Mechanism and Machine Theory 55 (2012) 34–50

Contents lists available at SciVerse ScienceDirect

Mechanism and Machine Theory

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

Design of a constant-force snap-fit mechanism for minimal

mating uncertainty
Yi-Ho Chen, Chao-Chieh Lan ⁎
Department of Mechanical Engineering, National Cheng Kung University, No. 1, University Road, Tainan City 701, Taiwan, ROC

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

Article history: For assembling injection-molded plastic parts having complicated geometry, snap-fit
Received 5 December 2011 mechanisms are preferable to screws and chemical adhesives. The use of snap-fits relies on
received in revised form 5 April 2012 accurate calculation of interference and the associated mating force. The mating force is
accepted 20 April 2012
proportional to the interference, which is very sensitive to plastic part dimensional error.
Available online 19 May 2012
Insufficient interference would result in loose assembly, whereas excessive interference would
impede assembly and possibly damage the thin-walled plastic parts. The possibility of
Keywords: unqualified interference impairs the high-value application of snap-fits to electronic appliances
Snap-fit connector
and automobiles. To alleviate the requirement for precise interference and to improve the
Constant-force mechanism
assembly's robustness, this paper presents a constant-force snap-fit mechanism that maintains a
Compliant mechanism
Plastic part assembly regular mating force against a range of interference uncertainty. We propose a design formulation
Contact analysis to find mechanism configurations that produce the most constant mating force. Illustrated
simulations and experiments show that the mating force of the constant-force snap-fit is less
sensitive to interference uncertainty than are typical snap-fits. Since uncertain mating force is
minimized without demanding precise interference, we expect this mechanism to provide a
ready alternative to existing snap-fit assembly applications.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction

A snap-fit connector (also known as a fastener or mechanism) is used to directly attach one part to another. It consists of a
cantilever-like structure with a hook at the free end. When a snap-fit connector is attached to one part, the interference of the
hook with another part to be assembled creates a mating force to deflect the cantilever and to interlock the two parts. The use of
snap-fit connectors involves far less assembly and disassembly than would screws, chemical adhesives, or other joining methods.
Snap-fit connectors can operate for many cycles and are cosmetically more attractive. They are often used to assemble complex
injection-molded parts in applications posing stringent space constraints, such as automobile interior parts [1], electronic
appliances [2], and building blocks [3]. For micro-electro-mechanical systems for which no other joining methods have been
devised, micro snap-fit connectors have been developed to join micro 3D structures [4–6].
Strength, constraint, compatibility, and robustness have been identified [7] as the key requirements of snap-fit design. The first
three are essential in the design phase. They can be easily met by following general design guidelines [7–10]. In contrast,
robustness must be met in the assembly phase. Robustness refers to the tolerance of dimensional variation. It is unpredictable
because dimensions vary according to fabrication and assembly conditions. The most critical dimension variation is the
interference between snap-fit hook and mating part. Although very small, the amount of interference determines the mating
force and hence the quality of the snap-fit. Since snap-fit connectors behave like cantilevers, the mating force increases with
increasing interference. Too little interference, and the associated mating force would cause loose assembly. Too much
interference, and the associated mating force would cause difficult assembly or possibly permanent deformation of parts and

⁎ Corresponding author. Tel.: + 886 6 2757575x62274; fax: + 886 6 2352973.

E-mail addresses: solzeta@gmail.com (Y.-H. Chen), cclan@mail.ncku.edu.tw (C.-C. Lan).

0094-114X/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmachtheory.2012.04.006
 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 35

connectors. The failure of parts is more costly. Like many contact-aided mechanisms [11–13], a successful design of snap-fit
connectors requires accurate calculation of interference-induced mating force. Mathematical models of this force are found in
Refs. [14,15].
Even if the interference and associated mating force are accurately reflected in the design, imperfect fabrication and assembly
introduce additional uncertainty. Injection-molded thin-walled parts are usually much larger than snap-fit connectors. Maintaining
their required dimensions is more difficult than snap-fit connectors. One common cause of dimensional error is shrinkage of the part
after molding. The shrinkage of common plastics is nearly 0.5–2% [16] of their original size. This percentage variation seems small, but
the part sizes are large enough to directly alter the amount of interference. For typical interference, a dimensional error of 0.05 mm is
already sufficient to affect the mating force. Thus, an unpredictable dimensional variation has a great influence on mating force.
Previous research has focused on the design of new connector geometry or material to improve assembly and disassembly
[17–19]. A connector geometry design to overcome interference-induced mating uncertainty has not yet been explored. Without
imposing costly precision requirements on injection-molded parts, this paper proposes a constant-force snap-fit connector. It exhibits
a constant mating force against a range of interference uncertainty. The idea is similar to the design of a constant-force mechanism
(e.g., Refs. [20–22]), where an originally bistable mechanism (e.g., Refs. [23–25]) is geometrically modified so that its force to
deflection curve shows an obvious range of constant force. Like constant-force mechanisms, the proposed snap-fit connector is a
closed structure that provides a range of constant reaction force when subjected to an uncertain input displacement. We focus on the
snap-fit connectors used to join vehicle-interior plastic parts, where the quality of assembly can strongly influence a vehicle's market
value.
In what follows, the snap-fit mating force is first characterized through a new analysis method. An optimization formulation is
then presented to design the shape of the constant-force snap-fit connector, the aim of which is to create a mating force that is least
sensitive to interference uncertainty. Finally, prototypes are fabricated and validated experimentally by comparison with typical
snap-fit connectors.

2. Characterization of mating force and analysis of typical snap-fits

2.1. Characterization of mating force

Fig. 1 shows typical snap-fit connectors (hereafter referred as clips) used to assemble automobile-interior parts. The clips can
be metallic or plastic. They are made of a single substance to allow fabrication by injection molding or metal forming. As shown in
Fig. 2(a–b), they are inserted onto the base to provide standard mating motion. The dimensional error (e.g., mold shrinkage) of
the large thin-walled base in the planar direction directly causes interference to deviate. The interference deviation results in
overly loose mating (Fig. 2(c)) or overly tight mating (Fig. 2(d)), depending on the mating configuration.
Fig. 3 shows detailed views of target and clip. The interference δ measures the distance from the tip of the clip to the bottom
surface of the target. During the insertion process, the reaction force from the clip to the target may be decomposed into the x and
y directions. The force in the x direction is denoted as the insertion force (Fi), and the force in the y direction is denoted as the
insertion gripping force (Fgi). Similarly for the retention process, the force in the x direction is denoted as the retention force (Fr),
and the force in the y direction is denoted as the retention gripping force (Fgr). These four are the mating forces. For a specific
interference, they are functions of the relative positions of target and clip. Fig. 4 illustrates the force curves of the retention
process. At the assembled position, there is sufficient Fr and Fgr to lock the parts together despite external disturbance. The forces
Fr and Fgr increase to maximums as the clip moves away from the target. The maximums are denoted as Frm and Fgrm, respectively.
The Fi and Fgi force curves of the insertion process will similarly have maximums, denoted as Fim and Fgim, respectively. A sound
design requires that the maximums of these force curves be kept within a specific range.

Metal

Metal
Bottom
view

Plastic
Top
view
Fig. 1. Typical snap-fit connectors used for vehicle-part assembly.
36 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

a) b)
Base
Target Target
Clip
Dimensional Dimensional
c) error d) error

Fig. 2. (a–b) Standard mating, (c) uncertain mating: too loose, and (d) uncertain mating: too tight.

δ Target
Fi
y Fgi
Bending
Insertion

x
Base

Base Clip

Target δ
Fr
Bending Fgr
Retention
Base

Base Clip

Fig. 3. Insertion and retention processes.

Among the four maximal forces (Frm, Fgrm, Fim, and Fgim), the maximal insertion force Fim should be small so as not to jam the
insertion process. The maximal retention force Frm has to be large enough to prevent separation of the assembled parts. For both
insertion and retention, the maximal gripping forces (Fgrm and Fgim) are along the y direction. They directly cause base bending.
Since the thin-walled base undergoes a larger bending moment than does the clip, it is more vulnerable to failure than the clip.
Base failure is the most common problem in snap-fit assembly, due to excessive gripping force. On the other hand, insufficient
gripping force results in loose assembly.
The four maximal forces depend on δ. Fig. 5(a) shows the maximal force to interference curve of a typical clip. Since a typical
clip is cantilever-like, its maximal forces are proportional to interference. The fabrication and assembly must be very precise in
order to maintain the desired interference and corresponding maximal forces. If there is an interference uncertainty, the maximal
forces will change accordingly. On the other hand, the maximal force to interference curve of a constant-force clip is shown in
Fig. 5(b). There is a constant mating force magnitude over a range of interference. The desired interference can be designed to be

a) F b) F
r gr

Frm Fgrm

Assembled Clip Assembled Clip

position displacement position displacement
Retention force Retention gripping force
Fig. 4. Force curves during retention.
 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 37

a) Frm Fgrm b) Frm Fgrm

Fim F gim Fim F gim

F uncertainty

F uncertainty
Desired Desired

δ δ
δ uncertainty δ uncertainty

Fig. 5. Maximal force to interference curves of (a) a typical clip and (b) a constant-force clip.

located at the midpoint of the constant-force region. As the slope is nearly zero, the maximal force can be kept nearly constant
within a certain variation of interference. Thus, this type of clip is least sensitive to interference uncertainties.

2.2. Analysis of a typical snap-fit mechanism

We consider the solid model of a typical clip in Fig. 6(a). This clip is adopted from those in Fig. 1. To describe its shape, the clip
is first represented as a line model by using its neutral axis, as shown in Fig. 6(b). It is then geometrically divided into seven
segments (B1–B7) separated by eight nodes (n1–n8). Segments B1, B3, B5, and B7 are straight. Every two straight segments are
continuously connected at an arc segment with radius r measured at the neutral axis. The arc segments (B2, B4, and B6) have the
same radius to represent the corner fillets in the solid model, which are required in the manufacturing process. Each of the seven
segments has a rectangular cross-section with constant in-plane thickness w and out-of-plane thickness t. The positions of the
eight nodes are determined by five reference nodes p1–p5. Fig. 6(a) shows the five reference nodes. Nodes p1 and p5 are identical
to n1 and n8, respectively. Nodes p2–p4 determine the positions of n2–n7. The subfigure in Fig. 6(b) shows the relationship
between p4 and n6–n7. By designating the reference node positions, the complete geometry of the clip can be determined.
Detailed dimension and material properties of the typical clip are listed in Table 1.
To compute the maximal forces in insertion and retention, we used the Generalized Multiple Shooting Method (GMSM) [26]. A
detailed formulation is in Appendix A. Fig. 7 shows the computed maximal forces versus interference. Among the four maximal
forces, Fgrm and Fgim are almost linear. They increase with increasing interference. Given that δd = 0.8 mm produces the desired
interference, a ±0.3 mm deviation would cause a ±3.43 N deviation in Fgrm and a ±2.25 N deviation in Fgim, both forces deviating
by nearly ±40% from those at δd = 0.8 mm. The Fgrm and Fgim curves have larger magnitudes, are steeper, and are more sensitive
to interference than are the Frm and Fim curves.
Concluding the discussion on mating forces, the magnitudes of the maximal gripping forces (Fgrm and Fgim) dominate the
quality of snap-fit assembly. However, maintaining desired gripping forces is difficult because they are sensitive to interference. A
robust snap-fit design should minimize the uncertainty of maximal gripping forces by making them insensitive to interference.

a) p4 b) n 7 B6 n6
B7 B5
p5 n8
p4 n5
p3 n6 B4
y n7 n4
r
w B3
n1 n3
p1 x p2 B1 n 2B 2
Solid model Line model
Fig. 6. Model of a typical clip.

Table 1
Parameters of the typical clip.

E = 2 GPa (POM); t = 10 mm; w = 1 mm

r = 1 mm; μ = 0.2
p1 = [0, 0]; p2 = [10, 0] mm; p3 = [11, 3] mm
p4 = [5, 7.5] mm; p5 = [2.5, 5.5] mm
38 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

10
F grm
8

Force (N)
F gim
6
F rm
4

2 F im
0
0.4 0.6 0.8 1 1.1
Interference (mm)

Fig. 7. Maximal-force curves of a typical clip.

3. Design of a constant-force snap-fit mechanism

3.1. Design concept

To obtain Fgrm and Fgim that are as insensitive to interference as possible, they must behave like the curve in Fig. 5(b). We
introduce a constant-force mechanism topology as shown in Fig. 8(a). The two torsional springs provide the positive stiffness, and
the two linear springs provide the negative stiffness. They combine to show a range of constant reaction force when a
displacement Δ is applied. Other topologies of constant-force mechanisms can be found in Ref. [27]. Fig. 8(b) shows the
monolithic analogy of Fig. 8(a) in order for miniaturization. This concept has been used for keyboard dome mechanisms. Fig. 8(c)
shows the F-Δ curve of a constant-force mechanism. Without violating the stress limit, a given input displacement Δ includes the
preload portion from 0 to a1 and the constant-force region from a1 to a2. In the constant-force region, the average force is denoted
as the constant force Fc, and flatness is defined as the percentage of the minimal force Fmin divided by the maximal force Fmax. The
purpose here is to design a constant-force mechanism that can be applied to snap-fit connectors to generate a constant Fgrm and
Fgim in an interference uncertainty range.
Derived from the topology in Fig. 8(b), the proposed constant-force clip is illustrated in Fig. 9. It consists of a closed-loop
homogenous structure enclosed in a rectangular design boundary. As the line model in Fig. 10 shows, the structure is composed of
alternate straight and curved beams. Similar to the typical clip in Fig. 6(b), the shape of the constant-force clip is determined by
seven reference nodes (p1–p7). There are corner fillets at p2–p6. The seven reference nodes determine the positions of nodes n1–
n12 and lengths L1–L11 of segments B1–B11. Like the mechanism in Fig. 8(b), there is an extruded portion (enclosing B5–B7) at the
center of the structure. The extruded portion has height Lh, defined as


Lh ¼ p4y − max p2y ; p3y ; p5y ; p6y ð1Þ

The height Lh needs to be large enough to provide sufficient mating forces. During the mating process, the left side (B5) of the extruded
portion is designated for retention, and the right side (B7) is for insertion.
As shown in Fig. 11, a Δ displacement in the y direction is applied to the clip to represent its maximal deflection due to a
specified interference. Without loss of generality, this displacement is applied on a surface point of B5. Given a clip shape and
desired interference δd = 0.8 mm, the distance between this point and the clip's tip is δd. Its coordinate can be geometrically
obtained from Fig. 11. This is the point of initial contact during the retention process. When given a Δ displacement at this point,

a) Δ c) F Δ=a a2
1

F
Δ
+ +
Δ Constant-force region:
b) a1 ~ a2
Preload region:
F 0 ~ a1
Flatness
Deflection = Fmin/Fmax×100%
Fig. 8. Concept of constant-force mechanisms.
 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 39

Lx
t
p4
Ly
w Lh r
p3 p5
y p2 p6
x p i = [ pix, piy] p7
p1

Fig. 9. Schematic of the constant-force clip design.

n6 n7
B5 B6 B7
B3 B4 B8
B2 n5 n8 B9 B
n3 n4 n9 n 10 10
n2 n11
B1 n i = [nix , n iy] B11
n1 n12

Fig. 10. Node and segment definitions (line model).

the clip responds with reaction force F in the y direction. The force F represents the maximal gripping forces Fgrm and Fgim.
By giving a range of Δ displacement, different interference between the target and clip can be tested. Displacement from Δ = 0.4
to 1.1 mm is then applied to this surface point to obtain the corresponding F. By designing a constant F-Δ curve, we expect
the retention Fgrm-δ and insertion Fgim-δ curves of a constant-force clip to show their force insensitivity to interference.

3.2. Optimization formulation

To make the F-Δ curve of a clip as constant as possible, we need to optimize the clip's initial shape. The optimization is
formulated in Eqs. (2)–(3). The objective in Eq. (2) is to match the values of Fmax and Fmin such that the F-Δ curve is as flat as
possible during Δ = a1–a2. The reference node positions (p1–p7) are the design variables. There are nine constraints in Eq. (3) to
ensure convergence and that the converged optimal shape is feasible. To confine the clip to within the design boundary,
constraints g1–g2 limit the reference node positions. Constraint g3 restricts the angle difference of neighboring straight segments
at each reference node, so as to avoid manufacturing difficulties. Constraints g4–g5 limit the angles of segments B5 and B7, thereby
limiting the retention and insertion forces. To further avoid manufacturing difficulties and ensure computation convergence,
constraint g6 imposes a prescribed length limit on each straight segment. Constraint g7 specifies the minimal height of the
extruded portion. For constraint g8, the maximal insertion force must be no greater than two thirds of the maximal retention
force, in order for the clip to be practical. For constraint g9, the maximal stress σm during the insertion and retention processes
cannot exceed the allowable stress σy/SF, where SF is the safety factor. In each optimization iteration, the force and stress values
required in constraints g8–g9 are obtained by using the computational model (GMSM) presented in Appendix A. The optimiza-
tion is realized by using fmincon() in MATLAB®.

minimize ðF max =F min −1Þ ð2Þ

Tip
n6 n7
In
Retention se
face Δ B6 fa rtio
ce
δd
Target B5 B7
F

Fig. 11. Definition of displacement point.

40 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

Table 2
Constant-force clip design parameters.

E = 2 GPa (Young's modulus of POM) [a1, a2] = [0.5, 1.1] mm (Constant force region)
δd = 0.8 mm (Desired interference) [Lx, Ly] = [20, 10] mm (Size of design boundary)
μ = 0.2 (Friction coefficient) t = 5 mm (Out-of-plane thickness)
SF = 1.5 (Safety factor) w = 0.5 mm (In-plane thickness)
σy = 76 MPa (Yield stress of POM) r = 0.5 mm (Radius of fillet)

8
>
> g1 : 0≤pix ≤Lx ; g 2 : 0≤piy ≤Ly ði ¼ 2e 7Þ
>
>
> −1         
>
>
> g 3 : cos cos ηi ðh1
Þ cos ηiþ2 ð0Þ
þ sin ηi ð1iÞ sin ηiþ2 ð0Þ ≥30∘ði ¼ 1; 3; 5; 7; 9Þ
>
>
>
> −1
> where ηi ¼ tan
> niþ1;y −niy = niþ1;x −nix
>
< h
 i
subject to g4 : 60∘≤ tan −1
n6y −n5y =ðn6x −n5x Þ ≤80∘ ð3Þ
>
> h
 i
>
>
>
> g5 : 135∘≤ tan−1 n7y −n8y =ðn7x −n8x Þ ≤170∘
>
>
> g6 : Li ≥1 mm ði ¼ 1; 3; 5; 7; 9; 11Þ
>
>
>
>
> g7 : Lh > 1:6; 1:8; or 2:0 mm
>
: g : F b2F =3; g : σ bσ =SF
8 im rm 9 m y

3.3. Results and discussion

Table 2 lists the design parameters. Three cases are tested: Lh > 1.6 mm, Lh > 1.8 mm, and Lh > 2.0 mm. Fig. 12 shows the
optimal clip shapes for the three cases. Fig. 13 shows the optimal F-Δ curves, and the respective numerical values are in Table 3.
The initial guess has flatness 76.65% for Δ = 0.5–1.1 mm. The flatnesses of using Lh > 1.6, Lh > 1.8, and Lh > 2.0 mm improve to
94.9%, 95.2%, and 85.0%, respectively. Using Lh > 1.8 mm gains the greatest flatness. The results of using Lh > 1.6 mm are
satisfactory as well, but the height Lh may be short enough to cause manufacturing difficulties. The constant force, maximal stress,
and active constraints are also listed in Table 3. In particular, the flatnesses are limited by constraint g1, which limits the x
coordinates of p6 and p7. This constraint may be relaxed by increasing Lx to obtain a smoother constant-force region, but it would
require a larger clip width.
The discussion of constant-force clip hereafter is based on the case of Lh > 1.8 mm. The coordinates of the reference nodes are
listed in Table 4. The y coordinates of p2, p3, and p5 are identical, due to the active constraint of g7. To verify that the clip indeed
has constant Fgrm-δ and Fgim-δ curves as required, we perform an interference analysis. Fig. 14 shows the maximal-force curves
with respect to different target interference. The typical-clip curves shown are copied from Fig. 7. For the constant-force clip, the
slopes of the maximal gripping-force curves are significantly lowered, remaining almost constant for δ = 0.5–1.1 mm. Table 5
gives the flatnesses of the curves in Fig. 14. The flatness definition is the same as that in Fig. 8(c). By using the constant-force clip,
the flatnesses are more than twice that of the typical clip. At δ = 0.8 mm, Fig. 15 further shows the deformed clip shapes during
insertion and retention. The deformation mainly occurs on B3 and B9. The extruded portion (B5–B7) hardly deforms and can be
treated as if it is rigid. Since this portion is nearly rigid, the optimization can be carried out by considering the Δ displacement to
be applied on any surface point of this portion. The optimal results will be very similar to those formulated in Section 3.2.
Comparing Tables 1 and 4, the constant-force clip is longer (p1 to p7 of 20 mm) than the typical clip (p1 to p3 of 11 mm). Still,
its tip-to-base distance (p1 to p4) is 6 mm, which is very close to that of the typical clip (p1 to p4 of 5 mm). Thus, the effects of Fgrm
and Fgim on the base bending moment are comparable.

2
Lh
0
4
2 3 4 5
2 6
1 Lh 7
4

2
Lh
0
0 5 10 15 20

Fig. 12. Optimal shapes of the constant-force clip.

 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 41

Reaction force (N)

4
Lh > 1.6 mm

2 Lh > 1.8 mm
Lh > 2.0 mm

0
0 0.2 0.4 0.6 0.8 1 1.2
Displacement (mm)

Fig. 13. Optimal F-Δ curves of the constant-force clip.

Table 3
Comparison of optimal results.

Lh (mm) Flatness Fc (N) σm (MPa) Active constraints

Initial guess 76.6% 3.18 43.2

> 1.6 94.9% 5.92 48.1 g1, g7
> 1.8 95.2% 5.04 50.7 g1, g4, g7, g9
> 2.0 85.0% 4.61 50.7 g1, g4, g7, g9

Table 4
Optimal design variables of Lh > 1.8 mm (unit: mm).

p1 = [0, 0] p3 = [4.92, 1.50] p5 = [12.77, 1.50] p7 = [20.0, 0]

p2 = [1.68, 1.50] p4 = [5.96, 3.30] p6 = [20.0, 1.28]

3.4. Effect of friction coefficient

Due to different environmental conditions, the friction coefficient may vary from the nominal 0.2. To study the effect of friction
coefficient, we simulated the maximal-force curves of the typical and constant-force clips using μ = 0, 0.1, and 0.2. As shown in
Figs. 16–17, the Fgim and Fgrm curves of the constant-force clip are insensitive to friction coefficient variations. The flatness and
magnitude are nearly unchanged. For the Fgim curves, the magnitude increases only 0.40 N from μ = 0 to 0.2. In contrast, the Fgim
and Fgrm curves of the typical clip are very sensitive to friction. They increase with increasing friction coefficient. The results in

Fgrm (typ.)
10

8 Fgim (typ.)
Force (N)

6
Fgim (const.)
4
Fgrm (const.)
2
0.4 0.6 0.8 1 1.1
Interference (mm)

Fig. 14. Fgrm-δ and Fgim-δ curves (μ = 0.2).

Table 5
Comparison of flatnesses (δ = 0.5–1.1 mm).

Type Fgim-δ Fgrm-δ

Typical clip 46.07% 35.54%

Constant-force clip 96.36% 92.06%
42 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

Target Insertion

Target Retention

Fig. 15. Constant-force clip deformation (δ = 0.8 mm).

Figs. 16–17 indicate that the constant-force clip is insensitive not only to interference but also to friction coefficients under
various conditions.

3.5. General design guidelines

To use the proposed formulation to design a different constant-force clip, the size of the clip must first be determined
according to the application. The desired interference and reasonable uncertainty range are then specified. Eqs. (2)–(3) are then
used to find the optimal clip shape for producing constant mating forces in the uncertainty range. If the computed or measured
mating force is not as anticipated, the flexural rigidity EI can be adjusted to obtain the required mating force. This is done by
varying the in-plane thickness w, out-of-plane thickness t, or material (E). Fig. 18 illustrates the constant-force clips for vehicle-
interior part assembly. As with the typical clips illustrated in Fig. 1, the constant-force clips are inserted onto the base. Their
working principles are the same except that they provide greater mating-force robustness than do typical clips.

μ=0 μ = 0.1 μ = 0.2

8
Reaction force (N)

F gim of typical clip

6

4
F gim of constant-force clip

2
0.4 0.6 0.8 1 1.2
Interference (mm)

Fig. 16. Effect of friction coefficient on the Fgim curves.

12
μ=0 μ = 0.1 μ = 0.2
10
Reaction force (N)

8 F grm of typical clip

4
F grm of constant-force clip
2
0.4 0.6 0.8 1 1.2
Interference (mm)

Fig. 17. Effect of friction coefficient on the Fgrm curves.

 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 43

Constant-force
snap-fit connector

Base

Fig. 18. Illustration of constant-force clips for vehicle-part assembly.

4. Experiments

4.1. Verification of optimization results

The constant-force clip of case Lh > 1.8 mm is converted into a solid model in Fig. 19. It is then fabricated as a prototype in order
to obtain the actual F-Δ curve. The dimensions and material (POM) used are exactly the same as those described in Tables 2 and 4.
Fig. 19 shows the experimental setup. A target with a force sensor (FUTEK LSB200) approaches the retention face of the clip. Along
the y direction, the target deflects the clip from Δ = 0 to 1.1 mm in order to measure the reaction force F. Fig. 19 shows the target
at the initial contact point (Δ = 0). This loading condition exactly matches the design requirement in Section 3.2. Fig. 20 compares
the experimental and simulation results. The experiment is conducted three times to obtain the error bars. In the constant-force
region, the experimental curve is roughly 0.7 N higher than the simulation curve. The inconsistency between simulation and
experiment is primarily a result of imperfections in fabrication and material. Nevertheless, the experimental curve is not sensitive
to these imperfections and still shows high flatness (91.59%). In practice, the clip will move laterally along a target. The experiment in
Section 4.2 studies this constant-force behavior subject to different interference.

4.2. Validation of insertion and retention processes

To validate the constant-force clips in the insertion and retention processes, a prototype with two mirror-imaged constant-
force clips was fabricated, as shown in Fig. 21(a). The two clips, same as that in Section 4.1, are joined to the base in order to
conveniently measure the Fgim-δ and Fgrm-δ curves. In practice, the clips can be made detachable from the base. For comparison, a
prototype with two mirror-imaged typical clips was also made, as shown in Fig. 21(b). The typical-clip dimensions are based on
the parameters in Table 1. Figs. 22 and 23 show the experimental setups of the typical and constant-force clips, respectively. The

Force
sensor δd = 0.8 mm
Target

y
x Constant-force clip

Fig. 19. Prototype of constant-force clip.

Reaction force (N)

2 GMSM
Experiment
0
0 0.2 0.4 0.6 0.8 1 1.1
Displacement (mm)

Fig. 20. Experimental F-Δ curve of the constant-force clip.

44 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

a) b)

Base Base

7.5 mm
6 mm
20 mm
11 mm
Fig. 21. Prototypes of constant-force and typical clips.

Linear stage Linear stage

Target

Target
Gripping
force
sensor

Typical clip
Insertion and
x retention force
sensor Motorized
y stage

Fig. 22. Experimental setup for the typical clip.

clip-base assembly is connected to two force sensors (5 lb each) to measure the insertion and retention forces. They are mounted
together on a motorized stage. The left and right targets are separately mounted on a manual linear stage. The two manual linear
stages move perpendicularly to the motorized stage in order to provide different interference. Two force sensors (25 lb each) are
connected to the left target to measure the gripping force. The experiment begins by setting a proper interference between the
targets and clip. The motorized stage then moves the clip against the targets, emulating the insertion and retention processes. The
configurations shown in Figs. 22–23 are at the beginning of insertion, with δ = 0.8 mm.
Figs. 24–25 show the gripping forces during the retention process. Both the typical clip and the constant-force clip were tested
for δ = 0.6, 0.8, and 1.0 mm. For the typical clip, the gripping-force curves significantly vary by interference, especially at the

Linear stage Linear stage

Gripping
Target
Target

force
sensor

Constant-force clip

Insertion and
x retention force
sensor Motorized
y stage

Fig. 23. Experimental setup for the constant-force clip.

 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 45

12
F grm
10

Gripping force (N)

F grm
8
6 F grm
4
δ = 1.0 mm
2 δ = 0.8 mm
0 δ = 0.6 mm

0 1 2 3 4 5 6 7
Clip displacement (mm)

Fig. 24. Retention process for typical clip.

maximums. The Fgrm of δ = 1.0 mm is almost twice that of δ = 0.6 mm. As for the constant-force clip in Fig. 25, the gripping-force
curves are almost the same regardless of interference. Hence, not only does the constant-force clip makes the maximal gripping
force constant, but the gripping forces at other clip displacements are also made invariant of target interference.
The same tests were conducted for δ = 0.4–1.1 mm, with 0.05 mm step size. The results are plotted in Figs. 26–29. Specifically,
Figs. 26 and 27 show the maximal-force curves of the typical clip in the insertion and retention process, respectively. They are
compared with the results computed using the GMSM. In the GMSM simulation, the experimentally determined friction
coefficients are 0.178 for the insertion and 0.197 for the retention process. The experimental curves are slightly higher than the
simulation curves. Nonetheless, the Fgim and Fgrm curves are still linear, showing sensitivity to interference. The maximal
difference between δ = 0.4 and 1.1 mm is nearly 5.10 N for Fgim and 8.32 N for Fgrm. Thus, the gripping forces experience major
variation with uncertain interference. Unlike the maximal gripping-force curves, the maximal retention and insertion force curves
are less sensitive to interference. Their magnitudes are also smaller than the maximal gripping-force curves.
Figs. 28–29 show the maximal-force curves of the constant-force clip. They are also compared with GMSM curves, in which the
experimentally determined friction coefficients are 0.200 for the insertion and 0.218 for the retention process. The experimental
Fim and Frm curves closely match the simulation. The experimental Fgim and Fgrm curves are 0.9 N higher than their simulated
values. Nevertheless, they show very good flatness between δ = 0.4 and 1.1 mm. The force variation is merely 1.38 N for Fgim and
1.11 N for Fgrm. The variations mainly occur between δ = 0.4 and 0.6 mm. The forces remain nearly constant during δ = 0.6 and
1.1 mm.
Without the constant-force clip, extra manufacturing and assembly cost is required to control the precise interference
between target and clip. In contrast, our proposed constant-force clip ensures the functionality of part mating by providing a
constant gripping force under a sufficient range of interference. Even when the clip surface friction condition varies, the constant-
force property is still valid as long as the interference variation is within the design range.

5. Conclusions

This paper has presented a constant-force clip to overcome snap-fit mating uncertainty. The mating forces of typical clips are
sensitive to interference. As uncertain interference is unavoidable, the use of typical clips may result in overly tight or overly loose
assembly. Instead of interference control, the proposed constant-force clip inherently provides constant mating forces against a
range of interference uncertainty. The success of the clip relies on shape optimization of its cantilever-like structure. Based on the
developed analysis and optimization formulation, the optimized clip can exhibit nearly constant mating forces within ±0.3 mm
of interference variation. The clip has been verified by comparing the simulation and experimental results with those of a typical

6
F grm
Gripping force (N)

2 δ = 1.0 mm
δ = 0.8 mm
δ = 0.6 mm
0
0 2 4 6 8
Clip displacement (mm)

Fig. 25. Retention process for constant-force clip.

46 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

8 F gim (Exp.)

Force (N)
F gim (GMSM)

4
F im (Exp.)

2
F im (GMSM)

0.4 0.6 0.8 1 1.1

Interference (mm)

Fig. 26. Maximal-force curves during insertion (typical clip).

12
F grm (Exp.)
10
Force (N)

8 F grm (GMSM)

6 F rm (GMSM)

4 F rm (Exp.)

2
0.4 0.6 0.8 1 1.1
Interference (mm)

Fig. 27. Maximal-force curves during retention (typical clip).

6 F gim (Exp.)
Force (N)

5 F gim (GMSM)

4
F im (Exp.)
3
F im (GMSM)
2
0.4 0.6 0.8 1 1.1
Interference (mm)

Fig. 28. Maximal-force curves during insertion (constant-force clip).

7 F rm (Exp.)

6
Force (N)

F rm (GMSM)
5
F grm (Exp.)
4
F grm (GMSM)
3

2
0.4 0.6 0.8 1 1.1
Interference (mm)

Fig. 29. Maximal-force curves during retention (constant-force clip).

 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 47

clip. Having similar dimensions and force magnitudes to a typical clip, the proposed constant-force clip can readily replace typical
clips to provide better robustness for snap-fit assembly.

Acknowledgments

This work was supported by the Ministry of Education, Taiwan, R.O.C. under the NCKU Project of Promoting Academic Excellence &
Developing World Class Research Centers, and by Yen Tjing-Ling Industrial Technology Research and Development Center.

Appendix A. Beam model and contact-force computation

The clip contact forces required in the analyses in Sections 2 and 3 are computed using the GMSM [26]. This method is capable of
accurate and efficient beam large deflection computation. Without loss of generality, we illustrate this method by considering the
typical clip in the insertion process. The clip beam model will be presented first, followed by the formulation of constraint equations
that obtain the contact forces. The formulation can be readily extended to the retention process and clips of other shapes.

A.1. Beam model and constraint equations

For the typical clip in Fig. 6, each segment is modeled as a homogenous beam. As an illustration, Fig. 30 shows the beam
models of segments B1 and B2. Each beam has a rectangular cross-section with in-plane thickness w and out-of-plane thickness
t. The flexural rigidity is denoted as EI, where E is Young's modulus and I = tw 3/12 is the second moment of area. The un-
deflected beam is described by a shape function η(u), where u ∈ [0 1] is a non-dimensional arc length along its neutral axis and
η measures the angle of rotation (or slope, in radians) along u. The shape functions of B1 (straight) and B2 (arc) are respectively
represented as

B1 : η1 ðuÞ ¼ c10 ; B2 : η2 ðuÞ ¼ c20 þ c21 u ðA1Þ

where the first subscript denotes the segment number, coefficient c10 is the angle of the straight beam, and coefficients c20–c21
determine the shape of the arc beam with radius of curvature L2/c21.
External loads change the shape of a beam from the initial η(u) to the deflected ψ(u). The beam reacts with internal force h
(horizontal), internal force v (vertical), and bending moment M to balance external loads at the two ends (u = 0 and u = 1). The
differential equations governing the beam deflection are

½ ψ ψ ′ x y ′ ¼ ½ ψ ′ ðA2Þ
/
E I ðh sinψ−v cosψÞ þ η ″ L cosψ L sinψ
L2

where a prime denotes derivative with respect to u, [x y] is the coordinate of a point on the beam, and L is the beam length. For the
typical clip, each intermediate node (n2–n7) has six constraint equations that must be satisfied. For example, the constraint equations at
n2 are

Angle continuity : ψ1 ð1Þ−η

01 ð1Þ ¼ ψ 2 ð0
Þ−η2 ð0Þ
0 
0 0
Moment balance : EI1  ψ1 ð1Þ−η1 ð1Þ =L1 ¼ EI 2  ψ2 ð0Þ−η2 ð0Þ =L2
ðA3Þ
Displacement continuity : x1 ð1Þ−x2 ð0Þ ¼ 0; y1 ð1Þ ¼ y2 ð0Þ ¼ 0
Force balance : h1 ¼ h2 ; v1 ¼ v2

The constraint equations of the grounded node n1 and free node n8 are

n1 : ψ1 ð0Þ−η1 ð0Þ ¼ 0; x1 ð0Þ ¼ 0; y1 ð0Þ ¼ 0

0 0 ðA4Þ
n8 : ψ7 ð1Þ−η7 ð1Þ ¼ 0; h7 ¼ 0; v7 ¼ 0

Corresponding to the constraint equations at the nodes, each beam has six unknowns. For example, the unknowns for B2 are x2(0),
y2(0), ψ2(0), ψ′2(0), h2, and v2. The total number of unknowns must match the total number of constraint equations. The numerical
technique to solve for the unknowns using the constraint equations is in Ref. [26]. After obtaining the unknowns, the deflected

Fig. 30. Beam models of segments B1 and B2.

48 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

shape of the clip can be obtained by integrating the differential equations in Eq. (A2). The bending moment and maximal bending
stress are further calculated as follows.

EI dðψ−ηÞ
M¼ ðA5Þ
L du

Ew dðψ−ηÞ
σm ¼ ðA6Þ
2L du

A.2. Contact computation

During the insertion process, the clip deflects when it contacts the target. Since the target can be represented as a point, the
contact analysis is simplified to a point-contact analysis. At the point of contact, corresponding constraint equations must be
formulated. Since clips are made of homogenous slender beams, we assume that the local deformation on the surface of the beam
is negligible when compared with the global beam deflection. Thus, contact point and force are considered to occur on the rigid
surface of the clip.
Consider the case where the contact point is right on the surface of the infinitesimal segment at n6. Fig. 31 shows the free-body
diagram. The normal and tangential contact forces from the clip to the target are denoted as Fn and Ft, respectively. To balance the
contact force from the target, they are resultants of the beam internal forces (h5, h6, v5, and v6).
  −1

Fn cosðψ5 ð1Þ−π=2Þ cosðψ5 ð1Þ−πÞ h6 −h5
¼ ðA7Þ
Ft sinðψ5 ð1Þ−π=2Þ sinðψ5 ð1Þ−π Þ v6 −v5

The frictional torque induced by Ft on the infinitesimal segment is very small and is neglected. The normal and tangential contact
forces must satisfy the following equations.

F t ¼ μF n ðA8aÞ

2
Ft ¼ s ðA8bÞ

B6
v6
Fn Target
v7
M6(1 ) h5 F
n7 h7
n6 t h
h6 6
w
M5(1) M6(0)
M7 (0)
v7 v5 B5
v6 Fn
B7
n8 Ft u
h7 w ψ 5(1)-π /2 n5
M7(1) n6
Infinitesimal segment at n6

Fig. 31. Free-body diagram of the insertion process.

5 Fi (GMSM) F gim
Fgi (GMSM)
4
Fi (FEA)
Force (N)

3 Fgi (FEA)

2
F im
1

0
0 0.5 1 1.5 2
Target displacement (mm)

Fig. 32. Insertion-force curves.

 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50 49

a) Target b) Target

Maximal Maximal insertion

insertion force gripping force

Fig. 33. Contact deformation during insertion.

Table 6
Comparison of computation results (δ = 0.8 mm, μ = 0).

Method Fim (N) Fgim (N) Time (s)

GMSM 1.66 4.96 5.9

FEA 1.60 4.91 1738

Eq. (A8a) is Coulomb's dry friction law, where μ is the coefficient of kinetic friction. Eq. (A8b) uses variable s to ensure that the
direction of the tangential contact force Ft is opposite the known direction of the moving target. The interference condition
requires that the surface point of n6 be lowered to be the same height as the bottom surface of the target.

y5 ð1Þ þ w sinðψ5 ð1Þ−π=2Þ=2

  ðA9Þ
¼ n6y þ r  sin η5 ð1Þ þ π=2 þ 1 þ w=2−δ

Except for the contact-related equations in Eqs. (A8)–(A9), continuity and moment balance equations must also be satisfied on
the infinitesimal segment at n6.

x5 ð1Þ ¼ x6 ð0Þ; y5 ð1Þ ¼
y6 ð0Þ; ψ5 ð1Þ−η5 ð1Þ ¼ ψ6 ð0Þ−η6 ð0Þ; M 5

ð1Þ ¼ M 6 ð0Þ

ðA10Þ
where M5 ð1Þ ¼ EI 5  ψ′ 5 ð1Þ−η′ 5 ð1Þ =L5 and M6 ð0Þ ¼ EI 6  ψ′ 6 ð0Þ−η′ 6 ð0Þ =L6

Eqs. (A8)–(A10) form the seven constraint equations required at n6. This is one more equation than would be associated with the
typical non-contact node, in order to solve for s. Together with the constraint equations at other nodes, they can be solved by
using the GMSM. After obtaining the internal forces, the insertion force and insertion gripping force are calculated as

F i ¼ h6 −h5 ; F gi ¼ v6 −v5 ðA11Þ

When the contact point is on B5 or B6 and not on the surface of n6, the corresponding segment needs to be divided into two sub-
segments at the contact point, in order to create an extra node. Similar constraint equations can then be formulated at that node.
Using the formulation in Eqs. (A1)–(A11), we perform a simulation where the target approaches the clip with δ = 0.8 mm and
μ = 0. This is to emulate the insertion process. Fig. 32 shows the computed insertion and insertion gripping-force curves. The
insertion force increases to its maximum Fim and then decreases. The maximum occurs when the contact point is on the surface
point of n6. Fig. 33(a) shows the simulation of this point. The insertion gripping force increases and reaches its maximum Fgim. It
stays at the maximum while the bottom surface of the target slides through the clip. The inception of Fgim occurs when the contact
point is on the surface of B6. Fig. 33(b) shows the simulation of the inception point. For comparison, finite element analysis (FEA)
results are also provided. The FEA is implemented in ANSYS® with PLANE82 as the deformation element and with TARGE169 and
CONTA172 as the contact elements. The FEA curves are very close to the GMSM curves. Table 6 lists the maximal forces computed
by using the GMSM and FEA. The minor difference is due to the use of solid geometry (FEA) and line geometry (GMSM) to model
deformation. However, the computation using the GMSM is significantly more efficient than FEA. To obtain the Fim-δ and Fgim-δ
curves, one simply has to compute the maximal forces at every δ.
By using a similar formulation, the retention and retention gripping forces (Fr and Fgr) during the retention process can also be
obtained. The contact point during the retention process may occur on the surface of B6 or B7.

References

[1] P.A. Tres, Designing Plastic Parts for Assembly, Hanser Gardner Publications, Cincinnati, OH, USA, 2006.
[2] M. Shalaby, K. Saitou, High-stiffness, lock-and-key heat-reversible locator-snap systems for the design for disassembly, ASME Journal of Mechanical Design
131 (4) (2009) 041005.
[3] P. Limayea, G. Ramua, S. Pamulapati, G.K. Ananthasuresh, A compliant mechanism kit with flexible beams and connectors along with analysis and optimal
synthesis procedures, Mechanism and Machine Theory (2011), http://dx.doi.org/10.1016/j.mechmachtheory.2011.07.008.
50 Y.-H. Chen, C.-C. Lan / Mechanism and Machine Theory 55 (2012) 34–50

[4] N. Dechev, L. Ren, W. Liu, L. Cleghorn, J.K. Mills, Development of a 6 degree of freedom robotic micromanipulator for use in 3D MEMS microassembly, IEEE
International Conference on Robotics and Automation, Orlando, FL, 2006, pp. 281–288.
[5] S. Bargiel, K. Rabenorosoa, C. Clévy, C. Gorecki, P. Lutz, Towards micro-assembly of hybrid MOEMS components on a reconfigurable silicon free-space micro-
optical bench, Journal of Micromechanics and Microengineering 20 (4) (2010) 045012.
[6] B.-H. Jang, H.-Y. Huang, W. Fang, A novel zero-insertion-force (ZIF) micro (μ)-connector: design, fabrication, and measurements, IEEE Transactions on
Industrial Electronics 56 (4) (2009) 1040–1047.
[7] P.R. Bonenberger, The First Snap-Fit Handbook: Creating Attachments for Plastics Parts, Hanser Gardner Publications, Cincinnati, OH, USA, 2005.
[8] S. Genc, R.W. Messier Jr., G.A. Gabriele, A systematic approach to integral snap-fit attachment design, Research in Engineering Design 10 (1998) 84–93.
[9] G. Suri, A.F. Luscher, Evaluation metrics for the rating and optimization of snap-fits, Research in Engineering Design 12 (2000) 191–203.
[10] J.M. Brock, P.K. Wright, Design tool for injection molded snap fits in consumer products, Journal of Manufacturing Systems 21 (1) (2002) 32–39.
[11] J.R. Cannon, L.L. Howell, A compliant contact-aided revolute joint, Mechanism and Machine Theory 40 (11) (2005) 1273–1293.
[12] Y.-M. Moon, Bio-mimetic design of finger mechanism with contact aided compliant mechanism, Mechanism and Machine Theory 42 (5) (2007) 600–611.
[13] V. Mehta, M. Frecker, G.A. Lesieutre, Stress relief in contact-aided compliant cellular mechanisms, ASME Journal of Mechanical Design 131 (9) (2009)
091009.
[14] C.-C. Lan, K.-M. Lee, An analytical contact model for design of compliant fingers, ASME Journal of Mechanical Design 130 (1) (2008) 011008.
[15] K. Jorabchi, K. Suresh, Nonlinear algebraic reduction for snap-fit simulation, ASME Journal of Mechanical Design 131 (6) (2009) 061004.
[16] E.A. Campo, The Complete Part Design Handbook: For Injection Molding of Thermoplastics, Hanser Gardner Publications, Cincinnati, OH, USA, 2006.
[17] B. Willems, W. Dewulf, J.R. Duflou, Active snap-fit development using topology optimization, International Journal of Production Research 45 (18) (2007)
4163–4187.
[18] K.-N. Chen, F.-T. Gau, C.-C. Chu, H.-C. Cheng, Structural optimization of contact springs of electrical connectors using response surface methodology,
International Conference on Electronic Materials and Packaging (EMAP), 2008, pp. 239–242.
[19] J. Carrell, D. Tate, S. Wang, H.-C. Zhang, Shape memory polymer snap-fits for active disassembly, Journal of Cleaner Production 19 (17-18) (2011)
2066–2074.
[20] C. Boyle, L.L. Howell, S.P. Magleby, M.S. Evans, Dynamic modeling of compliant constant-force compression mechanisms, Mechanism and Machine Theory 38
(12) (2003) 1469–1487.
[21] J.C. Meaders, C.A. Mattson, Optimization of near-constant force springs subject to mating uncertainty, Structural and Multidisciplinary Optimization 41 (1)
(2010) 1–15.
[22] C.-C. Lan, J.-H. Wang, Y.-H. Chen, A compliant constant-force mechanism for adaptive robot end-effector operations, IEEE International Conference on
Robotics and Automation, Anchorage, AK, USA, 2010, pp. 2131–2136.
[23] J. Tsay, H.-A. Chang, C.-K. Sung, Design and experiments of fully compliant bistable micromechanisms, Mechanism and Machine Theory 40 (1) (2005) 17–31.
[24] P.A. Halverson, L.L. Howell, S.P. Magleby, Tension-based multi-stable compliant rolling-contact elements, Mechanism and Machine Theory 45 (2) (2010)
147–156.
[25] M.A. Pucheta, A. Cardona, Design of bistable compliant mechanisms using precision-position and rigid-body replacement methods, Mechanism and Machine
Theory 45 (2) (2010) 304–326.
[26] C.-C. Lan, Y.-J. Cheng, Distributed shape optimization of compliant mechanisms using intrinsic functions, ASME Journal of Mechanical Design 130 (7) (2008)
072304.
[27] P. Alabuzhev, A. Gritchin, L. Kim, G. Migirenko, V. Chon, P. Stepanov, Vibration Protecting and Measuring Systems with Quasi-zero Stiffness, Hemisphere
Publishing, New York, 1989.