Applied Energy 189 (2017) 460–479

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

System integration, durability and reliability of fuel cells: Challenges and

solutions
Junye Wang
Faculty of Science and Technology, Athabasca University, 1 University Drive, Athabasca, AB T9S 3A3, Canada

h i g h l i g h t s

 Technical challenges and obstacles broken down during the scaling-up of fuel cells.
 Analysis of durability and reliability of fuel cell scaling-up using life cycle procedure.
 Three operating windows for different stages: components, individual cells, and stack.
 Operating window narrowing due to uneven flow distribution and dynamic load.
 Connection points among components, flow fields, cells, stack and system control.

a r t i c l e i n f o a b s t r a c t

Article history: The technical challenges and obstacles to scaling-up of fuel cells are diverse, including such issues as
Received 30 October 2016 water, heat, materials, catalyst, and flow fields because of multiple chemical and physical interactions
Received in revised form 12 December 2016 at the atomic level and stack system level. The current results and data, even assumptions and guidelines
Accepted 15 December 2016
are separated, inconsistent or unconnected. The unconnected data is partly the result of different disci-
plines. This paper is a first attempt toward understanding and analyzing the massive but spread-out
work, which has been done and reported in the literature on fuel cell performance, reliability and dura-
Keywords:
bility. In this, we analyze the procedure of fuel cell research and development, and break down the bar-
Flow field designs
Barriers
riers of scaling-up into four different stages: component, individual cell, stack and system control. We
Reliability and durability find that there are three different operating windows at each stage of the components, individual cells,
Process and system control and stack. While the operating window of components (e.g., membrane) are defined as ranges of temper-
Scaling-up ature and relative humidity (RH), the operating window of a cell must include channel velocity and pres-
sure drop within the cell. The operating window of a stack becomes narrower than that of its individual
cells due to uneven flow distribution and load change. We have also found that there are knowledge gaps
in the different stages of development. A solution for fuel cell scaling-up and a connection can be built
among the components, cells, stack, process and system control through the operating windows and flow
fields. The concepts of the three operating windows and flow field designs can build a connection among
properties of the material and structures of components (e.g., wettability, porosity, and hydrophobicity),
flow field, cells and performance of a stack and macro operation conditions (e.g., pressure, humidity and
flow rates). This clarifies key ambiguities and converges our future directions on how to bridge different
stages or disciplines of research and development. These can provide a new insight for future research to
address the key issues of durability and reliability that remain unsolved.
Ó 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
2. Life cycle procedure of research and development (R&D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
3. Materials and components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
3.1. Membrane electrode assemblies (MEA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
3.1.1. Membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
3.1.2. Gas diffusional layer (GDL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
3.1.3. Catalyst layers (CL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466

E-mail address: junyew@athabascau.ca

http://dx.doi.org/10.1016/j.apenergy.2016.12.083
0306-2619/Ó 2016 Elsevier Ltd. All rights reserved.
 J. Wang / Applied Energy 189 (2017) 460–479 461

3.2. Sealing, gasket and cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467

3.3. Properties of materials and water transport in diffusion media (DM). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
3.4. Summary and outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
4. Bipolar plates and single cell design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
4.1. Water and thermal management and operating conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
4.2. Theory of flow field designs for bipolar plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
4.3. Contact resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
4.4. Summary and outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
5. Flow field and stack design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
5.1. Performance test of fuel cell stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
5.2. Transient load and start/stop cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
5.3. Summary and outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
6. Control of system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
6.1. Stack operating windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
6.2. Startup/shutdown cycles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
6.3. Gas and air purging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
6.4. Summary and outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
7. Significance with application to fuel cell scaling-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
7.1. Knowledge gaps and challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
7.2. Three operating windows with application to the scaling-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
7.3. Summary and outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
8. Concluded remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476

1. Introduction Durability and degradation of materials and components have

Fuel cells are a low carbon technology because of their high effi-
ciency, low noise, and clean operation. Fuel cells could be great
sources of primary power if demonstrated to provide reliable
power for long continuous periods of time. High-quality power
and efficiency of fuel cells could be the most important marketing
drivers for many applications, such as backup power, material han-
dling and submarine engines. Ensuring their longevity and reliabil-
ity could greatly advance fuel cell technology. The US Department
of Energy (DOE) has established a durability target for fuel cells of
mobile (5000 h, equivalent to 150,000 miles of driving) and sta-
tionary (40,000 h). However, the world-wide commercialization
of fuel cells has not yet taken off [1–3]. Only a few fuel cell systems
with adequate long-term performance and reliability have been
demonstrated to the market. Their durability and cost have been
noted as the two greatest barriers [4,5]. Therefore, it is a top prior-
ity that fuel cell systems designers must solve the issues of durabil-
ity to function over the full range of operating conditions with less
than 10% loss of performance [6,7].
Durability is defined as the lifetime within the repair rate and
cost of planned repairs, overhaul and maintenance. However, this
durability target is not sufficient as the key target for acceptance
by end-users [1]. The end-user may focus more on reliability or
availability. Reliability is different from durability. Reliability is
defined as the ability of a product to perform the required function
under stated conditions for a certain period of time, which is often
measured by the probability of failure, frequency of failure, or in
terms of availability, which describes the ability of a system or
component to function well under stated conditions for a specified
period of time. The reliability and availability of fuel cells may be
more important than the durability for acceptance of end-users
as unplanned repair and maintenance can cause delays in home,
work or business activities, but this is not recognized in the strate-
gic targets and communities [1,2,8]. In fact, for commercialization
of any products, low reliability may result in decreasing market
share and penetration. Therefore, the principles of reliability and
validity are fundamental cornerstones of any products, systems
or manufacturer since they directly influence the acceptance of
products and capital return on investment [1].
been extensively studied at various scales from single cell to stack
with consideration to decreasing the costs and increasing durability
and efficiency of the overall system design. Degradation and aging of
any materials is unavoidable over time, but it is of concern that some
materials exceed their designed degradation rates. The conse-
quences of materials degradation are multiple: (1) the loss of system
gross performance and efficiency, (2) the loss of the system avail-
ability, (3) increase of cost due to increasing maintenance and repair
[1,9]. In an integrated system or stack, the malfunction of one com-
ponent may affect another, and the malfunction of a cell may affect
the operation of another cell. This type of interaction among compo-
nents in a stack system leads to a difficulty in maintenance and
repair. If one component fails, the whole stack system has to be
de-assembled, leading to a substantial repair costs. For example:
one has to disassemble a whole stack to replace a failed membrane
in an individual cell even though one can identify which membrane
failed. A single component failure could substantially increase the
cost of the whole stack, i.e.: a composite stack failure rate of 10%
could increase of the stack cost by 60% [10]. In practice, the cost of
materials and components has been reduced. For example, the cost
of the catalyst no longer dominates the price of most fuel-cell sys-
tems, although it is still significant [11]. Therefore, the study of reli-
ability is not only for the improvement of stack quality and
efficiency but also for substantial reduction of costs.
Research and development (R&D) and improvement of systems
must enhance the durability and reliability of fuel cells. The speci-
fic R&D issues are classified into: (1) Innovative structures of com-
ponents, such as a thinner catalyst layer and dimensions for easier
water removal [12,13], a thinner but stronger membrane that not
only facilitates easier water removal but also improves the reliabil-
ity of the thin membrane [14,15], catalyst layer (CL) [16,17] and
membrane electrode assembly (MEA) [18–21]; (2) New properties,
such as porosity and hydrophobicity modifications to improve
water and gas transport in the gas diffusion layers (GDL) [22–
24]; (3) Flow field designs and water and heat management
[2,25–29]; and (4) System modeling and control, especially of the
role of the start-stop processes [30–34].
Although the failures in modes and mechanisms of fuel cells
have been well realized [35], it is still difficult to communicate
462 J. Wang / Applied Energy 189 (2017) 460–479
between researchers and engineers because these functions are
performed separately and there is a lack of a systematic relation-
ship and connection among these groups to reach a consensus of
strategies due to the inconsistent and unreliable data and an
absence of guidelines for collaboration [36,37]. This problem is
compounded by a lack of understanding of the uncertainty within
the data from different disciplines and a misunderstanding of the
connection and interaction of the relative contribution of each
component’s degradation to the degradation of a whole fuel cell
stack. Since the components interact in a stack, the success of the
stack system depends to a large extent on the ability to evaluate,
analyze and integrate these interactions and connections. DOE
[4,5] has made massive efforts to develop a methodology of assess-
ment of the maturity of fuel cell technologies. DOE [38] has also
established a Durability Test Protocol. Some technology maturity/
readiness assessments for the degree of difficulty have been sug-
gested for research and development [39,40]. In recent years, there
have been many reviews of performance degradation and durabil-
ity of fuel cells [9,21,30,35,38,41–44]. However, no specific
methodology has been developed to evaluate the risk of readiness
of technology development.
Therefore, assessments of the maturity of fuel cell technologies
currently have limitations: (1) the lack of objective standard guide-
lines for implementation; (2) the lack of uncertainty metrics that
limits evaluation of the ‘‘actual” degree of maturity and develop-
ment risks; (3) the lack of stack system integration that addresses
technology readiness concerns of the end-users. The assessment
may impair the approach to achieve fuel cell commercialization.
There is still a need to achieve scientific and engineering break-
throughs of key knowledge gaps of the stack system integration
and reliability of fuel cell technologies/systems. In practice, there
is a lack of well documented reliability results and tools/methods
for the assessment of key technical barriers.
How can exiting knowledge and emerging technologies be
leveraged to address the issues of durability and reliability of fuel
cells? How will different scales from components to individual
cells and stacks be connected during scaling-up of fuel cells? The
primary objective of this paper is to analyze the processes of design
and development to break down technical barriers (e.g., reliability
and durability) into different stages and identify knowledge gaps
between the stages for an analysis of the likelihood of achieving
the desired set of performance and functional objectives. The sec-
ondary goal is to explore an approach on how to integrate/connect
different stages or scales. Emphasis is on connection points among
the different stages of scaling-up from components to stack sys-
tem. Although modeling of materials and catalyst and first-
principle modeling of transport will be mentioned, they will not
be included. The reader can refer to many good overviews of these
fields [45,46] for examples. An exhaustive review is difficult
because a significant part of the fuel cell scaling-up activities is
performed by Industry, and their outcome is not publicly available.
Alternatively, based on existing consensus in terms of experimen-
tal data and flow field designs, this paper identifies the key connec-
tion points required to move a component, a cell, or a stack from
one technology readiness stage to another through an analysis of
the scaling-up procedure and key design parameters. We will focus
on a strategy to address these issues systematically and potential
methods to connect different stages in R&D and designs of fuel
cells. This review is limited to proton exchange membrane fuel
cells (PEMFC) and direct methanol fuel cells (DMFCs), although it
will also be useful as information for other fuel cell types. Further-
more, auxiliary equipment, such as compressors, pumps and
hydrogen storage, are also not included as this paper is limited to
integration processes within a stack itself. This paper is organized
as follows: we analyze the procedure in the R&D of fuel cells and
identify four main stages of R&D: Components, cells, stack and con-
trol system as control points in Section 2. Then we analyze each
stage and identify the knowledge gap between these two stages.
In Section 3 we review the degradation mechanism of materials
and components of a fuel cell with emphasis on the concept of
the operating windows of materials and components. We explore
possibilities for widening the operating windows of components
through development of new materials and structure. In Section 4,
we present bipolar plates and a single fuel cell as a basic assembly.
We stress that the operating window of a single cell is different
from that of all combined components due to the change of two
threshold velocities despite there being no difference in material
properties. The operating window of an individual cell is always
dependent on flow velocity in channels that determine the capac-
ity to remove water. Therefore, the operating window of a cell can
be adjusted by material properties of the membrane and gas diffu-
sion layer, such as porosity, wettability and thickness. Flow field
and stack design is presented in Section 5. We examine the over-
lapping of the operating windows of components, cells, and stack
due to uneven flow distribution, which narrows the stack operat-
ing window. In Section 6, we discuss the last stage of operating
conditions and the program design of controlling a stack under
start/stop operations. There is a knowledge gap when considering
the effect of uneven flow distribution during designs of control sys-
tems. In Section 7, we discuss the significance of the operating win-
dows and their application to the fuel cell scaling-up design.
Finally we conclude this paper in Section 8. Further challenges
and observations concerning possible future directions for each
stage have been pointed out at the end of each section.
2. Life cycle procedure of research and development (R&D)
A single fuel cell is essentially reliable, but has to be integrated
into systems to meet industrial power requirements. Once fuel
cells connect together into a system or stack, in series or in parallel,
the stack often creates uncertainties in reliability due to issues in
the flow field design, system control design and system integra-
tion. A fuel cell stack consists of a large number of components
such as bipolar plates (BP), membrane electrode assemblies
(MEA), and gaskets. If any one of these components fails during
operations, it will result in the failure of the whole system. The fail-
ure rate of the series system is a sum of failure rates of its elements
in the system or the reliability of the system is the product of the
reliabilities of its elements. Therefore, the way that elements are
connected can affect the reliability of the whole system. To be reli-
able, the components in the fuel cell stack must have longer dura-
bility than that of individual fuel cells within the stack.
Different types of fuel cells have different causes of failure.
PEMFC often fail because of water and heat buildup, resulting in
catalyst particle growth, corrosion and dissolution, and mem-
brane/MEA chemical degradation, while the solid oxide fuel cell
(SOFC) can fail because of high operating temperatures, resulting
in electrode oxidation and cell-sealing problems due to mis-
matched thermal expansion of materials. However, the procedure
for R&D and commercialization of fuel cells is similar. They can
be categorized into three stages: components, individual cells
and stack (Fig. 1).
Reliability of a product or system is strongly affected by the sys-
tem integration and its components. Therefore, reliability design
activities are an ‘‘integrated process” during design and develop-
ment stage, and during subsequent verification (e.g., tolerance
analysis based on components selected). As mentioned above,
materials and components will be stable if they work under the
operating window of materials. However, the operating window
of the components is different from that of the channels. Fig. 2a
shows an operating window of materials and components. This
 J. Wang / Applied Energy 189 (2017) 460–479
Fig. 1. Three stages of fuel cell commercialization life cycle: component, single cell
and stack.
window is determined by the operating temperature and humidity
of materials while the operating window of the channels (Fig. 2b) is
determined by partial gas pressure and flow velocity given mate-
rial properties and structures [26]. Generally speaking, the operat-
ing window of the materials and components includes
temperature, humidity and composition. The operating window
of a single cell must include the velocity in channels. There are
two velocity thresholds to determine an operating range of a cell.
The maximum flow velocity in the channels is the threshold to
avoid drying for a given MEA and channel structure; and the min-
imum flow velocity in the channels is the threshold to avoid
flooding.
It is essential to design all the components and the system to
operate within their operating windows (Fig. 3). It should be
remembered that the operating windows are a range of operating
parameters. A stack will work properly inside the window range
whereas it may not work, or could cause failure, outside the range
for factors such as freezing and flooding. Operating conditions (e.g.,
pressure, humidity and temperature) and configuration are related
to the amount of mechanical stress and thermal deformation
caused by any high pressure drop or thermal expansion. Thus,
membranes CL and GDL must meet the requirement for strength
and stiffness of their materials if the cell operates under a high
pressure drop. Generally, the better the material strength and stiff-
ness, the higher the cost and the longer the material will last.
Fig. 2. Operating windows: (a) components and (b) channels [25] (Reproduced by permission of Elsevier).
463
Therefore, design is often a balance between cost and reliability.
In all cases, the development process of a product is an iterative
method of component selection and verification process. Any pro-
posed design solution should be verified against stated require-
ments before the fuel cell production can commence (Fig. 3b). If
the verification finds any design weaknesses, the design has to
be improved, and verified again before production can continue
(Fig. 3). An iteration process of the design, fabrication and verifica-
tion is to conform to all performance requirements, including effi-
ciency, reliability and durability.
3. Materials and components
The mechanisms of component degradation have been studied
using a combination of in situ tests and ex situ experiments to iso-
late and understand the different degradation modes. Some mech-
anisms of degradation and aging have been identified, such as
surface activity loss due to catalyst dissolution, and particle growth
and agglomeration, corrosion, membrane cracking and pinholes,
voltage loss due to contact resistance, and contamination [4]. The
strategy to address reliability and durability involves identifying
degradation mechanisms and developing approaches for mitigat-
ing their effects. The main components include MEA, CL, GDL and
BP.
3.1. Membrane electrode assemblies (MEA)
The MEA is the core component of a fuel cell, which consists of a
polymer electrolyte membrane, catalyst layers (CL) and gas diffu-
sion layers (GDL) (Fig. 4). Generally, the MEA can be fabricated
through two methods according to the catalyst layer coating: 3-
layer and 5-layer. One is that CL is coated on the surface of the
polymer electrolyte membrane as two electrodes. The catalyst-
coated membrane (CCM) is then sandwiched between two GDL’s.
Thus, the MEA has 5-layers which is then ready to be installed in
a cell. Another method is when the CL and the GDL are combined
to form an electrode, known as a 3-layer MEA. The 3-layer CCM
requires an attachment of a pair of GDLs making it a 5-layer config-
uration for installation. The carbon micro-porous layers (MPL) have
been typically used for these types of GDL.
Roll-to-Roll (R2R) technology is an important type of substrate-
based manufacturing process in which additive and subtractive
processes are used to construct structures in a continuous manner
[47]. R2R processing is commonly used for manufacturing PEM
MEA. Registration of the anode and cathode coatings on PEM cells
(the process of aligning the different layers) during R2R processing
464 J. Wang / Applied Energy 189 (2017) 460–479
Fig. 3. Design, fabrication, assembly and verification prior to commercialization.
Fig. 4. Schematic of MEA structure [2] (Reproduced by permission of Elsevier).
are complicated because the two electrode coatings are not adja-
cent. This is of critical importance to mitigating degradation during
operation [19]. Kang and Kim [48] found separation and significant
deterioration of the interface among the membrane, CL and GDL
compared with fresh MEA after dry/wet cycles. Further develop-
ment of direct coatings of electrodes onto membranes is still
required for highly uniform lamination processes for MEA compo-
nents as well as development needs for in-line quality control (QC)
of all MEA components [49].
Influence of hydro and humidity on MEA degradation has been
studied by many researchers [19,41,49–56]. It was reported that
the degradation of the MEA occurred under low humidity condi-
tions or without humidification of reactant gases, whereas no sig-
nificant degradation was observed under saturation humidification
of the feed stream [57–60]. Influence of nano-scale heterogeneity
of the catalyst layer was also studied on the interfacial strength
between catalyst layer and membrane [16,17,61–64].
3.1.1. Membrane
A perfluorosulfonic acid–based membrane (PFSA), developed by
DuPont (called Nafion) in the 1970s, is a notably progress, which
has been commonly used in PEMFC [65] and direct ethanol fuel
cells (DEFC) [66]. The degradation of a membrane can be attributed
to mechanical, thermal and chemical causes such as membrane
thinning, pinhole and cracking. Many factors cause the degradation
of a membrane, such as wet/dry cycles, mechanical and thermal
stresses, and drying.
Operating humidity of membranes are very important parame-
ters. A lower humidity leads to lower conductivity in the mem-
brane [67]. When the humidification is too high, water
condenses and the gas diffusion electrodes are flooded. This flood-
ing blocks the diffusion of the reaction gas in the gas diffusion layer
under water saturation conditions, which are produced by the pro-
duct water. Furthermore, flooding causes corrosion of the elec-
trodes, the catalyst layers, the gas GDL and the membrane [68].
On the other hand, continuous operation at a relative humidity
lower than 100% can lead to an increase in membrane degradation
[58]. The drier the operating conditions, the shorter the life of the
cell due to the membrane dehydration.
Relative humidity (RH) cycling or thermal cycling can cause
mechanical degradation and affect the membrane lifetime and fati-
gue resistance [69]. The swelling-shrinkage caused by water-
uptake can be the primary source of mechanical degradation
[56,70–72]. The membrane response strongly depends on its
 J. Wang / Applied Energy 189 (2017) 460–479
swelling anisotropy, where the stress amplitude decreases with
increasing anisotropy. These results suggest that it may be possible
to optimize a membrane with respect to swelling anisotropy to
achieve better fatigue resistance, potentially enhancing the dura-
bility of fuel cell membranes. Dillet et al. [73] mitigated the degra-
dation induced by startup and shutdown by tuning the MEA
specifications.
The hydration requirement of Nafion limits the operation of fuel
cells under 80 °C and hydration [74]. A higher temperature can
increase the conductivity of the membrane. However, conventional
PFSA hydrates for proton conductivity when the temperature is
below 80 °C. When the temperature rises above 80 °C, the
PEMFC operates at close to the boiling point of water. Sone et al.
[75] investigated the proton conductivity of Nafion 117 under var-
ious conditions of humidity and temperature from 21 to 80 °C. The
conductivity of this membrane varied strongly with the humidity
and heat-treatment of the membrane. From 21 to 45 °C, its conduc-
tivity decreased with increasing temperature at a given RH, while
it increased with temperature rise from 45 to 80 °C [75].
Dehydration causes stress concentration of the membrane,
leading to membrane pinholes, delamination and cracking, and
acceleration of the chemical degradation. Some high temperature
membranes have been developed [76–78]. For example, PBI-
based polymer membrane works on higher temperatures which
are attractive because of the drier operation and good CO tolerance
of the platinum catalyst [76]. Silica and zirconium phosphate can
be incorporated into Nafion through in situ chemical reactions that
increase the working temperature to above 100 °C [77,79,80].
Highly ordered mesoporous Nafion membranes possess substan-
tially higher tolerances toward RH fluctuation as compared to the
pristine Nafion membranes, and can be operated under dry condi-
tions. Thermal and mechanical degradation can occur at elevated
temperatures due to the relatively low glass transition tempera-
ture (Tg) of the hydrated Nafion (in the range of 90–120 °C)
[74,81,82]. Copolymer, P (HFa-co-BI), membranes are thermally
and mechanically stable although the proton conductivity values
of P (HFa-co-BI) membranes are smaller than that of the PBI mem-
branes. Their cell performance was 0.68 V at 0.2 A cm2 at 150 °C,
which is close to that of PBI membranes but their long-term dura-
bility (ca. 3116 cycles on in situ accelerated lifetime mode of load
cycling testing) is found to be far superior to the PBI membrane
[74,83]. On the other hand, a high temperature membrane requires
a high water vapor pressure to avoid possible high concentration
levels due to a low reactant-gas partial pressure, resulting in a high
total pressure as well as high parasite power. For example, a speci-
fic partial pressure of 0.5 atm of the reactant gases requires pres-
surization of at least 8 atm to keep 90% of the HR at 150 °C in a
water-saturated feed stream [74].
Besides operating conditions and properties of membrane
materials, the degradation of the membrane can be affected by
the thickness of the membranes. Yuan et al. [84,85] compared
Nafion membranes of four different thicknesses, including N117,
N115, NR212, and NR211, for 1000 h under the same operating
conditions. They found that the thicker membranes had much bet-
ter performance and less degradation after 800 h of operation. The
degradation of the thinner membranes were due to a dramatic
increase in hydrogen crossover. The pinholes were detected, using
IR image technology, with a high level of reactant crossover due to
heat output from the chemical reaction.
The aforementioned highly exothermal reaction between H2
and O2 can possibly lead to pinholes in the membrane, destroying
the membrane and causing catastrophic failure. Humidity and
temperature cycle can affect the mechanical properties of perfluo-
rosulfonic acid membranes, leading to stress concentration [54,55].
Highly uneven reactions lead to large temperature gradients and
local drying, which leads to cracking. Lim et al. [86] tested the
465
degradation of a standard MEA with PFSA ionomer membrane
using an accelerated stress test (AST). They found that the primary
cause of failure was uniform membrane thinning with pinhole for-
mation. The membrane became stiffer and more brittle during cyc-
lic humidity induced stress. High stress concentration can lead to
cracking and detaching due to the fatigue of load cycling and high
inlet/outlet pressure drops [2,87]. Some reinforced membranes
have been developed to tolerate high stresses. The lifetime of a
reinforced membrane can be many times longer than a non-
reinforced membrane [88]. The different types of the PBI have been
studied by Ossiander et al. [15]. They observed a distinct influence
of molecular weight distribution and reinforcement on the resis-
tance of stresses on the cathode layer and the mass transport resis-
tance of the cathode microporous layer. The stresses and plastic
deformation were lower than that of an unreinforced PFSA mem-
brane. Various perfluorosulfonated membranes exhibited better
mechanical fracture-resistance than polytetrafluoroethylene-rein
forced membranes. They showed better resistance to pinhole for-
mation and cracks due to their high resistance to cracking initia-
tion and propagation [9]. The PFSA backbone may be poisoned by
peroxide radical, leading to the degradation of Nafion membranes.
This can be improved through the incorporating of Ce++ and Mn3
in Nafion membrane [89,90].
3.1.2. Gas diffusional layer (GDL)
The GDL is commonly made of porous carbon materials, such as
carbon paper and carbon cloth [91,92] as a substrate covered by a
thinner microporous layer (MPL) with carbon black powder and a
hydrophobic agent. The GDL serves five functions: (1) reactant gas
permeability from flow-field channels to CL; (2) water removal
from the CL to flow channels; (3) electronic and ion conductivity;
(4) removal of reactive heat; and (5) mechanical support to the
membrane [29,93].
Operational conditions of a fuel cell, such as temperature, RH
and concentration vary as its load changes. The operating parame-
ters may exceed the operating window of the GDL, leading to var-
ious failures, such as hydrophobicity loss, carbon corrosion and
pore size changes due to compression or deformation. The chang-
ing of these properties can affect capacity of reactant and water
transport due to changing of the pore size and hydrophobicity
and loss of active electrochemical areas. In particular, the perfor-
mance of a fuel cell is more sensitive to the GDL changing when
the fuel cell operates in the higher current density region. This is
because more water will be generated at the high current density.
Furthermore, the freeze/thaw processes of the PEMFC occur mainly
within the pores of the GDLs in cold climates. Degradation of per-
formance was observed due to poor mass transport problems at
the freeze GDLs [94]. The freeze/thaw cycles can cause mechanical
stress. Carbon supports degradation occurs during start/stop due to
large fluctuations of potential, or during fuel starvation. In these
situations, the cathode or anode potential drastically increases,
causing carbon corrosion. Lim et al. [95] compared performance
of three different GDLs: cloth, paper and felt, during freeze/thaw
cycling conditions between 20 to 70 °C. The performance degra-
dation was negligible for the carbon felt based GDLs during
freeze/thaw cycling conditions while the cloth and paper GDLs
showed a marked performance loss [91]. Delamination of the
MPL from the GDL substrate may occur during freeze-thaw cycles,
as can also occur with catalyst-layer delamination from the mem-
brane [96] and mud crack under fatigue [18].
Arvay et al. [93] reviewed characterization techniques for GDLs.
They pointed out that the optimum characteristics of GDLs such as
structure, pore size, porosity, gas permeability, wettability, ther-
mal and electrical conductivities, surface morphology and water
management can improve the fuel cell performance. Serincan
and Pasaogullari [97] studied the influence of GDL anisotropy on
466 J. Wang / Applied Energy 189 (2017) 460–479
mechanical stresses of membranes. GDL with appropriate thick-
ness can optimize the stack operation performance. The membrane
thickness endures some changes: swelling and shrinking due to
changes in relative humidity and thermal expansion due to
changes in temperature. To avoid poor electrical and thermal con-
tact between the GDL and bipolar plate, a higher pressure may be
applied on the GDL. The effect of compression on the effective ther-
mal conductivity and thermal contact resistance were also studied
by Sadeghi et al. [98], Faydi et al. [99] and Mason et al. [100,101].
Because the compressive stress can easily reach 10 MPa or more
due to changes in water and heat transport, the GDL functions,
especially the interfacial electrical and thermal conductivities
between the bipolar plates and catalyst layers, depend significantly
on the GDL compression behavior [92,102]. Because the GDL
intrudes likely channels, the mechanical strength of the GDL is
required to be high enough to avoid intruding into the gas channels
of the bipolar plates.
Material design involves changing the material properties and
structures of the GDL and the CCL to function in the presence of liq-
uid water. Over the last 15 years, several strategies to change of
material properties in order to address water management in
PEM fuel cells have become common practice., such as integrating
hydrophobic materials (such as PTFE) into the GDL to reject water
using a hydrophobic surface, and adding a hydrophobic MPL
between the GDL and the catalyst layer to prevent liquid water
blockage of the reaction zones [103]. Thinner membranes, as thin
as 10 lm, have been used to shorten the distance of the water
back-diffusion process while, at the same time, to reduce the need
for anode humidification and to lower the ionic resistance or ohmic
losses through the membrane. However, thinner membranes are
often associated with poorer durability and higher gas crossover
rates due to their lower strength and stiffness, which limits the
practical membrane thickness to about 25–40 lm for fuel cell
applications [104,105]. Carbon corrosion can be mitigated by opti-
mizing the opening size of the cathode outlet in a PEMFC with
dead-ended anodes. Corrosion of the carbon support in both the
cathode and anode can be reduced with increasing the outlet size
at the cathode outlet. Carbon corrosion causes not only a loss of
weight and thickness but also degradation of mechanical strength
in the GDL.
3.1.3. Catalyst layers (CL)
The basic structure of electrodes is similar to that shown in
Fig. 4. Carbon paper or cloth is used for the basic mechanical struc-
ture of the electrode. Platinum (Pt) or a platinum alloy can be
printed using screen-printer technology [106] or sprayed using
ink on the carbon surface for catalysts for both anode and cathode
sides [107]. Catalyst failures can be categorized into three mecha-
nisms: (1) catalyst dissolution under non-equilibrium and sinter-
ing under load-cycling, (2) particle agglomeration via crystal
migration, and (3) detaching of catalyst due to carbon corrosion
[24,35,108].
The acceleration of degradations are related to operating con-
ditions, specially loading cycle. Catalysts degrade under large
fluctuations of potential or temperature during the start/stop pro-
cess, or under fuel starvation. Thus, numerous studies have been
devoted to understanding the degradation mechanisms of cata-
lysts occurring during PEMFCs operation and start/stop cycles to
develop efficient mitigation strategies. The cathode or anode
potential drastically increases, causing carbon corrosion under
dynamic load changes. Kinoshita et al. [109] found that the
amount of Pt dissolved per potential cycle. The dissolution reac-
tion is dependent on the number of oxidation-reduction cycles
rather than on the length of the cycle or time at oxidizing poten-
tials. Diloyan et al. [61] conducted a 300 h accelerated test with
potential cycling to study the effect of mechanical vibration on
Pt particle agglomeration and growth in the catalyst layer. The
average diameter of Pt particles under vibration is 10% smaller
than the ones without vibration conditions. Potential cycling sub-
stantially accelerates the Pt particle agglomeration process, in
which the Pt particles in the order of 2–2.5 nm have grown in
the pristine state to approximately 6 nm. The aggregated particles
may not be electrochemically active, leading to a decrease of the
active area. Tintula et al. [110] studied durability of Pt/C and Pt/
MC-PEDOT catalysts under 75 start-stop cycles. They found that
carbon oxidation is the main cause for degradation of the cata-
lysts. The Pt particle accumulation and loss of Pt from the carbon
support in the form of Pt ions and/or Pt particles are another
cause. Khorasany et al. [51] studied the sensitivity of the CCM
fatigue lifetime to the applied stress. They found that CCM was
more sensitive than that of the pure membrane, and varies signif-
icantly with environmental conditions.
Mayrhofer et al. [63] showed that whole Pt particles detach
from the carbon support and dissolve into the electrolyte with-
out redeposition after an accelerated degradation test of a
carbon-supported Pt catalyst. The increase of pore volume in
MEA can optimize the pore structure of catalyst layers and
improve the fuel cell performance, but it also causes a negative
effect on the lifetime of MEA. In the case of CCM, this treatment
can accelerate the detaching of CL from membranes causing
lower CCM durability. The detaching of the CL could be attribu-
ted to the degradation of both the recast Nafion film in catalyst
layers and the Nafion membrane. However, the increased pore
volume in CL accelerates this degradation [17]. Schulenburg
et al. [111] studied the pore structure of the Pt/C CL after 1000
start/stop cycls. They found significant loss of the active Pt sur-
face area in the cathode CL at the fuel cell cathode due to corro-
sion of the carbon support and a decrease in porosity and pore
size of the carbon support.
Poor water management can cause dehydration or flooding. The
operation under drying conditions could damage CL, such as a
decrease of the catalyst active area and an increase of charge trans-
fer resistance, whereas flooding facilitates corrosion of the elec-
trodes, Pt dissolution, blocking of the water and gas transport in
GDL and channels and gas starvation [68]. The three mechanisms
of CL detaching were proposed: (1) Separation of the catalyst layer
from the membrane due to creep deformation, (2) Separation of
the outer Nafion layer film from the core PTFE/Nafion membrane
due to creep deformation, and (3) Degradation of the Nafion plane
(or Nafion dissolution) from the PTFE surface [60,87]. Zhang et al.
[64] reviewed Pt-based CL degradation in PEMFC. Zamel [13] stud-
ied the CL and its dimensionality. Changes in the morphology,
structure or material properties significantly affects the catalyst
activity and durability.
Catalysts are likely poisoned by impurities of the feed gases
because all catalysts are susceptible to poisoning [112]. The cata-
lyst poisoning in both the anode and the cathode may reduce sub-
stantially the performance of a PEMFC. The most problematic
contaminants for Pt are sulfur species [113] and carbon monoxide
(CO) [114]. Two approaches have been used for poisoning resis-
tance: purification of fuel and the Pt alloy with other metals to
reduce susceptibility to poisoning. The nature of impurities in the
fuel stream depends on the source of the fuel. However Pt Alloy can
improve the tolerance of catalysts to poisoning to avoid any irre-
versible damage to PEMFCs. For complicated systems such as
alloys, there is no simple way to predict in advance the precise sus-
ceptibility to sulfur poisoning. Many studies [115] have been per-
formed on the interaction of sulfur species with elemental
catalysts (32–41). Unsupported or supported Pt-based alloy cata-
lysts have shown high tolerances to CO poisoning [116–118].
 J. Wang / Applied Energy 189 (2017) 460–479
3.2. Sealing, gasket and cooling
Sealing of fuel cells and good conductivity need high compres-
sive pressure to ensure sealing and contact for PEMFC. Mason et al.,
[101,102] studied the influence of clamping pressure on ohmic
resistance and compression of GDLs. High compressive stress can
cause fractures and deformation of the gasket. Jung et al. [119]
detected a higher hydrogen concentration and crossover rate at
the elevated clamping pressure and relative humidity. The stresses
on the seals can also result from thermal expansion due to temper-
ature gradients and from pressure gradients. For high temperature
fuel cells, such as SOFC, the sealing issues are more serious due to
the creep and damage under its high operating temperature. Zhang
et al. [120,121] studied creep and damage of bonded compliant
seals (BCS) in the bonded compliant seal of SOFC. They demon-
strated that the BCS structure met the long-term operation at the
high temperature of 600 °C with a proper braze bonding process.
These results show that the failure location is not in the region of
maximum creep deformation because of the influence of high level
multi-axial stress, which drastically decreases the multi-axial
ductility.
When the heat removal by the reactant and the product streams
is not sufficient, the temperature of the cell increases, causing the
membrane to dehydrate, reducing the proton conductivity of the
membrane. This leads to poor performance of the fuel cell and irre-
versible degradation. Therefore, waste heat must be removed by a
cooling system to maintain the fuel cell operation in a favorable
temperature. Many studies have been performed to improve the
cooling performance, such as optimization of the coolant flow field
design, and development of high effective coolants [122–129]. Sev-
eral cooling methods are commonly used for heat removal: (1)
using high thermal conductivity materials or heat pipes, (2) sepa-
rate air flow water or antifreeze coolant, and (3) phase change
(evaporative cooling and cooling through boiling) [122]. For exam-
ple, Özden et al. [123] designed a heat exchanger for passive cool-
ing of PEM fuel cells. Wen and Huang [124] used pyrolytic graphite
sheets to increase heat diffusion. Their results improved the tem-
perature uniformity in both the single cell and stack and increased
the stack performance while alleviating the flooding problem. It is
common to use fluids for cooling. Matian et al. [125] compared
three different cooling plates with separate channels for cooling
air to extract excess heat and maintain the desired temperature
distribution. They found that a cooling plate is an effective
approach to alleviate temperature variation among cells and min-
imize thermal problems and the temperature difference can be
minimized by implementing cooling plates with wider cooling
channels. Because liquid has much higher heat transfer coefficients
than that air, liquid coolants have widely been used to fuel cell
cooling in high power PEMFC stacks. Baek et al. [128] evaluated
the performances of six different cooling flow field designs in
terms of the uniformity of temperature distribution and pressure
drop. They found that multiple serpentine cooling flow field had
more uniform temperature distribution than the conventional
designs under same coolant pressure drops. Numerical simulation
is employed to investigate the coolant flow distribution, pressure
drop and thermal behavior of different designs. Cooling flow
fields were numerical studied by Alizadeh et al. [129] to evaluate
non-uniformity of temperature distribution. They found that an
appreciate flow field configuration can reduce much as a 50%
temperature difference.
3.3. Properties of materials and water transport in diffusion media
(DM)
The life span of a fuel cell stack using Nafion 117 has reached
continuous operations of 15,000 h at 80 °C [130]. PSFA membranes
467
Fig. 5. Schematic of droplet diffusion in DM and detachment in a channel.
are relatively strong and stable in both oxidative and reductive
conditions. Its durability of 60,000 h was reported as early as
2003 [65]. Some PFSA membranes have lasted more than wet-
dry cycles of 20,000 without failure [131]. It is clear that all com-
ponents, including MEA, CL and GDL, suffer from degradation over
time. However, under proper operating windows, the materials are
stable and degradation rates are low. Therefore, the best way to
ensure a long lifetime is for all components to operate under their
ideal operating windows.
As discussed above, there is much consensus about water and
heat transport in DM. Water and heat can accelerate degradation
of materials significantly. Water accumulation or flooding is
caused by poor capacity to transport through the diffusion media
(e.g., GDL and MPL) and channels. There are four key factors to con-
trol water transport and accumulation: (1) material properties,
such as porosity, wettability and hydrophobicity, (2) thickness of
GDL and MPL, (3) liquid adhesion and detachment in channels,
and (4) gas velocity and pressure in the gas channel.
When water generated in the CL builds up sufficient pressure,
the water enters pores in the DM (Fig. 5). Liquid water is trans-
ported by capillary and diffusion across the DM through the pores.
Once the water penetrates the pores of the DM, the liquid water is
allowed to flow into the gas flow channel if the water has not been
accumulated in the channel. However, the water drain from the
channel is depending on the force balance in the channel (Fig. 5).
If the pressure and shear force is greater than the adhesive force,
the droplet is removed by the gas. Otherwise, the water builds
up in the channel. Only a few percentages of the void fraction of
the GDL is required for liquid water transport and the smaller
pores remain free for gas diffusion of reactants from the gas flow
channel to the electrolyte/electrode interface. Once the water pen-
etrates the largest pores the liquid can drain freely due to the
applied pressure [132]. Therefore, there are two transports of
water: (1) water diffusion in the DM, and (2) water removal in
the channel. The former is dependent on the properties of materi-
als and the thickness of the DM, calculated using Darcy’s law. The
latter is depending on the gas velocity and pressure drop in the
channel for a given material of the DM.
However, the properties and thickness of DM vary widely:
thickness of carbon cloth or carbon paper of 110–500 lm, porosity
of 30–90% and air permeability of 0.5–85 cm3/(cm2 s) [133]. There
is still little quantitative and systematic information about the
range of the operating velocity and pressure drop for any specific
material and component structure under different operating condi-
tions [2].
3.4. Summary and outlook
The integration design process between components and single
cell stages is a key challenge for fuel cell scaling-up because the
individual inputs are provided by different disciplines or industries
within the scaling-up program. The designs of component struc-
ture are important, such as optimal thickness of the membrane
and GDL [93]. Some new materials and technologies have been
developed, such as high temperature membrane and multilayer
membrane fabrication, such as hot-pressing, casting and
dip-coating [14]. The state-of-the-art fuel cell membranes have
468 J. Wang / Applied Energy 189 (2017) 460–479
doubled the maximum proton conductivity and extended the oper-
ational longevity [44]. There are many experimental and modeling
studies of water transports in porous media [12,13,45,46,134].
However, the challenge for system and technology developers is
to be able to make clear, well-documented linkage of properties
and thickness of DM to the gas velocity and pressure drop in a
channel. Specifically, much work remains to be done to determine
the thresholds of the gas velocity and pressure drop in a channel
for the given properties and thickness of DM and water production
of CL because the thresholds of the velocity and pressure is neces-
sary to avoid flooding and drying for flow field designs of a single
cell.
4. Bipolar plates and single cell design
The operating parameters (e.g., temperature, humidity, concen-
tration, and flow rates) are the conditions at which the electrical or
mechanical components run. The components will operate effec-
tively within a specified temperature and humidity range from
the minimum operating temperature/RH to the maximum operat-
ing temperature/RH (or peak operating temperature or humidity).
Outside this range of safe operating temperatures/RH, degradation
of some components will accelerate, even fail, resulting in the fail-
ure of the fuel cell system. If the expected operating temperature
for a fuel cell deviates from the typical 10 °C to 25 °C range, then
this operating temperature will affect the type of materials and
components used in the fuel cell [135]. For example, if the operat-
ing temperature increases from 80 °C to 90 °C, membrane conduc-
tivity increases but deviates from optimal operating conditions of
PEMFC, leading to a faster degradation.
Single cell design is basically a design of flow field in bipolar
plates. The goals of the flow field design of a fuel cell are to: (1) find
an optimal configuration, including channel shape and length, with
uniform flow distribution and a low pressure drop to maximize the
power output and durability for given operating conditions, or (2)
find the optimal operating conditions to maximize the power out-
put and durability for a given configuration and structure [2]. A
single cell degradation is still the degradation of its components,
such as, membrane, GDL, BP, CL or sealing. As discussed in Sec-
tion 3, the failures of these components are caused mainly because
they operate outside the optimal operating windows although, in
some exceptions, due to others, such as material defects or impu-
rities of gas. Given materials and structures of MEA (e.g., CL, GDL
and membrane), the operating window of the components has
been determined according to temperature, flow rates and RH.
Therefore, a single cell design is to integrate the operating window
of materials and components with macro-operating parameters
(e.g., temperature, humidity and pressure). The design principles
of a single cell are to provide uniform gas reactant intake of each
channel across the active area in the cell and a reasonably low
and uniform pressure drop and to determine the thresholds of
gas velocity and pressure in channels to avoid water accumulation
or drying.
4.1. Water and thermal management and operating conditions
Flooding and drying can both cause significant component
degradation, as discussed in Section 3. Failure of any components
will affect the fuel cell’s life time. Thus, water management is
referred to as balancing membrane hydration with flooding
avoidance. Ishigami et al. [136] studied the uneven distribution
of oxygen partial pressures at the anode along the gas-flow chan-
nel from the inlet to the outlet using a real-time/space visualiza-
tion system. The MEA degradation mainly occurred near the inlet
and outlet of reactant gases after 1500 cycles. Furthermore, a high
pressure drop leads to high stress concentration that causes frac-
tures of the membrane. On the other hand, short channels must
use multi-channel configurations, such as parallel straight. Paral-
lel channels can reduce the pressure drop across the flow field
better than the serpentine channels by orders of magnitude
[25,26]). Furthermore, a parallel channel design is easier to man-
ufacture. However, whether parallel or multi-serpentine, channel
configurations may lead to uneven flow distribution in spite of
low pressure drop. An uneven flow distribution is a root cause
of water and thermal issues [1]. Therefore, to prevent failure of
the MEA, the flow field structure must be carefully designed to
avoid local drying or flooding of the membrane, especially at
the reactant inlet or outlet area.
The uneven liquid water production on active areas have been
experimentally studied by many researchers using various visual-
ization technologies [137]. Yoshioka et al. [59] observed a decline
of the cell voltage due to deterioration of the materials and hetero-
geneous reaction in the cell. Wang and Liu [138] measured seg-
ment current density under the channel and the shoulder in PEM
fuel cells. Reum et al. [139] and Weng et al. [140] measured fuel
cell segments to study the effect of heterogeneous humidification
on the local aging of the MEA. They divided the active area into
several segments along the flow channels and measured the influ-
ence of the local humidification of the MEA. A fluctuation in the
current distribution was observed while the current density was
higher than 400 mA cm2. More severely uneven current distribu-
tion would be observed at a lower hydrogen stoichiometry with
air starvation [42,64,138]. This is mainly because of the mixed
hydrogen and air on the CL in cases of hydrogen starvation. Lai
and Fly [141] developed stress sensors with high spatial resolution
to diagnose PEM degradation using a segmented fuel cell. Perez
et al. [142] reviewed segmented PEMFC (SFC) to study the factors
responsible for uneven electrochemical performance over the
MEA active area. All of their results demonstrated uneven reactions
on the active area.
Wu and Djilali [143] used quantitative flow visualization and a
transparent microfluidic platform replicating the salient features of
the cathode channel of a PEMFC. They investigated the dynamic
evolution of water droplets emerging from a single 50 lm pore
of a GDL into a 250 lm  250 lm air channel. The flow regimes
were found to be primarily dependent on the air flow Reynolds
numbers which ranged from 50 to 1200. At low Re, the slug flow
blocks the air flow through the channel. At higher Re, a periodic
pattern of droplet emergence, growth and detachment appears.
Further increases in air velocity induce wavy water film pattern.
Iranzo et al. [27,28] measured the liquid water distribution of the
channels of a multiple serpentine 50 cm2 PEMFC using neutron
radiographs. They found that, although there were differences in
cell voltage, when the total volumetric cathode gas flow was main-
tained for air and oxygen feed, no significant differences were
observed in the cell water content and water distribution. Further-
more, it is clear that the higher the flow rate of a channel, the less
water accumulates in the same bipolar plates and MEA. This con-
firmed that the gas flow rates are a primary factor for water accu-
mulation in a given flow field configuration, and components (e.g.,
material properties and structures). The influence of the electrode
permeability on the PEMFC performance depends strongly upon
the flow-field configuration [144]. Flow field design was an effec-
tive way to improve the performance and durability of PEM fuel
cells [1,2,145,146].
Thermal management is also important. Generally, PEMFC sys-
tems should run at a maximum of 80 °C because operation above
that temperature requires too much system support of the mem-
brane (e.g., pressure, humidification) to maintain good proton con-
duction [131]. Moreover, the glass transition temperatures of
conventional membranes are in the range of 80–120 °C. Above that
 J. Wang / Applied Energy 189 (2017) 460–479
Fig. 6. Six flow fields by Wang and Wang: (a) single serpentine, (b) two-path serpentine, (c) three-path serpentine, (d) six-path serpentine, (e) parallel and (f) interdigitate
[26] (Reproduced by permission of Wiley).
Fig. 7. Comparisons of dimensionless flow distributions and pressure drops among
different configurations: (a) flow distribution and (b) pressure drop [26] (Repro-
duced by permission of Wiley).
469
temperature range, the membranes are subject to creep and
pinhole-formation.
4.2. Theory of flow field designs for bipolar plates
Despite the many models of flow fields that have been devel-
oped using analytical and computational fluid dynamics
[45,46,147–149], these models are not a theory or tool for flow
field designs of bipolar plates. A theory and tool needs a strong
capacity to meet the goals of bipolar plate and single cell design.
They must: (1) find the optimal structure for the given operating
conditions and (2) find the optimal operating conditions for the
given structure, as mentioned above [2]. Wang [1,2,25,26,150–
155] has done a series of researches to develop such a theory or
tool and remarkable progress has been made. His theory estab-
lishes a relationship between flow conditions, configurations and
performance (e.g., flow distribution and pressure drop) and can
be used to evaluate different design alternatives. The comprehen-
sive reviews of flow field designs have been carried out by Wang
[2].
Flow field design of a bipolar plate includes two areas: gas flow
channels and flow field layout of bipolar plates. The design of gas
flow channels is to assess how high a flow velocity is needed to
remove the water in the GDL and channel and how low a velocity
needs to be to avoid drying. This threshold flow velocity and pres-
sure are used for determining the operating velocity and pressure
drop in the design of bipolar plates [2].
The theory of flow field design has demonstrated that low pres-
sure drops and uniform flow distributions are two of the key per-
formance index needed to achieve the above goals [25,26]. For a
given active area, so-called optimal designs are a compromise of
pressure drop and flow distribution, which is related to the number
and length of multiple parallel channels. Wang [1,2] pointed out
that various failures of scaling-up are likely from the root cause
of uneven flow distribution or high pressure drop. Wang and Wang
[25,26] have compared flow distribution and pressure drop of six
different configurations (Figs. 6 and 7). Long channels will lead to
high pressure drop and therefore are highly effective at removing
liquid water but require high pumping power. This may cause
470 J. Wang / Applied Energy 189 (2017) 460–479
issues of inlet drying and outlet flooding and lead high concentra-
tion differences between inlet and outlet. On the other hand, short
flow channels like parallel configuration have a small pressure
drop between the inlet and outlet but have uneven flow distribu-
tion, resulting in accumulation of liquid water and flooding. Some
experiments have confirmed these results [145,146,156,157].
4.3. Contact resistance
As the current conductors between cells, the BP requires high
current and thermal conductivity to conduct electrical current
from cell to cell, to remove heat from the active area, and prevent
leakage of gases and coolant [158–161]. A graphite composite is
commonly used for the BP materials because of its high corrosion
resistance and good surface contact resistance. However, due to
the problems in their manufacturability, permeability, and durabil-
ity for shock and vibration, many metal BPs have been developed
[161]. For the metal BPs, metallic corrosion and metal oxide forma-
tion can cause unacceptable increases in contact resistance. Metal
corrosion and dissolution lead to possible fouling and contamina-
tion of the catalyst and the ionomer [160–162]. The contact resis-
tance between the BP and the GDL may constitute a significant
fraction of the overall fuel cell electrical power. For example, stain-
less steel is an attractive material for bipolar plates. However, it is
prone to having a high contact resistance due to oxide formation
and metal corrosion, which is a major concern. Studies show that
contact resistance may improve tenfold with measures, such as
surface topography improvement and coating, depending on sub-
strate surface treatment and roughness [160,162]. The clamping
force can affect significantly the interfacial contact resistance and
the porosity of GDL [163]. The void volume is changed with the
deformation of the GDLs as the clamping force changes. A low con-
tact resistance can be achieved under a constant contact pressure
distribution and good porosity for the GDL or optimal rib width
of the bipolar plates [164].
4.4. Summary and outlook
Single cell design integrates materials, components, flow field
and performance (e.g., flow distribution and pressure drop). It is
central to controlling a single cell to operate within the operating
window of materials and components. Although new material
property and structure improvement of components can widen
the operating window in component stages, flow field designs
could be a more effective step in avoiding local water and heat
issues through high pressure drop and non-uniform flow distribu-
tion. An uneven flow distribution and high pressure drop can cause
local deviation of the operating conditions from that of its compo-
nents in a cell. Thus, even if a fuel cell is designed to operate under
the operating window of the components, some its channels may
exceed the designed window of the cell, leading to serious water
and heat issues as well as performance degradation [1,2]. In spite
of many studies of CL, MEA, and GDLs and water and heat manage-
ment [45,46,136–140], there is still little quantitative information
of a systematic relationship among water and heat, structures, flow
rates in channels and power density [1,2]. More fundamental
research of liquid water formation, transport and removal by gas
streams in GDL, MPL and channel remains to be done to determine
the maximum and minimum flow velocities thresholds of flooding
and drying for a given MEA and channel structure. The two thresh-
old values can be used to determine the maximum velocity and
pressure drop of a system to avoid drying and the minimum veloc-
ity and pressure drop to avoid flooding. The two thresholds provide
an operating range (or an operating window of the single cell). It
should be noted that the operating window of the single cell is dif-
ferent from that of all its components.
A technological breakthrough will likely occur through such a
system integration that is well grounded in the quantitative and
systematic knowledge of these two operating windows. This can
establish a connection among the operating windows of compo-
nents and single cells, flow field design and program of a system
control.
5. Flow field and stack design
Many studies have been carried out at the stack scale. As an
assembly of identical cells, the principle of a stack design is to
achieve two goals: (1) the maximum power output of a fuel cell
stack (ideally a linear sum of all individual cells in the stack), and
an identical lifetime and durability as that of its worst individual
cell [2]. Such a design of duplicate cells is based on a supposition
that all cells will perform identically since they use the same mate-
rials, seals, catalyst, structures, electrochemical processes and
work under same operating conditions [1]. However, in practice,
all stacks show more or less uneven performance decay of individ-
ual cells in a stack. There are two causes of non-uniformities in the
stack scale: (1) uneven flow distribution and high pressure drop
and, (2) transient loading cycle.
The theory of flow field designs for the bipolar plates is suitable
for the design of stack manifolds since they operate the same as
that of flow distribution in manifold systems.
5.1. Performance test of fuel cell stacks
Because it is difficult to test a fuel cell under its practical situa-
tions for several thousand hours to determine durability of main
components and stacks, accelerated tests have been developed
for fuel cell development. The joint research council (JRC) [165]
has recommended four different sub-tests: (1) normal efficiency
test, (2) thermal load cycling, (3) electrical load cycling, and, (4)
start-up/shutdown [38,83,166–169]. U.S. DOE has established a
test protocol for durability [38]. A typical accelerated test operates
at an open circuit voltage, low RH (25–50%), and elevated temper-
ature (80–95 °C).
Aside from the non-uniformities of the active area in a single
cell, the stack suffers from non-uniformities among the individual
cells [129]. Eckl et al. [170] measured performance of a 300 W PEM
fuel cell under extreme operating conditions, including complete
MEA drying-out or flooding. They found that the degradation of
the individual cells in their stack were not equally affected by
dehydration. The differences became more pronounced over time
and the cell voltage distribution became increasingly less uniform
with the aging process. The degradation issues were attributed to
uneven flow and temperature distributions. Their results showed
that membrane modification and cell structure adjustment can
reduce the methanol crossover in the cathode and improve the cell
performance. Mocotéguy et al. [171,172] carried out both steady-
state and dynamic tests using a 24-cell stack. They found uneven
performance decay as well and the non-uniformity of the cell volt-
age distribution increased in the aging process. The start/stop
cycling caused a higher degradation rate than was found in
steady-state operation. Chen et al. [173] studied the effect of
inlet/outlet flow distribution of a stack with 10 cells. Manifold
radius of greater than 5 mm resulted in more uniform flow distri-
bution within the stack. Yeh et al. [174] measured power output of
26-cell DMFC stack for a long term durability of 5000 h. Pei et al.
[175] tested performance degradation of a PEMFC stack of 100 cells
after a 500-h accelerated lifetime test. After 500 h of accelerated
lifetime test under conditions of idling cycle, load changing cycle,
high power cycle, and start-stop cycle, the different performance
degradation of individual cells were clearly observed. The average
 J. Wang / Applied Energy 189 (2017) 460–479
Fig. 8. Mechanical failure process of the swelling-induced stresses: (a) MEA in single cell with hydrothermal boundaries, and (b) mechanical failure mode [71] (Reproduced
by permission of The Electrochemical Society).
cell voltage curve of the first fourteen cells was lower than other
cells. Giddey et al. [176] studied 1 kW PEM fuel cell stack in stages
of 2-, 4-, 8- and 15-cells. The 15-cell stack reached an electric effi-
ciency of 41%. But, with external thermal insulation, it produced
combined heat and power efficiency of 80%. The average ohmic
resistance of cells was measured after 40 cold start/shut down
thermal cycles and 300 h accumulated operation over a period
of more than one year. The ohmic resistance values increased from
0.328 to 0.428 O cm2 while the 2- and 4-cell stacks achieved lower
values of 0.251 and 0.228 O cm2, respectively. Bonnet et al. [177]
found that the polarization curves were similar for both a single
cell and a 5-cell pilot stack before and after operation at driving
cycle conditions for 550 h. The influence of both air and hydrogen
flows on single cells and pilot stack were similar. The gas flows did
not affect the cell voltage but did affect water management. Chu
and Jiang [178] compared a single cell and a 30-cell stack. They
also found similar polarization curves for 10% and 90% RH.
Humidity has a stronger influence on the mechanical response
than temperature because swelling due to water uptake in GDLs
and membrane is much higher than thermal expansion. Schmit-
tinger and Vahidi [68] showed that the main causes of short life
and performance degradation were poor water management, fuel
and oxidant starvation, corrosion and chemical reactions of cell
components. Rodatz et al. [179] attributed variations in cell volt-
ages to several factors, including uneven flow distributions of reac-
tant gases, temperature, and cell defects. It was more difficult to
achieve water balance in a stack than in a single cell [175]. Ishi-
gami et al. [136] pointed out that single cell performance could
not represent alone to fully derive the expected performance of a
PEMFC stack, due to the non-uniformity in potential, temperature,
and reactant and product flow distributions. Rodatz et al. [179]
recognized that monitoring stack voltage did not allow them to
distinguish between the slight deterioration in all cells and the
failure of a single cell. They observed that the failure of a single cell
in a 100-cell stack reduced the stack voltage by only 1%. Although
the effect of the non-uniformities on the efficiency is small,
non-uniformities in single cell potential along a stack are indica-
tors of localized defects. Specifically, the effect may be amplified
due to blocking channels and GDLs. Wang [1,2] concluded that
the root cause of water and heat problems was the uneven flow
distribution and dynamic loads.
471
5.2. Transient load and start/stop cycle
For automobile applications, it is unavoidable that fuel cells
operate under load changing cycles, such as start/stop. Therefore,
steady-state degradation rates may not be truly representative
for transient systems. Under dynamic conditions, fuel cells will suf-
fer from load cycling between high- and low-power densities. This
can accelerate significantly degradation. Particularly, the start/stop
cycle can lead to membranes and GDL swelling when soaked in
water and subsequently shrunk during drying. Due to mechanical
constrains of the membrane and GDLs from bipolar plates, the
membranes and GDLs undertake in-plane compression during
swelling when wet conditions and in-plane tension due to shrink-
age during dry conditions. The repetitive processes of swelling and
shrinking lead to fatigue that may eventually lead to cracking of
the membrane. Kusoglu et al. [56,70,71] carried out numerical sim-
ulations of the mechanical response of a PEM during fuel cell oper-
ation (Fig. 8). Their results demonstrate that compressive stresses
developed as hydrothermal loading changed. This stress can even
exceed the yield strength, causing permanent deformation.
When the fuel cell operates at sub-zero temperatures for an
extended period of time, the residual water in membrane, CL, GDLs
and channels freezes. Due to the expansion of the freezing water,
this process causes thermal and mechanical stress. The repetitive
freezing and thaw of water in these components causes creep
and deformation of the cell components or may even cause physi-
cal breakdown. The repetitive cycles on the MEA can delaminate
the CL from both the membrane and the GDLs [180–182]. More-
over, the volume change causes change in the pore structure and
increase the contact ohmic resistance and decrease the electro-
chemical active surface area of MEA. Additionally, membranes
exhibit viscoelastic creep to relieve these stresses after several
freeze/thaw cycles (from 30 to 20 °C) [68]. This may eventually
lead to pinholes caused by local thinning and cracking of fully
hydrated membranes. As a demonstration of the purely mechani-
cal failure at extreme conditions, Mathias et al. [131] tested a
50 cm2 MEA under a humidity cycle between 150% RH for 2 min
and 0% RH for 2 min at 80 °C under air/air to investigate mechan-
ical degradation of membranes without chemical degradation.
Gas crossover leaks due to membrane failure was measured using
air permeation at an applied differential pressure of 20 kPa. Cracks
472 J. Wang / Applied Energy 189 (2017) 460–479

in the membrane cause gas crossover. The gas crossover causes

uncontrolled reaction between hydrogen and oxygen, leading to
subsequent pinhole formation and damage to the membrane,
reducing the life of the cell. While localized fuel starvation can
happen during normal operation due to poor distribution of reac-
tants or to water flooding, the start/stop cycles lead to the similar
local gas starvation. Therefore, mitigation of performance degrada-
tion during start/stop cycles is essentially water and thermal man-
agement. The lower the hydration state of the membrane after
shut-down, the less serious the freezing.

5.3. Summary and outlook

Knights et al. [183] reported an endurance a 17-cell stack com-

prised of 250 kW over 13,000 h operation. Verhage et al. [184]
demonstrated the life time of a fuel cell operating over 30,000 h
as a fixed power sources without replacement of components. Smit
[185] reported 40,000 h of operation without replacement of major Fig. 10. Narrowing of operating windows due to uneven flow distribution [2]
system components. Despite the fact that many stacks have (Reproduced by permission of Elsevier).
reached the DOE targets for durability, these stack system are sel-
dom in operation in the open market. A possible explanation for
this is that only few of the numerous, promising theoretically pro- As discussed above, many experimental results show that cells in
posals have led to practical solutions for real systems. Therefore, it a stack do suffer from uneven degradation [130,136,170–172,175,
may be difficult to reproduce their high performance if the uneven 179]. These experimental results are in strong agreement with
flow distributions remain unsolved. However, fuel cells for station- the theoretical ones [1,2].
ary applications demonstrate a possibility of achieving 5000 h for
automotive applications, although those applications are more
6. Control of system
complex due to continual start/stop cycles and load changes. Under
the operation of dynamic loads, a fuel cell stack also causes uneven
Any failure of a fuel cell system is caused by the failure of one or
flow distribution along the stack, leading to a narrower operating
several components within the stack. The failures of the compo-
window than that of its components or its individual cells because
nents happen primarily because the operating parameters exceed
of the superposition of uneven flow distribution among several
the local operating window of materials. As the PEMFC stack itself
manifold levels (Fig. 9) and load change cycles [2]. These issues
forms a complex system with a magnitude of interactions, each
can be amplified by water blockage in some channels or cells [1,2].
operating parameter must be considered in a systematic context
For given materials and structures of components, a maximum
[170]. Therefore, the objective of a control system is to ensure that
velocity to avoid drying and a minimum velocity to avoid flooding
all materials (e.g., membrane, GDL and CL) work within their oper-
can be determined in channel scale tests. Although the two thresh-
ating windows (e.g., under the ranges of temperature, humidity
old values do not vary for the same materials and structures of
and pressure). However, it is difficult to keep all components in
components, uneven flow distributions make some channels in a
the stack working within the operating window because of uneven
cell or some cells in a stack exceed their operation thresholds
flow distribution and load change. The range of the narrowed oper-
(Fig. 10). Fig. 10 shows the flow fields of three types of bipolar
ating window will be decreased by the overlapping of uneven flow
plates and the two threshold velocities. A proper design of bipolar
distributions among channels and individual cells or long channel,
plates can significantly improve flow fields and keep all the chan-
leading to a deviation from the operating window of the compo-
nels working within the operating window, such as 4-U type flow
nents (Figs. 8 and 9). Therefore, special attention should be paid
field design. Thus, a maximum pressure drop can be determined in
to the heterogeneous aging and performance degradation due to
the stack scale to avoid drying and a minimum pressure drop to
the uneven flow distribution for design of control programs.
avoid flooding. The operating window of a stack is different from
that of its individual cells due to its uneven flow distribution. It
should be noted that the operating window of the single cell is 6.1. Stack operating windows
much narrower than that of all its components and the operating
window of the stack is much narrower than that of all its cells. In order to maintain good proton conductivity, the water con-
tent in the membrane must be kept at a certain level to ensure that
the membrane remains fully hydrated. Thus, all the cells in a stack
must operate within the optimal operating windows that are based
on good proton conductivity as well as the thermal and mechanical
strengths of the membrane and GDL and CL. However, the water in
the pores of the CL, GDL and the membrane may accumulate
locally once the flow velocity in the channel is lower than the min-
imum threshold or starts drying once the velocity is higher than
the maximum threshold. The difficulty of a system control results
from the heterogeneous distribution. Therefore, favorable operat-
ing conditions can vary widely from cell to cell and depend on
the stack architecture and applications.
As long as the system operates outside the optimal operating
window, the degradation rates can increase by orders of magnitude
Fig. 9. Narrowing operating windows from components, individual cells to stack. when conditions include some of the following: load cycling,
 J. Wang / Applied Energy 189 (2017) 460–479
start–stop cycles, and uneven gas intakes. It is key to develop a
dynamic multi-physical fuel cell stack model, which should
encompass electrical, fluidic, and thermal domains [186]. Various
mitigation strategies have been developed that can improve the
degradation caused by cyclic loading operations, however, such
mitigation strategies can cause other durability issues and they
are challenging to implement, especially when the fuel cell stops
for an extended period of time [72]. Therefore, it is necessary to
take both uneven flow distribution and load change into account
during the development of programs of a system control. Such a
control is possible for optimized operation of the stack only if
excessive dryness or humidity is prevented by carefully adjusting
the operating conditions and procedures. Many researchers have
developed various process modeling, optimization and control pro-
grams for current management and system performance [187–
191]. But, in spite of many successes, the current design for stack
control system is mostly approached from an empirical angle due
to the lack of understanding of the mechanisms of failure pro-
cesses. However, these common empirical assumptions are chal-
lenged and disproven. We need to solve the problem of how
uneven flow distributions can be mitigated through flow field
architecture and system design based on an operating window,
and how a multidisciplinary system integration and operation con-
trol can help in making fuel cell stack reliability a reality. How are
the key factors and procedures identified when putting design the-
ory and control system to practice?
This requires the evaluation of how far a fuel cell deviates from
its design operating conditions. Therefore, Wang’s model
[2,25,26,155] is based on the first principle of mass and momen-
tum conservation to build a relationship among configuration
and performance (e.g., flow distribution and pressure drop) and
is a general design tool for manifold systems. The main design phi-
losophy is to ensure all the channels in a cell or all the cells in a
stack work in the design operating conditions through a design
of uniform flow distribution and proper pressure drop. Because
the design of uniform flow distribution can calculate the velocity
and pressure drop in every channel, the velocity and pressure drop
in each channel can compare with the threshold velocities and
pressure drops to water removing and drying for specific struc-
tures and properties of the components (e.g., CL, GDL and MPL).
If the velocity and pressure drop in channels deviates from the
threshold velocities and pressure drops, the flow fields must be
resigned until they meet the design tolerance. Even if the control
program allows for setting the operating parameters only within
a relatively narrow range, MEA drying-out and flooding can be
avoided if we know the range of the operating window of the stack
better. The current models can be specifically designed so that they
can be used to determine the fundamental thermal–physical
behavior of a fuel cell stack for program design of process and sys-
tem control.
6.2. Startup/shutdown cycles
The practical conditions for automotive uses are more complex
due to cyclic power demands such as startup/shutdown, and
freeze/thaw, which increase the severity of degradation due to
water and thermal problems and transient starvation in fuel [5].
Such transients are stressful to the cathode catalyst layer, espe-
cially for the carbon substrate supporting the Pt-based nanoparti-
cles [9]. The designed operating conditions are greatly
exacerbated by the start/stop cyclic change of the operating condi-
tions, leading to physical and chemical changes, sometimes with
catastrophic results. Yu et al. [42] reviewed performance degrada-
tion of PEMFC during startup and shutdown processes. During
start/stop cycles, some areas became transiently starved for fuel
or air. Such dynamics are especially harsh for the cathode CL with
473
the carbon-substrate of the Pt-based electro-catalyst particles. This
results in local potentials on the air electrode significantly higher
than 1 V and, thereby, induce irreversible corrosion into CO2 of
the carbon supports. This results in permanent loss of electrochem-
ically active area [9,181]. The cell potential can remain in the range
of expected conditions even as this condition persists, and this ‘‘re-
verse” current mechanism can induce damage to the cathode with-
out being directly observable [35]. The start/stop cycles also cause
uneven current distribution at a high current density while provid-
ing a uniform current distribution at a low current density [42,64].
In case of a prolonged shutdown, air can cross over the membrane
into the hydrogen channels in the anode chamber unless the stack
is continually provided with hydrogen. In this situation, the start-
ing flow of hydrogen will cause a gradient of hydrogen in which
hydrogen exists at the inlet but the exit is still hydrogen-starved.
As a result, starting and stopping the fuel cell can induce consider-
able damage to the cell [35]. Load change is one of the main causes
of gas starvation of fuel cells in vehicular application [192].
When at cold climate conditions, the residual water in the pores
of the CL, GDL and the membrane may freeze once the cell temper-
ature decreases below the freezing point [193]. The freezing water
needs to be thawed if the cell or stack start-up under the freezing
temperature. Even a little freezing water in the pores of MEA can
cause a decrease of active area due to blocking of gas transport
and can lead to severe degradation of the fuel cell. Methods that
could enable a fuel cell startup with no or only slight performance
degradation at subfreezing temperature have been studied
[42,194–196] carried out dynamic modeling, optimization and con-
trol of power density in a PEM fuel cell. The optimal operating range
can encompass operating temperatures from well below the freez-
ing point to above the boiling point of water and operating humidity
levels ranging from dry to wet, such as low humidification or humid-
ification cycling, temperatures of 90 °C or higher and fuel and air
starvation due to uneven flow distribution [41].How then do we
control a fuel cell system at its optimal operating conditions with
uneven flow distribution, dynamic load following, start/stop opera-
tion, road vibration/shock, and ambient conditions?
6.3. Gas and air purging
Repetitive freezing processes can significantly accelerate per-
formance degradation [171–180]. Three main methods to solve
the flooding and freezing problem during shutdown at cold climate
conditions are: (1) opening the pulse valve, allowing the hydrogen
gas to purge the liquid water out of flow channels, (2) raising the
operating temperature to decrease the relative humidity of the
reactant gas, and (3) anti freezing solutions.
Any freezing is because of residual water in the CL, GDLs, mem-
brane and channels after shutdown. Thus, since gas purging is cap-
able of removing the residual water from the CL, GDLs and
channels during shutdown, the gas purging by feeding dry nitrogen
and air gases to the anode and the cathode chambers, respectively
is an effective approach, prior to the cell temperature falling down
below 0 °C. This approach has been adopted by many engineers
[197,198] to prevent performance degradation in fuel cells. Some
control procedures of start/stop have been suggested [198,199].
Arthur et al. [198] suggested a fuel cell operational program for
additional hydrogen after stop. Reiser et al. [199] performed a
long-term durability test. The results showed an average voltage
loss of 0.2 V without air purging after about 250 start and stop
cycles. However, the performance decay was not so severe by using
air purging, with only about 0.055 V voltage loss after nearly 500
cycles.
Anti-freezing solutions, such as 30% methanol or 35% ethylene
glycol solution, can be used for purging fluid instead of the dry
gases. Using anti-freezing solutions reduced the performance
474 J. Wang / Applied Energy 189 (2017) 460–479
degradation, as reported by Cho et al. [179,180]. Ang et al. [200]
carried out fuel cell systems optimization. Successful cold startup
and stable operation at 20 °C should be taken into account
[201]. A passive cold start-up may not be viable approach. An
unprotected startup can cause local potentials on the cathode in
excess of 1.8 V relative to a hydrogen reference electrode.
6.4. Summary and outlook
As discussed above, the current system controls have consid-
ered startup/shutdown processes. It is clear that the layout of the
flow field can have a significant influence on the water removal
capability of a PEM fuel cell for the given materials and structures
of MEA. However, the degree that the layout affects water removal
capability depends on the flow field configurations and operating
conditions. Therefore, manipulation of the operating conditions is
a very common strategy to mitigate flooding. These strategies
include: increasing reactant gas flow rate well above stoichiomet-
ric levels to promote the capacity of water removal and advection,
increasing temperature for evaporation, and purging the cathode
periodically with momentarily high air flow rate. However, these
strategies use significant parasitic power that is directly linked to
operating conditions, (e.g., humidity, volume flow rate and pres-
sure drop), or that adds to the fuel cell system complexity. For a
control program design, it is still not sufficient to achieve the max-
imum power output since local water and thermal issues may not
be considered due to uneven flow distribution. Because of a lack of
quantitative understanding of how uneven flow distributions over-
lap among channels, cells and stack, it is difficult to take them into
account during the program design of process and system control.
All these are required to fundamentally understand how configura-
tion and operating conditions affect flow distributions and pres-
sure drop in a stack to optimize the operating control conditions
for a specific flow field layout. In recent years, Wang
[1,2,25,26,152–155,202] established a theory of flow field designs.
This theory provides a potential approach to link the operating
window of components, flow field designs and programs of operat-
ing and control. Much work remains to be done in this field.
7. Significance with application to fuel cell scaling-up
This paper breaks down the technical barriers of durability and
reliability into component, cell, stack and system control stages
and identifies knowledge gaps among stages of the fuel cell devel-
opment. These knowledge gaps and technical challenges are differ-
ent from the existing understanding while the existing challenges
in literature are rooted to highly complex mechanisms, involving
multiple chemical and physical interactions at the atomic-or
nano-level [36,203]. These critical issues and challenges that
remain unresolved has received little attention but provides
insights where the current research should head or change through
analysis of system integration and technology readiness to this
problem of durability and reliability in the near and far future. Spe-
cially, although the interdisciplinary collaborations have been real-
ized extensively, it has been lack where/how scientists and
engineers can collaborate from different disciplines and different
development stages. Fully understanding of these challenges pro-
vides directions through the connection points towards to durabil-
ity and reliability issues and can make it effective and efficient to
collaborate.
7.1. Knowledge gaps and challenges
There are knowledge gaps between the stages of components
and individual cells. The flooding or drying conditions of mem-
brane and DM are dependent on properties and structures of mate-
rials such as GDL porosity and wettability and GDL thickness. The
higher the porosity and the thinner the DM, the easier the water
diffuses in the DM. However, the detachment of droplets in a chan-
nel are determined by the balance of shear force and adhesion of
the wall. The main force of droplet shear is flow velocity and pres-
sure drop in the channel for a given water production. Thus, the
maximum pressure drop and gas velocity in a channel needs to
be determined to avoid channel and membrane drying and the
minimum pressure drop to avoid channel and GDL flooding. Partic-
ularly, these two thresholds will increase as water production
increases. However, how high is the maximum flow rate in a chan-
nel to avoid membrane drying and how low is the minimum flow
rate to avoid channel and GDL flooding for any given channel sizes
and material properties and structures of components? Although
many models and experiments have been performed on two-
phase transport in porous media [12,45,46,204–207] and two-
phase flow in a channel [45,46,147,208,157,209,210], there are
no any answer to the above question. Much work remains to be
done to establish such quantitative relationships to determine
these two thresholds between component and single cell stages.
Knowledge gaps exist between stages of individual cells and
stacks. The two thresholds depend on flow velocity and material
properties and temperature and humidity. The two thresholds will
change as the operating conditions of temperature and humidity
vary. Therefore, a manipulation of the operating conditions is an
effective approach to mitigate water and thermal problems, and
some strategies to address durability and reliability issues are suc-
cessful, such as increasing reactant gas flow rate well above stoi-
chiometric levels to promote the capacity of water removal and
advection, increasing temperature for evaporation, and purging
the cathode periodically with momentarily high air flow rate. A
proper design of bipolar plates can significantly improve flow fields
and keep all the channels working within the operating window,
such as a better design of 4-U type flow field (Fig. 10) [2]. However,
there is currently little quantitative and systematic information
about how superposition of uneven flow distributions causes the
deviations of local flow rates from the designed ones.
Knowledge gaps exist between the stages of stack integration
and system control stages. For example, how do the uneven flow
distributions of channels, individual cells and stacks overlap and
affect the two threshold velocities? Does a change of operating
parameters affect the local uniformity? Therefore, the operation
program of the current system controls are still not sufficient to
address issues of durability due to uneven flow distribution even
though optimal power output can be obtained. This is essential
for the next stage of the process and system simulation and
control.
7.2. Three operating windows with application to the scaling-up
We propose the three operating windows that build connection
points among components, flow field design, individual cells, stack
and program design of process and system control to address dura-
bility and reliability issues. It should be noted that the operating
window of a single cell is much narrower than that of all its com-
ponents and the operating window of a stack is much narrower
than that of all its cells. The concepts of the three operating win-
dows and theory of the flow field designs clarifies key ambiguities
and converges our future directions to address the issues of dura-
bility and reliability. These different windows have implications
how to build a connection among properties of material and struc-
tures of components, such as wettability, and porosity, perfor-
mance of a stack, and macro operation conditions, such as
pressure, humidity and flow rates.
 J. Wang / Applied Energy 189 (2017) 460–479
The results can be applied to R&D of fuel cells through four
interactive stages. In Stage 1, new materials and structures of com-
ponents are developed to widen the operating windows of the
components (e.g., membrane, GDL and CL). For example, the GDL
thickness could be reduced to transport water if its stiffness is suf-
ficient. Alloying catalyst could be resistant to poisoning. However,
it is more important to determine the threshold velocity and pres-
sure drop of water removing and membrane drying for any given
properties and structures. However, much effort is required toward
developing experiments and a coupled model of water and heat
transports occurring simultaneously in different DL materials and
structure and different channel shapes. For specific materials and
structures of CL, GDL and MPL, the threshold velocity and pressure
can be determined using experiments and high resolution models
such as lattice Boltzmann methods (LBM) and computational fluid
dynamics (CFD). Such models or experiments are extremely useful
for the operation regimes of the fuel cell system to maintain
healthy water balance, and improving high-power density perfor-
mance. It is expected that the model development efforts of water
transport in a combined structure of DL and channel will be direc-
ted more toward determining the threshold velocity and pressure
in the channel for various properties and structures of components
and different channel shapes. However, it is still difficult to build
relationships among properties and structures of components
and the threshold velocities and pressures under different proper-
ties, structures and power densities using these high resolution
models (e.g., LBM and CFD) or experiments. The relationships are
necessary to solve water and heat problems and are of interests
for fuel cell developers. An alternative approach is to build tables
or graphs of these relationships among properties and structures
of components and the threshold velocities and pressures under
different power densities using experiments.
In Stage 2, the threshold velocities and pressures provide a
range for the operating window of single cell and are constraint
to every channel in a bipolar plate and single cell. It is clear that
some channels may deviate from the design operating window
due to uneven flow distribution [1]. Design and development of
an individual cell should ensure uniform flow distribution and
low pressure drop to maintain the operating windows as good as
possible under their range of the operation and to avoid local devi-
ations. Wang’s model [2] can calculate the velocity and pressure
drop in each channel and, therefore, provides a tool to evaluate if
the operating parameters are within the operating windows and
if any design tolerance exists. Wang’s models also provide a direc-
tion of how to adjust structures of bipolar plates and operating
parameters to fit the channel velocity and pressure. However, this
model has not been exploited because no any threshold velocities
and pressures with material properties and structures of DL in a
channel are available now.
In Stage 3, because an absolute uniform distribution is not pos-
sible at all channels of a single cell, the uneven flow distribution of
a stack can superpose with that of its cells, resulting in a narrower
operating window. The superposition of uneven flow distributions
in a stack could be used for determining the operating range in the
stack design. Finally, the design tolerance of the flow field design in
the stack can provide a range of operating parameters and unifor-
mity for the program design of process and system simulation and
control. The system control can consider both optimal power out-
put and issues of the uneven flow distribution.
7.3. Summary and outlook
The components development is required to correlate the mea-
surable properties structure and the performance of cells through
the threshold velocities and pressures in channels. The threshold
velocities and pressure need to be determined in terms of the
475
material properties and interfaces of the GDL/gas channel which
significantly influence the overall water and heat transport and cell
behavior. These relationships and connection points occurring in
different stages of fuel cells can be directly utilized to address
the issues of durability and reliability in cell design and product
development only if they are available. Iteration among four stages
can create technical measures and design criteria to substantially
improve durability, and reliability and accelerate the commercial-
ization of fuel cells in the near future. Furthermore, the accuracy of
Wang’s model is depending on the sensitivity of the frictional and
the momentum factors due to the flow branching these coefficients
[25,26]. The resulting uncertainty needs to be evaluated due to the
superposition of uneven flow distributions. Therefore, much
remains to be done to build these quantitative relationships.
8. Concluded remarks
In the last decade, the durability and reliability of fuel cells have
been studied extensively from component, individual cell, and
stack scale. However, the current results, data, even assumptions,
and guidelines are separated and unconnected, even inconsistent.
This paper is a first attempt toward understanding and analyzing
the massive but wide spread work, which has been done and
reported in the literature, on fuel cell performance, reliability and
durability to identify main macroscopic factors that affect degrada-
tion and performance, such as properties (e.g., porosity, hydropho-
bicity and wettability), structures (e.g., thickness and shapes) and
operating parameters (e.g., velocity, pressure and humidity). We
breaks down the technical barriers of durability and reliability into
component, cell, stack and system control stage and propose three
operating windows of components, cell and stack through an anal-
ysis of the scaling-up procedure and key design parameters. We
found that the operating window of a single cell is narrower than
that of all its components and the operating window of a stack is
much narrower than that of all its cells.
In particular, we identify connection points among different
stages through the threshold velocity and pressure of the channels
to connect the properties and structures of components with indi-
vidual cells. This points out a way that are currently feasible in
science and technology to determine the range of the threshold
velocities and pressures of the three operating windows associated
with their relationships in design, optimization, and control of fuel
cell systems. Iteration between models, experiments, chemistry,
materials, control, design and manufacturing can create technical
measures, thereby significantly moving a component, a cell, or a
stack from one technology readiness stage to another.
It is expected that these connections or relationships among dif-
ferent stages can be best built through a combination of funda-
mental modeling, detailed experimental diagnostics and
fabrication to determine the range of operating windows to ensure
the fuel cell design to work in the operating windows. However,
much remains to be done to build these general relationships
and design criteria because of a huge combinations of properties,
structure and channel shape. Materialists and chemists who
develop components would collaborate closely with engineers
and developers who design and develop flow fields, structures
and system since only strong collaboration can bridge the knowl-
edge gaps connected with progress in this multidisciplinary field.
At a national or regional level, funding agencies, scientific acade-
mies, designers and engineers need to bring together expertise to
support this interactive and iterative process. For some academies
and industries, it is possible to build such relationships and design
criteria of some specific materials and structures for their own fuel
cells, thereby accelerating the improvement of durability and reli-
ability. Further research directions and procedures have been
pointed out at the end of each section for fuel cell scaling-up.
476 J. Wang / Applied Energy 189 (2017) 460–479
Acknowledgements
The author would like to thank the support of the Alberta
Economic Development and Trade for Campus Alberta Innovation
Program (CAIP) Research Chair (No. RCP-12-001-BCAIP). The
author would also like to thank the editor and three anonymous
reviewers for your constructive comments and suggestions and
thanks Jim Sellers for English proof-reading.
References
[1] Wang J. Barriers of scaling-up fuel cells: cost, durability and reliability. Energy
2015;80:509–21.
[2] Wang J. Theory and practice of flow field designs for fuel cell scaling-up: a
critical review. Appl Energy 2015;157:640–63.
[3] Upreti G et al. Fuel cells for non-automotive uses: status and prospects. Int J
Hydrogen Energy 2012;37(8):6339–48.
[4] Department of Energy. An integrated strategic plan for the research,
development, and demonstration of hydrogen and fuel cell technologies;
2011 <http://www.hydrogen.energy.gov/guiding_documents.html>
[accessed 29.05.16].
[5] Department of Energy. Technical plan – production. <https://www1.eere.
energy.gov/hydrogenandfuelcells/mypp/pdfs/production.pdf> [accessed
29.05.16].
[6] Department of Energy. FreedomCAR Fuel Cell Technical Team Roadmap,
<http://energy.gov/sites/prod/files/2014/02/f8/fctt_roadmap_june2013.pdf>
[accessed 29.5.16].
[7] Department of Energy. Fuel pathway integration technical team roadmap,
<https://www1.eere.energy.gov/vehiclesandfuels/pdfs/program/fpitt_
roadmap_june2013.pdf> [accessed 29.5.16].
[8] Zegers P. Fuel cell commercialization: the key to a hydrogen economy. J
Power Sources 2006;154(2):497–502.
[9] Dubau L et al. A review of PEM fuel cell durability: materials degradation,
local heterogeneities of aging and possible mitigation strategies. Energy
Environ 2014;3(6):540–60.
[10] http://www.nrel.gov/continuum/sustainable_transportation/fuel_cell_evs.
html [accessed 29.05.16].
[11] Mench MM. Mechanical engineers’ handbook: energy and power. In: Kutz
Myer, editor. Chapter 6 fuel cells. John Wiley & Sons Inc.; 2006.
[12] Zhang X, Gao Y. Impact of liquid water on oxygen reaction in cathode catalyst
layer of proton exchange membrane fuel cell: a simple and physically sound
model. J Power Sources 2016;318:251–63.
[13] Zamel N. The catalyst layer and its dimensionality – a look into its ingredients
and how to characterize their effects. J Power Sources 2016(309):141–59.
[14] Branco CM et al. New approaches towards novel composite and multilayer
membranes for intermediate temperature-polymer electrolyte fuel cells and
direct methanol fuel cells. J Power Sources 2016;316:139–59.
[15] Ossiander T et al. Influence of membrane type and molecular weight
distribution on the degradation of PBI-based HTPEM fuel cells. J Membr Sci
2016;509:27–35.
[16] Borup R et al. PEM fuel cell electro-catalyst durability measurements. J Power
Sources 2006;163:76–81.
[17] Mu S et al. Accelerated durability tests of catalyst layers with various pore
volume for catalyst coated membranes applied in PEM fuel cells. Int J
Hydrogen Energy 2010;35:2872–6.
[18] Pestrak M et al. The effect of mechanical fatigue on the lifetimes of membrane
electrode assemblies. J Fuel Cell Sci Technol 2010;7:041009.
[19] Sompalli B et al. Membrane degradation at catalyst layer edges in PEMFC
MEAs. J Electrochem Soc 2007;154:B1349–57.
[20] Kundu S et al. Morphological features (defecrs) in fuel cell membrane
electrode assemblies. J Power Sources 2006;157:650–6.
[21] El-kharouf A et al. Proton exchange membrane fuel cell degradation and
testing: review. J Energy Inst 2012;85(4):188–200.
[22] Andisheh-Tadbir M, Orfino FP, Kjeang E. Three-dimensional phase
segregation of micro-porous layers for fuel cells by nano-scale X-ray
computed tomography. J Power Sources 2016;310:61–9.
[23] Dicks AL. The role of carbon in fuel cells. J Power Sources 2006;156:128–41.
[24] Yin K-M, Cheng B-S, Chiang K-W. Non-uniform agglomerate cathode catalyst
layer model on the performance of PEMFC with consideration of water effect.
Renewable Energy 2016;95:191–201.
[25] Wang J, Wang H. Discrete approach for flow-field designs of parallel channel
configurations in fuel cells. Int J Hydrogen Energy 2012;37(14):10881–97.
[26] Wang J, Wang H. Flow field designs of bipolar plates in PEM fuel cells: theory
and applications. Fuel Cells 2012;12(6):989–1003.
[27] Iranzo A et al. PEM fuel cell operation under air and O2 feed: analysis of cell
performance and liquid water distributions. Fuel Cells 2016;16(4):463–8.
[28] Iranzo A et al. Liquid water preferential accumulation in channels of PEM fuel
cells with multiple serpentine flow fields. Int J Hydrogen Energy
2014;39:15687–95.
[29] Li EH et al. A review of water flooding issues in the proton exchange
membrane fuel cell. J Power Sources 2008;178:103–17.
[30] Jahnke T et al. Performance and degradation of proton exchange membrane
fuel cells: state of the art in modeling from atomistic to system scale. J Power
Sources 2016;304:207–33.
[31] Morimoto Y, Yamakawa S. Computational modeling aspects of PEFC
durability. In: Mench MM, Kumbur EC, Veziroğlu TN, editors. Polymer
electrolyte fuel cell degradation. Elsevier Inc.; 2012. p. 423–41.
[32] Ozden E, Tari I. Proton exchange membrane fuel cell degradation: a
parametric analysis using computational fluid dynamics. J Power Sources
2016;304:64–73.
[33] Pukrushpan JT, Stefanopoulou AG, Peng H. Control of fuel cell power systems:
principles, modeling, analysis, and feedback design. Springer; 2004.
[34] Chen CH, Jung SP, Yen SC. Flow distribution in the manifold of PEM fuel cell
stack. J Power Sources 2007;173(1):249–63.
[35] Borup R et al. Scientific aspects of polymer electrolyte fuel cell durability and
degradation. Chem Rev 2007;107(10):3904–51.
[36] Behling N. Solving the fuel cell dilemma. Fuel Cells Bull 2012;11:12–4.
[37] Rajashekara K. Present status and future trends in electric vehicle propulsion
technologies. IEEE J Emerg Selected Top Power Electron 2013;1(1):3–10.
[38] Department of Energy, Durability test Protocol for PEM Fuel Cells. <http://
www1.eere.energy.gov/hydrogenandfuelcells/fuelcells/
pdfs/component_durability_profile.pdf> [accessed 29.05.16].
[39] Mankins JC. Technology readiness and risk assessments: a new approach.
Acta Astronaut 2009;65(9–10):1208–15.
[40] U.S. Government Accountability Office. Technology readiness assessment
guide. GAO-16-4 10G; August 2016.
[41] De Bruijn FA, Dam VAT, Janssen GJM. Review: Durability and degradation
issues of PEM fuel cell components. Fuel Cells 2008;8:3–22.
[42] Yu Y et al. A review on performance degradation of proton exchange
membrane fuel cells during startup and shutdown processes: causes,
consequences, and mitigation strategies. J Power Sources 2012;205:10–23.
[43] Wu J et al. A review of PEM fuel cell durability: degradation mechanisms and
mitigation strategies. J Power Sources 2008;184:104–19.
[44] Zhang L et al. Recent advances in proton exchange membranes for fuel cell
applications. Chem Eng J 2012;204(206):87–97.
[45] Weber AZ et al. A critical review of modeling transport phenomena in
polymer-electrolyte fuel cells. J Electrochem Soc 2014;161(12):F1254–99.
[46] Wang CY. Fundamental models for fuel cell engineering. Chem Rev
2004;104:4727–66.
[47] Fuel Cell Technologies Office. ‘‘Manufacturing R&D. Chapter 3.5. Multi-Year
Program Plan, draft update; 2015 <http://energy.gov/sites/prod/files/2015/
06/f22/fcto_myrdd_manufacturing.pdf> [accessed 29.5.16].
[48] Kang J, Kim J. Membrane electrode assembly degradation by dry/wet gas on a
PEM fuel cell. Int J Hydrogen Energy 2010;35(23):13125–30.
[49] Quadrennial Technology Review; 2015, Chapter 6: innovating clean energy
technologies in advanced manufacturing. <http://energy.gov/sites/prod/files/
2015/09/f26/QTR2015-06-Manufacturing.pdf> [accessed 29.5.16].
[50] Khorasany RMH et al. Mechanical degradation of fuel cell membranes under
fatigue fracture tests. J Power Sources 2015;274:1208–16.
[51] Khorasany RMH et al. Fatigue properties of catalyst coated membranes for
fuel cells: ex-situ measurements supported by numerical simulations. Int J
Hydrogen Energy 2016;41(21):8992–9003.
[52] Serincan MF, Pasaogullari U. Mechanical behavior of the membrane during
the polymer electrolyte fuel cell operation. J Power Sources 2011;196:
1303–13.
[53] Luo Z et al. Degradation behavior of membrane–electrode-assembly
materials in10-cell PEMFC stack. Int J Hydrogen Energy 2006;31:1831–7.
[54] Tang Y et al. Stresses in proton exchange membranes due to hygro-thermal
loading. J Fuel Cell Sci Technol 2006;3(2):119–24.
[55] Tang Y et al. An experimental investigation of humidity and temperature
effects on the mechanical properties of perfluorosulfonic acid membrane.
Mater Sci Eng A 2006;425(1):297–304.
[56] Kusoglu A et al. Stress-strain behavior of perfluorosulfonic acid membranes at
various temperatures and humidities: experiments and phenomenological
modelling. J Fuel Cell Sci Technol 2009;6:011012.
[57] Yu PT, et al. Method of mitigating fuel cell degradation due to startup and
shutdown via hydrogen/nitrogen storage. U.S. Patent US8492046 B2; 2013.
[58] Endoh E et al. Degradation study of MEA for PEMFCs under low humidity
conditions. Electrochem Solid-State Lett 2004;7:A209–11.
[59] Yoshioka S et al. Development of a PEM fuel cell under low humidified
conditions. Fuel Cells Bull 2005;3:11–5.
[60] Jao T-C et al. Degradation mechanism study of PTFE/Nafion membrane in
MEA utilizing an accelerated degradation technique. Int J Hydrogen Energy
2012;37(18):13623–30.
[61] Diloyan G et al. Effect of mechanical vibration on platinum particle
agglomeration and growth in polymer electrolyte membrane fuel cell
catalyst layers. J Power Sources 2012;214:59–67.
[62] Poornesh KK et al. Gradation of mechanical properties in gas diffusion
electrode. Part 1: influence of nano-scale heterogeneity in catalyst layer on
interfacial strength between catalyst layer and membrane. J Power Sources
2010;195:2709–17.
[63] Mayrhofer K et al. Fuel cell catalyst degradation on the nanoscale.
Electrochem Commun 2008;10(8):1144–7.
[64] Zhang S et al. A review of platinum-based catalyst layer degradation in proton
exchange membrane fuel cells. J Power Sources 2009;194(2):588–600.
[65] Mehta V, Cooper JS. Review and analysis of PEM fuel cell design and
manufacturing. J Power Sources 2003;114(1):32–53.
 J. Wang / Applied Energy 189 (2017) 460–479
[66] Zakaria Z et al. Membranes for direct ethanol fuel cells: an overview. Appl
Energy 2016;163:334–42.
[67] Bouchet R, Siebert E. Proton conduction in acid doped polybenzimidazole.
Solid State Ionics 1999;118(3/4):287–99.
[68] Schmittinger W, Vahidi A. A review of the main parameters influencing long-
term performance and durability of PEM fuel cells. J Power Sources
2008;180:1–14.
[69] Khattra NS et al. Time-dependent mechanical response of a composite PFSA
membrane. J Power Sources 2013;228:256–69.
[70] Kusoglu A et al. Mechanical behavior of fuel cell membranes under humidity
cycles and effect of swelling anisotropy on the fatigue stresses. J Power
Sources 2007;170:345–58.
[71] Kusoglu A et al. Numerical investigation of mechanical durability in polymer
electrolyte membrane fuel cells. J Electrochem Soc 2010;157(5):B705–13.
[72] Tang HL, Pan M, Wang F. A mechanical durability comparison of various
perfluocarbon proton exchange membranes. J Appl Polym Sci 2008;109
(4):2671–8.
[73] Dillet J et al. Impact of flow rates and electrode specifications on degradations
during repeated startups and shutdowns in polymer-electrolyte membrane
fuel cells. J Power Sources 2014;250:68–79.
[74] Li Q et al. Approaches and recent development of polymer electrolyte
membranes for fuel cells operating above 100 C. Chem Mater
2003;15:4896–915.
[75] Sone Y, Ekdunge P, Simonsson D. Proton conductivity of Nafion 117 as
measured by a four-electrode AC impedance method. J Electrochem Soc
1996;143(4):1254–9.
[76] Cheddie D, Munroe N. Parametric model of an intermediate temperature
PEMFC. J Power Sources 2006;156:414–23.
[77] Kim S-K et al. Highly durable polymer electrolyte membranes at elevated
temperature: cross-linked copolymer structure consisting of poly
(benzoxazine) and poly (benzimidazole). J Power Sources 2013;226:346–53.
[78] Zhang J et al. Polybenzimidazole membrane-based PEM fuel cell in the
temperature range of 120–200 0C. J Power Sources 2007;172:163–71.
[79] Zhang J et al. Comprehensive strategy to design highly ordered mesoporous
Nafion membranes for fuel cells under low humidity conditions. J Mater
Chem A 2014;2:20578–87.
[80] Li Q et al. The CO poisoning effect in PEMFCs operational at temperatures up
to 200 °C. J Electrochem Soc 2003;150:A1599–605.
[81] Bauer F, Denneler S, Willert-Porada M. Influence of temperature and
humidity on the mechanical properties of NafionÒ 117 polymer electrolyte
membrane. J Polym Sci B: Polym Phys 2005;43(7):786–95.
[82] Young SK, Mauritz KA. Dynamic mechanical analyses of NafionÒ/organically
modified silicate nanocomposites. J Polym Sci B: Polym Phys
2001;39:1282–95.
[83] Lee JH, Lalk TR. Modeling fuel cell stack systems. J Power Sources
1998;73:229–41.
[84] Yuan X-Z et al. Degradation of a PEM fuel cell stack with NafionÒ membranes
of different thicknesses. Part I: in situ diagnosis. J Power Sources
2010;195:7594–9.
[85] Yuan X-Z et al. Degradation of a PEM fuel cell stack with NafionÒ membranes
of different thicknesses. Part II: Ex situ diagnosis. J Power Sources
2012;205:324–34.
[86] Lim C et al. Membrane degradation during combined chemical and
mechanical accelerated stress testing of polymer electrolyte fuel cells. J
Power Sources 2014;257:102–10.
[87] Banan R, Zu J, Bazylak A. Humidity and temperature cycling effects on cracks
and delaminations in PEMFCs. Fuel Cells 2015;15(2):327–36.
[88] Liu W, Ruth K, Rusch G. Membrane durability in PEM fuel cells. J New Mater
Electrochem Syst 2001;4:227–31.
[89] Trogadas P, Parrondo J, Ramani V. Degradation mitigation in polymer
electrolyte membranes using cerium oxide as a regenerative free-radical
scavenger. Electrochem Solid-State Lett 2008;11:B113–6.
[90] Trogadas P, Parrondo J, Ramani V. CeO2 surface oxygen vacancy concentration
governs in situ free radical scavenging efficacy in polymer electrolytes. ACS
Appl Mater Interfaces 2012;4(10):5098–102.
[91] Gamburzev S, Appleby AJ. Recent progress in performance improvement of
the proton exchange membrane fuel cell (PEMFC). J Power Sources 2002;107
(1):5–12.
[92] Mathias MF et al., editors. Handbook of fuel cells—fundamentals, technology
and applications. New York: John Wiley & Sons; 2003 [3, Chapter 42].
[93] Arvay A et al. Characterization techniques for gas diffusion layers for proton
exchange membrane fuel cells? A review. J Power Sources 2012;213:317–37.
[94] Mukundan R et al. Performance of PEM fuel cells at sub-freezing
temperatures. ECS Trans 2007;11(1):543–52.
[95] Lim S-J et al. Investigation of freeze/thaw durability in polymer electrolyte
fuel cells. Int J Hydrogen Energy 2010;35(23):13111–7.
[96] Ha T et al. Experimental study of the effect of dissolution on the gas diffusion
layer in polymer electrolyte membrane fuel cells. Int J Hydrogen Energy
2011;36(19):12436–43.
[97] Serincan MF, Pasaogullari U. Effect of gas diffusion layer anisotropy on
mechanical stresses in a polymer electrolyte membrane. J Power Sources
2011;196:1314–20.
[98] Sadeghi E, Djilali N, Bahrami M. Effect of compression on the effective
thermal conductivity and thermal contact resistance in PEM fuel cell gas
diffusion layers. In: Proceedings of the ASME international mechanical
477
engineering congress & exposition IMECE2010 Vancouver, Canada.
IMECE2010-39284, November 12–18, 2010.
[99] Faydi Y, Lachat R, Meyer Y. Thermomechanical characterisation of
commercial gas diffusion layers of a proton exchange membrane fuel cell
for high compressive pre-loads under dynamic excitation. Fuel
2016;182:124–30.
[100] Mason TJ et al. Effect of clamping pressure on ohmic resistance and
compression of gas diffusion layers for polymer electrolyte fuel cells. J
Power Sources 2012;219:52–9.
[101] Mason TJ et al. A study of the effect of compression on the performance of
polymer electrolyte fuel cells using electrochemical impedance spectroscopy
and dimensional change analysis. Int J Hydrogen Energy 2013;38
(18):7414–22.
[102] Montanini R, Squadrito G, Giacoppo G. Measurement of the clamping
pressure distribution in polymer electrolyte fuel cells using piezoresistive
sensor arrays and digital image correlation techniques. J Power Sources
2011;196(20):8484–93.
[103] Pei P, Chen H. Main factors affecting the lifetime of Proton Exchange
Membrane fuel cells in vehicle applications: a review. Appl Energy
2014;125:60–75.
[104] Freire TJP, Gonzalez ER. Effect of membrane characteristics and
humidification conditions on the impedance response of polymer
electrolyte fuel cells. J Electroanal Chem 2001;503:57–68.
[105] Peck DH et al. Preparation and performance evaluation of membrane
electrode assemblies for polymer electrolyte fuel cell. J New Mater
Electrochem Syst 1999;2(2):121–4.
[106] Renedo OD, Alonso-Lomillo MA, Martinez MJA. Recent developments in the
field of screen-printed electrodes and their related applications. Talanta
2007;73(2):202–19.
[107] Bang JH et al. Porous carbon supports prepared by ultrasonic spray pyrolysis
for direct methanol fuel cell electrodes. J Phys Chem C 2007;111:10959–64.
[108] Rice CA et al. Platinum dissolution in fuel cell electrodes: enhanced
degradation from surface area assessment in automotive accelerated stress
tests. J Electrochem Soc 2015;162(10):F1175–80.
[109] Kinoshita K, Lundquist JT, Stonehart P. Potential cycling effects on platinum
electrocatalyst surfaces. J Electroanal Chem 1973;48(2):157–66.
[110] Tintula KK et al. Durability of Pt/C and Pt/MC-PEDOT catalysts under
simulated start-stop cycles. Fuel Cells 2013;13(2):158–66.
[111] Schulenburg H et al. 3D imaging of catalyst support corrosion in polymer
electrolyte fuel cells. J Phys Chem C 2011;115(29):14236–43.
[112] Bartholomew CH. Mechanisms of catalyst deactivation. Appl Catal A
2001;212:17–60.
[113] Sethuraman VA, Weidner JW. Analysis of sulfur poisoning on a PEM fuel cell
electrode. J Power Sources 2010;55(20):5683–94.
[114] Shah AA et al. A transient PEMFC model with CO poisoning and mitigation by
O2 bleeding and Ru-containing catalyst. J Power Sources 2007;166(1):1–21.
[115] Kuhn M, Rodriguez JA. Interaction of sulfur with bimetallic surfaces:
coadsorption of sulfur and noble metals on Ru (001). J Phys Chem 1994;98
(46):12059–66.
[116] Pereira LGS et al. Investigation of the CO tolerance mechanism at several Pt-
based bimetallic anode electrocatalysts in a PEM fuel cell. Electrochim Acta
2009;54(7):1992–8.
[117] Garcia AC, Paganin VA, Ticianelli EA. CO tolerance of PdPt/C and PdPtRu/C
anodes for PEMFC. Electrochim Acta 2008;53(12):4309–15.
[118] Baschuk J, Li X. Carbon monoxide poisoning of proton exchange membrane
fuel cells. J Int J Energy Res 2001;25(8):695–713.
[119] Jung A et al. An experimental study on the hydrogen crossover in polymer
electrolyte membrane fuel cells for various current densities. Appl Energy
2016;175:212–7.
[120] Zhang Y-C et al. Simulation of creep and damage in the bonded compliant
seal of planar solid oxide fuel cell. Int J Hydrogen Energy 2014;39:17941–51.
[121] Zhang Y-C et al. Using short-time creep relaxation effect to decrease the
residual stress in the bonded compliant seal of planar solid oxide fuel cell: a
finite element simulation. J Power Sources 2014;255:108–15.
[122] Zhang G, Kandlikar SG. A critical review of cooling techniques in proton
exchange membrane fuel cell stacks. Int J Hydrogen Energy
2012;37:2412–29.
[123] Özden E, Tolj I, Barbir F. Designing heat exchanger with variable surface area
for passive cooling of PEM fuel cell. Appl Therm Eng 2013;51:1339–44.
[124] Wen CY, Huang GW. Application of a thermally conductive pyrolytic graphite
sheet to thermal management of a PEM fuel cell. J Power Sources 2008;178
(1):132–40.
[125] Matian M, Marquis A, Brandon N. Model based design and test of cooling
plates for an air-cooled polymer electrolyte fuel cell stack. Int J Hydrogen
Energy 2011;36(10):6051–166.
[126] Yu SH et al. Numerical study to examine the performance of multi-pass
serpentine flow fields for cooling plates in polymer electrolyte membrane
fuel cells. J Power Sources 2009;194:697–703.
[127] Chen FC et al. Analysis of optimal heat transfer in a PEM fuel cell cooling
plate. Fuel Cells 2003;3(4):181–8.
[128] Baek SM et al. A numerical study on uniform cooling of large-scale PEMFCs
with different coolant flow field designs. Appl Therm Eng 2011;31(8–
9):1427–34.
[129] Alizadeh E et al. A novel cooling flow field design for polymer electrolyte
membrane fuel cell stack. Int J Hydrogen Energy 2016;41(20):8525–32.
478 J. Wang / Applied Energy 189 (2017) 460–479
[130] Stucki et al. PEM water electrolyzers: evidence for membrane failure in 100
kW demonstration plants. J Appl Electrochem 1998;28:1041–9.
[131] Mathias MF et al. Two fuel cell cars in every garage? Electrochem Soc Interf
2005;14:24–35.
[132] Benziger J, Nehlsen J, Blackwell D, Brennan T, Itescu J. Water flow in the gas
diffusion layer of PEM fuel cells. J Membr Sci 2005;261:98–106.
[133] http://fuelcellsetc.com/helpful-tools/gas-diffusion-layer-gdl-comparison-
chart/ [accessed 28.10.16].
[134] Wu HW. A review of recent development: transport and performance
modeling of PEM fuel cells. Appl Energy 2016;165:81–106.
[135] Wu J, Liu Q, Fang H. Toward the optimization of operating conditions for
hydrogen polymer electrolyte fuel cells. J Power Sources 2006;156:388–99.
[136] Ishigami Y et al. Corrosion of carbon supports at cathode during hydrogen/air
replacement at anode studied by visualization of oxygen partial pressures in
a PEFC – start-up/shut-down simulation. J Power Sources 2011;196:3003–8.
[137] Bazylak A. Liquid water visualization in PEM fuel cells: a review. Int J
Hydrogen Energy 2009;34:3845–57.
[138] Wang L, Liu H. Separate measurement of current density under the channel
and the shoulder in PEM fuel cells. J Power Sources 2008;180:365–72.
[139] Reum M, Wokaun A, Büchi FN. Measuring the current distribution with
submillimeter resolution in PEFCs: III. Influence of the flow field geometry. J
Electrochem Soc 2009;156:B1225–31.
[140] Weng F, Hsu C, Li C. Experimental investigation of PEM fuel cell aging under
current cycling using segmented fuel cell. Int J Hydrogen Energy 2010;35
(8):3664–75.
[141] Lai Y-H, Fly GW. In-situ diagnostics and degradation mapping of a mixed-
mode accelerated stress test for proton exchange membranes. J Power
Sources 2015;274:1162–72.
[142] Perez LC et al. Segmented polymer electrolyte membrane fuel cell – a review.
Renew Sustain Energy Rev 2011;15(1):169–85.
[143] Wu TC, Djilali N. Experimental investigation of water droplet emergence in a
model polymer electrolyte membrane fuel cell microchannel. J Power
Sources 2012;208:248–56.
[144] Soler J, Hontanon E, Daza L. Electrode permeability and flow-field
configuration: influence on the performance of a PEMFC. J Power Sources
2003;118:172–8.
[145] Lim BH et al. Numerical analysis of modified parallel flow field designs for
fuel cells. Int J Hydrogen Energy 2016. doi: http://dx.doi.org/10.1016/j.
ijhydene.2016.03.189.
[146] Lim BH et al. Effects of flow field design on water management and reactant
distribution in PEMFC: a review. Ionics 2016;22(3):301–16.
[147] Kumbur EC, Sharp KV, Mench MM. Liquid droplet behaviour and instability in
a polymer electrolyte fuel cell flow channel. J Power Sources
2006;161:333–45.
[148] Maharudrayya S, Jayanti S, Deshpande AP. Flow distribution and pressure
drop in parallel-channel configurations of planar fuel cells. J Power Sources
2005;144:94–106.
[149] Wang X, Zhou B. Liquid water flooding process in proton exchange
membrane fuel cell cathode with straight parallel channels and porous
layer. J Power Sources 2011;196:1776–94.
[150] Wang J, Priestman GH, Wu D. A theoretical analysis of uniform flow
distribution for the admission of high-energy fluids to steam surface
condenser. J. Eng Gas Turbine Power, Trans ASME 2001;123(2):472–5.
[151] Wang J et al. Analytical solution of flow coefficients for a uniformly
distributed porous channel. Chem Eng J 2001;84(1):1–6.
[152] Wang J. Pressure drop and flow distribution in parallel-channel fuel cell
stacks: U-type arrangement. Int J Hydrogen Energy 2008;33(21):6339–50.
[153] Wang J. Comments on Flow distribution in U-type layers or stacks of planar
fuel cells, by W.H. Huang and Q.S. Zhu [J. of Power Sources, 2008; 178: 353–
362]. J Power Sources 2009;190(2):511–2.
[154] Wang J. Pressure drop and flow distribution in parallel-channel
configurations of fuel cells: Z-type arrangement. Int J Hydrogen Energy
2010;35(11):5498–509.
[155] Wang J, Wang H. Discrete method for design of flow distribution in
manifolds. Appl Therm Eng 2015;89:927–45.
[156] Bachman J et al. Experimental investigation of the effect of channel length on
performance and water accumulation in a PEMFC parallel flow field. Int J
Hydrogen Energy 2012;37(22):17172–9.
[157] Heidary H, Kermani MJ, Advani SG, Prasad AK. Experimental investigation of
in-line and staggered blockages in parallel flowfield channels of PEM fuel
cells. Int J Hydrogen Energy 2016;41(16):6885–93.
[158] Hermann A, Chaudhuri T, Spagnol P. Bipolar plates for PEM fuel cells: a
review. Int J Hydrogen Energy 2005;30(12):1297–302.
[159] Li XG, Sabir I. Review of bipolar plates in PEM fuel cells: flow-field designs. Int
J Hydrogen Energy 2005;30:359–71.
[160] Joseph S et al. Conducting polymer-coated stainless steel bipolar plates for
proton exchange membrane fuel cells (PEMFC). Int J Hydrogen Energy
2005;30(2):1339–44.
[161] Tawfik H, Hung Y, Mahajan D. Metal bipolar plates for PEM fuel cell? A
review. J Power Sources 2007;163(2):755–67.
[162] Bai C-Y, Ger M-D, Wu M-S. Corrosion behaviors and contact resistances of the
low-carbon steel bipolar plate with a chromized coating containing carbides
and nitrides. Int J Hydrogen Energy 2009;34(16):6778–89.
[163] Kraytsberg A, Auinat M, Ein-Eli Y. Reduced contact resistance of PEM fuel
cell’s bipolar plates via surface texturing. J Power Sources 2007;164
(2):697–703.
[164] Zhou P, Wu CW, Ma GJ. Contact resistance prediction and structure
optimization of bipolar plates. J Power Sources 2006;159(2):1115–22.
[165] JRC (Joint Research Center). FCTESQA testing procedures, PEFC power stack
performance testing procedure: measuring voltage, power and efficiency as
function of pressure for a water-cooled PEFC stack <http://iet.jrc.ec.europa.
eu/fuel-cells/downloads-0> [accessed 29.10.16].
[166] Yuan X-Z et al. A review of polymer electrolyte membrane fuel cell durability
test protocols. J Power Sources 2011;196:9107–16.
[167] Yuan X-Z et al. A review of accelerated conditioning for a polymer electrolyte
membrane fuel cell. J Power Sources 2011;196:9097–106.
[168] Garland NL, Benjamin TG, Kopasz JP. DOE fuel cell program: durability
technical targets and testing protocols. ECS Trans 2007;11:923–31.
[169] Miller M, Bazylak A. A review of polymer electrolyte membrane fuel cell stack
testing. J Power Sources 2011;196(2):601–13.
[170] Eckl R et al. Experimental analysis of water management in a self-
humidifying polymer electrolyte fuel cell stack. J Power Sources 2004;138
(1/2):137–44.
[171] Mocotéguy P et al. Long-term testing in dynamic mode of HT-PEMFC H3PO4/
PBI Celtec-P MEAs for l-CHP applications. Fuel Cells 2009;9:325–48.
[172] Mocotéguy P et al. Long-term testing in dynamic mode of HT-PEMFC H3PO 4/
PBI celtec-P based membrane electrode assemblies for micro-CHP
applications. Fuel Cells 2010;10(2):299–311.
[173] Chen D et al. Geometric optimization of a 10-cell modular plan, first ed.
Washington (DC); 2006. p. 205–27.
[174] Yeh P et al. Durability and efficiency tests for direct methanol fuel cell’s long-
term performance assessment. Energy 2016;107:716–24.
[175] Pei P et al. Analysis on the PEM fuel cells after accelerated life experiment. Int
J Hydrogen Energy 2010;35:3147–51.
[176] Giddey S, Ciacchi FT, Badwal SPS. Design, assembly and operation of polymer
electrolyte membrane fuel cell stacks to 1 kWe capacity. J Power Sources
2004;125:155–65.
[177] Bonnet C et al. Design of an 80 kWe PEM fuel cell system: scale up effect
investigation. J Power Sources 2008;182:441–8.
[178] Chu D, Jiang R. Comparative studies of polymer electrolyte membrane fuel
cell stack and single cell. J Power Sources 1999;80(1/2):226–34.
[179] Rodatz P et al. Operational aspects of a large PEFC stack under practical
conditions. J Power Sources 2004;128:208–17.
[180] Cho E et al. Characteristics of the PEMFC repetitively brought to temperatures
below 0 °C. J Electrochem Soc 2003;150(12):A1667–70.
[181] Cho EA et al. Effects of water removal on the performance degradation of
PEMFCs repetitively brought to 0 °C. J Electrochem Soc 2004;151(5):A661–5.
[182] Pesaran A, Kim G-H, Gonder J. Proceedings of the fuel cells durability, first ed.
Washington (DC); 2006. p. 205–27.
[183] Knights SD et al. Aging mechanisms and lifetime of PEFC and DMFC. J Power
Sources 2004;127(1/2):127–34.
[184] Verhage AJ et al. 30,000 h operation of a 70 kW stationary PEM fuel cell
system using hydrogen from a chlorine factory. Int J Hydrogen Energy
2013;38(11):4714–24.
[185] Smit M. Towards 40 000 hours of operation for Nedstack’s FCS XXL PEM fuel
cell stacks. Fuel Cells Bull 2014;2014(8):12–5.
[186] Gao F et al. PEM fuel cell stack modeling for real-time emulation in
hardware-in-the-loop applications. IEEE Trans Energy Convers 2011;26
(1):184–94.
[187] Meidanshahi V, Karimi G. Dynamic modeling, optimization and control of
power density in a PEM fuel cell. Appl Energy 2012;93:98–105.
[188] Tang Y et al. Experimental investigation of dynamic performance and
transient responses of kW-class PEM fuel cell stack under various loads
changes. Appl Energy 2012;87:68–76.
[189] Tirnovan R, Giurgea S, Miraoui A. Strategies for optimizing the opening of the
outlet air circuits nozzle to improve the efficiency of the PEMFC generator.
Appl Energy 2011;88:1197–204.
[190] Vahidi A, Stefanopoulou A, Peng H. Current management in a hybrid fuel cell
power system: a model-predictive control approach. IEEE Trans Control Syst
Technol 2006;14:1047–57.
[191] Sasmito AP et al. Optimization of an open-cathode polymer electrolyte fuel
cells stack utilizing Taguchi method. Appl Energy 2017;185:1225–32.
[192] Reiser CA et al. A reverse current decay mechanism for fuel cells. Electrochem
Solid-State Lett 2005;8:A273–6.
[193] Pei H et al. Operation characteristics of air-cooled proton exchange
membrane fuel cell stacks under ambient pressure. Appl Therm Eng
2014;63:227–33.
[194] Park J et al. A review of the gas diffusion layer in proton exchange membrane
fuel cells: durability and degradation. Appl Energy 2015;155:866–80.
[195] Wan Z et al. A review on cold start of proton exchange membrane fuel cells.
Energies 2014;7:3179–203.
[196] Kunusch C et al. Sliding-mode control of PEM fuel cells, advances in industrial
control, Chapter 2 PEM fuel cell systems. Springer-Verlag London Limited;
2012.
[197] Tabe Y et al. Cold start characteristics and freezing mechanism dependence
on start-up temperature in a polymer electrolyte membrane fuel cell. J Power
Sources 2012;208:366–73.
[198] Arthur DA et al. Fuel cell operational methods for hydrogen addition after
shutdown. United States Patent, US 8,232,014 B2; 2012.
[199] Reiser CA, Yang D, Sawyer RD. Procedure for starting up a fuel cell system
using a fuel purge. U.S. Patent 6,887,599 B2; 2005.
 J. Wang / Applied Energy 189 (2017) 460–479
[200] Ang SMC et al. Fuel cell systems optimization: methods and strategies. Int J
Hydrogen Energy 2011;36(22):14678–703.
[201] Oszcipok M et al. Portable proton exchange membrane fuel-cell systems for
outdoor applications. J Power Sources 2006;157:666–73.
[202] Wang J. Theory of flow distributions in manifolds. Chem Eng J
2011;168:1331–45.
[203] Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC. A review of polymer
electrolyte membrane fuel cells: technology, applications, and needs on
fundamental research. Appl Energy 2011;88:981–1007.
[204] Zhang X et al. Modelling water intrusion and oxygen diffusion in a
reconstructed microporous layer of PEM fuel cells. Int J Hydrogen Energy
2014;39:17222–30.
[205] Wang JY et al. Domain-decomposition method for parallel lattice Boltzmann
simulation of incompressible flow in porous media. Phys Rev E 2005;72(7):
art no 016706.
479
[206] Ashrafi M, Mehrzad S. Effects of heterogeneous surface of gas diffusion layers
on droplet transport in microchannels of PEM fuel cells. Int J Hydrogen
Energy 2016;41(3):1974–89.
[207] Quesnel C et al. Dynamic Percolation and Droplet Growth Behavior in Porous
Electrodes of Polymer Electrolyte Fuel Cells. J Phys Chem C 2015;119
(40):22934–44.
[208] Anderson R et al. A critical review of two-phase flow in gas flow channels of
proton exchange membrane fuel cells. J Power Sources 2010;195
(15):4531–53.
[209] Zhu X et al. Three-dimensional numerical simulations of water droplet
dynamics in a PEMFC gas channel. J Power Sources 2008;181(1):101–15.
[210] Shimpalee S, Lilavivat V. Study of water droplet removal on etched-metal
surfaces for proton exchange membrane fuel cell flow channel. J Electrochem
Energy Convers Storage 2016;13:art no 011003.