Professional Documents
Culture Documents
A New Analytical Method For Press-Fit Curve Prediction of Interference Fitting Parts
A New Analytical Method For Press-Fit Curve Prediction of Interference Fitting Parts
net/publication/317693147
CITATIONS READS
9 2,434
4 authors, including:
All content following this page was uploaded by Xingyuan Wang on 04 April 2019.
A new analytical method for press-fit curve prediction of interference fitting MARK
parts
⁎
Xingyuan Wanga, Zhifeng Loua,b, Xiaodong Wanga,b, , Chonglin Xua
a
Key Laboratory for Micro/Nano Technology and System of Liaoning Province, Dalian University of Technology, Dalian 116024, China
b
Key Laboratory for Precision and Non-traditional Machining of Ministry of Education, Dalian University of Technology, Dalian 116024, China
A R T I C L E I N F O A B S T R A C T
Keywords: Press-fit assembly is one of the traditional methods for assembly of interference fitting parts. But the connection
Press-fit assembly quality cannot be obtained directly from this method. At present, the press-fit curve is employed for quality
Interference fitting parts estimation, and thick-walled cylinder theory (TCT) is used for standard press-fit curve prediction. However, the
Press-fit curve prediction results have a large relative error which may reach to larger than 18% for neglecting the influence of
Resistant force
non-contact regions. This paper aims to build an accurate analytical method to predict the press-fit curve under
Analytical method
the linear elastic hypothesis. The finite element method (FEM) was used to analyze the influence of non-contact
regions, and the results show that the non-contact regions generate a constant resistant force in the press-fit
process. To establish the analytical method, a simplified model was proposed. The calculation method of re-
sistant force was deduced based on the simplified model, and the analytical method was established based on
both of TCT and the resistant force calculation method. The analytical results were compared with numerical
results and validated by experimental results. The results show that the new analytical method can predict the
press-fit curve accurately, and it is more efficient than FEM in design of interference fitting parts.
1. Introduction the container, and interference were investigated in detail. For the
friction coefficient is also an important factor for load bearing ability,
An interference fit is obtained by inserting a shaft into a hole with a Seifi and Abbasi (2015) estimated the friction coefficient in shaft/bush
slightly smaller diameter. To accommodate this interference, the shaft interference fit by an updated FE model. However, in practice, there is a
must contract radially, and the hole expands, producing a radial contact form error in the contact surface which will lead to a weak interference
stress at the interface (Selvage, 1978). Recently, interference fit is fit. The research of Boutoutaou et al. (2011) also supports that form
commonly used in various fields including aerospace, energy, agri- defects have a negative effect on the assembly strength and can't be
culture, transportation and medicine for its high ability of transmitting neglected in intensive tightening. Furthermore, connection perfor-
an axial force or torque between a shaft and hub with the help of mance is also affected by operating conditions such as separation fre-
friction effects (Yang et al., 2002). quencies. Kovan (2011) founds that the separation frequency has a
Press-fit assembly is one of the traditional methods for assembly of great influence on stress distribution of interference fit assembly be-
interference fitting parts, and its assembly quality has attracted the tween hollow shaft and hub, and the FEM gives more accurate results
attention of many researchers. Generally, the strength of assemblies than thick-walled cylinder theory.
depends on various parameters such as amount of interference, material The researches mentioned above all focus on a finished interference
properties, physical dimensions, geometrical irregularities of mating fit. But in practice, it’s unreasonable to measure the roughness and form
members and friction coefficient of contact surfaces (Seifi and Abbasi, error of each part, and then estimate the actual interference value and
2015). Hence, many studies have been performed to investigate the simulate the contact status. Therefore, it’s necessary to pay attention to
effect of various parameters on the load bearing ability of interference the assembly process to estimate the assembly quality. Kleditzsch et al.
fits. Application of finite element methods (FEM) to simulate and in- (2014, 2015) made a detailed numerical investigation on the joining
vestigate interference joints is widely spread. Teng et al. (2012) ana- process and load characteristics of steel-aluminum knurled interference
lyzed the stress distribution of cylindrical interference fit by FEM. The fits, and the results showed that the joining process is significant for
influencing factors such as joint diameter, joint width, outer diameter of axial strength and the transmission behavior of the connection.
⁎
Corresponding author at: Key Laboratory for Micro/Nano Technology and System of Liaoning Province, Dalian University of Technology, Dalian 116024, China.
E-mail address: xdwang@dlut.edu.cn (X. Wang).
http://dx.doi.org/10.1016/j.jmatprotec.2017.06.022
Received 29 January 2017; Received in revised form 16 June 2017; Accepted 17 June 2017
Available online 19 June 2017
0924-0136/ © 2017 Elsevier B.V. All rights reserved.
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
2.1. Finite element model number of the elements and nodes are 68859 and 299345 respectively.
The maximum skewness is 0.5278. Boundary conditions are shown in
To predict the press-fit curve, a 3D FE model was set up. This model Fig. 5.
was realized with the finite element software, ANSYS Workbench. For the presented FE model, the interference value is defined by
The design of the simulation conforms to the principle of the axial setting an offset of mating surface. The materials of part A and part B
press-fit assembly (Fig. 2). The geometric dimension of part A and part are Ni36CrTiAl (3J1) and 50Ni-50Fe (1J50) respectively. The material
B is shown in Fig. 3. In order to obtain accurate results, it is necessary to parameters of Ni36CrTiAl and 50Ni-50Fe are shown in Table 1 (You
mesh the model with small element size (0.1 mm). Considering the et al., 2015). The friction coefficient between part A and part B is as-
small mesh size and axisymmetric characteristic, only a quarter of the sumed as 0.09 which is calculated based on experimental results and
connection was modeled with hexahedral elements (Fig. 4). The
17
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Goodier, 1951):
Where r1, internal radius of part A; r2, contact radius; r3, external radius
of part B; E1, Elastic Modulus of part A; μ1, Poisson’s ratio of part A; E2,
Elastic Modulus of part B; μ2, Poisson’s ratio of part B.
Since the interference value equals the total deformation of mating
Fig. 4. A quarter of finite element model. surface of part A and part B. So the following equation can be obtained.
Z = u1 + u2 (3)
Material Young’s Poisson’s ratio Yield Tangential As shown in Fig. 7, press-fit curves under different interference
modulus strength modulus (MPa)
values were obtained by TCT and numerical simulation. Obviously the
(GPa) (MPa)
press-mounting force obtained by numerical simulation has a sudden
Ni36CrTiAl 157 0.34 686 942 change at the initial stage. Because at the beginning stage of press-fit
50Ni-50Fe 186 0.3 882 1116 process, part A and part B transit from non-contact to contact state, so
the press-mounting force increases sharply to a certain value. Then the
press-mounting force increases almost linearly with displacement. In
TCT, and the calculating process will be introduced in Section 4.2.
the final stage, the increment of press-mounting force becomes smaller.
Furthermore, another important phenomenon that a basically un-
2.2. Thick-walled cylinder theory changed deviation exists between the TCT and numerical results were
observed, and the relative error is larger than 18%.
For the parts are pressed together quasi-statically under constant In the press-fit process, the flange of part B may have an effect on
velocity (0.5 mm/s), the press-fit curve is usually predicted by thick- the press-mounting force. So it’s necessary to analyze the influence of
walled cylinder theory (TCT). For example, the external radius of part A the flange. The press-fit curves with and without flange were obtained
is larger than the internal radius of part B by the amount Z. According from numerical simulation and are shown in Fig. 8. Obviously, there is
to the definitions and parameters (Fig. 6), the contact pressure P, the a small deviation between these two curves. The relative error of the
decrement of external radius of part A u1, and the increment of internal maximum press-mounting force is about 3%, so the influence of the
radius of part B u2, u1 and u2 can be calculated by TCT (Timoshenko and flange is small enough to be ignored, and the unchanged deviation is
Fig. 6. Contact behavior of part A and part B. Fig. 7. Comparison between the results of TCT and numerical simulation.
18
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Fig. 8. Comparison between press-fit curves with and without flange. Fig. 10. Normal stress comparison of the contact surface of part A (interference value
12 μm, contact length 2.3 mm).
19
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
wedge (Fig. 12c). The stress components in any points of the wedge is as For the symmetry condition, the deformation on the x-axis in the
follows (Chen, 2007) direction of ϕ is zero.
(uφ)φ = 0 = 0 (12)
⎪ ρ (
⎧ σρ = − 2Fc cos β cos ϕ + sin β sin ϕ
α + sin α α − sin α )
σ ϕ = 0 Substituting Eq. (12) into (11), H and K can be obtained.
⎨
⎪ τρϕ = τϕρ = 0 (6) H = 0; K = 0
⎩
Where, Fc is two times of linear load. So Eq. (11) becomes as follows:
When the central angle of the wedge equals π, the wedge will turn 2F (1 − μ ′) F
⎧uρ = − πEc′ cos ϕ ln ρ − πE′ c ϕ sin ϕ + I cos ϕ
into a semi-infinitely large plate. Since the concentrated force is per-
pendicular to the straight boundary, β equals to zero. So Eq. (6) will ⎨uϕ = 2Fc sin ϕ ln ρ + (1 + μ ′) Fc sin ϕ − (1 − μ ′) Fc ϕ cos ϕ − I sin ϕ
⎩ πE′ πE′ πE′
becomes as
(13)
2F cos ϕ
⎧ σρ = − πc ρ The second function of Eq. (13) can be used to calculate the vertical
⎪
σϕ = 0 deformation of any point on the boundary. uϕ is positive in the direction
⎨
⎪ τρϕ = τϕρ = 0 of ϕ.
⎩ (7)
2Fc (1 + μ′) Fc
Physical equations and geometric equations under polar coordinates − (uφ)φ = π = − ln ρ − +I
2 πE′ πE ′ (14)
are as follows (Chen, 2007):
uϕ is unable to be calculated for unknown I. But the relative deforma-
1
⎧ ερ =E' ρ
(σ − μ′σϕ) tion between two points can be calculated by Eq. (14). As shown in
⎪ 1 Fig. 13, the y coordinates of point B and M are s and ρ respectively. B is
εϕ = E ' (σϕ − μ′σρ)
⎨ set as the reference point. So the relative deformation between B and M
⎪ γ = 1 τρϕ = 2(1 + μ ′) τρϕ in the vertical direction can be obtained.
⎩ ρϕ G E' (8)
2Fc (1 − μ2 ) s
⎧ ερ = ∂ρ
ρ ∂u η= ln
πE ρ (15)
⎪
⎪ uρ 1 ∂u ϕ
εϕ = ρ + ρ ∂ϕ According to Eq. (15), it is necessary to determine the optimal s. For
⎨
⎪ 1 ∂uρ ∂u ϕ uϕ different material, geometry and load, the reference point is different.
⎪ γρϕ = ρ ∂ϕ + ∂ρ − ρ (9) So the reference points of part A and part B need to be determined
⎩
respectively.
In Eq. (8), E′ and μ′ are elastic modulus and Poisson's ratio in plane In order to determine the reference point of part B, a numerical
strain respectively. ερ, εφ and γρφ are strain components of point A. simulation was implemented (Fig. 14). As shown in Fig. 15, the de-
E μ formation distribution under different s was obtained by Eq. (15) and
E′ = ; μ′ =
1 − μ2 1−μ compared with numerical results. Since the resistant region has a lim-
ited length, the deformation distribution within ρ = 1.5 mm was used
Substituting Eqs. (7) and (8) into Eq. (9), the following equation can
to determine the optimal reference point. From the comparison be-
be obtained.
tween the analytical and numerical results, the optimal reference point
∂uρ 2F cos ϕ is sB = 14 mm. With the same method, the reference point of part A was
⎧
∂ρ
= - πEc′ ρ
⎪
⎪ uρ 1 ∂u ϕ 2μ ′ F cos ϕ
+ ρ ∂ϕ = πE ′c ρ
⎨ρ
⎪ 1 ∂uρ ∂u ϕ uϕ
⎪ ρ ∂ϕ + ∂ρ − ρ = 0 (10)
⎩
Deformation components of arbitrary point can be obtained by
solving Eq. (10):
2F (1 − μ ′) F
⎧uρ = − πE′ cos ϕ ln ρ − πE′ c ϕ sin ϕ + I cos ϕ + K sin ϕ
⎪ (1 + μ ′) F (1 − μ ′) F
2F
u = sin ϕ ln ρ + πE c sin ϕ − πE c ϕ cos ϕ + Hρ
⎨ ϕ πE′ ′ ′
⎪ − I sin ϕ + K cos ϕ
⎩
(11)
In Eq. (11), I, H, K are constants. Fig. 13. The principle of relative deformation.
20
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Fig. 14. Simplified model of deformation simulation: (a) part B, (b) part A.
πEη
ρmin = s⋅exp ⎛⎜− ⎞⎟
⎝ 2Fc (1 − μ2 ) ⎠ (16)
ρmin of part A and part B under different concentrated forces are
shown in Table 2 and Table 3 respectively.
As shown in Tables 2 and 3, the position of equivalent load point
under different concentrated forces is constant. ρmin_A and ρmin_B are
2.2 μm and 3.3 μm respectively. Therefore the concentrated force
generated by the non-contact regions of part A and part B can be cal-
culated by the following equations.
1 πηA E1
FLA = Fc A = sA
2 4(1 − μ12 )ln ρ (17)
min A
1 πηB E2
FLB = Fc B = sB
2 4(1 − μ 22 )ln ρ (18)
min B
F = Ff + Fr (20)
4. Results
21
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Table 4
Friction coefficients under different interference value.
22
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Fig. 23. Comparison of the resistant forces between experimental and analytical results.
Fig. 20. Comparison between experimental and analytical results.
Fig. 23. The results show that the resistant force increases with the
increase of interference value, and the analytical results are consistent
with experimental results with the relative error less than 12%. This
error may be caused by the form error and stress concentration. Because
the form error will affect the real interference value; the stress con-
centration may cause the damage of the chamfer and finally makes the
real interference value smaller.
Therefore, the new analytical method can be used to calculate the
resistant force with an acceptable relative error and has a high accuracy
to predict the press-fit curve.
5. Conclusion
References
Fig. 22. Comparison between different constraint types (Z = 8 μm). Bae, J.H., Kim, J.S., Hwang, B.C., Bae, W.B., Kim, M.S., Kim, C., 2012. Prediction of the
dimensional deformation of the addendum and dedendum after the warm shrink
results. In practice, the maximum press-mounting force can be reduced fitting process using a correction coefficient. Int. J. Automot. Tech-Kor 13, 285–291.
Benuzzi, D., Donzella, G., 2004. Prediction of the press-fit curve in the assembly of a
by reducing the friction coefficient at the locating surface. railway axle and wheel. Proc. Inst. Mech. Eng. F–J. Rail Rapid Transit 218, 51–65.
To verify the validity of the new analytical method, the resistant Boutoutaou, H., Bouaziz, M., Fontaine, J.F., 2011. Modeling of interference fits taking
force was also obtained by experiments. Since the resistant force re- from defects of the surfaces in contact into account. Mater. Des. 21, 3692–3701.
Chen, M.X., 2007. Elasticity and Plasticity. Science Press Co.Ltd, Beijing, pp. 110–111.
mains unchanged in the assembly process, the press-mounting force is Kleditzsch, S., Lätzer, M., Awiszus, B., Leidich, E., 2014. Numerical investigation of
mainly caused by resistant force when the contact region has a small knurled shaft-hub connections and especially of the joining process. Mater. Sci.
length. At the beginning of the assembly process, part A and part B Forum 773–774, 18–27.
Kleditzsch, S., Awiszus, B., Lätzer, M., Leidich, E., 2015. Numerical and analytical in-
transit from non-contact to contact region and the contact length starts vestigation of steel-aluminum knurled interference fits: joining process and load
to increase. So when the contact length is equal to 0.04 mm, the press- characteristics. J. Mater. Process. Technol. 219, 286–294.
mounting force is regarded as resistant force. The comparison of re- Kovan, V., 2011. Separation frequency analysis of interference fitted hollow shaft-hub
connections by finite element method. Adv. Eng. Softw. 42, 644–648.
sistant forces between experimental and analytical results is shown in
23
X. Wang et al. Journal of Materials Processing Tech. 250 (2017) 16–24
Lou, Z.F., Wang, X.D., You, B., Xu, Y., 2015. Pressing-fitting technology and instrument method of cylindrical interference fit. J. Mech. Eng. 48, 160–166.
for precision small parts. Opt. Precis. Eng. 23 (6), 16605–16611. Timoshenko, S., Goodier, J.N., 1951. Theory of Elasticity, second ed. McGraw-Hill Book
Ozturk, F., 2011. Finite-element modelling of two-disc shrink fit assembly and an eva- Company, New York, pp. 58–60.
luation of material pairs of discs. Proc. Inst. Mech. Eng. Part C: J. Mech. Eng. Sci. 225, Yang, G.M., Coquille, J.C., Fontaine, J.F., Lambertin, M., 2002. Contact pressure between
263–273. two rough surfaces of a cylindrical fit. J. Mater. Process. Technol. 123, 490–497.
Seifi, R., Abbasi, K., 2015. Friction coefficient estimation in shaft/bush interference using You, B., Luo, Y., Wang, X.D., 2014. The application of computing methods for analysis of
finite element model updating. Eng. Fail. Anal. 57, 310–322. press-fit assembly in elasto-plastic field. J. Inf. Comput. Sci. 11 (9), 3157–3166.
Selvage, C.C., 1978. Assembly of Interference Fits by Impact and Constant Force Method. You, B., Lou, Z.F., Luo, Y., Xu, Y., Wang, X.D., 2015. Prediction of pressing quality for
Massachusetts Institute of Technology, Massachusetts, pp. 9–10. press-fit assembly based on press-fit curve and maximum press-mounting force. Int. J.
Teng, R.J., Zhang, Y.B., Zhou, X.J., Xu, Z.N., 2012. Mechanical properties and design Aerosp. Eng. 2015, 823019.
24