International Journal of Pressure Vessels and Piping 188 (2020) 104247
Contents lists available at ScienceDirect
International Journal of Pressure Vessels and Piping
journal homepage: http://www.elsevier.com/locate/ijpvp
Analytical modeling of FRP hubbed flanges
Ali Khazraiyan Vafadar, Abdel-Hakim Bouzid *, Anh Dung Ngô
Ecole de Technologie Superieure, 1100 Notre-Dame Ouest, Montreal (Quebec), Canada, H3C 1K3
A R T I C L E I N F O
A B S T R A C T
Keywords:
Analytical modelling
Composite flanges
FRP flanges
Structural integrity
Experimentation
Stress analysis
In spite of the increasing use of FRP composites in bolted flange joints and the good knowledge of these structures
and their material behavior the procedure used for their design is that of metallic flanges. There is a major
concern to appropriately address the anisotropic behavior of composite materials in FRP flange design. There
fore, it is necessary to revisit this procedure by implementing a more suitable model to analyse the stresses and
strains of FRP bolted flange joints.
This paper presents an analytical model that treats FRP bolted flange joints integrity and tightness based on
anisotropy and a flexibility analysis of all joint elements including the gasket, bolts and flanges. The theories that
treats all parts of the flange are described in details. Finally, the developed analytical model is validated with
experimental testing and numerical modeling of different size FRP flanges.
1. Introduction
bolted joint can only be expected if all the system components work
together in harmony.
To understand the behavior of fiber reinforced plastic bolted flange
joints, one of the early work was conducted by Sun in his thesis project
[4–6]. The author focused on the integrity of FRP flanges by proposing
an analytical study validated numerically using finite element modeling
(FEM). But it turns out that the analytical part was not reliable due to
some errors in his analytical flange ring model. The numerical model
was based on anisotropy material as the laminated shell theory was not
yet incorporated in the available software to allow a better handling of
the problem in 1995. Blach [7] used this model to propose a method for
a stress analysis of fiber reinforced plastic flanged connections with full
face gaskets. They used a few assumptions to simplify their proposed
method. Although they used a rigorous mathematical analysis to
describe all the complexity of the laminate materials, the physical con
straints related to the presence of the hub and the resulting two junctions
with the ring and shell was not considered in the analysis. Kurz and Roos
[8] studied the behavior of GRPF bolted flange joints, analytically and
experimentally. In their study, they divided the flange into several parts
each considered as a beam clamped at one end and subjected to a
bending moment at the other end. Consequently, their model only il
lustrates the effective portion of the flange subjected to a virtual radial
stress neglecting the stress in the tangential direction. The analytical
model that Estrada and Parsons proposed in their study, considers the
circumferential stresses and the rotation of the FRP stub flange. The
results obtained from FE models and experimental tests show a good
Fiber reinforced plastic composites have recently experienced a
spectacular development in the areas of pressure vessels and piping.
They are used in applications ranging from water and gas services to
nuclear and petrochemical industries. Their special properties such as
weight, chemical resistance, resistance to fatigue, low maintenance cost,
and even aesthetics have let these composites take precedence over
traditional materials [1,2]. Because of the ability to resist corrosive
environment, they are increasingly used in the chemical and petro
chemical process plants. They are a good alternative to nickel (Ni)
molybdenum (Mo) chromium (Cr) alloys at low temperatures [3]. Fluid
handling and transportation require safe bolted flange assemblies and
pressurized equipment. Indeed, the main purpose of a bolted flange joint
is to ensure the containment fluid and thus protect the immediate
environment against contamination from leakage of harmful fluids or
fluid escapes that are nauseating, toxic, and dangerous. Leakage may
result in a reduction of the efficiency of the installation and cause haz
ards to the workers or the public. In addition, the none-compliance to
the strict regulations on fugitive emission, the cost in the loss of revenue
and repair can be substantial. Therefore, the proper functioning of
bolted flange joints in a particular process is an important responsibility
of the user and the designer, as they must ensure compliance with the
laws, regulations and standards. The performance of a bolted flange
joint depends on its various components that it is composed of namely
the bolt gasket and the flanges. The satisfactory performance of the
* Corresponding author.
E-mail addresses: Ali.khazraiyanvafadar.1@ens.etsmtl.ca (A.K. Vafadar), hakim.bouzid@etsmtl.ca (A.-H. Bouzid), AnhDung.Ngo@etsmtl.ca (A.D. Ngô).
https://doi.org/10.1016/j.ijpvp.2020.104247
Received 21 June 2019; Received in revised form 21 August 2020; Accepted 28 October 2020
Available online 19 November 2020
0308-0161/© 2020 Elsevier Ltd. All rights reserved.
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
agreement with the analytical results [9]. Campbell proposed a new
technique to fabricate FRP flanges by adopting a one-piece integral
flange [10]. His design was based on the metal flange counterpart given
by ASME B16.5. In spite of the fact that a simple design method of FRP
flanges with full face gaskets is suggested in Ref. [11] the flange, bolts
and gasket flexibility as well as the elastic interaction between them are
neglected [12,13]. It is worth noting that in the ASME code designs of
FRP flanges and metal flanges, the effect of temperature on the redis
tribution of load and stresses is not addressed [14].
An overview of all researches done on FRP bolted flanged joints,
shows that most of the developed models are based on metallic flange
design considerations. In order to achieve accuracy in evaluating the
stresses in FRP flanges, it is necessary to adopt reliable design models
that describe the true material behavior and in particular the anisotropic
properties of the composite flanges [15]. Finally, it is worth noting that
very few papers discuss the leak rate performance of FRP bolted flange
joints [16].
This study concentrates mainly on the development of an analytical
model for fiber reinforced plastic bolted flanged joints based on lami
nation theory. The latter is applied to composite materials to account for
the anisotropic properties such as those encountered in FRP flanges. The
developed analytical model is based on the force and moment equilib
rium, the compatibility of the radial displacements and the rotations of
the flange different elements at their junctions as well as the axial
compatibility necessary to solve the statically undeterminate problem to
give the stresses at the inside and outside surfaces of the flange. The
proposed analytical model is validated by comparison with experi
mental data and finite element modeling conducted on NPS 3 and 12
class 150 FRP hubbed flanges.
2. Analytical model
In developing the analytical model of FRP bolted flange joints, one of
the objectives was that the approach follows the ASME code design
philosophy [3]. Therefore, the methodology is based on the two
important factors: structural integrity and leakage tightness. Although
the paper focus on structural integrity, the leakage verification can
easily be achieved by conducting experimental leak tests such as those
specified by ASTM F2836 or EN13555 that link the gasket load and
pressure with leakage. The developed model takes into account the
flexibility of the gasket, the bolts and the flanges as well as the elastic
interaction between them. It covers the two operating conditions:
Fig. 1. Analytical model for the FRP flange with the hub (a) bolt-up (b)
pressurization.
following assumptions are made:
• The tightening torque during phase i, is applied to all bolts of the
flange equally and simultaneously.
• The axial force resulting from tightening all the bolts is considered to
act as a ring load on a circle of constant radius of the flange known as
the bolt circle radius. This assumption makes it possible to consider
the flange as an axisymmetric case to simplify the analysis.
• A plane of symmetry is considered at the midplane of the joint. This
limits the model to the case of identical pair of flanges.
• Initial tightening or phase i: in this first stage the bolt-up of the flange
is conducted to properly seat the gasket.
• Pressurization of final phase f: an internal pressure is applied to the
inside of the vessel or pipe and consequently the flange. The inside
pressure acts on the flange wall in the radial direction but also cre
ates a hydrostatic end effect, which is represented by an axial stress
acting on the flat surfaces at the end of the cylindrical shell.
The analytical model of the FRP bolted flange joint with the hub is
illustrated in Fig. 1. In order to model the bolted flange joint analyti
cally, the flange is divided into three distinct parts which are connected:
these are the flange ring or plate, the hub and the shell. Fig. 1a shows the
flange in phase n = i or bolt-up and Fig. 1b shows the flange in phase n =
f or pressurization. Point 1 shown in Fig. 1 is the junction between the
flange ring and the hub and point 2 is the junction between the hub and
the shell.
The flange ring is treated as a thin composite laminated plate sub
jected to bending in the tangential direction and as a thick laminate
cylinder subjected to lateral pressure. The shell is treated as long lami
nate cylinder and the hub as short laminate cylinder subjected to pres
sure and edge loads for which the beam on elastic foundation theory is
applied.
To simplify the analytical model of the FRP bolted flange joint the
2.1. Flange ring model
The theory of plates based on the Kirchhoff-Love hypothesis char
acterize well the behavior of the composite flange ring. It is considered
to be a circular plate subjected to a moment in the tangential direction
and ring loads in axial direction on its free inner edge and balanced by a
simple support on its outer edge. In addition, the theory of thick-walled
cylinder is used to obtain the radial displacement of the ring produced
by pressure and the radial edge load.
Applying equilibrium of forces in x and r directions and moment
about θ to the plate element of volume rdθ by dr by t shown in Fig. 2 and
using the governing equations of the classic laminate plate theory the
following is obtained.
2
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
Nr =
du0
dψ
u0
ψ
Ap,rr + Bp,rr + Ap,rθ + Bp,rθ
dr
dr
r
r
(14)
and therefore
Nr =
du0
u0
Ap,rr + Ap,rθ
dr
r
(15)
similarly
∫t/2
du0
u0
Ap,rθ + Ap,θθ
dr
r
(16)
σ r xdx =
dψ
ψ
Dp,rr + Dp,rθ
dr
r
(17)
σθ xdx =
dψ
ψ
Dp,rθ + Dp,θθ
dr
r
(18)
σθ dx =
Nθ =
− t/2
∫t/2
Mr =
− t/2
Fig. 2. Plate element model.
∫t/2
Mθ =
− t/2
d(rNr )
=0
dr
Nθ −
(1)
(2)
d
(rQrx ) = 0
dr
(
)
1
dw
ψ + 0 Ap,rx
2
dr
(19)
− t/2
Noting that the FRP plate stiffnesses are given by
(3)
(
with the following boundary conditions:
At r = ri :
∫t/2
)
Ap,rr , Ap,rθ , Aθθ =
(20)
(Qrr , Qrθ , Qθθ )dx
− t/2
Nr = − pt + Qn1
(4)
Qn t
+ 1
2
(5)
Mr = M1n
τr dx =
Qrx =
d
(rMr ) = 0
dr
Mθ + rQrx −
∫t/2
(
(Qrr , Qrθ , Qθθ )xdx = (0, 0, 0)
(21)
− t/2
(6)
Qrx = − F
∫t/2
)
Bp,rr , Bp,rθ , Bp,θθ =
(
)
Dp,rr , Dp,rθ , Dp,θθ =
and at = r0 :
∫t/2
(Qrr , Qrθ , Qθθ )x2 dx
(22)
− t/2
(7)
Nr = 0
∫t/2
(23)
Mr = 0
(8)
Ap,rx =
w0 = 0
(9)
Now, substituting Eqs. (15) and (16) into the equilibrium equation
(1) gives the differential equation governing the radial displacement of
the flange ring at its mid-thickness.
According to the strain displacement field, we have:
εr =
du0
dψ
+x
dr
dr
(10)
εθ =
u0
ψ
+x
r
r
(11)
(
εrx =
1
dw
ψ+ 0
2
dr
r2
(12)
∫t/2
− t/2
+ Qrθ
∫t/2
σr dx =
(Qrr εr + Qrθ εθ )dx =
− t/2
r
(24)
d2 ψ dψ Dp,θθ ψ
ri F
+
−
=−
Dp,rr
dr2 dr
Dp,rr r
is:
)
∫t/2 [ (
du0
dψ
Qrr
+z
dr
dr
(25)
After application of the boundary conditions, the solution of Eq. (24)
(
)
(
)
pt − Qn1 (r0 )m2 − 1
pt − Qn1 (r0 )m1 − 1
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) rm1 − (
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) rm2
uf0 = (
4 Ap,rθ + Ap,θθ Ap,rr Δ
4 Ap,rθ −
Ap,θθ Ap,rr Δ
− t/2
]
)
]
∫t/2 [ (
(u
(u
ψ)
du0
dψ
ψ)
0
0
Qrr
+z
dx =
+z
+ Qrθ
+z
dx
r
r
dr
dr
r
r
d2 u0
du0 Ap,θθ
+r
u0 = 0
−
dr2
dr
Ap,rr
and substituting Eqs. (17)–(19) into the equilibrium equation (2) and
noting that at r = ri , rQrx = − ri F, gives the differential equation gov
erning the rotation of the flange ring at its mid-thickness.
)
u0 and w0 are the radial and transverse displacement of the ring mid
plane.
The membrane force is given by
Nr =
Qrx dx
− t/2
where: m1 =
(13)
√̅̅̅̅̅̅̅
Ap,θθ
Ap,rr , m2 = − m1
Δ = (2ro )m1 − 1 (2ri )m2 − 1 − (2ro )m2 − 1 (2ri )m1 −
− t/2
In terms of plate stiffness the membrane force is
(26)
1
(27)
In the case of equivalent elastic properties of the ring in the radial
3
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
and tangential directions, we have Ap,θθ = Ap,rr and m1 = − m2 = 1 and
therefore
Δ=
ro2 − ri2
2ro2 − ri2
(28)
Therefore, the radial displacement of the ring is given by:
(
)
(
)
pt − Qn1 ri2
pt − Qn1 ro2 ri2
1
)(
)
)(
)×
(
u0 (r) = (
×
r
−
r
Ap,rθ + Ap,rr ro2 − ri2
4 Ap,rθ − Ap,rr ro2 − ri2
(29)
The rotation about a tangential axis of the ring is obtained from the
solution of the differential equation (25) noting that the transformed
lamellae stiffness Dp,rr and Dp,θθ are equal
(
( )
)
dw
Fr
r
Cn C n r
(30)
ψ(r) = =
2 ln
− 1 − 2− 1
dr 8πDp,rr
ro
2
r
The transverse displacement w of the flange ring in the transverse
direction subjected to a concentrated force F and edge loads Mn1 and Qn1
at junction 1, can be obtained by integrating Eq. (30):
( ( )
)
( )
Fr2
r
r
C n r2
ln
− 1 − C2n ln
− 1 + C3n
w(r) =
(31)
ro
ro
8πDp,rr
4
where F is the gasket force applied on the flange ring. The constants Cn1 ,
Cn2 and Cn3 are determined from the following boundary conditions:
at r = ri ,
w=0
at r = ri ,
Mr = M1n −
at r = ro ,
Fig. 3. Shell element model.
(32)
tp n
Q
2 1
Mr = 0
(33)
(34)
⎫ ⎡
⎧
A11
Nx ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎢ A12
Nθ ⎪
⎪
⎪
⎪
⎪
⎢
⎬ ⎢
⎨
Nxθ
A16
=⎢
⎢
M
⎪
⎪
x
⎪ ⎢ B11
⎪
⎪
⎪
⎪
M ⎪
⎪ ⎣ B12
⎪
⎪
⎭
⎩ θ⎪
Mxθ
B16
where n is equal to i for bolt-up and f for pressurization.
Appendix A gives the detail constants used for the flange ring.
2.2. Hub and cylinder model
The cylinder part of the FRP flange is modeled as an axisymmetric
laminated shell subjected to internal pressure and edge loads. It is
considered as a long cylinder of semi-infinite length i.e. the length is
√̅̅̅̅̅̅
greater than π/βs = 2.45 ri ts . The thin shell theory will apply in this
case. At the finite end, the shell is subjected to the discontinuity shear
force Q and edge moment M. The long thin cylinder equations for the
displacements, rotation, bending moments, and shearing forces in terms
of conditions at any locations x are given by the theory of beams on
elastic foundation.
The hub can also be considered as a thin cylinder subjected to
axisymmetric loading. However, the length of the hub is considered
short and less than π/β. In this case, both ends of the hub are subjected to
the discontinuity shear force Q and moment M in addition to the hy
drostatic end force Nx . Applying the equilibrium of forces in the axial x
and radial r directions and moment about the tangential direction θ to
the shell or hub element of volume rdθ by dx by t shown in Fig. 3, the
following governing equations are derived:
dNx
=0
dx
Q−
dMx
=0
dx
dQ Nθ
−
=− p
dx
r
A12
A22
A26
B12
B22
B26
A16
A26
A66
B16
B26
B66
B11
B12
B16
D11
D12
D16
B12
B22
B26
D12
D22
D26
⎧ 0 dw ⎫
εx =
⎪
⎪
⎪
dx ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
u
⎪
⎪
0
⎪
ε
=
⎪
⎪
⎤⎪
θ
⎪
⎪
⎪
r ⎪
B16 ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
B26 ⎥
⎪
⎪
⎪
dv
⎥⎪
0
⎨ γ =
⎬
⎥
B66 ⎥
xθ
dx
⎥
D16 ⎥⎪
⎪
⎪
⎪
⎪
⎪
⎪
D26 ⎦⎪
d2 u ⎪
⎪
0
⎪
⎪
⎪
⎪
κ
=
−
⎪
⎪
x
2
⎪
D66 ⎪
dx
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 0
⎪
⎪
⎪
⎪
⎪
κ
=
0
⎪
⎪
θ
⎪
⎪
⎩
⎭
(38)
κ0xθ = 0
where:
Aij , Bij and Dij are the stiffness coefficients. According to the sym
metrical laminate shell A16 = A26 = D16 = D26 = 0 and [B] = 0. A
subscript s is added to these stiffness coefficients when referring to the
cylinder and h to the hub.
The stress resultants related to the shell displacement are:
Nx = A11
Nθ =
dw
u
+ A12
dx
r
)
(
u
A2
A12
A22 − 12 +
Nx
r
A11
A11
(39)
(40)
(35)
Mx = − D11
d2 u
dx2
(41)
(36)
Mθ = − D12
d2 u
D12
=−
Mx
dx2
D11
(42)
(37)
Mxθ = −
D16
Mx
D11
(43)
Substituting Eq. (39) and Eq. (40) into Eq. (36) and Eq. (37) and
combining the last two gives the following:
(
)
d4 u u
A2
A12 Nx
D11 4 + 2 A22 − 12 +
=p
(44)
dx
r
A11
A11 r
The relation between the stress resultants and strains are:
4
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
The displacement of the shell after integration is:
β x
− β x
u = e [C1 cos(β x) + C2 sin(β x)] + e
[C3 cos(β x) + C4 sin(β x)]
requirement. The final bolt load after pressurization should then be
verified for leakage after calculation. Consequently, during bolt-up the
system of equations is reduced to eight unknowns and the axial
compatibility equation is not required.
The radial displacements, rotations, radial shear force and moment
of the hub are equal to those of the flange ring at junction 1 and to those
of the shell at junction 2:
At junction 1
(45)
The rotation ψ, Moment M and shear forces Q are given as follows:
ψ=
du
dx
Mx = − D11
(46)
d2 u
dx2
(47)
un1,p = un1,h
(49)
(48)
ψ n1,p = ψ n1,h
(50)
The integral constants C1 , C2 , C3 and C4 can be determined from the
following boundary conditions for both bolt-up and pressurization, for
the hub:
At x = 0, then Mx = Mn1 and. Q = Qn1
At x = lh , then Mx = Mn2 and Q = Qn2 . Appendix B gives the detailed
expressions for the radial displacement, rotation, moment and shear
force for the hub.
for the shell:
Since the cylinder is considered as a long and infinite shell, the
displacement, rotation, moment and shear force of the shell are zero
when x tends to infinity. This gives C1 = 0 and C2 = 0. Also, at x = 0,
then, M = Mn2 and Q = Qn2 . Appendix C gives the detailed expressions for
the radial displacement, rotation, moment and shear force for the shell.
Qn1,h = Qn1,p
(51)
n
n
M1,h
= M1,p
(52)
2
Q=
dMx
d w
= − D11 2
dx
dx
At junction 2
un2,h
(53)
= un2,s
ψ n2,h = ψ n2,s
(54)
Qn2,s = Qn2,h
(55)
n
n
M2,s
= M2,h
(56)
Fig. 4 illustrates the compatibility of displacement based on the axial
distance traveled by the nut during the tightening process. It is to be
noted that this distance is the sum of the displacement of the flange due
to rotation, the elongation of the bolt and the compression of the gasket
all in the axial direction [2].
Considering the axial compatibility:
2.3. Compatibility equations
The equations of compatibility requiring continuity of rotation and
radial displacement at the two junctions can be used with the equations
of equilibrium of radial shear forces and moments to form a system of
equations with eight unknowns in the seating condition and nine un
3
∑
f
knowns for the pressurization condition because the final bolt force FB is
not known. For the operating condition this load can be obtained from
the geometric compatibility consideration in the axial direction. The
distance traveled by the nut during tightening remains constant and is
therefore equal in both the initial bolt-up and pressurization states.
Before applying the internal pressure, the bolt initial force FiB , which is
the minimum required bolt load that mus satisfy the seating
i=1
wii =
3
∑
wfi
(57)
i=1
or
wib + wig + wip = wfb + wfg + wfp
In terms of stiffness and angular rotation Eq. (58) becomes:
Fig. 4. Axial compatibility: a) hand tightening, b) bolt-up), c) pressurization.
5
(58)
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
FBf FGf
Fi Fi
+
+ (rB − rG )ψ f1 = B + G + (rB − rG )ψ i1
KB KG
KB KG
⎡
(59)
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
[V] = ⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢(
⎣ 1
Considering the axial equilibrium.
Initial pre-tightening state (i)
(60)
FiB = FiG
Final pressurization state (f)
(61)
FfB = FfG + pAp
Substitution of Eqs. (60) and (61) into Eq. (59) gives
(
) (
)
)
(
1
1
1
1
AP
FBf
+
+
p
=
F i + 2(rB − rG ) ψ I1 − ψ f1 +
KB KG
KB KG B
KG
(62)
After assembling the above equations, a system of 9 unknowns is
obtained. The equations are put into a matrix form to solve for the
unknowns.
0
1
Ah,12
2
p
4
4βh Ah,1 Dh,11
Ah,11 −
−
0
1
As,12
2
−
p
4
4βs As,11 Ds,11
As,11 −
0
− (L21 − L22 )tp p
) L27
(
− (rG − ri ) rG2 + ri2
p
4rm
)
1
AP
F i + 2(rB − rG )ψ I1 +
+
p
KB KG B
KG
(63)
[U] = [C]\[V]
1
Ah,12
2
−
p
4
4βh Ah,1 Dh,11
Ah,11 −
[U] is a vector composed of the 9 unknown elements: the two edges
forces and two edge moments, the radial displacements and the rotations
at both junctions and the bolt force.
⎡
⎤
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎦
(65)
⎤
− B11
⎢ 2β3 D
⎢ h h,11
⎢
⎢ − B
⎢
12
⎢ 2
⎢ 2βh Dh,11
⎢
⎢
⎢ − G11
⎢ 3
⎢ 2βh Dh,11
⎢
⎢
⎢ G12
⎢ 2
⎢ 2β Dh,11
⎢ h
⎢
⎢
1
[C] = ⎢
⎢ 2β3 D
⎢ s s,11
⎢
⎢ − 1
⎢
⎢ 2
⎢ 2βs Ds,11
⎢
⎢
⎢
0
⎢
⎢
⎢
⎢
⎢
0
⎢
⎢
⎢
⎣
0
B12
2β2h Dh,11
G11
2β3h Dh,11
G12
2β2h Dh,11
− 1
0
0
0
B22
2βh Dh,11
G12
2β2h Dh,11
G22
2βh Dh,11
0
− 1
0
0
G12
2β2h Dh,11
B11
2β3h Dh,1
B12
2β2h Dh,11
0
0
− 1
0
− G22
2βh Dh,11
− B12
2β2h Dh,11
− B22
2βh Dh,11
0
0
0
− 1
1
2β2s Ds,11
0
0
− 1
0
0
0
− 1
βs Ds,11
0
0
0
− 1
0
0
0
− (L21 − L22 )
0
0
0
− 1
tf
2
0
tf
L26
2
− L26
0
0
0
− 1
0
0
0
0
0
0
2(rB − rG )
[
]T
[U] = Qn2 , M2n , Qn1 , M1n , un2 , ψ n2,h , un1 , ψ n1 , FBn
0
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
⎥
⎥
0
⎥
⎥
⎥
rB − rG ⎥
L27 ⎥
⎥
2 π rm
⎥
⎥
1
1 ⎦
KB
+
(66)
KG
To obtain the final bolt force FBf which depends on the flange pa
rameters as well as the inside pressure, first the unknown vector [U] has
to be solved for the initial bolt-up condition with a known initial bolt
force FBi . The solution gives in particular the rotation of the flange at the
initial tightening condition that is used in the last row in [U] in the final
(64)
[C] is a square matrix and [V] is a vector defined as follows:
condition to solve for the bolt force FBf .
2.4. Stress calculation
To calculate the stresses and displacements in the hub and shell, the
solutions of the two vectors [Ui ] and [Uf ] relative to the initial and final
conditions n = i and n = f are first obtained. The longitudinal and
tangential stresses at the inside and outside surfaces of the hub are given
6
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
σ nh,l (x) =
n
Nh,l
(x) 6Mhn (x)
±
th2
th
(69)
σ nh,t (x) =
n
Nh,l
t(x) 6Dh,12 Mhn (x)
±
th
th2
Dh,11
(70)
The longitudinal and tangential stresses at the inside and outside
surfaces of the shell for the initial and final conditions n = i and n = f are
given by:
i
Ns,l
(x) = 0
and
(
As,11 −
⎡
n
⎢us (x)
n
Ns,t
(x) = ⎣
rs
f
Ns,l
(x) =
+
p rs
2
)
As,12
2
4β4s As,11 Ds,11
(71)
⎤
(
)
A2s,12
p⎥
As,12 n
+
N (x)
⎦ As,22 −
rs
As,11
As,11 s,l
(72)
σ ns,l (x) =
n
Ns,l
(x) 6Msn (x)
±
ts2
ts
(73)
σ ns,t (x) =
n
Ns,t
(x) 6Ds,12 Msn (x)
±
ts
Ds,11
ts2
(74)
3. FEM modeling
Two finite element models representing the NPS 3 and 12 class 150
FRP bolted flanged joints were developed under the finite element
software ANSYS® R16.2 in order to validate the proposed analytical
model. The 3D solid models are divided into three parts; the flange, the
gasket and the bolt. The model of the bolted flange assembly includes
only one flange and the gasket with half of its thickness due to the
symmetry of bolted flange joint with respect to a plane that passes
through the mid gasket thickness (Fig. 5). The FRP flanges are made out
of laminates of E-glass fibers for reinforcement and vinyl-ester as the
matrix. The three major parts (ring, hub and shell) have different
thickness and arrangement of plies but are all composed of mat and
woven roving laminates. It is worth noting that the laminates are hand
Fig. 5. Finite element model for the NPS 12 FRP flange.
in terms of the longitudinal and tangential membrane forces Nl and Nt :
i
Nh,l
(x) = 0
and
p rh
2
(
⎡
n
⎢uh (x)
n
Nh,t
(x) = ⎣
f
Nh,l
(x) =
rh
)
Ah,11 −
+
Ah,12
2
4β4h Ah,11 Dh,11
(67)
⎤
(
)
A2h,12
p⎥
Ah,12 n
+
N (x)
⎦ Ah,22 −
rh
Ah,11
Ah,11 h,l
(68)
Fig. 6. NPS 3 class 150 FRP Flange a) Experimental Test Rig. b) strain rosettes bounded to the hub.
7
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
Fig. 7. Gasket load-compression data curve.
Fig. 10. Longitudinal stress distribution of NPS 12 class 150 flange after
pressurization.
Fig. 8. Longitudinal stress distribution of NPS 12 class 150 flange after bolt-up.
Fig. 11. Tangential stress distribution of NPS 12 class 150 flange after
pressurization.
fundamental building blocks of a good simulation to reproduce the real
behavior of the bolted flange joint is to employ the correct element type
of the contact surface in the FE model. The contact element used in all
deformable contact surfaces is CONTA174 element that has 8 nodes and
is used in conjunction with 3D target interface surface element
TARGE170.
The flanged connection is assumed to be sufficiently far from the end
enclosures of the vessel, such that the bending and shear between the
shell and cap are only limited to their junction and does not affect the
length of the shell under consideration. Because of the axisymmetric
analysis, the boundary conditions applied in these finite element models
are the symmetry about a plane passing through gasket mid-thickness.
The nodes that belong to this plane are constrained from any move
ment in the axial direction. The bolt symmetrical plane is free to move in
the radial direction while the rotation about a plane perpendicular to the
flange axial direction is fixed. Since the flange is symmetrical in nature,
the cyclic symmetry is applied to the two sides of the cut section.
A standard friction model is used to simulate the contact and controls
sliding. The coefficients of static friction of 0.7 for rough surface and
0.15 for smoother surface are used. The pressure applied to the internal
surface of the NPS 3 and 12 flanges is about 1 MPa (150 psi). The lateral
pressure is applied to the flange internal wall while the hydrostatic end
load is applied to the end of the pipe. By applying a pre-stress equivalent
Fig. 9. Tangential stress distribution of NPS 12 class 150 flange after Bolt-up.
layered-up with midplane symmetry. Fig. 5 shows the mesh of the model
of the NPS 12 flange with the hub. In Ansys workbench, all parts of the
flange are modeled with SOLID185 elements while the gasket is meshed
with INTER195 elements. Both element types have 8 nodes. One of the
8
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
Fig. 12. Radial displacement in NPS 12 class 150 flange.
Fig. 14. Tangential stress distribution of NPS 3 class 150 flange after bolt-up.
Fig. 13. Longitudinal stress distribution of NPS 3 class 150 flange after bolt-up.
Fig. 15. Longitudinal stress distribution of NPS 3 class 150 flange after
pressurization.
to a load of 14600 and 10,000 N on the bolts of the two flanges
respectively, the initial gasket seating is achieved. In the ANSYS the
torque load is applied through the pre-existing command bolt
pretension.
In this model, the gasket deforms filling the irregularities on the
flange face, insuring full contact over its entire surface. The mechanical
behavior of the gasket is given by the nonlinear curves of gasket contact
stress versus axial displacement shown in Fig. 6.
hub locations on the flange as shown in Fig. 7b. The test was conducted
at room temperature (22 ◦ C) and all measuring sensors were calibrated
at this temperature. During the test, the bolt force, the gasket
displacement, the strains at selected locations of the flange surface, the
temperature and internal pressure were continuously monitored
through a data acquisition and control unit connected to a computer. A
special software working under LabView was written to monitor and
control the test parameters.
4. Experimental set-up
5. Results and discussion
The test rig used in the experimental investigation of FRP bolted
flanged joints is shown in Fig. 7a. The test rig is made of an ANSI B16.5
NPS 3 Class 150 FRP full face flange used with a metallic flange of the
same type and size and a 1/16 Teflon full face gasket. The FRP flange is
made of glass mat and woven roving impregnated with vinyl ester resin.
An initial bolt load of 14600 N was applied gradually by a torque wrench
to each instrumented bolt using a tightening procedure based on the
criss-cross pattern conducted with three load step levels. The load is
measured through a high temperature extensometer device made of
mainly a ceramic rod and tube that transmit the displacement to a straingaged beam located far from the heated flanges. The extension of the
displacement difference between the drilled bolt and the central rod
welded to its end gives the load applied with a good precision. A gas
pressure of 1 MPa was applied to the bolted joint. Three full bridge strain
rosettes were bonded with adhesive to the outside surface at the selected
For validation of the analytical model the results are presented and
compared to the numerical results obtained from FE modeling and
experimentation. The longitudinal and tangential stress distributions at
the inside and outside surfaces of the flange at bolt-up are illustrated in
Figs. 8 and 9 for the NPS 12 and 13 and 14 for NPS 3. These figures show
clearly that the analytical and FE stress distributions at the inside and
outside flange surfaces during bolt-up have a similar trend. The distri
butions of the longitudinal and tangential stresses in both cases are in a
good agreement and the difference is less than 4% far from the two
junctions where stress concentration is expected. These results match
well with the experimental stresses deduced from the measured strain
gages on the NPS 3 class 150 flange. As anticipated during bolt-up, due
to the lack of the pressure inside of the FRP bolted flange joints, the
9
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
stresses.
Although at the junctions, the stress at the inner and outer surfaces
show a higher difference, the distributions are consistent. In the case of
the NPS 12, there is a location where a drastic change of longitudinal
stress occurs. This is less visible in the case of the NPS 3 flange.
Figs. 13–16 show a relatively good agreement between the analytical
and FE models and the experimental data during pressurization except
at the junctions. Nevertheless, while the results of the tangential stresses
at the inner and outer flange surface are shown to have the same trend,
those at the hub show a discrepancy because the hub is considered as a
cylinder with an equivalent thickness rather than tapered. This simpli
fication is acceptable in the design of such a complex structure.
In addition, at the junction of the hub and the shell where stress
concentration exists, the stresses at this location increases drastically by
more than 92%. This phenomenon also occurs near the hub to the flange
ring junction. It is to be noted that since the small tapered portion of the
hub of the NPS 12 flange at the junction with the shell that is not
accounted for in the analytical model, the results would be affected.
Nonetheless, considering the fact that the analytical model is developed
to analyse general membranes stresses and not peak stresses generated
by stress concentration, the difference is not significant and is acceptable
for design purposes.
The radial displacement results obtained from the numerical and
analytical models of the two FRP bolted flange joints at bolt-up and
pressurization are compared in Figs. 12 and 17. As shown in these fig
ures, the general distributions have a remarkably similar trend. The
difference observed between the analytical and numerical results is once
again higher at the junctions but compares well elsewhere in both cases.
Finally, the radial displacement of the shell is shown to be much higher
than the flange ring and the hub during pressurization because these two
structures are much stiffer. The results of the radial displacements
confirm the robustness of the analytical developed model.
Fig. 16. Tangential stress distribution of NPS 3 class 150 flange after
pressurization.
6. Conclusion
This study presents a stress analysis methodology that handles the
anisotropy behavior of the complex structures of FRP bolted flange
joints. The modelling of the FRP flange composed of the three elements ;
the ring, the hub and the shell, is done using the different theories of
laminate materials. The analytical developed model was tested and
validated using FEM and experimental tests conducted on two different
size flanges. It follows the general ASME code design philosophy but
accounts for the anisotropy multilayer modelling as well as flexibility.
From the results obtained numerically and experimentally, it could
be stated the proposed analytical model for FRP flanges, has proved to
be efficient, accurate and reliable in predicting the longitudinal and
tangential stress distributions at inside and outside flange surfaces
including the radial displacement of the flange. However, at the junc
tions of the hub with the ring and shell, the predictions are less accurate
due to stress concentrations and laminate interconnections between the
elements. The tangential stresses at the hub inside surface are those that
show the highest difference. However, in terms of radial displacement,
the importance of considering anisotropy is demonstrated. The results of
this work have led to the conclusion that the presented model has great
potentials: it provides a new formulation that could be used as an
alternative to the ASME code section X standard while allowing accu
racy, reliability and thorough stress analysis of FRP bolted flange joints.
Additional experimental tests on FRP materials with different properties
in the two longitudinal and tangential directions are curently being
conducted to confirm these findings.
Fig. 17. Radial displacement in NPS 3 class 150 flange.
stresses in the shell far away from the hub and shell junction are equal to
zero.
Figs. 10 and 11 and 15 and 16 present the longitudinal and tangential
stress distributions of the NPS 12 and 3 flanges respectively obtained
after application of an internal pressure of 1 MPa. A similar stress dis
tribution trend along the hub and cylinder length is observed in both the
analytical and FE models. However, the FE numerical results show
higher stress values at the outside flange surface near the junctions of the
hub and the shell and the hub and ring flange. On comparing the
analytical and numerical results in the shell far away from the junctions,
the highest difference observed in the case of the NPS 12 flange is
around 22% and 18% for the longitudinal stress distribution at the
outside and inside diameters respectively. While this difference may
seem acceptable, for the case of the NPS 3 flange, the tangential stress at
the inside diameter of the hub with both methods are completely the
opposite, especially near the junction with the flange. It is difficult to
confirm with the experimental measurement since the bounding of
strain gages at the inside diameter is not an easy task because of the
complexity of introducing the wiring inside the vessel. Nevertheless,
such differences are anticipated because these junctions are smoother
and the laminates are not interconnected in the FE models. Moreover,
the presence of the high local stresses in the composite flanges affects
considerably the stress distribution particularly at the outside surfaces of
the hub and the shell. Since the material properties vary through the
thickness of the composite flanges, there is an even distribution of
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
10
A.K. Vafadar et al.
International Journal of Pressure Vessels and Piping 188 (2020) 104247
Nomenclature
Ab
Ag
Ap
Ai,mn
As
Bi,mn
C1 to C4
Di,mn
db
E
F
K
l
M
n
N
p
Q
Qrx
r
rb
t
u
w
x
Poisson ratio
Hub constant [mm− 1]
Shell constant [mm− 1]
Strain
Tangential direction
Normal stress [MPa]
Shear stress [MPa]
Rotation of the plate [rad]
Bolt area [mm2]
Full gasket contact area [mm2]
Pressure area [mm2]
In-plane laminate modulus [MPa.mm]
Section cut area of the shell [mm2]
Coupling laminate modulus [MPa.mm2]
Constants of integration
Flexural laminate modulus [MPa.mm3]
Diameter of the bolt [mm]
Young Modulus [MPa]
Force [N]
Uniaxial stiffness [N/mm]
Length [mm]
Discontinuity moment [N]
Number of bolts
Membrane force [N/mm]
Internal pressure [MPa]
Discontinuity or shear force [N]
Shear force in the plane rx [N]
Radial direction or radius [mm]
Bolt circle [mm]
Thickness [mm]
Radial displacement [mm]
Axial displacement [mm]
Axial direction
Subscript
0
1
2
b, B
f
g, G
h
i
l
r
s
t
o
p
x
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
refers
ν
βh
βs
ε
θ
σ
τ
ψ
to mid plane
to junction between ring and hub
to junction between hub and shell
to bolt
to flange
to gasket
to hub
to inside
to longitudinal direction
to radial
to shell
to tangential direction
to outside
to plate or ring
to axial
Superscript
i
Refers to initial tightening of the bolt condition
f
Refers to final condition (pressurization)
n
Refers to working condition, n=i,f
Appendix a. Flange ring Details
The constants of the flange ring axial displacement and rotation used in the analysis are obtained for the two conditions of bolt-up n=i and
pressurization n=f. The bolt force F and edge loads Q and M are also related to those two conditions:
11
A.K. Vafadar et al.
1
(
)(
4πDp,rr ro2 − ri2 Dp,rr
C1n = −
C2n =
International Journal of Pressure Vessels and Piping 188 (2020) 104247
4πDp,rr
(
( )
]
(
)(
)
(
)
(
r
t )
) Dp,12 − Dp,rr ro2 − ri2 F + 2 Dp,rθ + Dp,rr ri2 log i − 8πri2 Dp,rr M1n − f Qn1
ro
2
+ Dp,rθ
[
( ) ]
[
(
)
ro2 ri2
tf n ) (
ri
n
)(
)
π
D
−
+
D
M
−
D
F
log
4
Q
p,rr
p,rθ
p,rr
1
1
2
ro
ro2 − ri2 Dp,rr − Dp,12
(A1)
(A2)
[(
)(
(
)(
(
) (r )
)
− ri2
t )
) D2p,rθ − 3D2p,rr + 2Dp,rθ Dp,rr ro2 − ri2 F − 8πri2 D2p,rr − Dp,12 Dp,rr M1n − f Qn1 − 2ro2 D2p,rθ − D2p,rr log i F − 4ro2
(
)
2
ro
16πDp,rr ro2 − ri2 D2p,rr − D2p,rθ
(
)
)
(
]
(
)
(
)
ri
ri ( n tf n )
D2p,rθ + D2p,rr + 2Dp,rθ Dp,rr log 2
(A3)
F − 16π ro2 D2p,rr − Dp,rθ Dp,rr log
M1 − Q1
ro
ro
2
C3n =
(
In order to alleviate the system of equations the following is used.
To simplify the derivations the following coefficients are defined:
L21 = (
L22 = (
− r3
)( i
)
r2o − r2i Ap,12 + Ap,11
(A4)
− r r2
)( i o
)
Ap,12 − Ap,11
(A5)
r2o
−
r2i
( )
ri − 2ri log
L23 =
(A6)
8π Dp,11
(
L24 = −
ri
ro
( )
)(
)
)
(
r2o ri + r3i Dp,12 − Dp,11 + 2r3i log rroi Dp,11 + Dp,12
(
)(
)
8πDp,11 r2o − r2i Dp,11 + Dp,12
(A7)
( )
)
(
− r2o ri log rroi Dp,11 + Dp,12
(
)(
)
L25 =
4Dp,11 π r2o − r2i Dp,11 − Dp,12
(A8)
(A9)
L26 = L21 + L22
(A10)
L27 = L23 + L24 + L25
Appendix B. HUB DETAILS
Hub laminate:the equations of this appendix are used to calculate the displacement and stresses in the hub:
]1/4
[
Ah,22 Ah,11 − A2h,12
βh =
4rh2 Dh,11 Ah,11
1
F11 (βh x) = [cosh(βh x)sin(βh x) − sinh(βh x)cos(βh x)]
2
(B1)
F12 (βh x) = sin(βh x)sinh(βh x)
(B2)
1
F13 (βh x) = [cosh(βh x)sin(βh x) + sinh(βh x)cos(βh x)]
2
(B3)
F14 (βh x) = cosh(βh x)cos(βh x)
(B4)
/
1
B11 = [sinh(2βh lh ) − sin(2βh lh )] [sinh2 (βh x) − sin2 (βh x)]
2
(B5)
/
1
B12 = [cosh(2βh lh ) − cos(2βh lh )] [sinh2 (βh x) − sin2 (βh x)]
2
(B6)
B22 = [sinh(2βh lh ) + sin(2βh lh )] / [sinh2 (βh x) − sin2 (βh x)]
(B7)
G11 = −
cosh(βh lh )sin(βh lh ) − sinh(βh lh )cos(βh lh )
sinh2 (βh x) − sin2 (βh x)
(B8)
(B9)
G12 = [2 sinh(βh lh )sin(2βh lh )] / [sinh2 (βh x) − sin2 (βh x)]
12
International Journal of Pressure Vessels and Piping 188 (2020) 104247
A.K. Vafadar et al.
G22 = − 2
cosh(βh lh )sin(βh lh ) + sinh(βh lh )cos(βh lh )
sinh2 (βh x) − sin2 (βh x)
(B10)
Substituting the above equations, the lamina stresses and displacement in the hub can be determined from the following relations:
unh (x) =
Qn1
Mn
ψn
F11 (βh x) + 2 1 F12 (βh x) + 1 F13 (βh x) + un1 F14 (βh x)
βh
2β3h Dh,11
2βh Dh,11
[
ψ nh (x) = βh
Qn1
Mn
F12 (βh x) + 2 1 F13 (βh x) +
3
2βh Dh,11
βh Dh,11
ψ n1
βh
(B11)
]
F14 (βh x) − 2un1 F11 (βh x)
(B12)
]
Qn1
M1n
ψ n1
n
F
(β
x)
+
F
(β
x)
−
F
(β
x)
−
u
F
(β
x)
13
14
11
12
h
h
h
h
1
βh
2β3h Dh,11
2β2h Dh,11
[
Mhn (x) = 2β2h Dh,11
[
Qn1
F14 (βh x) −
2β3h Dh,11
Qn1 (x) = 2β3h Dh,11
M1n
F11 (βh x) −
β2h Dh,11
ψ n1
βh
(B13)
]
(B14)
F12 (βh x) − 2un1 F13 (βh x)
Appendix C. SHELL DETAILS
Shell laminate: the equations of this appendix are used to calculate the displacement and stresses in the shell:
]1/4
[
As,22 As,11 − A2s,12
βs =
4rs2 Ds,11 As,11
(C1)
f1 (βs x) = e−
βs x
[cos(βs x)]
(C2)
f2 (βs x) = e−
βs x
[cos(βs x) − sin(βs x)]
(C3)
f3 (βs x) = e−
βs x
[cos(βs x) + sin(βs x)]
(C4)
f4 (βs x) = e−
βs x
[sin(βs x)]
(C5)
substituting the above equations, the lamina stresses and displacement in the shell can be determined from the following relations:
uns (x) =
Qn2
Mn
f1 (βs x) + 2 2 f2 (βs x) +
3
2βs Ds,11
2βs Ds,11
[
ψ ns (x) = βs −
Qn2
f3 (βs x) −
2β3s Ds,11
As,11 − 12As,12
p
4β4s As,11 Ds,11
+
]
M2n
f1 (βs x)
β2s Ds,11
(C7)
[
Qn2
Mn
f4 (βs x) + 2 2 f3 (βs x)
3
2βs Ds,11
2βs Ds,11
Msn (x) = 2β2s Ds,11
[
Qns (x) = 2β3s Ds,11
(C6)
Qn2
f2 (βs x) −
2β3s Ds,11
M2n
f4 (βs x)
β2s Ds,11
]
(C8)
]
(C9)
References
[9] H. Estrada, I.D. Parsons, Strength and leakage finite element analysis of a GFRP
flange joint, Int. J. Pres. Ves. Pip. 76 (1999) 543–550.
[10] J.M. Campbell, FRP flange for process pipe and tanks,” Proc. FRP Symposium,
National Association of Corrosion Engineers, Niagara Frontier section, USA, 1990.
[11] A.E. Blach, FRP bolted flanged connections subjected to longitudinal bending
moments, in: Computer Aided Design in Composite Material Technology International Conference, 1996, pp. 11–20.
[12] A. Bouzid, A. Chaaban, Flanged joint analysis: a simplified method based on elastic
interaction, CSME Transactions 17 (2) (1993) 181–196.
[13] A. Bouzid, H. Galai, A new approach to model bolted flange joints with full face
gaskets, ASME Journal of Pressure Vessel Technology 133 (2) (2011), 021402-6.
[14] A. Bouzid, A. Nechache, An analytical solution for evaluating gasket stress change
in bolted flange connections subjected to high temperature loading, ASME Journal
of Pressure Vessel Technology 127 (4) (2005) 414–427.
[15] A.K. Vafadar, A. Bouzid, A.D. Ngo, Effect of Material Anisotropy on the Structural
Integrity of Composite Bolted Flanged Joints, Waikoloa, Hawaii, USA, 2017.
[16] J. Czerwinski, V.K. Garikipati, C.N. Jones, B. Pires, J.P. Ludman, D.H. Mccauley,
B. Linnemann, A collaborative approach to achieving a gas-tight seal using
expanded polytetrafluorethylene (EPTFE) gaskets in a fiberglass-reinforced plastic
(FRP) lap-joint flanges, ASME Pressure Vessels and Piping Division, PVP2012 2
(2012) 239–246, 2012.
[1] M. Aljuboury, M.J. Rizvi, S. Grove, R. Cullen, Bolted fibre-reinforced polymer
flange joints for pipelines: a review of current practice and future challenges, Proc
IMechE Part L: J Materials: Design and Applications 233 (8) (2019) 1698–1717.
[2] A.E. Blach, S.V. Hoa, Bolted flange connections for glass fiber reinforced plastic
pipes and pressure vessels, in: Proceedings 11th International Conference on Fluid
Sealing, Cannes, France, 1987, pp. 445–457.
[3] ASME, ASME Boiler and Pressure Vessel Code Section X, Fiberglass reinforced
plastic pressure vessels, New York, 2015.
[4] L. Sun, Bolted Flanged Connections Made of Fiber Reinforced Plastic Materials, Ph.
D. Dissertation, Concordia University, Montreal, Canada, 1995.
[5] A.E. Blach, L. Sun, Fiber reinforced plastic bolted flanged connections, in:
Proceedings 2nd International Symposium on Fluid Sealing, La Baule, France,
September, 1990, pp. 445–457.
[6] A.E. Blach, S.V. Hoa, Effects of pull-back on stresses in FRP flanges, Exp. Tech. 12
(11) (1988) 12s–16s.
[7] A.E. Blach, Fiber reinforced bolted flanged connections discussion of ASME code
rules, in: Proceedings of the International Conference on Pressure Vessel
Technology, ICPVT, Design and Analysis, vol. 2, 1996, pp. 301–306.
[8] H. Kurz, E. Roos, Design of floating type bolted flange connections with GRP
flanges, Int. J. Pres. Ves. Pip. 89 (2012) 1–8.
13