGMD, perhaps close to that of a DCMD system, for a given membrane area and feed flow rate, it rejects brine at a lower temperature and correspondingly achieves higher preheating of the cold stream. This could therefore result in a higher overall GOR.

In this study, we use numerical modeling to investigate the energy efficiency of PGMD and CGMD systems. The results are compared to previously published results for other conventional MD configurations under the same operating conditions. The effect of pure water flow direction in the gap is evaluated. The effect of gap conductivity and membrane material conductivity on GOR is studied.

Section snippets

Review

Several MD models with varying degrees of complexity have been developed to understand the effect of system parameters on flux [25], [26]. Fewer, however, have analyzed energy efficiency [6], [27], [28]. The modeling approach followed in this paper is very similar to that found in Summers et al. [5]. As a result, only the new features of the modeling are discussed in detail.

A one-dimensional model of the MD modules is studied, where properties vary along the length of the module, but are

Validation

The numerical modeling framework presented above has been validated for AGMD and DCMD in the past [5]. PGMD experiments were carried out using the AGMD apparatus described in detail elsewhere [18], [30]. The apparatus was used to study PGMD by collecting water from the top, hydrostatically forcing the gap region to be flooded with pure water. Experiments were conducted at different values of Tf,in(40,50,60,70°C) and Tc,in(17,20,25°C). For the numerical model predictions, an effective gap

GOR comparison, effect of gap flow

The configuration of pure water flow in the gap can affect the performance of a PGMD or CGMD system. Fig. 4 shows the different options for pure water flow in the gap. On the left, pure water in the gap flows countercurrent to the coolant fluid across the condensing plate. The opposite gap configuration is to have the pure water flow parallel to the coolant stream, as seen on the right. An intermediate design (middle) may be that of perpendicular or crossflow, where water flows in a direction

Conclusions

  • 1.

    Numerical modeling shows that PGMD systems have higher GOR than AGMD. The proposed CGMD configuration with a high thermal conductivity gap has two times higher GOR than even PGMD.

  • 2.

    Pure water flow in the gap countercurrent to the cold stream leads to highest energy efficiency followed in order of efficiency by crossflow and parallel configurations.

  • 3.

    An increase in gap conductivity improves permeate production and GOR, with diminishing returns beyond k10W/mK in the cases considered here.

  • 4.

    The main

Acknowledgments

This work was funded by the Cooperative Agreement Between the Masdar Institute of Science and Technology (Masdar Institute), Abu Dhabi, UAE and the Massachusetts Institute of Technology (MIT), Cambridge, MA, USA, Reference no. 02/MI/MI/CP/11/07633/GEN/G/00, and facilitated by the MIT Deshpande Center for Technological Innovation and the Masdar Institute Center for Innovation and Entrepreneurship (iInnovation).

The authors would like to thank MIT undergraduate students Ann M. Huston and Grace

References (32)

Cited by (132)

View all citing articles on Scopus
View full text