Woods2014 Membrane Processes For Heating, Ventilation, and Air Conditioning

Renewable and Sustainable Energy Reviews 33 (2014) 290–304
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser
Membrane processes for heating, ventilation, and air conditioning

Jason Woods n
National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO 80401, USA
art ic l e i nf o a b s t r a c t
Article history: This article reviews literature on using membranes in heating, ventilation, and air conditioning (HVAC)
Received 11 September 2013 applications. Membranes enable the separation of one species from another, and membranes allowing
Received in revised form the selective permeation of water vapor can be used to condition air in buildings, potentially more
22 January 2014
efficiently than conventional HVAC equipment. After a brief background on membrane technology, this
Accepted 31 January 2014
Available online 28 February 2014
review focuses on the following processes: vacuum membrane dehumidification; membrane energy
recovery ventilation; liquid desiccant dehumidification; liquid desiccant regeneration; evaporative
Keywords: cooling; and humidification. It highlights the design, modeling, and experimental research on these
HVAC topics, and suggests areas for further research.
Buildings
& 2014 Published by Elsevier Ltd.
Membrane
Air conditioning
Liquid desiccant
Energy recovery ventilator
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
2. Membrane technology background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
2.1. Membrane type: dense and porous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
2.1.1. Dense membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
2.1.2. Porous membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
2.2. Membrane form: flat sheets and hollow fibers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3. HVAC applications of membranes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3.1. Vacuum membrane dehumidification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3.1.1. Module design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
3.1.2. System design and analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
3.2. Membrane energy recovery ventilators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
3.2.1. Module design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
3.3. Liquid desiccant dehumidification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
3.3.1. Module design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
3.4. Membrane evaporative cooling and humidification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
3.4.1. Module design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
3.5. Other membrane contactors for HVAC applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
4. Modeling approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
4.1. Transport through the membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
4.1.1. Mass transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
4.1.2. Heat transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
4.2. Transport through the gas phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
4.3. Transport through the liquid phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
n
Tel.: þ 1 303 384 6189.
E-mail address: [email protected]
http://dx.doi.org/10.1016/j.rser.2014.01.092
1364-0321 & 2014 Published by Elsevier Ltd.
J. Woods / Renewable and Sustainable Energy Reviews 33 (2014) 290–304 291
5. Research needs and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
1. Introduction Then, it describes the HVAC applications, mostly for cooling and
dehumidification, and assesses research on designing, modeling,
The temperature and humidity of the air in buildings needs to and testing these membrane processes and devices. The review
be controlled to maintain human comfort, prevent mold growth, concludes with future research needs and directions.
and ensure building durability. This requires the addition and
removal of sensible and latent energy, meaning that the air in
buildings requires heating, cooling, dehumidifying, and humidify-
2. Membrane technology background
ing to varying degrees depending on the location of the building
and how the building is used. Research in the heating, ventilation,
A membrane is a selective barrier between two phases and is
and air conditioning (HVAC) field has improved the methods for
typically used to separate one species from another. Membranes
conditioning this air, but it is typically done in the same manner as
can be made of many materials, but polymers are most common.
it was decades ago: heating with direct combustion (usually
Fig. 1 shows a generalized schematic of a membrane process. The
natural gas in the United States), vapor compression heat pumps,
feed stream is supplied to one side of the membrane, the permeate
or electric resistance heaters; and cooling and dehumidifying with
permeates across the membrane, and the retentate is retained on
vapor compression air conditioners.
the feed side of the membrane. In some cases, a sweep stream
There are some concerns about these current practices; con-
helps carry away the permeate.
cerns that must be addressed in any sustainable energy future. In
The permeate is transferred across the membrane because
the United States, conditioning air accounts for 48% of the primary
of a chemical potential gradient, which can be from gradients in
energy used in buildings [1]. Cooling and dehumidification, mostly
pressure, concentration, temperature, or electric potential. The first
from vapor compression systems, accounts for a significant portion
two are most important for the processes discussed in this review.
of the peak electric demand in hot climates. Regulations are also
The key function of a membrane is selective separation. Mem-
phasing out refrigerants (CFCs, HCFCs) because they damage the
branes are characterized by their permeability and their selectivity.
ozone layer or contribute to climate change. Controlling humidity
The permeability is the amount of a species crossing the membrane,
with vapor compression systems is also becoming more difficult as
per unit area and per unit driving force. The selectivity is the amount
energy improvements often reduce the building's sensible load
of the more permeable species crossing the membrane relative to
(e.g., from improved insulation), but do not affect the latent load
others. A higher permeability means less membrane area for a given
(e.g., from required ventilation or internal gains) [2].
transfer rate, and a higher selectivity means a purer product stream,
Researchers are pursuing alternatives to these conventional
which can be either the permeate or the retentate.
practices, especially for cooling and dehumidification [3]. Advance-
For HVAC processes, water vapor is usually the permeate, and
ments in artificial membranes enable new possibilities in this area.
thus the membranes need to be permeable to water vapor and
While traditionally used for industrial separations, such as reverse
selective to water vapor over other species. We can achieve these
osmosis and gas separation [4–6], membranes provide a means to
characteristics with two membrane types: dense membranes and
selectively transfer water vapor from one fluid to another.
porous membranes. We can also distinguish membranes by their
The coupling between latent and sensible energy enables a
form: flat sheet or hollow fiber.
variety of potentially energy-efficient HVAC processes. While the
membrane itself does not save energy, it can enable or improve
processes that do. Membranes provide a means to remove moisture
2.1. Membrane type: dense and porous
from the air without cooling the air to the dew-point temperature.
This could mean the elimination of environmentally harmful
Membranes are typically separated into dense and porous
refrigerants from cooling systems. They also make energy recovery
categories based on their pore structure. Dense membranes have
processes possible, where moisture is exchanged between two
pores on the order of 0.1 nm and porous membranes have pores
separate airstreams. Membranes can also improve absorptive and
on the order of 0.1 μm. In dense membranes, water vapor adsorbs
evaporative processes, which are used in technologies like absorp-
onto the polymer and diffuses through the polymer matrix on the
tion chillers, liquid desiccant dehumidifiers, and evaporative cool-
molecular level. In porous membranes, the water vapor diffuses
ers; technologies that are energy-efficient, but have yet to reach
through the air/vapor mixture within the pore space. There is
their market potential.
some transition region between these two where both mechan-
Over the past 15–20 years, researchers have explored these uses
isms are important [8]. However, the two can be separated in this
of membranes in HVAC processes. Interest in this area remains high,
review because the pores used in the membrane processes of
as illustrated with recent research grants from the US Department
interest are not near this transition region.
of Energy through their Advanced Research Projects Agency focused
on Energy (ARPA-E) [7], which funds researchers working on high-
risk, innovative energy ideas. In 2010, 16 projects received funding
through an HVAC-focused program (Building Energy Efficiency “feed side”
retentate
feed
Through Innovative Thermodevices). Membranes were a key focus
in six of these 16 projects. permeate
sweep
The purpose of this review is to summarize the literature on “permeate side”
membrane HVAC processes. It begins with background on mem- permeate + sweep
brane technology, which is not meant to be comprehensive, but Fig. 1. Generic membrane process. Useful product stream can be retentate or
instead to highlight topics and literature relevant to this article. permeate. Sweep is optional.
292 J. Woods / Renewable and Sustainable Energy Reviews 33 (2014) 290–304
Nomenclature P permeability (kmol Pa 1 m 1 s 1)

q heat flux (W m 2)
A area available for heat and/or mass transfer (m2) R universal gas constant (8314 J kmol 1 K 1)
D diffusion coefficient for water vapor through dense RHT heat transfer resistance (m2 K W 1)
membrane (m2 s 1) RMT mass transfer resistance (Pa m2 s kmol 1)
Dwa diffusion coefficient for water vapor in air (m2 s 1) S water vapor solubility in membrane (kmol m 3 Pa 1)
DK Knudsen diffusion coefficient (m2 s 1) T temperature (1C)
DM effective molecular diffusion coefficient (m2 s 1) t membrane thickness (m)
dp mean pore size of membrane (m) yair mole fraction of air
Jv molar flux (kmol m 2 s 1) ε membrane porosity
Mw molar mass of water (18 kg kmol 1) τ membrane tortuosity
pv vapor pressure (Pa)
2.1.1. Dense membranes Composite membranes deserve some further explanation, as

Dense membranes (e.g., see [6]) are polymer films where the these are fairly common and significantly improve permeability.
intrinsic properties of the membrane material, specifically a Permeability is inversely proportional to the membrane thickness,
kinetic property (diffusivity) and a thermodynamic property so making very thin membranes is one way of creating highly
(solubility), determine the permeation rate. A molecule will adsorb permeable membranes. Composite membranes are created by
onto the polymer surface, diffuse through the bulk polymer due to combining a thin dense layer, on the order of 1 μm thick, to a
a concentration gradient, and then desorb from the other surface highly porous support structure, on the order of 100 μm thick. The
(Fig. 2). These membranes are used in gas separation processes, support layer provides mechanical strength, but adds little resis-
such as separating O2 from N2, CO2 from combustion gases, or tance to vapor transfer compared to the dense layer.
water vapor from air. Polyimide membranes are an example. They
have a high permeability to CO2 because of the physical and
chemical interactions between CO2 and polyimide. Polyimide 2.1.2. Porous membranes
interacts differently with methane (CH4), resulting in a lower Porous membranes have more open volume and larger pores
permeability. This enables polyimide membranes to separate CO2 than dense membranes. Researchers developing prototypes for
from CH4, with a CO2/CH4 selectivity on the order of 10–100. HVAC devices used porous membranes with pore sizes ranging
Membranes for transferring water vapor are hydrophilic, mean- from 0.03 to 1 μm, and porosities ranging from 40% to 70% [24–28].
ing they have a strong affinity to water. Water is very soluble in The porosity, which is between 0 and 1, is the volume of the pores
these membranes, resulting in high permeability and a water/air (i.e., open space) divided by the bulk membrane volume.
(or H2O/N2) selectivity of up to 100,000 [9–11]. This means that Porous membranes are used to contact a liquid with a gas, as
100,000 molecules of water pass through the membrane for every opposed to the gas/gas contact where dense membranes are used.
single molecule of nitrogen, assuming equal driving potentials. These liquid/air modules are a subset of membrane contactors,
Higher permeability means less area is needed for a given process, which have been reviewed by Drioli et al. [29]. Similar membranes
and high selectivity allows for removal of water vapor from an are also used in membrane distillation and osmotic distillation
airstream without removing much air. Increasing the water vapor [30–32], where volatile liquids evaporate from one liquid stream
permeability and the water/air selectivity of these membranes is and condense into another liquid, selectively separating them from
the focus of many researchers, who have looked at new polymer less volatile liquids and solids.
materials [9,12,13], ceramic materials [14], liquid membranes The separation mechanism for porous membranes differs from
[15–19], and composite membranes [6,20–22]. A 2009 ARPA-E dense membranes. The pores are larger so that water vapor
grant funded ADMA Products, Inc. to develop a coated metal diffuses through the pore space, instead of through the solid
membrane specifically for air conditioning applications [23]. polymer matrix. Unlike dense membranes, the membrane material
is hydrophobic (e.g., polypropylene); the water molecules are more
attracted to each other than they are to the solid material. This
dense membrane selective to water vapor
produces surface tension forces that keep the liquid from entering
the pores and creates a liquid/gas interface at the membrane
vapor transfer surface (Fig. 3). The required entry or breakthrough pressure, which
is the pressure where liquids enter the pores, is proportional to the
water/material surface tension and inversely proportional to the
pore size. It is typically on the order of 10–100 psi, but it is lowered
substantially if surfactants contaminate the liquid [30].
N2 or O2 Permeability can be increased by making the membranes
thinner, with larger pores, and with a higher porosity (see Eqs.
H2O
(8)–(10) in Section 4.1.1); however, there are trade-offs. Larger
pores increase the likelihood of pore breakthrough and liquid
entry, and thin membranes can be less durable. Membranes as
thin as 20 μm have been used [33], but there are questions about
their longevity. One possibility is to use asymmetric membranes,
where a relatively thick, highly porous layer is used to provide
structural support, and a thin layer with small pores (1–10 nm)
ensures no liquid entry. These can be made in different ways.
Fig. 2. Membrane–fluid interface for dense membrane. A thin porous membrane can be attached to a highly porous
substrate [26], a dense coating can be applied to a porous pressure is created with a compressor, or vacuum pump. The H2O/N2
membrane [34,35], or a membrane with an asymmetric structure selectivity of the membrane is important, so that the compressor
can be made in one step [36,37]. pulls primarily water vapor through the membrane.
A similar process, referred to as membrane drying, is used for
2.2. Membrane form: flat sheets and hollow fibers removing moisture from a variety of gas streams. Wang et al. [20]
discussed drying of compressed air, where the compressed air is
In addition to the microscopic structure of the membrane, we delivered to the lumen-side of hollow fibers, and ambient-pressure
can also distinguish between two macroscopic structures, or
forms: flat-sheet membranes and hollow-fiber membranes. Flat- header
sheet membranes are large sheets, usually on the order of 100 μm Fluid 1
thick. They are used to construct modules similar to plate-and-
frame heat exchangers (Fig. 4a).
Hollow fibers are tubes, typically around 500 μm in diameter,
where the wall of the tube is the membrane. They are used to
construct modules similar to shell-and-tube heat exchangers
(Fig. 4b). In these modules, one fluid flows inside the tubes (the
membrane
lumen side) and the other flows around the tubes (the shell side).
Both forms can be used in membrane HVAC devices. The
reasons for using one over the other are specific to the application,
so they are discussed in the next section.
Fluid 2
3. HVAC applications of membranes
This review focuses on five common membrane HVAC pro- Fluid 2

cesses. Table 1 lists the inlet and outlet streams, as they are labeled (shell side)
in Fig. 1, and also shows which processes require dense, hydro-
philic membranes, and which require porous, hydrophobic mem-
branes. Sections 3.1–3.4 describe these processes in more detail
(liquid desiccant regeneration and dehumidification are discussed Fluid 1
in the same section). Some less common processes are described (lumen side)
briefly in Section 3.5.
header Fluid 1
3.1. Vacuum membrane dehumidification membrane
Fluid 2
Vacuum membrane dehumidification uses a pressure gradient to
transfer vapor across a dense, hydrophilic membrane (Fig. 5). Water
vapor adsorbs into the membrane, diffuses through the membrane,
and desorbs from the membrane on the low-pressure side. This low
membrane material Cross section view

vapor transfer
Fig. 4. Schematic of (a) flat-sheet module and (b) hollow-fiber module.
liquid air/vapor mixture dry air

humid air (to conditioned space)
water vapor transfer
to ambient
liquid-gas interface water vapor
Fig. 3. Membrane–fluid interface for porous membrane. Pores are generally compressor
tortuous and interconnected, but they are shown as straight-through pores here
for simplicity. Fig. 5. Schematic of membrane vacuum drying.
Table 1
Summary of membrane HVAC applications. Note that evaporative cooling and humidification are combined since their characteristics are the same.
Feed Retentate Sweep inlet Sweep/permeate outlet Membrane type
a
Vacuum membrane dehumidification Humid air Dry air N/A Water vapor Dense, hydrophilic
Membrane energy recovery ventilator Hot, humid air Dry, cool aira Dry, cool air Hot, humid aira Dense, hydrophilic
Liquid desiccant air conditioning Warm, humid air Hot, dry aira Concentrated LD Diluted LD Porous, hydrophobic
Liquid desiccant regeneration Diluted LD Concentrated LDa Ambient air Humid air Porous, hydrophobic
Evaporative cooling/humidification Water Water Hot, dry air Cool, humid aira Porous, hydrophobic
a
Useful product stream or streams.
gas is delivered to the shell side, creating a pressure gradient minimize the energy required to create the pressure gradient
across the membrane. Compressing the air flow rates used in across the membrane.
space conditioning would be impractical and energy inefficient, While this process looks promising on paper, there are some
but using a compressor on the permeate side is an alternative way practical limitations. To illustrate this, consider the example
to apply a pressure gradient [38]. This means the compressor only shown in Fig. 6, which roughly corresponds to a 10-kW residential
needs to compress the water vapor (and the very small amount of air conditioner. The inlet and outlet humidity is indicated with the
air) that crosses the membrane. In 2000, El-Dessouky et al. [39] vapor pressure, pv, which correspond to typical inlet and supply
proposed this process for HVAC applications. They modeled their dew point temperatures of 16 1C and 10 1C, respectively. The
system with a membrane having a H2O/N2 selectivity of 400, but required water vapor flow rate (0.14 kg/min) is calculated based
as discussed in Section 2.1.1, a selectivity over 100,000 is possible. on this change in humidity and the assumed airstream flow rate.
To calculate the efficiency of this process, we use the same
dehumidification efficiency equation used by Scovazzo et al. [40]
3.1.1. Module design (see also [44]). Since the desired vapor pressure of water in the
A vacuum membrane dehumidifier is made by assembling exiting airstream is 1.25 kPa, the compressor must reduce the
highly permeable, highly selective membranes into a module. pressure to below 1.25 kPa. If the water vapor is discharged from
These can be flat-sheet or hollow-fiber membranes. El-Dessouky the compressor to ambient air, as it is in El-Dessouky [39], the
et al. [39] modeled a hollow-fiber module, but they did not build a compression ratio is at least 80 (101 kPa/1.25 kPa). To minimize
prototype, and although Scovazzo et al. [40] suggest using a the membrane module size, the membrane requires a finite
hollow-fiber module for this process, they built a small (10 cm2) driving force across it. Thus, the pressure will more likely be near
flat-sheet prototype for their experiments. Hollow fibers are more 1 kPa and the compression ratio 100. Using a compression ratio of
likely in commercial applications because their structure enables 100 and assuming a compressor efficiency of 60%, the calculated
them to withstand the large pressure differences between the dehumidification efficiency is 200%. This means that for every one
airstream ( 100 kPa) and compressor inlet ( 1 kPa). unit of electrical energy input, there are two units of latent energy
Supplying the air to the lumen side of the hollow fibers, as removed from the airstream.
done by Wang et al. [20], can lead to large pressure losses [40]. Regardless of the efficiency, this process requires large, com-
Supplying the air to the shell side instead has less pressure loss, plex compressors that are not suitable for building air condi-
but it can potentially result in poor flow distribution [41–43]. tioners. The flow rate of 0.14 kg/min is around 5.4 m3/min, or
Some researchers have suggested modules where the air flows 200 cfm, at that pressure. Pulling this much volume through the
around a structured arrangement of fibers on the shell side compressor at a compression ratio of 100 will require a large
[35,40], making distribution more uniform. compressor with at least two stages of compression [44]. It will
In addition to ensuring good distribution, the module should also need significant cooling (adiabatic compression will lead to
also minimize the mass transfer resistance in the gas boundary temperatures of several hundred degrees Celsius). This will be
layers next to the membrane. The concentration of water vapor at impractical for most air conditioning applications. But some
the membrane surface is not the same as the concentration in the researchers are modifying the process to try to address these
bulk flow. This concentration polarization affects performance, as issues.
discussed by several researchers [9,10,20]. It means that the actual Scovazzo et al. [40] attempt to solve these issues by using some
permeability is less than that of the membrane alone. Since of the air that has already been dehumidified as a sweep gas
concentration polarization increases as the permeation rate (Fig. 7), which is a common method for membrane drying of
increases, this effect is more important for highly permeable compressed air [20,38]. The air exiting the feed side is at ambient
membranes [10]. pressure with a water vapor pressure of 1.25 kPa. When the
The increase in concentration polarization as the permeation
rate increases also reduces the selectivity. The low transfer rate of air/vapor mixture
air leads to nearly negligible concentration polarization for air, air/vapor mixture (to conditioned space)
while the high transfer rate of water vapor results in high m = 41 kg/min m = 41 kg/min
concentration polarization. In other words, the largest resistance p = 1.8 kPa p = 1.25 kPa
to air transfer is the membrane, while the largest resistance to
water vapor transfer is the gas film next to the membrane. Data water vapor
from Wang et al. [20] illustrate this. They calculate the selectivity m = 0.14 kg/min p = 101 kPa
p < 1.25 kPa
of the membrane to be 30,000, but the actual measured selectivity
(including boundary layers) ranged from 30 for low airstream flow compressor
rates to 300 for higher flow rates. These drastic differences show Fig. 6. Illustrative case considered for vacuum membrane dehumidifier: m¼ mass
the importance of including the boundary layer resistances in flow rate, p¼ pressure, pv ¼vapor pressure. Flow rates, pressures, and vapor pressure
modeling this process. values are illustrative.
3.1.2. System design and analysis water vapor transfer

air/vapor mixture
(to conditioned space)
Researchers have designed, built, and tested many membrane air/vapor mixture
m = 31 kg/min
modules for gas separation and membrane drying, but most of it m = 41 kg/min
p = 1.25 kPa
p = 1.8 kPa
has focused on industrial separation processes as opposed to
comfort cooling for buildings. Thus, there is little research on air/vapor
mixture
system design for comfort cooling. The two metrics to consider are p = 101 kPa
m = 10 kg/min throttling valve
the energy efficiency of the process (i.e., operating costs) and the p ~ 20 kPa
size of the compressor and membrane module (i.e., initial costs). p < 1.25 kPa compressor
This process has the potential to be much more efficient than a Fig. 7. Schematic example of alternative vacuum membrane dehumidifier inves-
vapor-compression dehumidifier, which condenses water vapor tigated by Scovazzo et al. [40]. Flow rates, pressures, and vapor pressure values are
out of the air by cooling the air below the dew point. The key is to illustrative.
pressure is reduced through the throttling valve, the vapor pressure received funding from ARPA-E in 2009 to advance their concept
is also reduced (vapor pressure¼ pressure concentration). Thus, further [23], but there is little information about this technology
achieving a vapor pressure below 1.25 kPa is possible at much higher available in the open literature.
absolute pressures, reducing the compression ratio from 80 to around The Dais concept also includes a membrane evaporative cooler,
5–10. This eases the requirements for the compressor, and can also a concept which is explained in Section 3.4. The low relative
improve the compressor efficiency. It does not necessarily improve the humidity (RH) exiting the vacuum dehumidifier enables this
process efficiency because some of the dehumidified air is being evaporative cooling. This is not possible with vapor compression
discarded with the sweep and thus is not provided to the space. systems, which supply air near saturation. Combining these two
Another alternative design, investigated by Dais Analytic devices offers a way to decouple the latent and sensible loads, with
Corporation [45], is to pump the vapor to a lower vapor-pressure the vacuum dehumidifier providing latent cooling and the eva-
sink instead of the ambient pressure of 101 kPa. In this concept, porative cooler providing sensible cooling. El-Dossouky et al. [39]
the compressor discharges the water vapor into a second modeled a (non-membrane) indirect evaporative cooler at the exit
membrane module, which puts the low-pressure vapor in contact, of the membrane dehumidifier and illustrated the energy savings
through a membrane, with ambient air (Fig. 8). Thus, the com- potential of this concept.
pressor only needs to pump up to the ambient vapor pressure,
instead of the ambient absolute pressure. 3.2. Membrane energy recovery ventilators
The compression ratio in this concept, and therefore the dehu-
midification efficiency, depend on the ambient humidity (i.e., dew- Membrane-based energy recovery ventilators (ERVs) use dense
point temperature). The compression ratio is simply the ratio of the membranes similar to those used in vacuum membrane dehu-
pressures between the two membrane exchangers (3.5/1.25 in the midification, but the operating principle is different (Fig. 10). The
illustrative example in Fig. 8), and, as before, the dehumidification compressor is replaced by a low-vapor-pressure sweep stream, and
efficiency can be calculated based on the equation from Scovazzo the absolute pressure gradient across the membrane is replaced
[40]. The result is shown in Fig. 9 assuming a 0.25 kPa driving by a vapor pressure difference. Like heat exchangers, these devices
potential in each membrane exchanger, inlet and outlet conditions transfer sensible energy between the two airstreams, but they also
shown in Fig. 8, and a compressor with an efficiency of 60%. Note transfer latent energy via water-vapor diffusion. They are used to
that this does not include the electrical energy for the fans required exchange sensible and latent energy between a building's exhaust
to push the air through either of the membrane exchangers. Dais air and its incoming ventilation air. Thus, it supplies lower energy
ventilation air in the summer and higher energy ventilation air in
air/vapor mixture
the winter, reducing the required space conditioning.
air/vapor mixture (to conditioned space) Researchers converted sensible heat exchangers into these ERV
m = 41 kg/min m = 41 kg/min devices more than 30 years ago, by replacing the thermally-
pv = 1.8 kPa pv = 1.25 kPa
conductive heat exchange surfaces with vapor-permeable paper
sheets [11,46]. More recently, Zhang and Jiang [47] replaced these
water vapor paper sheets with dense, hydrophilic polymer membranes. They
p = 3.5 kPa
m = 0.14 kg/min modeled and measured the effectiveness of a cross-flow mem-
p < 1.25 kPa
brane ERV and compared it to the effectiveness of a sensible-only
compressor metal heat recovery ventilator (HRV) and a paper ERV. This
effectiveness can be split into sensible, latent, and total, which
air/vapor mixture air/vapor mixture are often used to characterize ERVs. Assuming equal flow rates,
(from outside) (back to outside) these are calculated with [48]
m = 41 kg/min pv ~ 3.5 kPa
pv = 2.5 kPa T 1;out T 1;in
εSENS ¼ ð1Þ
T 2;in T 1;in
Fig. 8. Schematic example of alternative vacuum membrane dehumidifier inves-
tigated by Dais Analytic [45]. Flow rates, pressures, and vapor pressure values are
ω1;out ω1;in
illustrative. εLAT ¼ ð2Þ
ω2;in ω1;in
21 3.5
Theoretical compressor dehumidification COP
h1;out h1;in
εTOT ¼ ð3Þ
18 3
h2;in h1;in
where T, ω, and h are respectively the air temperature, humidity

15 2.5 ratio, and enthalpy, the subscripts 1 and 2 indicate the two
Compression ratio
different airstreams.
12 2 Zhang and Jiang [47] showed that a membrane provides
roughly two times more latent recovery than paper under hot,
9 1.5 humid conditions. Since the HRV recovers only sensible energy,
the latent effectiveness of the HRV is zero. Because much of the
6 1 energy during this hot, humid condition is latent, the total energy
3 0.5
hot, humid air cool, dry air
ventilation airstream
0 0
10 12 14 16 18 20 22 24 26 hot, humid air cool, dry air
exhaust airstream
Ambient dew-point temperature (°C) water vapor transfer &
sensible heat transfer
Fig. 9. Theoretical compression ratio and dehumidification efficiency (COP ¼
coefficient of performance) versus ambient dew-point temperature for the design Fig. 10. Operating principle of membrane ERV in summer. In the winter, exhaust
shown in Fig. 8. and ventilation airstreams are switched.
transfer of the membrane ERV was 4.8 times higher than the HRV diffusivity of water vapor in air are roughly equal, the airside
and 1.7 times higher than the paper ERV. For cold conditions, the resistances for mass transfer and heat transfer should be roughly
sensible energy recovery generally dominates over the latent equal. Zhang and Jiang [47] showed that a membrane ERV
recovery; thus, the benefits for cold climates is smaller. Zhang provides roughly the same sensible energy transfer as a metal
et al. estimates that the membrane ERV has 1.3 times higher HRV, even though the membrane thermal conductivity is orders of
energy transfer than the HRV in Beijing (a cold climate) [49]. magnitude lower than that of the metal plates. This indicates that
Similar conclusions can be drawn from other studies looking at most of the heat transfer resistance must be coming from the air
different climates [50–54]. boundary layers. Indeed, Kistler and Cussler [51] estimated that
Over the last decade, several manufacturers have developed the air boundary layer accounts for more than 95% of the overall
membrane ERV products. These compete against other energy heat transfer resistance.
recovery technologies, as discussed in a recent review by The importance of the airside boundary layers for mass transfer
Mardiana-Idayu and Riffat [55]. Fig. 11 plots data from a standard is estimated to be much smaller; Zhang [57] calculated that they
test performed by the Air-Conditioning, Heating, and Refrigeration account for 25% of the total resistance, while Min and Su [58]
Institute (AHRI) on several commercially available membrane ERVs estimated that they account for 10–35% of the total. This means
from about a dozen manufacturers. The data shows that the latent the membrane mass transfer resistance is not negligible, and this
effectiveness is generally less than the sensible effectiveness. Some is why the latent effectiveness in Fig. 11 is always lower than the
of the reasons for this are discussed in the next section. sensible effectiveness. It is also why some points in Fig. 11 give
three times more latent transfer than others; the points with
higher latent transfer use more permeable membranes. This also
3.2.1. Module design means that, for some membranes, the airside resistances are not as
The dense, hydrophilic membranes used for membrane ERVs important. Using the membrane permeability from [59], the air-
typically have high water permeability and high water/air selec- side resistance becomes roughly 75% of the total. These modules
tivity, like the membranes for vacuum dehumidifiers. But the with high performing membranes are near the 451 line in Fig. 11.
absolute pressure difference across the membrane is orders of They provide nearly as much latent recovery as sensible, which
magnitude smaller, making the construction requirements less shows that in some modules the airside resistances for mass
stringent. This low pressure difference also means that it is not the transfer are becoming important.
water/air selectivity that is important, but rather the selectivity of These airside boundary layers are a function of the module
water vapor over various pollutants (e.g., CO, CO2, volatile organic construction. The modules can be made with either hollow-fiber
compounds (VOCs)) [56]. Unlike air, the concentration differences or flat-sheet membranes. Flat sheets are more common, and to the
of pollutants across the membrane can be large, and the mem- author's knowledge, all commercially available units use flat-sheet
brane must be selective to water vapor over these pollutants. membranes. One disadvantage of flat sheets is that they are not
Measurements have shown that selectivity of water over self supporting, so that a spacer must be used to reinforce the
pollutants is generally smaller than the H2O/N2 selectivity. Wang membrane. As several researchers have noted [60–62], an unsup-
et al. [20] measured the permeability of a particular dense ported membrane can deflect or bulge into the airstreams, which
membrane to H2O, CO2, and N2, and calculated an H2O/CO2 affects performance by reducing the effectiveness of the exchanger
selectivity that was 15 times smaller than the H2O/N2 selectivity. [60,63]. Larson et al. [60] modeled and measured mass flow rates
Zhang et al. [56] measured the permeability of water vapor and in the air channels for different amounts of membrane deflection.
several VOCs through membranes made of different materials. They recommended pre-tightening the membranes, which will
They found selectivity values in the range of 100–400. While these reduce their deflections. Even with pre-tightening, the membranes
are still adequate for removing most pollutants from buildings, it may still require a support spacer to maintain uniform channel
may be a poor assumption to assume no pollutants cross the thicknesses.
membrane along with the moisture. Regardless of whether or not spacers are needed, they can be
Like vacuum dehumidification, concentration polarization is useful for enhancing heat and mass transfer in the airside boundary
present in ERVs, and the module should be designed to minimize layer. However, they will also increase the pressure drop in the
these airside resistances. A comparison between a membrane channel. A few studies have focused on understanding this tradeoff
ERV and a sensible-only HRV illustrates the importance of these between heat/mass transport and pressure drop [57,62,64–66].
airside resistances. Since the air thermal diffusivity and the mass More extensive research has analyzed a similar tradeoff in liquid-
to-liquid membrane processes [67–71].
1 While flat-sheet modules are more common, hollow-fiber
modules can also be used as membrane ERVs. Kistler and Cussler
[51] looked at both types, but they were interested in hollow fibers
0.8
because of their ease of construction. One disadvantage of hollow
Latent effectiveness
fibers is their large pressure drop. Zhang [72] measured a pressure

0.6 drop for their 1.2-mm i.d. fiber lumens of around 350 Pa. The
pressure drop in the module used by Kistler and Cussler [51],
which had 0.6-mm i.d. fibers, were likely much higher, but they
0.4
were not reported. Pressure drops above 350 Pa are unlikely to be
used in these devices because fan initial cost generally increases
0.2
with the supplied pressure, and because of increased parasitic
energy use. For contrast, Mardiana-Idayu and Riffat [73] reported
pressure drops for their flat-sheet modules of 30 Pa, and pressure
0 drops for most commercially available units range from 100 Pa to
0 0.2 0.4 0.6 0.8 1
300 Pa.
Sensible effectiveness One final consideration in module design is the orientation of
Fig. 11. Latent and sensible effectiveness of membrane ERV units tested according the airstreams. Most are crossflow, due to easier construction, but
to AHRI standards [140]. some researchers have also looked at designing flat-sheet modules
to be at least partially counterflow [53,74], which offers better vapor pressure gradients set up by a low-activity aqueous solution
performance. (i.e., liquid desiccant), rather than the pressure gradients from a
compressor. Fig. 12 is a schematic of a liquid desiccant membrane
3.2.2. System design and analysis dehumidifier, which consists of two membrane exchangers: a
The energy savings of an ERV are directly proportional to the conditioner and a regenerator. In the conditioner, moisture passes
sensible and latent effectiveness. They are also impacted by the from the air through a porous, hydrophobic membrane and into
parasitic energy use by the fans that are needed to move the air. the desiccant. The desiccant is then heated, so that in the
So from an energy perspective, the optimal design will have high regenerator, water evaporates out of the desiccant and into a
effectiveness and low pressure drop. separate airstream. This concentrates the desiccant so it can re-
The effectiveness of membrane ERVs depends on the flow rate absorb moisture in the conditioner.
per membrane area. Prototypes and commercially available mod- Using a membrane offers several benefits, primarily because
ules are typically sized to give sensible and latent effectiveness the liquid desiccant is contained and separated from the air-
values of around 60–80% [59,61,62,75,76], although as mentioned streams. Many of these dehumidifiers spray a high flow rate of
before, the latent effectiveness is typically lower than the sensible liquid desiccant into the airstream [82], which often leads to liquid
effectiveness. The values in Fig. 11 are for a single condition, which desiccant entrainment in the air. Because the liquid desiccant is
is reasonable for the sensible effectiveness; any variation in the corrosive, this carryover can damage ductwork, other HVAC
conductivity of the membrane will have a small impact on equipment, and building components. Using a low flow rate of
performance because the airside resistances dominate. The latent liquid desiccant on a wicked surface reduces this carryover [83],
effectiveness, though, varies with the inlet air conditions; the but using a membrane eliminates it. For the regenerator, contain-
membrane's permeability varies more with temperature and ment is not as critical; the air is not entering the building.
humidity, and the membrane resistance is not negligible. However, containment still reduces the risk of corrosion, and
The dependency of permeability on the air temperature and using membranes in both the regenerator and conditioner pre-
humidity can be understood with the data from Zhang et al. [49]. vents some air pollutants from contaminating the desiccant [24].
Plotting their 15 measured latent effectiveness values versus the In addition to containing the desiccant, membranes also provide a
inlet RH indicates a strong positive correlation (plot not shown constant area, which can be maintained even at part load [24,84].
here for brevity). The slope of the best-fit line indicates approxi-
mately a 7 percentage point increase in latent effectiveness for 3.3.1. Module design
every 10 percentage point increase in RH, over a range of 28–50% The membrane modules for both the conditioner and regen-
RH. By contrast, there is no correlation between the sensible erator bring liquid desiccant and air into contact on opposite sides
effectiveness with either temperature or RH. of a porous membrane. The modules for both are similar; it is
The variation in membrane permeability with RH could be due primarily the operation that differs: high liquid desiccant tem-
to either the slope of the sorption curve (i.e., solubility) or the peratures lead to desorption (regenerator), low temperatures lead
diffusivity, but most researchers have focused on the former. to absorption (conditioner). One difference in design, though, is
Zhang et al. [59] calculated the slope of the sorption curve as a that the membrane in the regenerator must be able to withstand
function of RH, showing a steeper curve at higher RH. Min et al. [77] elevated temperatures (up to 90 1C) and the temperature swings
showed similar trends with their measurements on three mem- between these elevated temperatures and ambient. As noted by
branes. Niu and Zhang [78] modeled hypothetical materials with Conde [84], a membrane regenerator is susceptible to leaks
different sorption curves and quantified this effect. They showed because of thermal expansion mismatches.
that a linear sorption curve (i.e., constant slope) is generally In both the conditioner and regenerator, hollow fibers offer
preferred, but noted that for polymers the curve is typically some advantages over flat sheets. The structure of the hollow
concave, meaning steeper curves, and therefore higher perme- fibers inherently supports the liquid pressure, eliminating any
ability, at higher RH (see Eq. (7) in Section 4.1.1). Since pressure membrane deflection. Hollow fibers also provide a higher surface
drop is not a strong function of inlet condition, this variation in area to volume ratio, which is illustrated by comparing the
permeability changes the energy performance depending on the 0.015 m2 membrane area in the flat-sheet module used by Isetti
location and season. et al. [24] with the 1.13 m2 membrane area in the hollow-fiber
Other factors impact the performance of an ERV. Kosar [79] module used by Bergero and Chiari [25]. Hollow fibers are also
quantified the effect of unbalanced airflows, where the exhaust easier to seal, with all of the sealing taking place at the two
airflow is less than the incoming ventilation air because of air headers (Fig. 4b), instead of around the edges of each flat-sheet
being exhausted at other locations (e.g., a kitchen exhaust fan). membrane. A simple method for making small-scale, hollow-fiber
This reduces the capacity of the ERV. Zhang [63] looked at the modules was documented by researchers at the Georgia Institute
maldistribution of the airflows (e.g., from one channel to another)
due to improper installation within the ducting system. This also warm, humid air hot, dry air
degrades performance. Several researchers [50,80,81] looked at
how control methods can affect performance. For example, in the
summer there may be times when the indoor temperature and
diluted LD conditioner
humidity are higher than outdoors because of internal gains (e.g.,
equipment, people). Operating an ERV during these times increases
energy use because the exhaust air adds sensible and latent energy
to the incoming ventilation air. Although more research is needed, concentrated LD
some possible control strategies have been proposed to minimize
this undesirable operation [81].
humid air ambient air
3.3. Liquid desiccant dehumidification regenerator
Fig. 12. Schematic of liquid desiccant dehumidification. Break lines are used for
Liquid desiccant dehumidifiers perform the same function as simplicity; the process generally has components (e.g., tank, pumps, heat exchan-
membrane vacuum dehumidifiers, but the process is driven by gers) between the regenerator and conditioner that depend on the configuration.
of Technology, which can be found in the appendices of some adhesion of the membrane to the plate. Once assembled, the
theses (e.g., [85]). adhesive swelled, causing misalignment of some liquid manifolds
These hollow-fiber modules typically look like the module in and leaking, especially for the regenerator, which saw much larger
Fig. 4b, but there are two issues with this design, both related to temperature swings [84]. They suggest that the assembly and
the air flow on the shell side: high pressure drop [35], and poor reliability of the module could be improved with injection-molded
airflow distribution [41,42]. The airside pressure drops are gen- parts and thermal welding of membranes instead of adhesion [84].
erally higher than in flat-sheet modules, but this also means the Kozubal et al. [33,94] pursued a similar device, but with an
airside resistances are typically small, with the largest resistance evaporatively cooled airstream replacing the chilled water, which
typically coming from the membrane [36]. Kneifel et al. [35] eliminates the need for a cooling tower or chiller. They tested two
proposed a transversal flow hollow-fiber module to improve flow prototypes, one with 41 channel pairs and a surface area of
distribution, where frames of hollow fibers are stacked together, 10.2 m2 [33], and one with 36 channel pairs and a surface area
with each set perpendicular to the airstream. of 12.1 m2 [26]. Again, the sealing of the membrane edges was
Using flat sheets instead can reduce this pressure drop, but challenging.
they have a lower area to volume ratio and generally have lower
overall mass transfer coefficients due to slow airside transfer [86].
Similar to membrane ERVs, spacers or enhancements can be used 3.3.2. System design and analysis
to reduce the airside resistance, as in [26]. These spacers also help While using membranes in liquid desiccant air conditioners is
support the membrane [87]. Several researchers have looked at relatively new, liquid desiccant systems by themselves are not
flat sheet membrane contactors for desiccant-air modules [88–90]. new. The advantages and energy-savings potential of liquid
The primary advantage of flat-sheet membranes, though, is that desiccant systems have been reviewed previously [91,95], whether
they enable a third fluid stream that can integrally cool the they use membranes or not. Their primary advantage is that they
conditioner or integrally heat the regenerator, as shown in replace some electrical energy input with thermal energy, redu-
Fig. 13. If the membrane module contains only two fluids, the cing the peak electric demand and benefiting utilities. They also
liquid desiccant must be heated prior to entering the regenerator reduce energy use in buildings requiring strict humidity control, or
and cooled prior to entering the conditioner. Alternatively, a three- in supermarkets that have large refrigeration systems. For more
fluid module, which is easier to construct with flat-sheet mem- details, see [91,95–97].
branes, enables cooling of the desiccant internal to the conditioner Some researchers have specifically looked at system efficiencies
and heating of the desiccant internal to the regenerator. and benefits when using membrane modules. The membrane
This integral heating and cooling offers advantages over an contains the liquid desiccant, which is beneficial if the system
adiabatic dehumidifier. Internal cooling in the conditioner keeps design requires several fluids interacting in one heat and mass
the vapor pressure of the desiccant low, which leads to larger exchanger. One of these concepts, integrated cooling of the
driving forces [26]. Compared to a non-cooled conditioner, this conditioner, was discussed above. Another option is to combine
translates into either a lower required surface area or dryer air, or liquid desiccant dehumidification with another cooling device,
both. Liquid desiccant flow rates are lower because the cooling either evaporative cooling or vapor compression cooling.
fluid provides the heat capacity to carry away the heat of absorp- Combining liquid desiccant systems with evaporative cooling
tion; without integrated cooling, the liquid desiccant must provide can lead to significant energy savings because liquid desiccants
this heat capacity [91]. This also means lower pump energy. In the provide low RH air (similar to the vacuum dehumidifiers discussed
regenerator, the larger driving forces from internal heating enable in Section 3.1). This results in independent control of latent and
a smaller device and also make it easier to concentrate the sensible cooling. Conde and Weber [92] combined a membrane
desiccant to a higher concentration. The heating fluid maintains evaporative cooler with the liquid desiccant dehumidifier, both
the desiccant temperature, even for low desiccant flow rates. made with flat-sheet membranes (see Section 3.4 for more on
Two research groups recently built conditioners using mem- membrane evaporative cooling). The concept investigated by
branes with internal cooling. Conde and Weber [84,92,93] con- Kozubal et al. and Woods and Kozubal also included a separate
structed a prototype conditioner with internal cooling, and also a evaporative cooler [26,33,94].
regenerator with internal heating [84]. The conditioner was a flat- Liquid desiccant membrane modules can also be used in vapor
sheet device with 14 air channels and a total area of 4.6 m2. Conde compression air conditioners, which enable more efficient latent
and Weber [93] also attempted to construct a shell-and-tube cooling than a conventional vapor compression machine. This was
module, where the flat sheets were wrapped around a metal tube considered by Bergero and Chiari [98], who showed 50% energy
containing the chilled water. This assembly process was difficult savings over a conventional vapor compression air conditioner for
and damaged the membranes. The flat-sheet prototypes were also high latent conditions. In this concept, the evaporator cools the
difficult to assemble, particularly the detailed machining and the liquid desiccant entering the conditioner and the condenser heats
the liquid desiccant entering the regenerator. Isetti [99] looked at
combining the refrigerant, liquid desiccant, and air into one
warm, humid air warm, dry air module in a similar concept that used hollow fibers.
Some private companies are developing technology in this area,
diluted LD concentrated LD but the details are not available in the open literature. ARPA-E
warm water chilled water funded United Technologies Research Center from 2010 to 2013 [23]
Integrally cooled conditioner to develop a hybrid vapor compression machine using liquid
desiccants and membranes. 7AC technologies [100] received private
hot water warm water investment to pursue a similar concept.
diluted LD concentrated LD
humid air ambient air 3.4. Membrane evaporative cooling and humidification
Integrally heated regenerator
Fig. 13. Flat-sheet designs of integrally cooled conditioner and integrally heated Evaporative cooling and humidification differ only in the goal of
regenerator. the process: evaporative cooling is used to control temperatures,
dry air cool, humid air Note that scaling is not an issue for liquid desiccant systems,
since there is no municipal water. Biological growth is also not an
water water issue because liquid desiccants ensure a low RH environment.
Charles and Johnson [101] also investigated fouling on the
Fig. 14. Membrane humidification or evaporative cooling. airside. They found minimal effect from particle fouling, but they
think this could be because their particles were too small. This is
humidification to control humidity. These processes are shown in an important topic for all of the membrane devices discussed in
Fig. 14. this review, and deserves further research.
Evaporative cooling has long been known as an energy efficient
means of cooling air, but there are still some issues, real or 3.5. Other membrane contactors for HVAC applications
perceived, that limit their use. Membranes have the ability to
solve some of these issues. There are other HVAC applications of membranes that do not fit
As discussed by Johnson et al. [28], membranes have a large, into the categories above. This is not meant to detract from their
fixed surface area for improved heat and mass transfer, especially potential success. Rather, it simply indicates that they have been
when using hollow fibers. The small pores of the membrane could studied less than the concepts discussed above.
also eliminate the passage of microbes and bacteria, providing a The first concept is an ERV, like in Section 3.2, but instead of
sanitary evaporative cooler without the need for frequent cleaning contacting the two airstreams, it pumps a liquid desiccant
or the use of anti-bacterial agents. Using membranes can also between two liquid–air membrane contactors [102–105]. The
eliminate pumps and sumps by using mains-water pressure to liquid desiccant exchanges heat and moisture with the outgoing
provide low flow rates slightly above the rate of evaporation. exhaust air in one module and with the incoming ventilation air in
the other. This enables latent and sensible energy recovery even
when the exhaust and supply ventilation ducting are not co-
3.4.1. Module design located. While these are ERVs, the modules are similar to the
Researchers have focused primarily on hollow-fiber mem- liquid–air modules discussed in Section 3.3. This same research
branes [25,27,28,101], as they offer more surface area and easier group has also investigated a system using these same liquid–air
construction than flat sheets. Similar to the other membrane modules with a desiccant heater and a desiccant chiller so that
devices, flow distribution was an issue in some hollow-fiber these modules could be used as a stand-alone air conditioning
prototypes [28]. Much of these distribution problems were found system [106].
to be worse at high packing fractions [41,43], which are typically Researchers have also considered membranes for components
preferred based on theoretical performance [34]. in absorption heat pumps [107–114], which have long been looked
Using flat sheets makes it easier to use more than two fluids. at for their energy-savings potential by using solar or waste
While internal cooling or heating is not as helpful as it is for the thermal energy. The membranes can reduce the size of the heat
liquid desiccant dehumidifiers, there are other reasons to have and mass exchange components (generator, absorber condenser,
more than two fluids. Conde [92] used flat-sheet membranes to evaporator) by providing thin liquid films and high surface area.
build an indirect evaporative cooler, where the evaporation of A project was funded by ARPA-E from 2009 to 2012 to develop
water was used to cool a third fluid stream. Kozubal et al. [33] also these membrane components [23].
built an indirect evaporative cooler, but eventually abandoned Researchers have considered other membrane processes for
membranes in favor of flocked or wicked surfaces [26,33]. liquid desiccant regeneration: vacuum membrane distillation [115];
electrodialysis [116,117]; and reverse osmosis [118,119]. The latter
two operate at lower temperatures than the thermally driven
regeneration discussed in Section 3.3. They also are generally driven
3.4.2. System design and analysis by electricity, which is available in most buildings, but is generally
A key concern for evaporative systems is the treatment of the more expensive than direct combustion, and has more embodied
water prior to entering the membrane module. Using tap water energy [120].
can lead to fouling, which is the deposition of solute or particles on
the membrane surface [101]. While fouling can be an issue in any
evaporative system, fouling of membranes can substantially 4. Modeling approaches
reduce the evaporation rate by blocking or clogging pores. This
fouling can be caused by scaling or biological fouling. The behavior of these processes and their performance can be
Scaling occurs when the concentrations of certain dissolved predicted with numerical models. A review of appropriate model-
salts (primarily calcium carbonate) reach saturation. Charles and ing techniques and their equations is outside of the scope of this
Johnson [101] used experiments to indicate how much wasted review. Instead, this section gives a general overview and cites
water, which is the amount of water provided above the rate of relevant references.
evaporation, is required to prevent significant scaling. They found The most common modeling approach is to discretize the
that scaling will result in significant flux reduction for wasting governing equations in the flow direction using a one dimensional
flow rates under 30% (70% evaporated, 30% drained). However, finite-difference technique [25,26,47,104]. Some researchers have
their study was not conclusive on how membrane cleaning, where also looked at NTU-effectiveness approaches [48,51,72,121] where
periodic high flow rates of water are used to flood the membrane NTU is the calculated number of transfer units for either heat or
channels, will prevent scaling. It also did not consider water mass transfer.
pretreatment. In either approach, the key (and more difficult) task is under-
Charles and Johnson [101] also looked at biological growth. standing the heat and mass transfer between the fluid streams
They found that bacteria do not necessarily need liquid water for (normal to the flow direction). This is typically calculated with
growth, but can grow on the airside surface of the membrane, resistance networks. Fig. 15 shows hypothetical vapor pressure
which is warm and humid. This contradicts one of the supposed and temperature profiles, at steady state, through each layer
advantages of membranes; additional research is needed to between the fluids for four of the processes discussed above. Each
answer this question more definitively. of these layers resists heat and mass transfer.
where S is the solubility of water vapor in the membrane material,

AIR MEM VAPOR AIR MEM AIR and D is the water vapor diffusivity. Both of these properties depend
on the moisture content of the membrane. However, the solubility
pv pv (i.e., sorption curve slope) seems to be a stronger function of
moisture content (see Section 3.2.2).
This model requires an estimate for the permeability. This can
be measured directly using standard techniques from ASTM [122].
Alternatively, this can be calculated from the solubility and
diffusivity. Solubility can be measured by weighing a sample at
T different values of RH, as in [77]. The diffusivity can be measured
with a transient permeation experiment, the so-called time-lag
method [123].
T For porous membranes, mass transport is commonly modeled
with the dusty gas model [124,125]. The adaptation relevant to
transport through membranes for HVAC processes is the same as
that used for membrane distillation, and details on this equation
and its derivation can be found elsewhere [32]. For the processes
discussed in this review, there are two relevant transport mechan-
AIR MEM LD PLATE WATER AIR MEM WATER isms: Knudsen diffusion, which dominates in smaller pores, and
molecular diffusion, which dominates in larger pores. While
thermal and surface diffusion are not zero [126], they are small
pv and typically neglected [32]. Viscous flow is also negligible since
the absolute pressure gradient is small [32]. High pressure on the
liquid side, as long as it is below the breakthrough pressure, will
pv not affect transport. With these assumptions, the molar flux across
the membrane is
ω
1 1 yair 1 Δpvmem
Jv ¼ þ ð8Þ
RT DM DK t
T T
where R is the gas constant and yair is the mole fraction of air. The
transport coefficients for Knudsen flow (DK) and molecular diffu-
sion (DM) are
sffiffiffiffiffiffiffiffiffiffiffi
ε dp 8RT
DK ¼ ð9Þ
Fig. 15. Steady state temperature (T), vapor pressure (pv), and liquid desiccant mass τ 3 π Mw
fraction (ω) across the membrane and fluid boundary layers for: (a) vacuum
membrane dehumidifier; (b) ERV; (c) internally cooled liquid desiccant condi- ε
DM ¼ Dwa ð10Þ
tioner; and (d) evaporative cooler. MEM ¼membrane, LD ¼liquid desiccant. τ
where ε is the porosity (open volume/total volume), dp is the mean
The heat flux (q) and molar flux (J) across each of these pore size, Dwa is the molecular diffusion coefficient for water vapor
resistances, i, is modeled with in air, and Mw is the molar mass of water. The deviation of the
pores from being straight, cylindrical, and non-interconnected is
qi ¼ ΔT i =RHT;i ð4Þ typically assumed to be captured in the tortuosity factor (τ).
Often the mean pore size, thickness, and porosity can be
J v;i ¼ Δpv;i =RMT;i ð5Þ obtained from the membrane manufacturer, but they can also be
where RHT and RMT are respectively the heat and mass transfer estimated using experimental techniques [127,128]. The tortuosity
resistances, T is the temperature, and pv is the vapor pressure. We factor can be estimated as a function of the porosity depending on
must estimate these resistances for the membrane and for the how the membrane is formed [129]. A value of two is assumed by
phases on each side of the membrane. These are discussed briefly many researchers [130], but this can be low for low porosity
in the following sections. membranes. Instead of a single pore size, there is actually a
distribution of pore sizes to consider in modeling [131]. However,
for diffusion-based processes, using the mean pore size is ade-
4.1. Transport through the membrane quate unless the pore sizes are small enough for Knudsen diffusion
to dominate [132].
4.1.1. Mass transport
For describing mass transport through the membrane, we must
4.1.2. Heat transport
distinguish between transport through dense membranes and
The key property for heat transport is the thermal conductivity.
transport through porous membranes. For dense membranes, the
For dense membranes, this depends on the water content, and is a
transport across the membrane is calculated with
material property that needs to be measured or obtained from a
Δpvmem material property database. As discussed in Section 3, the mem-
Jv ¼ P ð6Þ
t brane is a small part of the overall heat transfer resistance; thus,
where P is the permeability, t is the membrane thickness, and this value is not critical.
Δpv-mem is the vapor pressure difference across the membrane. The The importance of membrane conductivity for processes using
permeability is defined by a solution–diffusion mechanism [6,8] porous membranes was also found to be small [86]. In porous
membranes, the conductivity of the membrane results from a
P ¼ SD ð7Þ combination of the conductivity of the air in the pores and the
solid membrane. The parallel and series models bound the product developers work to make them competitive with conven-
problem. The parallel model assumes two independent conduction tional technology. The focus of the research is starting to transition
paths through the air and the solid membrane; the series model from feasibility and proof of concept to cost and longevity. This is
assumes conduction through the solid membrane material and leading to design for manufacturing, accelerated life testing, and
then through the air. Experiments have shown actual values to be demonstrated performance in field installations. However, each
somewhere between these two extremes [133], and that inter- technology is at a different stage, and each has its own set of
mediate models are typically more appropriate. research needs.
Vacuum membrane dehumidification is just past the proof of
4.2. Transport through the gas phase concept phase. While similar modules have been used for indus-
trial drying applications, the largest module tested for this concept
Correlations for the non-dimensional heat and mass transfer was around 10 cm2 [40]. Tests on larger modules are needed,
coefficients (Nusselt and Sherwood numbers) are usually used to including hollow-fiber modules. Research is also needed to better
estimate transport through the gas phase. Correlations for heat understand how to size these devices, both the membrane module
exchangers are often used, meaning they neglect the effect of the and the compressor. Other configurations can also improve effi-
mass flux at the wall. The thermal and moisture diffusivities are ciency, primarily ones that lower the compression ratio of the
approximately equal for air, so these heat transfer correlations can compressor.
be used for mass transfer as well. Membrane ERVs are commercial products, but there are still
There are some differences, though, between the correlations some unanswered questions. For example, what is the perfor-
for heat exchangers and for these membrane devices. Zhang [134] mance of these units in the field, and how does this compare to
showed that neither constant temperature nor constant heat flux the performance measured in the laboratory? Specifically, the
correlations are appropriate for these devices; instead, some measured performance is needed on commercial products for a
combination of the two is best. The geometries of the airflow range of inlet temperatures and humidities. In-the-field monitor-
conduits can also make direct application of these heat-exchanger ing can also show how performance varies with inlet condition,
correlations inappropriate [135]. Below is a discussion about and also how it deteriorates with time. The latter can indicate the
transport in parallel plate channels, in hollow-fiber lumens, and useful life of the equipment.
around hollow fibers. Membrane liquid desiccant dehumidifiers require manufacturing
For empty channels in flat-sheet modules, the flow is typically methods that ensure leak-free modules, which can be checked with
fully developed and laminar. However, the channels are not accelerated life testing of the conditioner and regenerator. While
typically empty; they usually have some type of membrane there are still improvements needed in design and manufacturing,
support. Zhang studied several simple geometries that can be field tests of these devices can help determine their reliability.
analyzed theoretically [74,136,137]. Sometimes the spacers are Membrane prototypes for evaporative cooling and humidifica-
more complex, and they obstruct and mix the flow. In these cases, tion have shown the feasibility of this process. Questions about
it is difficult to estimate the heat and mass transfer coefficients. fouling persist. It is also unclear whether membranes, with their
Instead, they should be measured, as was done in [64]. added cost, will be preferred to wicked surfaces.
For hollow fibers, gas flow through the lumens is fully devel- In addition to these specific research needs, some research
oped and laminar. But gas flow through the lumens is uncommon would benefit all of these devices. The membrane can still be
due to high pressure losses. An exception is for the membrane improved, whether this means higher permeability, higher
vacuum dehumidifier, where the low-pressure vapor is likely to be strength, or reduced cost. But as this review discussed, the air
removed through the lumens. The shell side of a hollow-fiber boundary layer resistance is as important, if not more important,
module is more difficult to model, due primarily to non-uniform than the membrane resistance. Therefore, mass transfer enhance-
spacing of fibers. Some researchers have reviewed this modeling ments in the flow channels are needed.
problem and summarized correlations from several studies [41,43]. Research on the distribution of airflow is also important, which
is incomplete for flat-sheet modules. While the geometry is similar
4.3. Transport through the liquid phase to heat exchangers, unsupported membranes can lead to flow
maldistribution that would not be seen in an equivalent heat
The liquid phase in HVAC processes is either water or liquid exchanger with metal plates.
desiccant. For pure water, the transport resistance is small [27,138]. Another potential issue is membrane durability. The success of
There is no resistance to mass transfer, and since the Prandtl industrial membrane processes shows that it is possible to operate
number for water is an order of magnitude larger than that for membrane systems either without fouling or with controlled
air, the heat transfer resistance is usually smaller than that of air. fouling and scheduled maintenance. The HVAC processes dis-
Heat transfer for the liquid desiccant is also relatively fast, and cussed here should have less fouling issues than common mem-
this resistance is small [86]. The small diffusion coefficients mean brane filtration processes, where pressure forcibly pushes water
that mass transfer through the liquid desiccant is slower than heat through the membrane. However, more research is needed to
transfer, although researchers have still shown this to be small understand these potential fouling phenomena, and how environ-
compared to the air boundary layer [25,86]. If the flow rate is high mental conditions and pollutants will affect the life of the membrane.
enough, a developing-flow correlation is needed [114]. Once the This could be degradation of the material from exposure to tempera-
transport coefficient is known, the flux can be calculated with the ture and humidity cycles, airside particulate fouling on the membrane
stagnant film model [139]. surface, or degradation from chemical reactions or oxidation of the
membrane material.
The outlook for these membrane devices is uncertain. Advances
in membrane technology over the previous few decades have
5. Research needs and outlook enabled these unique devices for HVAC applications, which could
potentially lead to more energy efficient products. But more
The development of HVAC membrane processes comes hun- research and development is needed if these products are to
dreds of years after the invention of the vapor compression air compete against the inexpensive and established air conditioning
conditioner. These processes continue to evolve as researchers and technologies.
Acknowledgments [30] Lawson KW, Lloyd DR. Membrane distillation. J Membr Sci 1997;124:1–25.
[31] Gryta M. Osmotic MD and other membrane distillation variants. J Membr Sci
2005;246:145–56.
This work was supported by the US Department of Energy [32] Curcio E, Drioli E. Membrane distillation and related operations—a review.
under Contract no. DE-AC36-08-GO28308 with the National Sep Purif Rev 2005;34:35–86.
Renewable Energy Laboratory. The author would like to thank [33] Kozubal, E. Woods J. Judkoff, R. Development and analysis of desiccant
enhanced evaporative air conditioner prototype. National Renewable Energy
Eric Kozubal, Dane Christensen, Kim Trenbath, and Jay Burch for Laboratory, TP-5500-54755;2012.
their helpful comments. [34] Zhang LZ, Huang S-M. Coupled heat and mass transfer in a counter flow
hollow fiber membrane module for air humidification. Int J Heat Mass Transf
2011;54:1055–63.
[35] Kneifel K, Nowak S, Albrecht W, Hilke R, Just R, Peinemann KV. Hollow fiber
References membrane contactor for air humidity control: modules and membranes. J
Membr Sci 2006;276:241–51.
[1] US-DOE. Buildings energy data book: buildings technology program. US [36] Albrecht W, Hilke R, Kneifel K, Weigel T, Peinemann KN. Selection of
Department of Energy; 2011. microporous hydrophobic membranes for use in gas/liquid contactors: an
[2] Rudd A, Henderson H. Monitored indoor moisture and temperature condi- experimental approach. J Membr Sci 2005;263:66–76.
tions in humid-climate US residences. ASHRAE Trans 2007;113:435–49. [37] Zhang LZ. Coupled heat and mass transfer through asymmetric porous
[3] Hughes BR, Chaudhry HN, Ghani SA. A review of sustainable cooling membranes with finger-like macrovoids structure. Int J Heat Mass Transf
technologies in buildings. Renew Sustain Energy Rev 2011;15:3112–20. 2009;52:751–9.
[4] Ho WSW, Sirkar KK. Membrane handbook. Chapman & Hall, New York; 1992. [38] Vallieres C, Favre E. Vacuum versus sweeping gas operation for binary
[5] Li NN, Fane AG, Winston WS, Ho T Matsuura. Advanced membrane mixtures separation by dense membrane processes. J Membr Sci 2004;244:
technology and applications. Hoboken, NJ: John Wiley & Sons; 2008. 17–23.
[6] Mulder M. Basic principles of membrane technology. 2nd ed.. Dordrecht, [39] El-Dessouky HT, Ettouney HM, Bouhamra W. A novel air conditioning system
Boston: Kluwer Academic; 1996. —membrane air drying and evaporative cooling. Chem Eng Res Des
[7] US-DOE. Advanced research projects agency—energy. US Department of 2000;78:999–1009.
Energy. Available from: 〈http://arpa-e.energy.gov/〉 [cited 11.03.13]. [40] Scovazzo P, Scovazzo AJ. Isothermal dehumidification or gas drying using
[8] Wijmans JG, Baker RW. The solution-diffusion model: a review. J Membr Sci vacuum sweep dehumidification. Appl Therm Eng 2013;50:225–33.
1995;107:1–21. [41] Lipscomb GG, Sonalkar S. Sources of non-ideal flow distribution and their
[9] Metz SJ, van de Ven WJC, Potreck J, Mulder MHV, Wessling M. Transport of effect on the performance of hollow fiber gas separation modules. Sep Purif
water vapor and inert gas mixtures through highly selective and highly Rev 2004;33:41–76.
permeable polymer membranes. J Membr Sci 2005;251:29–41. [42] Chen H, Cao C, Xu LL, Xiao TH, Jiang GL. Experimental velocity measurements
[10] He GH, Mi YL, Yue PL, Chen GH. Theoretical study on concentration and effect of flow maldistribution on predicted permeator performances.
polarization in gas separation membrane processes. J Membr Sci 1999;153: J Membr Sci 1998;139:259–68.
243–58. [43] Wickramasinghe SR, Semmens MJ, Cussler EL. Mass transfer in various
[11] Tanaka O. An analysis of simultaneous heat and water-vapor exchange hollow fiber geometries. J Membr Sci 1992;69:235–50.
through a flat paper plate cross-flow total heat-exchanger. Int J Heat Mass [44] Hanlon P, editor. Compressor handbook. McGraw-Hill, New York; 2001.
Transf 1984;27:2259–65. [45] Dais, NanoAir. Dais Analtyic Inc. Available from: 〈http://www.daisanalytic.
[12] Metz SJ, van de Ven WJC, Mulder MHV, Wessling M. Mixed gas water vapor/ com/applications/nanoair.html〉 [cited 13.03.13].
N-2 transport in poly(ethylene oxide) poly(butylene terephthalate) block [46] Field AA. Reclaiming latent heat in flue gases. J Heat, Pip, Air Cond 1975;46
copolymers. J Membr Sci 2005;266:51–61. (11):85–7.
[13] Zhang LZ, Wang YY, Wang CL, Xiang H. Synthesis and characterization of a [47] Zhang LZ, Jiang Y. Heat and mass transfer in a membrane-based energy
PVA/LiCl blend membrane for air dehumidification. J Membr Sci 2008;308: recovery ventilator. J Membr Sci 1999;163:29–38.
198–206. [48] Zhang LZ, Niu JL. Effectiveness correlations for heat and moisture transfer
[14] Asaeda M, Du LD, Ikeda K. Experimental studies of dehumidification of air by processes in an enthalpy exchanger with membrane cores. J Heat Transf
an improved ceramic membrane. J Chem Eng Jpn 1986;19:238–40. 2002;124:922–9.
[15] Ito A. Dehumidification of air by a hygroscopic liquid membrane supported [49] Zhang YP, Jiang Y, Zhang LZ, Deng YC, Jin ZF. Analysis of thermal perfor-
on surface of a hydrophobic microporous membrane. J Membr Sci 2000;175: mance and energy savings of membrane based heat recovery ventilator.
35–42. Energy 2000;25:515–27.
[16] Li J, Ito A. Dehumidification and humidification of air by surface-soaked [50] Zhou YP, Wu JY, Wang RZ. Performance of energy recovery ventilator with
liquid membrane module with triethylene glycol. J Membr Sci 2008;325: various weathers and temperature set-points. Energy Build 2007;39:1202–10.
1007–12. [51] Kistler KR, Cussler EL. Membrane modules for building ventilation. Chem Eng
[17] Krull FF, Fritzmann C, Melin T. Liquid membranes for gas/vapor separation. Res Des 2002;80:53–64.
J Membr Sci 2008;325:509–19. [52] Zhang LZ. Energy performance of independent air dehumidification systems
[18] Noble RD, Gin DL. Perspective on ionic liquids and ionic liquid membranes. with energy recovery measures. Energy 2006;31:1228–42.
J Membr Sci 2011;369:1–4. [53] Nasif M, Al-Waked R, Morrison G, Behnia M. Membrane heat exchanger in
[19] Zhang LZ. Fabrication of a lithium chloride solution based composite HVAC energy recovery systems, systems energy analysis. Energy Build
supported liquid membrane and its moisture permeation analysis. J Membr 2010;42:1833–40.
Sci 2006;276:91–100. [54] Zhang LZ, Niu JL. Energy requirements for conditioning fresh air and the
[20] Wang KL, McCray SH, Newbold DD, Cussler EL. Hollow fiber air drying. long-term savings with a membrane-based energy recovery ventilator in
J Membr Sci 1992;72:231–44. Hong Kong. Energy 2001;26:119–35.
[21] Zhang XR, Zhang LZ, Liu HM, Pei LX. One-step fabrication and analysis of an [55] Mardiana-Idayu A, Riffat SB. Review on heat recovery technologies for
asymmetric cellulose acetate membrane for heat and moisture recovery. building applications. Renew Sustain Energy Rev 2012;16:1241–55.
J Membr Sci 2011;366:158–65. [56] Zhang LZ, Zhang XR, Miao QZ, Pei LX. Selective permeation of moisture and
[22] Zhang LZ. Numerical study of heat and mass transfer in an enthalpy VOCs through polymer membranes used in total heat exchangers for indoor
exchanger with a hydrophobic–hydrophilic composite membrane core. air ventilation. Indoor Air 2012;22:321–30.
Numer Heat Transf: Part A Appl 2007;51:697–714. [57] Zhang LZ. Heat and mass transfer in a total heat exchanger: cross-corrugated
[23] US-DOE. Building energy efficiency through innovative thermodevices, triangular ducts with composite supported liquid membrane. Numer Heat
advanced research projects agency—energy. US Department of Energy. Transf: Part A Appl 2008;53:1195–210.
Available from: 〈http://arpa-e.energy.gov/?q=arpa-e-programs/beetit〉 [cited [58] Min J, Su M. Performance analysis of a membrane-based enthalpy exchanger:
11.03.13]. effects of the membrane properties on the exchanger performance. J Membr
[24] Isetti C, Nannei E, Magrini A. On the application of a membrane air–liquid Sci 2010;348:376–82.
contactor for air dehumidification. Energy Build 1997;25:185–93. [59] Zhang LZ, Liang CH, Pei LX. Heat and moisture transfer in application scale
[25] Bergero S, Chiari A. Experimental and theoretical analysis of air humidifica- parallel-plates enthalpy exchangers with novel membrane materials.
tion/dehumidification processes using hydrophobic capillary contactors. J Membr Sci 2008;325:672–82.
Appl Therm Eng 2001;21:1119–35. [60] Larson MD, Besant RW, Simonson CJ. The effect of membrane deflections on
[26] Woods J, Kozubal E. A desiccant-enhanced evaporative air conditioner: flow rate in crossflow air-to-air exchangers. HVAC&R Res 2008;14:275–88.
numerical model and experiments. Energy Convers Manag 2013;65:208–20. [61] Hilmersson A. Paulsson U. Analysis of an energy recovery ventilator (M.S.
[27] Chiari A. Air humidification with membrane contactors: experimental and thesis). Halmstad University, Technical report IDE0621; 2006.
theoretical results. Int J Ambient Energy 2000;21:187–95. [62] Dobbs, G.M. Lemcoff, N.O. Cogswell F.J. Benoit J.T. Development of a high
[28] Johnson DW, Yavuzturk C, Pruis J. Analysis of heat and mass transfer latent effectiveness energy recovery ventilator with integration into rooftop
phenomena in hollow fiber membranes used for evaporative cooling. package equipment. United Technologies Research Center for US Department
J Membr Sci 2003;227:159–71. of Energy, UTRC Report R2006-6.400.0005-F-FR01; 2006.
[29] Drioli E, Criscuoli A, Curcio E. Membrane contactors: fundamentals, applica- [63] Zhang LZ. Flow maldistribution and performance deteriorations in
tions and potentialities. Membrane science and technology series. 1st ed.. membrane-based heat and mass exchangers. J Heat Transf: Trans ASME
Amsterdam, Boston: Elsevier; 2006. 2009;131 11801–1-11801–7.
[64] Woods J, Kozubal E. Heat transfer and pressure drop in spacer filled channels [96] Mohammad AT, Bin Mat S, Sulaiman MY, Sopian K, Al-abidi AA. Survey of
for membrane energy recovery ventilators. Appl Therm Eng 2013;50: hybrid liquid desiccant air conditioning systems. Renew Sustain Energy Rev
868–76. 2013;20:186–200.
[65] Zhang LZ. Convective mass transport in cross-corrugated membrane exchan- [97] Daou K, Wang RZ, Xia ZZ. Desiccant cooling air conditioning: a review.
gers. J Membr Sci 2005;260:75–83. Renew Sustain Energy Rev 2006;10:55–77.
[66] Kwak K, Bai C. A study on performance improvement of corrugated type total [98] Bergero S, Chiari A. Performance analysis of a liquid desiccant and mem-
heat exchanger considering the structure of flow passage on surface. J Mech brane contactor hybrid air-conditioning system. Energy Build 2010;42:
Sci Technol 2009;23:1528–35. 1976–86.
[67] Schwinge J, Wiley DE, Fane AG. Novel spacer design improves observed flux. [99] Isetti C, Nannei E, Orlandini B. Three-fluid membrane contactors for
J Membr Sci 2004;229:53–61. improving the energy efficiency of refrigeration and air-handling systems.
[68] Schwinge J, Wiley DE, Fletcher DF. Simulation of unsteady flow and vortex Int J Ambient Energy 2012;34:1–14.
shedding for narrow spacer-filled channels. Ind Eng Chem Res 2003;42: [100] 7AC. Ultra high efficiency HVAC solutions. 7AC Technologies Inc. Available
4962–77. from: 〈http://dx.doi.org//www.7actech.com〉 [cited 24.03.13].
[69] Li F, Meindersma W, de Haan AB, Reith T. Novel spacers for mass transfer [101] Charles NT, Johnson DW. The occurrence and characterization of fouling
enhancement in membrane separations. J Membr Sci 2005;253:1–12. during membrane evaporative cooling. J Membr Sci 2008;319:44–53.
[70] Fimbres-Weihs GA, Wiley DE. Review of 3D CFD modeling of flow and mass [102] Fan HS, Simonson CJ, Besant RW, Shang W. Performance of a run-around
transfer in narrow spacer-filled channels in membrane modules. Chem Eng system for HVAC heat and moisture transfer applications using cross-flow
Process 2010;49:759–81. plate exchangers coupled with aqueous lithium bromide. HVAC&R Res
[71] Shrivastava A, Kumar S, Cussler EL. Predicting the effect of membrane 2006;12:313–36.
spacers on mass transfer. J Membr Sci 2008;323:247–56. [103] Hemingson HB, Simonson CJ, Besant RW. Steady-state performance of a run-
[72] Zhang LZ. An analytical solution for heat mass transfer in a hollow fiber around membrane energy exchanger (RAMEE) for a range of outdoor air
membrane based air-to-air heat mass exchanger. J Membr Sci 2010;360: conditions. Int J Heat Mass Transf 2011;54:1814–24.
217–25. [104] Seyed-Ahmadi M, Erb B, Simonson CJ, Besant RW. Transient behavior of run-
[73] Mardiana-Idayu A, Riffat SB. An experimental study on the performance of around heat and moisture exchanger system. Part I: model formulation and
enthalpy recovery system for building applications. Energy Build 2011;43: verification. Int J Heat Mass Transf 2009;52:6000–11.
2533–8. [105] Seyed-Ahmadi M, Erb B, Simonson CJ, Besant RW. Transient behavior of run-
[74] Zhang LZ. Heat and mass transfer in plate-fin sinusoidal passages with vapor- around heat and moisture exchanger system. Part II: sensitivity studies for a
permeable wall materials. Int J Heat Mass Transf 2008;51:618–29. range of initial conditions. Int J Heat Mass Transf 2009;52:6012–20.
[75] DAIS ConsERV. Why ConsERV. Available from: 〈http://www.conserv.com/ [106] Abdel-Salam AH, Ge G, Simonson CJ. Performance analysis of a membrane
commercial/why.html〉 [cited 19.03.13]. liquid desiccant air-conditioning system. Energy Build 2013;62:559–69.
[76] dPoint Technologies. HE series ExR ERV core. Available from: 〈http://www. [107] Drost, M.K. Friedrich, M. Martin, C. Martin J. Cameron, R. Recent develop-
dpoint.ca/images/docs/specsheets/dPoint_HE-ExR_2013.pdf〉 [cited 19.03.13]. ments in microtechnology-based chemical heat pumps. In: Proceedings of
[77] Min J, Hu T, Liu X. Evaluation of moisture diffusivities in various membranes. the 3rd international conference on micro-reaction technology. Frankfurt,
J Membr Sci 2010;357:185–91. Germany; 1999.
[78] Niu JL, Zhang LZ. Membrane-based enthalpy exchanger: material considera- [108] Drost, M.K. Enhancement of heat and mass transfer in mechanically con-
tions and clarification of moisture resistance. J Membr Sci 2001;189:179–91. strained ultra-thin films. US Department of Energy. DE-FC36-01GO11049;
[79] Kosar, D. Laboratory evaluation of energy recovery ventilators. Building 2005.
America Partnership for Improved Residential Construction (BA-PIRC). [109] Shaal, F. Weimer, T. Stroh, N. Walitza, E. Mattes H. Hasse, H. Membrane
DOE/GO 102013-3888; 2013. contactors for absorption refrigeration. In: Proceedings of the 10th Aachen
[80] Lam, K.P. Lee S.R. Dobbs, G.M. Simulation of the effect of an energy recovery membrane colloquium. Aachen, Germany; 2005.
ventilator on indoor thermal conditions and system performance. In: [110] Ali AHH, Schwerdt P. Characteristics of the membrane utilized in a compact
Proceedings of the ninth international building performance simulation absorber for lithium bromide–water absorption chillers. Int J Refrig 2009;32:
association conference. Montreal; 2005. 1886–96.
[81] Rasouli M, Simonson CJ, Besant RW. Applicability and optimum control [111] Ali AHH. Design of a compact absorber with a hydrophobic membrane
strategy of energy recovery ventilators in different climatic conditions. contactor at the liquid–vapor interface for lithium bromide–water absorption
Energy Build 2010;42:1376–85. chillers. Appl Energy 2010;87:1112–21.
[82] Lowenstein, A. Slayzak, S. Ryan J. Pesaran, A. Advanced commercial liquid- [112] Chen JH, Chang H, Chen SR. Simulation study of a hybrid absorber—heat
desiccant technology development study. National Renewable Energy exchanger using hollow fiber membrane module for the ammonia–water
Laboratory. NREL/TP-550-24688; 1998. absorption cycle. Int J Refrig 2006;29:1043–52.
[83] Lowenstein, A. Slayzak S. Kozubal, E. A zero carryover liquid-desiccant air [113] Woods J, Pellegrino J, Kozubal E, Slayzak S, Burch J. Modeling of a membrane-
conditioner for solar applications. In: Proceedings of the ASME 2006 based absorption heat pump. J Membr Sci 2009;337:113–24.
international solar energy conference. Denver, Colorado; 2006. [114] Woods J, Pellegrino J, Kozubal E, Burch J. Design and experimental character-
[84] Conde M. Weber, R. Open absorption system for cooling and air conditioning ization of a membrane-based absorption heat pump. J Membr Sci 2011;378:
using membrane contactors. Eidgenössisches Departement für Umwelt; 85–94.
2008. [115] Wang ZS, Gu ZL, Feng SY, Li Y. Applications of membrane distillation
[85] Ma C. Optimization of asymmetric hollow fiber membranes for natural gas technology in energy transformation process-basis and prospect. Chin Sci
separation. Masters of Science, Chemical and Biomolecular Engineering, Bull 2009;54:2766–80.
Georgia Institute of Technology, Atlanta, GA. 2011. [116] Li XW, Zhang XS. Photovoltaic–electrodialysis regeneration method for liquid
[86] Woods J, Kozubal E. Combining liquid desiccant dehumidification with a dew- desiccant cooling system. Sol Energy 2009;83:2195–204.
point evaporative cooler: a design analysis. HVAC&R Res. 2013;19:663–75. [117] Li XW, Zhang X-S, Quan S. Single-stage and double-stage photovoltaic driven
[87] Larson MD, Simonson CJ, Besant RW, Gibson PW. The elastic and moisture regeneration for liquid desiccant cooling system. Appl Energy 2011;88:
transfer properties of polyethylene and polypropylene membranes for use in 4908–17.
liquid-to-air energy exchangers. J Membr Sci 2007;302:136–49. [118] Al-Sulaiman FA, Gandhidasan P, Zubair SM. Liquid desiccant based two-stage
[88] Jain S, Tripathi S, Das RS. Experimental performance of a liquid desiccant evaporative cooling system using reverse osmosis (RO) process for regenera-
dehumidification system under tropical climates. Energy Convers Manag tion. Appl Therm Eng 2007;27:2449–54.
2011;52:2461–6. [119] Al-Farayedhi AA, Gandhidasan P, Ahmed SY. Regeneration of liquid desiccants
[89] Das RS, Jain S. Experimental performance of indirect air–liquid membrane using membrane technology. Energy Convers Manag 1999;40:1405–11.
contactors for liquid desiccant cooling systems. Energy 2013;57:319–25. [120] Deru M. Torcellini, P. Source energy and emission factors for energy use in
[90] Huang SM, Zhang LZ, Tang K, Pei LX. Fluid flow and heat mass transfer in buildings. National Renewable Energy Laboratory. NREL Report TP-550-3861;
membrane parallel-plates channels used for liquid desiccant air dehumidi- 2007.
fication. Int J Heat Mass Transf 2012;55:2571–80. [121] Kadylak D, Cave P, Merida W. Effectiveness correlations for heat and mass
[91] Lowenstein A. Review of liquid desiccant technology for HVAC applications. transfer in membrane humidifiers. Int J Heat Mass Transf 2009;52:1504–9.
HVAC&R Res 2008;14:819–39. [122] ASTM. E96: standard test methods for water vapor transmission of materials.
[92] Conde M. Weber. R. Open absorption system for cooling and air conditioning American Society for Testing and Materials; 2012.
using membrane contactors. Eidgenössisches Departement für Umwelt; [123] Barrer RM. Permeation, diffusion, and solution of gases in organic polymers.
2006. Trans Faraday Soc 1939;35:628–43.
[93] Conde M. Weber. R. Open absorption system for cooling and air conditioning [124] Mason EA, Malinauskas AP. Gas transport in porous media: the dusty-gas
using membrane contactors. Eidgenössisches Departement für Umwelt; model. Chemical engineering monographs. Amsterdam: Elsevier Scientific
2005. Pub. Co.; 1983.
[94] Kozubal, KE. Woods, J. Burch, J. Boranian A. Merrigan, T. Desiccant enhanced [125] Maxwell JC. Illustrations of the dynamical theory of gases. Philos Mag
evaporative air-conditioning (DEVap): evaluation of a new concept in ultra 1860;20:19–32.
efficient air conditioning. National Renewable Energy Laboratory. TP-5500- [126] Findley ME, Tanna VV, Rao YB, Yeh CL. Mass and heat transfer relations in
49722; 2011. evaporation through porous membranes. AIChE J 1969;15:483–9.
[95] Conde, M. Liquid desiccant-based air-conditioning systems—LDACS. In: [127] Iversen SB, Bhatia VK, DamJohansen K, Jonsson G. Characterization of
Proceedings of the European Conference on Polygeneration. Tarragona, microporous membranes for use in membrane contactors. J Membr Sci
Spain; 2007. 1997;130:205–17.
[128] Khayet M, Matsuura T. Preparation and characterization of polyvinylidene [135] Gabelman A, Hwang ST. Hollow fiber membrane contactors. J Membr Sci
fluoride membranes for membrane distillation. Ind Eng Chem Res 2001;40: 1999;159:61–106.
5710–8. [136] Zhang LZ. Laminar flow and heat transfer in plate-fin triangular ducts in
[129] Mackie JS, Meares P. The diffusion of electrolytes in a cation-exchange resin thermally developing entry region. Int J Heat Mass Transf 2007;50:1637–40.
membrane. I. Theoretical. Proc R Soc A 1955;232:498–509. [137] Zhang LZ. Thermally developing forced convection and heat transfer in
[130] El-Bourawi MS, Ding Z, Ma R, Khayet M. A framework for better under- rectangular plate-fin passages under uniform plate temperature. Numer
standing membrane distillation separation process. J Membr Sci 2006;285: Heat Transf: Part A Appl 2007;52:549–64.
4–29. [138] Woods J, Pellegrino J. Heat and mass transfer in liquid-to-liquid membrane
[131] Mochizuki S, Zydney AL. Theoretical-analysis of pore-size distribution effects contactors: design approach and model applicability. Int J Heat Mass Transf
on membrane-transport. J Membr Sci 1993;82:211–27. 2013;59:46–57.
[132] Woods J, Pellegrino J, Burch J. Generalized guidance for considering pore-size [139] Zydney AL. Stagnant film model for concentration polarization in membrane
distribution in membrane distillation. J Membr Sci 2011;368:124–33.
systems. J Membr Sci 1997;130:275–81.
[133] Garcia-Payo MC, Izquierdo-Gil MA. Thermal resistance technique for measur-
[140] AHRI. ANSI/AHRI standard 1060 2011 standard for performance rating of air-
ing the thermal conductivity of thin microporous membranes. J Phys D: Appl
to-air heat exchangers for energy recovery ventilation equipment. Arlington:
Phys 2004;37:3008–16.
VA. Air-Conditioning, Heating, and Refrigeration Institute; 2011.
[134] Zhang LZ. Heat and mass transfer in a cross-flow membrane-based enthalpy
exchanger under naturally formed boundary conditions. Int J Heat Mass
Transf 2007;50:151–62.

Woods2014 Membrane Processes For Heating, Ventilation, and Air Conditioning

Uploaded by

Copyright:

Available Formats

Woods2014 Membrane Processes For Heating, Ventilation, and Air Conditioning

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Woods2014 Membrane Processes For Heating, Ventilation, and Air Conditioning

Uploaded by

Copyright:

Available Formats

Renewable and Sustainable Energy Reviews 33 (2014) 290–304

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

Membrane processes for heating, ventilation, and air conditioning

5. Research needs and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301

Nomenclature P permeability (kmol Pa 1 m 1 s 1)

2.1.1. Dense membranes Composite membranes deserve some further explanation, as

This review focuses on ﬁve common membrane HVAC pro- Fluid 2

membrane material Cross section view

liquid air/vapor mixture dry air

water vapor transfer

Feed Retentate Sweep inlet Sweep/permeate outlet Membrane type

3.1.2. System design and analysis water vapor transfer

where T, ω, and h are respectively the air temperature, humidity

ﬁbers is their large pressure drop. Zhang [72] measured a pressure

where S is the solubility of water vapor in the membrane material,

You might also like