i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

Study on the diaphragm fracture in a diaphragm

compressor for a hydrogen refueling station

Xiaohan Jia, Jiahao Chen, Han Wu, Xueyuan Peng*

School of Energy and Power Engineering, Xi'an Jiaotong University, No. 28 Xianning West Road, Xi'an 710049, PR
China

article info abstract

Article history: When a diaphragm compressor applied in the hydrogen refueling stations, the diaphragm
Received 2 November 2015 fracture occurred frequently in the edge and middle region of the outer circular groove of
Received in revised form the perforated plate. The unstable operation of the plunger pump caused the oil shortage in
26 February 2016 the oil cylinder, so the diaphragm would cling to the perforated plate while the piston
Accepted 26 February 2016 reached the bottom dead center. Thus, the diaphragm in the groove region would endure
Available online 25 March 2016 the radial stress caused by the large deflection deformation and the additional stress
caused by the perforated plate's groove. In this paper, a theoretical calculation method
Keywords: combined with the thin-plate large deflection theory and the thin-plate small deflection
Hydrogen refueling station theory was proposed to analyze the two kinds of the radial stresses. The experimental and
Diaphragm compressor numerical methods also were conducted for mutual verification. Furthermore, the influ-
Diaphragm fracture encing factors of the diaphragm radial stress were investigated, including the radius of the
Thin-plate theory groove fillet, the groove width and the diaphragm thickness. The results indicated that the
radial stress caused by the large deflection deformation is similar to that caused by the
small deflection deformation. The maximum radial stress of the diaphragm increased
slowly with the increase in the radius of the groove fillet. But with increasing the dia-
phragm thickness or decreasing the groove width, both the deflection and the radial stress
of the diaphragm had an obviously decrease.
Copyright 2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

advantage of the electrochemical hydrogen compressor is that

Introduction
With the rapid progress of fuel cell vehicle in the world, a
number of hydrogen fueling infrastructure stations have been
built to provide hydrogen fuel for vehicles [1e3]. In a hydrogen
fueling station, hydrogen gas is compressed to achieve a high-
pressure of 5000 psi to 10,000 psi and is stored in high-pressure
vessels [4e10]. There are mainly two methods for the
hydrogen compression, the mechanical hydrogen compressor
and the electrochemical hydrogen compressor. The
hydrogen separation and compression are achieved in a single
one step process, so the electrochemical hydrogen
compressor in a polymer-electrolyte-membrane (PEM)-cell is
more effective than mechanical hydrogen compressor just for
some specific applications, such as low power applications,
and the mechanical hydrogen compressors are more suitable
and efficient for normal applications [11e13]. However, the
leakage of mechanical compressor and dispenser are
currently the main contributors to the total risk of a hydrogen

* Corresponding author.
E-mail address: xypeng@xjtu.edu.cn (X. Peng).
http://dx.doi.org/10.1016/j.ijhydene.2016.02.106
0360-3199/Copyright 2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1 6413

profile of a diaphragm compressor has a significant influence

Nomenclature on the diaphragm's fatigue life. The traditional generatrix is a
single exponential polynomial. Wu et al. [19] analyzed the
C integral constant
stress distribution in the diaphragm as it came into contact
D bending rigidity, N/m2
with the cavity surface and provided an optimization algo-
E Young's modulus, MPa
rithm for the traditional generatrix equation of a cavity
F average reacting pressure, MPa
profile. Ban [20] analyzed the stress variation in the dia-
H deflection of the diaphragm, mm
phragm during the compression process using an iteration
H0 deflection of the diaphragm in the center or
method. Wang et al. [21] designed a new generatrix that was
deflection of the perforated plate, mm
composed of two different polynomials. The stress distribu-
Mi the radial moment of the diaphragm, Nm
tion of the diaphragm in the cavity with the new generatrix
p suction pressure, MPa
was more uniform than with the traditional one, although
Qi shear stress of the cross section of a circle with
the cavity volume hardly increased and the compressor's
radius ri, MPa
flow rate was still extremely low. Altukhov et al. [22,23]
q support reaction, N
found that the diameter of the discharge holes in the cav-
r radius of the diaphragm, mm
ity was the significant factor that influences diaphragm's
r1 radius (length) of the OA, mm
fatigue life. Li et al. [18,24] indicated that exceeding the radial
r1 inside radius of outer groove (length of the OB),
stress is the root cause of diaphragm failure. A new gener-
mm
atrix of the cavity profile of a diaphragm compressor,
r1 outside radius of outer groove (length of the
together with an optimization algorithm, was developed. The
OC), mm
research results indicated that the diaphragm exhibited low
r1 radius (length) of the OA, mm
radial stress, as it clung to the cavity when the new gener-
R radius of the perforated plate, mm
atrix was employed. However, little research has been con-
t thickness of the diaphragm, mm
ducted on the diaphragm's fracture occurred in the third
w1 deflection of part AB
position, which is caused by the outer groove of the perfo-
w2 deflection of part BC
rated plate.
w3 deflection of part CD
Thin-plate large deflection theory and thin-plate small
Z deflection exponent of the perforated plate's
deflection theory are the main basis for the theoretical anal-
generatrix
ysis of the diaphragm radial stress as it clung to the perforated
r radial strain of the diaphragm
plate [21,25]. Von Kaman's equations describe the thin circular
t circumferential strain of the diaphragm
plate under uniform loads [26]. Many valid methods have been
q deflection angle of the diaphragm or deflection
subsequently developed for thin plate, including the finite
angle of the perforated plate's generatrix,

element method [27], perturbation and variation methods [28],
m Poisson's ratio
the locally transversal linearization method [29], and the
sa allowable radial stress of the diaphragm, MPa
modified iteration method [30].
sGr radial stress of the diaphragm's lower surface,
Previous studies and diaphragm failure cases indicate that
MPa
exceeding the radial stress is the root cause of diaphragm
sFr radial stress of the diaphragm's upper surface,
failure. The purpose of this study was to research the dia-
MPa
phragm fracture caused by the outer groove of the perforated
sr radial stress of the diaphragm, MPa
plate. A theoretical calculation method combined with the
st circumferential stress of the diaphragm, MPa
thin-plate large deflection theory and the thin-plate small
deflection theory was proposed to analyze the diaphragm
fueling station [14,15]. So the diaphragm compressor has radial stress as the diaphragm clung to the perforated plate
become the best choice for compressing hydrogen because it due to the lack of hydraulic oil. Meanwhile, the numerical
ensures gas purity without any leaking [16]. Moreover, the simulation also was applied to research on the diaphragm
mature leak-detecting system of diaphragm compressor also radial stress, and the experiments had been carried out to
can reduce the risk to a minimum. Compared with other types verify the theoretical and numerical methods. Furthermore,
of compressors, the special structure of the diaphragm the influencing factors of the diaphragm radial stress were
compressor provides a lower compression index and higher investigated, including the fillet of groove, the groove width
volumetric efficiency, which are both beneficial for energy and the diaphragm thickness. The results indicated that the
saving. Meanwhile, the periodic maintenance of the seals and maximum radial stress of the diaphragm could be decreased
diaphragms of the compressor is relatively easy [17]. obviously with increasing the diaphragm thickness or
However, the frequent diaphragm failure is a shortcoming decreasing the groove width when the diaphragm clung to the
of a diaphragm compressor. A fracture is the most common perforated plate.
mode for a diaphragm failure. A diaphragmatic fracture
usually occurs in three positions: the edge of the diaphragm,
the center of the diaphragm and the region contacting the Analysis of diaphragm fracture
outer groove of the perforated plate [18]. The first two are
caused by the cavity profile of a diaphragm and researched The structure of the diaphragm compressor is shown in Fig. 1.
in previous studies. The generatrix equation for the cavity When the piston is at the top dead center, the oil pressure
6414 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1
Fig. 1 e Structure of diaphragm compressor.
reaches its maximum, and the diaphragm comes into contact
with the cavity surface. When the piston moves to the bottom
dead center, the diaphragm deforms downward, thus
enlarging the gas space and starting the suction process.
When the piston reaches the bottom dead center, part of the
diaphragm comes into contact with the perforated plate,
which ends the suction process. When the piston moves to the
top dead center, the diaphragm is then pushed upward by the
hydraulic oil, and the compression process begins. The
discharge valve opens until the gas reaches the discharge
pressure. Finally, the diaphragm comes into contact with the
cavity surface at the completion of the operation cycle.
The design profile of the perforated plate should assure
that only part of diaphragm comes into contact with the
perforated plate while the piston reaches the bottom dead
center to prevent the diaphragm's additional deformation. In
every operation cycle, the hydraulic oil in the cylinder would
decreased due to the leakage from the piston rings. Thus, a
plunger pump is applied to replenish the hydraulic oil to the
cylinder. However, the shortage of the hydraulic oil in the
Fig. 2 e Indentations of the diaphragm.
cylinder often appears due to the plunger pump's unstable
operation in practical work process.
Fig. 2 shows indentations of the diaphragm around the
outer groove. The additional stress occurred when the dia-
phragm clung to the perforated plate due to the lack of the
hydraulic oil. Thus, a combination of radial and additional
stresses caused the diaphragmatic fracture in the region of the
outer groove of the perforated plate, as shows in Fig. 3. The
radius of the diaphragm is usually more than hundred milli-
meters, and the thickness of the diaphragm is less than one
millimeter. The stress caused by the large deflection of the
diaphragm when it contacted to the perforated plate, and the
additional stress caused by the small deflection of the dia-
phragm when it clung to the groove of the perforated plate
tightly. Thus, the thin-plate large and small theories are per-
fect for the theoretical analysis of the diaphragm radial
stresses.
Research on the diaphragm radial stress
Theoretical calculation of the diaphragm radial stress
Large deflection
The generatrix function of the perforated plate profile can be
expressed as:

 Z1  r 2 
1 r
Hr H0 2  Z 1 Z  1 ; (1)
Z1 R R
where H is the diaphragm's deflection, H0 is the diaphragm's
deflection in the center, and Z is the deflection exponent. The
deformation of the diaphragm in the compressor, being circular
and symmetric, can be regarded as a combination of stretching
and bending. The tensile stress is uniform in the thickness di-
rection, whereas the bending stress is linearly distributed. The
bending stress is zero in the middle plate of the diaphragm, thus
achieving the maximal value on both side of the diaphragm. The
maximal radial and circumferential stresses exist on both sides
of the diaphragm. When the diaphragm clings to the perforated
plate, the radial stresses of the diaphragm's lower surface
(contact surface) sGr and the diaphragm's upper surface (non-
contact surface) sFr can be calculated as follows:
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1 6415

Fig. 3 e Diaphragm fracture cases.

8
 The deflection of the diaphragm is w, the deflection angle

>
> EH20 t of diaphragm is q. The equation of the diaphragm's bending
>
< sGr br  ar
R2 H0 deformation can be expressed as follow:
> 2
; (2)
>
> EH t

: sFr 20 br ar d 1 d Q
R H0 rq  ; (4)
dr r dr D
where t is the thickness of the diaphragm, E is the Young
modulus, br and ar can be calculated as follows: where Q is the shear stress of the cross section of a circle with

 2

z1 3m 2z 2  m 2z 1  m r2 2rz1 r2z
 br 2 z  1   2
81  m 1  mz 1z 3 4z1  mz 1 8R

z 1z 3Rz1
4zz 1R2z

 ; (3)
z1 z mrz1
sFr  1 m
z  11  m2 Rz1

radius r. D is the bending rigidity, which can be obtained from

where m is the Poisson's ratio. follow:

Small deflection Eh3

D ; (5)
121  m2
Fig. 4 shows the deformation of the diaphragm around the
perforated plate's groove. There is a suction gas force which For the part AB, Q can be obtained from the equilibrium
acts on the diaphragm's upper surface, but no oil force acts on equation,
the diaphragm's lower surface due to the lack of the hydraulic
oil. Therefore, the diaphragm clings to the perforated plate ppr2  Fpr21 pr Fr21
Q  ; (6)
tightly, and the additional stress occurs around the perforated 2pr 2 2r
plate's groove. where F is the average reacting pressure of the part OA. Thus,
In the Fig. 4, O is the center point of the perforated plate, B- the deflection angle of the part AB can be calculated as follow:
C is the perforated plate's groove, where the diaphragm has a
small deflection. Points A and D are the ends of the deflection. Fr21 2 ln r  1r pr3 Cr C1
q1  ; (7)
8D 16D 2 r
The lengths of OA, OB, OC and OD are r1, r2, r3 and r4,
respectively. and the deflection and deflection angle relations is expressed
as follow:

dw
q ; (8)
dr
So, the deflection of the part AB can be calculated as follow:

Fr21 r2 ln r  1 pr4 Cr2

w1  C1 ln r C2 ; (9)
8D 64D 4
where C, C1 and C2 are integral constants. Analogously, for the
part BC:
Fig. 4 e Deformation of the diaphragm around the groove.
6416 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1

2 
Ppr2  Fpr21  2pq1 r2 pr Fr21 2q1 r2 Fr1 2 ln r2 1 3pr22 C C1 m Fr21 2 ln r2  1r2 pr32
Q  ; (10) D   2 
2pr 2 2r 8D 16D 2 r2 r2 8D 16D


Cr2 C1
Fr21 2q1 r2 2 ln r  1r pr3 C3 r C4
2 r2
q2  ; (11)

2
8D 16D 2 r Fr1 2q1 r2 2 ln r2 1 3pr22 C C1
D  

2 8D 16D 2 r22
Fr1 2q1 r2 r2 ln r  1 pr4 C3 r2 

w2  C4 ln r C5 ; (12) m Fr21 2q1 r2 2 ln r2  1r2 pr32 Cr2 C1
8D 64D 4  :
r2 8D 16D 2 r2
where C3, C4 and C5 are integral constants, q1 is the support (23)
reaction. For the part CD:
If r r3, then q2 q3, w2 0, w3 0 and M2 M3, thus:
Ppr2  Fpr21  2pq1 r2  2pq2 r3 pr Fr21 2q1 r2 2q2 r3
Q  ; Fr21 2 ln r1  1r1 pr31 Cr1 C1
2pr 2 2r 
(13) 8D 16D 2 r1

2
Fr1 2q1 r2 2 ln r1  1r1 pr31 C3 r1 C4

 ; (24)
Fr21 2q1 r2 2q2 r3 2 ln r  1r pr3 C6 r C7 8D 16D 2 r1
q3  ; (14)
8D 16D 2 r

Fr21 2q1 r2 r23 ln r3  1 pr43 C3 r23

2  C4 ln r3 C5 0; (25)
Fr1 2q1 r2 2q2 r3 r2 ln r  1 pr4 C6 r2 8D 64D 4
w3  C7 ln r C8 ;
8D 64D 4

(15) Fr21 2q1 r2 2q2 r3 r23 ln r3  1 pr43 C6 r23
 C7 ln r3 C8 0;
8D 64D 4
where C6, C7 and C8 are integral constants, q2 is the support (26)
reaction.
Applying the boundary conditions,


Fr21 2q1 r2 2 ln r2 1 3pr23 C3 C4
if r r1, then q1 0 and w1 0, thus: D  
8D 16D 2 r23


Fr21 2 ln r1  1r1 pr31 Cr1 C1 m Fr21 2q1 r2 2 ln r2  1r2 pr33 C3 r3 C4
 0; (16) 
8D 16D 2 r1 r3 8D 16D 2 r3


2
Fr1 2q1 r2 2q2 r3 2 ln r2 1 3pr23 C6 C7
D  
Fr21 r21 ln r1  1 pr41 Cr21 8D 16D 2 r23
 C1 ln r1 C2 0; (17) 
2 
8D 64D 4 m Fr1 2q1 r2 2q2 r3 2 ln r2  1r2 pr33 C6 r3 C7
 ;
meanwhile, the radial moment of the diaphragm M1 is zero for r3 8D 16D 2 r3
r r1. According to the thin-plate theory, the M1 can be (27)
expressed as follow:
If r r4, then q3 0, w3 0 and M3 0, thus:
 
dq m

M1 D q ; (18) Fr21 2q1 r2 2q2 r3 2 ln r4  1r4 pr34 C6 r4 C7
dr r  0; (28)
8D 16D 2 r4
substituting Eq. (7) into Eq. (18):



2  Fr21 2q1 r2 2q2 r3 r24 ln r4  1 pr44 C6 r24
Fr 2 ln r1 1 3pr21 C C1 m Fr21 2 ln r1  1r1 pr31  C7 ln r4 C8 0;
D 1   2  8D 64D 4
8D 16D 2 r1 r1 8D 16D
 (29)
Cr1 C1
0:
2 r1

2
Fr1 2q1 r2 2q2 r3 2 ln r4 1 3pr24 C C1
(19) D  
8D 16D 2 r24

2 
If r r2, then q1 q2, w1 0, w2 0 and M1 M2, thus: m Fr1 2q1 r2 2q2 r3 2 ln r4  1r4 pr34 Cr4 C1
 0;
r4 8D 16D 2 r4
Fr21 2 ln r1  1r1 pr31 Cr1 C1
 (30)
8D 16D 2 r1

2
Fr1 2q1 r2 2 ln r1  1r1 pr31 C3 r1 C4 there are 14 equations (Eq. (16), Eq. (17) and Eq. (19) -30) with 14
 ; (20) unknowns (F, r1, r4, q1, q2, C, C1, C2, C3, C4, C5, C6, C7, C8). Thus,
8D 16D 2 r1
the Quasi-Newton method and the global optimization algo-
Fr21 r22 ln r2  1 pr42 Cr22 rithm are used to solve these equations. After that, the
 C1 ln r2 C2 0; (21)
8D 64D 4 maximum radial stress along the thickness direction in the
diaphragm can be obtained as follow:

2
Fr1 2q1 r2 r22 ln r2  1 pr42 C3 r22
 C4 ln r2 C5 0; (22) 6Mr
8D 64D 4 sr : (31)
t2
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1 6417

Numerical simulation of diaphragm radial stress

Table 1 e Computational radial stress of three grid levels
(suction pressure 1.7 MPa).
Under the working condition, the diaphragm was under uni-
The length Radial Radial Radial
form load on both sides and the periphery was clamped by the
of grid/mm stress of the stress of the stress of the
cavity and the perforated plates. Thus, it could be considered point B/MPa point C/MPa point G/MPa
as a problem of thin circular plate with edge-clamped.
0.02 266.1 270.2 217.3
Meanwhile, the deformation of the diaphragm was axisym-
0.01 272.4 283.3 220.1
metric, so the modeling could be simplified to a quarter of the 0.005 276.3 281.7 222.4
diaphragm, as shown in Fig. 5. % Difference 3.8 4.8 2.2
Following actual operation conditions, a constant suction
pressure force was applied on the upper surface of the dia-
phragm, a fixed constraint was applied in the periphery of the strain gauges used are temperature-compensated, single-
diaphragm, and the two sides of the quarter of the diaphragm element, foil strain gauges' type BA120-2AA150(11) with a grid
only could move in xy plane. Points B and C were the contact length of 2 mm, 120 0.1 U electrical resistance, a gauge factor
points of diaphragm, point G was the midpoint of the B-C. of 2.08 1%, a maximum measurement range of 2000 m and a
The grid independence study was performed by comparing resolution of 1 m. Meanwhile, one displacement sensor was
the simulation results obtained under different grid numbers, installed near the discharge valve on each cylinder to measure
as shown in Table 1 and Fig. 6. The maximum difference be- the displacement of the diaphragm. The displacement sen-
tween the results of the coarser and finer grids was less than sors used were contact-type, resistance-type and self-
5%, suggesting that grid independency results could be ach- recovery displacement sensors type JSC-25-5 V with a stroke
ieved in the coarser mesh with a grid length of 0.02 mm. length of 25 mm, a linearity of 0.3% F.S and infinite resolu-
However, to eliminate any uncertainty, simulations were tion. The schematic diagram of this test rig and the arrange-
performed using a grid length of 0.005 mm. ment of the strain gauges on the diaphragm are shown in
Fig. 6.

Experiment validation

The relationship between the stress and strain in the dia-

Results and discussions
phragm could be presented as the following equations:
Analysis of the diaphragm radial stress
8
>
> E
< sr
>
1  m2
r mt
The structure parameters of the diaphragm compressor
; (32)
>
> E perforated plate applied in the experiment are shown in Table
>
: st t mr
1  m2 2, and the material properties of the metallic diaphragm are
shown in Table 3.
where, sr and st are the radial and circumferential stress, r
When the diaphragm clung to the perforated plate, the
and t are the radial and circumferential strain. Thus, the
radial stress distribution of the diaphragm with small deflec-
radial stresses on the sides of the diaphragm could be gained
tion deformation could obtained by using the theoretical
by measuring the radial and circumferential strains. To obtain
method mentioned in Section Theoretical calculation of the
the radial stresses of the diaphragms when it clung to the
diaphragm radial stress. As shown in Fig. 7, the peak values
perforated plate due to the lack of hydraulic oil, a test rig was
of radial stress occurred at three points B, C and G, where
built by utilizing a V-type diaphragm compressor, it is a single-
points B and C were the contact points between the dia-
stage compressor with double-cylinder. The nominal suction
phragm and the diaphragm perforated plate around the small
and discharge pressure are 0.1 MPa and 1.7 MPa, respectively.
deformation area, point G was the midpoint of the B-C. The
The rotational speed is 420 rpm, the gas flow-rate of the dia-
peak values of radial stress of points B and C occurred on the
phragm compressor is 20 Nm3/h. And other parameters of the
upper surface of the diaphragm as 140.1 MPa and 138.2 MPa,
diaphragm compressor could be obtained from Table 2. The
respectively. Meanwhile, the peak value of radial stress of
point G occurred on the lower surface of the diaphragm as
107.4 MPa.
The radial stress of the diaphragm with large deflection
deformation also could be calculated by Eq. (3). The radial
stresses of points B, C and G with larger deflection deforma-
tion were 132.5 MPa, 137.5 MPa and 113.0 MPa, respectively.
Therefore, the value of radial stress caused by the large
deflection deformation was similar to that caused by the small
deflection deformation. By adding two kinds of the stresses,
the total radial stresses of points B, C and G were 272.6 MPa,
275.7 MPa and 220.4 MPa, respectively. The diaphragm's
allowable stress was only 170 MPa, so the diaphragmatic
fracture frequently occurred in these three positions, as
Fig. 5 e Modeling of the diaphragm and perforated plate. shown in Fig. 3.
6418 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1

Fig. 6 e A Schematic diagram of the rest rig; B Arrangement of the strain gauges on the diaphragm.

Table 2 e Parameters of the perforated plate.

Parameters Values
Discharge pressure p(A)/MPa 1.7
Inside radius of outer groove r2/mm 366
Outside radius of outer groove r3/mm 378
Z of the perforated plate 10
H0 of the perforated plate 13

Table 3 e Parameters and material properties of the

metallic diaphragm.
Parameters/Material properties Values
Radius of the diaphragm R/mm 435
Thickness of the diaphragm t/mm 0.5
Poisson's ratio m 0.3
Young modulus E/MPa 203,000 Fig. 7 e Radial stress distribution of the diaphragm with
Allowable stress sa/MPa 170
small deflection deformation.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1 6419

Table 4 shows the comparison of the radial stresses of

points B, C and G by using the theoretical, experimental and
numerical methods. The experimental radial stress of point G
could be obtained directly by arranging a strain gauge, and
that of point B and C could be obtained with three strain
gauges by using the extrapolation method [31,32]. The
maximum difference of the radial stress between theoretical
and experimental value was less than 8.5%, and that between
numerical and experimental value was less than 10%. Thus,
the accuracies of the theoretical and the simulation method of
the diaphragm radial stress were verified by the experiments.

Influencing factors of diaphragm radial stress

From the above research, it could be see that the maximum

radial stress of the diaphragm had far exceeded the allowable
stress when the diaphragm clung to the perforated plate due
to the lack of the hydraulic oil. In generally, there are two
types of the factors influencing the diaphragm radial stress
during working processing, the operation conditions and the
structure parameters. The operation conditions include the
suction pressure, discharge pressure and rotational speed.
However, the influence of the suction pressure and discharge
pressure can be regarded as the influence of pressure differ-
ence between the diaphragm upper surface and lower surface.
Obviously, the diaphragm radial stress increases with the in-
crease in the pressure difference between the diaphragm
upper surface and lower surface. Meanwhile, the diaphragm
compressors generally work at a low rotational speed, and the
speed of the diaphragm is low while it clings to the perforated
plate, so the influence of the rotational speed on the dia-
phragm radial stress can be ignored. Therefore, the influences
of the structure parameters on the diaphragm radial stress
had been discussed by using the numerical simulation
method which has been verified by the theoretical calculation
and the experiments.
Various fillets radii
When a fillet was applied to the perforated plate's groove, the
peak values of the diaphragm radial stress also occurred at
same points B, C and G. Fig. 8 shows the comparison of the
radial stresses of points B, C and G as the diaphragm clung to
the perforated plate at various fillet radii. The displacement of
point G was the maximum diaphragm's deflection in the
perforated plate's groove. It could be seen from Fig. 8 that the
displacement of point G increased from 0.07 mm to 0.21 mm
when the radius of the groove fillet changed from 0 mm to
2.0 mm. Meanwhile, the radial stresses of points B, C and G
increased slowly with increasing the fillet radius, but the
Fig. 8 e Comparison of the radial stresses of points B, C and
G as the diaphragm clung to the perforated plate at various
fillet radii.
radial stress increase rates of points B, C and G increased
obviously with the increase in the radius of the groove fillet.
This could be explained by the fact that the radial stress
caused by the small deflection deformation increased with
increasing the displacement of point G, but the radial stress
caused by large deflection deformation remained almost same
due to the same generatrix of the perforated plate. Hence, the
radial stress caused by small deflection deformation became
more and more influential in the diaphragm total radial stress,
and the radial stress increase rate of the points B, C and G
became larger and larger.
Various diaphragm thickness
The thicknesses of the diaphragms were set at 0.5, 0.8 and
1.0 mm, and the other parameters are showing Table 2.
Similarly, the peak values of the diaphragm radial stress still
occurred at same points B, C and G. The comparison of the
radial stresses of points B, C and G as the diaphragm clung to
the perforated plate at various diaphragm thicknesses is
presented in Fig. 9. It could be seen from Fig. 9 that the

Table 4 e Comparison of the radial stresses of points B, C

and G as the diaphragm clung to the perforated plate by
using different method.
Position Theoretical Experiment Numerical
calculation/MPa data/MPa simulation/MPa
Point B 272.6 251.5 276.3 Fig. 9 e Comparison of the radial stresses of point B, C and
Point C 275.7 255.9 281.7 G as the diaphragm clung to the perforated plate at various
Point G 220.4 207.0 222.4
diaphragm thicknesses.
6420 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1
displacement of point G decreased from 0.071 to 0.012 mm
when the diaphragm thickness changed from 0.5 to 1.0 mm.
The radial stresses of points B, C and G decreased obviously
with the increase in the diaphragm thickness, and the
maximum radial stress decrease rate was more than 36%. The
diaphragm bending rigidity increased with increasing the
diaphragm thickness, so the maximum diaphragm deflection
in the perforated plate's groove region (the displacement of
point G) decreased rapidly and the radial stresses of points B, C
and G decreased correspondingly.
Various groove widths
The widths of the perforated plate's groove were set as 8, 10 and
12 mm, and the other parameters are showing Table 2. As same
as above, the peak values of the diaphragm radial stress
occurred at points B, C and G. Fig. 10 shows the comparison of
the radial stresses of point B, C and G as the diaphragm clung to
the perforated plate at various groove widths. When the width
of the groove increased from 8 to 12 mm, the radial stress of
point B increased from 188.6 to 276.3 MPa, the radial stress of
point C increased from 194.1 to 281.7 MPa and that of point G
increased from 155.7 to 222.4 MPa. Meanwhile, the displace-
ment of the point G increased from 0.039 to 0.071 mm with
increasing the groove widths. This could be explained by the
fact that the maximum diaphragm deflection in the perforated
plate's groove region (the displacement of point G) increased
obviously due to the increase in the groove width, so the radial
stresses of points B, C and G became larger and larger.
Conclusions
A theoretical calculation method combined with the thin-
plate large deflection theory and the thin-plate small deflec-
tion theory was proposed to analyze the diaphragm radial
stress when the diaphragm clung to the perforated plate due
to the lack of the hydraulic oil. Based on the theoretical and
experimental investigations presented, the following conclu-
sions are drawn:
Fig. 10 e Comparison of the radial stresses of point B, C and
G as the diaphragm clung to the perforated plate at various
groove widths.
(1) While the piston reached the bottom dead center, the
diaphragm nearby the groove would endure the radial
stress caused by the large deflection deformation and the
additional stress caused by the perforated plate's groove.
(2) The maximal values of the diaphragm radial stress
during the working processing occurred at three points,
two contact points between the diaphragm and the
perforated plate's groove and the midpoint between the
two contact points, which explained why the dia-
phragmatic fracture frequently occurred at these posi-
tions. Meanwhile, for the general diaphragm
compressor, the value of radial stress caused by the
large deflection deformation was close to that caused by
the small deflection deformation at these points.
(3) The structure parameters of the diaphragm compressor
had a significant influence on the diaphragm radial
stress. Applying a fillet to the perforated plate's groove
increased the maximal radial stress of the diaphragm
while it clung to the perforated plate tightly. And
decreasing the groove width or increasing the dia-
phragm thickness could decrease the diaphragm
maximal radial stress efficiently.
references
[1] Kang JE, Brown T, Recker WW, Samuelsen GS. Refueling
hydrogen fuel cell vehicles with 68 proposed refueling
stations in California: measuring deviations from daily travel
patterns. Int J Hydrogen Energy 2014;39(7):3444e9.
[2] Li Z, Pan X, Sun K, Zhou W, Gao D, Liu S, et al. Development
of safety standard for mobile hydrogen refueling facilities in
China. Int J Hydrogen Energy 2014;39(25):13935e9.
[3] Brey JJ, Brey R, Contreras I, Carazo AF. Roll-out of hydrogen
fueling stations in Spain through a procedure based on data
envelopment analysis. Int J Hydrogen Energy
2014;39(8):4116e22.
[4] Weinert JX, Shaojun L, Ogden JM, Jianxin M. Hydrogen
refueling station costs in Shanghai. Int J Hydrogen Energy
2009;32(16):4089e100.
[5] Elgowainy A, Reddi K, Sutherland E, Joseck F. Tube-trailer
consolidation strategy for reducing hydrogen refueling
station costs. Int J Hydrogen Energy 2014;39(35):20197e206.
[6] Bhuiya MMH, Lee CY, Hwang T, Munira S, Hopkins R, Yoon H,
et al. Experimentally tuned dual stage hydrogen compressor
for improved compression ratio. Int J Hydrogen Energy
2014;39(24):12924e33.
[7] Khaitan SK, Raju M, McCalley JD. Design of a novel and
efficient hydrogen compressor for wind energy based storage
systems. Int J Hydrogen Energy 2015;40(3):1379e87.
[8] Reddi K, Elgowainy A, Sutherland E. Hydrogen refueling
station compression and storage optimization with tube-
trailer deliveries. Int J Hydrogen Energy 2014;39(33):19169e81.
[9] Worldwide Hydrogen Fueling Stations. Fuel cells. 2000. www.
fuelcells.org/info/charts/h2fuelingstations.pdf.
[10] Yu W, Dianbo X, Jianmei F, Xueyuan P. Research on sealing
performance and self-acting valve reliability in high-
pressure oil-free hydrogen compressors for hydrogen
refueling stations. Int J Hydrogen Energy
2010;35(15):8063e70.
[11] Rohland B, Eberle K, Stro bel R, Scholta J, Garche J.
Electrochemical hydrogen compressor. Electrochim acta
1998;43(24):3841e6.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 4 1 2 e6 4 2 1
[12] Grigoriev SA, Shtatniy IG, Millet P, Porembsky VI, Fateev VN.
Description and characterization of an electrochemical
hydrogen compressor/concentrator based on solid polymer
electrolyte technology. Int J Hydrogen Energy
2011;36(6):4148e55.
[13] Stro bel R, Oszcipok M, Fasil M, Rohland B, Jo
rissen L,
Grache J. The compression of hydrogen in an
electrochemical cell based on a PE fuel cell design. J Power
Sources 2002;105(2):208e15.
[14] Sun K, Pan X, Li Z, Ma J. Risk analysis on mobile hydrogen
refueling stations in Shanghai. Int J Hydrogen Energy
2014;39(35):20411e9.
[15] Zhiyong L, Xiangmin P, Jianxin M. Quantitative risk
assessment on 2010 Expo hydrogen station. Int J Hydrogen
Energy 2011;36(6):4079e86.
[16] Manfred D. Hofer diaphragm compressors. 2012. http://www.
andreas-hofer.de/E/downloads/
diaphragmcompressor10e02.pdf.
[17] Roy A, Watson S, Infield D. Comparison of electrical energy
efficiency of atmospheric and high-pressure electrolysers.
Int J Hydrogen Energy 2006;31(14):1964e79.
[18] Li J, Jia X, Wu Z, Peng X. The cavity profile of a diaphragm
compressor for a hydrogen refueling station. Int J Hydrogen
Energy 2014;39(8):3926e35.
[19] Wu B, Wu LZ, Wang XG, Yunfeng X. Analysis and study of the
diaphragm cavity profile of diaphragm compressors.
Compress Technol 2008;4(1):1e4.
[20] Ban L. Optimization research on concave surface of
diaphragm compressor. MsD Dissertation. China: Institute of
Mechanics, Chinese Academy of Sciences; 1998.
[21] Wang DS, Feng HP. Design of diaphragm chamber curve
surface shape with double function of a diaphragm
compressor. J Xi'an Jiaot Univ 1989;23(3):25e32.
6421
[22] Altukhov SM, Zhukov AV. Determining the geometric
parameters of the compression chamber of a diaphragm
compressor with a hydraulic drive. Chem Pet Eng
1986;22(7):309e12.
[23] Altukhov SM. Service life of compressor diaphragms. Chem
Pet Eng 1965;1(5):351e5.
[24] Li J, Zhao X, Jia X, Peng X. Design and experimental
verification for diaphragm compressor cavity profile by two
exponential terms generatrix. J Xi'an Jiaot Univ
2014;48(4):127e32.
[25] Kalshnikov VF, Kuznetsov VD, Antonova LI. Calculation of
the reliability of a compressor diaphragm under variable
loading. Chem Pet Eng 1987;23(4):168e70.
[26] Turvey GJ. A study of the behaviour of square plates at large
deflections according to the theories of Foeppl and Von
Karman. J Strain Anal Eng 1978;13(1):11e6.
[27] DaSilva PP, Krauth W. Numerical solution of the Von
Karman equations for a thin plate. Int J Mod Phys C
1997;8(2):427e34.
[28] Chien WZ, Yeh KY. On the large defection of circular plates.
China Sci 1954;10(3):209e38.
[29] Ramachandra LS, Roy D. A novel technique in the solution of
axisymmetric large deflection analysis of a circular plate. J
Appl Mech 2001;68(5):814e6.
[30] Liu RH. Large deflection of corrugated circular plate with
plane boundary region. Int J Nonlin Mech 1984;9(4):383e406.
[31] Dally JW, Riley WF. Experimental stress analysis. 3rd ed.
McGraw-Hill Higer Education; 1991.
[32] Jin GX, Su H. Determination of stress concentration factors
from measured strains. J Jiangsu Inst Technol 1993;14(1):19e24.