Professional Documents
Culture Documents
Tribology International
journal homepage: www.elsevier.com/locate/triboint
a r t i c l e in f o a b s t r a c t
Article history: Although stuffing boxes are old systems used to ensure stem valve sealing, the analytical developments
Received 2 December 2007 of the stresses and the displacements generated during assembly and operation are very limited and
Received in revised form seldom verified and the studies carried out on these devices are either restricted or not accessible.
16 November 2008
Moreover, even with the evolution of calculation and simulation means, studies based on numerical
Accepted 2 February 2009
Available online 12 February 2009
models are rare.
This work proposes a simplified analytical approach, using the theory of thick-walled cylinders to
Keywords: analyse the stresses and displacements in stuffing box systems. The magnitude and distribution of the
Packed glands lateral contact pressures generated at the housing–packing–stem interfaces as a result of the application
Contact pressure
of the gland axial stress are determined as a function of the radial flexibility of the different components
Analytical modeling
involved. The results of the developed approach are compared and validated against the more accurate
finite elements axisymmetric models.
& 2009 Elsevier Ltd. All rights reserved.
0301-679X/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.triboint.2009.02.002
ARTICLE IN PRESS
friction coefficients at the two interfaces: the flexibility of the different components that interact in a
stuffing box packing assembly. The packing and the housing are
Ki m0 R2 3R2 þ 5R1
¼ (1) treated using thick cylinder theory whereas the stem is con-
K0 mi R1 5R2 þ 3R1
sidered as a solid cylinder subjected to compression. Fig. 2 shows
A more rigorous calculation [7] of the moments gives Eq. (2) the axisymetrical model used to treat the problem.
which is slightly different from Eq. (1): The stem and the housing box are made from steel with the
characteristics as summarized in Table 1. While for the FEA study
Ki m0 R2 5R2 þ 7R1
¼ (2) the packing made of graphite material is modeled with a hyper
K0 mi R1 7R2 þ 5R1
elastic behaviour characterized by the Mooney–Rivlin constants
Both methods do not allow calculating Ki and K0 separately, but C1 and C2, calculated from the experimental compression curves
compare their values. This is independent of the load, the given in [8], the analytical model considers an elastic behaviour
dimensions and the mechanical characteristics of the stem and with an elasticity modulus given by Eq. (3) [9]:
the housing.
In this paper, it is proposed to calculate the lateral pressure E1 ¼ 2ð1 þ n1 ÞG1 ¼ 4ð1 þ n1 ÞðC 1 þ C 2 Þ (3)
coefficients and their ratio using an approach based on the theory
of thick walled cylinders. These coefficients will be used to For the stem subjected to a radial pressure qi, the stresses are
determine the distributions of axial stress and axial displacements given by
of the packing. The effect of width and number of the packing
ssl ðR1 Þ ¼ 0
rings on the packing behaviour will be examined. The analytical
ssr ðR1 Þ ¼ qi
solution will be compared to the more accurate FEA.
ssy ðR1 Þ ¼ qi (4)
3. Analytical model Using Hooke’s law, the radial displacement at the stem outside
diameter is
An analytical model that calculates the lateral pressure
R1 ð1 n2 Þ
coefficients Ki and K0 is presented in this section. It considers us ðR1 Þ ¼ qi (5)
E2
Stem Axis
qi qo
σx
qi qo
σ Gland
Housing
Packing
x
dx
R2
Packings
(1) Housing
Stem
R1 (2) (3)
Table 1 or
Stuffing-box components material characteristics. " ! #
E1 Y 21 1
Stem Packing Housing 1 þ Y 21 n1 Y 21 þ n1 þ ð1 þ Y 2
3 þ n Y
3 3
2
n3 q0
Þ
E3 Y 23 1
Subscript 2 1 3 þ 2qi þ n1 ðY 21 1Þsx ¼ 0 (13)
Ri (mm) 14.29 23.81
R0 (mm) 14.29 23.81 33.75 The system of Eqs. (12) and (13) can be written in a matrix form
Y 1.67 1.42
a11 2Y 21 n1 ðY 21 1Þsx
" #" # " #
E (MPa) 200 000 126 200 000 qi
n 0.3 0.4 0.3 ¼ (14)
2 a22 q0 n1 ðY 21 1Þsx
where
For the housing considered as a cylinder subjected to an internal
pressure, the stresses at the inside diameter are given by E1
a11 ¼ 1 n1 þ Y 21 þ n1 Y 21 þ ð1 n2 ÞðY 21 1Þ
E2
shl ðR2 Þ ¼ 0
and
shr ðR2 Þ ¼ q0 !
1 þ Y 23 E1 Y2 1
h
sy ðR2 Þ ¼ q0 (6) a22 ¼ 1 þ n1 þ Y 21 n1 Y 21 þ ð1 n3 þ Y 23 þ n3 Y 23 Þ 12
E3 Y3 1
Y 23 1
The solution of this system is
Using Hooke’s law, the housing radial displacement at inside
diameter is a22 2Y 21
qi ¼ n1 ðY 21 1Þsx (15)
2
4Y 21 a11 a22
R2 ð1 n3 þ ð1 þ n 3 ÞY 3 Þ
uh ðR2 Þ ¼ q0 (7) and
ðY 23 1ÞE3
2 a11
For the packing considered as a cylinder subjected to internal and
q0 ¼ n1 ðY 21 1Þsx (16)
4Y 21 a11 a22
external pressures, the stresses at the inside and outside radii are
given by Lamé as The lateral pressure coefficients are, therefore,
2Y 21 1 þ Y 21 and
spy ðR1 Þ ¼ q0 þ qi (8)
Y 21 1 Y 21 1 2 a11
K0 ¼ n1 ðY 21 1Þ (18)
4Y 21 a11 a22
spl ðR2 Þ ¼ sx Finally, the lateral pressure coefficient ratio is given by
spr ðR2 Þ ¼ q0 Ki q a22 2Y 21
1þ Y 21 2 ¼ i ¼ (19)
spy ðR2 Þ ¼ q0 þ qi (9) K 0 q0 2 a11
Y 21 1 Y 21 1
For the calculation of the packing axial displacements, an approxi-
The displacements of the packing at the inside and outside radii mate method based on an elastic packing behaviour is proposed.
using Hooke’s law are Knowing the packing stresses at any given axial position x,
Hooke’s law may be used to give the axial displacement as
R1
up ðR1 Þ ¼ ½2Y 21 q0 þ ð1 þ Y 21 þ n1 Y 21 n1 Þqi dw 1
ðY 21 1ÞE1 l ¼ ¼ ½s n1 ðsy þ sr Þ (20)
2
dx E1 l
þn 1 ðY 1 1Þsx (10)
At the packing ID, using the conditions of Eqs. (8), this
deformation is given by
R2
up ðR2 Þ ¼ ½ð1 þ Y 21 n1 Y 21 þ n1 Þq0 þ 2qi " !#
ðY 21 1ÞE1 dw 1 1 þ Y 21 2Y 21
l ¼ ¼ sx n1 qi þ 2 qi 2 q0 (21)
þ n1 ðY 21 1Þsx (11) dx E1 Y1 1 Y1 1
or Therefore,
E1 dw
2Y 21 qo þ ½1 þ Y 21 þ n1 Y 21 n1 þ ð1 n2 ÞðY 21 1Þqi ¼ asD ebx (23)
E2 dx
2
þn 1 ðY 1 1Þsx ¼ 0 (12)
with
At R2: " #
1 2n1
a¼ 1þ 2 ðK i Y 21 K 0 Þ (24)
up ðR2 Þ ¼ uh ðR2 Þ E1 Y1 1
ARTICLE IN PRESS
asD Ki K0 Ki/K0
wðxÞ ¼ ½ebl ebx (25)
b
Eq. (1) – – 1.470
Eq. (2) – – 1.533
Analytical model 0.649 0.653 0.994
MEF 0.600 0.591 1.016
4. Finite element model
Housing
0.6
0.55 35 MPa
Packing
0.5 50 MPa
0.45
0 1 2 3 4 5 6 7
Number of packing rings
The values of Ki and K0 calculated by the proposed model are Figs. 7 and 8 show the effect of the gland stress and the friction
also validated with those obtained experimentally by Bartonicek coefficient on the axial displacement using FE analysis and the
et al. [10]. analytical model for 5 and 6 packings, respectively. These figures
The finite element analysis allowed a further look onto the show that for a given vertical position, when the gland stress
effect of the applied gland stress, the friction coefficient and the increases or the friction coefficient decreases, the displacement
number of packing rings on the lateral pressure coefficients. Fig. 5 increases. The displacement increases from almost zero at
presents the results of such effects on the inner lateral pressure the lower packing surface to reach a maximum value at the
coefficient Ki. For the same axial stress, the effect of the friction upper surface. Although these figures show a similar trend,
coefficient is less significant since at most less than 10% is the comparison with the analytical model is clearly indicated in
observed with a 50 MPa gland stress. Of particular importance is Figs. 9 and 10 for gland stresses of 35 and 50 MPa. It is to be noted
the lateral pressure coefficient decrease with increased gland that the difference between the two methods depends on the
stress. Nevertheless, in all cases the lateral pressure coefficient friction coefficient and the gland stress. The scatter is wider at
ratio was found to approach unity as indicated in Fig. 6. This ratio the mid-axial position and is about 20% when sD ¼ 50 MPa and
remains constant at all times even after some relaxation has taken m ¼ 0.25. For all other lower friction coefficients and axial packing
place [11]. stresses, the difference is not significant.
Although the hyper elastic material defined by the Moon-
ey–Rivlin model is the nearest available Ansys material that
1.01
Friction coefficient
Axial stress
1.005 0.15 7 σD
0.175 Rectangle : 35 MPa
Lateral pressure coefficients ratio, Ki/Ko
Triangle : 10 MPa
0.20 Star : 50 MPa
1 6
0.25
Axial displacement, w (mm)
50 MPa
0.995 5 Friction coefficient
Solid line : 0.15
0.99 Dotted line : 0.20
4
Dashed line : 0.25 35 MPa
0.985
3
0.98
2
0.975
1 10 MPa
0.97
0 1 2 3 4 5 6 7 0
Number of packing rings 20 30 40 50 60 70 80 90
Axial position (mm)
Fig. 6. Lateral pressure coefficient ratio.
Fig. 8. Analytical axial displacements for six packing rings.
6 σD 5
σD
4.5
5
Axial displacement, w (mm)
1
1 10 MPa
0.5
0 0
20 30 40 50 60 70 80 20 30 40 50 60 70 80 90
Axial position (mm) Axial position (mm)
Fig. 7. FE axial displacements for five packing rings. Fig. 9. Analytical and FE packing axial displacements for sD ¼ 50 MPa.
ARTICLE IN PRESS
20
6 σD
Friction coefficient 10 qi
4 0.20 5 μ = 0.15
0.25 35 MPa
σD
3 0
Solid Line : FEA
Dashed line : analytical model
Number of packing rings
-5
2
2 3 4
qo
-10 5 6 7
1 8
-15
0
20 30 40 50 60 70 80 90 -20
Axial position (mm) 20 40 60 80 100 120
Axial position (mm)
Fig. 10. Analytical and FE packing axial displacements for sD ¼ 35 MPa.
Fig. 12. Radial contact stresses qi and q0.
6 20
σD σD
5 σD = 35 MPa 18
Radial contact pressure, qi (MPa)
Axial displacement, w (mm)
Friction coefficient
16
4 σD = 35 MPa
Solid line : 0.15 μ = 0.15
Dotted line : 0.2
Dashed line : 0.25 14
3
Number of packing rings
12
8
2
6 10 Numbers packing rings
3
2 3
1 4 5
8
6 7
8
0 6
20 40 60 80 100 20 40 60 80 100
Axial position (mm) Axial position (mm)
Fig. 11. FE packing axial displacement for sD ¼ 35 MPa. Fig. 13. Superimposed radial contact pressure distributions.
characterizes packing behaviour, its constants are determined by a The effect the number of packing rings on the contact pressure
tri-axial stress test and is only considered for the loading case. distributions of qi and q0 along the vertical position is also shown
Notwithstanding, it is believed that the shear effect not accounted in Fig. 12. It is evident that, for a given position, both interface
for by the analytical model is what causes this difference. In contact stresses increase with a decrease of the number of packing
addition, the representativeness of the packing material behaviour rings. The curves show a substantial decrease of contact stress at
by Mooney–Rivlin model can be questionable to a certain extend. the opposite side of the gland in the case of high number of
Recent study [12,13] suggests that porous bodies such as packing rings. Besides being a source of leakage paths, this low
exfoliated graphite packing are modeled using modified Cam–Clay compressed region may provide space for relaxation to take place.
material behaviour which in general can describe soil and rock Figs. 13 and 14 show the superposition of the contact pressure
because such materials consolidate under pressure and are prone distribution curves of different number of packing configurations.
to shear. Nevertheless, for the purpose of validation of the A change in the coordinate system is conducted to bring the
developed analytical model during initial loading, the Moon- contact surface between the gland and the first packing at the
ey–Rivlin model is considered acceptable. Fig. 11 shows the axial same axial position. The resulting outcome is a perfect overlap
displacement for deferent number of packing rings and friction of the eight curve cases with all the points laying on one curve
coefficients for the same axial stress. with the exception of a few points that are the subject of end
ARTICLE IN PRESS
5
σD = 10 MPa
μ = 0.20 Acknowledgements
4
The authors wish to express their gratitude to the Natural
Sciences and Engineering Research Council of Canada for partially
3
financing this study. They also wish to thank Dr. Le Van Ngan and
Henry Champliaud for their help and generosity in sharing their
2 scientific knowledge.
Numbers packing rings
2 3 References
1
5 6
7 8 [1] ISO 15848-1. Industrial valves—measurement, test and qualification proce-
dures for fugitive emissions—part 1: classification system and qualification
0 procedures for type testing of valves; 2006.
10 30 50 70 90 110 [2] ISO 15848-2. Industrial valves—measurement, test and qualification proce-
Axial position (mm) dures for fugitive emissions—part 2: production acceptance test of valves;
2003.
Fig. 14. Radial contact stress at packing-housing interface for sD ¼ 10 MPa and [3] API 622. Type testing of process valve packing for fugitive emissions, 1st ed.
m ¼ 0.20. 2006.
[4] Denny DF. A force analyse of the stuffing-box seal. Research report 550, BHRA,
the Fluid Engineering Centre, Harlow, England; 1957.
[5] Ochonski W. Radial stress distribution and friction forces in a soft-packed
effects. These curves could be used to select the number of stuffing-box seal. Tribology International 1988;21(1):31–8.
packing rings required to guaranty the sealability of the stuffing [6] Pengyun S, Kuangmin C, et al. Theoretical analysis of the lateral pressure
box packing system based on an acceptable criterion. For example, coefficients in a soft-packed stuffing-box seal. Tribology International
1997;30(10):759–65.
the number of packing can be limited to those having a contact [7] Diany M, Bouzid A. Evaluation of contact stress in stuffing box packings. In:
stress above a multiple value of the fluid pressure. 2006 ASME-PVP conference, paper no. PVP2006-ICPVT11-93083, Vancouver,
British Columbia; 2006.
[8] Klenk T, Kockelmann H, Roos E, Bartonicek J, Schoeckle F. Characteristics and
testing techniques for stuffing box packings, vol. 382. 1999 ASME Pressure
6. Conclusion
Vessels and Piping Division, PVP; 1999. p. 135–43.
[9] Ansys. Standard manual, version 10. Ansys Inc.; 2005.
An analytical model based on thick cylinder theory has been [10] Bartonicek J, Schoeckle F. Approach to a correct function of stuffing boxes.
Computer technology—1996: applications and methodology, vol. 326.
developed to model the stem-packing-housing interaction. The American Society of Mechanical Engineers, Pressure Vessels and Piping
model gives reliable contact pressure distributions at the inter- Division (publication) PVP; 1996. p. 115–21.
faces and can predict the deformations involved. The results from [11] Diany M, Bouzid A. Analytical modelling of creep relaxation of stuffing box
packing. In: 2007 ASME-PVP conference, paper no. PVP2007-26094, San
the developed model agree with those of FEA and experimental
Antonio, TX, USA; 2007.
studies of the literature. The model suggests that the contact [12] Nı́ Ruaidhe M, Torrance AA. Developing a finite element model of exfoliated
pressure ratio is close to one, while the interface contact pressure graphite valve packings. In: Proceeding of the 17th international conference
depends on several parameters including geometry, materials and on fluid sealing; 2003. p. 95–109.
[13] Roe M, Torrance A. Performance prediction of exfoliated graphite seals: 1.
friction. The model has potential in predicting contact stress as a Determination of parameters and verification of model. Sealing Technology
function of geometry, materials, friction coefficient and load and 2008;4:6–13.